Set LD_LIBRARY_PATH globally in all workflow jobs/steps

This should also fix installation problem of node-gyp in CI

.github/workflows/build-all-versions.yml

@@ -14,7 +14,7 @@ jobs:
     strategy:
       matrix:
         os: [ubuntu-latest, macos-latest, windows-latest]
-        cabal: ["3.2"]
+        cabal: ["latest"]
         ghc:
           - "8.6.5"
           - "8.8.3"
@@ -33,7 +33,7 @@ jobs:
     - uses: actions/checkout@v2
       if: github.event.action == 'opened' || github.event.action == 'synchronize' || github.event.ref == 'refs/heads/master'
 
-    - uses: actions/setup-haskell@v1.1.4
+    - uses: haskell/actions/setup@v1
       id: setup-haskell-cabal
       name: Setup Haskell
       with:
@@ -65,7 +65,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        stack: ["2.3.3"]
+        stack: ["latest"]
         ghc: ["7.10.3","8.0.2", "8.2.2", "8.4.4", "8.6.5", "8.8.4"]
         # ghc: ["8.8.3"]
 
@@ -73,11 +73,12 @@ jobs:
     - uses: actions/checkout@v2
       if: github.event.action == 'opened' || github.event.action == 'synchronize' || github.event.ref == 'refs/heads/master'
 
-    - uses: actions/setup-haskell@v1.1.4
+    - uses: haskell/actions/setup@v1
       name: Setup Haskell Stack
       with:
-        # ghc-version: ${{ matrix.ghc }}
-        stack-version: ${{ matrix.stack }}
+        ghc-version: ${{ matrix.ghc }}
+        stack-version: 'latest'
+        enable-stack: true
 
     - uses: actions/cache@v1
       name: Cache ~/.stack
@@ -90,6 +91,6 @@ jobs:
         stack build --system-ghc --stack-yaml stack-ghc${{ matrix.ghc }}.yaml
         # stack build --system-ghc --test --bench --no-run-tests --no-run-benchmarks
 
-    # - name: Test
-    #   run: |
-    #     stack test --system-ghc
+    - name: Test
+      run: |
+        stack test --system-ghc --stack-yaml stack-ghc${{ matrix.ghc }}.yaml

.github/workflows/build-binary-packages.yml

@@ -2,7 +2,8 @@ name: Build Binary Packages
 
 on:
   workflow_dispatch:
-  release:
+  release: 
+    types: ["created"]
 
 jobs:
 
@@ -10,11 +11,13 @@ jobs:
 
   ubuntu:
     name: Build Ubuntu package
-    runs-on: ubuntu-18.04
-    # strategy:
-    #   matrix:
-    #     ghc: ["8.6.5"]
-    #     cabal: ["2.4"]
+    strategy:
+      matrix:
+        os:
+        - ubuntu-18.04
+        - ubuntu-20.04
+
+    runs-on: ${{ matrix.os }}
 
     steps:
     - uses: actions/checkout@v2
@@ -53,19 +56,33 @@ jobs:
     - name: Upload artifact
       uses: actions/upload-artifact@v2
       with:
-        name: gf-${{ github.sha }}-ubuntu
+        name: gf-${{ github.event.release.tag_name }}-${{ matrix.os }}.deb
         path: dist/gf_*.deb
         if-no-files-found: error
 
+    - name: Rename package for specific ubuntu version
+      run: |
+        mv dist/gf_*.deb dist/gf-${{ github.event.release.tag_name }}-${{ matrix.os }}.deb
+
+    - uses: actions/upload-release-asset@v1.0.2
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+      with:
+        upload_url: ${{ github.event.release.upload_url }}
+        asset_path: dist/gf-${{ github.event.release.tag_name }}-${{ matrix.os }}.deb
+        asset_name: gf-${{ github.event.release.tag_name }}-${{ matrix.os }}.deb
+        asset_content_type: application/octet-stream
+
 # ---
 
   macos:
     name: Build macOS package
-    runs-on: macos-10.15
     strategy:
       matrix:
         ghc: ["8.6.5"]
         cabal: ["2.4"]
+        os: ["macos-10.15"]
+    runs-on: ${{ matrix.os }}
 
     steps:
     - uses: actions/checkout@v2
@@ -92,19 +109,33 @@ jobs:
     - name: Upload artifact
       uses: actions/upload-artifact@v2
       with:
-        name: gf-${{ github.sha }}-macos
+        name: gf-${{ github.event.release.tag_name }}-macos
         path: dist/gf-*.pkg
         if-no-files-found: error
+    
+    - name: Rename package
+      run: |
+        mv dist/gf-*.pkg dist/gf-${{ github.event.release.tag_name }}-macos.pkg
+
+    - uses: actions/upload-release-asset@v1.0.2
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+      with:
+        upload_url: ${{ github.event.release.upload_url }}
+        asset_path: dist/gf-${{ github.event.release.tag_name }}-macos.pkg
+        asset_name: gf-${{ github.event.release.tag_name }}-macos.pkg
+        asset_content_type: application/octet-stream
 
 # ---
 
   windows:
     name: Build Windows package
-    runs-on: windows-2019
     strategy:
       matrix:
         ghc: ["8.6.5"]
         cabal: ["2.4"]
+        os: ["windows-2019"]
+    runs-on: ${{ matrix.os }}
 
     steps:
     - uses: actions/checkout@v2
@@ -136,16 +167,18 @@ jobs:
         cp /mingw64/bin/libpgf-0.dll /c/tmp-dist/c
         cp /mingw64/bin/libgu-0.dll /c/tmp-dist/c
 
+    # JAVA_HOME_8_X64 = C:\hostedtoolcache\windows\Java_Adopt_jdk\8.0.292-10\x64
     - name: Build Java bindings
       shell: msys2 {0}
       run: |
-        export PATH="${PATH}:/c/Program Files/Java/jdk8u275-b01/bin"
+        export JDKPATH=/c/hostedtoolcache/windows/Java_Adopt_jdk/8.0.292-10/x64
+        export PATH="${PATH}:${JDKPATH}/bin"
         cd src/runtime/java
         make \
-          JNI_INCLUDES="-I \"/c/Program Files/Java/jdk8u275-b01/include\" -I \"/c/Program Files/Java/jdk8u275-b01/include/win32\" -I \"/mingw64/include\" -D__int64=int64_t" \
+          JNI_INCLUDES="-I \"${JDKPATH}/include\" -I \"${JDKPATH}/include/win32\" -I \"/mingw64/include\" -D__int64=int64_t" \
           WINDOWS_LDFLAGS="-L\"/mingw64/lib\" -no-undefined"
         make install
-        cp .libs//msys-jpgf-0.dll /c/tmp-dist/java/jpgf.dll
+        cp .libs/msys-jpgf-0.dll /c/tmp-dist/java/jpgf.dll
         cp jpgf.jar /c/tmp-dist/java
 
     - name: Build Python bindings
@@ -157,7 +190,7 @@ jobs:
         cd src/runtime/python
         python setup.py build
         python setup.py install
-        cp /usr/lib/python3.8/site-packages/pgf* /c/tmp-dist/python
+        cp /usr/lib/python3.9/site-packages/pgf* /c/tmp-dist/python
 
     - name: Setup Haskell
       uses: actions/setup-haskell@v1
@@ -180,6 +213,18 @@ jobs:
     - name: Upload artifact
       uses: actions/upload-artifact@v2
       with:
-        name: gf-${{ github.sha }}-windows
+        name: gf-${{ github.event.release.tag_name }}-windows
         path: C:\tmp-dist\*
         if-no-files-found: error
+
+    - name: Create archive
+      run: |
+        Compress-Archive C:\tmp-dist C:\gf-${{ github.event.release.tag_name }}-windows.zip
+    - uses: actions/upload-release-asset@v1.0.2
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+      with:
+        upload_url: ${{ github.event.release.upload_url }}
+        asset_path: C:\gf-${{ github.event.release.tag_name }}-windows.zip
+        asset_name: gf-${{ github.event.release.tag_name }}-windows.zip
+        asset_content_type: application/zip

.github/workflows/build-majestic.yml

@@ -0,0 +1,162 @@
+name: Build majestic runtime
+
+on: push
+
+env:
+  LD_LIBRARY_PATH: /usr/local/lib
+
+jobs:
+
+  ubuntu-runtime:
+    name: Runtime (Ubuntu)
+    runs-on: ubuntu-20.04
+
+    steps:
+    - uses: actions/checkout@v2
+
+    - name: Build runtime
+      working-directory: ./src/runtime/c
+      run: |
+        autoreconf -i
+        ./configure
+        make
+        sudo make install
+
+    - name: Upload artifact
+      uses: actions/upload-artifact@master
+      with:
+        name: libpgf-ubuntu
+        path: |
+          /usr/local/lib/libpgf*
+          /usr/local/include/pgf
+
+  ubuntu-haskell:
+    name: Haskell (Ubuntu)
+    runs-on: ubuntu-20.04
+    needs: ubuntu-runtime
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Download artifact
+      uses: actions/download-artifact@master
+      with:
+        name: libpgf-ubuntu
+    - run: |
+        sudo mv lib/* /usr/local/lib/
+        sudo mv include/* /usr/local/include/
+
+    - name: Setup Haskell
+      uses: haskell/actions/setup@v1
+
+    - name: Build & run testsuite
+      working-directory: ./src/runtime/haskell
+      run: |
+        cabal test --extra-lib-dirs=/usr/local/lib
+
+  ubuntu-python:
+    name: Python (Ubuntu)
+    runs-on: ubuntu-20.04
+    needs: ubuntu-runtime
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Download artifact
+      uses: actions/download-artifact@master
+      with:
+        name: libpgf-ubuntu
+    - run: |
+        sudo mv lib/* /usr/local/lib/
+        sudo mv include/* /usr/local/include/
+
+    - name: Install bindings
+      working-directory: ./src/runtime/python
+      run: |
+        python setup.py build
+        sudo python setup.py install
+
+    - name: Run testsuite
+      working-directory: ./src/runtime/python
+      run: |
+        pip install pytest
+        pytest
+
+  ubuntu-javascript:
+    name: JavaScript (Ubuntu)
+    runs-on: ubuntu-20.04
+    needs: ubuntu-runtime
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Download artifact
+      uses: actions/download-artifact@master
+      with:
+        name: libpgf-ubuntu
+    - run: |
+        sudo mv lib/* /usr/local/lib/
+        sudo mv include/* /usr/local/include/
+
+    - name: Install dependencies
+      working-directory: ./src/runtime/javascript
+      run: |
+        npm ci
+
+    - name: Run testsuite
+      working-directory: ./src/runtime/javascript
+      run: |
+        npm run test
+
+# ----------------------------------------------------------------------------
+
+  macos-runtime:
+    name: Runtime (macOS)
+    runs-on: macOS-11
+
+    steps:
+    - uses: actions/checkout@v2
+
+    - name: Install build tools
+      run: |
+        brew install \
+          autoconf \
+          automake \
+          libtool \
+
+    - name: Build runtime
+      working-directory: ./src/runtime/c
+      run: |
+        glibtoolize
+        autoreconf -i
+        ./configure
+        make
+        sudo make install
+
+    - name: Upload artifact
+      uses: actions/upload-artifact@master
+      with:
+        name: libpgf-macos
+        path: |
+          /usr/local/lib/libpgf*
+          /usr/local/include/pgf
+
+  macos-haskell:
+    name: Haskell (macOS)
+    runs-on: macOS-11
+    needs: macos-runtime
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Download artifact
+      uses: actions/download-artifact@master
+      with:
+        name: libpgf-macos
+    - run: |
+        sudo mv lib/* /usr/local/lib/
+        sudo mv include/* /usr/local/include/
+
+    - name: Setup Haskell
+      uses: haskell/actions/setup@v1
+
+    - name: Build & run testsuite
+      working-directory: ./src/runtime/haskell
+      run: |
+        cabal test --extra-lib-dirs=/usr/local/lib

.github/workflows/build-python-package.yml

@@ -25,7 +25,7 @@ jobs:
 
     - name: Install cibuildwheel
       run: |
-        python -m pip install git+https://github.com/joerick/cibuildwheel.git@master
+        python -m pip install git+https://github.com/joerick/cibuildwheel.git@main
 
     - name: Install build tools for OSX
       if: startsWith(matrix.os, 'macos')

.gitignore

@@ -5,6 +5,7 @@
 *.jar
 *.gfo
 *.pgf
+*.ngf
 debian/.debhelper
 debian/debhelper-build-stamp
 debian/gf
@@ -46,6 +47,8 @@ src/runtime/c/sg/.dirstamp
 src/runtime/c/stamp-h1
 src/runtime/java/.libs/
 src/runtime/python/build/
+src/runtime/python/**/__pycache__/
+src/runtime/python/**/.pytest_cache/
 .cabal-sandbox
 cabal.sandbox.config
 .stack-work
@@ -53,6 +56,10 @@ DATA_DIR
 
 stack*.yaml.lock
 
+# Output files for test suite
+*.out
+gf-tests.html
+
 # Generated documentation (not exhaustive)
 demos/index-numbers.html
 demos/resourcegrammars.html

.travis.yml

@@ -1,14 +0,0 @@
-sudo: required
-
-language: c
-
-services:
-  - docker
-
-before_install:
-  - docker pull odanoburu/gf-src:3.9
-
-script:
-  - |
-    docker run --mount src="$(pwd)",target=/home/gfer,type=bind odanoburu/gf-src:3.9 /bin/bash -c "cd /home/gfer/src/runtime/c &&
-    autoreconf -i && ./configure && make && make install ; cd /home/gfer ; cabal install -fserver -fc-runtime --extra-lib-dirs='/usr/local/lib'"

CHANGELOG.md

@@ -0,0 +1,11 @@
+### New since 3.11 (WIP)
+
+- Added a changelog!
+
+### 3.11
+
+See <https://www.grammaticalframework.org/download/release-3.11.html>
+
+### 3.10
+
+See <https://www.grammaticalframework.org/download/release-3.10.html>

Makefile

@@ -1,31 +1,48 @@
-.PHONY: all build install doc clean gf html deb pkg bintar sdist
+.PHONY: all build install doc clean html deb pkg bintar sdist
 
 # This gets the numeric part of the version from the cabal file
 VERSION=$(shell sed -ne "s/^version: *\([0-9.]*\).*/\1/p" gf.cabal)
 
+# Check if stack is installed
+STACK=$(shell if hash stack 2>/dev/null; then echo "1"; else echo "0"; fi)
+
+# Check if cabal >= 2.4 is installed (with v1- and v2- commands)
+CABAL_NEW=$(shell if cabal v1-repl --help >/dev/null 2>&1 ; then echo "1"; else echo "0"; fi)
+
+ifeq ($(STACK),1)
+  CMD=stack
+else
+  CMD=cabal
+  ifeq ($(CABAL_NEW),1)
+    CMD_PFX=v1-
+  endif
+endif
+
 all: build
 
 dist/setup-config: gf.cabal Setup.hs WebSetup.hs
-	cabal configure
+ifneq ($(STACK),1)
+	cabal ${CMD_PFX}configure
+endif
 
 build: dist/setup-config
-	cabal build
+	${CMD} ${CMD_PFX}build
 
 install:
-	cabal copy
-	cabal register
+ifeq ($(STACK),1)
+	stack install
+else
+	cabal ${CMD_PFX}copy
+	cabal ${CMD_PFX}register
+endif
 
 doc:
-	cabal haddock
+	${CMD} ${CMD_PFX}haddock
 
 clean:
-	cabal clean
+	${CMD} ${CMD_PFX}clean
 	bash bin/clean_html
 
-gf:
-	cabal build rgl-none
-	strip dist/build/gf/gf
-
 html::
 	bash bin/update_html
 
@@ -35,7 +52,7 @@ html::
 deb:
 	dpkg-buildpackage -b -uc
 
-# Make an OS X Installer package
+# Make a macOS installer package
 pkg:
 	FMT=pkg bash bin/build-binary-dist.sh
 
@@ -48,6 +65,6 @@ bintar:
 
 # Make a source tar.gz distribution using git to make sure that everything is included.
 # We put the distribution in dist/ so it is removed on `make clean`
-sdist:
-	test -d dist || mkdir dist
-	git archive --format=tar.gz --output=dist/gf-${VERSION}.tar.gz HEAD
+# sdist:
+# 	test -d dist || mkdir dist
+# 	git archive --format=tar.gz --output=dist/gf-${VERSION}.tar.gz HEAD

README.md

@@ -1,4 +1,4 @@
-![GF Logo](doc/Logos/gf1.svg)
+![GF Logo](https://www.grammaticalframework.org/doc/Logos/gf1.svg)
 
 # Grammatical Framework (GF)
 
@@ -30,13 +30,16 @@ GF particularly addresses four aspects of grammars:
 
 ## Compilation and installation
 
-The simplest way of installing GF is with the command:
+The simplest way of installing GF from source is with the command:
 ```
 cabal install
 ```
+or:
+```
+stack install
+```
 
-For more details, see the [download page](http://www.grammaticalframework.org/download/index.html)
-and [developers manual](http://www.grammaticalframework.org/doc/gf-developers.html).
+For more information, including links to precompiled binaries, see the [download page](https://www.grammaticalframework.org/download/index.html).
 
 ## About this repository
 

RELEASE.md

@@ -45,11 +45,16 @@ but the generated _artifacts_ must be manually attached to the release as _asset
 
 ### 4. Upload to Hackage
 
-1. Run `make sdist`
+In order to do this you will need to be added the [GF maintainers](https://hackage.haskell.org/package/gf/maintainers/) on Hackage.
+
+1. Run `stack sdist --test-tarball` and address any issues.
 2. Upload the package, either:
-    1. **Manually**: visit <https://hackage.haskell.org/upload> and upload the file `dist/gf-X.Y.tar.gz`
-    2. **via Cabal (≥2.4)**: `cabal upload dist/gf-X.Y.tar.gz`
-3. If the documentation-building fails on the Hackage server, do:
+    1. **Manually**: visit <https://hackage.haskell.org/upload> and upload the file generated by the previous command.
+    2. **via Stack**: `stack upload . --candidate`
+3. After testing the candidate, publish it:
+    1. **Manually**: visit <https://hackage.haskell.org/package/gf-X.Y.Z/candidate/publish>
+    1. **via Stack**: `stack upload .`
+4. If the documentation-building fails on the Hackage server, do:
 ```
 cabal v2-haddock --builddir=dist/docs --haddock-for-hackage --enable-doc
 cabal upload --documentation dist/docs/*-docs.tar.gz

WebSetup.hs

@@ -26,6 +26,14 @@ import Distribution.PackageDescription(PackageDescription(..))
   so users won't see this message unless they check the log.)
 -}
 
+-- | Notice about contrib grammars
+noContribMsg :: IO ()
+noContribMsg = putStr $ unlines
+  [ "Example grammars are no longer included in the main GF repository, but have moved to gf-contrib."
+  , "If you want them to be built, clone the following repository in the same directory as gf-core:"
+  , "https://github.com/GrammaticalFramework/gf-contrib.git"
+  ]
+
 example_grammars :: [(String, String, [String])] -- [(pgf, subdir, source modules)]
 example_grammars =
    [("Letter.pgf","letter",letterSrc)
@@ -50,11 +58,8 @@ buildWeb gf flags (pkg,lbi) = do
   contrib_exists <- doesDirectoryExist contrib_dir
   if contrib_exists
   then mapM_ build_pgf example_grammars
-  else putStr $ unlines
-    [ "Example grammars are no longer included in the main GF repository, but have moved to gf-contrib."
-    , "If you want these example grammars to be built, clone this repository in the same top-level directory as GF:"
-    , "https://github.com/GrammaticalFramework/gf-contrib.git"
-    ]
+  -- else noContribMsg
+  else return ()
   where
     gfo_dir = buildDir lbi </> "examples"
 

debian/changelog

@@ -1,3 +1,9 @@
+gf (3.11) bionic focal; urgency=low
+
+  * GF 3.11
+
+ -- Inari Listenmaa <inari@digitalgrammars.com>  Sun, 25 Jul 2021 10:27:40 +0800
+
 gf (3.10.4-1) xenial bionic cosmic; urgency=low
 
   * GF 3.10.4

debian/rules

@@ -16,9 +16,9 @@ override_dh_shlibdeps:
 override_dh_auto_configure:
 	cd src/runtime/c && bash setup.sh configure --prefix=/usr
 	cd src/runtime/c && bash setup.sh build
-	cabal update
-	cabal install --only-dependencies
-	cabal configure --prefix=/usr -fserver -fc-runtime --extra-lib-dirs=$(CURDIR)/src/runtime/c/.libs --extra-include-dirs=$(CURDIR)/src/runtime/c
+	cabal v1-update
+	cabal v1-install --only-dependencies
+	cabal v1-configure --prefix=/usr -fserver -fc-runtime --extra-lib-dirs=$(CURDIR)/src/runtime/c/.libs --extra-include-dirs=$(CURDIR)/src/runtime/c
 
 SET_LDL=LD_LIBRARY_PATH=$$LD_LIBRARY_PATH:$(CURDIR)/src/runtime/c/.libs
 
@@ -26,10 +26,10 @@ override_dh_auto_build:
 	cd src/runtime/python && EXTRA_INCLUDE_DIRS=$(CURDIR)/src/runtime/c EXTRA_LIB_DIRS=$(CURDIR)/src/runtime/c/.libs python setup.py build
 	cd src/runtime/java && make CFLAGS="-I$(CURDIR)/src/runtime/c -L$(CURDIR)/src/runtime/c/.libs" INSTALL_PATH=/usr
 	echo $(SET_LDL)
-	-$(SET_LDL) cabal build
+	-$(SET_LDL) cabal v1-build
 
 override_dh_auto_install:
-	$(SET_LDL) cabal copy --destdir=$(CURDIR)/debian/gf
+	$(SET_LDL) cabal v1-copy --destdir=$(CURDIR)/debian/gf
 	cd src/runtime/c && bash setup.sh copy prefix=$(CURDIR)/debian/gf/usr
 	cd src/runtime/python && python setup.py install --prefix=$(CURDIR)/debian/gf/usr
 	cd src/runtime/java && make INSTALL_PATH=$(CURDIR)/debian/gf/usr install

doc/gf-developers-old-cabal.t2t

@@ -0,0 +1,201 @@
+GF Developer's Guide: Old installation instructions with Cabal
+
+
+This page contains the old installation instructions from the [Developer's Guide ../doc/gf-developers.html].
+We recommend Stack as a primary installation method, because it's easier for a Haskell beginner, and we want to keep the main instructions short.
+But if you are an experienced Haskeller and want to keep using Cabal, here are the old instructions using ``cabal install``.
+
+Note that some of these instructions may be outdated. Other parts may still be useful.
+
+== Compilation from source with Cabal ==
+
+The build system of GF is based on //Cabal//, which is part of the
+Haskell Platform, so no extra steps are needed to install it. In the simplest
+case, all you need to do to compile and install GF, after downloading the
+source code as described above, is
+
+```
+$ cabal install
+```
+
+This will automatically download any additional Haskell libraries needed to
+build GF. If this is the first time you use Cabal, you might need to run
+``cabal update`` first, to update the list of available libraries.
+
+If you want more control, the process can also be split up into the usual
+//configure//, //build// and //install// steps.
+
+=== Configure ===
+
+During the configuration phase Cabal will check that you have all
+necessary tools and libraries needed for GF. The configuration is
+started by the command:
+
+```
+$ cabal configure
+```
+
+If you don't see any error message from the above command then you
+have everything that is needed for GF. You can also add the option
+``-v`` to see more details about the configuration.
+
+You can use ``cabal configure --help`` to get a list of configuration options.
+
+=== Build ===
+
+The build phase does two things. First it builds the GF compiler from
+the Haskell source code and after that it builds the GF Resource Grammar
+Library using the already build compiler.  The simplest command is:
+
+```
+$ cabal build
+```
+
+Again you can add the option ``-v`` if you want to see more details.
+
+==== Parallel builds ====
+
+If you have Cabal>=1.20 you can enable parallel compilation by using
+
+```
+$ cabal build -j
+```
+
+or by putting a line
+```
+jobs: $ncpus
+```
+in your ``.cabal/config`` file. Cabal
+will pass this option to GHC when building the GF compiler, if you
+have GHC>=7.8.
+
+Cabal also passes ``-j`` to GF to enable parallel compilation of the
+Resource Grammar Library. This is done unconditionally to avoid
+causing problems for developers with Cabal<1.20. You can disable this
+by editing the last few lines in ``WebSetup.hs``.
+
+=== Install ===
+
+After you have compiled GF you need to install the executable and libraries
+to make the system usable.
+
+```
+$ cabal copy
+$ cabal register
+```
+
+This command installs the GF compiler for a single user, in the standard
+place used by Cabal.
+On Linux and Mac this could be ``$HOME/.cabal/bin``.
+On Mac it could also be ``$HOME/Library/Haskell/bin``.
+On Windows this is ``C:\Program Files\Haskell\bin``.
+
+The compiled GF Resource Grammar Library will be installed
+under the same prefix, e.g. in
+``$HOME/.cabal/share/gf-3.3.3/lib`` on Linux and
+in ``C:\Program Files\Haskell\gf-3.3.3\lib`` on Windows.
+
+If you want to install in some other place then use the ``--prefix``
+option during the configuration phase.
+
+=== Clean ===
+
+Sometimes you want to clean up the compilation and start again from clean
+sources. Use the clean command for this purpose:
+
+```
+$ cabal clean
+```
+
+
+%=== SDist ===
+%
+%You can use the command:
+%
+%% This does *NOT* include everything that is needed // TH 2012-08-06
+%```
+%$ cabal sdist
+%```
+%
+%to prepare archive with all source codes needed to compile GF.
+
+=== Known problems with Cabal ===
+
+Some versions of Cabal (at least version 1.16) seem to have a bug that can
+cause the following error:
+
+```
+Configuring gf-3.x...
+setup: Distribution/Simple/PackageIndex.hs:124:8-13: Assertion failed
+```
+
+The exact cause of this problem is unclear, but it seems to happen
+during the configure phase if the same version of GF is already installed,
+so a workaround is to remove the existing installation with
+
+```
+ghc-pkg unregister gf
+```
+
+You can check with ``ghc-pkg list gf`` that it is gone.
+
+== Compilation with make ==
+
+If you feel more comfortable with Makefiles then there is a thin Makefile
+wrapper arround Cabal for you. If you just type:
+```
+$ make
+```
+the configuration phase will be run automatically if needed and after that
+the sources will be compiled.
+
+%% cabal build rgl-none does not work with recent versions of Cabal
+%If you don't want to compile the resource library
+%every time then you can use:
+%```
+%$ make gf
+%```
+
+For installation use:
+```
+$ make install
+```
+For cleaning:
+```
+$ make clean
+```
+%and to build source distribution archive run:
+%```
+%$ make sdist
+%```
+
+
+== Partial builds of RGL ==
+
+**NOTE**: The following doesn't work with recent versions of ``cabal``. //(This comment was left in 2015, so make your own conclusions.)//
+%% // TH 2015-06-22
+
+%Sometimes you just want to work on the GF compiler and don't want to
+%recompile the resource library after each change. In this case use
+%this extended command:
+
+%```
+%$ cabal build rgl-none
+%```
+
+The resource grammar library can be compiled in two modes: with present
+tense only and with all tenses. By default it is compiled with all
+tenses. If you want to use the library with only present tense you can
+compile it in this special mode with the command:
+
+```
+$ cabal build present
+```
+
+You could also control which languages you want to be recompiled by
+adding the option ``langs=list``. For example the following command
+will compile only the English and the Swedish language:
+
+```
+$ cabal build langs=Eng,Swe
+```

doc/gf-developers.t2t

@@ -1,6 +1,6 @@
 GF Developers Guide
 
-2018-07-26
+2021-07-15
 
 %!options(html): --toc
 
@@ -15,388 +15,287 @@ you are a GF user who just wants to download and install GF
 == Setting up your system for building GF ==
 
 To build GF from source you need to install some tools on your
-system: the //Haskell Platform//, //Git// and the //Haskeline library//.
+system: the Haskell build tool //Stack//, the version control software //Git// and the //Haskeline// library.
 
-**On Linux** the best option is to install the tools via the standard
-software distribution channels, i.e. by using the //Software Center//
-in Ubuntu or the corresponding tool in other popular Linux distributions.
-Or, from a Terminal window, the following command should be enough:
+%**On Linux** the best option is to install the tools via the standard
+%software distribution channels, i.e. by using the //Software Center//
+%in Ubuntu or the corresponding tool in other popular Linux distributions.
 
-- On Ubuntu: ``sudo apt-get install haskell-platform git libghc6-haskeline-dev``
-- On Fedora: ``sudo dnf install haskell-platform git ghc-haskeline-devel``
+%**On Mac OS and Windows**, the tools can be downloaded from their respective
+%web sites, as described below.
+
+=== Stack ===
+The primary installation method is via //Stack//.
+(You can also use Cabal, but we recommend Stack to those who are new to Haskell.)
+
+To install Stack:
+
+- **On Linux and Mac OS**, do either
+
+    ``$ curl -sSL https://get.haskellstack.org/ | sh``
+
+  or
+
+    ``$ wget -qO- https://get.haskellstack.org/ | sh``
 
 
-**On Mac OS and Windows**, the tools can be downloaded from their respective
-web sites, as described below.
+- **On other operating systems**, see the [installation guide https://docs.haskellstack.org/en/stable/install_and_upgrade].
 
-=== The Haskell Platform ===
 
-GF is written in Haskell, so first of all you need
-the //Haskell Platform//, e.g. version 8.0.2 or 7.10.3. Downloads
-and installation instructions are available from here:
+%If you already have Stack installed, upgrade it to the latest version by running: ``stack upgrade``
 
-    http://hackage.haskell.org/platform/
-
-Once you have installed the Haskell Platform, open a terminal
-(Command Prompt on Windows) and try to execute the following command:
-```
-$ ghc --version
-```
-This command should show you which version of GHC you have. If the installation
-of the Haskell Platform	 was successful you should see a message like:
-
-```
-The Glorious Glasgow Haskell Compilation System, version 8.0.2
-```
-
-Other required tools included in the Haskell Platform are
-[Cabal http://www.haskell.org/cabal/],
-[Alex http://www.haskell.org/alex/]
-and
-[Happy http://www.haskell.org/happy/].
 
 === Git ===
 
-To get the GF source code, you also need //Git//.
-//Git// is a distributed version control system, see
-https://git-scm.com/downloads for more information.
+To get the GF source code, you also need //Git//, a distributed version control system.
 
-=== The haskeline library ===
+- **On Linux**, the best option is to install the tools via the standard
+software distribution channels:
+
+  - On Ubuntu: ``sudo apt-get install git-all``
+  - On Fedora: ``sudo dnf install git-all``
+
+
+- **On other operating systems**, see
+https://git-scm.com/book/en/v2/Getting-Started-Installing-Git for installation.
+
+
+
+=== Haskeline ===
 
 GF uses //haskeline// to enable command line editing in the GF shell.
-This should work automatically on Mac OS and Windows, but on Linux one
-extra step is needed to make sure the C libraries (terminfo)
-required by //haskeline// are installed.  Here is one way to do this:
 
-- On Ubuntu: ``sudo apt-get install libghc-haskeline-dev``
-- On Fedora: ``sudo dnf install ghc-haskeline-devel``
+- **On Mac OS and Windows**, this should work automatically.
+
+- **On Linux**, an extra step is needed to make sure the C libraries (terminfo)
+required by //haskeline// are installed:
+
+  - On Ubuntu: ``sudo apt-get install libghc-haskeline-dev``
+  - On Fedora: ``sudo dnf install ghc-haskeline-devel``
 
 
-== Getting the source ==
+== Getting the source ==[getting-source]
 
-Once you have all tools in place you can get the GF source code. If you
-just want to compile and use GF then it is enough to have read-only
-access. It is also possible to make changes in the source code but if you
-want these changes to be applied back to the main source repository you will
-have to send the changes to us.  If you plan to work continuously on
-GF then you should consider getting read-write access.
+Once you have all tools in place you can get the GF source code from
+[GitHub https://github.com/GrammaticalFramework/]:
 
-=== Read-only access ===
+- https://github.com/GrammaticalFramework/gf-core for the GF compiler
+- https://github.com/GrammaticalFramework/gf-rgl for the Resource Grammar Library
 
-==== Getting a fresh copy for read-only access ====
 
-Anyone can get the latest development version of GF by running:
+=== Read-only access: clone the main repository ===
+
+If you only want to compile and use GF, you can just clone the repositories as follows:
 
 ```
-$ git clone https://github.com/GrammaticalFramework/gf-core.git
-$ git clone https://github.com/GrammaticalFramework/gf-rgl.git
+  $ git clone https://github.com/GrammaticalFramework/gf-core.git
+  $ git clone https://github.com/GrammaticalFramework/gf-rgl.git
 ```
 
-This will create directories ``gf-core`` and ``gf-rgl`` in the current directory.
-
-
-==== Updating your copy ====
-
-To get all new patches from each repo:
-```
-$ git pull
-```
-This can be done anywhere in your local repository.
-
-
-==== Recording local changes ====[record]
-
-Since every copy is a repository, you can have local version control
-of your changes.
-
-If you have added files, you first need to tell your local repository to
-keep them under revision control:
+To get new updates, run the following anywhere in your local copy of the repository:
 
 ```
-$ git add file1 file2 ...
+  $ git pull
 ```
 
-To record changes, use:
+=== Contribute your changes: fork the main repository ===
+
+If you want the possibility to contribute your changes,
+you should create your own fork, do your changes there,
+and then send a pull request to the main repository.
+
++ **Creating and cloning a fork  —**
+See GitHub documentation for instructions how to [create your own fork https://docs.github.com/en/get-started/quickstart/fork-a-repo]
+of the repository. Once you've done it, clone the fork to your local computer.
 
 ```
-$ git commit file1 file2 ...
+  $ git clone https://github.com/<YOUR_USERNAME>/gf-core.git
 ```
 
-This creates a patch against the previous version and stores it in your
-local repository. You can record any number of changes before
-pushing them to the main repo. In fact, you don't have to push them at
-all if you want to keep the changes only in your local repo.
-
-Instead of enumerating all modified files on the command line,
-you can use the flag ``-a`` to automatically record //all// modified
-files. You still need to use ``git add`` to add new files.
-
-
-=== Read-write access ===
-
-If you are a member of the GF project on GitHub, you can push your
-changes directly to the GF git repository on GitHub.
++ **Updating your copy —**
+Once you have cloned your fork, you need to set up the main repository as a remote:
 
 ```
-$ git push
+  $ git remote add upstream https://github.com/GrammaticalFramework/gf-core.git
 ```
 
-It is also possible for anyone else to contribute by
+Then you can get the latest updates by running the following:
 
-- creating a fork of the GF repository on GitHub,
-- working with local clone of the fork (obtained with ``git clone``),
-- pushing changes to the fork,
-- and finally sending a pull request.
+```
+  $ git pull upstream master
+```
+
++ **Recording local changes —**
+See Git tutorial on how to [record and push your changes https://git-scm.com/book/en/v2/Git-Basics-Recording-Changes-to-the-Repository] to your fork.
+
++ **Pull request —**
+When you want to contribute your changes to the main gf-core repository,
+[create a pull request https://docs.github.com/en/github/collaborating-with-pull-requests/proposing-changes-to-your-work-with-pull-requests/creating-a-pull-request]
+from your fork.
 
 
 
-== Compilation from source with Cabal ==
+If you want to contribute to the RGL as well, do the same process for the RGL repository.
 
-The build system of GF is based on //Cabal//, which is part of the
-Haskell Platform, so no extra steps are needed to install it. In the simplest
-case, all you need to do to compile and install GF, after downloading the
-source code as described above, is
+
+== Compilation from source ==
+
+By now you should have installed Stack and Haskeline, and cloned the Git repository on your own computer, in a directory called ``gf-core``.
+
+=== Primary recommendation: use Stack ===
+
+Open a terminal, go to the top directory (``gf-core``), and type the following command.
+
+```
+$ stack install
+```
+
+It will install GF and all necessary tools and libraries to do that.
+
+
+=== Alternative: use Cabal ===
+You can also install GF using Cabal, if you prefer Cabal to Stack. In that case, you may need to install some prerequisites yourself.
+
+The actual installation process is similar to Stack: open a terminal, go to the top directory (``gf-core``), and type the following command.
 
 ```
 $ cabal install
 ```
 
-This will automatically download any additional Haskell libraries needed to
-build GF. If this is the first time you use Cabal, you might need to run
-``cabal update`` first, to update the list of available libraries.
+//The old (potentially outdated) instructions for Cabal are moved to a [separate page ../doc/gf-developers-old-cabal.html]. If you run into trouble with ``cabal install``, you may want to take a look.//
 
-If you want more control, the process can also be split up into the usual
-//configure//, //build// and //install// steps.
+== Compiling GF with C runtime system support ==
 
-=== Configure ===
-
-During the configuration phase Cabal will check that you have all
-necessary tools and libraries needed for GF. The configuration is
-started by the command:
-
-```
-$ cabal configure
-```
-
-If you don't see any error message from the above command then you
-have everything that is needed for GF. You can also add the option
-``-v`` to see more details about the configuration.
-
-You can use ``cabal configure --help`` to get a list of configuration options.
-
-=== Build ===
-
-The build phase does two things. First it builds the GF compiler from
-the Haskell source code and after that it builds the GF Resource Grammar
-Library using the already build compiler.  The simplest command is:
-
-```
-$ cabal build
-```
-
-Again you can add the option ``-v`` if you want to see more details.
-
-==== Parallel builds ====
-
-If you have Cabal>=1.20 you can enable parallel compilation by using
-
-```
-$ cabal build -j
-```
-
-or by putting a line
-```
-jobs: $ncpus
-```
-in your ``.cabal/config`` file. Cabal
-will pass this option to GHC when building the GF compiler, if you
-have GHC>=7.8.
-
-Cabal also passes ``-j`` to GF to enable parallel compilation of the
-Resource Grammar Library. This is done unconditionally to avoid
-causing problems for developers with Cabal<1.20. You can disable this
-by editing the last few lines in ``WebSetup.hs``.
-
-
-==== Partial builds ====
-
-**NOTE**: The following doesn't work with recent versions of ``cabal``.
-%% // TH 2015-06-22
-
-Sometimes you just want to work on the GF compiler and don't want to
-recompile the resource library after each change. In this case use
-this extended command:
-
-```
-$ cabal build rgl-none
-```
-
-The resource library could also be compiled in two modes: with present
-tense only and with all tenses. By default it is compiled with all
-tenses. If you want to use the library with only present tense you can
-compile it in this special mode with the command:
-
-```
-$ cabal build present
-```
-
-You could also control which languages you want to be recompiled by
-adding the option ``langs=list``. For example the following command
-will compile only the English and the Swedish language:
-
-```
-$ cabal build langs=Eng,Swe
-```
-
-=== Install ===
-
-After you have compiled GF you need to install the executable and libraries
-to make the system usable.
-
-```
-$ cabal copy
-$ cabal register
-```
-
-This command installs the GF compiler for a single user, in the standard
-place used by Cabal.
-On Linux and Mac this could be ``$HOME/.cabal/bin``.
-On Mac it could also be ``$HOME/Library/Haskell/bin``.
-On Windows this is ``C:\Program Files\Haskell\bin``.
-
-The compiled GF Resource Grammar Library will be installed
-under the same prefix, e.g. in
-``$HOME/.cabal/share/gf-3.3.3/lib`` on Linux and
-in ``C:\Program Files\Haskell\gf-3.3.3\lib`` on Windows.
-
-If you want to install in some other place then use the ``--prefix``
-option during the configuration phase.
-
-=== Clean ===
-
-Sometimes you want to clean up the compilation and start again from clean
-sources. Use the clean command for this purpose:
-
-```
-$ cabal clean
-```
-
-
-%=== SDist ===
-%
-%You can use the command:
-%
-%% This does *NOT* include everything that is needed // TH 2012-08-06
-%```
-%$ cabal sdist
-%```
-%
-%to prepare archive with all source codes needed to compile GF.
-
-=== Known problems with Cabal ===
-
-Some versions of Cabal (at least version 1.16) seem to have a bug that can
-cause the following error:
-
-```
-Configuring gf-3.x...
-setup: Distribution/Simple/PackageIndex.hs:124:8-13: Assertion failed
-```
-
-The exact cause of this problem is unclear, but it seems to happen
-during the configure phase if the same version of GF is already installed,
-so a workaround is to remove the existing installation with
-
-```
-ghc-pkg unregister gf
-```
-
-You can check with ``ghc-pkg list gf`` that it is gone.
-
-== Compilation with make ==
-
-If you feel more comfortable with Makefiles then there is a thin Makefile
-wrapper arround Cabal for you. If you just type:
-```
-$ make
-```
-the configuration phase will be run automatically if needed and after that
-the sources will be compiled.
-
-%% cabal build rgl-none does not work with recent versions of Cabal
-%If you don't want to compile the resource library
-%every time then you can use:
-%```
-%$ make gf
-%```
-
-For installation use:
-```
-$ make install
-```
-For cleaning:
-```
-$ make clean
-```
-%and to build source distribution archive run:
-%```
-%$ make sdist
-%```
-
-== Compiling GF with C run-time system support ==
-
-The C run-time system is a separate implementation of the PGF run-time services.
+The C runtime system is a separate implementation of the PGF runtime services.
 It makes it possible to work with very large, ambiguous grammars, using
-probabilistic models to obtain probable parses. The C run-time system might
-also be easier to use than the Haskell run-time system on certain platforms,
+probabilistic models to obtain probable parses. The C runtime system might
+also be easier to use than the Haskell runtime system on certain platforms,
 e.g. Android and iOS.
 
-To install the C run-time system, go to the ``src/runtime/c`` directory
-%and follow the instructions in the ``INSTALL`` file.
-and use the ``install.sh`` script:
-```
-bash setup.sh configure
-bash setup.sh build
-bash setup.sh install
-```
-This will install
-the C header files and libraries need to write C programs that use PGF grammars.
-Some example C programs are included in the ``utils`` subdirectory, e.g.
-``pgf-translate.c``.
+To install the C runtime system, go to the ``src/runtime/c`` directory.
 
-When the C run-time system is installed, you can install GF with C run-time
-support by doing
+- **On Linux and Mac OS —**
+  You should have autoconf, automake, libtool and make.
+  If you are missing some of them, follow the
+  instructions in the [INSTALL https://github.com/GrammaticalFramework/gf-core/blob/master/src/runtime/c/INSTALL] file.
+
+  Once you have the required libraries, the easiest way to install the C runtime is to use the ``install.sh`` script. Just type
+
+  ``$ bash install.sh``
+
+  This will install the C header files and libraries need to write C programs
+  that use PGF grammars.
+
+%  If this doesn't work for you, follow the manual instructions in the [INSTALL https://github.com/GrammaticalFramework/gf-core/blob/master/src/runtime/c/INSTALL] file under your operating system.
+
+- **On other operating systems —** Follow the instructions in the
+[INSTALL https://github.com/GrammaticalFramework/gf-core/blob/master/src/runtime/c/INSTALL] file under your operating system.
+
+
+
+Depending on what you want to do with the C runtime, you can follow one or more of the following steps.
+
+=== Use the C runtime from another programming language ===[bindings]
+
+% **If you just want to use the C runtime from Python, Java, or Haskell, you don't need to change your GF installation.**
+
+- **What —**
+This is the most common use case for the C runtime: compile
+your GF grammars into PGF with the standard GF executable,
+and manipulate the PGFs from another programming language,
+using the bindings to the C runtime.
+
+
+- **How —**
+The Python, Java and Haskell bindings are found in the
+``src/runtime/{python,java,haskell-bind}`` directories,
+respecively. Compile them by following the instructions
+in the ``INSTALL`` or ``README`` files in those directories.
+
+The Python library can also be installed from PyPI using ``pip install pgf``.
+
+
+//If you are on Mac and get an error about ``clang`` version, you can try some of [these solutions https://stackoverflow.com/questions/63972113/big-sur-clang-invalid-version-error-due-to-macosx-deployment-target]—but be careful before removing any existing installations.//
+
+
+=== Use GF shell with C runtime support ===
+
+- **What —**
+If you want to use the GF shell with C runtime functionalities, then you need to (re)compile GF with special flags.
+
+The GF shell can be started with ``gf -cshell`` or ``gf -crun`` to use
+the C run-time system instead of the Haskell run-time system.
+Only limited functionality is available when running the shell in these
+modes (use the ``help`` command in the shell for details).
+
+(Re)compiling your GF with these flags will also give you
+Haskell bindings to the C runtime, as a library called ``PGF2``,
+but if you want Python or Java bindings, you need to do [the previous step #bindings].
+
+% ``PGF2``: a module to import in Haskell programs, providing a binding to the C run-time system.
+
+- **How —**
+If you use cabal, run the following command:
 
 ```
-cabal install -fserver -fc-runtime
+cabal install -fc-runtime
 ```
-from the top directory. This give you three new things:
 
-- ``PGF2``: a module to import in Haskell programs, providing a binding to
-   the C run-time system.
+from the top directory (``gf-core``).
 
-- The GF shell can be started with ``gf -cshell`` or ``gf -crun`` to use
-  the C run-time system instead of the Haskell run-time system.
-  Only limited functionality is available when running the shell in these
-  modes (use the ``help`` command in the shell for details).
+If you use stack, uncomment the following lines in the ``stack.yaml`` file:
 
-- ``gf -server`` mode is extended with new requests to call the C run-time
-   system, e.g. ``c-parse``, ``c-linearize`` and ``c-translate``.
+```
+flags:
+   gf:
+     c-runtime: true
+extra-lib-dirs:
+   - /usr/local/lib
+```
+and then run ``stack install`` from the top directory (``gf-core``).
 
 
-=== Python and Java bindings ===
+//If you get an "``error while loading shared libraries``" when trying to run GF with C runtime, remember to declare your ``LD_LIBRARY_PATH``.//
+//Add ``export LD_LIBRARY_PATH="/usr/local/lib"`` to either your ``.bashrc`` or ``.profile``. You should now be able to start GF with C runtime.//
+
+
+=== Use GF server mode with C runtime ===
+
+- **What —**
+With this feature, ``gf -server`` mode is extended with new requests to call the C run-time
+system, e.g. ``c-parse``, ``c-linearize`` and ``c-translate``.
+
+- **How —**
+If you use cabal, run the following command:
+
+```
+cabal install -fc-runtime -fserver
+```
+from the top directory.
+
+If you use stack, add the following lines in the ``stack.yaml`` file:
+
+```
+flags:
+   gf:
+     c-runtime: true
+     server: true
+extra-lib-dirs:
+   - /usr/local/lib
+```
+
+and then run ``stack install``, also from the top directory.
 
-The C run-time system can also be used from Python and Java. Python and Java
-bindings are found in the ``src/runtime/python`` and ``src/runtime/java``
-directories, respecively. Compile them by following the instructions in
-the ``INSTALL`` files in those directories.
 
-The Python library can also be installed from PyPI using `pip install pgf`.
 
 == Compilation of RGL ==
 
 As of 2018-07-26, the RGL is distributed separately from the GF compiler and runtimes.
 
+To get the source, follow the previous instructions on [how to clone a repository with Git #getting-source].
+
+After cloning the RGL, you should have a directory named ``gf-rgl`` on your computer.
+
 === Simple ===
 To install the RGL, you can use the following commands from within the ``gf-rgl`` repository:
 ```
@@ -418,103 +317,68 @@ If you do not have Haskell installed, you can use the simple build script ``Setu
 
 == Creating binary distribution packages ==
 
-=== Creating .deb packages for Ubuntu ===
+The binaries are generated with Github Actions. More details can be viewed here:
 
-This was tested on Ubuntu 14.04 for the release of GF 3.6, and the
-resulting ``.deb`` packages appears to work on Ubuntu 12.04, 13.10 and 14.04.
-For the release of GF 3.7, we generated ``.deb`` packages on Ubuntu 15.04 and
-tested them on Ubuntu 12.04 and 14.04.
+https://github.com/GrammaticalFramework/gf-core/actions/workflows/build-binary-packages.yml
 
-Under Ubuntu, Haskell executables are statically linked against other Haskell
-libraries, so the .deb packages are fairly self-contained.
 
-==== Preparations ====
+== Running the test suite ==
+
+The GF test suite is run with one of the following commands from the top directory:
 
 ```
-sudo apt-get install dpkg-dev debhelper
+  $ cabal test
 ```
 
-==== Creating the package ====
-
-Make sure the ``debian/changelog`` starts with an entry that describes the
-version you are building. Then run
+or
 
 ```
-make deb
+  $ stack test
 ```
 
-If get error messages about missing dependencies
-(e.g. ``autoconf``, ``automake``, ``libtool-bin``, ``python-dev``,
-``java-sdk``, ``txt2tags``)
-use ``apt-get intall`` to install them, then try again.
-
-
-=== Creating OS X Installer packages ===
-
-Run
-
-```
-make pkg
-```
-
-=== Creating binary tar distributions ===
-
-Run
-
-```
-make bintar
-```
-
-=== Creating .rpm packages for Fedora ===
-
-This is possible, but the procedure has not been automated.
-It involves using the cabal-rpm tool,
-
-```
-sudo dnf install cabal-rpm
-```
-
-and following the Fedora guide
-[How to create an RPM package http://fedoraproject.org/wiki/How_to_create_an_RPM_package].
-
-Under Fedora, Haskell executables are dynamically linked against other Haskell
-libraries, so ``.rpm`` packages for all Haskell libraries that GF depends on
-are required. Most of them are already available in the Fedora distribution,
-but a few of them might have to be built and distributed along with
-the GF ``.rpm`` package.
-When building ``.rpm`` packages for GF 3.4, we also had to build ``.rpm``s for
-``fst`` and ``httpd-shed``.
-
-== Running the testsuite ==
-
-**NOTE:** The test suite has not been maintained recently, so expect many
-tests to fail.
-%% // TH 2012-08-06
-
-GF has testsuite. It is run with the following command:
-```
-$ cabal test
-```
 The testsuite architecture for GF is very simple but still very flexible.
 GF by itself is an interpreter and could execute commands in batch mode.
 This is everything that we need to organize a testsuite. The root of the
-testsuite is the testsuite/ directory. It contains subdirectories which
-themself contain GF batch files (with extension .gfs). The above command
-searches the subdirectories of the testsuite/ directory for files with extension
-.gfs and when it finds one it is executed with the GF interpreter.
-The output of the script is stored in file with extension .out and is compared
-with the content of the corresponding file with extension .gold, if there is one.
-If the contents are identical the command reports that the test was passed successfully.
-Otherwise the test had failed.
+testsuite is the ``testsuite/`` directory. It contains subdirectories
+which themselves contain GF batch files (with extension ``.gfs``).
+The above command searches the subdirectories of the ``testsuite/`` directory
+for files with extension ``.gfs`` and when it finds one, it is executed with
+the GF interpreter. The output of the script is stored in file with extension ``.out``
+and is compared with the content of the corresponding file with extension ``.gold``, if there is one.
 
-Every time when you make some changes to GF that have to be tested, instead of
-writing the commands by hand in the GF shell, add them to one .gfs file in the testsuite
-and run the test. In this way you can use the same test later and we will be sure
-that we will not incidentaly break your code later.
+Every time when you make some changes to GF that have to be tested,
+instead of writing the commands by hand in the GF shell, add them to one ``.gfs``
+file in the testsuite subdirectory where its ``.gf`` file resides and run the test.
+In this way you can use the same test later and we will be sure that we will not
+accidentally break your code later.
+
+**Test Outcome - Passed:** If the contents of the files with the ``.out`` extension
+are identical to their correspondingly-named files with the extension ``.gold``,
+the command will report that the tests passed successfully, e.g.
 
-If you don't want to run the whole testsuite you can write the path to the subdirectory
-in which you are interested. For example:
 ```
-$ cabal test testsuite/compiler
+ Running 1 test suites...
+ Test suite gf-tests: RUNNING...
+ Test suite gf-tests: PASS
+ 1 of 1 test suites (1 of 1 test cases) passed.
 ```
-will run only the testsuite for the compiler.
+
+**Test Outcome - Failed:**  If there is a contents mismatch between the files
+with the ``.out`` extension and their corresponding files with the extension ``.gold``,
+the test diagnostics will show a fail and the areas that failed. e.g.
+
+```
+  testsuite/compiler/compute/Records.gfs: OK
+  testsuite/compiler/compute/Variants.gfs: FAIL
+  testsuite/compiler/params/params.gfs: OK
+  Test suite gf-tests: FAIL
+  0 of 1 test suites (0 of 1 test cases) passed.
+```
+
+The fail results overview is available in gf-tests.html which shows 4 columns:
+
++ __Results__ - only areas that fail will appear. (Note: There are 3 failures in the gf-tests.html which are labelled as (expected). These failures should be ignored.)
++ __Input__ - which is the test written in the .gfs file
++ __Gold__ - the expected output from running the test set out in the .gfs file. This column refers to the contents from the .gold extension files.
++ __Output__ - This column refers to the contents from the .out extension files which are generated as test output.
+After fixing the areas which fail, rerun the test command. Repeat the entire process of fix-and-test until the test suite passes before  submitting a pull request to include your changes.

doc/gf-editor-modes.t2t

@@ -15,6 +15,12 @@ instructions inside.
 ==Atom==
 [language-gf https://atom.io/packages/language-gf], by John J. Camilleri
 
+==Visual Studio Code==
+
+[Grammatical Framework Language Server https://marketplace.visualstudio.com/items?itemName=anka-213.gf-vscode] by Andreas Källberg.
+
+This provides syntax highlighting and a client for the Grammatical Framework language server. Follow the installation instructions in the link.
+
 ==Eclipse==
 
 [GF Eclipse Plugin https://github.com/GrammaticalFramework/gf-eclipse-plugin/], by John J. Camilleri

doc/gf-people.md

@@ -7,7 +7,6 @@ title: "Grammatical Framework: Authors and Acknowledgements"
 The current maintainers of GF are
 
 [Krasimir Angelov](http://www.chalmers.se/cse/EN/organization/divisions/computing-science/people/angelov-krasimir),
-[Thomas Hallgren](http://www.cse.chalmers.se/~hallgren/),
 [Aarne Ranta](http://www.cse.chalmers.se/~aarne/),
 [John J. Camilleri](http://johnjcamilleri.com), and
 [Inari Listenmaa](https://inariksit.github.io/).
@@ -22,6 +21,7 @@ and
 
 The following people have contributed code to some of the versions:
 
+- [Thomas Hallgren](http://www.cse.chalmers.se/~hallgren/) (University of Gothenburg)
 - Grégoire Détrez (University of Gothenburg)
 - Ramona Enache (University of Gothenburg)
 - [Björn Bringert](http://www.cse.chalmers.se/alumni/bringert) (University of Gothenburg)

doc/hackers-guide/CompilationOverview.md

@@ -0,0 +1,11 @@
+# Compilation
+
+The GF language is designed to be easy for the programmers to use but be able to run it efficiently we need to reduce it to a more low-level language. The goal of this chapter is to give an overview of the different steps in the compilation. The program transformation goes throught the following phases:
+
+- renaming - here all identifiers in the grammar are made explicitly qualified. For example, if you had used the identifier PredVP somewhere, the compiler will search for a definition of that identifier in either the current module or in any of the modules imported from the current one. If a definition is found in, say in a module called Sentence, then the unqualified name PredVP will be replaced with the explicit qualification Sentence.PredVP. On the other hand, if the source program is already using an explicit qualification like Sentence.PredVP, then the compiler will check whether PredVP is indeed defined in the module Sentence. 
+
+- type checking - here the compiler will check whether all functions and variables are used correctly with respect to their types. For each term that the compiler checks it will also generate a new version of the term after the type checking. The input and output terms may not need to be the same. For example, the compiler may insert explicit type information. It might fill-in implicit arguments, or it may instantiate meta variables.
+
+- partial evaluation - here is where the real compilation starts. The compiler will fully evaluate the term for each linearization to a normal. In the process, all uses of operations will be inlined. This is part of reducing the GF language to a simpler language which does not support operations.
+
+- PMCFG generation - the language that the GF runtime understands is an extension of the PMCFG formalism. Not all features permitted in the GF language are allowed on that level. Most of the uses for that extra features have been eliminated via partial evaluation. If there are any left, then the compilation will abort. The main purpose of the PMCFG generation is to get rid of most of the parameter types in the source grammar. That is possible by generating several specialized linearization rules from a single linearization rule in the source.

doc/hackers-guide/DESIDERATA.md

@@ -0,0 +1,51 @@
+This is an experiment to develop **a majestic new GF runtime**.
+
+The reason is that there are several features that we want to have and they all require a major rewrite of the existing C runtime.
+Instead of beating the old code until it starts doing what we want, it is time to start from scratch.
+
+# New Features
+
+The features that we want are:
+
+- We want to support **even bigger grammars that don't fit in the main memory** anymore. Instead, they should reside on the disc and parts will be loaded on demand.
+The current design is that all memory allocated for the grammars should be from memory-mapped files. In this way the only limit for the grammar size will
+be the size of the virtual memory, i.e. 2^64 bytes. The swap file is completely circumvented, while all of the available RAM can be used as a cache for loading parts
+of the grammar.
+
+- We want to be able to **update grammars dynamically**. This is a highly desired feature since recompiling large grammars takes hours.
+Instead, dynamic updates should happen instantly.
+
+- We want to be able to **store additional information in the PGF**. For example that could be application specific semantic data.
+Another example is to store the source code of the different grammar rules, to allow the compiler to recompile individual rules.
+
+- We want to **allow a single file to contain slightly different versions of the grammar**. This will be a kind of a version control system,
+which will allow different users to store their own grammar extensions while still using the same core content.
+
+- We want to **avoid the exponential explosion in the size of PMCFG** for some grammars. This happens because PMCFG as a formalism is too low-level.
+By enriching it with light-weight variables, we can make it more powerful and hopefully avoid the exponential explosion.
+
+- We want to finally **ditch the old Haskell runtime** which has long outlived its time.
+
+There are also two bugs in the old C runtime whose fixes will require a lot of changes, so instead of fixing the old runtime we do it here:
+
+- **Integer literals in the C runtime** are implemented as 32-bit integers, while the Haskell runtime used unlimited integers.
+Python supports unlimited integers too, so it would be nice to support them in the new runtime as well.
+
+- The old C runtime assumed that **String literals are terminated with the NULL character**. None of the modern languages (Haskell, Python, Java, etc) make
+that assumption, so we should drop it too.
+
+# Consequences
+
+The desired features will have the following implementation cosequences.
+
+- The switch from memory-based to disc-based runtime requires one big change. Before it was easy to just keep a pointer from one object to another.
+Unfortunately this doesn't work with memory-mapped files, since every time when you map a file into memory it may end up at a different virtual address.
+Instead we must use file offsets. In order to make programming simpler, the new runtime will be **implemented in C++ instead of C**. This allows us to overload
+the arrow operator (`->`) which will dynamically convert file offsets to in-memory pointers.
+
+- The choice of C++ also allows us to ditch the old `libgu` library and **use STL** instead.
+
+- The content of the memory mapped files is platform-specific. For that reason there will be two grammar representations:
+     - **Native Grammar Format** (`.ngf`) - which will be instantly loadable by just mapping it to memory, but will be platform-dependent.
+     - **Portable Grammar Format** (`.pgf`) - which will take longer to load but will be more compact and platform independent.
+  The runtime will be able to load `.pgf` files and convert them to `.ngf`. Conversely `.pgf` can be exported from the current `.ngf`.

doc/hackers-guide/README.md

@@ -0,0 +1,20 @@
+# The Hacker's Guide to GF
+
+This is the hacker's guide to GF, for the guide to the galaxy, see the full edition [here](https://en.wikipedia.org/wiki/The_Hitchhiker%27s_Guide_to_the_Galaxy).
+Here we will limit outselves to the vastly narrower domain of the [GF](https://www.grammaticalframework.org) runtime. This means that we will not meet
+any [Vogons](https://en.wikipedia.org/wiki/Vogon), but we will touch upon topics like memory management, databases, transactions, compilers,
+functional programming, theorem proving and sometimes even languages. Subjects that no doubt would interest any curious hacker.
+
+So, **Don't Panic!** and keep reading. This is a live document and will develop together with the runtime itself.
+
+**TABLE OF CONTENTS**
+
+1. Compilation
+   1. [Overview](CompilationOverview.md)
+   1. [Lambda Calculus](LambdaCalculus.md)
+   2. [Parallel Multiple Context-Free Grammars](PMCFG.md)
+2. Runtime
+   1. [Desiderata](DESIDERATA.md)
+   2. [Memory Model](memory_model.md)
+   3. [Abstract Expressions](abstract_expressions.md)
+   4. [Transactions](transactions.md)

doc/hackers-guide/abstract_expressions.md

@@ -0,0 +1,192 @@
+# Data Marshalling Strategies
+
+The runtime is designed to be used from a high-level programming language, which means that there are frequent foreign calls between the host language and C. This also implies that all the data must be frequently marshalled between the binary representations of the two languages. This is usually trivial and well supported for primitive types like numbers and strings but for complex data structures we need to design our own strategy.
+
+The most central data structure in GF is of course the abstract syntax expression. The other two secondary but closely related structures are types and literals. These are complex structures and no high-level programming language will let us to manipulate them directly unless if they are in the format that the runtime of the language understands. There are three main strategies to deal with complex data accross a language boundry:
+
+1. Keep the data in the C world and provide only an opaque handle to the host language. This means that all operations over the data must be done in C via foreign calls.
+2. Design a native host-language representation. For each foreign call the data is copied from the host language to the C representation and vice versa. Copying is obviously bad, but not too bad if the data is small. The added benefit is that now both languages have first-class access to the data. As a bonus, the garbage collector of the host language now understands the data and can immediately release it if part of it becomes unreachable.
+3. Keep the data in the host language. The C code has only an indirect access via opaque handles and calls back to the host language. The program in the host language has first-class access and the garbage collector can work with the data. No copying is needed.
+
+The old C runtime used option 1. Obviously, this means that abstract expressions cannot be manipulated directly, but this is not the only problem. When the application constructs abstract expressions from different pieces, a whole a lot of overhead is added. First, the design was such that data in C must always be allocated from a memory pool. This means that even if we want to make a simple function application, we first must allocate a pool which adds memory overhead. In addition, the host language must allocate an object which wraps arround the C structure. The net effect is that while the plain abstract function application requires the allocation of only two pointers, the actually allocated data may be several times bigger if the application builds the expression piece by piece. The situation is better if the expression is entirely created from the runtime and the application just needs to keep a reference to it.
+
+Another problem is that when the runtime has to create a whole bunch of expressions, for instance as a result from parsing or random and exhaustive generation, then all the expressions are allocated in the same memory pool. The application gets separate handles to each of the produced expressions, but the memory pool is released only after all of the handles become unreachable. Obviously the problem here is that different expressions share the same pool. Unfortunately this is hard to avoid since although the expressions are different, they usually share common subexpression. Identifying the shared parts would be expensive and at the end it might mean that each expression node must be allocated in its own pool.
+
+The path taken in the new runtime is a combination of strategies 2 and 3. The abstract expressions are stored in the heap of the host language and use a native for that language representation.
+
+# Abstract Expressions in Different Languages
+
+In Haskell, abstract expressions are represented with an algebraic data type:
+```Haskell
+data Expr =
+   EAbs BindType Var Expr
+ | EApp Expr Expr
+ | ELit Literal
+ | EMeta  MetaId
+ | EFun   Fun
+ | EVar   Int
+ | ETyped Expr Type
+ | EImplArg Expr
+```
+while in Python and all other object-oriented languages an expression is represented with objects of different classes:
+```Python
+class Expr: pass
+class ExprAbs(Expr): pass
+class ExprApp(Expr): pass
+class ExprLit(Expr): pass
+class ExprMeta(Expr): pass
+class ExprFun(Expr): pass
+class ExprVar(Expr): pass
+class ExprTyped(Expr): pass
+class ExprImplArg(Expr): pass
+```
+
+The runtime needs its own representation as well but only when an expression is stored in a .ngf file. This happens for instance with all types in the abstract syntax of the grammar. Since the type system allows dependent types, some type signature might contain expressions too. Another appearance for abstract expressions is in function definitions, i.e. in the def rules.
+
+Expressions in the runtime are represented with C structures which on the other hand may contain tagged references to other structures. The lowest four bits of each reference encode the type of structure that it points to, while the rest contain the file offsets in the memory mapped file. For example, function application is represented as:
+```C++
+struct PgfExprApp {
+  static const uint8_t tag = 1;
+
+  PgfExpr fun;
+  PgfExpr arg;
+};
+```
+Here the constant `tag` says that any reference to a PgfExprApp structure must contain the value 1 in its lowest four bits. The fields `fun` and `arg` refer to the function and the argument for that application. The type PgfExpr is defined as:
+```C++
+typedef uintptr_t object;
+typedef object PgfExpr;
+```
+In order to dereference an expression, we first neeed to pattern match and then obtain a `ref<>` object:
+```C++
+switch (ref<PgfExpr>::get_tag(e)) {
+...
+case PgfExprApp::tag: {
+    auto eapp = ref<PgfExprApp>::untagged(e);
+    // do something with eapp->fun and eapp->arg
+    ...
+    break;
+}
+...
+}
+```
+
+The representation in the runtime is internal and should never be exposed to the host language. Moreover, these structures live in the memory mapped file and as we discussed in Section "[Memory Model](memory_model.md)" accessing them requires special care. This also means that occasionally the runtime must make a copy from the native representation to the host representation and vice versa. For example, function:
+```Haskell
+functionType :: PGF -> Fun -> Maybe Type
+```
+must look up the type of an abstract syntax function in the .ngf file and return its type. The type, however, is in the native representation and it must first be copied in the host representation. The converse also happens. When the compiler wants to add a new abstract function to the grammar, it creates its type in the Haskell heap, which the runtime later copies to the native representation in the .ngf file. This is not much different from any other database. The database file usually uses a different data representation than what the host language has.
+
+In most other runtime operations, copying is not necessary. The only thing that the runtime needs to know is how to create new expressions in the heap of the host and how to pattern match on them. For that it calls back to code implemented differently for each host language. For example in:
+```Haskell
+readExpr :: String -> Maybe Expr
+```
+the runtime knows how to read an abstract syntax expression, while for the construction of the actual value it calls back to Haskell. Similarly:
+```Haskell
+showExpr :: [Var] -> Expr -> String
+```
+uses code implemented in Haskell to pattern match on the different algebraic constructors, while the text generation itself happens inside the runtime.
+
+# Marshaller and Unmarshaller
+
+The marshaller and the unmarshaller are the two key data structures which bridge together the different representation realms for abstract expressions and types. The structures have two equivalent definitions, one in C++:
+```C++
+struct PgfMarshaller {
+    virtual object match_lit(PgfUnmarshaller *u, PgfLiteral lit)=0;
+    virtual object match_expr(PgfUnmarshaller *u, PgfExpr expr)=0;
+    virtual object match_type(PgfUnmarshaller *u, PgfType ty)=0;
+};
+
+struct PgfUnmarshaller {
+    virtual PgfExpr eabs(PgfBindType btype, PgfText *name, PgfExpr body)=0;
+    virtual PgfExpr eapp(PgfExpr fun, PgfExpr arg)=0;
+    virtual PgfExpr elit(PgfLiteral lit)=0;
+    virtual PgfExpr emeta(PgfMetaId meta)=0;
+    virtual PgfExpr efun(PgfText *name)=0;
+    virtual PgfExpr evar(int index)=0;
+    virtual PgfExpr etyped(PgfExpr expr, PgfType typ)=0;
+    virtual PgfExpr eimplarg(PgfExpr expr)=0;
+    virtual PgfLiteral lint(size_t size, uintmax_t *v)=0;
+    virtual PgfLiteral lflt(double v)=0;
+    virtual PgfLiteral lstr(PgfText *v)=0;
+    virtual PgfType dtyp(int n_hypos, PgfTypeHypo *hypos,
+                         PgfText *cat,
+                         int n_exprs, PgfExpr *exprs)=0;
+    virtual void free_ref(object x)=0;
+};
+```
+and one in C:
+```C
+typedef struct PgfMarshaller PgfMarshaller;
+typedef struct PgfMarshallerVtbl PgfMarshallerVtbl;
+struct PgfMarshallerVtbl {
+    object (*match_lit)(PgfUnmarshaller *u, PgfLiteral lit);
+    object (*match_expr)(PgfUnmarshaller *u, PgfExpr expr);
+    object (*match_type)(PgfUnmarshaller *u, PgfType ty);
+};
+struct PgfMarshaller {
+    PgfMarshallerVtbl *vtbl;
+};
+
+typedef struct PgfUnmarshaller PgfUnmarshaller;
+typedef struct PgfUnmarshallerVtbl PgfUnmarshallerVtbl;
+struct PgfUnmarshallerVtbl {
+    PgfExpr (*eabs)(PgfUnmarshaller *this, PgfBindType btype, PgfText *name, PgfExpr body);
+    PgfExpr (*eapp)(PgfUnmarshaller *this, PgfExpr fun, PgfExpr arg);
+    PgfExpr (*elit)(PgfUnmarshaller *this, PgfLiteral lit);
+    PgfExpr (*emeta)(PgfUnmarshaller *this, PgfMetaId meta);
+    PgfExpr (*efun)(PgfUnmarshaller *this, PgfText *name);
+    PgfExpr (*evar)(PgfUnmarshaller *this, int index);
+    PgfExpr (*etyped)(PgfUnmarshaller *this, PgfExpr expr, PgfType typ);
+    PgfExpr (*eimplarg)(PgfUnmarshaller *this, PgfExpr expr);
+    PgfLiteral (*lint)(PgfUnmarshaller *this, size_t size, uintmax_t *v);
+    PgfLiteral (*lflt)(PgfUnmarshaller *this, double v);
+    PgfLiteral (*lstr)(PgfUnmarshaller *this, PgfText *v);
+    PgfType (*dtyp)(PgfUnmarshaller *this,
+                    int n_hypos, PgfTypeHypo *hypos,
+                    PgfText *cat,
+                    int n_exprs, PgfExpr *exprs);
+    void (*free_ref)(PgfUnmarshaller *this, object x);
+};
+struct PgfUnmarshaller {
+    PgfUnmarshallerVtbl *vtbl;
+};
+```
+Which one you will get, depends on whether you import `pgf/pgf.h` from C or C++.
+
+As we can see, most of the arguments for the different methods are of type `PgfExpr`, `PgfType` or `PgfLiteral`. These are all just type synonyms for the type `object`, which on the other hand is nothing else but a number with enough bits to hold an address if necessary. The interpretation of the number depends on the realm in which the object lives. The following table shows the interpretations for four languages as well as the one used internally in the .ngf files:
+|          | PgfExpr        | PgfLiteral        | PgfType        |
+|----------|----------------|-------------------|----------------|
+| Haskell  | StablePtr Expr | StablePtr Literal | StablePtr Type |
+| Python   | ExprObject *   | PyObject *        | TypeObject *   |
+| Java     | jobject        | jobject           | jobject        |
+| .NET     | GCHandle       | GCHandle          | GCHandle       |
+| internal | file offset    | file offset       | file offset    |
+
+The marshaller is the structure that lets the runtime to pattern match on an expression. When one of the match methods is executed, it checks the kind of expr, literal or type and calls the corresponding method from the unmarshaller which it gets as an argument. The method on the other hand gets as arguments the corresponding sub-expressions and attributes.
+
+Generally the role of an unmarshaller is to construct things. For example, the variable `unmarshaller` in `PGF2.FFI` is an object which can construct new expressions in the Haskell heap from the already created children. Function `readExpr`, for instance, passes that one to the runtime to instruct it that the result must be in the Haskell realm.
+
+Constructing objects is not the only use of an unmarshaller. The implementation of `showExpr` passes to `pgf_print_expr` an abstract expression in Haskell and the `marshaller` defined in PGF2.FFI. That marshaller knows how to pattern match on Haskell expressions and calls the right methods from whatever unmarhaller is given to it. What it will get in that particular case is a special unmarshaller which does not produce new representations of abstract expressions, but generates a string.
+
+
+# Literals
+
+Finally, we should have a few remarks about how values of the literal types `String`, `Int` and `Float` are represented in the runtime.
+
+`String` is represented as the structure:
+```C
+typedef struct {
+    size_t size;
+    char text[];
+} PgfText;
+```
+Here the first field is the size of the string in number of bytes. The second field is the string itself, encoded in UTF-8. Just like in most modern languages, the string may contain the zero character and that is not an indication for end of string. This means that functions like `strlen` and `strcat` should never be used when working with PgfText. Despite that the text is not zero terminated, the runtime always allocates one more last byte for the text content and sets it to zero. That last byte is not included when calculating the field `size`. The purpose is that with that last zero byte the GDB debugger knows how to show the string properly. Most of the time, this doesn't incur any memory overhead either since `malloc` always allocates memory in size divisible by the size of two machine words. The consequence is that usually there are some byte left unused at the end of every string anyway.
+
+`Int` is like the integers in Haskell and Python and can have arbitrarily many digits. In the runtime, the value is represented as an array of `uintmax_t` values. Each of these values contains as many decimal digits as it is possible to fit in `uintmax_t`. For example on a 64-bit machine, 
+the maximal value that fits is 18446744073709551616. However, the left-most digit here is at most 1, this means that if we want to represend an arbitrary sequence of digits, the maximal length of the sequence must be at most 19. Similarly on a 32-bit machine each value in the array will store 9 decimal digits. Finally the sign of the number is stored as the sign of the first number in the array which is always threated as `intmax_t`.
+
+Just to have an example, the number `-774763251095801167872` is represented as the array `{-77, 4763251095801167872}`. Note that this representation is not at all suitable for implementing arithmetics with integers, but is very simple to use for us since the runtime only needs to to parse and linearize numbers.
+
+`Float` is trivial and is just represented as the type `double` in C/C++. This can also be seen in the type of the method `lflt` in the unmarshaller.
+

doc/hackers-guide/memory_model.md

@@ -0,0 +1,136 @@
+# The different storage files
+
+The purpose of the `.ngf` files is to be used as on-disk databases that store grammars. Their format is platform-dependent and they should not be copied from
+one platform to another. In contrast the `.pgf` files are platform-independent and can be moved around. The runtime can import a `.pgf` file and create an `.ngf` file.
+Conversely a `.pgf` file can be exported from an already existing `.ngf` file.
+
+The internal relation between the two files is more interesting. The runtime uses its own memory allocator which always allocates memory from a memory mapped file.
+The file may be explicit or an anonymous one. The `.ngf` is simply a memory image saved in a file. This means that loading the file is always immediate.
+You just create a new mapping and the kernel will load memory pages on demand.
+
+On the other hand a `.pgf` file is a version of the grammar serialized in a platform-independent format. This means that loading this type of file is always slower.
+Fortunately, you can always create an `.ngf` file from it to speed up later reloads.
+
+The runtime has three ways to load a grammar:
+
+#### 1. Loading a `.pgf`
+```Haskell
+readPGF :: FilePath -> IO PGF
+```
+This loads the `.pgf` into an anonymous memory-mapped file. In practice, this means that instead of allocating memory from an explicit file, the runtime will still
+use the normal swap file.
+
+#### 2. Loading a `.pgf` and booting a new `.ngf`
+```Haskell
+bootPGF :: FilePath -> FilePath -> IO PGF
+```
+The grammar is loaded from a `.pgf` (the first argument) and the memory is mapped to an explicit `.ngf` (second argument). The `.ngf` file is created by the function
+and a file with the same name should not exist before the call.
+
+#### 3. Loading an existing memory image
+```Haskell
+readNGF :: FilePath -> IO PGF
+```
+Once an `.ngf` file exists, it can be mapped back to memory by using this function. This call is always guaranteed to be fast. The same function can also
+create new empty `.ngf` files. If the file does not exist, then a new one will be created which contains an empty grammar. The grammar could then be extended
+by dynamically adding functions and categories.
+
+# The content of an `.ngf` file
+
+The `.ngf` file is a memory image but this is not the end of the story. The problem is that there is no way to control at which address the memory image would be
+mapped. On Posix systems, `mmap` takes as hint the mapping address but the kernel may choose to ignore it. There is also the flag `MAP_FIXED`, which makes the hint
+into a constraint, but then the kernel may fail to satisfy the constraint. For example that address may already be used for something else. Furthermore, if the
+same file is mapped from several processes (if they all load the same grammar), it would be difficult to find an address which is free in all of them.
+Last but not least using `MAP_FIXED` is considered a security risk.
+
+Since the start address of the mapping can change, using traditional memory pointers withing the mapped area is not possible. The only option is to use offsets
+relative to the beginning of the area. In other words, if normally we would have written `p->x`, now we have the offset `o` which we must use like this:
+```C++
+((A*) (current_base+o))->x
+```
+
+Writing the explicit pointer arithmetics and typecasts, each time when we dereference a pointer, is not better than Vogon poetry and it
+becomes worse when using a chain of arrow operators. The solution is to use the operator overloading in C++.
+There is the type `ref<A>` which wraps around a file offset to a data item of type `A`. The operators `->` and `*`
+are overloaded for the type and they do the necessary pointer arithmetics and type casts.
+
+This solves the problem with code readability but creates another problem. How do `->` and `*` know the address of the memory mapped area? Obviously,
+`current_base` must be a global variable and there must be a way to initialize it. More specifically it must be thread-local to allow different threads to
+work without collisions.
+
+A database (a memory-mapped file) in the runtime is represented by the type `DB`. Before any of the data in the database is accessed, the database must
+be brought into scope. Bringing into scope means that `current_base` is initialized to point to the mapping area for that database. After that any dereferencing
+of a reference will be done relative to the corresponding database. This is how scopes are defined:
+```C++
+{
+    DB_scope scope(db, READER_SCOPE);
+    ...
+}
+```
+Here `DB_scope` is a helper type and `db` is a pointer to the database that you want to bring into scope. The constructor for `DB_scope` saves the old value
+for `current_base` and then sets it to point to the area of the given database. Conversely, the destructor restores the previous value.
+
+The use of `DB_scope` is reentrant, i.e. you can do this:
+```C++
+{
+    DB_scope scope(db1, READER_SCOPE);
+    ...
+    {
+        DB_scope scope(db2, READER_SCOPE);
+        ...
+    }
+    ...
+}
+```
+What you can't do is to have more than one database in scope simultaneously. Fortunately, that is not needed. All API functions start a scope
+and the internals of the runtime always work with the current database in scope.
+
+Note the flag `READER_SCOPE`. You can use either `READER_SCOPE` or `WRITER_SCOPE`. In addition to selecting the database, the `DB_scope` also enforces
+the single writer/multiple readers policy. The main problem is that a writer may have to enlarge the current file, which consequently may mean
+that the kernel should relocate the mapping area to a new address. If there are readers at the same time, they may break since they expect that the mapped
+area is at a particular location.
+
+# Developing writers
+
+There is one important complication when developing procedures modifying the database. Every call to `DB::malloc` may potentially have to enlarge the mapped area
+which sometimes leads to changing `current_base`. That would not have been a problem if GCC was not sometimes caching variables in registers. Look at the following code:
+```C++
+p->r = foo();
+```
+Here `p` is a reference which is used to access another reference `r`. On the other hand, `foo()` is a procedure which directly or indirectly calls `DB::malloc`.
+GCC compiles assignments by first computing the address to modify, and then it evaluates the right hand side. This means that while `foo()` is being evaluated the address computed on the left-hand side is saved in a register or somewhere in the stack. But now, if it happens that the allocation in `foo()` has changed
+`current_base`, then the saved address is no longer valid.
+
+That first problem is solved by overloading the assignment operator for `ref<A>`:
+```C++
+ref<A>& operator= (const ref<A>& r) {
+    offset = r.offset;
+    return *this;
+}
+```
+On first sight, nothing special happens here and it looks like the overloading is redundant. However, now the assignments are compiled in a very different way.
+The overloaded operator is inlined, so there is no real method call and we don't get any overhead. The real difference is that now, whatever is on the left-hand side of the assignment becomes the value of the `this` pointer, and `this` is always the last thing to be evaluated in a method call. This solves the problem.
+`foo()` is evaluated first and if it changes `current_base`, the change will be taken into account when computing the left-hand side of the assignment.
+
+Unfortunately, this is not the only problem. A similar thing happens when the arguments of a function are calls to other functions. See this:
+```C++
+foo(p->r,bar(),q->r)
+```
+Where now `bar()` is the function that performs allocation. The compiler is free to keep in a register the value of `current_base` that it needs for the evaluation of
+`p->r`, while it evaluates `bar()`. But if `current_base` has changed, then the saved value would be invalid while computing `q->r`. There doesn't seem to be
+a work around for this. The only solution is to:
+
+**Never call a function that allocates as an argument to another function**
+
+Instead we call allocating functions on a separate line and we save the result in a temporary variable.
+
+
+# Thread-local variables
+
+A final remark is the compilation of thread-local variables. When a thread-local variable is compiled in a position-dependent code, i.e. in executables, it is
+compiled efficiently by using the `fs` register which points to the thread-local segment. Unfortunately, that is not the case by default for shared
+libraries like our runtime. In that case, GCC applies the global-dynamic model which means that access to a thread local variable is internally implemented
+with a call to the function `__tls_get_addr`. Since `current_base` is used all the time, this adds overhead.
+
+The solution is to define the variable with the attribute `__attribute__((tls_model("initial-exec")))` which says that it should be treated as if it is defined
+in an executable. This removes the overhead, but adds the limitation that the runtime should not be loaded with `dlopen`.

doc/hackers-guide/transactions.md

@@ -0,0 +1,131 @@
+# Transactions
+
+The `.ngf` files that the runtime creates are actual databases which are used to get quick access to the grammars. Like in any database, we also make it possible to dynamically change the data. In our case this means that we can add and remove functions and categories at any time. Moreover, any changes happen in transactions which ensure that changes are not visible until the transaction is commited. The rest of the document describes how the transactions are implemented.
+
+# Databases and Functional Languages
+
+The database model of the runtime is specifically designed to be friendly towards pure functional languages like Haskell. In a usual database, updates happen constantly and therefore executing one and the same query at different times would yield different results. In our grammar databases, queries correspond to operations like parsing, linearization and generation. This means that if we had used the usual database model, all these operations would have to be bound to the IO monad. Consider this example:
+```Haskell
+main = do
+  gr <- readNGF "Example.ngf"
+  functionType gr "f" >>= print
+  -- modify the grammar gr
+  functionType gr "f" >>= print
+```
+Here we ask for the type of a function before and after an arbitrary update in the grammar `gr`. Obviously if we allow that then `functionType` would have to be in the IO monad, e.g.:
+
+```Haskell
+functionType :: PGF -> Fun -> IO Type
+```
+
+Although this is a possible way to go, it would mean that the programmer would have to do all grammar related work in the IO. This is not nice and against the spirit of functional programming. Moreover, all previous implementations of the runtime have assumed that most operations are pure. If we go along that path then this will cause a major breaking change.
+
+Fortunately there is an alternative. Read-only operations remain pure functions, but any update should create a new revision of the database rather than modifying the existing one. Compare this example with the previous:
+```Haskell
+main = do
+  gr <- readNGF "Example.ngf"
+  print (functionType gr "f")
+  gr2 <- modifyPGF gr $ do
+           -- do all updates here
+  print (functionType gr2 "f")
+```
+Here `modifyPGF` allows us to do updates but the updates are performed on a freshly created clone of the grammar `gr`. The original grammar is never ever modified. After the changes the variable `gr2` is a reference to the new revision. While the transaction is in progress we cannot see the currently changing revision, and therefore all read-only operations can remain pure. Only after the transaction is complete do we get to use `gr2`, which will not change anymore.
+
+Note also that above `functionType` is used with its usual pure type:
+```Haskell
+functionType :: PGF -> Fun -> Type
+```
+This is safe since the API never exposes database revisions which are not complete. Furthermore, the programmer is free to keep several revisions of the same database simultaneously. In this example:
+```Haskell
+main = do
+  gr <- readNGF "Example.ngf"
+  gr2 <- modifyPGF gr $ do
+           -- do all updates here
+  print (functionType gr "f", functionType gr2 "f")
+```
+The last line prints the type of function `"f"` in both the old and the new revision. Both are still available.
+
+The API as described so far would have been complete if all updates were happening in a single thread. In reality we can expect that there might be several threads or processes modifying the database. The database ensures a multiple readers/single writer exclusion but this doesn't mean that another process/thread cannot modify the database while the current one is reading an old revision. In a parallel setting, `modifyPGF` first merges the revision which the process is using with the latest revision in the database. On top of that the specified updates are performed. The final revision after the updates is returned as a result.
+
+**TODO: Interprocess synhronization is still not implemented**
+
+**TODO: Merges are still not implemented.**
+
+The process can also ask for the latest revision by calling `checkoutPGF`, see bellow.
+
+# Databases and Imperative Languages
+
+In imperative languages, the state of the program constantly changes and the considerations in the last section do not apply. All read-only operations always work with the latest revision. Bellow is the previous example translated to Python:
+```Python
+gr = readNGF("Example.ngf")
+print(functionType(gr,"f"))
+with gr.transaction() as t:
+  # do all updates here by using t
+print(functionType(gr,"f"))
+```
+Here the first call to `functionType` returns the old type of "f", while the second call retrives the type after the updates. The transaction itself is initiated by the `with` statement. Inside the with statement `gr` will still refer to the old revision since the new one is not complete yet. If the `with` statement is finished without exceptions then `gr` is updated to point to the new one. If an exception occurs then the new revision is discarded, which corresponds to a transaction rollback. Inside the `with` block, the object `t` of type `Transaction` provides methods for modifying the data.
+
+# Branches
+
+Since the database already supports revisions, it is a simple step to support branches as well. A branch is just a revision with a name. When you open a database with `readNGF`, the runtime looks up and returns the revision (branch) with name `master`. There might be other branches as well. You can retrieve a specific branch by calling:
+```Haskell
+checkoutPGF :: PGF -> String -> IO (Maybe PGF)
+```
+Here the string is the branch name. New branches can be created by using:
+```Haskell
+branchPGF :: PGF -> String -> Transaction a -> IO PGF
+```
+Here we start with an existing revision, apply a transaction and store the result in a new branch with the given name.
+
+# Implementation
+
+The low-level API for transactions consists of only four functions:
+```C
+PgfRevision pgf_clone_revision(PgfDB *db, PgfRevision revision,
+                               PgfText *name,
+                               PgfExn *err);
+
+void pgf_free_revision(PgfDB *pgf, PgfRevision revision);
+
+void pgf_commit_revision(PgfDB *db, PgfRevision revision,
+                         PgfExn *err);
+
+PgfRevision pgf_checkout_revision(PgfDB *db, PgfText *name,
+                                  PgfExn *err);
+```
+Here `pgf_clone_revision` makes a copy of an existing revision and — if `name` is not `NULL` — changes its name. The new revision is transient and exists only until it is released with `pgf_free_revision`. Transient revisions can be updated with the API for adding functions and categories. To make a revision persistent, call `pgf_commit_revision`. After the revision is made persistent it will stay in the database even after you call `pgf_free_revision`. Moreover, it will replace the last persistent revision with the same name. The old revision will then become transient and will exist only until all clients call `pgf_free_revision` for it.
+
+Persistent revisions can never be updated. Instead you clone it to create a new transient revision. That one is updated and finally it replaces the existing persistent revision.
+
+This design for transactions may sound unusual but it is just another way to present the copy-on-write strategy. There instead of transaction logs, each change to the data is written in a new place and the result is made available only after all changes are in place. This is for instance what the [LMDB](http://www.lmdb.tech/doc/) (Lightning Memory-Mapped Database) does and it has also served as an inspiration for us.
+
+## Functional Data Structures
+
+From an imperative point of view, it may sound wasteful that a new copy of the grammar is created for each transaction. Functional programmers on the other hand know that with a functional data structure, you can make a copy which shares as much of the data with the original as possible. Each new version copies only those bits that are different from the old one. For example the main data structure that we use to represent the abstract syntax of a grammar is a size-balanced binary tree as described by:
+
+- Stephen Adams, "Efficient sets: a balancing act", Journal of Functional Programming 3(4):553-562, October 1993, http://www.swiss.ai.mit.edu/~adams/BB/.
+
+- J. Nievergelt and E.M. Reingold, "Binary search trees of bounded balance", SIAM journal of computing 2(1), March 1973.
+
+
+## Garbage Collection
+
+We use reference counting to keep track of which objects should be kept alive. For instance, `pgf_free_revision` knows that a transient revision should be removed only when its reference count reaches zero. This means that there is no process or thread using it. The function also checks whether the revision is persistent. Persistent revisions are never removed since they can always be retrieved with `checkoutPGF`.
+
+Clients are supposed to correctly use `pgf_free_revision` to indicate that they don't need a revision any more. Unfortunately, this is not always possible to guarantee. For example many languages with garbage collection will call `pgf_free_revision` from a finalizer method. In some languages, however, the finalizer is not guaranteed to be executed if the process terminates before the garbage collection is done. Haskell is one of those languages. Even in languages with reference counting like Python, the process may get killed by the operating system and then the finalizer may still not be executed.
+
+The solution is that we count on the database clients to correctly report when a revision is not needed. However, on a fresh database restart we explictly clean all left over transient revisions. This means that even if a client is killed or if it does not correctly release its revisions, the worst that can happen is a memory leak until the next restart.
+
+
+## Atomicity
+
+The transactions serve two goals. First they make it possible to isolate readers from seeing unfinished changes from writers. Second, they ensure atomicity. A database change should be either completely done or not done at all. The use of transient revisions ensures the isolation but the atomicity is only partly taken care of.
+
+Think about what happens when a writer starts updating a transient revision. All the data is allocated in a memory mapped file. From the point of view of the runtime, all changes happen in memory. When all is done, the runtime calls `msync` which tells the kernel to flush all dirty pages to disk. The problem is that the kernel is also free to flush pages at any time. For instance, if there is not enough memory, it may decide to swap out pages earlier and reuse the released physical space to swap in other virtual pages. This would be fine if the transaction eventually succeeds. However, if this doesn't happen then the image in the file is already changed.
+
+We can avoid the situation by calling [mlock](https://man7.org/linux/man-pages/man2/mlock.2.html) and telling the kernel that certain pages should not be swapped out. The question is which pages to lock. We can lock them all, but this is too much. That would mean that as soon as a page is touched it will never leave the physical memory. Instead, it would have been nice to tell the kernel -- feel free to swap out clean pages but, as soon as they get dirty, keep them in memory until further notice. Unfortunately there is no way to do that directly.
+
+The work around is to first use [mprotect](https://man7.org/linux/man-pages/man2/mprotect.2.html) and keep all pages as read-only. Any attempt to change a page will cause segmentation fault which we can capture. If the change happens during a transaction then we can immediate lock the page and add it to the list of modified pages. When a transaction is successful we sync all modified pages. If an attempt to change a page happens outside of a transaction, then this is either a bug in the runtime or the client is trying to change an address which it should not change. In any case this prevents unintended changes in the data.
+
+
+**TODO: atomicity is not implemented yet**

download/index-3.11.md

@@ -1,8 +1,9 @@
 ---
 title: Grammatical Framework Download and Installation
-...
+date: 25 July 2021
+---
 
-**GF 3.11** was released on ... December 2020.
+**GF 3.11** was released on 25 July 2021.
 
 What's new? See the [release notes](release-3.11.html).
 
@@ -24,22 +25,25 @@ Binary packages are available for Debian/Ubuntu, macOS, and Windows and include:
 
 Unlike in previous versions, the binaries **do not** include the RGL.
 
-[Binary packages on GitHub](https://github.com/GrammaticalFramework/gf-core/releases/tag/RELEASE-3.11)
+[Binary packages on GitHub](https://github.com/GrammaticalFramework/gf-core/releases/tag/3.11)
 
 #### Debian/Ubuntu
 
+There are two versions: `gf-3.11-ubuntu-18.04.deb` for Ubuntu 18.04 (Cosmic), and `gf-3.11-ubuntu-20.04.deb` for Ubuntu 20.04 (Focal).
+
 To install the package use:
+
 ```
-sudo dpkg -i gf_3.11.deb
+sudo apt-get install ./gf-3.11-ubuntu-*.deb
 ```
 
-The Ubuntu `.deb` packages should work on Ubuntu 16.04, 18.04 and similar Linux distributions.
+<!-- The Ubuntu `.deb` packages should work on Ubuntu 16.04, 18.04 and similar Linux distributions. -->
 
 #### macOS
 
 To install the package, just double-click it and follow the installer instructions.
 
-The packages should work on at least 10.13 (High Sierra) and 10.14 (Mojave).
+The packages should work on at least Catalina and Big Sur.
 
 #### Windows
 
@@ -49,24 +53,39 @@ You will probably need to update the `PATH` environment variable to include your
 
 For more information, see [Using GF on Windows](https://www.grammaticalframework.org/~inari/gf-windows.html) (latest updated for Windows 10).
 
-## Installing the latest release from source
+## Installing from Hackage
+
+_Instructions applicable for macOS, Linux, and WSL2 on Windows._
 
 [GF is on Hackage](http://hackage.haskell.org/package/gf), so under
 normal circumstances the procedure is fairly simple:
 
-1.  Install a recent version of the [Haskell Platform](http://hackage.haskell.org/platform) (see note below)
-2.  `cabal update`
-3.  On Linux: install some C libraries from your Linux distribution (see note below)
-4.  `cabal install gf`
-
-You can also download the source code release from [GitHub](https://github.com/GrammaticalFramework/gf-core/releases),
-and follow the instructions below under **Installing from the latest developer source code**.
+```
+cabal update
+cabal install gf-3.11
+```
 
 ### Notes
 
+**GHC version**
+
+The GF source code is known to be compilable with GHC versions 7.10 through to 8.10.
+
+**Obtaining Haskell**
+
+There are various ways of obtaining Haskell, including:
+
+- ghcup
+    1.  Install from https://www.haskell.org/ghcup/
+    2.  `ghcup install ghc 8.10.4`
+    3.  `ghcup set ghc 8.10.4`
+- Haskell Platform https://www.haskell.org/platform/
+- Stack https://haskellstack.org/
+
+
 **Installation location**
 
-The above steps installs GF for a single user.
+The above steps install GF for a single user.
 The executables are put in `$HOME/.cabal/bin` (or on macOS in `$HOME/Library/Haskell/bin`),
 so you might want to add this directory to your path (in `.bash_profile` or similar):
 
@@ -74,47 +93,38 @@ so you might want to add this directory to your path (in `.bash_profile` or simi
 PATH=$HOME/.cabal/bin:$PATH
 ```
 
-**Build tools**
-
-In order to compile GF you need the build tools **Alex** and **Happy**.
-These can be installed via Cabal, e.g.:
-
-```
-cabal install alex happy
-```
-
-or obtained by other means, depending on your OS.
-
 **Haskeline**
 
 GF uses [`haskeline`](http://hackage.haskell.org/package/haskeline), which
 on Linux depends on some non-Haskell libraries that won't be installed
-automatically by cabal, and therefore need to be installed manually.
+automatically by Cabal, and therefore need to be installed manually.
 Here is one way to do this:
 
 - On Ubuntu: `sudo apt-get install libghc-haskeline-dev`
 - On Fedora: `sudo dnf install ghc-haskeline-devel`
 
-**GHC version**
+## Installing from source code
 
-The GF source code has been updated to compile with GHC versions 7.10 through to 8.8.
+**Obtaining**
 
-## Installing from the latest developer source code
+To obtain the source code for the **release**,
+download it from [GitHub](https://github.com/GrammaticalFramework/gf-core/releases).
 
-If you haven't already, clone the repository with:
+Alternatively, to obtain the **latest version** of the source code:
 
+1. If you haven't already, clone the repository with:
 ```
 git clone https://github.com/GrammaticalFramework/gf-core.git
 ```
-
-If you've already cloned the repository previously, update with:
-
+2. If you've already cloned the repository previously, update with:
 ```
 git pull
 ```
 
-Then install with:
 
+**Installing**
+
+You can then install with:
 ```
 cabal install
 ```
@@ -125,7 +135,7 @@ or, if you're a Stack user:
 stack install
 ```
 
-The above notes for installing from source apply also in these cases.
+<!--The above notes for installing from source apply also in these cases.-->
 For more info on working with the GF source code, see the
 [GF Developers Guide](../doc/gf-developers.html).
 

download/index.html

@@ -1,8 +1,8 @@
 <html>
 <head>
-  <meta http-equiv="refresh" content="0; URL=/download/index-3.10.html" />
+  <meta http-equiv="refresh" content="0; URL=/download/index-3.11.html" />
 </head>
 <body>
-  You are being redirected to <a href="index-3.10.html">the current version</a> of this page.
+  You are being redirected to <a href="index-3.11.html">the current version</a> of this page.
 </body>
 </html>

download/release-3.11.md

@@ -1,7 +1,7 @@
 ---
 title: GF 3.11 Release Notes
-date: ... December 2020
-...
+date: 25 July 2021
+---
 
 ## Installation
 
@@ -12,24 +12,27 @@ See the [download page](index-3.11.html).
 From this release, the binary GF core packages do not contain the RGL.
 The RGL's release cycle is now completely separate from GF's. See [RGL releases](https://github.com/GrammaticalFramework/gf-rgl/releases).
 
-Over 400 changes have been pushed to GF core
+Over 500 changes have been pushed to GF core
 since the release of GF 3.10 in December 2018.
 
 ## General
 
 - Make the test suite work again.
 - Compatibility with new versions of GHC, including multiple Stack files for the different versions.
-- Updates to build scripts and CI.
-- Bug fixes.
+- Support for newer version of Ubuntu 20.04 in the precompiled binaries.
+- Updates to build scripts and CI workflows.
+- Bug fixes and code cleanup.
 
 ## GF compiler and run-time library
 
-- Huge improvements in time & space requirements for grammar compilation (pending [#87](https://github.com/GrammaticalFramework/gf-core/pull/87)).
 - Add CoNLL output to `visualize_tree` shell command.
 - Add canonical GF as output format in the compiler.
 - Add PGF JSON as output format in the compiler.
 - Deprecate JavaScript runtime in favour of updated [TypeScript runtime](https://github.com/GrammaticalFramework/gf-typescript).
+- Improvements in time & space requirements when compiling certain grammars.
 - Improvements to Haskell export.
+- Improvements to the GF shell.
+- Improvements to canonical GF compilation.
 - Improvements to the C runtime.
 - Improvements to `gf -server` mode.
 - Clearer compiler error messages.

gf.cabal

@@ -1,19 +1,19 @@
 name: gf
-version: 3.10.4-git
+version: 3.11.0-git
 
-cabal-version: >= 1.22
+cabal-version: 1.22
 build-type: Custom
 license: OtherLicense
 license-file: LICENSE
 category: Natural Language Processing, Compiler
 synopsis: Grammatical Framework
 description: GF, Grammatical Framework, is a programming language for multilingual grammar applications
-homepage: http://www.grammaticalframework.org/
+homepage: https://www.grammaticalframework.org/
 bug-reports: https://github.com/GrammaticalFramework/gf-core/issues
-maintainer: Thomas Hallgren
-tested-with: GHC==7.10.3, GHC==8.0.2, GHC==8.2.2, GHC==8.4.3
+tested-with: GHC==7.10.3, GHC==8.0.2, GHC==8.10.4
 
 data-dir: src
+extra-source-files: WebSetup.hs
 data-files:
   www/*.html
   www/*.css
@@ -41,39 +41,34 @@ data-files:
 
 custom-setup
   setup-depends:
-    base,
-    Cabal >=1.22.0.0,
-    directory,
-    filepath,
-    process >=1.0.1.1
+    base >= 4.9.1 && < 4.15,
+    Cabal >= 1.22.0.0,
+    directory >= 1.3.0 && < 1.4,
+    filepath >= 1.4.1 && < 1.5,
+    process >= 1.0.1.1 && < 1.7
 
 source-repository head
-  type:     git
+  type: git
   location: https://github.com/GrammaticalFramework/gf-core.git
 
 flag interrupt
   Description: Enable Ctrl+Break in the shell
-  Default:     True
+  Default: True
 
 flag server
   Description: Include --server mode
-  Default:     True
+  Default: True
 
 flag network-uri
    description: Get Network.URI from the network-uri package
    default: True
 
---flag new-comp
---  Description: Make -new-comp the default
---  Default:     True
-
-flag c-runtime
-  Description: Include functionality from the C run-time library (which must be installed already)
-  Default:     False
-
-Library
+executable gf
+  hs-source-dirs: src/programs, src/compiler
+  main-is: gf-main.hs
   default-language:    Haskell2010
-  build-depends: base >= 4.6 && <5,
+  build-depends: pgf2,
+                 base >= 4.6 && <5,
                  array,
                  containers,
                  bytestring,
@@ -82,81 +77,18 @@ Library
                  pretty,
                  mtl,
                  exceptions,
-                 fail,
-                 -- For compatability with ghc < 8
-                 -- We need transformers-compat >= 0.6.3, but that is only in newer snapshots where it is redundant.
-                 transformers-compat,
-                 ghc-prim
-  hs-source-dirs: src/runtime/haskell
-
-  other-modules:
-    -- not really part of GF but I have changed the original binary library
-    -- and we have to keep the copy for now.
-    Data.Binary
-    Data.Binary.Put
-    Data.Binary.Get
-    Data.Binary.Builder
-    Data.Binary.IEEE754
-
---ghc-options: -fwarn-unused-imports
---if impl(ghc>=7.8)
---  ghc-options: +RTS -A20M -RTS
-  ghc-prof-options: -fprof-auto
-
-  exposed-modules:
-    PGF
-    PGF.Internal
-    PGF.Haskell
-
-  other-modules:
-    PGF.Data
-    PGF.Macros
-    PGF.Binary
-    PGF.Optimize
-    PGF.Printer
-    PGF.CId
-    PGF.Expr
-    PGF.Generate
-    PGF.Linearize
-    PGF.Morphology
-    PGF.Paraphrase
-    PGF.Parse
-    PGF.Probabilistic
-    PGF.SortTop
-    PGF.Tree
-    PGF.Type
-    PGF.TypeCheck
-    PGF.Forest
-    PGF.TrieMap
-    PGF.VisualizeTree
-    PGF.ByteCode
-    PGF.OldBinary
-    PGF.Utilities
-
-  if flag(c-runtime)
-    exposed-modules: PGF2
-    other-modules:   PGF2.FFI PGF2.Expr PGF2.Type
-                     GF.Interactive2 GF.Command.Commands2
-    hs-source-dirs:  src/runtime/haskell-bind
-    build-tools:     hsc2hs
-    extra-libraries: pgf gu
-    c-sources:       src/runtime/haskell-bind/utils.c
-    cc-options:      -std=c99
-
----- GF compiler as a library:
-
-  build-depends: filepath, directory>=1.2, time,
+                 ghc-prim,
+                 filepath, directory>=1.2, time,
                  process, haskeline, parallel>=3, json
+  ghc-options: -threaded
 
-  hs-source-dirs: src/compiler
-  exposed-modules:
+  other-modules:
     GF
     GF.Support
     GF.Text.Pretty
     GF.Text.Lexing
     GF.Grammar.Canonical
 
-  other-modules:
     GF.Main GF.Compiler GF.Interactive
 
     GF.Compile GF.CompileInParallel GF.CompileOne  GF.Compile.GetGrammar
@@ -177,26 +109,21 @@ Library
     GF.Command.TreeOperations
     GF.Compile.CFGtoPGF
     GF.Compile.CheckGrammar
-    GF.Compile.Compute.ConcreteNew
-    GF.Compile.Compute.Predef
-    GF.Compile.Compute.Value
+    GF.Compile.Compute.Concrete
     GF.Compile.ExampleBased
     GF.Compile.Export
     GF.Compile.GenerateBC
     GF.Compile.GeneratePMCFG
     GF.Compile.GrammarToPGF
     GF.Compile.Multi
-    GF.Compile.Optimize
+    GF.Compile.OptimizePGF
     GF.Compile.PGFtoHaskell
     GF.Compile.PGFtoJava
     GF.Haskell
     GF.Compile.ConcreteToHaskell
     GF.Compile.GrammarToCanonical
     GF.Grammar.CanonicalJSON
-    GF.Compile.PGFtoJS
     GF.Compile.PGFtoJSON
-    GF.Compile.PGFtoProlog
-    GF.Compile.PGFtoPython
     GF.Compile.ReadFiles
     GF.Compile.Rename
     GF.Compile.SubExOpt
@@ -206,7 +133,6 @@ Library
     GF.Compile.TypeCheck.Concrete
     GF.Compile.TypeCheck.ConcreteNew
     GF.Compile.TypeCheck.Primitives
-    GF.Compile.TypeCheck.RConcrete
     GF.Compile.TypeCheck.TC
     GF.Compile.Update
     GF.Data.BacktrackM
@@ -264,92 +190,38 @@ Library
     GF.System.Directory
     GF.System.Process
     GF.System.Signal
+    GF.System.NoSignal
     GF.Text.Clitics
     GF.Text.Coding
+    GF.Text.Lexing
     GF.Text.Transliterations
     Paths_gf
-
-  if flag(c-runtime)
-    cpp-options: -DC_RUNTIME
-
-  if flag(server)
-    build-depends: httpd-shed>=0.4.0.3, network>=2.3 && <2.7,
-                   cgi>=3001.2.2.0
-    if flag(network-uri)
-      build-depends: network-uri>=2.6, network>=2.6
-    else
-      build-depends: network<2.6
-
-    cpp-options: -DSERVER_MODE
-    other-modules:
-      GF.Server
-      PGFService
-      RunHTTP
-      SimpleEditor.Convert
-      SimpleEditor.JSON
-      SimpleEditor.Syntax
-      URLEncoding
-      CGI
-      CGIUtils
-      Cache
-      Fold
-      ExampleDemo
-      ExampleService
-    hs-source-dirs: src/server src/server/transfer src/example-based
-
-  if flag(interrupt)
-    cpp-options: -DUSE_INTERRUPT
-    other-modules: GF.System.UseSignal
-  else
-    other-modules: GF.System.NoSignal
-
-  if impl(ghc>=7.8)
-    build-tools: happy>=1.19, alex>=3.1
---  ghc-options: +RTS -A20M -RTS
-  else
-    build-tools: happy, alex>=3
-
-  ghc-options: -fno-warn-tabs
+    
+    -- not really part of GF but I have changed the original binary library
+    -- and we have to keep the copy for now.
+    Data.Binary
+    Data.Binary.Put
+    Data.Binary.Get
+    Data.Binary.Builder
+    Data.Binary.IEEE754
 
   if os(windows)
-    build-depends: Win32
+    build-depends:
+      Win32 >= 2.3.1.1 && < 2.7
   else
-    build-depends: unix, terminfo>=0.4
-
-  if impl(ghc>=8.2)
-    ghc-options: -fhide-source-paths
-
-Executable gf
-  hs-source-dirs: src/programs
-  main-is: gf-main.hs
-  default-language:    Haskell2010
-  build-depends: gf, base
-  ghc-options: -threaded
---ghc-options: -fwarn-unused-imports
-
-  if impl(ghc>=7.0)
-    ghc-options: -rtsopts -with-rtsopts=-I5
-    if impl(ghc<7.8)
-      ghc-options: -with-rtsopts=-K64M
-
-  ghc-prof-options: -auto-all
-
-  if impl(ghc>=8.2)
-    ghc-options: -fhide-source-paths
-
-executable pgf-shell
---if !flag(c-runtime)
-  buildable: False
-  main-is:             pgf-shell.hs
-  hs-source-dirs:      src/runtime/haskell-bind/examples
-  build-depends:       gf, base, containers, mtl, lifted-base
-  default-language:    Haskell2010
-  if impl(ghc>=7.0)
-    ghc-options: -rtsopts
+    build-depends:
+      terminfo >=0.4.0 && < 0.5,
+      unix >= 2.7.2 && < 2.8
 
 test-suite gf-tests
-  type:           exitcode-stdio-1.0
-  main-is:        run.hs
+  type: exitcode-stdio-1.0
+  main-is: run.hs
   hs-source-dirs: testsuite
-  build-depends:  base>=4.3 && <5, Cabal>=1.8, directory, filepath, process
-  default-language:    Haskell2010
+  build-depends:
+    base >= 4.9.1 && < 4.15,
+    Cabal >= 1.8,
+    directory >= 1.3.0 && < 1.4,
+    filepath >= 1.4.1 && < 1.5,
+    process >= 1.4.3 && < 1.7
+  build-tool-depends: gf:gf
+  default-language: Haskell2010

index.html

@@ -8,7 +8,7 @@
 
   <meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no">
   <link rel="stylesheet" href="https://stackpath.bootstrapcdn.com/bootstrap/4.1.3/css/bootstrap.min.css" integrity="sha384-MCw98/SFnGE8fJT3GXwEOngsV7Zt27NXFoaoApmYm81iuXoPkFOJwJ8ERdknLPMO" crossorigin="anonymous">
-  <link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.4.2/css/all.css" integrity="sha384-/rXc/GQVaYpyDdyxK+ecHPVYJSN9bmVFBvjA/9eOB+pb3F2w2N6fc5qB9Ew5yIns" crossorigin="anonymous">
+  <link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.15.4/css/all.css" crossorigin="anonymous">
 
   <link rel="alternate" href="https://github.com/GrammaticalFramework/gf-core/" title="GF GitHub repository">
 </head>
@@ -85,10 +85,27 @@
   <div class="col-sm-6 col-md-3 mb-4">
     <h3>Contribute</h3>
     <ul class="mb-2">
-      <li><a href="http://groups.google.com/group/gf-dev">Mailing List</a></li>
+      <li>
+        <a href="https://web.libera.chat/?channels=#gf">
+          <i class="fas fa-hashtag"></i>
+          IRC
+        </a>
+        /
+        <a href="https://discord.gg/EvfUsjzmaz">
+          <i class="fab fa-discord"></i>
+          Discord
+        </a>
+      </li>
+      <li>
+        <a href="https://stackoverflow.com/questions/tagged/gf">
+          <i class="fab fa-stack-overflow"></i>
+          Stack Overflow
+        </a>
+      </li>
+      <li><a href="https://groups.google.com/group/gf-dev">Mailing List</a></li>
       <li><a href="https://github.com/GrammaticalFramework/gf-core/issues">Issue Tracker</a></li>
-      <li><a href="doc/gf-people.html">Authors</a></li>
       <li><a href="//school.grammaticalframework.org/2020/">Summer School</a></li>
+      <li><a href="doc/gf-people.html">Authors</a></li>
     </ul>
     <a href="https://github.com/GrammaticalFramework/" class="btn btn-primary ml-3">
       <i class="fab fa-github mr-1"></i>
@@ -154,7 +171,7 @@ least one, it may help you to get a first idea of what GF is.
 <div class="row">
 
   <div class="col-md-6">
-    <h2>Applications & Availability</h2>
+    <h2>Applications & availability</h2>
     <p>
     GF can be used for building
     <a href="//cloud.grammaticalframework.org/translator/">translation systems</a>,
@@ -214,56 +231,45 @@ least one, it may help you to get a first idea of what GF is.
     </p>
 
     <p>
-    We run the IRC channel <strong><code>#gf</code></strong> on the Freenode network, where you are welcome to look for help with small questions or just start a general discussion.
-    You can <a href="https://webchat.freenode.net/?channels=gf">open a web chat</a>
-    or <a href="/irc/">browse the channel logs</a>.
+    We run the IRC channel <strong><code>#gf</code></strong> on the Libera network, where you are welcome to look for help with small questions or just start a general discussion.
+    You can <a href="https://web.libera.chat/?channels=#gf">open a web chat</a>
+    or <a href="https://www.grammaticalframework.org/irc/?C=M;O=D">browse the channel logs</a>.
     </p>
     <p>
-    If you have a larger question which the community may benefit from, we recommend you ask it on the <a href="http://groups.google.com/group/gf-dev">mailing list</a>.
+    There is also a <a href="https://discord.gg/EvfUsjzmaz">GF server on Discord</a>.
+    </p>
+
+    <p>
+    For bug reports and feature requests, please create an issue in the
+    <a href="https://github.com/GrammaticalFramework/gf-core/issues">GF Core</a> or
+    <a href="https://github.com/GrammaticalFramework/gf-rgl/issues">RGL</a> repository.
+
+    For programming questions, consider asking them on <a href="https://stackoverflow.com/questions/tagged/gf">Stack Overflow with the <code>gf</code> tag</a>.
+    If you have a more general question to the community, we recommend you ask it on the <a href="http://groups.google.com/group/gf-dev">mailing list</a>.
     </p>
 
   </div>
 
   <div class="col-md-6">
     <h2>News</h2>
-
     <dl class="row">
+      <dt class="col-sm-3 text-center text-nowrap">2021-07-25</dt>
+      <dd class="col-sm-9">
+        <strong>GF 3.11 released.</strong>
+        <a href="download/release-3.11.html">Release notes</a>
+      </dd>
+      <dt class="col-sm-3 text-center text-nowrap">2021-05-05</dt>
+      <dd class="col-sm-9">
+        <a href="https://cloud.grammaticalframework.org/wordnet/">GF WordNet</a> now supports languages for which there are no other WordNets. New additions: Afrikaans, German, Korean, Maltese, Polish, Somali, Swahili.
+      </dd>
       <dt class="col-sm-3 text-center text-nowrap">2021-03-01</dt>
       <dd class="col-sm-9">
-        <a href="//school.grammaticalframework.org/2020/">Seventh GF Summer School</a>, in Singapore and online, 26 July &ndash; 8 August 2021.
+        <a href="//school.grammaticalframework.org/2020/">Seventh GF Summer School</a>, in Singapore and online, 26 July &ndash; 6 August 2021.
       </dd>
       <dt class="col-sm-3 text-center text-nowrap">2020-09-29</dt>
       <dd class="col-sm-9">
         <a href="https://www.mitpressjournals.org/doi/pdf/10.1162/COLI_a_00378">Abstract Syntax as Interlingua</a>: Scaling Up the Grammatical Framework from Controlled Languages to Robust Pipelines. A paper in Computational Linguistics (2020) summarizing much of the development in GF in the past ten years.
       </dd>
-      <dt class="col-sm-3 text-center text-nowrap">2018-12-03</dt>
-      <dd class="col-sm-9">
-        <a href="//school.grammaticalframework.org/2018/">Sixth GF Summer School</a> in Stellenbosch (South Africa), 3–14 December 2018
-      </dd>
-      <dt class="col-sm-3 text-center text-nowrap">2018-12-02</dt>
-      <dd class="col-sm-9">
-        <strong>GF 3.10 released.</strong>
-        <a href="download/release-3.10.html">Release notes</a>
-      </dd>
-      <dt class="col-sm-3 text-center text-nowrap">2018-07-25</dt>
-      <dd class="col-sm-9">
-        The GF repository has been split in two:
-        <a href="https://github.com/GrammaticalFramework/gf-core">gf-core</a> and
-        <a href="https://github.com/GrammaticalFramework/gf-rgl">gf-rgl</a>.
-        The original <a href="https://github.com/GrammaticalFramework/GF">GF</a> repository is now archived.
-      </dd>
-      <dt class="col-sm-3 text-center text-nowrap">2017-08-11</dt>
-      <dd class="col-sm-9">
-        <strong>GF 3.9 released.</strong>
-        <a href="download/release-3.9.html">Release notes</a>
-      </dd>
-      <dt class="col-sm-3 text-center text-nowrap">2017-06-29</dt>
-      <dd class="col-sm-9">
-        GF is moving to <a href="https://github.com/GrammaticalFramework/GF/">GitHub</a>.</dd>
-      <dt class="col-sm-3 text-center text-nowrap">2017-03-13</dt>
-      <dd class="col-sm-9">
-        <a href="//school.grammaticalframework.org/2017/">GF Summer School</a> in Riga (Latvia), 14-25 August 2017
-      </dd>
     </dl>
 
     <h2>Projects</h2>
@@ -333,7 +339,7 @@ least one, it may help you to get a first idea of what GF is.
     Libraries are at the heart of modern software engineering. In natural language
     applications, libraries are a way to cope with thousands of details involved in
     syntax, lexicon, and inflection. The
-    <a href="lib/doc/synopsis/index.html">GF resource grammar library</a> has
+    <a href="lib/doc/synopsis/index.html">GF resource grammar library</a> (RGL) has
     support for an increasing number of languages, currently including
     Afrikaans,
     Amharic (partial),

src/compiler/Data/Binary/Builder.hs

@@ -68,7 +68,7 @@ import qualified Data.ByteString.Lazy as L
 import Data.ByteString.Base (inlinePerformIO)
 import qualified Data.ByteString.Base as S
 #else
-import Data.ByteString.Internal (inlinePerformIO)
+import Data.ByteString.Internal (accursedUnutterablePerformIO)
 import qualified Data.ByteString.Internal as S
 --import qualified Data.ByteString.Lazy.Internal as L
 #endif
@@ -199,7 +199,7 @@ defaultSize = 32 * k - overhead
 
 -- | Sequence an IO operation on the buffer
 unsafeLiftIO :: (Buffer -> IO Buffer) -> Builder
-unsafeLiftIO f =  Builder $ \ k buf -> inlinePerformIO $ do
+unsafeLiftIO f =  Builder $ \ k buf -> accursedUnutterablePerformIO $ do
     buf' <- f buf
     return (k buf')
 {-# INLINE unsafeLiftIO #-}

src/compiler/Data/Binary/Get.hs

@@ -423,7 +423,7 @@ readN n f = fmap f $ getBytes n
 getPtr :: Storable a => Int -> Get a
 getPtr n = do
     (fp,o,_) <- readN n B.toForeignPtr
-    return . B.inlinePerformIO $ withForeignPtr fp $ \p -> peek (castPtr $ p `plusPtr` o)
+    return . B.accursedUnutterablePerformIO $ withForeignPtr fp $ \p -> peek (castPtr $ p `plusPtr` o)
 {- INLINE getPtr -}
 
 ------------------------------------------------------------------------

src/compiler/GF/Command/Abstract.hs

@@ -1,6 +1,6 @@
 module GF.Command.Abstract(module GF.Command.Abstract,Expr,showExpr,Term) where
 
-import PGF(CId,mkCId,Expr,showExpr)
+import PGF2(Expr,showExpr)
 import GF.Grammar.Grammar(Term)
 
 type Ident = String
@@ -11,7 +11,7 @@ type Pipe = [Command]
 
 data Command
    = Command Ident [Option] Argument
-   deriving (Eq,Ord,Show)
+   deriving Show
 
 data Option
   = OOpt Ident
@@ -29,13 +29,7 @@ data Argument
   | ATerm Term
   | ANoArg
   | AMacro Ident
-  deriving (Eq,Ord,Show)
-
-valCIdOpts :: String -> CId -> [Option] -> CId
-valCIdOpts flag def opts =
-  case [v | OFlag f (VId v) <- opts, f == flag] of
-    (v:_) -> mkCId v
-    _     -> def
+  deriving Show
 
 valIntOpts :: String -> Int -> [Option] -> Int
 valIntOpts flag def opts =
@@ -49,6 +43,18 @@ valStrOpts flag def opts =
     v:_ -> valueString v
     _   -> def
 
+maybeIntOpts :: String -> a -> (Int -> a) -> [Option] -> a
+maybeIntOpts flag def fn opts =
+  case [v | OFlag f (VInt v) <- opts, f == flag] of
+    (v:_) -> fn v
+    _     -> def
+
+maybeStrOpts :: String -> a -> (String -> a) -> [Option] -> a
+maybeStrOpts flag def fn opts =
+  case listFlags flag opts of
+    v:_ -> fn (valueString v)
+    _   -> def
+
 listFlags flag opts = [v | OFlag f v <- opts, f == flag]
 
 valueString v =

src/compiler/GF/Command/CommandInfo.hs

@@ -3,8 +3,7 @@ import GF.Command.Abstract(Option,Expr,Term)
 import GF.Text.Pretty(render)
 import GF.Grammar.Printer() -- instance Pretty Term
 import GF.Grammar.Macros(string2term)
-import qualified PGF as H(showExpr)
-import qualified PGF.Internal as H(Literal(LStr),Expr(ELit)) ----
+import PGF2(mkStr,unStr,showExpr)
 
 data CommandInfo m = CommandInfo {
   exec     :: [Option] -> CommandArguments -> m CommandOutput,
@@ -38,21 +37,19 @@ class Monad m => TypeCheckArg m where typeCheckArg :: Expr -> m Expr
 
 --------------------------------------------------------------------------------
 
-data CommandArguments = Exprs [Expr] | Strings [String] | Term Term
+data CommandArguments = Exprs [(Expr,Float)] | Strings [String] | Term Term
 
 newtype CommandOutput = Piped (CommandArguments,String) ---- errors, etc
 
 -- ** Converting command output
 fromStrings ss         = Piped (Strings ss, unlines ss)
-fromExprs   es         = Piped (Exprs es,unlines (map (H.showExpr []) es))
+fromExprs show_p es    = Piped (Exprs es,unlines (map (\(e,p) -> (if show_p then (++) ("["++show p++"] ") else id) (showExpr [] e)) es))
 fromString  s          = Piped (Strings [s], s)
 pipeWithMessage es msg = Piped (Exprs es,msg)
 pipeMessage msg        = Piped (Exprs [],msg)
 pipeExprs   es         = Piped (Exprs es,[]) -- only used in emptyCommandInfo
 void                   = Piped (Exprs [],"")
 
-stringAsExpr = H.ELit . H.LStr -- should be a pattern macro
-
 -- ** Converting command input
 
 toStrings args =
@@ -61,23 +58,23 @@ toStrings args =
       Exprs es -> zipWith showAsString (True:repeat False) es
       Term t -> [render t]
   where
-    showAsString first t =
-      case t of
-        H.ELit (H.LStr s) -> s
-        _ -> ['\n'|not first] ++
-             H.showExpr [] t ---newline needed in other cases than the first
+    showAsString first (e,p) =
+      case unStr e of
+        Just s  -> s
+        Nothing -> ['\n'|not first] ++
+                   showExpr [] e ---newline needed in other cases than the first
 
 toExprs args =
   case args of
-    Exprs es -> es
-    Strings ss -> map stringAsExpr ss
-    Term t -> [stringAsExpr (render t)]
+    Exprs es -> map fst es
+    Strings ss -> map mkStr ss
+    Term t -> [mkStr (render t)]
 
 toTerm args =
   case args of
     Term t -> t
     Strings ss -> string2term $ unwords ss -- hmm
-    Exprs es -> string2term $ unwords $ map (H.showExpr []) es -- hmm
+    Exprs es -> string2term $ unwords $ map (showExpr [] . fst) es -- hmm
 
 -- ** Creating documentation
 

src/compiler/GF/Command/Commands.hs

@@ -1,16 +1,11 @@
-{-# LANGUAGE FlexibleInstances, UndecidableInstances #-}
+{-# LANGUAGE FlexibleInstances, UndecidableInstances, CPP #-}
 module GF.Command.Commands (
-  PGFEnv,HasPGFEnv(..),pgf,mos,pgfEnv,pgfCommands,
+  HasPGF(..),pgfCommands,
   options,flags,
   ) where
-import Prelude hiding (putStrLn,(<>)) -- GHC 8.4.1 clash with Text.PrettyPrint
+import Prelude hiding (putStrLn,(<>))
 
-import PGF
-
-import PGF.Internal(lookStartCat,functionsToCat,lookValCat,restrictPGF,hasLin)
-import PGF.Internal(abstract,funs,cats,Expr(EFun)) ----
-import PGF.Internal(ppFun,ppCat)
-import PGF.Internal(optimizePGF)
+import PGF2
 
 import GF.Compile.Export
 import GF.Compile.ToAPI
@@ -28,28 +23,28 @@ import GF.Command.TreeOperations ---- temporary place for typecheck and compute
 
 import GF.Data.Operations
 
-import PGF.Internal (encodeFile)
+import Data.Char
 import Data.List(intersperse,nub)
 import Data.Maybe
 import qualified Data.Map as Map
 import GF.Text.Pretty
 import Data.List (sort)
+import Control.Monad(mplus)
 import qualified Control.Monad.Fail as Fail
 --import Debug.Trace
 
 
-data PGFEnv = Env {pgf::PGF,mos::Map.Map Language Morpho}
+class (Functor m,Monad m,MonadSIO m) => HasPGF m where getPGF :: m (Maybe PGF)
 
-pgfEnv pgf = Env pgf mos
-  where mos = Map.fromList [(la,buildMorpho pgf la) | la <- languages pgf]
+instance (Monad m,HasPGF m,Fail.MonadFail m) => TypeCheckArg m where
+  typeCheckArg e = do mb_pgf <- getPGF
+                      case mb_pgf of
+                        Just pgf -> either fail
+                                           (return . fst)
+                                           (inferExpr pgf e)
+                        Nothing  -> fail "Import a grammar before using this command"
 
-class (Functor m,Monad m,MonadSIO m) => HasPGFEnv m where getPGFEnv :: m PGFEnv
-
-instance (Monad m,HasPGFEnv m,Fail.MonadFail m) => TypeCheckArg m where
-  typeCheckArg e = (either (fail . render . ppTcError) (return . fst)
-                    . flip inferExpr e . pgf) =<< getPGFEnv
-
-pgfCommands :: HasPGFEnv m => Map.Map String (CommandInfo m)
+pgfCommands :: HasPGF m => Map.Map String (CommandInfo m)
 pgfCommands = Map.fromList [
   ("aw", emptyCommandInfo {
      longname = "align_words",
@@ -62,7 +57,7 @@ pgfCommands = Map.fromList [
        "by the view flag. The target format is png, unless overridden by the",
        "flag -format. Results from multiple trees are combined to pdf with convert (ImageMagick)."
        ],
-     exec = getEnv $ \ opts arg (Env pgf mos) -> do
+     exec = needPGF $ \ opts arg pgf -> do
          let es = toExprs arg
          let langs = optLangs pgf opts
          if isOpt "giza" opts
@@ -74,7 +69,7 @@ pgfCommands = Map.fromList [
              let grph = if null es then [] else lsrc ++ "\n--end_source--\n\n"++ltrg++"\n-end_target--\n\n"++align
              return $ fromString grph
            else do
-             let grphs = map (graphvizAlignment pgf langs) es
+             let grphs = map (graphvizWordAlignment langs graphvizDefaults) es
              if isFlag "view" opts || isFlag "format" opts
                then do
                  let view = optViewGraph opts
@@ -96,6 +91,7 @@ pgfCommands = Map.fromList [
        ("view",  "program to open the resulting file")
        ]
     }),
+
   ("ca", emptyCommandInfo {
      longname = "clitic_analyse",
      synopsis = "print the analyses of all words into stems and clitics",
@@ -106,16 +102,17 @@ pgfCommands = Map.fromList [
        "by the flag '-clitics'. The list of stems is given as the list of words",
        "of the language given by the '-lang' flag."
        ],
-     exec  = getEnv $ \opts ts env -> case opts of
-               _ | isOpt "raw" opts ->
-                    return . fromString .
-                    unlines . map (unwords . map (concat . intersperse "+")) .
-                    map (getClitics (isInMorpho (optMorpho env opts)) (optClitics opts)) .
-                    concatMap words $ toStrings ts
-               _ ->
-                    return . fromStrings .
-                    getCliticsText (isInMorpho (optMorpho env opts)) (optClitics opts) .
-                    concatMap words $ toStrings ts,
+     exec  = needPGF $ \opts ts pgf -> do
+               concr <- optLang pgf opts
+               case opts of
+                 _ | isOpt "raw" opts ->
+                       return . fromString .
+                       unlines . map (unwords . map (concat . intersperse "+")) .
+                       map (getClitics (not . null . lookupMorpho concr) (optClitics opts)) .
+                       concatMap words $ toStrings ts
+                 _ -> return . fromStrings .
+                      getCliticsText (not . null . lookupMorpho concr) (optClitics opts) .
+                      concatMap words $ toStrings ts,
      flags = [
        ("clitics","the list of possible clitics (comma-separated, no spaces)"),
        ("lang",   "the language of analysis")
@@ -147,19 +144,19 @@ pgfCommands = Map.fromList [
        ],
      flags = [
        ("file","the file to be converted (suffix .gfe must be given)"),
-       ("lang","the language in which to parse"),
-       ("probs","file with probabilities to rank the parses")
+       ("lang","the language in which to parse")
        ],
-     exec = getEnv $ \ opts _ env@(Env pgf mos) -> do
+     exec = needPGF $ \opts _ pgf -> do
        let file = optFile opts
-       pgf <- optProbs opts pgf
        let printer = if (isOpt "api" opts) then exprToAPI else (showExpr [])
-       let conf = configureExBased pgf (optMorpho env opts) (optLang pgf opts) printer
+       concr <- optLang pgf opts
+       let conf = configureExBased pgf concr printer
        (file',ws) <- restricted $ parseExamplesInGrammar conf file
        if null ws then return () else putStrLn ("unknown words: " ++ unwords ws)
        return (fromString ("wrote " ++ file')),
      needsTypeCheck = False
      }),
+
   ("gr", emptyCommandInfo {
      longname = "generate_random",
      synopsis = "generate random trees in the current abstract syntax",
@@ -174,54 +171,53 @@ pgfCommands = Map.fromList [
      explanation = unlines [
        "Generates a list of random trees, by default one tree.",
        "If a tree argument is given, the command completes the Tree with values to",
-       "all metavariables in the tree. The generation can be biased by probabilities,",
-       "given in a file in the -probs flag."
+       "all metavariables in the tree. The generation can be biased by probabilities",
+       "if the grammar was compiled with option -probs"
+       ],
+     options = [
+       ("show_probs", "show the probability of each result")
        ],
      flags = [
        ("cat","generation category"),
        ("lang","uses only functions that have linearizations in all these languages"),
-       ("number","number of trees generated"),
-       ("depth","the maximum generation depth"),
-       ("probs", "file with biased probabilities (format 'f 0.4' one by line)")
+       ("number","number of trees generated")
        ],
-     exec = getEnv $ \ opts arg (Env pgf mos) -> do
-       pgf <- optProbs opts (optRestricted opts pgf)
+     exec = needPGF $ \opts arg pgf -> do
        gen <- newStdGen
-       let dp = valIntOpts "depth" 4 opts
        let ts  = case mexp (toExprs arg) of
-                   Just ex -> generateRandomFromDepth gen pgf ex (Just dp)
-                   Nothing -> generateRandomDepth     gen pgf (optType pgf opts) (Just dp)
-       returnFromExprs $ take (optNum opts) ts
+                   Just ex -> generateRandomFrom gen pgf ex
+                   Nothing -> generateRandom     gen pgf (optType pgf opts)
+       returnFromExprs (isOpt "show_probs" opts) $ take (optNum opts) ts
      }),
+
   ("gt", emptyCommandInfo {
      longname = "generate_trees",
      synopsis = "generates a list of trees, by default exhaustive",
      explanation = unlines [
-       "Generates all trees of a given category. By default, ",
-       "the depth is limited to 4, but this can be changed by a flag.",
+       "Generates all trees of a given category.",
        "If a Tree argument is given, the command completes the Tree with values",
        "to all metavariables in the tree."
        ],
+     options = [
+       ("show_probs", "show the probability of each result")
+       ],
      flags = [
        ("cat","the generation category"),
-       ("depth","the maximum generation depth"),
        ("lang","excludes functions that have no linearization in this language"),
        ("number","the number of trees generated")
        ],
      examples = [
-       mkEx "gt                     -- all trees in the startcat, to depth 4",
+       mkEx "gt                     -- all trees in the startcat",
        mkEx "gt -cat=NP -number=16  -- 16 trees in the category NP",
-       mkEx "gt -cat=NP -depth=2    -- trees in the category NP to depth 2",
        mkEx "gt (AdjCN ? (UseN ?))  -- trees of form (AdjCN ? (UseN ?))"
        ],
-     exec = getEnv $ \ opts arg (Env pgf mos) -> do
-       let pgfr = optRestricted opts pgf
-       let dp = valIntOpts "depth" 4 opts
-       let ts = case mexp (toExprs arg) of
-                  Just ex -> generateFromDepth pgfr ex (Just dp)
-                  Nothing -> generateAllDepth pgfr (optType pgf opts) (Just dp)
-       returnFromExprs $ take (optNumInf opts) ts
+     exec = needPGF $ \opts arg pgf -> do
+       let es = case mexp (toExprs arg) of
+                  Just ex -> generateAllFrom pgf ex
+                  Nothing -> generateAll     pgf (optType pgf opts)
+       returnFromExprs (isOpt "show_probs" opts) $ takeOptNum opts es
      }),
+
   ("i", emptyCommandInfo {
      longname = "import",
      synopsis = "import a grammar from source code or compiled .pgf file",
@@ -242,33 +238,28 @@ pgfCommands = Map.fromList [
        ("probs","file with biased probabilities for generation")
        ],
      options = [
-       -- ["gfo", "src", "no-cpu", "cpu", "quiet", "verbose"]
        ("retain","retain operations (used for cc command)"),
        ("src",   "force compilation from source"),
        ("v",     "be verbose - show intermediate status information")
        ],
      needsTypeCheck = False
      }),
+
   ("l", emptyCommandInfo {
      longname = "linearize",
      synopsis = "convert an abstract syntax expression to string",
      explanation = unlines [
-       "Shows the linearization of a Tree by the grammars in scope.",
+       "Shows the linearization of a tree by the grammars in scope.",
        "The -lang flag can be used to restrict this to fewer languages.",
        "A sequence of string operations (see command ps) can be given",
        "as options, and works then like a pipe to the ps command, except",
-       "that it only affect the strings, not e.g. the table labels.",
-       "These can be given separately to each language with the unlexer flag",
-       "whose results are prepended to the other lexer flags. The value of the",
-       "unlexer flag is a space-separated list of comma-separated string operation",
-       "sequences; see example."
+       "that it only affect the strings, not e.g. the table labels."
        ],
      examples = [
        mkEx "l -lang=LangSwe,LangNor no_Utt   -- linearize tree to LangSwe and LangNor",
-       mkEx "gr -lang=LangHin -cat=Cl | l -table -to_devanagari -- hindi table",
-       mkEx "l -unlexer=\"LangAra=to_arabic LangHin=to_devanagari\" -- different unlexers"
+       mkEx "gr -lang=LangHin -cat=Cl | l -table -to_devanagari -- hindi table"
        ],
-     exec = getEnv $ \ opts ts (Env pgf mos) -> return . fromStrings . optLins pgf opts $ toExprs ts,
+     exec = needPGF $ \ opts ts pgf -> return . fromStrings . optLins pgf opts $ toExprs ts,
      options = [
        ("all",    "show all forms and variants, one by line (cf. l -list)"),
        ("bracket","show tree structure with brackets and paths to nodes"),
@@ -276,33 +267,13 @@ pgfCommands = Map.fromList [
        ("list","show all forms and variants, comma-separated on one line (cf. l -all)"),
        ("multi","linearize to all languages (default)"),
        ("table","show all forms labelled by parameters"),
-       ("tabtreebank","show the tree and its linearizations on a tab-separated line"),
-       ("treebank","show the tree and tag linearizations with language names")
-       ] ++ stringOpOptions,
-     flags = [
-       ("lang","the languages of linearization (comma-separated, no spaces)"),
-       ("unlexer","set unlexers separately to each language (space-separated)")
-       ]
-     }),
-  ("lc", emptyCommandInfo {
-     longname = "linearize_chunks",
-     synopsis = "linearize a tree that has metavariables in maximal chunks without them",
-     explanation = unlines [
-       "A hopefully temporary command, intended to work around the type checker that fails",
-       "trees where a function node is a metavariable."
-       ],
-     examples = [
-       mkEx "l -lang=LangSwe,LangNor -chunks ? a b (? c d)"
-       ],
-     exec = getEnv $ \ opts ts (Env pgf mos) -> return . fromStrings $ optLins pgf (opts ++ [OOpt "chunks"]) (toExprs ts),
-     options = [
        ("treebank","show the tree and tag linearizations with language names")
        ] ++ stringOpOptions,
      flags = [
        ("lang","the languages of linearization (comma-separated, no spaces)")
-       ],
-     needsTypeCheck = False
+       ]
      }),
+
   ("ma", emptyCommandInfo {
      longname = "morpho_analyse",
      synopsis = "print the morphological analyses of all words in the string",
@@ -310,18 +281,20 @@ pgfCommands = Map.fromList [
        "Prints all the analyses of space-separated words in the input string,",
        "using the morphological analyser of the actual grammar (see command pg)"
        ],
-     exec  = getEnv $ \opts ts env -> case opts of
-               _ | isOpt "missing" opts ->
-                    return . fromString . unwords .
-                    morphoMissing (optMorpho env opts) .
-                    concatMap words $ toStrings ts
-               _ | isOpt "known" opts ->
-                    return . fromString . unwords .
-                    morphoKnown (optMorpho env opts) .
-                    concatMap words $ toStrings ts
-               _ -> return . fromString . unlines .
-                    map prMorphoAnalysis . concatMap (morphos env opts) .
-                    concatMap words $ toStrings ts,
+     exec  = needPGF $ \opts ts pgf -> do
+               concr <- optLang pgf opts
+               case opts of
+                 _ | isOpt "missing" opts ->
+                      return . fromString . unwords .
+                      morphoMissing concr .
+                      concatMap words $ toStrings ts
+                 _ | isOpt "known" opts ->
+                      return . fromString . unwords .
+                      morphoKnown concr .
+                      concatMap words $ toStrings ts
+                 _ -> return . fromString . unlines .
+                      map prMorphoAnalysis . concatMap (morphos pgf opts) .
+                      concatMap words $ toStrings ts,
      flags = [
        ("lang","the languages of analysis (comma-separated, no spaces)")
        ],
@@ -335,18 +308,16 @@ pgfCommands = Map.fromList [
      longname = "morpho_quiz",
      synopsis = "start a morphology quiz",
      syntax   = "mq (-cat=CAT)? (-probs=FILE)? TREE?",
-     exec = getEnv $ \ opts arg (Env pgf mos) -> do
-         let lang = optLang pgf opts
+     exec = needPGF $ \ opts arg pgf -> do
+         lang <- optLang pgf opts
          let typ  = optType pgf opts
-         pgf <- optProbs opts pgf
          let mt = mexp (toExprs arg)
          restricted $ morphologyQuiz mt pgf lang typ
          return void,
      flags = [
        ("lang","language of the quiz"),
        ("cat","category of the quiz"),
-       ("number","maximum number of questions"),
-       ("probs","file with biased probabilities for generation")
+       ("number","maximum number of questions")
        ]
      }),
 
@@ -357,24 +328,25 @@ pgfCommands = Map.fromList [
        "Shows all trees returned by parsing a string in the grammars in scope.",
        "The -lang flag can be used to restrict this to fewer languages.",
        "The default start category can be overridden by the -cat flag.",
-       "See also the ps command for lexing and character encoding.",
-       "",
-       "The -openclass flag is experimental and allows some robustness in ",
-       "the parser. For example if -openclass=\"A,N,V\" is given, the parser",
-       "will accept unknown adjectives, nouns and verbs with the resource grammar."
+       "See also the ps command for lexing and character encoding."
+       ],
+     exec = needPGF $ \opts ts pgf ->
+              return $
+                foldr (joinPiped . fromParse1 opts) void 
+                  (concat [
+                              [(s,parse concr (optType pgf opts) s) |
+                                            concr <- optLangs pgf opts] 
+                           | s <- toStrings ts]),
+     options = [
+       ("show_probs", "show the probability of each result")
        ],
-     exec = getEnv $ \ opts ts (Env pgf mos) ->
-              return $ fromParse opts (concat [map ((,) s) (par pgf opts s) | s <- toStrings ts]),
      flags = [
        ("cat","target category of parsing"),
        ("lang","the languages of parsing (comma-separated, no spaces)"),
-       ("openclass","list of open-class categories for robust parsing"),
-       ("depth","maximal depth for proof search if the abstract syntax tree has meta variables")
-       ],
-     options = [
-       ("bracket","prints the bracketed string from the parser")
+       ("number","limit the results to the top N trees")
        ]
      }),
+
   ("pg", emptyCommandInfo { -----
      longname = "print_grammar",
      synopsis = "print the actual grammar with the given printer",
@@ -394,9 +366,8 @@ pgfCommands = Map.fromList [
         " " ++ opt ++ "\t\t" ++ expl |
            ((opt,_),expl) <- outputFormatsExpl, take 1 expl /= "*"
        ]),
-     exec  = getEnv $ \opts _ env -> prGrammar env opts,
+     exec  = needPGF $ \opts _ pgf -> prGrammar pgf opts,
      flags = [
-       --"cat",
        ("file",   "set the file name when printing with -pgf option"),
        ("lang",   "select languages for the some options (default all languages)"),
        ("printer","select the printing format (see flag values above)")
@@ -416,6 +387,7 @@ pgfCommands = Map.fromList [
        mkEx ("pg -funs | ? grep \" S ;\"  -- show functions with value cat S")
        ]
      }),
+
   ("pt", emptyCommandInfo {
      longname = "put_tree",
      syntax = "pt OPT? TREE",
@@ -429,11 +401,12 @@ pgfCommands = Map.fromList [
      examples = [
        mkEx "pt -compute (plus one two)                               -- compute value"
        ],
-     exec = getEnv $ \ opts arg (Env pgf mos) ->
-            returnFromExprs . takeOptNum opts . treeOps pgf opts $ toExprs arg,
+     exec = needPGF $ \opts arg pgf ->
+            returnFromExprs False .  takeOptNum opts . map (flip (,) 0) . treeOps pgf opts $ toExprs arg,
      options = treeOpOptions undefined{-pgf-},
      flags = [("number","take at most this many trees")] ++ treeOpFlags undefined{-pgf-}
      }),
+
   ("rf",  emptyCommandInfo {
      longname = "read_file",
      synopsis = "read string or tree input from a file",
@@ -446,10 +419,9 @@ pgfCommands = Map.fromList [
        ],
      options = [
        ("lines","return the list of lines, instead of the singleton of all contents"),
-       ("paragraphs","return the list of paragraphs, as separated by empty lines"),
        ("tree","convert strings into trees")
        ],
-     exec = getEnv $ \ opts _ (Env pgf mos) -> do
+     exec = needPGF $ \ opts _ pgf -> do
        let file = valStrOpts "file" "_gftmp" opts
        let exprs []         = ([],empty)
            exprs ((n,s):ls) | null s
@@ -458,12 +430,12 @@ pgfCommands = Map.fromList [
                                 Just e  -> let (es,err) = exprs ls
                                            in case inferExpr pgf e of
                                                 Right (e,t) -> (e:es,err)
-                                                Left tcerr  -> (es,"on line" <+> n <> ':' $$ nest 2 (ppTcError tcerr) $$ err)
+                                                Left err    -> (es,"on line" <+> n <> ':' $$ nest 2 err $$ err)
                                 Nothing -> let (es,err) = exprs ls
                                            in (es,"on line" <+> n <> ':' <+> "parse error" $$ err)
            returnFromLines ls = case exprs ls of
                                   (es, err) | null es   -> return $ pipeMessage $ render (err $$ "no trees found")
-                                            | otherwise -> return $ pipeWithMessage es (render err)
+                                            | otherwise -> return $ pipeWithMessage (map (flip (,) 0) es) (render err)
 
        s <- restricted $ readFile file
        case opts of
@@ -472,56 +444,26 @@ pgfCommands = Map.fromList [
          _ | isOpt "tree" opts ->
                returnFromLines [(1::Int,s)]
          _ | isOpt "lines" opts -> return (fromStrings $ lines s)
-         _ | isOpt "paragraphs" opts -> return (fromStrings $ toParagraphs $ lines s)
          _ -> return (fromString s),
      flags = [("file","the input file name")]
      }),
-  ("rt", emptyCommandInfo {
-     longname = "rank_trees",
-     synopsis = "show trees in an order of decreasing probability",
-     explanation = unlines [
-       "Order trees from the most to the least probable, using either",
-       "even distribution in each category (default) or biased as specified",
-       "by the file given by flag -probs=FILE, where each line has the form",
-       "'function probability', e.g. 'youPol_Pron  0.01'."
-       ],
-     exec = getEnv $ \ opts arg (Env pgf mos) -> do
-         let ts = toExprs arg
-         pgf <- optProbs opts pgf
-         let tds = rankTreesByProbs pgf ts
-         if isOpt "v" opts
-           then putStrLn $
-                  unlines [showExpr []  t ++ "\t--" ++ show d | (t,d) <- tds]
-           else return ()
-         returnFromExprs $ map fst tds,
-     flags = [
-       ("probs","probabilities from this file (format 'f 0.6' per line)")
-       ],
-     options = [
-       ("v","show all trees with their probability scores")
-       ],
-     examples = [
-      mkEx "p \"you are here\" | rt -probs=probs | pt -number=1 -- most probable result"
-      ]
-     }),
+
   ("tq", emptyCommandInfo {
      longname = "translation_quiz",
      syntax   = "tq -from=LANG -to=LANG (-cat=CAT)? (-probs=FILE)? TREE?",
      synopsis = "start a translation quiz",
-     exec = getEnv $ \ opts arg (Env pgf mos) -> do
-         let from = optLangFlag "from" pgf opts
-         let to   = optLangFlag "to" pgf opts
+     exec = needPGF $ \ opts arg pgf -> do
+         from <- optLangFlag "from" pgf opts
+         to   <- optLangFlag "to"   pgf opts
          let typ  = optType pgf opts
          let mt   = mexp (toExprs arg)
-         pgf <- optProbs opts pgf
          restricted $ translationQuiz mt pgf from to typ
          return void,
      flags = [
        ("from","translate from this language"),
        ("to","translate to this language"),
        ("cat","translate in this category"),
-       ("number","the maximum number of questions"),
-       ("probs","file with biased probabilities for generation")
+       ("number","the maximum number of questions")
        ],
      examples = [
        mkEx ("tq -from=Eng -to=Swe                               -- any trees in startcat"),
@@ -529,7 +471,6 @@ pgfCommands = Map.fromList [
        ]
      }),
 
-
   ("vd", emptyCommandInfo {
      longname = "visualize_dependency",
      synopsis = "show word dependency tree graphically",
@@ -547,7 +488,7 @@ pgfCommands = Map.fromList [
        "flag -format. Results from multiple trees are combined to pdf with convert (ImageMagick).",
        "See also 'vp -showdep' for another visualization of dependencies." 
        ],
-     exec = getEnv $ \ opts arg (Env pgf mos) -> do
+     exec = needPGF $ \ opts arg pgf -> do
          let absname = abstractName pgf
          let es = toExprs arg
          let debug = isOpt "v" opts
@@ -560,8 +501,8 @@ pgfCommands = Map.fromList [
          mclab <- case cnclabels of
            "" -> return Nothing
            _  -> (Just . getCncDepLabels) `fmap` restricted (readFile cnclabels)
-         let lang = optLang pgf opts
-         let grphs = map (graphvizDependencyTree outp debug mlab mclab pgf lang) es
+         concr <- optLang pgf opts
+         let grphs = map (graphvizDependencyTree outp debug mlab mclab concr) es
          if isOpt "conll2latex" opts
            then return $ fromString $ conlls2latexDoc $ stanzas $ unlines $ toStrings arg
            else if isFlag "view" opts && valStrOpts "output" "" opts == "latex"
@@ -596,7 +537,6 @@ pgfCommands = Map.fromList [
        ]
     }),
 
-
   ("vp", emptyCommandInfo {
      longname = "visualize_parse",
      synopsis = "show parse tree graphically",
@@ -608,9 +548,8 @@ pgfCommands = Map.fromList [
        "by the view flag. The target format is png, unless overridden by the",
        "flag -format. Results from multiple trees are combined to pdf with convert (ImageMagick)."
        ],
-     exec = getEnv $ \ opts arg (Env pgf mos) -> do
-         let es = toExprs arg
-         let lang = optLang pgf opts
+     exec = needPGF $ \opts arg pgf -> do
+         let es        = toExprs arg
          let gvOptions = GraphvizOptions {noLeaves = isOpt "noleaves" opts && not (isOpt "showleaves" opts),
                                           noFun = isOpt "nofun" opts || not (isOpt "showfun" opts),
                                           noCat = isOpt "nocat" opts && not (isOpt "showcat" opts),
@@ -623,10 +562,11 @@ pgfCommands = Map.fromList [
                                           leafEdgeStyle = valStrOpts "leafedgestyle" "dashed" opts
                                          }
          let depfile = valStrOpts "file" "" opts
+         concr  <- optLang pgf opts
          mlab <- case depfile of
            "" -> return Nothing
            _  -> (Just . getDepLabels) `fmap` restricted (readFile depfile)
-         let grphs = map (graphvizParseTreeDep mlab pgf lang gvOptions) es
+         let grphs = map (graphvizDependencyTree "dot" False mlab Nothing concr) es
          if isFlag "view" opts || isFlag "format" opts
            then do
              let view = optViewGraph opts
@@ -661,7 +601,6 @@ pgfCommands = Map.fromList [
        ]
     }),
 
-
   ("vt", emptyCommandInfo {
      longname = "visualize_tree",
      synopsis = "show a set of trees graphically",
@@ -674,7 +613,7 @@ pgfCommands = Map.fromList [
        "flag -format. Results from multiple trees are combined to pdf with convert (ImageMagick).",
        "With option -mk, use for showing library style function names of form 'mkC'."
        ],
-     exec = getEnv $ \ opts arg (Env pgf mos) ->
+     exec = needPGF $ \opts arg pgf ->
       let es = toExprs arg in
        if isOpt "mk" opts
        then return $ fromString $ unlines $ map (tree2mk pgf) es
@@ -686,7 +625,7 @@ pgfCommands = Map.fromList [
        else do
          let funs = not (isOpt "nofun" opts)
          let cats = not (isOpt "nocat" opts)
-         let grphs = map (graphvizAbstractTree pgf (funs,cats)) es
+         let grphs = map (graphvizAbstractTree pgf (graphvizDefaults{noFun=funs,noCat=cats})) es
          if isFlag "view" opts || isFlag "format" opts
            then do
              let view = optViewGraph opts
@@ -708,6 +647,7 @@ pgfCommands = Map.fromList [
        ("view","program to open the resulting file (default \"open\")")
        ]
      }),
+
   ("ai", emptyCommandInfo {
      longname = "abstract_info",
      syntax = "ai IDENTIFIER  or  ai EXPR",
@@ -720,205 +660,150 @@ pgfCommands = Map.fromList [
        "If a whole expression is given it prints the expression with refined",
        "metavariables and the type of the expression."
        ],
-     exec = getEnv $ \ opts arg (Env pgf mos) -> do
+     exec = needPGF $ \opts arg pgf -> do
        case toExprs arg of
-         [EFun id] -> case Map.lookup id (funs (abstract pgf)) of
-                        Just fd -> do putStrLn $ render (ppFun id fd)
-                                      let (_,_,_,prob) = fd
-                                      putStrLn ("Probability: "++show prob)
-                                      return void
-                        Nothing -> case Map.lookup id (cats (abstract pgf)) of
-                                     Just cd   -> do putStrLn $
-                                                        render (ppCat id cd $$
-                                                                if null (functionsToCat pgf id)
-                                                                  then empty
-                                                                  else ' ' $$
-                                                                       vcat [ppFun fid (ty,0,Just ([],[]),0) | (fid,ty) <- functionsToCat pgf id] $$
-                                                                       ' ')
-                                                     let (_,_,prob) = cd
-                                                     putStrLn ("Probability: "++show prob)
-                                                     return void
-                                     Nothing   -> do putStrLn ("unknown category of function identifier "++show id)
-                                                     return void
-         [e]         -> case inferExpr pgf e of
-                          Left tcErr   -> error $ render (ppTcError tcErr)
-                          Right (e,ty) -> do putStrLn ("Expression:  "++showExpr [] e)
-                                             putStrLn ("Type:        "++showType [] ty)
-                                             putStrLn ("Probability: "++show (probTree pgf e))
-                                             return void
+         [e] -> case unApp e of
+                  Just (id, []) -> case functionType pgf id of
+                                     Just ty -> do putStrLn (showFun pgf id ty)
+                                                   putStrLn ("Probability: "++show (exprProbability pgf e))
+                                                   return void
+                                     Nothing -> case categoryContext pgf id of
+                                                  Just hypos -> do putStrLn ("cat "++id++if null hypos then "" else ' ':showContext [] hypos)
+                                                                   let ls = [showFun pgf fn ty | fn <- functionsByCat pgf id, Just ty <- [functionType pgf fn]]
+                                                                   if null ls
+                                                                     then return ()
+                                                                     else putStrLn (unlines ("":ls))
+                                                                   putStrLn ("Probability: "++show (categoryProbability pgf id))
+                                                                   return void
+                                                  Nothing    -> do putStrLn ("unknown category of function identifier "++show id)
+                                                                   return void
+                  _             -> case inferExpr pgf e of
+                                     Left err     -> error err
+                                     Right (e,ty) -> do putStrLn ("Expression:  "++showExpr [] e)
+                                                        putStrLn ("Type:        "++showType [] ty)
+                                                        putStrLn ("Probability: "++show (exprProbability pgf e))
+                                                        return void
          _           -> do putStrLn "a single identifier or expression is expected from the command"
                            return void,
      needsTypeCheck = False
      })
   ]
  where
-   getEnv exec opts ts = liftSIO . exec opts ts =<< getPGFEnv
-
-   par pgf opts s = case optOpenTypes opts of
-                  []        -> [parse_ pgf lang (optType pgf opts) (Just dp) s | lang <- optLangs pgf opts]
-                  open_typs -> [parseWithRecovery pgf lang (optType pgf opts) open_typs (Just dp) s | lang <- optLangs pgf opts]
-     where
-       dp = valIntOpts "depth" 4 opts
-
-   fromParse opts = foldr (joinPiped . fromParse1 opts) void
+   needPGF exec opts ts = do
+     mb_pgf <- getPGF
+     case mb_pgf of
+       Just pgf -> liftSIO $ exec opts ts pgf
+       _ -> fail "Import a grammar before using this command"
 
    joinPiped (Piped (es1,ms1)) (Piped (es2,ms2)) = Piped (jA es1 es2,ms1+++-ms2)
      where
        jA (Exprs es1) (Exprs es2) = Exprs (es1++es2)
-       -- ^ fromParse1 always output Exprs
 
-   fromParse1 opts (s,(po,bs))
-     | isOpt "bracket" opts = pipeMessage (showBracketedString bs)
-     | otherwise            =
-         case po of
-           ParseOk ts      -> fromExprs ts
-           ParseFailed i   -> pipeMessage $ "The parser failed at token "
-                                             ++ show i ++": "
-                                             ++ show (words s !! max 0 (i-1))
-                                          -- ++ " in " ++ show s
-           ParseIncomplete -> pipeMessage "The sentence is not complete"
-           TypeError errs  ->
-             pipeMessage . render $
-               "The parsing is successful but the type checking failed with error(s):"
-               $$ nest 2 (vcat (map (ppTcError . snd) errs))
+   fromParse1 opts (s,po) =
+     case po of
+       ParseOk ts      -> fromExprs (isOpt "show_probs" opts) (takeOptNum opts ts)
+       ParseFailed i t -> pipeMessage $ "The parser failed at token "
+                                        ++ show i ++": "
+                                        ++ show t
+       ParseIncomplete -> pipeMessage "The sentence is not complete"
 
-   optLins pgf opts ts = case opts of
-     _ | isOpt "groups" opts ->
-       concatMap snd $ groupResults
-         [[(lang, s) | lang <- optLangs pgf opts,s <- linear pgf opts lang t] | t <- ts]
-     _ -> concatMap (optLin pgf opts) ts
+   optLins pgf opts ts = concatMap (optLin pgf opts) ts
    optLin pgf opts t =
      case opts of
-       _ | isOpt "treebank" opts && isOpt "chunks" opts ->
-         (showCId (abstractName pgf) ++ ": " ++ showExpr [] t) :
-         [showCId lang ++ ": " ++ li | (lang,li) <- linChunks pgf opts t] --linear pgf opts lang t | lang <- optLangs pgf opts]
        _ | isOpt "treebank" opts ->
-         (showCId (abstractName pgf) ++ ": " ++ showExpr [] t) :
-         [showCId lang ++ ": " ++ s | lang <- optLangs pgf opts, s<-linear pgf opts lang t]
-       _ | isOpt "tabtreebank" opts ->
-         return $ concat $ intersperse "\t" $ (showExpr [] t) :
-                   [s | lang <- optLangs pgf opts, s <- linear pgf opts lang t]
-       _ | isOpt "chunks" opts -> map snd $ linChunks pgf opts t   
-       _ -> [s | lang <- optLangs pgf opts, s<-linear pgf opts lang t]
-   linChunks pgf opts t = 
-     [(lang, unwords (intersperse "<+>" (map (unlines . linear pgf opts lang) (treeChunks t)))) | lang <- optLangs pgf opts]
+         (abstractName pgf ++ ": " ++ showExpr [] t) :
+         [concreteName concr ++ ": " ++ s | concr <- optLangs pgf opts, s<-linear opts concr t]
+       _ -> [s | concr <- optLangs pgf opts, s <- linear opts concr t]
 
-   linear :: PGF -> [Option] -> CId -> Expr -> [String]
-   linear pgf opts lang = let unl = unlex opts lang in case opts of
-       _ | isOpt "all"     opts -> concat . -- intersperse [[]] .
-                                   map (map (unl . snd)) . tabularLinearizes pgf lang
+   linear :: [Option] -> Concr -> Expr -> [String]
+   linear opts concr = case opts of
+       _ | isOpt "all"     opts -> concat .
+                                   map (map snd) . tabularLinearizeAll concr
        _ | isOpt "list"    opts -> (:[]) . commaList . concat .
-                                   map (map (unl . snd)) . tabularLinearizes pgf lang
-       _ | isOpt "table"   opts -> concat . -- intersperse [[]] .
-                    map (map (\(p,v) -> p+++":"+++unl v)) . tabularLinearizes pgf lang
-       _ | isOpt "bracket" opts -> (:[]) . unwords . map showBracketedString . bracketedLinearize pgf lang
-       _                        -> (:[]) . unl . linearize pgf lang
+                                   map (map snd) . tabularLinearizeAll concr
+       _ | isOpt "table"   opts -> concat .
+                                    map (map (\(p,v) -> p+++":"+++v)) . tabularLinearizeAll concr
+       _ | isOpt "bracket" opts -> (:[]) . unwords . map showBracketedString . bracketedLinearize concr
+       _                        -> (:[]) . linearize concr
 
    -- replace each non-atomic constructor with mkC, where C is the val cat
    tree2mk pgf = showExpr [] . t2m where
      t2m t = case unApp t of
-       Just (cid,ts@(_:_)) -> mkApp (mk cid) (map t2m ts)
-       _ -> t
-     mk = mkCId . ("mk" ++) . showCId . lookValCat (abstract pgf)
-
-   unlex opts lang = stringOps Nothing (getUnlex opts lang ++ map prOpt opts) ----
-
-   getUnlex opts lang = case words (valStrOpts "unlexer" "" opts) of
-     lexs -> case lookup lang
-               [(mkCId la,tail le) | lex <- lexs, let (la,le) = span (/='=') lex, not (null le)] of
-       Just le -> chunks ',' le
-       _ -> []
+               Just (cid,ts@(_:_)) -> mkApp (mk cid) (map t2m ts)
+               _                   -> t
+     mk f = case functionType pgf f of
+              Just ty -> let (_,cat,_) = unType ty
+                         in "mk" ++ cat
+              Nothing -> f
 
    commaList [] = []
    commaList ws = concat $ head ws : map (", " ++) (tail ws)
 
--- Proposed logic of coding in unlexing:
---   - If lang has no coding flag, or -to_utf8 is not in opts, just opts are used.
---   - If lang has flag coding=utf8, -to_utf8 is ignored.
---   - If lang has coding=other, and -to_utf8 is in opts, from_other is applied first.
--- THIS DOES NOT WORK UNFORTUNATELY - can't use the grammar flag properly
-{-
-   unlexx pgf opts lang = {- trace (unwords optsC) $ -} stringOps Nothing optsC where ----
-     optsC = case lookConcrFlag pgf (mkCId lang) (mkCId "coding") of
-       Just (LStr "utf8") -> filter (/="to_utf8") $ map prOpt opts
-       Just (LStr other) | isOpt "to_utf8" opts ->
-                      let cod = ("from_" ++ other)
-                      in cod : filter (/=cod) (map prOpt opts)
-       _ -> map prOpt opts
--}
-   optRestricted opts pgf =
-     restrictPGF (\f -> and [hasLin pgf la f | la <- optLangs pgf opts]) pgf
-
    optLang  = optLangFlag "lang"
    optLangs = optLangsFlag "lang"
 
-   optLangsFlag f pgf opts = case valStrOpts f "" opts of
-     "" -> languages pgf
-     lang -> map (completeLang pgf) (chunks ',' lang)
-   completeLang pgf la = let cla = (mkCId la) in
-     if elem cla (languages pgf)
-       then cla
-       else (mkCId (showCId (abstractName pgf) ++ la))
+   optLangFlag flag pgf opts =
+     case optLangsFlag flag pgf opts of
+       []     -> fail "no language specified"
+       (l:ls) -> return l
 
-   optLangFlag f pgf opts = head $ optLangsFlag f pgf opts ++ [wildCId]
+   optLangsFlag flag pgf opts =
+     case valStrOpts flag "" opts of
+       ""  -> Map.elems langs
+       str -> mapMaybe (completeLang pgf) (chunks ',' str)
+     where
+       langs = languages pgf
 
-   optOpenTypes opts = case valStrOpts "openclass" "" opts of
-     ""   -> []
-     cats -> mapMaybe readType (chunks ',' cats)
-
-   optProbs opts pgf = case valStrOpts "probs" "" opts of
-     ""   -> return pgf
-     file -> do
-       probs <- restricted $ readProbabilitiesFromFile file pgf
-       return (setProbabilities probs pgf)
+       completeLang pgf la =
+         mplus (Map.lookup la                       langs)
+               (Map.lookup (abstractName pgf ++ la) langs)
 
    optFile opts = valStrOpts "file" "_gftmp" opts
 
    optType pgf opts =
-     let str = valStrOpts "cat" (showCId $ lookStartCat pgf) opts
-     in case readType str of
-          Just ty -> case checkType pgf ty of
-                       Left tcErr -> error $ render (ppTcError tcErr)
-                       Right ty   -> ty
-          Nothing -> error ("Can't parse '"++str++"' as a type")
+     let readOpt str = case readType str of
+                         Just ty -> case checkType pgf ty of
+                                      Left err -> error err
+                                      Right ty -> ty
+                         Nothing -> error ("Can't parse '"++str++"' as a type")
+     in maybeStrOpts "cat" (startCat pgf) readOpt opts
    optViewFormat opts = valStrOpts "format" "png" opts
    optViewGraph opts = valStrOpts "view" "open" opts
    optNum opts = valIntOpts "number" 1 opts
    optNumInf opts = valIntOpts "number" 1000000000 opts ---- 10^9
    takeOptNum opts = take (optNumInf opts)
 
-   returnFromExprs es = return $ case es of
-     [] -> pipeMessage "no trees found"
-     _  -> fromExprs es
+   returnFromExprs show_p es =
+     return $ 
+       case es of
+         [] -> pipeMessage "no trees found"
+         _  -> fromExprs show_p es
 
-   prGrammar (Env pgf mos) opts
+   prGrammar pgf opts
      | isOpt "pgf"      opts = do
-          let pgf1 = if isOpt "opt" opts then optimizePGF pgf else pgf
-          let outfile = valStrOpts "file" (showCId (abstractName pgf) ++ ".pgf") opts
-          restricted $ encodeFile outfile pgf1
+          let outfile = valStrOpts "file" (abstractName pgf ++ ".pgf") opts
+          restricted $ writePGF outfile pgf
           putStrLn $ "wrote file " ++ outfile
           return void
-     | isOpt "cats"     opts = return $ fromString $ unwords $ map showCId $ categories pgf
-     | isOpt "funs"     opts = return $ fromString $ unlines $ map showFun $ funsigs pgf
-     | isOpt "fullform" opts = return $ fromString $ concatMap (morpho mos "" prFullFormLexicon) $ optLangs pgf opts
-     | isOpt "langs"    opts = return $ fromString $ unwords $ map showCId $ languages pgf
+     | isOpt "cats"     opts = return $ fromString $ unwords $ categories pgf
+     | isOpt "funs"     opts = return $ fromString $ unlines [showFun pgf f ty | f <- functions pgf, Just ty <- [functionType pgf f]]
+     | isOpt "fullform" opts = return $ fromString $ concatMap prFullFormLexicon $ optLangs pgf opts
+     | isOpt "langs"    opts = return $ fromString $ unwords $ Map.keys $ languages pgf
 
-     | isOpt "lexc"     opts = return $ fromString $ concatMap (morpho mos "" prLexcLexicon) $ optLangs pgf opts
-     | isOpt "missing"  opts = return $ fromString $ unlines $ [unwords (showCId la:":": map showCId cs) |
-                                                                  la <- optLangs pgf opts, let cs = missingLins pgf la]
-     | isOpt "words" opts = return $ fromString $ concatMap (morpho mos "" prAllWords) $ optLangs pgf opts
+     | isOpt "lexc"     opts = return $ fromString $ concatMap prLexcLexicon $ optLangs pgf opts
+     | isOpt "missing"  opts = return $ fromString $ unlines $ [unwords (concreteName concr:":":[f | f <- functions pgf, not (hasLinearization concr f)]) |
+                                                                  concr <- optLangs pgf opts]
+     | isOpt "words" opts = return $ fromString $ concatMap prAllWords $ optLangs pgf opts
      | otherwise             = do fmt <- readOutputFormat (valStrOpts "printer" "pgf_pretty" opts)
                                   return $ fromString $ concatMap snd $ exportPGF noOptions fmt pgf
 
-   funsigs pgf = [(f,ty) | (f,(ty,_,_,_)) <- Map.assocs (funs (abstract pgf))]
-   showFun (f,ty) = showCId f ++ " : " ++ showType [] ty ++ " ;"
+   showFun pgf id ty = kwd++" "++ id ++ " : " ++ showType [] ty
+                       where
+                         kwd | functionIsConstructor pgf id = "data"
+                             | otherwise                    = "fun"
 
-   morphos (Env pgf mos) opts s =
-     [(s,morpho mos [] (\mo -> lookupMorpho mo s) la) | la <- optLangs pgf opts]
-
-   morpho mos z f la = maybe z f $ Map.lookup la mos
-
-   optMorpho (Env pgf mos) opts = morpho mos (error "no morpho") id (head (optLangs pgf opts))
+   morphos pgf opts s =
+     [(s,lookupMorpho concr s) | concr <- optLangs pgf opts]
 
    optClitics opts = case valStrOpts "clitics" "" opts of
      "" -> []
@@ -931,18 +816,28 @@ pgfCommands = Map.fromList [
    -- ps -f -g s returns g (f s)
    treeOps pgf opts s = foldr app s (reverse opts) where
      app (OOpt  op)         | Just (Left  f) <- treeOp pgf op = f
-     app (OFlag op (VId x)) | Just (Right f) <- treeOp pgf op = f (mkCId x)
+     app (OFlag op (VId x)) | Just (Right f) <- treeOp pgf op = f x
      app _                                                    = id
 
+morphoMissing  :: Concr -> [String] -> [String]
+morphoMissing = morphoClassify False
+
+morphoKnown    :: Concr -> [String] -> [String]
+morphoKnown = morphoClassify True
+
+morphoClassify :: Bool -> Concr -> [String] -> [String]
+morphoClassify k concr ws = [w | w <- ws, k /= null (lookupMorpho concr w), notLiteral w] where
+  notLiteral w = not (all isDigit w)
+
 treeOpOptions pgf = [(op,expl) | (op,(expl,Left  _)) <- allTreeOps pgf]
 treeOpFlags   pgf = [(op,expl) | (op,(expl,Right _)) <- allTreeOps pgf]
 
-translationQuiz :: Maybe Expr -> PGF -> Language -> Language -> Type -> IO ()
+translationQuiz :: Maybe Expr -> PGF -> Concr -> Concr -> Type -> IO ()
 translationQuiz mex pgf ig og typ = do
   tts <- translationList mex pgf ig og typ infinity
   mkQuiz "Welcome to GF Translation Quiz." tts
 
-morphologyQuiz :: Maybe Expr -> PGF -> Language -> Type -> IO ()
+morphologyQuiz :: Maybe Expr -> PGF -> Concr -> Type -> IO ()
 morphologyQuiz mex pgf ig typ = do
   tts <- morphologyList mex pgf ig typ infinity
   mkQuiz "Welcome to GF Morphology Quiz." tts
@@ -951,30 +846,28 @@ morphologyQuiz mex pgf ig typ = do
 infinity :: Int
 infinity = 256
 
-prLexcLexicon :: Morpho -> String
-prLexcLexicon mo =
-  unlines $ "Multichar_Symbols":multichars:"":"LEXICON Root" : [prLexc l p ++ ":" ++ w  ++ " # ;" | (w,lps) <- morpho, (l,p) <- lps] ++ ["END"]
+prLexcLexicon :: Concr -> String
+prLexcLexicon concr =
+  unlines $ "Multichar_Symbols":multichars:"":"LEXICON Root" : [prLexc l p ++ ":" ++ w  ++ " # ;" | (w,lps) <- morpho, (l,p,_) <- lps] ++ ["END"]
  where
-  morpho = fullFormLexicon mo
-  prLexc l p = showCId l ++ concat (mkTags (words p))
+  morpho = fullFormLexicon concr
+  prLexc l p = l ++ concat (mkTags (words p))
   mkTags p = case p of
     "s":ws -> mkTags ws   --- remove record field
     ws -> map ('+':) ws
 
-  multichars = unwords $ nub $ concat [mkTags (words p) | (w,lps) <- morpho, (l,p) <- lps]
-  -- thick_A+(AAdj+Posit+Gen):thick's # ;
+  multichars = unwords $ nub $ concat [mkTags (words p) | (w,lps) <- morpho, (l,p,_) <- lps]
 
-prFullFormLexicon :: Morpho -> String
-prFullFormLexicon mo =
-  unlines (map prMorphoAnalysis (fullFormLexicon mo))
+prFullFormLexicon :: Concr -> String
+prFullFormLexicon concr =
+  unlines (map prMorphoAnalysis (fullFormLexicon concr))
 
-prAllWords :: Morpho -> String
-prAllWords mo =
-  unwords [w | (w,_) <- fullFormLexicon mo]
+prAllWords :: Concr -> String
+prAllWords concr =
+  unwords [w | (w,_) <- fullFormLexicon concr]
 
-prMorphoAnalysis :: (String,[(Lemma,Analysis)]) -> String
 prMorphoAnalysis (w,lps) =
-  unlines (w:[showCId l ++ " : " ++ p | (l,p) <- lps])
+  unlines (w:[l ++ " : " ++ p ++ show prob | (l,p,prob) <- lps])
 
 viewGraphviz :: String -> String -> String -> [String] -> SIO CommandOutput
 viewGraphviz view format name grphs = do
@@ -1019,3 +912,7 @@ stanzas = map unlines . chop . lines where
   chop ls = case break (=="") ls of
     (ls1,[])  -> [ls1]
     (ls1,_:ls2) -> ls1 : chop ls2
+
+#if !(MIN_VERSION_base(4,9,0))
+errorWithoutStackTrace = error
+#endif

src/compiler/GF/Command/Commands2.hs

@@ -1,831 +0,0 @@
-{-# LANGUAGE FlexibleInstances, UndecidableInstances #-}
-module GF.Command.Commands2 (
-  PGFEnv,HasPGFEnv(..),pgf,concs,pgfEnv,emptyPGFEnv,pgfCommands,
-  options, flags,
-  ) where
-import Prelude hiding (putStrLn,(<>)) -- GHC 8.4.1 clash with Text.PrettyPrint
-
-import PGF2
-import qualified PGF as H
-import GF.Compile.ToAPI(exprToAPI)
-import GF.Infra.UseIO(writeUTF8File)
-import GF.Infra.SIO(MonadSIO,liftSIO,putStrLn,restricted,restrictedSystem)
-import GF.Command.Abstract
-import GF.Command.CommandInfo
-import GF.Data.Operations
-import Data.List(intersperse,intersect,nub,sortBy)
-import Data.Maybe
-import qualified Data.Map as Map
-import GF.Text.Pretty
-import Control.Monad(mplus)
-import qualified Control.Monad.Fail as Fail
-
-
-data PGFEnv = Env {pgf::Maybe PGF,concs::Map.Map ConcName Concr}
-
-pgfEnv pgf = Env (Just pgf) (languages pgf)
-emptyPGFEnv = Env Nothing Map.empty
-
-class (Fail.MonadFail m,MonadSIO m) => HasPGFEnv m where getPGFEnv :: m PGFEnv
-
-instance (Monad m,HasPGFEnv m) => TypeCheckArg m where
-  typeCheckArg e = do env <- getPGFEnv
-                      case pgf env of
-                        Just gr -> either fail
-                                          (return . hsExpr . fst)
-                                          (inferExpr gr (cExpr e))
-                        Nothing  -> fail "Import a grammar before using this command"
-
-pgfCommands :: HasPGFEnv m => Map.Map String (CommandInfo m)
-pgfCommands = Map.fromList [
-  ("aw", emptyCommandInfo {
-     longname = "align_words",
-     synopsis = "show word alignments between languages graphically",
-     explanation = unlines [
-       "Prints a set of strings in the .dot format (the graphviz format).",
-       "The graph can be saved in a file by the wf command as usual.",
-       "If the -view flag is defined, the graph is saved in a temporary file",
-       "which is processed by graphviz and displayed by the program indicated",
-       "by the flag. The target format is postscript, unless overridden by the",
-       "flag -format."
-       ],
-     exec = needPGF $ \opts es env -> do
-         let cncs = optConcs env opts
-         if isOpt "giza" opts
-           then if length cncs == 2
-                  then let giz = map (gizaAlignment pgf (snd (cncs !! 0)) (snd (cncs !! 1)) . cExpr) (toExprs es)
-                           lsrc = unlines $ map (\(x,_,_) -> x) giz
-                           ltrg = unlines $ map (\(_,x,_) -> x) giz
-                           align = unlines $ map (\(_,_,x) -> x) giz
-                           grph = if null (toExprs es) then [] else lsrc ++ "\n--end_source--\n\n"++ltrg++"\n-end_target--\n\n"++align
-                       in return (fromString grph)
-                  else error "For giza alignment you need exactly two languages"
-           else let gvOptions=graphvizDefaults{leafFont = valStrOpts "font" "" opts,
-                                               leafColor = valStrOpts "color" "" opts,
-                                               leafEdgeStyle = valStrOpts "edgestyle" "" opts
-                                              }
-                    grph = if null (toExprs es) then [] else graphvizWordAlignment (map snd cncs) gvOptions (cExpr (head (toExprs es)))
-                in if isFlag "view" opts || isFlag "format" opts
-                     then do let file s = "_grph." ++ s
-                             let view = optViewGraph opts
-                             let format = optViewFormat opts
-                             restricted $ writeUTF8File (file "dot") grph
-                             restrictedSystem $ "dot -T" ++ format ++ " " ++ file "dot" ++ " > " ++ file format
-                             restrictedSystem $ view ++ " " ++ file format
-                             return void
-                     else return (fromString grph),
-     examples = [
-       ("gr | aw"                         , "generate a tree and show word alignment as graph script"),
-       ("gr | aw -view=\"open\""          , "generate a tree and display alignment on Mac"),
-       ("gr | aw -view=\"eog\""           , "generate a tree and display alignment on Ubuntu"),
-       ("gt | aw -giza | wf -file=aligns" , "generate trees, send giza alignments to file")
-       ],
-     options = [
-       ("giza",  "show alignments in the Giza format; the first two languages")
-       ],
-     flags = [
-       ("format","format of the visualization file (default \"png\")"),
-       ("lang",  "alignments for this list of languages (default: all)"),
-       ("view",  "program to open the resulting file"),
-       ("font",  "font for the words"),
-       ("color", "color for the words"),
-       ("edgestyle", "the style for links between words")
-       ]
-    }),
-{-
-  ("eb", emptyCommandInfo {
-     longname = "example_based",
-     syntax = "eb (-probs=FILE | -lang=LANG)* -file=FILE.gfe",
-     synopsis = "converts .gfe files to .gf files by parsing examples to trees",
-     explanation = unlines [
-       "Reads FILE.gfe and writes FILE.gf. Each expression of form",
-       "'%ex CAT QUOTEDSTRING' in FILE.gfe is replaced by a syntax tree.",
-       "This tree is the first one returned by the parser; a biased ranking",
-       "can be used to regulate the order. If there are more than one parses",
-       "the rest are shown in comments, with probabilities if the order is biased.",
-       "The probabilities flag and configuration file is similar to the commands",
-       "gr and rt. Notice that the command doesn't change the environment,",
-       "but the resulting .gf file must be imported separately."
-       ],
-     options = [
-       ("api","convert trees to overloaded API expressions (using Syntax not Lang)")
-       ],
-     flags = [
-       ("file","the file to be converted (suffix .gfe must be given)"),
-       ("lang","the language in which to parse"),
-       ("probs","file with probabilities to rank the parses")
-       ],
-     exec = \env@(pgf, mos) opts _ -> do
-       let file = optFile opts
-       pgf <- optProbs opts pgf
-       let printer = if (isOpt "api" opts) then exprToAPI else (H.showExpr [])
-       let conf = configureExBased pgf (optMorpho env opts) (optLang pgf opts) printer
-       (file',ws) <- restricted $ parseExamplesInGrammar conf file
-       if null ws then return () else putStrLn ("unknown words: " ++ unwords ws)
-       return (fromString ("wrote " ++ file')),
-     needsTypeCheck = False
-     }),
--}
-{-
-  ("gr", emptyCommandInfo {
-     longname = "generate_random",
-     synopsis = "generate random trees in the current abstract syntax",
-     syntax = "gr [-cat=CAT] [-number=INT]",
-     examples = [
-       mkEx "gr                     -- one tree in the startcat of the current grammar",
-       mkEx "gr -cat=NP -number=16  -- 16 trees in the category NP",
-       mkEx "gr -lang=LangHin,LangTha -cat=Cl  -- Cl, both in LangHin and LangTha",
-       mkEx "gr -probs=FILE         -- generate with bias",
-       mkEx "gr (AdjCN ? (UseN ?))  -- generate trees of form (AdjCN ? (UseN ?))"
-       ],
-     explanation = unlines [
-       "Generates a list of random trees, by default one tree.",
-       "If a tree argument is given, the command completes the Tree with values to",
-       "all metavariables in the tree. The generation can be biased by probabilities,",
-       "given in a file in the -probs flag."
-       ],
-     flags = [
-       ("cat","generation category"),
-       ("lang","uses only functions that have linearizations in all these languages"),
-       ("number","number of trees generated"),
-       ("depth","the maximum generation depth"),
-       ("probs", "file with biased probabilities (format 'f 0.4' one by line)")
-       ],
-     exec = \env@(pgf, mos) opts xs -> do
-       pgf <- optProbs opts (optRestricted opts pgf)
-       gen <- newStdGen
-       let dp = valIntOpts "depth" 4 opts
-       let ts  = case mexp xs of
-                   Just ex -> H.generateRandomFromDepth gen pgf ex (Just dp)
-                   Nothing -> H.generateRandomDepth     gen pgf (optType pgf opts) (Just dp)
-       returnFromExprs $ take (optNum opts) ts
-     }),
--}
-  ("gt", emptyCommandInfo {
-     longname = "generate_trees",
-     synopsis = "generates a list of trees, by default exhaustive",
-     flags = [("cat","the generation category"),
-              ("number","the number of trees generated")],
-     examples = [
-       mkEx "gt                     -- all trees in the startcat",
-       mkEx "gt -cat=NP -number=16  -- 16 trees in the category NP"],
-     exec = needPGF $ \ opts _ env@(pgf,_) ->
-            let ts = map fst (generateAll pgf cat)
-                cat = optType pgf opts
-            in returnFromCExprs (takeOptNum opts ts),
-     needsTypeCheck = False
-     }),
-  ("i", emptyCommandInfo {
-     longname = "import",
-     synopsis = "import a grammar from a compiled .pgf file",
-     explanation = unlines [
-       "Reads a grammar from a compiled .pgf file.",
-       "Old modules are discarded.",
-{-
-       "The grammar parser depends on the file name suffix:",
-
-       "  .cf    context-free (labelled BNF) source",
-       "  .ebnf  extended BNF source",
-       "  .gfm   multi-module GF source",
-       "  .gf    normal GF source",
-       "  .gfo   compiled GF source",
--}
-       "  .pgf   precompiled grammar in Portable Grammar Format"
-       ],
-     flags = [
---     ("probs","file with biased probabilities for generation")
-       ],
-     options = [
-       -- ["gfo", "src", "no-cpu", "cpu", "quiet", "verbose"]
---     ("retain","retain operations (used for cc command)"),
---     ("src",   "force compilation from source"),
---     ("v",     "be verbose - show intermediate status information")
-       ],
-     needsTypeCheck = False
-     }),
-  ("l", emptyCommandInfo {
-     longname = "linearize",
-     synopsis = "convert an abstract syntax expression to string",
-     explanation = unlines [
-       "Shows the linearization of a Tree by the grammars in scope.",
-       "The -lang flag can be used to restrict this to fewer languages.",
-       "A sequence of string operations (see command ps) can be given",
-       "as options, and works then like a pipe to the ps command, except",
-       "that it only affect the strings, not e.g. the table labels.",
-       "These can be given separately to each language with the unlexer flag",
-       "whose results are prepended to the other lexer flags. The value of the",
-       "unlexer flag is a space-separated list of comma-separated string operation",
-       "sequences; see example."
-       ],
-     examples = [
-       mkEx "l -lang=LangSwe,LangNor no_Utt   -- linearize a tree to LangSwe and LangNor",
-       mkEx "gr -lang=LangHin -cat=Cl | l -table -to_devanagari -- hindi table",
-       mkEx "l -unlexer=\"LangAra=to_arabic LangHin=to_devanagari\" -- different unlexers"
-       ],
-     exec = needPGF $ \ opts arg env ->
-                      return . fromStrings . optLins env opts . map cExpr $ toExprs arg,
-     options = [
-       ("all",    "show all forms and variants, one by line (cf. l -list)"),
-       ("bracket","show tree structure with brackets and paths to nodes"),
-       ("groups", "all languages, grouped by lang, remove duplicate strings"),
-       ("list","show all forms and variants, comma-separated on one line (cf. l -all)"),
-       ("multi","linearize to all languages (default)"),
-       ("table","show all forms labelled by parameters"),
-       ("treebank","show the tree and tag linearizations with language names")
-       ],
-     flags = [
-       ("lang","the languages of linearization (comma-separated, no spaces)")
-       ]
-     }),
-  ("ma", emptyCommandInfo {
-     longname = "morpho_analyse",
-     synopsis = "print the morphological analyses of the (multiword) expression in the string",
-     explanation = unlines [
-       "Prints all the analyses of the (multiword) expression in the input string,",
-       "using the morphological analyser of the actual grammar (see command pg)"
-       ],
-     exec  = needPGF $ \opts args env -> 
-               return ((fromString . unlines .
-                        map prMorphoAnalysis . concatMap (morphos env opts) . toStrings) args),
-     flags = [
-       ("lang","the languages of analysis (comma-separated, no spaces)")
-       ]
-     }),
-{-
-  ("mq", emptyCommandInfo {
-     longname = "morpho_quiz",
-     synopsis = "start a morphology quiz",
-     syntax   = "mq (-cat=CAT)? (-probs=FILE)? TREE?",
-     exec = \env@(pgf, mos) opts xs -> do
-         let lang = optLang pgf opts
-         let typ  = optType pgf opts
-         pgf <- optProbs opts pgf
-         let mt = mexp xs
-         restricted $ morphologyQuiz mt pgf lang typ
-         return void,
-     flags = [
-       ("lang","language of the quiz"),
-       ("cat","category of the quiz"),
-       ("number","maximum number of questions"),
-       ("probs","file with biased probabilities for generation")
-       ]
-     }),
--}
-  ("p", emptyCommandInfo {
-     longname = "parse",
-     synopsis = "parse a string to abstract syntax expression",
-     explanation = unlines [
-       "Shows all trees returned by parsing a string in the grammars in scope.",
-       "The -lang flag can be used to restrict this to fewer languages.",
-       "The default start category can be overridden by the -cat flag.",
-       "See also the ps command for lexing and character encoding."
-       ],
-     flags = [
-       ("cat","target category of parsing"),
-       ("lang","the languages of parsing (comma-separated, no spaces)"),
-       ("number","maximum number of trees returned")
-       ],
-     examples = [
-         mkEx "p  \"this fish is fresh\" | l -lang=Swe  -- try parsing with all languages and translate the successful parses to Swedish"
-       ],
-     exec = needPGF $ \ opts ts env -> return . cParse env opts $ toStrings ts
-     }),
-  ("pg", emptyCommandInfo {
-     longname = "print_grammar",
-     synopsis = "prints different information about the grammar",
-     exec  = needPGF $ \opts _ env -> prGrammar env opts,
-     options = [
-       ("cats",   "show just the names of abstract syntax categories"),
-       ("fullform", "print the fullform lexicon"),
-       ("funs",   "show just the names and types of abstract syntax functions"),
-       ("langs",  "show just the names of top concrete syntax modules"),
-       ("lexc", "print the lexicon in Xerox LEXC format"),
-       ("missing","show just the names of functions that have no linearization"),
-       ("words", "print the list of words")
-       ],
-     flags = [
-       ("lang","the languages that need to be printed")
-       ],
-     examples = [
-       mkEx "pg -langs -- show the names of top concrete syntax modules",
-       mkEx "pg -funs | ? grep \" S ;\"  -- show functions with value cat S"
-       ]
-     }),
-
-{-
-  ("pt", emptyCommandInfo {
-     longname = "put_tree",
-     syntax = "pt OPT? TREE",
-     synopsis = "return a tree, possibly processed with a function",
-     explanation = unlines [
-       "Returns a tree obtained from its argument tree by applying",
-       "tree processing functions in the order given in the command line",
-       "option list. Thus 'pt -f -g s' returns g (f s). Typical tree processors",
-       "are type checking and semantic computation."
-       ],
-     examples = [
-       mkEx "pt -compute (plus one two)                               -- compute value",
-       mkEx "p \"4 dogs love 5 cats\" | pt -transfer=digits2numeral | l -- four...five..."
-       ],
-     exec = \env@(pgf, mos) opts ->
-            returnFromExprs . takeOptNum opts . treeOps pgf opts,
-     options = treeOpOptions undefined{-pgf-},
-     flags = [("number","take at most this many trees")] ++ treeOpFlags undefined{-pgf-}
-     }),
--}
-  ("rf",  emptyCommandInfo {
-     longname = "read_file",
-     synopsis = "read string or tree input from a file",
-     explanation = unlines [
-       "Reads input from file. The filename must be in double quotes.",
-       "The input is interpreted as a string by default, and can hence be",
-       "piped e.g. to the parse command. The option -tree interprets the",
-       "input as a tree, which can be given e.g. to the linearize command.",
-       "The option -lines will result in a list of strings or trees, one by line."
-       ],
-     options = [
-       ("lines","return the list of lines, instead of the singleton of all contents"),
-       ("tree","convert strings into trees")
-       ],
-     exec = needPGF $ \opts _ env@(pgf, mos) -> do
-       let file = optFile opts
-       let exprs []         = ([],empty)
-           exprs ((n,s):ls) | null s
-                            = exprs ls
-           exprs ((n,s):ls) = case readExpr s of
-                                Just e  -> let (es,err) = exprs ls
-                                           in case inferExpr pgf e of
-                                                Right (e,t) -> (e:es,err)
-                                                Left msg    -> (es,"on line" <+> n <> ':' $$ msg $$ err)
-                                Nothing -> let (es,err) = exprs ls
-                                           in (es,"on line" <+> n <> ':' <+> "parse error" $$ err)
-           returnFromLines ls = case exprs ls of
-                                  (es, err) | null es   -> return $ pipeMessage $ render (err $$ "no trees found")
-                                            | otherwise -> return $ pipeWithMessage (map hsExpr es) (render err)
-
-       s <- restricted $ readFile file
-       case opts of
-         _ | isOpt "lines" opts && isOpt "tree" opts ->
-               returnFromLines (zip [1::Int ..] (lines s))
-         _ | isOpt "tree" opts ->
-               returnFromLines [(1::Int,s)]
-         _ | isOpt "lines" opts -> return (fromStrings $ lines s)
-         _ -> return (fromString s),
-     flags = [("file","the input file name")]
-     }),
-  ("rt", emptyCommandInfo {
-     longname = "rank_trees",
-     synopsis = "show trees in an order of decreasing probability",
-     explanation = unlines [
-       "Order trees from the most to the least probable, using either",
-       "even distribution in each category (default) or biased as specified",
-       "by the file given by flag -probs=FILE, where each line has the form",
-       "'function probability', e.g. 'youPol_Pron  0.01'."
-       ],
-     exec = needPGF $ \opts es env@(pgf, _) -> do
-         let tds = sortBy (\(_,p) (_,q) -> compare p q)
-                          [(t, treeProbability pgf t) | t <- map cExpr (toExprs es)]
-         if isOpt "v" opts
-           then putStrLn $
-                  unlines [PGF2.showExpr []  t ++ "\t--" ++ show d | (t,d) <- tds]
-           else return ()
-         returnFromExprs $ map (hsExpr . fst) tds,
-     flags = [
-       ("probs","probabilities from this file (format 'f 0.6' per line)")
-       ],
-     options = [
-       ("v","show all trees with their probability scores")
-       ],
-     examples = [
-      mkEx "p \"you are here\" | rt -probs=probs | pt -number=1 -- most probable result"
-      ]
-     }),
-{-
-  ("tq", emptyCommandInfo {
-     longname = "translation_quiz",
-     syntax   = "tq -from=LANG -to=LANG (-cat=CAT)? (-probs=FILE)? TREE?",
-     synopsis = "start a translation quiz",
-     exec = \env@(pgf, mos) opts xs -> do
-         let from = optLangFlag "from" pgf opts
-         let to   = optLangFlag "to" pgf opts
-         let typ  = optType pgf opts
-         let mt   = mexp xs
-         pgf <- optProbs opts pgf
-         restricted $ translationQuiz mt pgf from to typ
-         return void,
-     flags = [
-       ("from","translate from this language"),
-       ("to","translate to this language"),
-       ("cat","translate in this category"),
-       ("number","the maximum number of questions"),
-       ("probs","file with biased probabilities for generation")
-       ],
-     examples = [
-       mkEx ("tq -from=Eng -to=Swe                               -- any trees in startcat"),
-       mkEx ("tq -from=Eng -to=Swe (AdjCN (PositA ?2) (UseN ?))  -- only trees of this form")
-       ]
-     }),
-  ("vd", emptyCommandInfo {
-     longname = "visualize_dependency",
-     synopsis = "show word dependency tree graphically",
-     explanation = unlines [
-       "Prints a dependency tree in the .dot format (the graphviz format, default)",
-       "or the CoNLL/MaltParser format (flag -output=conll for training, malt_input",
-       "for unanalysed input).",
-       "By default, the last argument is the head of every abstract syntax",
-       "function; moreover, the head depends on the head of the function above.",
-       "The graph can be saved in a file by the wf command as usual.",
-       "If the -view flag is defined, the graph is saved in a temporary file",
-       "which is processed by graphviz and displayed by the program indicated",
-       "by the flag. The target format is png, unless overridden by the",
-       "flag -format."
-       ],
-     exec = \env@(pgf, mos) opts es -> do
-         let debug = isOpt "v" opts
-         let file = valStrOpts "file" "" opts
-         let outp = valStrOpts "output" "dot" opts
-         mlab <- case file of
-           "" -> return Nothing
-           _  -> (Just . H.getDepLabels . lines) `fmap` restricted (readFile file)
-         let lang = optLang pgf opts
-         let grphs = unlines $ map (H.graphvizDependencyTree outp debug mlab Nothing pgf lang) es
-         if isFlag "view" opts || isFlag "format" opts then do
-           let file s = "_grphd." ++ s
-           let view = optViewGraph opts
-           let format = optViewFormat opts
-           restricted $ writeUTF8File (file "dot") grphs
-           restrictedSystem $ "dot -T" ++ format ++ " " ++ file "dot" ++ " > " ++ file format
-           restrictedSystem $ view ++ " " ++ file format
-           return void
-          else return $ fromString grphs,
-     examples = [
-       mkEx "gr | vd              -- generate a tree and show dependency tree in .dot",
-       mkEx "gr | vd -view=open   -- generate a tree and display dependency tree on a Mac",
-       mkEx "gr -number=1000 | vd -file=dep.labels -output=malt      -- generate training treebank",
-       mkEx "gr -number=100 | vd -file=dep.labels -output=malt_input -- generate test sentences"
-       ],
-     options = [
-       ("v","show extra information")
-       ],
-     flags = [
-       ("file","configuration file for labels per fun, format 'fun l1 ... label ... l2'"),
-       ("format","format of the visualization file (default \"png\")"),
-       ("output","output format of graph source (default \"dot\")"),
-       ("view","program to open the resulting file (default \"open\")"),
-       ("lang","the language of analysis")
-       ]
-    }),
--}
-
-  ("vp", emptyCommandInfo {
-     longname = "visualize_parse",
-     synopsis = "show parse tree graphically",
-     explanation = unlines [
-       "Prints a parse tree in the .dot format (the graphviz format).",
-       "The graph can be saved in a file by the wf command as usual.",
-       "If the -view flag is defined, the graph is saved in a temporary file",
-       "which is processed by graphviz and displayed by the program indicated",
-       "by the flag. The target format is png, unless overridden by the",
-       "flag -format."
-       ],
-     exec = needPGF $ \opts arg env@(pgf, concs) ->
-      do let es = toExprs arg
-         let concs = optConcs env opts
-
-         let gvOptions=graphvizDefaults{noLeaves = isOpt "noleaves" opts && not (isOpt "showleaves" opts),
-                                        noFun = isOpt "nofun" opts || not (isOpt "showfun" opts),
-                                        noCat = isOpt "nocat" opts && not (isOpt "showcat" opts),
-                                        nodeFont = valStrOpts "nodefont" "" opts,
-                                        leafFont = valStrOpts "leaffont" "" opts,
-                                        nodeColor = valStrOpts "nodecolor" "" opts,
-                                        leafColor = valStrOpts "leafcolor" "" opts,
-                                        nodeEdgeStyle = valStrOpts "nodeedgestyle" "solid" opts,
-                                        leafEdgeStyle = valStrOpts "leafedgestyle" "dashed" opts
-                                       }
-
-         let grph= if null es || null concs
-                   then []
-                   else graphvizParseTree (snd (head concs)) gvOptions (cExpr (head es))
-         if isFlag "view" opts || isFlag "format" opts then do
-           let file s = "_grph." ++ s
-           let view = optViewGraph opts
-           let format = optViewFormat opts
-           restricted $ writeUTF8File (file "dot") grph
-           restrictedSystem $ "dot -T" ++ format ++ " " ++ file "dot" ++ " > " ++ file format
-           restrictedSystem $ view ++ " " ++ file format
-           return void
-          else return $ fromString grph,
-     examples = [
-       mkEx "p -lang=Eng \"John walks\" | vp  -- generate a tree and show parse tree as .dot script",
-       mkEx "gr | vp -view=\"open\" -- generate a tree and display parse tree on a Mac"
-       ],
-     options = [
-       ("showcat","show categories in the tree nodes (default)"),
-       ("nocat","don't show categories"),
-       ("showfun","show function names in the tree nodes"),
-       ("nofun","don't show function names (default)"),
-       ("showleaves","show the leaves of the tree (default)"),
-       ("noleaves","don't show the leaves of the tree (i.e., only the abstract tree)")
-       ],
-     flags = [
-       ("lang","the language to visualize"),
-       ("format","format of the visualization file (default \"png\")"),
-       ("view","program to open the resulting file (default \"open\")"),
-       ("nodefont","font for tree nodes (default: Times -- graphviz standard font)"),
-       ("leaffont","font for tree leaves (default: nodefont)"),
-       ("nodecolor","color for tree nodes (default: black -- graphviz standard color)"),
-       ("leafcolor","color for tree leaves (default: nodecolor)"),
-       ("nodeedgestyle","edge style between tree nodes (solid/dashed/dotted/bold, default: solid)"),
-       ("leafedgestyle","edge style for links to leaves (solid/dashed/dotted/bold, default: dashed)")
-       ]
-    }),
-
-  ("vt", emptyCommandInfo {
-     longname = "visualize_tree",
-     synopsis = "show a set of trees graphically",
-     explanation = unlines [
-       "Prints a set of trees in the .dot format (the graphviz format).",
-       "The graph can be saved in a file by the wf command as usual.",
-       "If the -view flag is defined, the graph is saved in a temporary file",
-       "which is processed by graphviz and displayed by the program indicated",
-       "by the flag. The target format is postscript, unless overridden by the",
-       "flag -format."
-       ],
-     exec = needPGF $ \opts arg env@(pgf, _) ->
-       let es = toExprs arg in
-       if isOpt "api" opts
-       then do
-         mapM_ (putStrLn . exprToAPI) es
-         return void
-       else do
-         let gvOptions=graphvizDefaults{noFun = isOpt "nofun" opts,
-                                        noCat = isOpt "nocat" opts,
-                                        nodeFont = valStrOpts "nodefont" "" opts,
-                                        nodeColor = valStrOpts "nodecolor" "" opts,
-                                        nodeEdgeStyle = valStrOpts "nodeedgestyle" "solid" opts
-                                       }
-         let grph = unlines (map (graphvizAbstractTree pgf gvOptions . cExpr) es)
-         if isFlag "view" opts || isFlag "format" opts then do
-           let file s = "_grph." ++ s
-           let view = optViewGraph opts
-           let format = optViewFormat opts
-           restricted $ writeUTF8File (file "dot") grph
-           restrictedSystem $ "dot -T" ++ format ++ " " ++ file "dot" ++ " > " ++ file format
-           restrictedSystem $ view  ++ " " ++ file format
-           return void
-          else return $ fromString grph,
-     examples = [
-       mkEx "p \"hello\" | vt              -- parse a string and show trees as graph script",
-       mkEx "p \"hello\" | vt -view=\"open\" -- parse a string and display trees on a Mac"
-       ],
-     options = [
-       ("api", "show the tree with function names converted to 'mkC' with value cats C"),
-       ("nofun","don't show functions but only categories"),
-       ("nocat","don't show categories but only functions")
-       ],
-     flags = [
-       ("format","format of the visualization file (default \"png\")"),
-       ("view","program to open the resulting file (default \"open\")"),
-       ("nodefont","font for tree nodes (default: Times -- graphviz standard font)"),
-       ("nodecolor","color for tree nodes (default: black -- graphviz standard color)"),
-       ("nodeedgestyle","edge style between tree nodes (solid/dashed/dotted/bold, default: solid)")
-       ]
-     }),
-
-  ("ai", emptyCommandInfo {
-     longname = "abstract_info",
-     syntax = "ai IDENTIFIER  or  ai EXPR",
-     synopsis = "Provides an information about a function, an expression or a category from the abstract syntax",
-     explanation = unlines [
-       "The command has one argument which is either function, expression or",
-       "a category defined in the abstract syntax of the current grammar. ",
-       "If the argument is a function then its type is printed out.",
-       "If it is a category then the category definition is printed.",
-       "If a whole expression is given it prints the expression with refined",
-       "metavariables and the type of the expression."
-       ],
-     exec = needPGF $ \opts args env@(pgf,cncs) ->
-       case map cExpr (toExprs args) of
-         [e] -> case unApp e of
-                  Just (id,[]) -> return (fromString 
-                                            (case functionType pgf id of
-                                               Just ty -> showFun id ty
-                                               Nothing -> let funs = functionsByCat pgf id
-                                                          in showCat id funs))
-                                  where
-                                    showCat c funs = "cat "++c++
-                                                     " ;\n\n"++
-                                                     unlines [showFun f ty| f<-funs,
-                                                                            Just ty <- [functionType pgf f]]
-                                    showFun f ty = "fun "++f++" : "++showType [] ty++" ;"
-                  _  -> case inferExpr pgf e of
-                          Left msg     -> error msg
-                          Right (e,ty) -> do putStrLn ("Expression:  "++PGF2.showExpr [] e)
-                                             putStrLn ("Type:        "++PGF2.showType [] ty)
-                                             putStrLn ("Probability: "++show (treeProbability pgf e))
-                                             return void
-         _           -> do putStrLn "a single function name or category name is expected"
-                           return void,
-     needsTypeCheck = False
-     })
-  ]
- where
-   cParse env@(pgf,_) opts ss = 
-        parsed [ parse cnc cat s | s<-ss,(lang,cnc)<-cncs]
-     where
-       cat = optType pgf opts
-       cncs = optConcs env opts
-       parsed rs = Piped (Exprs ts,unlines msgs)
-          where
-            ts = [hsExpr t|ParseOk ts<-rs,(t,p)<-takeOptNum opts ts]
-            msgs = concatMap mkMsg rs
-
-            mkMsg (ParseOk ts) = (map (PGF2.showExpr [] . fst).takeOptNum opts) ts
-            mkMsg (ParseFailed _ tok) = ["Parse failed: "++tok]
-            mkMsg (ParseIncomplete)   = ["The sentence is incomplete"]
-
-   optLins env opts ts = case opts of
-     _ | isOpt "groups" opts ->
-       concatMap snd $ groupResults
-         [[(lang, s) | (lang,concr) <- optConcs env opts,s <- linear opts lang concr t] | t <- ts]
-     _ -> concatMap (optLin env opts) ts
-   optLin env@(pgf,_) opts t =
-     case opts of
-       _ | isOpt "treebank" opts ->
-         (abstractName pgf ++ ": " ++ PGF2.showExpr [] t) :
-         [lang ++ ": " ++ s | (lang,concr) <- optConcs env opts, s<-linear opts lang concr t]
-       _ -> [s | (lang,concr) <- optConcs env opts, s<-linear opts lang concr t]
-
-   linear :: [Option] -> ConcName -> Concr -> PGF2.Expr -> [String]
-   linear opts lang concr = case opts of
-       _ | isOpt "all"     opts -> concat . map (map snd) . tabularLinearizeAll concr
-       _ | isOpt "list"    opts -> (:[]) . commaList .
-                                   concatMap (map snd) . tabularLinearizeAll concr
-       _ | isOpt "table"   opts -> concatMap (map (\(p,v) -> p+++":"+++v)) . tabularLinearizeAll concr
-       _ | isOpt "bracket" opts -> (:[]) . unwords . map showBracketedString . bracketedLinearize concr
-       _                        -> (:[]) . linearize concr
-
-   groupResults :: [[(ConcName,String)]] -> [(ConcName,[String])]
-   groupResults = Map.toList . foldr more Map.empty . start . concat
-     where
-       start ls = [(l,[s]) | (l,s) <- ls]
-       more (l,s) = 
-          Map.insertWith (\ [x] xs -> if elem x xs then xs else (x : xs)) l s
-
-   optConcs = optConcsFlag "lang"
-
-   optConcsFlag f (pgf,cncs) opts =
-       case valStrOpts f "" opts of
-         "" -> Map.toList cncs
-         lang -> mapMaybe pickLang (chunks ',' lang)
-     where
-       pickLang l = pick l `mplus` pick fl
-         where
-           fl = abstractName pgf++l
-           pick l = (,) l `fmap` Map.lookup l cncs
-
-{-
-   -- replace each non-atomic constructor with mkC, where C is the val cat
-   tree2mk pgf = H.showExpr [] . t2m where
-     t2m t = case H.unApp t of
-       Just (cid,ts@(_:_)) -> H.mkApp (mk cid) (map t2m ts)
-       _ -> t
-     mk = H.mkCId . ("mk" ++) . H.showCId . H.lookValCat (H.abstract pgf)
-
-   unlex opts lang = stringOps Nothing (getUnlex opts lang ++ map prOpt opts) ----
-
-   getUnlex opts lang = case words (valStrOpts "unlexer" "" opts) of
-     lexs -> case lookup lang
-               [(H.mkCId la,tail le) | lex <- lexs, let (la,le) = span (/='=') lex, not (null le)] of
-       Just le -> chunks ',' le
-       _ -> []
--}
-   commaList [] = []
-   commaList ws = concat $ head ws : map (", " ++) (tail ws)
-
-   optFile opts = valStrOpts "file" "_gftmp" opts
-
-   optType pgf opts =
-     case listFlags "cat" opts of
-       v:_ -> let str = valueString v
-              in case readType str of
-                   Just ty -> case checkType pgf ty of
-                                Left msg -> error msg
-                                Right ty -> ty
-                   Nothing -> error ("Can't parse '"++str++"' as a type")
-       _   -> startCat pgf
-
-   optViewFormat opts = valStrOpts "format" "png" opts
-   optViewGraph opts = valStrOpts "view" "open" opts
-{-
-   optNum opts = valIntOpts "number" 1 opts
--}
-   optNumInf opts = valIntOpts "number" 1000000000 opts ---- 10^9
-   takeOptNum opts = take (optNumInf opts)
-
-   returnFromCExprs = returnFromExprs . map hsExpr
-   returnFromExprs es =
-      return $ case es of
-                 [] -> pipeMessage "no trees found"
-                 _  -> fromExprs es
-
-   prGrammar env@(pgf,cncs) opts
-     | isOpt "langs" opts = return . fromString . unwords $ (map fst (optConcs env opts))
-     | isOpt "cats" opts = return . fromString . unwords $ categories pgf
-     | isOpt "funs" opts = return . fromString . unwords $ functions pgf
-     | isOpt "missing" opts = return . fromString . unwords $
-                                 [f | f <- functions pgf, not (and [hasLinearization concr f | (_,concr) <- optConcs env opts])]
-     | isOpt "fullform" opts = return $ fromString $ concatMap (prFullFormLexicon . snd) $ optConcs env opts
-     | isOpt "words"    opts = return $ fromString $ concatMap (prAllWords . snd) $ optConcs env opts
-     | isOpt "lexc"     opts = return $ fromString $ concatMap (prLexcLexicon . snd) $ optConcs env opts
-     | otherwise = return void
-
-   gizaAlignment pgf src_cnc tgt_cnc e =
-     let src_res   = alignWords src_cnc e
-         tgt_res   = alignWords tgt_cnc e
-         alignment = [show i++"-"++show j | (i,(_,src_fids)) <- zip [0..] src_res, (j,(_,tgt_fids)) <- zip [0..] tgt_res, not (null (intersect src_fids tgt_fids))]
-     in (unwords (map fst src_res), unwords (map fst tgt_res), unwords alignment)
-
-   morphos env opts s =
-     [(s,res) | (lang,concr) <- optConcs env opts, let res = lookupMorpho concr s, not (null res)]
-{-
-   mexp xs = case xs of
-     t:_ -> Just t
-     _   -> Nothing
--}
-   -- ps -f -g s returns g (f s)
-{-
-   treeOps pgf opts s = foldr app s (reverse opts) where
-     app (OOpt  op)         | Just (Left  f) <- treeOp pgf op = f
-     app (OFlag op (VId x)) | Just (Right f) <- treeOp pgf op = f (H.mkCId x)
-     app _                                                    = id
-
-treeOpOptions pgf = [(op,expl) | (op,(expl,Left  _)) <- allTreeOps pgf]
-treeOpFlags   pgf = [(op,expl) | (op,(expl,Right _)) <- allTreeOps pgf]
-
-translationQuiz :: Maybe H.Expr -> H.PGF -> H.Language -> H.Language -> H.Type -> IO ()
-translationQuiz mex pgf ig og typ = do
-  tts <- translationList mex pgf ig og typ infinity
-  mkQuiz "Welcome to GF Translation Quiz." tts
-
-morphologyQuiz :: Maybe H.Expr -> H.PGF -> H.Language -> H.Type -> IO ()
-morphologyQuiz mex pgf ig typ = do
-  tts <- morphologyList mex pgf ig typ infinity
-  mkQuiz "Welcome to GF Morphology Quiz." tts
-
--- | the maximal number of precompiled quiz problems
-infinity :: Int
-infinity = 256
--}
-prLexcLexicon :: Concr -> String
-prLexcLexicon concr =
-  unlines $ "Multichar_Symbols":multichars:"":"LEXICON Root" : [prLexc l p ++ ":" ++ w  ++ " # ;" | (w,lps) <- morpho, (l,p,_) <- lps] ++ ["END"]
- where
-  morpho = fullFormLexicon concr
-  prLexc l p = l ++ concat (mkTags (words p))
-  mkTags p = case p of
-    "s":ws -> mkTags ws   --- remove record field
-    ws -> map ('+':) ws
-
-  multichars = unwords $ nub $ concat [mkTags (words p) | (w,lps) <- morpho, (l,p,_) <- lps]
-  -- thick_A+(AAdj+Posit+Gen):thick's # ;
-
-prFullFormLexicon :: Concr -> String
-prFullFormLexicon concr =
-  unlines (map prMorphoAnalysis (fullFormLexicon concr))
-
-prAllWords :: Concr -> String
-prAllWords concr =
-  unwords [w | (w,_) <- fullFormLexicon concr]
-
-prMorphoAnalysis :: (String,[MorphoAnalysis]) -> String
-prMorphoAnalysis (w,lps) =
-  unlines (w:[fun ++ " : " ++ cat | (fun,cat,p) <- lps])
-
-hsExpr c =
-  case unApp c of
-    Just (f,cs) -> H.mkApp (H.mkCId f) (map hsExpr cs)
-    _ -> case unStr c of
-           Just str -> H.mkStr str
-           _ -> case unInt c of
-                  Just n -> H.mkInt n
-                  _ -> case unFloat c of
-                         Just d -> H.mkFloat d
-                         _ -> error $ "GF.Command.Commands2.hsExpr "++show c
-
-cExpr e =
-  case H.unApp e of
-    Just (f,es) -> mkApp (H.showCId f) (map cExpr es)
-    _ -> case H.unStr e of
-           Just str -> mkStr str
-           _ -> case H.unInt e of
-                  Just n -> mkInt n
-                  _ -> case H.unFloat e of
-                         Just d -> mkFloat d
-                         _ -> error $ "GF.Command.Commands2.cExpr "++show e
-
-needPGF exec opts ts =
-  do Env mb_pgf cncs <- getPGFEnv
-     case mb_pgf of
-       Just pgf -> liftSIO $ exec opts ts (pgf,cncs)
-       _ -> fail "Import a grammar before using this command"

src/compiler/GF/Command/CommonCommands.hs

@@ -3,7 +3,6 @@
 -- elsewhere
 module GF.Command.CommonCommands where
 import Data.List(sort)
-import Data.Char (isSpace)
 import GF.Command.CommandInfo
 import qualified Data.Map as Map
 import GF.Infra.SIO
@@ -15,8 +14,9 @@ import GF.Command.Abstract --(isOpt,valStrOpts,prOpt)
 import GF.Text.Pretty
 import GF.Text.Transliterations
 import GF.Text.Lexing(stringOp,opInEnv)
+import Data.Char (isSpace)
 
-import qualified PGF as H(showCId,showExpr,toATree,toTrie,Trie(..))
+import PGF2(showExpr)
 
 extend old new = Map.union (Map.fromList new) old -- Map.union is left-biased
 
@@ -102,9 +102,7 @@ commonCommands = fmap (mapCommandExec liftSIO) $ Map.fromList [
        "To see transliteration tables, use command ut."
        ],
      examples = [
---       mkEx "l (EAdd 3 4) | ps -code         -- linearize code-like output",
        mkEx "l (EAdd 3 4) | ps -unlexcode    -- linearize code-like output",
---       mkEx "ps -lexer=code | p -cat=Exp     -- parse code-like input",
        mkEx "ps -lexcode | p -cat=Exp        -- parse code-like input",
        mkEx "gr -cat=QCl | l | ps -bind      -- linearization output from LangFin",
        mkEx "ps -to_devanagari \"A-p\"         -- show Devanagari in UTF8 terminal",
@@ -117,13 +115,11 @@ commonCommands = fmap (mapCommandExec liftSIO) $ Map.fromList [
                let (os,fs) = optsAndFlags opts
                trans <- optTranslit opts
 
-               case opts of
-                 _ | isOpt "lines" opts -> return $ fromStrings $ map (trans . stringOps (envFlag fs) (map prOpt os)) $ toStrings x
-                 _ | isOpt "paragraphs" opts -> return $ fromStrings $ map (trans . stringOps (envFlag fs) (map prOpt os)) $ toParagraphs $ toStrings x
-                 _ -> return ((fromString . trans . stringOps (envFlag fs) (map prOpt os) . toString) x),
+               if isOpt "lines" opts 
+                  then return $ fromStrings $ map (trans . stringOps (envFlag fs) (map prOpt os)) $ toStrings x
+                  else return ((fromString . trans . stringOps (envFlag fs) (map prOpt os) . toString) x),
      options = [
-       ("lines","apply the operation separately to each input line, returning a list of lines"),
-       ("paragraphs","apply separately to each input paragraph (as separated by empty lines), returning a list of lines")
+       ("lines","apply the operation separately to each input line, returning a list of lines")
        ] ++
        stringOpOptions,
      flags = [
@@ -170,7 +166,8 @@ commonCommands = fmap (mapCommandExec liftSIO) $ Map.fromList [
        restrictedSystem $ syst ++ " <" ++ tmpi ++ " >" ++ tmpo
        fmap fromString $ restricted $ readFile tmpo,
        -}
-       fmap fromString . restricted . readShellProcess syst $ toString arg,
+       fmap (fromStrings . lines) . restricted . readShellProcess syst . unlines . map (dropWhile (=='\n')) $ toStrings $ arg,
+
      flags = [
        ("command","the system command applied to the argument")
        ],
@@ -178,12 +175,6 @@ commonCommands = fmap (mapCommandExec liftSIO) $ Map.fromList [
        mkEx "gt | l | ? wc  -- generate trees, linearize, and count words"
        ]
      }),
-  ("tt", emptyCommandInfo {
-     longname = "to_trie",
-     syntax = "to_trie",
-     synopsis = "combine a list of trees into a trie",
-     exec = \ _ -> return . fromString . trie . toExprs
-    }),
   ("ut", emptyCommandInfo {
      longname = "unicode_table",
      synopsis = "show a transliteration table for a unicode character set",
@@ -231,7 +222,6 @@ envFlag fs =
     _ -> Nothing
 
 stringOpOptions = sort $ [
-       ("bind","bind tokens separated by Prelude.BIND, i.e. &+"),
        ("chars","lexer that makes every non-space character a token"),
        ("from_cp1251","decode from cp1251 (Cyrillic used in Bulgarian resource)"),
        ("from_utf8","decode from utf8 (default)"),
@@ -256,27 +246,6 @@ stringOpOptions = sort $ [
         ("to_"   ++ p, "from GF " ++ n ++ " transliteration to unicode")] |
                                     (p,n) <- transliterationPrintNames]
 
-trie = render . pptss . H.toTrie . map H.toATree
-  where
-    pptss [ts] = "*"<+>nest 2 (ppts ts)
-    pptss tss  = vcat [i<+>nest 2 (ppts ts)|(i,ts)<-zip [(1::Int)..] tss]
-
-    ppts = vcat . map ppt
-
-    ppt t =
-      case t of
-        H.Oth e     -> pp (H.showExpr [] e)
-        H.Ap f [[]] -> pp (H.showCId f)
-        H.Ap f tss  -> H.showCId f $$ nest 2 (pptss tss)
-
 -- ** Converting command input
 toString  = unwords . toStrings
 toLines = unlines . toStrings
-
-toParagraphs = map (unwords . words) . toParas
-  where
-    toParas ls = case break (all isSpace) ls of
-      ([],[])   -> []
-      ([],_:ll) -> toParas ll
-      (l, [])   -> [unwords l]
-      (l, _:ll) -> unwords l : toParas ll

src/compiler/GF/Command/Importing.hs

@@ -1,7 +1,7 @@
 module GF.Command.Importing (importGrammar, importSource) where
 
-import PGF
-import PGF.Internal(optimizePGF,unionPGF,msgUnionPGF)
+import PGF2
+import PGF2.Internal(unionPGF)
 
 import GF.Compile
 import GF.Compile.Multi (readMulti)
@@ -17,14 +17,16 @@ import GF.Data.ErrM
 
 import System.FilePath
 import qualified Data.Set as Set
+import qualified Data.Map as Map
+import Control.Monad(foldM)
 
 -- import a grammar in an environment where it extends an existing grammar
-importGrammar :: PGF -> Options -> [FilePath] -> IO PGF
-importGrammar pgf0 _ [] = return pgf0
+importGrammar :: Maybe PGF -> Options -> [FilePath] -> IO (Maybe PGF)
+importGrammar pgf0 _    []    = return pgf0
 importGrammar pgf0 opts files =
   case takeExtensions (last files) of
-    ".cf"   -> importCF opts files getBNFCRules bnfc2cf
-    ".ebnf" -> importCF opts files getEBNFRules ebnf2cf
+    ".cf"   -> fmap Just $ importCF opts files getBNFCRules bnfc2cf
+    ".ebnf" -> fmap Just $ importCF opts files getEBNFRules ebnf2cf
     ".gfm"  -> do
       ascss <- mapM readMulti files
       let cs = concatMap snd ascss
@@ -36,14 +38,17 @@ importGrammar pgf0 opts files =
         Bad msg -> do putStrLn ('\n':'\n':msg)
                       return pgf0
     ".pgf" -> do
-      pgf2 <- mapM readPGF files >>= return . foldl1 unionPGF
-      ioUnionPGF pgf0 pgf2
+      mapM readPGF files >>= foldM ioUnionPGF pgf0
+    ".ngf" -> do
+      mapM readNGF files >>= foldM ioUnionPGF pgf0
     ext -> die $ "Unknown filename extension: " ++ show ext
 
-ioUnionPGF :: PGF -> PGF -> IO PGF
-ioUnionPGF one two = case msgUnionPGF one two of
-  (pgf, Just msg) -> putStrLn msg >> return pgf
-  (pgf,_)         -> return pgf
+ioUnionPGF :: Maybe PGF -> PGF -> IO (Maybe PGF)
+ioUnionPGF Nothing    two = return (Just two)
+ioUnionPGF (Just one) two =
+  case unionPGF one two of
+    Nothing         -> putStrLn "Abstract changed, previous concretes discarded." >> return (Just two)
+    Just pgf        -> return (Just pgf)
 
 importSource :: Options -> [FilePath] -> IO SourceGrammar
 importSource opts files = fmap (snd.snd) (batchCompile opts files)
@@ -56,7 +61,6 @@ importCF opts files get convert = impCF
       startCat <- case rules of
                     (Rule cat _ _ : _) -> return cat
                     _                  -> fail "empty CFG"
-      let pgf = cf2pgf (last files) (mkCFG startCat Set.empty rules)
-      probs <- maybe (return . defaultProbabilities) readProbabilitiesFromFile (flag optProbsFile opts) pgf
-      return $ setProbabilities probs 
-             $ if flag optOptimizePGF opts then optimizePGF pgf else pgf
+      probs <- maybe (return Map.empty) readProbabilitiesFromFile (flag optProbsFile opts)
+      let pgf = cf2pgf opts (last files) (mkCFG startCat Set.empty rules) probs
+      return pgf

src/compiler/GF/Command/Interpreter.hs

@@ -6,8 +6,8 @@ module GF.Command.Interpreter (
 import GF.Command.CommandInfo
 import GF.Command.Abstract
 import GF.Command.Parse
-import PGF.Internal(Expr(..))
 import GF.Infra.UseIO(putStrLnE)
+import PGF2
 
 import Control.Monad(when)
 import qualified Data.Map as Map
@@ -56,17 +56,8 @@ interpretPipe env cs = do
 -- | macro definition applications: replace ?i by (exps !! i)
 appCommand :: CommandArguments -> Command -> Command
 appCommand args c@(Command i os arg) = case arg of
-  AExpr e -> Command i os (AExpr (app e))
+  AExpr e -> Command i os (AExpr (exprSubstitute e (toExprs args)))
   _       -> c
- where
-  xs = toExprs args
-
-  app e = case e of
-    EAbs b x e -> EAbs b x (app e)
-    EApp e1 e2 -> EApp (app e1) (app e2)
-    ELit l     -> ELit l
-    EMeta i    -> xs !! i
-    EFun x     -> EFun x
 
 -- | return the trees to be sent in pipe, and the output possibly printed
 --interpret :: CommandEnv -> [Expr] -> Command -> SIO CommandOutput
@@ -113,4 +104,4 @@ getCommandTrees env needsTypeCheck a args =
     ATerm t -> return (Term t)
     ANoArg  -> return args  -- use piped
   where
-    one e = return (Exprs [e]) -- ignore piped
+    one e = return (Exprs [(e,0)]) -- ignore piped

src/compiler/GF/Command/SourceCommands.hs

@@ -8,9 +8,11 @@ import qualified Data.Map as Map
 
 import GF.Infra.SIO(MonadSIO(..),restricted)
 import GF.Infra.Option(modifyFlags,optTrace) --,noOptions
-import GF.Data.Operations (chunks,err,raise)
-import GF.Text.Pretty(render)
+import GF.Infra.Dependencies(depGraph)
+import GF.Infra.CheckM
+import GF.Text.Pretty(render,pp)
 import GF.Data.Str(sstr)
+import GF.Data.Operations (chunks,err,raise)
 
 import GF.Grammar hiding (Ident,isPrefixOf)
 import GF.Grammar.Analyse
@@ -18,10 +20,8 @@ import GF.Grammar.Parser (runP, pExp)
 import GF.Grammar.ShowTerm
 import GF.Grammar.Lookup (allOpers,allOpersTo)
 import GF.Compile.Rename(renameSourceTerm)
-import qualified GF.Compile.Compute.ConcreteNew as CN(normalForm,resourceValues)
-import GF.Compile.TypeCheck.RConcrete as TC(inferLType,ppType)
-import GF.Infra.Dependencies(depGraph)
-import GF.Infra.CheckM(runCheck)
+import GF.Compile.Compute.Concrete(normalForm)
+import GF.Compile.TypeCheck.Concrete as TC(inferLType,ppType)
 
 import GF.Command.Abstract(Option(..),isOpt,listFlags,valueString,valStrOpts)
 import GF.Command.CommandInfo
@@ -162,12 +162,11 @@ sourceCommands = Map.fromList [
       do sgr <- getGrammar
          liftSIO (exec opts (toStrings ts) sgr)
 
-    compute_concrete opts ws sgr =
+    compute_concrete opts ws sgr = fmap fst $ runCheck $
       case runP pExp (UTF8.fromString s) of
         Left (_,msg) -> return $ pipeMessage msg
-        Right t      -> return $ err pipeMessage
-                                     (fromString . showTerm sgr style q)
-                                 $ checkComputeTerm opts sgr t
+        Right t      -> do t <- checkComputeTerm opts sgr t
+                           return (fromString (showTerm sgr style q t))
       where
         (style,q) = pOpts TermPrintDefault Qualified opts
         s = unwords ws
@@ -200,16 +199,16 @@ sourceCommands = Map.fromList [
                 | otherwise = unwords $ map prTerm ops
           return $ fromString printed
 
-    show_operations os ts sgr =
+    show_operations os ts sgr = fmap fst $ runCheck $
       case greatestResource sgr of
-        Nothing -> return $ fromString "no source grammar in scope; did you import with -retain?"
+        Nothing -> checkError (pp "no source grammar in scope; did you import with -retain?")
         Just mo -> do
           let greps = map valueString (listFlags "grep" os)
           let isRaw = isOpt "raw" os
           ops <- case ts of
              _:_ -> do
                let Right t = runP pExp (UTF8.fromString (unwords ts))
-               ty <- err error return $ checkComputeTerm os sgr t
+               ty <- checkComputeTerm os sgr t
                return $ allOpersTo sgr ty
              _   -> return $ allOpers sgr
           let sigs = [(op,ty) | ((mo,op),ty,pos) <- ops]
@@ -254,14 +253,12 @@ sourceCommands = Map.fromList [
          return void
 
 checkComputeTerm os sgr t =
-  do mo <- maybe (raise "no source grammar in scope") return $
+  do mo <- maybe (checkError (pp "no source grammar in scope")) return $
            greatestResource sgr
-     ((t,_),_) <- runCheck $ do t <- renameSourceTerm sgr mo t
-                                inferLType sgr [] t
+     t <- renameSourceTerm sgr mo t
+     (t,_) <- inferLType sgr [] t                                
      let opts = modifyFlags (\fs->fs{optTrace=isOpt "trace" os})
-         t1 = CN.normalForm (CN.resourceValues opts sgr) (L NoLoc identW) t
-         t2 = evalStr t1
-     checkPredefError t2
+     fmap evalStr (normalForm sgr t)
   where
     -- ** Try to compute pre{...} tokens in token sequences
     evalStr t =

src/compiler/GF/Command/TreeOperations.hs

@@ -1,18 +1,17 @@
 module GF.Command.TreeOperations (
   treeOp,
   allTreeOps,
-  treeChunks
   ) where
 
-import PGF(Expr,PGF,CId,compute,mkApp,unApp,unapply,unMeta,exprSize,exprFunctions)
+import PGF2(Expr,PGF,Fun,compute,mkApp,unApp,unMeta,exprSize,exprFunctions)
 import Data.List
 
 type TreeOp = [Expr] -> [Expr]
 
-treeOp :: PGF -> String -> Maybe (Either TreeOp (CId -> TreeOp))
+treeOp :: PGF -> String -> Maybe (Either TreeOp (Fun -> TreeOp))
 treeOp pgf f = fmap snd $ lookup f $ allTreeOps pgf
 
-allTreeOps :: PGF -> [(String,(String,Either TreeOp (CId -> TreeOp)))]
+allTreeOps :: PGF -> [(String,(String,Either TreeOp (Fun -> TreeOp)))]
 allTreeOps pgf = [
    ("compute",("compute by using semantic definitions (def)",
       Left  $ map (compute pgf))),
@@ -34,16 +33,6 @@ largest = reverse . smallest
 smallest :: [Expr] -> [Expr]
 smallest = sortBy (\t u -> compare (exprSize t) (exprSize u))
 
-treeChunks :: Expr -> [Expr]
-treeChunks = snd . cks where
-  cks t = 
-    case unapply t of
-      (t, ts) -> case unMeta t of
-                   Just _  -> (False,concatMap (snd . cks) ts)
-                   Nothing -> case unzip (map cks ts) of
-                                (bs,_) | and bs      -> (True, [t])
-                                (_,cts)              -> (False,concat cts)
-
 subtrees :: Expr -> [Expr]
 subtrees t = t : case unApp t of
   Just (f,ts) -> concatMap subtrees ts

src/compiler/GF/Compile.hs

@@ -1,6 +1,6 @@
 module GF.Compile (compileToPGF, link, batchCompile, srcAbsName) where
 
-import GF.Compile.GrammarToPGF(mkCanon2pgf)
+import GF.Compile.GrammarToPGF(grammar2PGF)
 import GF.Compile.ReadFiles(ModEnv,getOptionsFromFile,getAllFiles,
                             importsOfModule)
 import GF.CompileOne(compileOne)
@@ -14,7 +14,7 @@ import GF.Infra.UseIO(IOE,FullPath,liftIO,getLibraryDirectory,putIfVerb,
                       justModuleName,extendPathEnv,putStrE,putPointE)
 import GF.Data.Operations(raise,(+++),err)
 
-import Control.Monad(foldM,when,(<=<),filterM,liftM)
+import Control.Monad(foldM,when,(<=<))
 import GF.System.Directory(doesFileExist,getModificationTime)
 import System.FilePath((</>),isRelative,dropFileName)
 import qualified Data.Map as Map(empty,insert,elems) --lookup
@@ -22,8 +22,7 @@ import Data.List(nub)
 import Data.Time(UTCTime)
 import GF.Text.Pretty(render,($$),(<+>),nest)
 
-import PGF.Internal(optimizePGF)
-import PGF(PGF,defaultProbabilities,setProbabilities,readProbabilitiesFromFile)
+import PGF2(PGF,readProbabilitiesFromFile)
 
 -- | Compiles a number of source files and builds a 'PGF' structure for them.
 -- This is a composition of 'link' and 'batchCompile'.
@@ -36,11 +35,10 @@ link :: Options -> (ModuleName,Grammar) -> IOE PGF
 link opts (cnc,gr) =
   putPointE Normal opts "linking ... " $ do
     let abs = srcAbsName gr cnc
-    pgf <- mkCanon2pgf opts gr abs
-    probs <- liftIO (maybe (return . defaultProbabilities) readProbabilitiesFromFile (flag optProbsFile opts) pgf)
+    probs <- liftIO (maybe (return Map.empty) readProbabilitiesFromFile (flag optProbsFile opts))
+    pgf <- grammar2PGF opts gr abs probs
     when (verbAtLeast opts Normal) $ putStrE "OK"
-    return $ setProbabilities probs 
-           $ if flag optOptimizePGF opts then optimizePGF pgf else pgf
+    return pgf
 
 -- | Returns the name of the abstract syntax corresponding to the named concrete syntax
 srcAbsName gr cnc = err (const cnc) id $ abstractOfConcrete gr cnc
@@ -78,14 +76,10 @@ compileModule opts1 env@(_,rfs) file =
   do file <- getRealFile file
      opts0 <- getOptionsFromFile file
      let curr_dir = dropFileName file
-     lib_dirs <- getLibraryDirectory (addOptions opts0 opts1)
-     let opts = addOptions (fixRelativeLibPaths curr_dir lib_dirs opts0) opts1
---     putIfVerb opts $ "curr_dir:" +++ show curr_dir ----
---     putIfVerb opts $ "lib_dir:" +++ show lib_dirs ----
+     lib_dir  <- getLibraryDirectory (addOptions opts0 opts1)
+     let opts = addOptions (fixRelativeLibPaths curr_dir lib_dir opts0) opts1
      ps0 <- extendPathEnv opts
      let ps = nub (curr_dir : ps0)
---     putIfVerb opts $ "options from file: " ++ show opts0
---     putIfVerb opts $ "augmented options: " ++ show opts
      putIfVerb opts $ "module search path:" +++ show ps ----
      files <- getAllFiles opts ps rfs file
      putIfVerb opts $ "files to read:" +++ show files ----
@@ -98,17 +92,13 @@ compileModule opts1 env@(_,rfs) file =
       if exists
         then return file
         else if isRelative file
-               then do
-                       lib_dirs <- getLibraryDirectory opts1
-                       let candidates = [ lib_dir </> file | lib_dir <- lib_dirs ]
-                       putIfVerb opts1 (render ("looking for: " $$ nest 2 candidates))
-                       file1s <- filterM doesFileExist candidates
-                       case length file1s of
-                         0 -> raise (render ("Unable to find: " $$ nest 2 candidates))
-                         1 -> do return $ head file1s
-                         _ -> do putIfVerb opts1 ("matched multiple candidates: " +++ show file1s)
-                                 return $ head file1s
-               else raise (render ("File" <+> file <+> "does not exist"))
+               then do lib_dir <- getLibraryDirectory opts1
+                       let file1 = lib_dir </> file
+                       exists <- doesFileExist file1
+                       if exists
+                         then return file1
+                         else raise (render ("None of these files exists:" $$ nest 2 (file $$ file1)))
+               else raise (render ("File" <+> file <+> "does not exist."))
 
 compileOne' :: Options -> CompileEnv -> FullPath -> IOE CompileEnv
 compileOne' opts env@(gr,_) = extendCompileEnv env <=< compileOne opts gr

src/compiler/GF/Compile/CFGtoPGF.hs

@@ -1,99 +1,110 @@
-{-# LANGUAGE FlexibleContexts #-}
+{-# LANGUAGE FlexibleContexts, ImplicitParams #-}
 module GF.Compile.CFGtoPGF (cf2pgf) where
 
 import GF.Grammar.CFG
 import GF.Infra.UseIO
+import GF.Infra.Option
+import GF.Compile.OptimizePGF
 
-import PGF
-import PGF.Internal
+import PGF2
+import PGF2.Internal
 
 import qualified Data.Set as Set
 import qualified Data.Map as Map
 import qualified Data.IntMap as IntMap
 import Data.Array.IArray
 import Data.List
+import Data.Maybe(fromMaybe)
 
 --------------------------
 -- the compiler ----------
 --------------------------
 
-cf2pgf :: FilePath -> ParamCFG -> PGF
-cf2pgf fpath cf = 
- let pgf = PGF Map.empty aname (cf2abstr cf) (Map.singleton cname (cf2concr cf))
- in updateProductionIndices pgf
+cf2pgf :: Options -> FilePath -> ParamCFG -> Map.Map Fun Double -> PGF
+cf2pgf opts fpath cf probs = error "TODO: cf2pgf" {-
+ build (let abstr = cf2abstr cf probs
+        in newPGF [] aname abstr [(cname, cf2concr opts abstr cf)])
  where
    name = justModuleName fpath
-   aname = mkCId (name ++ "Abs")
-   cname = mkCId name
+   aname = name ++ "Abs"
+   cname = name
 
-cf2abstr :: ParamCFG -> Abstr
-cf2abstr cfg = Abstr aflags afuns acats
+cf2abstr :: (?builder :: Builder s) => ParamCFG -> Map.Map Fun Double -> B s AbstrInfo
+cf2abstr cfg probs = newAbstr aflags acats afuns
   where
-    aflags = Map.singleton (mkCId "startcat") (LStr (fst (cfgStartCat cfg)))
+    aflags = [("startcat", LStr (fst (cfgStartCat cfg)))]
 
-    acats  = Map.fromList [(cat, ([], [(0,mkRuleName rule) | rule <- rules], 0))
-                            | (cat,rules) <- (Map.toList . Map.fromListWith (++))
-                                                [(cat2id cat, catRules cfg cat) | 
-                                                     cat <- allCats' cfg]]
-    afuns  = Map.fromList [(mkRuleName rule, (cftype [cat2id c | NonTerminal c <- ruleRhs rule] (cat2id (ruleLhs rule)), 0, Nothing, 0))
-                            | rule <- allRules cfg]
+    acats  = [(c', [], toLogProb (fromMaybe 0 (Map.lookup c' probs))) | cat <- allCats' cfg, let c' = cat2id cat]
+    afuns  = [(f', dTyp [hypo Explicit "_" (dTyp [] (cat2id c) []) | NonTerminal c <- ruleRhs rule] (cat2id (ruleLhs rule)) [], 0, [], toLogProb (fromMaybe 0 (Map.lookup f' funs_probs)))
+                            | rule <- allRules cfg
+                            , let f' = mkRuleName rule]
 
-    cat2id = mkCId . fst
+    funs_probs = (Map.fromList . concat . Map.elems . fmap pad . Map.fromListWith (++))
+                     [(cat,[(f',Map.lookup f' probs)]) | rule <- allRules cfg,
+                                                         let cat = cat2id (ruleLhs rule),
+                                                         let f' = mkRuleName rule]
+      where
+        pad :: [(a,Maybe Double)] -> [(a,Double)]
+        pad pfs = [(f,fromMaybe deflt mb_p) | (f,mb_p) <- pfs]
+          where
+            deflt = case length [f | (f,Nothing) <- pfs] of
+                      0 -> 0
+                      n -> max 0 ((1 - sum [d | (f,Just d) <- pfs]) / fromIntegral n)
 
-cf2concr :: ParamCFG -> Concr
-cf2concr cfg = Concr Map.empty Map.empty
-                     cncfuns lindefsrefs lindefsrefs
-                     sequences productions
-                     IntMap.empty Map.empty
-                     cnccats
-                     IntMap.empty
-                     totalCats
+    toLogProb = realToFrac . negate . log
+
+    cat2id = fst
+
+cf2concr :: (?builder :: Builder s) => Options -> B s AbstrInfo -> ParamCFG -> B s ConcrInfo
+cf2concr opts abstr cfg = 
+  let (lindefs',linrefs',productions',cncfuns',sequences',cnccats') =
+           (if flag optOptimizePGF opts then optimizePGF (fst (cfgStartCat cfg)) else id)
+                (lindefsrefs,lindefsrefs,IntMap.toList productions,cncfuns,sequences,cnccats)
+  in newConcr abstr [] []
+              lindefs' linrefs'
+              productions' cncfuns'
+              sequences' cnccats' totalCats
   where
     cats  = allCats' cfg
     rules = allRules cfg
 
-    sequences0 = Set.fromList (listArray (0,0) [SymCat 0 0] : 
-                               map mkSequence rules)
-    sequences  = listArray (0,Set.size sequences0-1) (Set.toList sequences0)
+    idSeq = [SymCat 0 0]
 
-    idFun = CncFun wildCId (listArray (0,0) [seqid])
-      where
-        seq   = listArray (0,0) [SymCat 0 0]
-        seqid = binSearch seq sequences (bounds sequences)
+    sequences0 = Set.fromList (idSeq : 
+                               map mkSequence rules)
+    sequences  = Set.toList sequences0
+
+    idFun = ("_",[Set.findIndex idSeq sequences0])
     ((fun_cnt,cncfuns0),productions0) = mapAccumL (convertRule cs) (1,[idFun]) rules
     productions = foldl addProd IntMap.empty (concat (productions0++coercions))
-    cncfuns = listArray (0,fun_cnt-1) (reverse cncfuns0)
+    cncfuns = reverse cncfuns0
 
-    lbls = listArray (0,0) ["s"]
-    (fid,cnccats0) = (mapAccumL mkCncCat 0 . Map.toList . Map.fromListWith max)
-                              [(c,p) | (c,ps) <- cats, p <- ps]
+    lbls = ["s"]
+    (fid,cnccats) = (mapAccumL mkCncCat 0 . Map.toList . Map.fromListWith max)
+                         [(c,p) | (c,ps) <- cats, p <- ps]
     ((totalCats,cs), coercions) = mapAccumL mkCoercions (fid,Map.empty) cats
-    cnccats = Map.fromList cnccats0
 
-    lindefsrefs =
-       IntMap.fromList (map mkLinDefRef cats)
+    lindefsrefs = map mkLinDefRef cats
 
     convertRule cs (funid,funs) rule =
       let args   = [PArg [] (cat2arg c) | NonTerminal c <- ruleRhs rule]
           prod   = PApply funid args
-          seqid  = binSearch (mkSequence rule) sequences (bounds sequences)
-          fun    = CncFun (mkRuleName rule) (listArray (0,0) [seqid])
+          seqid  = Set.findIndex (mkSequence rule) sequences0
+          fun    = (mkRuleName rule, [seqid])
           funid' = funid+1
       in funid' `seq` ((funid',fun:funs),let (c,ps) = ruleLhs rule in [(cat2fid c p, prod) | p <- ps])
 
-    mkSequence rule = listArray (0,length syms-1) syms
+    mkSequence rule = snd $ mapAccumL convertSymbol 0 (ruleRhs rule)
       where
-        syms   = snd $ mapAccumL convertSymbol 0 (ruleRhs rule)
-
         convertSymbol d (NonTerminal (c,_)) = (d+1,if c `elem` ["Int","Float","String"] then SymLit d 0 else SymCat d 0)
         convertSymbol d (Terminal t)        = (d,  SymKS t)
 
     mkCncCat fid (cat,n)
-      | cat == "Int"    = (fid, (mkCId cat, CncCat fidInt    fidInt    lbls))
-      | cat == "Float"  = (fid, (mkCId cat, CncCat fidFloat  fidFloat  lbls))
-      | cat == "String" = (fid, (mkCId cat, CncCat fidString fidString lbls))
+      | cat == "Int"    = (fid, (cat, fidInt,    fidInt,    lbls))
+      | cat == "Float"  = (fid, (cat, fidFloat,  fidFloat,  lbls))
+      | cat == "String" = (fid, (cat, fidString, fidString, lbls))
       | otherwise       = let fid' = fid+n+1
-                          in fid' `seq` (fid', (mkCId cat,CncCat fid (fid+n) lbls))
+                          in fid' `seq` (fid', (cat, fid, fid+n, lbls))
 
     mkCoercions (fid,cs) c@(cat,[p]) = ((fid,cs),[])
     mkCoercions (fid,cs) c@(cat,ps ) =
@@ -102,25 +113,16 @@ cf2concr cfg = Concr Map.empty Map.empty
 
     mkLinDefRef (cat,_) =
       (cat2fid cat 0,[0])
-      
+
     addProd prods (fid,prod) =
       case IntMap.lookup fid prods of
-        Just set -> IntMap.insert fid (Set.insert prod set) prods
-        Nothing  -> IntMap.insert fid (Set.singleton prod)  prods
-
-    binSearch v arr (i,j)
-      | i <= j    = case compare v (arr ! k) of
-                      LT -> binSearch v arr (i,k-1)
-                      EQ -> k
-                      GT -> binSearch v arr (k+1,j)
-      | otherwise = error "binSearch"
-      where
-        k = (i+j) `div` 2
+        Just set -> IntMap.insert fid (prod:set) prods
+        Nothing  -> IntMap.insert fid [prod]     prods
 
     cat2fid cat p =
-      case Map.lookup (mkCId cat) cnccats of
-        Just (CncCat fid _ _) -> fid+p
-        _                     -> error "cat2fid"
+      case [start | (cat',start,_,_) <- cnccats, cat == cat'] of
+        (start:_) -> fid+p
+        _         -> error "cat2fid"
 
     cat2arg c@(cat,[p]) = cat2fid cat p
     cat2arg c@(cat,ps ) =
@@ -131,4 +133,5 @@ cf2concr cfg = Concr Map.empty Map.empty
 mkRuleName rule =
   case ruleName rule of
 	CFObj n _ -> n
-	_         -> wildCId
+	_         -> "_"
+-}

src/compiler/GF/Compile/CheckGrammar.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/11/11 23:24:33 $ 
+-- > CVS $Date: 2005/11/11 23:24:33 $
 -- > CVS $Author: aarne $
 -- > CVS $Revision: 1.31 $
 --
@@ -21,15 +21,15 @@
 -----------------------------------------------------------------------------
 
 module GF.Compile.CheckGrammar(checkModule) where
-import Prelude hiding ((<>)) -- GHC 8.4.1 clash with Text.PrettyPrint
 
+import Prelude hiding ((<>))
 import GF.Infra.Ident
 import GF.Infra.Option
 
 import GF.Compile.TypeCheck.Abstract
-import GF.Compile.TypeCheck.RConcrete
-import qualified GF.Compile.TypeCheck.ConcreteNew as CN
-import qualified GF.Compile.Compute.ConcreteNew as CN
+import GF.Compile.TypeCheck.Concrete(checkLType,inferLType,ppType)
+import qualified GF.Compile.TypeCheck.ConcreteNew as CN(checkLType,inferLType)
+import GF.Compile.Compute.Concrete(normalForm)
 
 import GF.Grammar
 import GF.Grammar.Lexer
@@ -54,11 +54,7 @@ checkModule opts cwd sgr mo@(m,mi) = do
                              checkCompleteGrammar opts cwd gr (a,abs) mo
           _            -> return mo
   infoss <- checkInModule cwd mi NoLoc empty $ topoSortJments2 mo
-  foldM updateCheckInfos mo infoss
-  where
-    updateCheckInfos mo = fmap (foldl update mo) . parallelCheck . map check
-      where check (i,info) = fmap ((,) i) (checkInfo opts cwd sgr mo i info)
-    update mo@(m,mi) (i,info) = (m,mi{jments=Map.insert i info (jments mi)})
+  foldM (foldM (checkInfo opts cwd sgr)) mo infoss
 
 -- check if restricted inheritance modules are still coherent
 -- i.e. that the defs of remaining names don't depend on omitted names
@@ -74,9 +70,9 @@ checkRestrictedInheritance cwd sgr (name,mo) = checkInModule cwd mo NoLoc empty
      let (incl,excl) = partition (isInherited mi) (Map.keys (jments m))
      let incld c   = Set.member c (Set.fromList incl)
      let illegal c = Set.member c (Set.fromList excl)
-     let illegals = [(f,is) | 
+     let illegals = [(f,is) |
            (f,cs) <- allDeps, incld f, let is = filter illegal cs, not (null is)]
-     case illegals of 
+     case illegals of
        [] -> return ()
        cs -> checkWarn ("In inherited module" <+> i <> ", dependence of excluded constants:" $$
                          nest 2 (vcat [f <+> "on" <+> fsep is | (f,is) <- cs]))
@@ -92,12 +88,12 @@ checkCompleteGrammar opts cwd gr (am,abs) (cm,cnc) = checkInModule cwd cnc NoLoc
 
   -- check that all abstract constants are in concrete; build default lin and lincats
   jsc <- foldM checkAbs jsc (Map.toList jsa)
-  
+
   return (cm,cnc{jments=jsc})
   where
    checkAbs js i@(c,info) =
      case info of
-       AbsFun (Just (L loc ty)) _ _ _ 
+       AbsFun (Just (L loc ty)) _ _ _
            -> do let mb_def = do
                        let (cxt,(_,i),_) = typeForm ty
                        info <- lookupIdent i js
@@ -120,8 +116,7 @@ checkCompleteGrammar opts cwd gr (am,abs) (cm,cnc) = checkInModule cwd cnc NoLoc
                                     return js
                    _ -> do
                      case mb_def of
-                       Ok def -> do (cont,val) <- linTypeOfType gr cm ty
-                                    let linty = (snd (valCat ty),cont,val)
+                       Ok def -> do linty <- linTypeOfType gr cm (L loc ty)
                                     return $ Map.insert c (CncFun (Just linty) (Just (L NoLoc def)) Nothing Nothing) js
                        Bad _  -> do noLinOf c
                                     return js
@@ -136,13 +131,12 @@ checkCompleteGrammar opts cwd gr (am,abs) (cm,cnc) = checkInModule cwd cnc NoLoc
            checkWarn ("no linearization type for" <+> c <> ", inserting default {s : Str}")
            return $ Map.insert c (CncCat (Just (L NoLoc defLinType)) Nothing Nothing Nothing Nothing) js
        _ -> return js
-     
+
    checkCnc js (c,info) =
      case info of
        CncFun _ d mn mf -> case lookupOrigInfo gr (am,c) of
-                             Ok (_,AbsFun (Just (L _ ty)) _ _ _) -> 
-                                        do (cont,val) <- linTypeOfType gr cm ty
-                                           let linty = (snd (valCat ty),cont,val)
+                             Ok (_,AbsFun (Just (L loc ty)) _ _ _) ->
+                                        do linty <- linTypeOfType gr cm (L loc ty)
                                            return $ Map.insert c (CncFun (Just linty) d mn mf) js
                              _       -> do checkWarn ("function" <+> c <+> "is not in abstract")
                                            return js
@@ -158,130 +152,123 @@ checkCompleteGrammar opts cwd gr (am,abs) (cm,cnc) = checkInModule cwd cnc NoLoc
 
        _ -> return $ Map.insert c info js
 
-
--- | General Principle: only Just-values are checked. 
--- A May-value has always been checked in its origin module.
-checkInfo :: Options -> FilePath -> SourceGrammar -> SourceModule -> Ident -> Info -> Check Info
-checkInfo opts cwd sgr (m,mo) c info = checkInModule cwd mo NoLoc empty $ do
+checkInfo :: Options -> FilePath -> SourceGrammar -> SourceModule -> (Ident,Info) -> Check SourceModule
+checkInfo opts cwd sgr sm (c,info) = checkInModule cwd (snd sm) NoLoc empty $ do
   checkReservedId c
   case info of
-    AbsCat (Just (L loc cont)) -> 
-      mkCheck loc "the category" $ 
+    AbsCat (Just (L loc cont)) ->
+      mkCheck loc "the category" $
         checkContext gr cont
 
-    AbsFun (Just (L loc typ0)) ma md moper -> do
-      typ <- compAbsTyp [] typ0   -- to calculate let definitions
+    AbsFun (Just (L loc typ)) ma md moper -> do
       mkCheck loc "the type of function" $
         checkTyp gr typ
+      typ <- compAbsTyp [] typ   -- to calculate let definitions
       case md of
         Just eqs -> mapM_ (\(L loc eq) -> mkCheck loc "the definition of function" $
-	                                        checkDef gr (m,c) typ eq) eqs
+                                          checkDef gr (fst sm,c) typ eq) eqs
         Nothing  -> return ()
-      return (AbsFun (Just (L loc typ)) ma md moper)
+      update sm c (AbsFun (Just (L loc typ)) ma md moper)
 
     CncCat mty mdef mref mpr mpmcfg -> do
       mty  <- case mty of
-                Just (L loc typ) -> chIn loc "linearization type of" $ 
-                                     (if False --flag optNewComp opts
-                                        then do (typ,_) <- CN.checkLType (CN.resourceValues opts gr) typ typeType
-                                                typ  <- computeLType gr [] typ
-                                                return (Just (L loc typ))
-                                        else do (typ,_) <- checkLType gr [] typ typeType
-                                                typ  <- computeLType gr [] typ
-                                                return (Just (L loc typ)))
+                Just (L loc typ) -> chIn loc "linearization type of" $ do
+                                      (typ,_) <- checkLType gr [] typ typeType
+                                      typ  <- normalForm gr typ
+                                      return (Just (L loc typ))
                 Nothing          -> return Nothing
       mdef <- case (mty,mdef) of
-                (Just (L _ typ),Just (L loc def)) -> 
+                (Just (L _ typ),Just (L loc def)) ->
                     chIn loc "default linearization of" $ do
                        (def,_) <- checkLType gr [] def (mkFunType [typeStr] typ)
                        return (Just (L loc def))
                 _ -> return Nothing
       mref <- case (mty,mref) of
-                (Just (L _ typ),Just (L loc ref)) -> 
+                (Just (L _ typ),Just (L loc ref)) ->
                     chIn loc "reference linearization of" $ do
                        (ref,_) <- checkLType gr [] ref (mkFunType [typ] typeStr)
                        return (Just (L loc ref))
                 _ -> return Nothing
       mpr  <- case mpr of
-                (Just (L loc t)) -> 
+                (Just (L loc t)) ->
                     chIn loc "print name of" $ do
                        (t,_) <- checkLType gr [] t typeStr
                        return (Just (L loc t))
                 _ -> return Nothing
-      return (CncCat mty mdef mref mpr mpmcfg)
+      update sm c (CncCat mty mdef mref mpr mpmcfg)
 
     CncFun mty mt mpr mpmcfg -> do
       mt <- case (mty,mt) of
-              (Just (cat,cont,val),Just (L loc trm)) -> 
+              (Just (_,cat,cont,val),Just (L loc trm)) ->
                   chIn loc "linearization of" $ do
                      (trm,_) <- checkLType gr [] trm (mkFunType (map (\(_,_,ty) -> ty) cont) val)  -- erases arg vars
                      return (Just (L loc trm))
               _ -> return mt
       mpr  <- case mpr of
-                (Just (L loc t)) -> 
+                (Just (L loc t)) ->
                     chIn loc "print name of" $ do
                        (t,_) <- checkLType gr [] t typeStr
                        return (Just (L loc t))
                 _ -> return Nothing
-      return (CncFun mty mt mpr mpmcfg)
+      update sm c (CncFun mty mt mpr mpmcfg)
 
     ResOper pty pde -> do
       (pty', pde') <- case (pty,pde) of
          (Just (L loct ty), Just (L locd de)) -> do
-           ty'     <- chIn loct "operation" $
-                         (if False --flag optNewComp opts
-                            then CN.checkLType (CN.resourceValues opts gr) ty typeType >>= return . CN.normalForm (CN.resourceValues opts gr) (L loct c) . fst -- !!
-                            else checkLType gr [] ty typeType >>= computeLType gr [] . fst)
+           ty'     <- chIn loct "operation" $ do
+                         (ty,_) <- checkLType gr [] ty typeType
+                         normalForm gr ty
            (de',_) <- chIn locd "operation" $
-                         (if False -- flag optNewComp opts
-                            then CN.checkLType (CN.resourceValues opts gr) de ty'
-                            else checkLType gr [] de ty')
+                         checkLType gr [] de ty'
            return (Just (L loct ty'), Just (L locd de'))
          (Nothing         , Just (L locd de)) -> do
            (de',ty') <- chIn locd "operation" $
-                          (if False -- flag optNewComp opts
-                            then CN.inferLType (CN.resourceValues opts gr) de
-                            else inferLType gr [] de)
+                          inferLType gr [] de
            return (Just (L locd ty'), Just (L locd de'))
          (Just (L loct ty), Nothing) -> do
            chIn loct "operation" $
              checkError (pp "No definition given to the operation")
-      return (ResOper pty' pde')
+      update sm c (ResOper pty' pde')
 
     ResOverload os tysts -> chIn NoLoc "overloading" $ do
       tysts' <- mapM (uncurry $ flip (\(L loc1 t) (L loc2 ty) -> checkLType gr [] t ty >>= \(t,ty) -> return (L loc1 t, L loc2 ty))) tysts  -- return explicit ones
-      tysts0 <- lookupOverload gr (m,c)  -- check against inherited ones too
-      tysts1 <- mapM (uncurry $ flip (checkLType gr [])) 
+      tysts0 <- lookupOverload gr (fst sm,c)  -- check against inherited ones too
+      tysts1 <- mapM (uncurry $ flip (checkLType gr []))
                   [(mkFunType args val,tr) | (args,(val,tr)) <- tysts0]
       --- this can only be a partial guarantee, since matching
       --- with value type is only possible if expected type is given
-      checkUniq $ 
+      checkUniq $
         sort [let (xs,t) = typeFormCnc x in t : map (\(b,x,t) -> t) xs | (_,x) <- tysts1]
-      return (ResOverload os [(y,x) | (x,y) <- tysts'])
+      update sm c (ResOverload os [(y,x) | (x,y) <- tysts'])
 
     ResParam (Just (L loc pcs)) _ -> do
-      ts <- chIn loc "parameter type" $ 
-              liftM concat $ mapM mkPar pcs
-      return (ResParam (Just (L loc pcs)) (Just ts))
+      (sm,cnt,ts) <- chIn loc "parameter type" $
+                        mkParamValues sm 0 [] pcs
+      update sm c (ResParam (Just (L loc pcs)) (Just (ts,cnt)))
 
-    _ ->  return info
+    _ ->  return sm
  where
-   gr = prependModule sgr (m,mo)
-   chIn loc cat = checkInModule cwd mo loc ("Happened in" <+> cat <+> c)
+   gr = prependModule sgr sm
+   chIn loc cat = checkInModule cwd (snd sm) loc ("Happened in" <+> cat <+> c)
 
-   mkPar (f,co) = do
-       vs <- liftM sequence $ mapM (\(_,_,ty) -> allParamValues gr ty) co
-       return $ map (mkApp (QC (m,f))) vs
+   mkParamValues sm         cnt ts []          = return (sm,cnt,[])
+   mkParamValues sm@(mn,mi) cnt ts ((f,co):fs) = do
+     sm <- case lookupIdent f (jments mi) of
+             Ok (ResValue ty _) -> update sm f (ResValue ty cnt)
+             Bad msg            -> checkError (pp msg)
+     vs <- liftM sequence $ mapM (\(_,_,ty) -> allParamValues gr ty) co
+     (sm,cnt,ts) <- mkParamValues sm (cnt+length vs) ts fs
+     return (sm,cnt,map (mkApp (QC (mn,f))) vs ++ ts)
 
    checkUniq xss = case xss of
-     x:y:xs 
+     x:y:xs
       | x == y    -> checkError $ "ambiguous for type" <+>
-                                  ppType (mkFunType (tail x) (head x))  
+                                  ppType (mkFunType (tail x) (head x))
       | otherwise -> checkUniq $ y:xs
      _ -> return ()
 
    mkCheck loc cat ss = case ss of
-     [] -> return info
+     [] -> return sm
      _ -> chIn loc cat $ checkError (vcat ss)
 
    compAbsTyp g t = case t of
@@ -296,6 +283,8 @@ checkInfo opts cwd sgr (m,mo) c info = checkInModule cwd mo NoLoc empty $ do
      Abs _ _ _ -> return t
      _ -> composOp (compAbsTyp g) t  
 
+   update (mn,mi) c info = return (mn,mi{jments=Map.insert c info (jments mi)})
+
 
 -- | for grammars obtained otherwise than by parsing ---- update!!
 checkReservedId :: Ident -> Check ()
@@ -306,25 +295,26 @@ checkReservedId x =
 -- auxiliaries
 
 -- | linearization types and defaults
-linTypeOfType :: Grammar -> ModuleName -> Type -> Check (Context,Type)
-linTypeOfType cnc m typ = do
-  let (cont,cat) = typeSkeleton typ
-  val  <- lookLin cat
-  args <- mapM mkLinArg (zip [0..] cont)
-  return (args, val)
+linTypeOfType :: Grammar -> ModuleName -> L Type -> Check ([Ident],Ident,Context,Type)
+linTypeOfType cnc m (L loc typ) = do
+  let (ctxt,res_cat) = typeSkeleton typ
+  val <- lookLin res_cat
+  lin_args <- mapM mkLinArg (zip [0..] ctxt)
+  let (args,arg_cats) = unzip lin_args
+  return (arg_cats, snd res_cat, args, val)
  where
    mkLinArg (i,(n,mc@(m,cat))) = do
      val  <- lookLin mc
      let vars = mkRecType varLabel $ replicate n typeStr
-	 symb = argIdent n cat i
+         symb = argIdent n cat i
      rec <- if n==0 then return val else
                        errIn (render ("extending" $$
                                       nest 2 vars $$
                                       "with" $$
                                       nest 2 val)) $
                              plusRecType vars val
-     return (Explicit,symb,rec)
+     return ((Explicit,symb,rec),cat)
    lookLin (_,c) = checks [ --- rather: update with defLinType ?
-      lookupLincat cnc m c >>= computeLType cnc []
+      lookupLincat cnc m c >>= normalForm cnc
      ,return defLinType
      ]

src/compiler/GF/Compile/Compute/Concrete.hs

@@ -1,3 +1,538 @@
-module GF.Compile.Compute.Concrete{-(module M)-} where
---import GF.Compile.Compute.ConcreteLazy as M -- New
---import GF.Compile.Compute.ConcreteStrict as M -- Old, inefficient
+{-# LANGUAGE RankNTypes, CPP #-}
+
+-- | Functions for computing the values of terms in the concrete syntax, in
+-- | preparation for PMCFG generation.
+module GF.Compile.Compute.Concrete
+           ( normalForm
+           , Value(..), Thunk, ThunkState(..), Env
+           , EvalM, runEvalM, evalError
+           , eval, apply, force, value2term, patternMatch
+           , newMeta,getMeta,setMeta
+           , newThunk,newEvaluatedThunk
+           , getResDef, getInfo, getAllParamValues
+           ) where
+
+import Prelude hiding ((<>)) -- GHC 8.4.1 clash with Text.PrettyPrint
+
+import GF.Grammar hiding (Env, VGen, VApp, VRecType)
+import GF.Grammar.Lookup(lookupResDef,lookupOrigInfo,allParamValues)
+import GF.Grammar.Predef
+import GF.Grammar.Lockfield(lockLabel)
+import GF.Grammar.Printer
+import GF.Data.Str(Str,glueStr,str2strings,str,sstr,plusStr,strTok)
+import GF.Data.Operations(Err(..),err,errIn,maybeErr,mapPairsM)
+import GF.Data.Utilities(mapFst,mapSnd)
+import GF.Infra.CheckM
+import GF.Infra.Option
+import Data.STRef
+import Data.Maybe(fromMaybe)
+import Data.List
+import Data.Char
+import Control.Monad
+import Control.Monad.ST
+import Control.Applicative
+import qualified Control.Monad.Fail as Fail
+import qualified Data.Map as Map
+import GF.Text.Pretty
+import PGF2.Transactions(LIndex)
+
+-- * Main entry points
+
+normalForm :: Grammar -> Term -> Check Term
+normalForm gr t =
+  fmap mkFV (runEvalM gr (eval [] t [] >>= value2term 0))
+  where
+    mkFV [t] = t
+    mkFV ts  = FV ts
+
+
+data ThunkState s
+  = Unevaluated (Env s) Term
+  | Evaluated (Value s)
+  | Unbound (Maybe Type) {-# UNPACK #-} !MetaId
+
+type Thunk s = STRef s (ThunkState s)
+type Env s = [(Ident,Thunk s)]
+
+data Value s
+  = VApp QIdent [Thunk s]
+  | VMeta (Thunk s) (Env s) [Thunk s]
+  | VSusp (Thunk s) (Env s) [Thunk s] (Thunk s -> EvalM s (Value s))
+  | VGen  {-# UNPACK #-} !Int [Thunk s]
+  | VClosure (Env s) Term
+  | VProd BindType Ident (Value s) (Env s) Term
+  | VRecType [(Label, Value s)]
+  | VR [(Label, Thunk s)]
+  | VP (Value s) Label [Thunk s]
+  | VExtR (Value s) (Value s)
+  | VTable (Value s) (Value s)
+  | VT TInfo (Env s) [Case]
+  | VV Type [Thunk s]
+  | VS (Value s) (Thunk s) [Thunk s]
+  | VSort Ident
+  | VInt Integer
+  | VFlt Double
+  | VStr String
+  | VC [Value s]
+  | VGlue (Value s) (Value s)
+  | VPatt Int (Maybe Int) Patt
+  | VPattType (Value s)
+  | VAlts (Value s) [(Value s, Value s)]
+  | VStrs [Value s]
+    -- This last constructor is only generated internally
+    -- in the PMCFG generator.
+  | VSymCat Int LIndex [(LIndex, Thunk s)]
+
+
+eval env (Vr x)         vs  = case lookup x env of
+                                Just tnk -> force tnk vs
+                                Nothing  -> evalError ("Variable" <+> pp x <+> "is not in scope")
+eval env (Sort s)       []  = return (VSort s)
+eval env (EInt n)       []  = return (VInt n)
+eval env (EFloat d)     []  = return (VFlt d)
+eval env (K t)          []  = return (VStr t)
+eval env Empty          []  = return (VC [])
+eval env (App t1 t2)    vs  = do tnk <- newThunk env t2
+                                 eval env t1 (tnk : vs)
+eval env (Abs b x t)    []  = return (VClosure env (Abs b x t))
+eval env (Abs b x t) (v:vs) = eval ((x,v):env) t vs
+eval env (Meta i)       vs  = do tnk <- newMeta Nothing i
+                                 return (VMeta tnk env vs)
+eval env (ImplArg t)    []  = eval env t []
+eval env (Prod b x t1 t2)[] = do v1 <- eval env t1 []
+                                 return (VProd b x v1 env t2)
+eval env (Typed t ty)   vs  = eval env t vs
+eval env (RecType lbls) []  = do lbls <- mapM (\(lbl,ty) -> fmap ((,) lbl) (eval env ty [])) lbls
+                                 return (VRecType lbls)
+eval env (R as)         []  = do as <- mapM (\(lbl,(_,t)) -> fmap ((,) lbl) (newThunk env t)) as
+                                 return (VR as)
+eval env (P t lbl)      vs  = do v <- eval env t []
+                                 case v of
+                                   VR as -> case lookup lbl as of
+                                              Nothing  -> evalError ("Missing value for label" <+> pp lbl $$
+                                                                     "in record" <+> pp t)
+                                              Just tnk -> force tnk vs
+                                   v     -> return (VP v lbl vs)
+eval env (ExtR t1 t2)   []  = do v1 <- eval env t1 []
+                                 v2 <- eval env t2 []
+                                 case (v1,v2) of
+                                   (VR       as1,VR       as2) -> return (VR       (foldl (\as (lbl,v) -> update lbl v as) as1 as2))
+                                   (VRecType as1,VRecType as2) -> return (VRecType (foldl (\as (lbl,v) -> update lbl v as) as1 as2))
+                                   _                           -> return (VExtR v1 v2)
+eval env (Table t1 t2)  []  = do v1 <- eval env t1 []
+                                 v2 <- eval env t2 []
+                                 return (VTable v1 v2)
+eval env (T i cs)       []  = return (VT i env cs)
+eval env (V ty ts)      []  = do tnks <- mapM (newThunk env) ts
+                                 return (VV ty tnks)
+eval env t@(S t1 t2)    vs  = do v1   <- eval env t1 []
+                                 tnk2 <- newThunk env t2
+                                 let v0 = VS v1 tnk2 vs
+                                 case v1 of
+                                   VT _  env cs -> patternMatch v0 (map (\(p,t) -> (env,[p],tnk2:vs,t)) cs)
+                                   VV ty tnks   -> do t2 <- force tnk2 [] >>= value2term (length env)
+                                                      ts <- getAllParamValues ty
+                                                      case lookup t2 (zip ts tnks) of
+                                                        Just tnk -> force tnk vs
+                                                        Nothing  -> return v0
+                                   v1      -> return v0
+eval env (Let (x,(_,t1)) t2) vs = do tnk <- newThunk env t1
+                                     eval ((x,tnk):env) t2 vs
+eval env (Q q@(m,id))   vs
+  | m == cPredef            = do vs' <- mapM (flip force []) vs
+                                 mb_res <- evalPredef id vs'
+                                 case mb_res of
+                                   Just res -> return res
+                                   Nothing  -> return (VApp q vs)
+  | otherwise               = do t <- getResDef q
+                                 eval env t vs
+eval env (QC q)         vs  = return (VApp q vs)
+eval env (C t1 t2)      []  = do v1 <- eval env t1 []
+                                 v2 <- eval env t2 []
+                                 case (v1,v2) of
+                                   (VC vs1,VC vs2) -> return (VC (vs1++vs2))
+                                   (VC vs1,v2    ) -> return (VC (vs1++[v2]))
+                                   (v1,    VC vs2) -> return (VC ([v1]++vs2))
+                                   (v1,    v2    ) -> return (VC [v1,v2])
+eval env t@(Glue t1 t2) []  = do v1 <- eval env t1 []
+                                 v2 <- eval env t2 []
+                                 case liftM2 (++) (value2string v1) (value2string v2) of
+                                   Just s  -> return (string2value s)
+                                   Nothing -> return (VGlue v1 v2)
+eval env (EPatt min max p) [] = return (VPatt min max p)
+eval env (EPattType t)  []  = do v <- eval env t []
+                                 return (VPattType v)
+eval env (ELincat c ty) []  = do v <- eval env ty []
+                                 let lbl = lockLabel c
+                                     lv  = VRecType []
+                                 case v of
+                                   (VRecType as) -> return (VRecType (update lbl lv as))
+                                   _             -> return (VExtR v (VRecType [(lbl,lv)]))
+eval env (ELin c t)     []  = do v <- eval env t []
+                                 let lbl = lockLabel c
+                                 tnk <- newEvaluatedThunk (VR [])
+                                 case v of
+                                   (VR as) -> return (VR (update lbl tnk as))
+                                   _       -> return (VExtR v (VR [(lbl,tnk)]))
+eval env (FV ts)        vs  = msum [eval env t vs | t <- ts]
+eval env (Alts d as)    []  = do vd <- eval env d []
+                                 vas <- forM as $ \(t,s) -> do
+                                           vt <- eval env t []
+                                           vs <- eval env s []
+                                           return (vt,vs)
+                                 return (VAlts vd vas)
+eval env (Strs ts)      []  = do vs <- mapM (\t -> eval env t []) ts
+                                 return (VStrs vs)
+eval env (TSymCat d r rs) []= do rs <- forM rs $ \(i,pv) ->
+                                         case lookup pv env of
+                                           Just tnk -> return (i,tnk)
+                                           Nothing  -> evalError ("Variable" <+> pp pv <+> "is not in scope")
+                                 return (VSymCat d r rs)
+eval env t              vs  = evalError ("Cannot reduce term" <+> pp t)
+
+apply (VMeta m env vs0)             vs  = do st <- getMeta m
+                                             case st of
+                                               Evaluated v -> apply v vs
+                                               Unbound _ _ -> return (VMeta m env (vs0++vs))
+apply (VApp f  vs0)                 vs  = return (VApp f (vs0++vs))
+apply (VGen i  vs0)                 vs  = return (VGen i (vs0++vs))
+apply (VClosure env (Abs b x t)) (v:vs) = eval ((x,v):env) t vs
+apply v                             []  = return v
+
+evalPredef id [v]
+  | id == cLength = return (fmap VInt (liftM genericLength (value2string v)))
+evalPredef id [v1,v2]
+  | id == cTake   = return (fmap string2value (liftM2 genericTake (value2int v1) (value2string v2)))
+evalPredef id [v1,v2]
+  | id == cDrop   = return (fmap string2value (liftM2 genericDrop (value2int v1) (value2string v2)))
+evalPredef id [v1,v2]
+  | id == cTk     = return (fmap string2value (liftM2 genericTk (value2int v1) (value2string v2)))
+  where
+    genericTk n = reverse . genericTake n . reverse
+evalPredef id [v1,v2]
+  | id == cDp     = return (fmap string2value (liftM2 genericDp (value2int v1) (value2string v2)))
+  where
+    genericDp n = reverse . genericDrop n . reverse
+evalPredef id [v]
+  | id == cIsUpper= return (fmap toPBool (liftM (all isUpper) (value2string v)))
+evalPredef id [v]
+  | id == cToUpper= return (fmap string2value (liftM (map toUpper) (value2string v)))
+evalPredef id [v]
+  | id == cToLower= return (fmap string2value (liftM (map toLower) (value2string v)))
+evalPredef id [v1,v2]
+  | id == cEqStr  = return (fmap toPBool (liftM2 (==) (value2string v1) (value2string v2)))
+evalPredef id [v1,v2]
+  | id == cOccur  = return (fmap toPBool (liftM2 occur (value2string v1) (value2string v2)))
+evalPredef id [v1,v2]
+  | id == cOccurs  = return (fmap toPBool (liftM2 occurs (value2string v1) (value2string v2)))
+evalPredef id [v1,v2]
+  | id == cEqInt  = return (fmap toPBool (liftM2 (==) (value2int v1) (value2int v2)))
+evalPredef id [v1,v2]
+  | id == cLessInt= return (fmap toPBool (liftM2 (<) (value2int v1) (value2int v2)))
+evalPredef id [v1,v2]
+  | id == cPlus   = return (fmap VInt (liftM2 (+) (value2int v1) (value2int v2)))
+evalPredef id [v]
+  | id == cError  = case value2string v of
+                      Just msg -> fail msg
+                      Nothing  -> return Nothing
+evalPredef id vs  = return Nothing
+
+toPBool True  = VApp (cPredef,cPTrue)  []
+toPBool False = VApp (cPredef,cPFalse) []
+
+occur s1 []          = False
+occur s1 s2@(_:tail) = check s1 s2
+  where
+    check xs        []  = False
+    check []        ys  = True
+    check (x:xs) (y:ys)
+      | x == y          = check xs ys
+    check _      _      = occur s1 tail
+
+occurs cs s2 = any (\c -> elem c s2) cs
+
+update lbl v []              = [(lbl,v)]
+update lbl v (a@(lbl',_):as)
+  | lbl==lbl'                = (lbl,v) : as
+  | otherwise                = a : update lbl v as
+
+
+patternMatch v0 []                      = fail "No matching pattern found"
+patternMatch v0 ((env0,ps,args0,t):eqs) = match env0 ps eqs args0
+  where
+    match env []              eqs      args  = eval env t args
+    match env (PT ty p   :ps) eqs      args  = match env (p:ps) eqs args
+    match env (PAlt p1 p2:ps) eqs      args  = match env (p1:ps) ((env,p2:ps,args,t):eqs) args
+    match env (PM q      :ps) eqs      args  = do t <- getResDef q
+                                                  case t of
+                                                    EPatt _ _ p -> match env (p:ps) eqs args
+                                                    _ -> evalError $ hang "Expected pattern macro:" 4
+                                                                          (pp t)
+    match env (PV v      :ps) eqs (arg:args) = match ((v,arg):env) ps eqs args
+    match env (PAs v p   :ps) eqs (arg:args) = match ((v,arg):env) (p:ps) eqs (arg:args)
+    match env (PW        :ps) eqs (arg:args) = match env ps eqs args
+    match env (PTilde _  :ps) eqs (arg:args) = match env ps eqs args
+    match env (p         :ps) eqs (arg:args) = do
+      v <- force arg []
+      case (p,v) of
+        (p,       VMeta i envi vs  ) -> return (VSusp i envi vs (\tnk -> match env (p:ps) eqs (tnk:args)))
+        (p,       VGen  i vs       ) -> return v0
+        (p,       VSusp i envi vs k) -> return (VSusp i envi vs (\tnk -> match env (p:ps) eqs (tnk:args)))
+        (PP q qs, VApp r tnks)
+          | q == r        -> match env (qs++ps) eqs (tnks++args)
+        (PR pas, VR as)   -> matchRec env pas as ps eqs args
+        (PString s1, VStr s2)
+          | s1 == s2      -> match env ps eqs args
+        (PString s1, VC [])
+          | null s1       -> match env ps eqs args
+        (PSeq min1 max1 p1 min2 max2 p2,v)
+                          -> case value2string v of
+                               Just s  -> do let n  = length s
+                                                 lo = min1 `max` (n-fromMaybe n max2)
+                                                 hi = (n-min2) `min` fromMaybe n max1
+                                                 (ds,cs) = splitAt lo s
+                                             eqs <- matchStr env (p1:p2:ps) eqs (hi-lo) (reverse ds) cs args
+                                             patternMatch v0 eqs
+                               Nothing -> return v0
+        (PRep minp maxp p, v)
+                          -> case value2string v of
+                               Just s  -> do let n = length s `div` (max minp 1)
+                                             eqs <- matchRep env n minp maxp p minp maxp p ps ((env,PString []:ps,(arg:args),t) : eqs) (arg:args)
+                                             patternMatch v0 eqs
+                               Nothing -> return v0
+        (PChar, VStr [_]) -> match env ps eqs args
+        (PChars cs, VStr [c])
+          | elem c cs     -> match env ps eqs args
+        (PInt n, VInt m)
+          | n == m        -> match env ps eqs args
+        (PFloat n, VFlt m)
+          | n == m        -> match env ps eqs args
+        _                 -> patternMatch v0 eqs
+
+    matchRec env []            as ps eqs args = match env ps eqs args
+    matchRec env ((lbl,p):pas) as ps eqs args =
+      case lookup lbl as of
+        Just tnk -> matchRec env pas as (p:ps) eqs (tnk:args)
+        Nothing  -> evalError ("Missing value for label" <+> pp lbl)
+
+    matchStr env ps eqs i ds []     args = do
+      arg1 <- newEvaluatedThunk (string2value (reverse ds))
+      arg2 <- newEvaluatedThunk (string2value [])
+      return ((env,ps,arg1:arg2:args,t) : eqs)
+    matchStr env ps eqs 0 ds cs     args = do
+      arg1 <- newEvaluatedThunk (string2value (reverse ds))
+      arg2 <- newEvaluatedThunk (string2value cs)
+      return ((env,ps,arg1:arg2:args,t) : eqs)
+    matchStr env ps eqs i ds (c:cs) args = do
+      arg1 <- newEvaluatedThunk (string2value (reverse ds))
+      arg2 <- newEvaluatedThunk (string2value (c:cs))
+      eqs  <- matchStr env ps eqs (i-1 :: Int) (c:ds) cs args
+      return ((env,ps,arg1:arg2:args,t) : eqs)
+
+    matchRep env 0 minp maxp p minq maxq q ps eqs args = do
+      return eqs
+    matchRep env n minp maxp p minq maxq q ps eqs args = do
+      matchRep env (n-1) minp maxp p (minp+minq) (liftM2 (+) maxp maxq) (PSeq minp maxp p minq maxq q) ps ((env,q:ps,args,t) : eqs) args
+
+value2term i (VApp q tnks) =
+  foldM (\e1 tnk -> fmap (App e1) (force tnk [] >>= value2term i)) (QC q) tnks
+value2term i (VMeta m env tnks) = do
+  res <- zonk m tnks
+  case res of
+    Right i -> foldM (\e1 tnk -> fmap (App e1) (force tnk [] >>= value2term i)) (Meta i) tnks
+    Left  v -> value2term i v
+value2term i (VSusp j env vs k) = do
+  tnk <- newEvaluatedThunk (VGen maxBound vs)
+  v <- k tnk
+  value2term i v
+value2term i (VGen j tnks) =
+  foldM (\e1 tnk -> fmap (App e1) (force tnk [] >>= value2term i)) (Vr (identS ('v':show j))) tnks
+value2term i (VClosure env (Abs b x t)) = do
+  tnk <- newGen i
+  v <- eval ((x,tnk):env) t []
+  t <- value2term (i+1) v
+  return (Abs b (identS ('v':show i)) t)
+value2term i (VProd b x v1 env t2)
+  | x == identW = do t1 <- value2term i v1
+                     v2 <- eval env t2 []
+                     t2 <- value2term i v2
+                     return (Prod b x t1 t2)
+  | otherwise   = do t1 <- value2term i v1
+                     tnk <- newGen i
+                     v2 <- eval ((x,tnk):env) t2 []
+                     t2 <- value2term (i+1) v2
+                     return (Prod b (identS ('v':show i)) t1 t2)
+value2term i (VRecType lbls) = do
+  lbls <- mapM (\(lbl,v) -> fmap ((,) lbl) (value2term i v)) lbls
+  return (RecType lbls)
+value2term i (VR as) = do
+  as <- mapM (\(lbl,tnk) -> fmap (\t -> (lbl,(Nothing,t))) (force tnk [] >>= value2term i)) as
+  return (R as)
+value2term i (VP v lbl tnks) = do
+  t <- value2term i v
+  foldM (\e1 tnk -> fmap (App e1) (force tnk [] >>= value2term i)) (P t lbl) tnks
+value2term i (VExtR v1 v2) = do
+  t1 <- value2term i v1
+  t2 <- value2term i v2
+  return (ExtR t1 t2) 
+value2term i (VTable v1 v2) = do
+  t1 <- value2term i v1
+  t2 <- value2term i v2
+  return (Table t1 t2)
+value2term i (VT ti _ cs) = return (T ti cs)
+value2term i (VV ty tnks) = do ts <- mapM (\tnk -> force tnk [] >>= value2term i) tnks
+                               return (V ty ts)
+value2term i (VS v1 tnk2 tnks) = do t1 <- value2term i v1
+                                    t2 <- force tnk2 [] >>= value2term i
+                                    foldM (\e1 tnk -> fmap (App e1) (force tnk [] >>= value2term i)) (S t1 t2) tnks
+value2term i (VSort s) = return (Sort s)
+value2term i (VStr tok) = return (K tok)
+value2term i (VInt n) = return (EInt n)
+value2term i (VFlt n) = return (EFloat n)
+value2term i (VC vs) = do
+  ts <- mapM (value2term i) vs
+  case ts of
+    []     -> return Empty
+    (t:ts) -> return (foldl C t ts)
+value2term i (VGlue v1 v2) = do
+  t1 <- value2term i v1
+  t2 <- value2term i v2
+  return (Glue t1 t2)
+value2term i (VPatt min max p) = return (EPatt min max p)
+value2term i (VPattType v) = do t <- value2term i v
+                                return (EPattType t)
+value2term i (VAlts vd vas) = do
+  d <- value2term i vd
+  as <- forM vas $ \(vt,vs) -> do
+           t <- value2term i vt
+           s <- value2term i vs
+           return (t,s)
+  return (Alts d as)
+value2term i (VStrs vs) = do
+  ts <- mapM (value2term i) vs
+  return (Strs ts)
+
+value2string (VStr s) = Just s
+value2string (VC vs)  = fmap unwords (mapM value2string vs)
+value2string _        = Nothing
+
+string2value s =
+  case words s of
+    []  -> VC []
+    [w] -> VStr w
+    ws  -> VC (map VStr ws)
+
+value2int (VInt n) = Just n
+value2int _        = Nothing
+
+-----------------------------------------------------------------------
+-- * Evaluation monad
+
+type MetaThunks s = Map.Map MetaId (Thunk s)
+type Cont s r = MetaThunks s -> r -> ST s (CheckResult r)
+newtype EvalM s a = EvalM (forall r . Grammar -> (a -> Cont s r) -> Cont s r)
+
+instance Functor (EvalM s) where
+  fmap f (EvalM g) = EvalM (\gr k -> g gr (k . f))
+
+instance Applicative (EvalM s) where
+  pure x = EvalM (\gr k -> k x)
+  (EvalM f) <*> (EvalM x) = EvalM (\gr k -> f gr (\f -> x gr (\x -> k (f x))))
+
+instance Monad (EvalM s) where
+  (EvalM f) >>= g = EvalM (\gr k -> f gr (\x -> case g x of
+                                                  EvalM g -> g gr k))
+#if !(MIN_VERSION_base(4,13,0))
+  -- Monad(fail) will be removed in GHC 8.8+
+  fail = Fail.fail
+#endif
+
+instance Fail.MonadFail (EvalM s) where
+  fail msg = EvalM (\gr k _ r -> return (Fail (pp msg)))
+
+instance Alternative (EvalM s) where
+  empty = EvalM (\gr k _ r -> return (Success r))
+  (EvalM f) <|> (EvalM g) = EvalM $ \gr k mt r -> do
+     res <- f gr k mt r
+     case res of
+       Fail msg  -> return (Fail msg)
+       Success r -> g gr k mt r
+
+instance MonadPlus (EvalM s) where
+
+runEvalM :: Grammar -> (forall s . EvalM s a) -> Check [a]
+runEvalM gr f =
+  case runST (case f of
+                EvalM f -> f gr (\x mt xs -> return (Success (x:xs))) Map.empty []) of
+    Fail msg   -> checkError msg
+    Success xs -> return (reverse xs)
+
+evalError :: Doc -> EvalM s a
+evalError msg = EvalM (\gr k _ r -> return (Fail msg))
+  
+getResDef :: QIdent -> EvalM s Term
+getResDef q = EvalM $ \gr k mt r -> do
+  case lookupResDef gr q of
+    Ok t    -> k t mt r
+    Bad msg -> return (Fail (pp msg))
+
+getInfo :: QIdent -> EvalM s (ModuleName,Info)
+getInfo q = EvalM $ \gr k mt r -> do
+  case lookupOrigInfo gr q of
+    Ok res  -> k res mt r
+    Bad msg -> return (Fail (pp msg))
+
+getAllParamValues :: Type -> EvalM s [Term]
+getAllParamValues ty = EvalM $ \gr k mt r ->
+  case allParamValues gr ty of
+    Ok ts   -> k ts mt r
+    Bad msg -> return (Fail (pp msg))
+
+newThunk env t = EvalM $ \gr k mt r -> do
+ tnk <- newSTRef (Unevaluated env t)
+ k tnk mt r
+
+newEvaluatedThunk v = EvalM $ \gr k mt r -> do
+ tnk <- newSTRef (Evaluated v)
+ k tnk mt r
+
+newMeta mb_ty i = EvalM $ \gr k mt r ->
+  if i == 0
+    then do tnk <- newSTRef (Unbound mb_ty i)
+            k tnk mt r
+    else case Map.lookup i mt of
+           Just tnk -> k tnk mt r
+           Nothing  -> do tnk <- newSTRef (Unbound mb_ty i)
+                          k tnk (Map.insert i tnk mt) r
+
+getMeta tnk    = EvalM $ \gr k mt r -> readSTRef tnk >>= \st -> k st mt r
+
+setMeta tnk st = EvalM $ \gr k mt r -> do
+  old <- readSTRef tnk
+  writeSTRef tnk st
+  r <- k () mt r
+  writeSTRef tnk old
+  return r
+
+newGen i = EvalM $ \gr k mt r -> do
+ tnk <- newSTRef (Evaluated (VGen i []))
+ k tnk mt r
+
+force tnk vs = EvalM $ \gr k mt r -> do 
+  s <- readSTRef tnk
+  case s of
+    Unevaluated env t -> case eval env t vs of
+                           EvalM f -> f gr (\v mt r -> do writeSTRef tnk (Evaluated v)
+                                                          r <- k v mt r
+                                                          writeSTRef tnk s
+                                                          return r) mt r
+    Evaluated v       -> case apply v vs of
+                           EvalM f -> f gr k mt r
+    Unbound _ _       -> k (VMeta tnk [] vs) mt r
+
+zonk tnk vs = EvalM $ \gr k mt r -> do
+  s <- readSTRef tnk
+  case s of
+    Evaluated v -> case apply v vs of
+                     EvalM f -> f gr (k . Left) mt r
+    Unbound _ i -> k (Right i) mt r

src/compiler/GF/Compile/Compute/ConcreteNew.hs

@@ -1,588 +0,0 @@
--- | Functions for computing the values of terms in the concrete syntax, in
--- | preparation for PMCFG generation.
-module GF.Compile.Compute.ConcreteNew
-           (GlobalEnv, GLocation, resourceValues, geLoc, geGrammar,
-            normalForm,
-            Value(..), Bind(..), Env, value2term, eval, vapply
-           ) where
-import Prelude hiding ((<>)) -- GHC 8.4.1 clash with Text.PrettyPrint
-
-import GF.Grammar hiding (Env, VGen, VApp, VRecType)
-import GF.Grammar.Lookup(lookupResDefLoc,allParamValues)
-import GF.Grammar.Predef(cPredef,cErrorType,cTok,cStr,cTrace,cPBool)
-import GF.Grammar.PatternMatch(matchPattern,measurePatt)
-import GF.Grammar.Lockfield(isLockLabel,lockRecType) --unlockRecord,lockLabel
-import GF.Compile.Compute.Value hiding (Error)
-import GF.Compile.Compute.Predef(predef,predefName,delta)
-import GF.Data.Str(Str,glueStr,str2strings,str,sstr,plusStr,strTok)
-import GF.Data.Operations(Err,err,errIn,maybeErr,mapPairsM)
-import GF.Data.Utilities(mapFst,mapSnd)
-import GF.Infra.Option
-import Control.Monad(ap,liftM,liftM2) -- ,unless,mplus
-import Data.List (findIndex,intersect,nub,elemIndex,(\\)) --,isInfixOf
---import Data.Char (isUpper,toUpper,toLower)
-import GF.Text.Pretty
-import qualified Data.Map as Map
-import Debug.Trace(trace)
-
--- * Main entry points
-
-normalForm :: GlobalEnv -> L Ident -> Term -> Term
-normalForm (GE gr rv opts _) loc = err (bugloc loc) id . nfx (GE gr rv opts loc)
-
-nfx env@(GE _ _ _ loc) t = do
-  v <- eval env [] t
-  case value2term loc [] v of
-    Left i  -> fail ("variable #"++show i++" is out of scope")
-    Right t -> return t
-
-eval :: GlobalEnv -> Env -> Term -> Err Value
-eval (GE gr rvs opts loc) env t = ($ (map snd env)) # value cenv t
-  where
-    cenv = CE gr rvs opts loc (map fst env)
-
---apply env = apply' env
-
---------------------------------------------------------------------------------
-
--- * Environments
-
-type ResourceValues = Map.Map ModuleName (Map.Map Ident (Err Value))
-
-data GlobalEnv = GE Grammar ResourceValues Options GLocation
-data CompleteEnv = CE {srcgr::Grammar,rvs::ResourceValues,
-                       opts::Options,
-                       gloc::GLocation,local::LocalScope}
-type GLocation = L Ident
-type LocalScope = [Ident]
-type Stack = [Value]
-type OpenValue = Stack->Value
-
-geLoc     (GE _ _ _ loc) = loc
-geGrammar (GE gr _ _ _)  = gr
-
-ext b env = env{local=b:local env}
-extend bs env = env{local=bs++local env}
-global env = GE (srcgr env) (rvs env) (opts env) (gloc env)
-
-var :: CompleteEnv -> Ident -> Err OpenValue
-var env x = maybe unbound pick' (elemIndex x (local env))
-  where
-    unbound = fail ("Unknown variable: "++showIdent x)
-    pick' i = return $ \ vs -> maybe (err i vs) ok (pick i vs)
-    err i vs = bug $ "Stack problem: "++showIdent x++": "
-                    ++unwords (map showIdent (local env))
-                    ++" => "++show (i,length vs)
-    ok v = --trace ("var "++show x++" = "++show v) $
-           v
-
-pick :: Int -> Stack -> Maybe Value
-pick 0 (v:_) = Just v
-pick i (_:vs) = pick (i-1) vs
-pick i vs = Nothing -- bug $ "pick "++show (i,vs)
-
-resource env (m,c) =
---  err bug id $
-    if isPredefCat c
-    then value0 env =<< lockRecType c defLinType -- hmm
-    else maybe e id $ Map.lookup c =<< Map.lookup m (rvs env)
-  where e = fail $ "Not found: "++render m++"."++showIdent c
-
--- | Convert operators once, not every time they are looked up
-resourceValues :: Options -> SourceGrammar -> GlobalEnv
-resourceValues opts gr = env
-  where
-    env = GE gr rvs opts (L NoLoc identW)
-    rvs = Map.mapWithKey moduleResources (moduleMap gr)
-    moduleResources m = Map.mapWithKey (moduleResource m) . jments
-    moduleResource m c _info = do L l t <- lookupResDefLoc gr (m,c)
-                                  let loc = L l c
-                                      qloc = L l (Q (m,c))
-                                  eval (GE gr rvs opts loc) [] (traceRes qloc t)
-
-    traceRes = if flag optTrace opts
-               then traceResource
-               else const id
-
--- * Tracing
-
--- | Insert a call to the trace function under the top-level lambdas
-traceResource (L l q) t =
-  case termFormCnc t of
-    (abs,body) -> mkAbs abs (mkApp traceQ [args,body])
-      where
-        args = R $ tuple2record (K lstr:[Vr x|(bt,x)<-abs,bt==Explicit])
-        lstr = render (l<>":"<>ppTerm Qualified 0 q)
-        traceQ = Q (cPredef,cTrace)
-
--- * Computing values
-
--- | Computing the value of a top-level term
-value0 :: CompleteEnv -> Term -> Err Value
-value0 env = eval (global env) []
-
--- | Computing the value of a term
-value :: CompleteEnv -> Term -> Err OpenValue
-value env t0 =
-  -- Each terms is traversed only once by this function, using only statically
-  -- available information. Notably, the values of lambda bound variables 
-  -- will be unknown during the term traversal phase.
-  -- The result is an OpenValue, which is a function that may be applied many
-  -- times to different dynamic values, but without the term traversal overhead
-  -- and without recomputing other statically known information.
-  -- For this to work, there should be no recursive calls under lambdas here.
-  -- Whenever we need to construct the OpenValue function with an explicit
-  -- lambda, we have to lift the recursive calls outside the lambda.
-  -- (See e.g. the rules for Let, Prod and Abs)
-{-
-  trace (render $ text "value"<+>sep [ppL (gloc env)<>text ":",
-                                      brackets (fsep (map ppIdent (local env))),
-                                      ppTerm Unqualified 10 t0]) $
---}
-  errIn (render t0) $
-  case t0 of
-    Vr x   -> var env x
-    Q x@(m,f)
-      | m == cPredef -> if f==cErrorType                -- to be removed
-                        then let p = identS "P"
-                             in const # value0 env (mkProd [(Implicit,p,typeType)] (Vr p) [])
-                        else if f==cPBool
-                               then const # resource env x
-                               else const . flip VApp [] # predef f
-      | otherwise    -> const # resource env x --valueResDef (fst env) x
-    QC x   -> return $ const (VCApp x [])
-    App e1 e2 -> apply' env e1 . (:[]) =<< value env e2
-    Let (x,(oty,t)) body -> do vb <- value (ext x env) body
-                               vt <- value env t
-                               return $ \ vs -> vb (vt vs:vs)
-    Meta i -> return $ \ vs -> VMeta i (zip (local env) vs) []
-    Prod bt x t1 t2 ->
-       do vt1 <- value env t1
-          vt2 <- value (ext x env) t2
-          return $ \ vs -> VProd bt (vt1 vs) x $ Bind $ \ vx -> vt2 (vx:vs)
-    Abs bt x t -> do vt <- value (ext x env) t
-                     return $ VAbs bt x . Bind . \ vs vx -> vt (vx:vs)
-    EInt n -> return $ const (VInt n)
-    EFloat f -> return $ const (VFloat f)
-    K s -> return $ const (VString s)
-    Empty -> return $ const (VString "")
-    Sort s | s == cTok -> return $ const (VSort cStr)        -- to be removed 
-           | otherwise -> return $ const (VSort s)
-    ImplArg t -> (VImplArg.) # value env t
-    Table p res -> liftM2 VTblType # value env p <# value env res
-    RecType rs -> do lovs <- mapPairsM (value env) rs
-                     return $ \vs->VRecType $ mapSnd ($vs) lovs
-    t@(ExtR t1 t2) -> ((extR t.)# both id) # both (value env) (t1,t2)
-    FV ts   -> ((vfv .) # sequence) # mapM (value env) ts
-    R as    -> do lovs <- mapPairsM (value env.snd) as
-                  return $ \ vs->VRec $ mapSnd ($vs) lovs
-    T i cs  -> valueTable env i cs
-    V ty ts -> do pvs <- paramValues env ty
-                  ((VV ty pvs .) . sequence) # mapM (value env) ts
-    C t1 t2 -> ((ok2p vconcat.) # both id) # both (value env) (t1,t2)
-    S t1 t2 -> ((select env.) # both id) # both (value env) (t1,t2)
-    P t l   -> --maybe (bug $ "project "++show l++" from "++show v) id $
-               do ov <- value env t
-                  return $ \ vs -> let v = ov vs
-                                   in maybe (VP v l) id (proj l v)
-    Alts t tts -> (\v vts -> VAlts # v <# mapM (both id) vts) # value env t <# mapM (both (value env)) tts
-    Strs ts    -> ((VStrs.) # sequence) # mapM (value env) ts
-    Glue t1 t2 -> ((ok2p (glue env).) # both id) # both (value env) (t1,t2)
-    ELin c r -> (unlockVRec (gloc env) c.) # value env r
-    EPatt p  -> return $ const (VPatt p) -- hmm
-    EPattType ty -> do vt <- value env ty
-                       return (VPattType . vt)
-    Typed t ty -> value env t
-    t -> fail.render $ "value"<+>ppTerm Unqualified 10 t $$ show t
-
-vconcat vv@(v1,v2) =
-    case vv of
-      (VString "",_) -> v2
-      (_,VString "") -> v1
-      (VApp NonExist _,_) -> v1
-      (_,VApp NonExist _) -> v2
-      _ -> VC v1 v2
-
-proj l v | isLockLabel l = return (VRec [])
-                ---- a workaround 18/2/2005: take this away and find the reason
-                ---- why earlier compilation destroys the lock field
-proj l v =
-    case v of
-      VFV vs -> liftM vfv (mapM (proj l) vs)
-      VRec rs -> lookup l rs
---    VExtR v1 v2 -> proj l v2 `mplus` proj l v1 -- hmm
-      VS (VV pty pvs rs) v2 -> flip VS v2 . VV pty pvs # mapM (proj l) rs
-      _ -> return (ok1 VP v l)
-
-ok1 f v1@(VError {}) _ = v1
-ok1 f v1 v2 = f v1 v2
- 
-ok2 f v1@(VError {}) _ = v1
-ok2 f _ v2@(VError {}) = v2
-ok2 f v1 v2 = f v1 v2
-
-ok2p f (v1@VError {},_) = v1
-ok2p f (_,v2@VError {}) = v2
-ok2p f vv = f vv
-
-unlockVRec loc c0 v0 = v0
-{-
-unlockVRec loc c0 v0 = unlockVRec' c0 v0
-  where
-    unlockVRec' ::Ident -> Value -> Value
-    unlockVRec' c v =
-        case v of
-    --    VClosure env t -> err bug (VClosure env) (unlockRecord c t)
-          VAbs bt x (Bind f) -> VAbs bt x (Bind $ \ v -> unlockVRec' c (f v))
-          VRec rs        -> plusVRec rs lock
-    --    _              -> VExtR v (VRec lock) -- hmm
-          _              -> {-trace (render $ ppL loc $ "unlock non-record "++show v0)-} v -- hmm
-    --    _              -> bugloc loc $ "unlock non-record "++show v0
-      where
-        lock = [(lockLabel c,VRec [])]
--}
-
--- suspicious, but backwards compatible
-plusVRec rs1 rs2 = VRec ([(l,v)|(l,v)<-rs1,l `notElem` ls2] ++ rs2)
-  where ls2 = map fst rs2
-
-extR t vv =
-    case vv of
-      (VFV vs,v2) -> vfv [extR t (v1,v2)|v1<-vs]
-      (v1,VFV vs) -> vfv [extR t (v1,v2)|v2<-vs]
-      (VRecType rs1, VRecType rs2) ->
-          case intersect (map fst rs1) (map fst rs2) of
-            [] -> VRecType (rs1 ++ rs2)
-            ls -> error $ "clash"<+>show ls
-      (VRec     rs1, VRec     rs2) -> plusVRec rs1 rs2
-      (v1          , VRec [(l,_)]) | isLockLabel l -> v1 -- hmm
-      (VS (VV t pvs vs) s,v2) -> VS (VV t pvs [extR t (v1,v2)|v1<-vs]) s
---    (v1,v2) -> ok2 VExtR v1 v2 -- hmm
-      (v1,v2) -> error $ "not records" $$ show v1 $$ show v2
-  where
-    error explain = ppbug $ "The term" <+> t
-                            <+> "is not reducible" $$ explain
-
-glue env (v1,v2) = glu v1 v2
-  where
-    glu v1 v2 =
-        case (v1,v2) of
-          (VFV vs,v2) -> vfv [glu v1 v2|v1<-vs]
-          (v1,VFV vs) -> vfv [glu v1 v2|v2<-vs]
-          (VString s1,VString s2) -> VString (s1++s2)
-          (v1,VAlts d vs) -> VAlts (glx d) [(glx v,c) | (v,c) <- vs]
-             where glx v2 = glu v1 v2
-          (v1@(VAlts {}),v2) ->
-           --err (const (ok2 VGlue v1 v2)) id $
-             err bug id $
-             do y' <- strsFromValue v2
-                x' <- strsFromValue v1
-                return $ vfv [foldr1 VC (map VString (str2strings (glueStr v u))) | v <- x', u <- y']
-          (VC va vb,v2) -> VC va (glu vb v2)
-          (v1,VC va vb) -> VC (glu v1 va) vb
-          (VS (VV ty pvs vs) vb,v2) -> VS (VV ty pvs [glu v v2|v<-vs]) vb
-          (v1,VS (VV ty pvs vs) vb) -> VS (VV ty pvs [glu v1 v|v<-vs]) vb
-          (v1@(VApp NonExist _),_) -> v1
-          (_,v2@(VApp NonExist _)) -> v2
---        (v1,v2) -> ok2 VGlue v1 v2
-          (v1,v2) -> if flag optPlusAsBind (opts env)
-                       then VC v1 (VC (VApp BIND []) v2)
-                       else let loc  = gloc env
-                                vt v = case value2term loc (local env) v of
-                                         Left i  -> Error ('#':show i)
-                                         Right t -> t
-                                originalMsg = render $ ppL loc (hang "unsupported token gluing" 4
-                                                               (Glue (vt v1) (vt v2)))
-                                term = render $ pp  $ Glue (vt v1) (vt v2)
-                            in error $ unlines
-                                [originalMsg
-                                ,""
-                                ,"There was a problem in the expression `"++term++"`, either:"
-                                ,"1) You are trying to use + on runtime arguments, possibly via an oper."
-                                ,"2) One of the arguments in `"++term++"` is a bound variable from pattern matching a string, but the cases are non-exhaustive."
-                                ,"For more help see https://github.com/GrammaticalFramework/gf-core/tree/master/doc/errors/gluing.md"
-                                ]
-
-
--- | to get a string from a value that represents a sequence of terminals
-strsFromValue :: Value -> Err [Str]
-strsFromValue t = case t of
-  VString s   -> return [str s]
-  VC s t -> do
-    s' <- strsFromValue s
-    t' <- strsFromValue t
-    return [plusStr x y | x <- s', y <- t']
-{-
-  VGlue s t -> do
-    s' <- strsFromValue s
-    t' <- strsFromValue t
-    return [glueStr x y | x <- s', y <- t']
--}
-  VAlts d vs -> do
-    d0 <- strsFromValue d
-    v0 <- mapM (strsFromValue . fst) vs
-    c0 <- mapM (strsFromValue . snd) vs
-  --let vs' = zip v0 c0
-    return [strTok (str2strings def) vars | 
-              def  <- d0,
-              vars <- [[(str2strings v, map sstr c) | (v,c) <- zip vv c0] | 
-                                                          vv <- sequence v0]
-           ]
-  VFV ts -> concat # mapM strsFromValue ts
-  VStrs ts -> concat # mapM strsFromValue ts
-
-  _ -> fail ("cannot get Str from value " ++ show t)
-
-vfv vs = case nub vs of
-           [v] -> v
-           vs -> VFV vs
-
-select env vv =
-    case vv of
-      (v1,VFV vs) -> vfv [select env (v1,v2)|v2<-vs]
-      (VFV vs,v2) -> vfv [select env (v1,v2)|v1<-vs]
-      (v1@(VV pty vs rs),v2) ->
-         err (const (VS v1 v2)) id $
-         do --ats <- allParamValues (srcgr env) pty
-            --let vs = map (value0 env) ats
-            i <- maybeErr "no match" $ findIndex (==v2) vs
-            return (ix (gloc env) "select" rs i)
-      (VT _ _ [(PW,Bind b)],_) -> {-trace "eliminate wild card table" $-} b []
-      (v1@(VT _ _ cs),v2) ->
-                 err (\_->ok2 VS v1 v2) (err bug id . valueMatch env) $
-                 match (gloc env) cs v2
-      (VS (VV pty pvs rs) v12,v2) -> VS (VV pty pvs [select env (v11,v2)|v11<-rs]) v12
-      (v1,v2) -> ok2 VS v1 v2
-
-match loc cs v = 
-  case value2term loc [] v of
-    Left i  -> bad ("variable #"++show i++" is out of scope")
-    Right t -> err bad return (matchPattern cs t)
-  where
-    bad = fail . ("In pattern matching: "++)
-
-valueMatch :: CompleteEnv -> (Bind Env,Substitution) -> Err Value
-valueMatch env (Bind f,env') = f # mapPairsM (value0 env) env'
-
-valueTable :: CompleteEnv -> TInfo -> [Case] -> Err OpenValue
-valueTable env i cs =
-    case i of
-      TComp ty -> do pvs <- paramValues env ty
-                     ((VV ty pvs .) # sequence) # mapM (value env.snd) cs
-      _ -> do ty <- getTableType i
-              cs' <- mapM valueCase cs
-              err (dynamic cs' ty) return (convert cs' ty)
-  where
-    dynamic cs' ty _ = cases cs' # value env ty
-
-    cases cs' vty vs = err keep ($vs) (convertv cs' (vty vs))
-      where
-        keep msg = --trace (msg++"\n"++render (ppTerm Unqualified 0 (T i cs))) $
-                   VT wild (vty vs) (mapSnd ($vs) cs')
-
-    wild = case i of TWild _ -> True; _ -> False
-
-    convertv cs' vty = 
-      case value2term (gloc env) [] vty of
-        Left i    -> fail ("variable #"++show i++" is out of scope")
-        Right pty -> convert' cs' =<< paramValues'' env pty
-
-    convert cs' ty = convert' cs' =<< paramValues' env ty
-
-    convert' cs' ((pty,vs),pvs) =
-      do sts  <- mapM (matchPattern cs') vs
-         return $ \ vs -> VV pty pvs $ map (err bug id . valueMatch env)
-                                           (mapFst ($vs) sts)
-
-    valueCase (p,t) = do p' <- measurePatt # inlinePattMacro p
-                         pvs <- linPattVars p'
-                         vt <- value (extend pvs env) t
-                         return (p',\vs-> Bind $ \bs-> vt (push' p' bs pvs vs))
-
-    inlinePattMacro p =
-        case p of
-          PM qc -> do r <- resource env qc
-                      case r of
-                        VPatt p' -> inlinePattMacro p'
-                        _ -> ppbug $ hang "Expected pattern macro:" 4
-                                          (show r)
-          _ -> composPattOp inlinePattMacro p
-
-
-paramValues env ty = snd # paramValues' env ty
-
-paramValues' env ty = paramValues'' env =<< nfx (global env) ty
-
-paramValues'' env pty = do ats <- allParamValues (srcgr env) pty
-                           pvs <- mapM (eval (global env) []) ats
-                           return ((pty,ats),pvs)
-
-push' p bs xs = if length bs/=length xs
-                then bug $ "push "++show (p,bs,xs)
-                else push bs xs
-
-push :: Env -> LocalScope -> Stack -> Stack
-push bs [] vs = vs
-push bs (x:xs) vs = maybe err id (lookup x bs):push bs xs vs
-  where  err = bug $ "Unbound pattern variable "++showIdent x
-
-apply' :: CompleteEnv -> Term -> [OpenValue] -> Err OpenValue
-apply' env t [] = value env t
-apply' env t vs =
-  case t of
-    QC x                   -> return $ \ svs -> VCApp x (map ($svs) vs)
-{-
-    Q x@(m,f) | m==cPredef -> return $
-                              let constr = --trace ("predef "++show x) .
-                                           VApp x 
-                              in \ svs -> maybe constr id (Map.lookup f predefs)
-                                          $ map ($svs) vs
-              | otherwise  -> do r <- resource env x
-                                 return $ \ svs -> vapply (gloc env) r (map ($svs) vs)
--}
-    App t1 t2              -> apply' env t1 . (:vs) =<< value env t2
-    _                      -> do fv <- value env t
-                                 return $ \ svs -> vapply (gloc env) (fv svs) (map ($svs) vs)
-
-vapply :: GLocation -> Value -> [Value] -> Value
-vapply loc v [] = v
-vapply loc v vs =
-  case v of
-    VError {} -> v
---  VClosure env (Abs b x t) -> beta gr env b x t vs
-    VAbs bt _ (Bind f) -> vbeta loc bt f vs
-    VApp pre vs1 -> delta' pre (vs1++vs)
-      where
-        delta' Trace (v1:v2:vs) = let vr = vapply loc v2 vs
-                                  in vtrace loc v1 vr
-        delta' pre vs = err msg vfv $ mapM (delta pre) (varyList vs)
-        --msg = const (VApp pre (vs1++vs))
-        msg = bug . (("Applying Predef."++showIdent (predefName pre)++": ")++)
-    VS (VV t pvs fs) s -> VS (VV t pvs [vapply loc f vs|f<-fs]) s
-    VFV fs -> vfv [vapply loc f vs|f<-fs]
-    VCApp f vs0 -> VCApp f (vs0++vs)
-    VMeta i env vs0 -> VMeta i env (vs0++vs)
-    VGen i vs0 -> VGen i (vs0++vs)
-    v -> bug $ "vapply "++show v++" "++show vs
-
-vbeta loc bt f (v:vs) =
-  case (bt,v) of
-    (Implicit,VImplArg v) -> ap v
-    (Explicit,         v) -> ap v
-  where
-    ap (VFV avs) = vfv [vapply loc (f v) vs|v<-avs]
-    ap v         = vapply loc (f v) vs
-
-vary (VFV vs) = vs
-vary v = [v]
-varyList = mapM vary
-
-{-
-beta env b x t (v:vs) =
-  case (b,v) of
-    (Implicit,VImplArg v) -> apply' (ext (x,v) env) t vs
-    (Explicit,         v) -> apply' (ext (x,v) env) t vs
--}
-
-vtrace loc arg res = trace (render (hang (pv arg) 4 ("->"<+>pv res))) res
-  where
-    pv v = case v of
-             VRec (f:as) -> hang (pf f) 4 (fsep (map pa as))
-             _ -> ppV v
-    pf (_,VString n) = pp n
-    pf (_,v) = ppV v
-    pa (_,v) = ppV v
-    ppV v = case value2term' True loc [] v of
-              Left i  -> "variable #" <> pp i <+> "is out of scope"
-              Right t -> ppTerm Unqualified 10 t
-
--- | Convert a value back to a term
-value2term :: GLocation -> [Ident] -> Value -> Either Int Term
-value2term = value2term' False
-value2term' stop loc xs v0 =
-  case v0 of
-    VApp pre vs    -> liftM (foldl App (Q (cPredef,predefName pre))) (mapM v2t vs)
-    VCApp f vs     -> liftM  (foldl App (QC f))   (mapM v2t vs)
-    VGen j vs      -> liftM2 (foldl App) (var j)  (mapM v2t vs)
-    VMeta j env vs -> liftM  (foldl App (Meta j)) (mapM v2t vs)
-    VProd bt v x f -> liftM2 (Prod bt x) (v2t v) (v2t' x f)
-    VAbs  bt   x f -> liftM  (Abs  bt x)         (v2t' x f)
-    VInt n         -> return (EInt n)
-    VFloat f       -> return (EFloat f)
-    VString s      -> return (if null s then Empty else K s)
-    VSort s        -> return (Sort s)
-    VImplArg v     -> liftM  ImplArg (v2t v)
-    VTblType p res -> liftM2 Table (v2t p) (v2t res)
-    VRecType rs    -> liftM  RecType (mapM (\(l,v) -> fmap ((,) l) (v2t v)) rs)
-    VRec as        -> liftM  R       (mapM (\(l,v) -> v2t v >>= \t -> return (l,(Nothing,t))) as)
-    VV t _ vs      -> liftM  (V t)   (mapM v2t vs)
-    VT wild v cs   -> v2t v >>= \t -> liftM (T ((if wild then TWild else TTyped) t)) (mapM nfcase cs)
-    VFV vs         -> liftM  FV (mapM v2t vs)
-    VC v1 v2       -> liftM2 C (v2t v1) (v2t v2)
-    VS v1 v2       -> liftM2 S (v2t v1) (v2t v2)
-    VP v l         -> v2t v >>= \t -> return (P t l)
-    VPatt p        -> return (EPatt p)
-    VPattType v    -> v2t v >>= return . EPattType
-    VAlts v vvs    -> liftM2 Alts (v2t v) (mapM (\(x,y) -> liftM2 (,) (v2t x) (v2t y)) vvs)
-    VStrs vs       -> liftM  Strs (mapM v2t vs)
---  VGlue v1 v2    -> Glue (v2t v1) (v2t v2)
---  VExtR v1 v2    -> ExtR (v2t v1) (v2t v2)
-    VError err     -> return (Error err)
-    _              -> bug ("value2term "++show loc++" : "++show v0)
-  where
-    v2t = v2txs xs
-    v2txs = value2term' stop loc
-    v2t' x f = v2txs (x:xs) (bind f (gen xs))
-
-    var j
-      | j<length xs = Right (Vr (reverse xs !! j))
-      | otherwise   = Left  j
-
-
-    pushs xs e = foldr push e xs
-    push x (env,xs) = ((x,gen xs):env,x:xs)
-    gen xs = VGen (length xs) []
-
-    nfcase (p,f) = liftM ((,) p) (v2txs xs' (bind f env'))
-      where (env',xs') = pushs (pattVars p) ([],xs)
-
-    bind (Bind f) x = if stop
-                      then VSort (identS "...") -- hmm
-                      else f x
-
-
-linPattVars p =
-    if null dups
-    then return pvs
-    else fail.render $ hang "Pattern is not linear. All variable names on the left-hand side must be distinct." 4 (ppPatt Unqualified 0 p)
-  where
-    allpvs = allPattVars p
-    pvs = nub allpvs
-    dups = allpvs \\ pvs
-
-pattVars = nub . allPattVars
-allPattVars p =
-    case p of
-      PV i    -> [i]
-      PAs i p -> i:allPattVars p
-      _       -> collectPattOp allPattVars p
-
----
-ix loc fn xs i =
-    if i<n
-    then xs !! i
-    else bugloc loc $ "(!!): index too large in "++fn++", "++show i++"<"++show n
-  where n = length xs
-
-infixl 1 #,<# --,@@
-
-f # x = fmap f x
-mf <# mx  = ap mf mx
---m1 @@ m2 = (m1 =<<) . m2
-
-both f (x,y) = (,) # f x <# f y
-
-bugloc loc s = ppbug $ ppL loc s
-
-bug msg = ppbug msg
-ppbug doc = error $ render $ hang "Internal error in Compute.ConcreteNew:" 4 doc

src/compiler/GF/Compile/Compute/Predef.hs

@@ -1,168 +0,0 @@
--- | Implementations of predefined functions
-{-# LANGUAGE TypeSynonymInstances, FlexibleInstances #-}
-module GF.Compile.Compute.Predef(predef,predefName,delta) where
-
-import qualified Data.Map as Map
-import Data.Array(array,(!))
-import Data.List (isInfixOf)
-import Data.Char (isUpper,toLower,toUpper)
-import Control.Monad(ap)
-
-import GF.Data.Utilities (apBoth) --mapSnd
-
-import GF.Compile.Compute.Value
-import GF.Infra.Ident (Ident,showIdent) --,varX
-import GF.Data.Operations(Err) -- ,err
-import GF.Grammar.Predef
-
---------------------------------------------------------------------------------
-class Predef a where
-  toValue :: a -> Value
-  fromValue :: Value -> Err a
-
-instance Predef Int where
-  toValue = VInt
-  fromValue (VInt i) = return i
-  fromValue v = verror "Int" v
-
-instance Predef Bool where
-  toValue = boolV
-
-instance Predef String where
-  toValue = string
-  fromValue v = case norm v of
-                  VString s -> return s
-                  _ -> verror "String" v
-
-instance Predef Value where
-  toValue = id
-  fromValue = return
-
-instance Predef Predefined where
-  toValue p = VApp p []
-  fromValue v = case v of
-                  VApp p _ -> return p
-                  _        -> fail $ "Expected a predefined constant, got something else"
-
-{-
-instance (Predef a,Predef b) => Predef (a->b) where
-  toValue f = VAbs Explicit (varX 0) $ Bind $ err bug (toValue . f) . fromValue
--}
-verror t v =
-  case v of
-    VError e -> fail e
-    VGen {}  -> fail $ "Expected a static value of type "++t
-                       ++", got a dynamic value"
-    _ -> fail $ "Expected a value of type "++t++", got "++show v
-
---------------------------------------------------------------------------------
-
-predef f = maybe undef return (Map.lookup f predefs)
-  where
-    undef = fail $ "Unimplemented predfined operator: Predef."++showIdent f
-
-predefs :: Map.Map Ident Predefined
-predefs = Map.fromList predefList
-
-predefName pre = predefNames ! pre
-predefNames = array (minBound,maxBound) (map swap predefList)
-
-predefList =
-    [(cDrop,Drop),(cTake,Take),(cTk,Tk),(cDp,Dp),(cEqStr,EqStr),
-     (cOccur,Occur),(cOccurs,Occurs),(cToUpper,ToUpper),(cToLower,ToLower),
-     (cIsUpper,IsUpper),(cLength,Length),(cPlus,Plus),(cEqInt,EqInt),
-     (cLessInt,LessInt),
-     -- cShow, cRead, cMapStr, cEqVal
-     (cError,Error),(cTrace,Trace),
-     -- Canonical values:
-     (cPBool,PBool),(cPFalse,PFalse),(cPTrue,PTrue),(cInt,Int),(cFloat,Float),
-     (cInts,Ints),(cNonExist,NonExist)
-     ,(cBIND,BIND),(cSOFT_BIND,SOFT_BIND),(cSOFT_SPACE,SOFT_SPACE)
-     ,(cCAPIT,CAPIT),(cALL_CAPIT,ALL_CAPIT)]
-    --- add more functions!!!
-
-delta f vs =
-    case f of
-      Drop    -> fromNonExist vs NonExist (ap2 (drop::Int->String->String))
-      Take    -> fromNonExist vs NonExist (ap2 (take::Int->String->String))
-      Tk      -> fromNonExist vs NonExist (ap2 tk)
-      Dp      -> fromNonExist vs NonExist (ap2 dp)
-      EqStr   -> fromNonExist vs PFalse   (ap2 ((==)::String->String->Bool))
-      Occur   -> fromNonExist vs PFalse   (ap2 occur)
-      Occurs  -> fromNonExist vs PFalse   (ap2 occurs)
-      ToUpper -> fromNonExist vs NonExist (ap1 (map toUpper))
-      ToLower -> fromNonExist vs NonExist (ap1 (map toLower))
-      IsUpper -> fromNonExist vs PFalse   (ap1 (all' isUpper))
-      Length  -> fromNonExist vs (0::Int) (ap1 (length::String->Int))
-      Plus    -> ap2 ((+)::Int->Int->Int)
-      EqInt   -> ap2 ((==)::Int->Int->Bool)
-      LessInt -> ap2 ((<)::Int->Int->Bool)
-    {- -- | Show | Read | ToStr | MapStr | EqVal -}
-      Error   -> ap1 VError
-      Trace   -> ap2 vtrace
-      -- Canonical values:
-      PBool   -> canonical
-      Int     -> canonical
-      Float   -> canonical
-      Ints    -> canonical
-      PFalse  -> canonical
-      PTrue   -> canonical
-      NonExist-> canonical
-      BIND    -> canonical
-      SOFT_BIND->canonical
-      SOFT_SPACE->canonical
-      CAPIT   -> canonical
-      ALL_CAPIT->canonical
-  where
-    canonical = delay
-    delay = return (VApp f vs) -- wrong number of arguments
-
-    ap1 f = case vs of
-              [v1] -> (toValue . f) `fmap` fromValue v1
-              _ -> delay
-
-    ap2 f = case vs of
-             [v1,v2] -> toValue `fmap` (f `fmap` fromValue v1 `ap` fromValue v2)
-             _ -> delay
-
-    fromNonExist vs a b
-      | null [v | v@(VApp NonExist _) <- vs] = b
-      | otherwise                            = return (toValue a)
-
-    vtrace :: Value -> Value -> Value
-    vtrace x y = y -- tracing is implemented elsewhere
-
---  unimpl id = bug $ "unimplemented predefined function: "++showIdent id
---  problem id vs = bug $ "unexpected arguments: Predef."++showIdent id++" "++show vs
-
-    tk i s = take (max 0 (length s - i)) s :: String
-    dp i s = drop (max 0 (length s - i)) s :: String
-    occur s t = isInfixOf (s::String) (t::String)
-    occurs s t = any (`elem` (t::String)) (s::String)
-    all' = all :: (a->Bool) -> [a] -> Bool
-
-boolV b = VCApp (cPredef,if b then cPTrue else cPFalse) []
-
-norm v =
-  case v of
-    VC v1 v2 -> case apBoth norm (v1,v2) of
-                  (VString s1,VString s2) -> VString (s1++" "++s2)
-                  (v1,v2) -> VC v1 v2
-    _ -> v
-{-
-strict v = case v of
-             VError err -> Left err
-             _ -> Right v
--}
-string s = case words s of
-             [] -> VString ""
-             ss -> foldr1 VC (map VString ss)
-
----
-
-swap (x,y) = (y,x)
-{-
-bug msg = ppbug msg
-ppbug doc = error $ render $
-                    hang "Internal error in Compute.Predef:" 4 doc
--}

src/compiler/GF/Compile/Compute/Value.hs

@@ -1,56 +0,0 @@
-module GF.Compile.Compute.Value where
-import GF.Grammar.Grammar(Label,Type,MetaId,Patt,QIdent)
-import PGF.Internal(BindType)
-import GF.Infra.Ident(Ident)
-import Text.Show.Functions()
-import Data.Ix(Ix)
-
--- | Self-contained (not quite) representation of values
-data Value
-  = VApp Predefined [Value] -- from Q, always Predef.x, has a built-in value
-  | VCApp QIdent [Value] -- from QC, constructors
-  | VGen Int [Value] -- for lambda bound variables, possibly applied
-  | VMeta MetaId Env [Value]
--- -- | VClosure Env Term -- used in Typecheck.ConcreteNew
-  | VAbs BindType Ident Binding -- used in Compute.ConcreteNew
-  | VProd BindType Value Ident Binding -- used in Compute.ConcreteNew
-  | VInt Int
-  | VFloat Double
-  | VString String
-  | VSort Ident
-  | VImplArg Value
-  | VTblType Value Value
-  | VRecType [(Label,Value)]
-  | VRec [(Label,Value)]
-  | VV Type [Value] [Value] -- preserve type for conversion back to Term
-  | VT Wild Value [(Patt,Bind Env)]
-  | VC Value Value
-  | VS Value Value
-  | VP Value Label
-  | VPatt Patt
-  | VPattType Value
-  | VFV [Value]
-  | VAlts Value [(Value, Value)]
-  | VStrs [Value]
--- -- | VGlue Value Value -- hmm
--- --  | VExtR Value Value -- hmm
-  | VError String
-  deriving (Eq,Show)
-
-type Wild = Bool
-type Binding = Bind Value
-data Bind a = Bind (a->Value) deriving Show
-
-instance Eq (Bind a) where x==y = False
-
-type Env = [(Ident,Value)]
-
--- | Predefined functions
-data Predefined = Drop | Take | Tk | Dp | EqStr | Occur | Occurs | ToUpper
-                | ToLower | IsUpper | Length | Plus | EqInt | LessInt 
-             {- | Show | Read | ToStr | MapStr | EqVal -}
-                | Error | Trace
-                -- Canonical values below:
-                | PBool | PFalse | PTrue | Int | Float | Ints | NonExist 
-                | BIND | SOFT_BIND | SOFT_SPACE | CAPIT | ALL_CAPIT
-                deriving (Show,Eq,Ord,Ix,Bounded,Enum)

src/compiler/GF/Compile/ConcreteToHaskell.hs

@@ -1,5 +1,7 @@
 -- | Translate concrete syntax to Haskell
 module GF.Compile.ConcreteToHaskell(concretes2haskell,concrete2haskell) where
+
+import PGF2(Literal(..))
 import Data.List(isPrefixOf,sort,sortOn)
 import qualified Data.Map as M
 import qualified Data.Set as S
@@ -7,7 +9,7 @@ import GF.Text.Pretty
 --import GF.Grammar.Predef(cPredef,cInts)
 --import GF.Compile.Compute.Predef(predef)
 --import GF.Compile.Compute.Value(Predefined(..))
-import GF.Infra.Ident(Ident,identS,identW,prefixIdent)
+import GF.Infra.Ident(Ident,identC,identS,identW,prefixIdent,showRawIdent,rawIdentS)
 import GF.Infra.Option
 import GF.Haskell as H
 import GF.Grammar.Canonical as C
@@ -16,13 +18,13 @@ import Debug.Trace(trace)
 
 -- | Generate Haskell code for the all concrete syntaxes associated with
 -- the named abstract syntax in given the grammar.
-concretes2haskell opts absname gr =
-  [(filename,render80 $ concrete2haskell opts abstr cncmod)
-     | let Grammar abstr cncs = grammar2canonical opts absname gr,
-       cncmod<-cncs,
-       let ModId name = concName cncmod
-           filename = name ++ ".hs" :: FilePath
-  ]
+concretes2haskell opts absname gr = do
+  Grammar abstr cncs <- grammar2canonical opts absname gr
+  return [(filename,render80 $ concrete2haskell opts abstr cncmod)
+             | cncmod<-cncs,
+               let ModId name = concName cncmod
+                   filename = showRawIdent name ++ ".hs" :: FilePath
+         ]
 
 -- | Generate Haskell code for the given concrete module.
 -- The only options that make a difference are
@@ -53,7 +55,7 @@ concrete2haskell opts
     labels = S.difference (S.unions (map S.fromList recs)) common_labels
     common_records = S.fromList [[label_s]]
     common_labels = S.fromList [label_s]
-    label_s = LabelId "s"
+    label_s = LabelId (rawIdentS "s")
 
     signature (CatDef c _) = TypeSig lf (Fun abs (pure lin))
       where
@@ -69,7 +71,7 @@ concrete2haskell opts
       where
      --funcats = S.fromList [c | FunDef f (C.Type _ (TypeApp c _))<-funs]
        allcats = S.fromList [c | CatDef c _<-cats]
-      
+
     gId :: ToIdent i => i -> Ident
     gId = (if haskellOption opts HaskellNoPrefix then id else prefixIdent "G")
           . toIdent
@@ -116,7 +118,7 @@ concrete2haskell opts
               where (ls,ts) = unzip $ sortOn fst [(l,t)|RecordRow l t<-rs]
             StrType -> tcon0 (identS "Str")
             TableType pt lt -> Fun (ppT pt) (ppT lt)
---          TupleType lts -> 
+--          TupleType lts ->
 
     lincatDef (LincatDef c t) = tsyn0 (lincatName c) (convLinType t)
 
@@ -126,7 +128,7 @@ concrete2haskell opts
     linDefs = map eqn . sortOn fst . map linDef
       where eqn (cat,(f,(ps,rhs))) = (cat,Eqn (f,ps) rhs)
 
-    linDef (LinDef f xs rhs0) = 
+    linDef (LinDef f xs rhs0) =
         (cat,(linfunName cat,(lhs,rhs)))
       where
         lhs = [ConP (aId f) (map VarP abs_args)]
@@ -144,7 +146,7 @@ concrete2haskell opts
           where
             vs = [(VarValueId (Unqual x),a)|(VarId x,a)<-zip xs args]
             env= [(VarValueId (Unqual x),lc)|(VarId x,lc)<-zip xs (map arglincat absctx)]
-            
+
         letlin a (TypeBinding _ (C.Type _ (TypeApp acat _))) =
           (a,Ap (Var (linfunName acat)) (Var (abs_arg a)))
 
@@ -181,13 +183,13 @@ concrete2haskell opts
 
         ppL l =
           case l of
-            FloatConstant x -> pure (lit x)
-            IntConstant n -> pure (lit n)
-            StrConstant s -> pure (token s)
+            LFlt x -> pure (lit x)
+            LInt n -> pure (lit n)
+            LStr s -> pure (token s)
 
         pId p@(ParamId s) =
           if "to_R_" `isPrefixOf` unqual s then toIdent p else gId p -- !! a hack
-          
+
         table cs =
             if all (null.patVars) ps
             then lets ds (LambdaCase [(ppP p,t')|(p,t')<-zip ps ts'])
@@ -315,13 +317,13 @@ instance Records rhs => Records (TableRow rhs) where
 
 -- | Record subtyping is converted into explicit coercions in Haskell
 coerce env ty t =
-  case (ty,t) of 
+  case (ty,t) of
     (_,VariantValue ts) -> VariantValue (map (coerce env ty) ts)
     (TableType ti tv,TableValue _ cs) ->
       TableValue ti [TableRow p (coerce env tv t)|TableRow p t<-cs]
     (RecordType rt,RecordValue r) ->
       RecordValue [RecordRow l (coerce env ft f) |
-                     RecordRow l f<-r,ft<-[ft|RecordRow l' ft<-rt,l'==l]]
+                     RecordRow l f<-r,ft<-[ft | RecordRow l' ft <- rt, l'==l]]
     (RecordType rt,VarValue x)->
       case lookup x env of
         Just ty' | ty'/=ty -> -- better to compare to normal form of ty'
@@ -334,18 +336,17 @@ coerce env ty t =
     _ -> t
   where
     app f ts = ParamConstant (Param f ts) -- !! a hack
-    to_rcon = ParamId . Unqual . to_rcon' . labels
+    to_rcon = ParamId . Unqual . rawIdentS . to_rcon' . labels
 
 patVars p = []
 
-labels r = [l|RecordRow l _<-r]
+labels r = [l | RecordRow l _ <- r]
 
 proj = Var . identS . proj'
-proj' (LabelId l) = "proj_"++l
+proj' (LabelId l) = "proj_" ++ showRawIdent l
 rcon = Var . rcon'
 rcon' = identS . rcon_name
-rcon_name ls = "R"++concat (sort ['_':l|LabelId l<-ls])
-
+rcon_name ls = "R"++concat (sort ['_':showRawIdent l | LabelId l <- ls])
 to_rcon' = ("to_"++) . rcon_name
 
 recordType ls =
@@ -400,17 +401,17 @@ linfunName c = prefixIdent "lin" (toIdent c)
 
 class ToIdent i where toIdent :: i -> Ident
 
-instance ToIdent ParamId where toIdent (ParamId q) = qIdentS q
-instance ToIdent PredefId where toIdent (PredefId s) = identS s
-instance ToIdent CatId   where toIdent (CatId s) = identS s
-instance ToIdent C.FunId where toIdent (FunId s) = identS s
-instance ToIdent VarValueId where toIdent (VarValueId q) = qIdentS q
+instance ToIdent ParamId where toIdent (ParamId q) = qIdentC q
+instance ToIdent PredefId where toIdent (PredefId s) = identC s
+instance ToIdent CatId   where toIdent (CatId s) = identC s
+instance ToIdent C.FunId where toIdent (FunId s) = identC s
+instance ToIdent VarValueId where toIdent (VarValueId q) = qIdentC q
 
-qIdentS = identS . unqual
+qIdentC = identS . unqual
 
-unqual (Qual (ModId m) n) = m++"_"++n
-unqual (Unqual n) = n
+unqual (Qual (ModId m) n) = showRawIdent m++"_"++ showRawIdent n
+unqual (Unqual n) = showRawIdent n
 
 instance ToIdent VarId where
   toIdent Anonymous = identW
-  toIdent (VarId s) = identS s
+  toIdent (VarId s) = identC s

src/compiler/GF/Compile/ExampleBased.hs

@@ -3,11 +3,7 @@ module GF.Compile.ExampleBased (
   configureExBased
   ) where
 
-import PGF
---import PGF.Probabilistic
---import PGF.Morphology
---import GF.Compile.ToAPI
-
+import PGF2
 import Data.List
 
 parseExamplesInGrammar :: ExConfiguration -> FilePath -> IO (FilePath,[String])
@@ -37,47 +33,38 @@ convertFile conf src file = do
       (ex,   end) = break (=='"') (tail exend)
     in ((unwords (words cat),ex), tail end)  -- quotes ignored
   pgf = resource_pgf conf
-  morpho = resource_morpho conf
   lang = language conf 
   convEx (cat,ex) = do
     appn "("
     let typ = maybe (error "no valid cat") id $ readType cat
-    ws <- case fst (parse_ pgf lang typ (Just 4) ex) of
-      ParseFailed _ -> do
-        let ws = morphoMissing morpho (words ex)
+    ws <- case parse lang typ ex of
+      ParseFailed _ _ -> do
         appv ("WARNING: cannot parse example " ++ ex) 
-        case ws of
-          [] -> return ()
-          _  -> appv ("  missing words: " ++ unwords ws)
-        return ws
-      TypeError _ ->
         return []
       ParseIncomplete ->
         return []
       ParseOk ts ->
-        case rank ts of
+        case ts of
           (t:tt) -> do
             if null tt 
               then return ()
               else appv ("WARNING: ambiguous example " ++ ex) 
-            appn t 
-            mapM_ (appn . ("  --- " ++)) tt
+            appn (printExp conf (fst t))
+            mapM_ (appn . ("  --- " ++) . printExp conf . fst) tt
             appn ")" 
             return [] 
     return ws
-  rank ts = [printExp conf t ++ "  -- " ++ show p | (t,p) <- rankTreesByProbs pgf ts]
   appf = appendFile file
   appn s = appf s >> appf "\n"
   appv s = appn ("--- " ++ s) >> putStrLn s
 
 data ExConfiguration = ExConf {
-  resource_pgf    :: PGF,
-  resource_morpho :: Morpho,
+  resource_pgf :: PGF,
   verbose  :: Bool,
-  language :: Language,
-  printExp :: Tree -> String
+  language :: Concr,
+  printExp :: Expr -> String
   }
 
-configureExBased :: PGF -> Morpho -> Language -> (Tree -> String) -> ExConfiguration
-configureExBased pgf morpho lang pr = ExConf pgf morpho False lang pr
+configureExBased :: PGF -> Concr -> (Expr -> String) -> ExConfiguration
+configureExBased pgf concr pr = ExConf pgf False concr pr
 

src/compiler/GF/Compile/Export.hs

@@ -1,14 +1,10 @@
 module GF.Compile.Export where
 
-import PGF
-import PGF.Internal(ppPGF)
+import PGF2
 import GF.Compile.PGFtoHaskell
 --import GF.Compile.PGFtoAbstract
 import GF.Compile.PGFtoJava
-import GF.Compile.PGFtoProlog
-import GF.Compile.PGFtoJS
 import GF.Compile.PGFtoJSON
-import GF.Compile.PGFtoPython
 import GF.Infra.Option
 --import GF.Speech.CFG
 import GF.Speech.PGFToCFG
@@ -22,6 +18,7 @@ import GF.Speech.SLF
 import GF.Speech.PrRegExp
 
 import Data.Maybe
+import qualified Data.Map as Map
 import System.FilePath
 import GF.Text.Pretty
 
@@ -35,15 +32,12 @@ exportPGF :: Options
           -> [(FilePath,String)] -- ^ List of recommended file names and contents.
 exportPGF opts fmt pgf = 
     case fmt of
-      FmtPGFPretty    -> multi "txt" (render . ppPGF)
+      FmtPGFPretty    -> multi "txt" (showPGF)
       FmtCanonicalGF  -> [] -- canon "gf" (render80 . abstract2canonical)
       FmtCanonicalJson-> []
-      FmtJavaScript   -> multi "js"  pgf2js
       FmtJSON         -> multi "json"  pgf2json
-      FmtPython       -> multi "py"  pgf2python
       FmtHaskell      -> multi "hs"  (grammar2haskell opts name)
       FmtJava         -> multi "java" (grammar2java opts name)
-      FmtProlog       -> multi "pl"  grammar2prolog
       FmtBNF          -> single "bnf"   bnfPrinter
       FmtEBNF         -> single "ebnf"  (ebnfPrinter opts)
       FmtSRGS_XML     -> single "grxml" (srgsXmlPrinter opts)
@@ -57,20 +51,13 @@ exportPGF opts fmt pgf =
       FmtRegExp       -> single "rexp" regexpPrinter
       FmtFA           -> single "dot"  slfGraphvizPrinter
  where
-   name = fromMaybe (showCId (abstractName pgf)) (flag optName opts)
+   name = fromMaybe (abstractName pgf) (flag optName opts)
 
    multi :: String -> (PGF -> String) -> [(FilePath,String)]
    multi ext pr = [(name <.> ext, pr pgf)]
 
 -- canon ext pr = [("canonical"</>name<.>ext,pr pgf)]
 
-   single :: String -> (PGF -> CId -> String) -> [(FilePath,String)]
-   single ext pr = [(showCId cnc <.> ext, pr pgf cnc) | cnc <- languages pgf]
+   single :: String -> (PGF -> Concr -> String) -> [(FilePath,String)]
+   single ext pr = [(concreteName cnc <.> ext, pr pgf cnc) | cnc <- Map.elems (languages pgf)]
 
-
--- | Get the name of the concrete syntax to generate output from.
--- FIXME: there should be an option to change this.
-outputConcr :: PGF -> CId
-outputConcr pgf = case languages pgf of
-                    []     -> error "No concrete syntax."
-                    cnc:_  -> cnc

src/compiler/GF/Compile/GenerateBC.hs

@@ -1,10 +1,11 @@
+{-# LANGUAGE CPP #-}
 module GF.Compile.GenerateBC(generateByteCode) where
 
 import GF.Grammar
 import GF.Grammar.Lookup(lookupAbsDef,lookupFunType)
 import GF.Data.Operations
-import PGF(CId,utf8CId)
-import PGF.Internal(CodeLabel,Instr(..),IVal(..),TailInfo(..),Literal(..))
+import PGF2(Literal(..))
+import PGF2.Internal(CodeLabel,Instr(..),IVal(..),TailInfo(..))
 import qualified Data.Map as Map
 import Data.List(nub,mapAccumL)
 import Data.Maybe(fromMaybe)
@@ -63,7 +64,7 @@ compileEquations gr arity st (i:is) eqs       fl bs = whilePP eqs Map.empty
 
     case_instr t =
       case t of
-        (Q (_,id)) -> CASE (i2i id)
+        (Q (_,id)) -> CASE (showIdent id)
         (EInt n)   -> CASE_LIT (LInt n)
         (K s)      -> CASE_LIT (LStr s)
         (EFloat d) -> CASE_LIT (LFlt d)
@@ -105,7 +106,7 @@ compileFun gr eval st vs (App e1 e2) h0 bs args =
 compileFun gr eval st vs (Q (m,id))  h0 bs args =
   case lookupAbsDef gr m id of
     Ok (_,Just _)
-       -> (h0,bs,eval st (GLOBAL (i2i id)) args)
+       -> (h0,bs,eval st (GLOBAL (showIdent id)) args)
     _  -> let Ok ty = lookupFunType gr m id
               (ctxt,_,_) = typeForm ty
               c_arity    = length ctxt
@@ -114,14 +115,14 @@ compileFun gr eval st vs (Q (m,id))  h0 bs args =
               diff   = c_arity-n_args
           in if diff <= 0
                then if n_args == 0
-                      then (h0,bs,eval st (GLOBAL (i2i id)) [])
+                      then (h0,bs,eval st (GLOBAL (showIdent id)) [])
                       else let h1  = h0 + 2 + n_args
-                           in (h1,bs,PUT_CONSTR (i2i id):is1++eval st (HEAP h0) [])
+                           in (h1,bs,PUT_CONSTR (showIdent id):is1++eval st (HEAP h0) [])
                else let h1  = h0 + 1 + n_args
                         is2 = [SET (FREE_VAR i) | i <- [0..n_args-1]] ++ [SET (ARG_VAR (i+1)) | i <- [0..diff-1]]
                         b   = CHECK_ARGS diff :
                               ALLOC (c_arity+2) : 
-                              PUT_CONSTR (i2i id) : 
+                              PUT_CONSTR (showIdent id) : 
                               is2 ++
                               TUCK (ARG_VAR 0) diff :
                               EVAL (HEAP h0) (TailCall diff) :
@@ -167,16 +168,16 @@ compileFun gr eval st vs e _ _ _ = error (show e)
 
 compileArg gr st vs (Q(m,id)) h0 bs =
   case lookupAbsDef gr m id of
-    Ok (_,Just _) -> (h0,bs,GLOBAL (i2i id),[])
+    Ok (_,Just _) -> (h0,bs,GLOBAL (showIdent id),[])
     _             -> let Ok ty = lookupFunType gr m id
                          (ctxt,_,_) = typeForm ty
                          c_arity    = length ctxt
                      in if c_arity == 0
-                          then (h0,bs,GLOBAL (i2i id),[])
+                          then (h0,bs,GLOBAL (showIdent id),[])
                           else let is2 = [SET (ARG_VAR (i+1)) | i <- [0..c_arity-1]]
                                    b   = CHECK_ARGS c_arity :
                                          ALLOC (c_arity+2) :
-                                         PUT_CONSTR (i2i id) :
+                                         PUT_CONSTR (showIdent id) :
                                          is2 ++
                                          TUCK (ARG_VAR 0) c_arity :
                                          EVAL (HEAP h0) (TailCall c_arity) :
@@ -224,12 +225,12 @@ compileArg gr st vs e           h0 bs =
                 diff   = c_arity-n_args
             in if diff <= 0
                  then let h2  = h1 + 2 + n_args
-                      in (h2,bs1,HEAP h1,is1 ++ (PUT_CONSTR (i2i id) : is2))
+                      in (h2,bs1,HEAP h1,is1 ++ (PUT_CONSTR (showIdent id) : is2))
                  else let h2  = h1 + 1 + n_args
                           is2 = [SET (FREE_VAR i) | i <- [0..n_args-1]] ++ [SET (ARG_VAR (i+1)) | i <- [0..diff-1]]
                           b   = CHECK_ARGS diff :
                                 ALLOC (c_arity+2) :
-                                PUT_CONSTR (i2i id) :
+                                PUT_CONSTR (showIdent id) :
                                 is2 ++
                                 TUCK (ARG_VAR 0) diff :
                                 EVAL (HEAP h0) (TailCall diff) :
@@ -298,9 +299,6 @@ freeVars xs (Vr x)
   | not (elem x xs)     = [x]
 freeVars xs e           = collectOp (freeVars xs) e
 
-i2i :: Ident -> CId
-i2i = utf8CId . ident2utf8
-
 push_is :: Int -> Int -> [IVal] -> [IVal]
 push_is i 0 is = is
 push_is i n is = ARG_VAR i : push_is (i-1) (n-1) is

src/compiler/GF/Compile/GeneratePMCFG.hs

@@ -10,631 +10,173 @@
 -----------------------------------------------------------------------------
 
 module GF.Compile.GeneratePMCFG
-    (generatePMCFG, pgfCncCat, addPMCFG, resourceValues
+    (generatePMCFG, pgfCncCat, addPMCFG
     ) where
 
---import PGF.CId
-import PGF.Internal as PGF(CncCat(..),Symbol(..),fidVar)
-
-import GF.Infra.Option
-import GF.Grammar hiding (Env, mkRecord, mkTable)
-import GF.Grammar.Lookup
+import GF.Grammar hiding (VApp)
 import GF.Grammar.Predef
-import GF.Grammar.Lockfield (isLockLabel)
-import GF.Data.BacktrackM
-import GF.Data.Operations
-import GF.Infra.UseIO (ePutStr,ePutStrLn) -- IOE,
-import GF.Data.Utilities (updateNthM) --updateNth
-import GF.Compile.Compute.ConcreteNew(normalForm,resourceValues)
-import qualified Data.Map as Map
-import qualified Data.Set as Set
-import qualified Data.List as List
---import qualified Data.IntMap as IntMap
-import qualified Data.IntSet as IntSet
+import GF.Grammar.Lookup
+import GF.Infra.CheckM
+import GF.Infra.Option
 import GF.Text.Pretty
-import Data.Array.IArray
-import Data.Array.Unboxed
---import Data.Maybe
---import Data.Char (isDigit)
-import Control.Applicative(Applicative(..))
+import GF.Compile.Compute.Concrete
+import GF.Data.Operations(Err(..))
+import PGF2.Transactions
+import qualified Data.Map.Strict as Map
 import Control.Monad
-import Control.Monad.Identity
---import Control.Exception
---import Debug.Trace(trace)
-import qualified Control.Monad.Fail as Fail
+import Data.List(mapAccumL)
 
-----------------------------------------------------------------------
--- main conversion function
+generatePMCFG :: Options -> FilePath -> SourceGrammar -> SourceModule -> Check SourceModule
+generatePMCFG opts cwd gr cmo@(cm,cmi) = do
+  let gr' = prependModule gr cmo
+  js <- mapM (addPMCFG opts cwd gr' cmi) (Map.toList (jments cmi))
+  return (cm,cmi{jments = (Map.fromAscList js)})
 
---generatePMCFG :: Options -> SourceGrammar -> Maybe FilePath -> SourceModule -> IOE SourceModule
-generatePMCFG opts sgr opath cmo@(cm,cmi) = do
-  (seqs,js) <- mapAccumWithKeyM (addPMCFG opts gr cenv opath am cm) Map.empty (jments cmi)
-  when (verbAtLeast opts Verbose) $ ePutStrLn ""
-  return (cm,cmi{mseqs = Just (mkSetArray seqs), jments = js})
+addPMCFG opts cwd gr cmi (id,CncFun mty@(Just (_,cat,ctxt,val)) mlin@(Just (L loc term)) mprn Nothing) =
+  checkInModule cwd cmi loc ("Happened in the PMCFG generation for" <+> id) $ do
+    rules <- pmcfgForm gr term ctxt val
+    return (id,CncFun mty mlin mprn (Just rules))
+addPMCFG opts cwd gr cmi id_info = return id_info
+
+pmcfgForm :: Grammar -> Term -> Context -> Type -> Check [PMCFGRule]
+pmcfgForm gr t ctxt ty =
+  runEvalM gr $ do
+    ((_,ms),args) <- mapAccumM (\(d,ms) (_,_,ty) -> do
+                                   let (ms',_,t) = type2metaTerm gr d ms 0 [] ty
+                                   tnk <- newThunk [] t
+                                   return ((d+1,ms'),tnk))
+                               (0,Map.empty) ctxt
+    sequence_ [newMeta (Just ty) i | (i,ty) <- Map.toList ms]
+    v <- eval [] t args
+    (lins,params) <- flatten v ty ([],[])
+    lins <- mapM str2lin lins
+    (r,rs,_) <- compute params
+    args <- zipWithM tnk2pmcfgcat args ctxt
+    return (PMCFGRule (PMCFGCat r rs) args (reverse lins))
+    where
+      tnk2pmcfgcat tnk (_,_,ty) = do
+        v <- force tnk []
+        (_,params) <- flatten v ty ([],[])
+        (r,rs,_) <- compute params
+        return (PMCFGCat r rs)
+
+      compute []     = return (0,[],1)
+      compute (v:vs) = do
+        (r, rs ,cnt ) <- param2int v
+        (r',rs',cnt') <- compute vs
+        return (r*cnt'+r',combine cnt' rs rs',cnt*cnt')
+
+type2metaTerm :: SourceGrammar -> Int -> Map.Map MetaId Type -> LIndex -> [(LIndex,Ident)] -> Type -> (Map.Map MetaId Type,Int,Term)
+type2metaTerm gr d ms r rs (Sort s) | s == cStr =
+  (ms,r+1,TSymCat d r rs)
+type2metaTerm gr d ms r rs (RecType lbls) =
+  let ((ms',r'),ass) = mapAccumL (\(ms,r) (lbl,ty) -> let (ms',r',t) = type2metaTerm gr d ms r rs ty
+                                                      in ((ms',r'),(lbl,(Just ty,t))))
+                                 (ms,r) lbls
+  in (ms',r',R ass)  
+type2metaTerm gr d ms r rs (Table p q) =
+  let pv = identS ('p':show (length rs))
+      (ms',r',t) = type2metaTerm gr d ms r ((r'-r,pv):rs) q
+      count = case allParamValues gr p of
+                Ok ts   -> length ts
+                Bad msg -> error msg
+  in (ms',(r'-r)*count,T (TTyped p) [(PV pv,t)])
+type2metaTerm gr d ms r rs ty@(QC q) =
+  let i = Map.size ms + 1
+  in (Map.insert i ty ms,r,Meta i)
+
+
+flatten (VSusp tnk env vs k) ty st = do
+  tnk_st <- getMeta tnk
+  case tnk_st of
+    Evaluated v             -> do v <- apply v vs
+                                  flatten v ty st
+    Unbound (Just (QC q)) _ -> do (m,ResParam (Just (L _ ps)) _) <- getInfo q
+                                  msum [bind tnk m p | p <- ps]
+                                  v <- k tnk
+                                  flatten v ty st
   where
-    cenv = resourceValues opts gr
-    gr = prependModule sgr cmo
-    MTConcrete am = mtype cmi
-
-mapAccumWithKeyM :: (Monad m, Ord k) => (a -> k -> b -> m (a,c)) -> a
-                                     -> Map.Map k b -> m (a,Map.Map k c)
-mapAccumWithKeyM f a m = do let xs = Map.toAscList m
-                            (a,ys) <- mapAccumM f a xs
-                            return (a,Map.fromAscList ys)
+    bind tnk m (p, ctxt) = do
+      tnks <- mapM (\(_,_,ty) -> newMeta (Just ty) 0) ctxt
+      setMeta tnk (Evaluated (VApp (m,p) tnks))
+flatten (VR as) (RecType lbls) st = do
+  foldM collect st lbls
   where
-    mapAccumM f a []          = return (a,[])
-    mapAccumM f a ((k,x):kxs) = do (a,y ) <- f a k x
-                                   (a,kys) <- mapAccumM f a kxs
-                                   return (a,(k,y):kys)
-
-
---addPMCFG :: Options -> SourceGrammar -> GlobalEnv -> Maybe FilePath -> Ident -> Ident -> SeqSet -> Ident -> Info -> IOE (SeqSet, Info)
-addPMCFG opts gr cenv opath am cm seqs id (GF.Grammar.CncFun mty@(Just (cat,cont,val)) mlin@(Just (L loc term)) mprn Nothing) = do
---when (verbAtLeast opts Verbose) $ ePutStr ("\n+ "++showIdent id++" ...")
-  let pres  = protoFCat gr res val
-      pargs = [protoFCat gr (snd $ catSkeleton ty) lincat | ((_,_,ty),(_,_,lincat)) <- zip ctxt cont]
-
-      pmcfgEnv0   = emptyPMCFGEnv
-  b             <- convert opts gr cenv (floc opath loc id) term (cont,val) pargs
-  let (seqs1,b1) = addSequencesB seqs b
-      pmcfgEnv1  = foldBM addRule
-                          pmcfgEnv0
-                          (goB b1 CNil [])
-                          (pres,pargs)
-      pmcfg      = getPMCFG pmcfgEnv1
-      
-      stats      = let PMCFG prods funs = pmcfg
-                       (s,e)            = bounds funs
-                       !prods_cnt       = length prods
-                       !funs_cnt        = e-s+1
-                   in (prods_cnt,funs_cnt)
-
-  when (verbAtLeast opts Verbose) $
-    ePutStr ("\n+ "++showIdent id++" "++show (product (map catFactor pargs)))
-  seqs1 `seq` stats `seq` return ()
-  when (verbAtLeast opts Verbose) $ ePutStr (" "++show stats)
-  return (seqs1,GF.Grammar.CncFun mty mlin mprn (Just pmcfg))
+    collect st (lbl,ty) =
+      case lookup lbl as of
+        Just tnk -> do v <- force tnk []
+                       flatten v ty st
+        Nothing  -> evalError ("Missing value for label" <+> pp lbl $$
+                               "among" <+> hsep (punctuate (pp ',') (map fst as)))
+flatten v@(VT _ env cs) (Table p q) st = do
+  ts <- getAllParamValues p
+  foldM collect st ts
   where
-    (ctxt,res,_) = err bug typeForm (lookupFunType gr am id)
-
-    addRule lins (newCat', newArgs') env0 =
-      let [newCat] = getFIds newCat'
-          !fun     = mkArray lins
-          newArgs  = map getFIds newArgs'
-      in addFunction env0 newCat fun newArgs
-
-addPMCFG opts gr cenv opath am cm seqs id (GF.Grammar.CncCat mty@(Just (L _ lincat)) 
-                                                             mdef@(Just (L loc1 def))
-                                                             mref@(Just (L loc2 ref))
-                                                             mprn
-                                                             Nothing) = do
-  let pcat = protoFCat gr (am,id) lincat
-      pvar = protoFCat gr (MN identW,cVar) typeStr
-
-      pmcfgEnv0  = emptyPMCFGEnv
-
-  let lincont    = [(Explicit, varStr, typeStr)]
-  b             <- convert opts gr cenv (floc opath loc1 id) def (lincont,lincat) [pvar]
-  let (seqs1,b1) = addSequencesB seqs b
-      pmcfgEnv1  = foldBM addLindef
-                          pmcfgEnv0
-                          (goB b1 CNil [])
-                          (pcat,[pvar])
-
-  let lincont    = [(Explicit, varStr, lincat)]
-  b             <- convert opts gr cenv (floc opath loc2 id) ref (lincont,typeStr) [pcat]
-  let (seqs2,b2) = addSequencesB seqs1 b
-      pmcfgEnv2  = foldBM addLinref
-                          pmcfgEnv1
-                          (goB b2 CNil [])
-                          (pvar,[pcat])
-
-  let pmcfg     = getPMCFG pmcfgEnv2
-
-  when (verbAtLeast opts Verbose) $ ePutStr ("\n+ "++showIdent id++" "++show (catFactor pcat))
-  seqs2 `seq` pmcfg `seq` return (seqs2,GF.Grammar.CncCat mty mdef mref mprn (Just pmcfg))
+    collect st t = do
+      tnk <- newThunk [] t
+      let v0 = VS v tnk []
+      v <- patternMatch v0 (map (\(p,t) -> (env,[p],[tnk],t)) cs)
+      flatten v q st
+flatten (VV _ tnks) (Table _ q) st = do
+  foldM collect st tnks
   where
-    addLindef lins (newCat', newArgs') env0 =
-      let [newCat] = getFIds newCat'
-          !fun     = mkArray lins
-      in addFunction env0 newCat fun [[fidVar]]
+    collect st tnk = do
+      v <- force tnk []
+      flatten v q st
+flatten v (Sort s) (lins,params) | s == cStr = do
+  return (v:lins,params)
+flatten v (QC q) (lins,params) = do
+  return (lins,v:params)
 
-    addLinref lins (newCat', [newArg']) env0 =
-      let newArg   = getFIds newArg'
-          !fun     = mkArray lins
-      in addFunction env0 fidVar fun [newArg]
-
-addPMCFG opts gr cenv opath am cm seqs id info = return (seqs, info)
-
-floc opath loc id = maybe (L loc id) (\path->L (External path loc) id) opath
-
-convert opts gr cenv loc term ty@(_,val) pargs =
-  case normalForm cenv loc (etaExpand ty term) of
-    Error s -> fail $ render $ ppL loc ("Predef.error: "++s)
-    term    -> return $ runCnvMonad gr (convertTerm opts CNil val term) (pargs,[])
+str2lin (VStr s)         = return [SymKS s]
+str2lin (VSymCat d r rs) = do (r, rs) <- compute r rs
+                              return [SymCat d r rs]
   where
-    etaExpand (context,val) = mkAbs pars . flip mkApp args
-      where pars = [(Explicit,v) | v <- vars]
-            args = map Vr vars
-            vars = map (\(bt,x,t) -> x) context
+    compute r' []                = return (r',[])
+    compute r' ((cnt',tnk):tnks) = do
+      (r, rs,_) <- force tnk [] >>= param2int
+      (r',rs' ) <- compute r' tnks
+      return (r*cnt'+r',combine cnt' rs rs')
+str2lin (VC vs)          = fmap concat (mapM str2lin vs)
+str2lin v                = do t <- value2term 0 v
+                              evalError ("the term" <+> ppTerm Unqualified 0 t $$
+                                         "cannot be evaluated at compile time.")
 
-pgfCncCat :: SourceGrammar -> Type -> Int -> CncCat
-pgfCncCat gr lincat index =
-  let ((_,size),schema) = computeCatRange gr lincat
-  in PGF.CncCat index (index+size-1)
-                      (mkArray (map (renderStyle style{mode=OneLineMode} . ppPath) 
-                                    (getStrPaths schema)))
+param2int (VApp q tnks) = do
+  (r ,    cnt ) <- getIdxCnt q
+  (r',rs',cnt') <- compute tnks
+  return (r*cnt' + r',rs',cnt*cnt')
   where
-    getStrPaths :: Schema Identity s c -> [Path]
-    getStrPaths = collect CNil []
-      where
-        collect path paths (CRec rs)   = foldr (\(lbl,Identity t) paths -> collect (CProj lbl path) paths t) paths rs
-        collect path paths (CTbl _ cs) = foldr (\(trm,Identity t) paths -> collect (CSel  trm path) paths t) paths cs
-        collect path paths (CStr _)    = reversePath path : paths
-        collect path paths (CPar _)    = paths
+    getIdxCnt q = do
+      (_,ResValue (L _ ty) idx) <- getInfo q
+      let QC p = valTypeCnc ty
+      (_,ResParam _ (Just (_,cnt))) <- getInfo p
+      return (idx,cnt)
 
-----------------------------------------------------------------------
--- CnvMonad monad
---
--- The branching monad provides backtracking together with
--- recording of the choices made. We have two cases
--- when we have alternative choices:
---
---      * when we have parameter type, then
---        we have to try all possible values
---      * when we have variants we have to try all alternatives
---
--- The conversion monad keeps track of the choices and they are
--- returned as 'Branch' data type.
+    compute []         = return (0,[],1)
+    compute (tnk:tnks) = do
+      (r, rs ,cnt ) <- force tnk [] >>= param2int
+      (r',rs',cnt') <- compute tnks
+      return (r*cnt'+r',combine cnt' rs rs',cnt*cnt')
+param2int (VMeta tnk _ _) = do
+  tnk_st <- getMeta tnk
+  case tnk_st of
+    Evaluated v         -> param2int v
+    Unbound (Just ty) j -> do let QC q = valTypeCnc ty
+                              (_,ResParam _ (Just (_,cnt))) <- getInfo q
+                              return (0,[(1,j)],cnt)
 
-data Branch a
-  = Case Int Path [(Term,Branch a)]
-  | Variant [Branch a]
-  | Return  a
+combine cnt'          []            rs' = rs'
+combine cnt'         rs              [] = [(r*cnt',pv) | (r,pv) <- rs]
+combine cnt' ((r,pv):rs) ((r',pv'):rs') =
+  case compare pv pv' of
+    LT -> (r*cnt',   pv ) : combine cnt' rs ((r',pv'):rs')
+    EQ -> (r*cnt'+r',pv ) : combine cnt' rs ((r',pv'):rs')
+    GT -> (       r',pv') : combine cnt' ((r,pv):rs) rs'
 
-newtype CnvMonad a = CM {unCM :: SourceGrammar
-                              -> forall b . (a -> ([ProtoFCat],[Symbol]) -> Branch b)
-                              -> ([ProtoFCat],[Symbol])
-                              -> Branch b}
+mapAccumM f a []     = return (a,[])
+mapAccumM f a (x:xs) = do (a, y) <- f a x
+                          (a,ys) <- mapAccumM f a xs
+                          return (a,y:ys)
 
-instance Fail.MonadFail CnvMonad where
-  fail = bug
-
-instance Applicative CnvMonad where
-  pure = return
-  (<*>) = ap
-
-instance Monad CnvMonad where
-    return a   = CM (\gr c s -> c a s)
-    CM m >>= k = CM (\gr c s -> m gr (\a s -> unCM (k a) gr c s) s)
-
-instance MonadState ([ProtoFCat],[Symbol]) CnvMonad where
-    get = CM (\gr c s -> c s s)
-    put s = CM (\gr c _ -> c () s)
-
-instance Functor CnvMonad where
-    fmap f (CM m) = CM (\gr c s -> m gr (c . f) s)
-
-runCnvMonad :: SourceGrammar -> CnvMonad a -> ([ProtoFCat],[Symbol]) -> Branch a
-runCnvMonad gr (CM m) s = m gr (\v s -> Return v) s
-
--- | backtracking for all variants
-variants :: [a] -> CnvMonad a
-variants xs = CM (\gr c s -> Variant [c x s | x <- xs])
-
--- | backtracking for all parameter values that a variable could take
-choices :: Int -> Path -> CnvMonad Term
-choices nr path = do (args,_) <- get
-                     let PFCat _ _ schema = args !! nr
-                     descend schema path CNil
-  where
-    descend (CRec rs)     (CProj lbl path) rpath = case lookup lbl rs of
-                                                     Just (Identity t) -> descend t path (CProj lbl rpath)
-    descend (CRec rs)     CNil             rpath = do rs <- mapM (\(lbl,Identity t) -> fmap (assign lbl) (descend t CNil (CProj lbl rpath))) rs
-                                                      return (R rs)
-    descend (CTbl pt cs)  (CSel  trm path) rpath = case lookup trm cs of
-                                                     Just (Identity t) -> descend t path (CSel trm rpath)
-    descend (CTbl pt cs)  CNil             rpath = do cs <- mapM (\(trm,Identity t) -> descend t CNil (CSel trm rpath)) cs
-                                                      return (V pt cs)
-    descend (CPar (m,vs)) CNil             rpath = case vs of
-                                                     [(value,index)] -> return value
-                                                     values          -> let path = reversePath rpath
-                                                                        in CM (\gr c s -> Case nr path [(value, updateEnv path value gr c s)
-                                                                                                                    | (value,index) <- values])
-    descend schema       path              rpath = bug $ "descend "++show (schema,path,rpath)
-
-    updateEnv path value gr c (args,seq) = 
-      case updateNthM (restrictProtoFCat path value) nr args of
-        Just args -> c value (args,seq)
-        Nothing   -> bug "conflict in updateEnv"
-
--- | the argument should be a parameter type and then
--- the function returns all possible values.
-getAllParamValues :: Type -> CnvMonad [Term]
-getAllParamValues ty = CM (\gr c -> c (err bug id (allParamValues gr ty)))
-
-mkRecord :: [(Label,CnvMonad (Schema Branch s c))] -> CnvMonad (Schema Branch s c)
-mkRecord xs = CM (\gr c -> foldl (\c (lbl,CM m) bs s -> c ((lbl,m gr (\v s -> Return v) s) : bs) s) (c . CRec) xs [])
-
-mkTable  :: Type -> [(Term ,CnvMonad (Schema Branch s c))] -> CnvMonad (Schema Branch s c)
-mkTable pt xs = CM (\gr c -> foldl (\c (trm,CM m) bs s -> c ((trm,m gr (\v s -> Return v) s) : bs) s) (c . CTbl pt) xs [])
-
-----------------------------------------------------------------------
--- Term Schema
---
--- The term schema is a term-like structure, with records, tables,
--- strings and parameters values, but in addition we could add
--- annotations of arbitrary types
-
--- | Term schema
-data Schema b s c
-  = CRec      [(Label,b (Schema b s c))]
-  | CTbl Type [(Term, b (Schema b s c))]
-  | CStr s
-  | CPar c
---deriving Show -- doesn't work
-
-instance Show s => Show (Schema b s c) where
-  showsPrec _ sch =
-      case sch of
-        CRec r   -> showString "CRec " . shows (map fst r)
-        CTbl t _ -> showString "CTbl " . showsPrec 10 t . showString " _"
-        CStr s   -> showString "CStr " . showsPrec 10 s
-        CPar c   -> showString "CPar{}"
-
--- | Path into a term or term schema
-data Path
-  = CProj Label Path
-  | CSel  Term  Path
-  | CNil
-  deriving (Eq,Show)
-
--- | The ProtoFCat represents a linearization type as term schema.
--- The annotations are as follows: the strings are annotated with
--- their index in the PMCFG tuple, the parameters are annotated
--- with their value both as term and as index.
-data ProtoFCat  = PFCat Ident Int (Schema Identity Int (Int,[(Term,Int)]))
-type Env        = (ProtoFCat, [ProtoFCat])
-
-protoFCat :: SourceGrammar -> Cat -> Type -> ProtoFCat
-protoFCat gr cat lincat =
-  case computeCatRange gr lincat of
-    ((_,f),schema) -> PFCat (snd cat) f schema
-
-getFIds :: ProtoFCat -> [FId]
-getFIds (PFCat _ _ schema) =
-  reverse (solutions (variants schema) ())
-  where
-    variants (CRec rs)         = fmap sum $ mapM (\(lbl,Identity t) -> variants t) rs
-    variants (CTbl _ cs)       = fmap sum $ mapM (\(trm,Identity t) -> variants t) cs
-    variants (CStr _)          = return 0
-    variants (CPar (m,values)) = do (value,index) <- member values
-                                    return (m*index)
-
-catFactor :: ProtoFCat -> Int
-catFactor (PFCat _ f _) = f
-
-computeCatRange gr lincat = compute (0,1) lincat
-  where
-    compute st (RecType  rs) = let (st',rs') = List.mapAccumL (\st (lbl,t) -> case lbl of
-                                                                                LVar _ -> let (st',t') = compute st t
-                                                                                          in (st ,(lbl,Identity t'))
-                                                                                _      -> let (st',t') = compute st t
-                                                                                          in (st',(lbl,Identity t'))) st rs
-                               in (st',CRec rs')
-    compute st (Table pt vt) = let vs        = err bug id (allParamValues gr pt)
-                                   (st',cs') = List.mapAccumL (\st v       -> let (st',vt') = compute st vt
-                                                                              in (st',(v,Identity vt'))) st vs
-                               in (st',CTbl pt cs')
-    compute st (Sort s)
-                   | s == cStr = let (index,m) = st
-                                 in ((index+1,m),CStr index)
-    compute st t               = let vs = err bug id (allParamValues gr t)
-                                     (index,m) = st
-                                 in ((index,m*length vs),CPar (m,zip vs [0..]))
-
-ppPath (CProj lbl path) = lbl <+> ppPath path
-ppPath (CSel  trm path) = ppU 5 trm   <+> ppPath path
-ppPath CNil             = empty
-
-reversePath path = rev CNil path
-  where
-    rev path0 CNil             = path0
-    rev path0 (CProj lbl path) = rev (CProj lbl path0) path
-    rev path0 (CSel  trm path) = rev (CSel  trm path0) path
-
-
-----------------------------------------------------------------------
--- term conversion
-
-type Value a = Schema Branch a Term
-
-convertTerm :: Options -> Path -> Type -> Term -> CnvMonad (Value [Symbol])
-convertTerm opts sel ctype (Vr x)       = convertArg opts ctype (getVarIndex x) (reversePath sel)
-convertTerm opts sel ctype (Abs _ _ t)  = convertTerm opts sel ctype t                -- there are only top-level abstractions and we ignore them !!!
-convertTerm opts sel ctype (R record)   = convertRec opts sel ctype record
-convertTerm opts sel ctype (P term l)   = convertTerm opts (CProj l sel) ctype term
-convertTerm opts sel ctype (V pt ts)    = convertTbl opts sel ctype pt ts
-convertTerm opts sel ctype (S term p)   = do v <- evalTerm CNil p
-                                             convertTerm opts (CSel v sel) ctype term
-convertTerm opts sel ctype (FV vars)    = do term <- variants vars
-                                             convertTerm opts sel ctype term
-convertTerm opts sel ctype (C t1 t2)    = do v1 <- convertTerm opts sel ctype t1
-                                             v2 <- convertTerm opts sel ctype t2
-                                             return (CStr (concat [s | CStr s <- [v1,v2]]))
-convertTerm opts sel ctype (K t)        = return (CStr [SymKS t])
-convertTerm opts sel ctype Empty        = return (CStr [])
-convertTerm opts sel ctype (Alts s alts)= do CStr s <- convertTerm opts CNil ctype s
-                                             alts <- forM alts $ \(u,alt) -> do
-                                                            CStr u <- convertTerm opts CNil ctype u
-                                                            Strs ps <- unPatt alt
-                                                            ps     <- mapM (convertTerm opts CNil ctype) ps
-                                                            return (u,map unSym ps)
-                                             return (CStr [SymKP s alts])
-  where
-    unSym (CStr [])        = ""
-    unSym (CStr [SymKS t]) = t
-    unSym _                = ppbug $ hang ("invalid prefix in pre expression:") 4 (Alts s alts)
-
-    unPatt (EPatt p) = fmap Strs (getPatts p)
-    unPatt u         = return u
-
-    getPatts p = case p of
-      PAlt a b  -> liftM2 (++) (getPatts a) (getPatts b)
-      PString s -> return [K s]
-      PSeq a b  -> do
-        as <- getPatts a
-        bs <- getPatts b
-        return [K (s ++ t) | K s <- as, K t <- bs]
-      _ -> fail (render ("not valid pattern in pre expression" <+> ppPatt Unqualified 0 p))
-
-convertTerm opts sel ctype (Q (m,f))
-  | m == cPredef &&
-    f == cBIND                          = return (CStr [SymBIND])
-  | m == cPredef &&
-    f == cSOFT_BIND                     = return (CStr [SymSOFT_BIND])
-  | m == cPredef &&
-    f == cSOFT_SPACE                    = return (CStr [SymSOFT_SPACE])
-  | m == cPredef &&
-    f == cCAPIT                         = return (CStr [SymCAPIT])
-  | m == cPredef &&
-    f == cALL_CAPIT                     = return (CStr [SymALL_CAPIT])
-  | m == cPredef &&
-    f == cNonExist                      = return (CStr [SymNE])
-{-
-convertTerm opts sel@(CProj l _) ctype (ExtR t1 t2@(R rs2))
-                    | l `elem` map fst rs2 = convertTerm opts sel ctype t2
-                    | otherwise            = convertTerm opts sel ctype t1
-
-convertTerm opts sel@(CProj l _) ctype (ExtR t1@(R rs1) t2)
-                    | l `elem` map fst rs1 = convertTerm opts sel ctype t1
-                    | otherwise            = convertTerm opts sel ctype t2
--}
-convertTerm opts CNil ctype t           = do v <- evalTerm CNil t
-                                             return (CPar v)
-convertTerm _    sel  _     t           = ppbug ("convertTerm" <+> sep [parens (show sel),ppU 10 t])
-
-convertArg :: Options -> Term -> Int -> Path -> CnvMonad (Value [Symbol])
-convertArg opts (RecType rs)  nr path =
-  mkRecord (map (\(lbl,ctype) -> (lbl,convertArg opts ctype nr (CProj lbl path))) rs)
-convertArg opts (Table pt vt) nr path = do
-  vs <- getAllParamValues pt
-  mkTable pt (map (\v -> (v,convertArg opts vt nr (CSel v path))) vs)
-convertArg opts (Sort _)      nr path = do
-  (args,_) <- get
-  let PFCat cat _ schema = args !! nr
-      l = index (reversePath path) schema
-      sym | CProj (LVar i) CNil <- path = SymVar nr i
-          | isLiteralCat opts cat       = SymLit nr l
-          | otherwise                   = SymCat nr l
-  return (CStr [sym])
-  where
-    index (CProj lbl path) (CRec   rs) = case lookup lbl rs of
-                                           Just (Identity t) -> index path t
-    index (CSel  trm path) (CTbl _ rs) = case lookup trm rs of
-                                           Just (Identity t) -> index path t
-    index CNil             (CStr idx)  = idx
-convertArg opts ty nr path              = do
-  value <- choices nr (reversePath path)
-  return (CPar value)
-
-convertRec opts CNil (RecType rs) record =
-    mkRecord [(lbl,convertTerm opts CNil ctype (proj lbl))|(lbl,ctype)<-rs]
-  where proj lbl = if isLockLabel lbl then R [] else projectRec lbl record
-convertRec opts (CProj lbl path) ctype record =
-  convertTerm opts path ctype (projectRec lbl record)
-convertRec opts _ ctype _ = bug ("convertRec: "++show ctype)
-
-convertTbl opts CNil (Table _ vt) pt ts = do
-  vs <- getAllParamValues pt
-  mkTable pt (zipWith (\v t -> (v,convertTerm opts CNil vt t)) vs ts)
-convertTbl opts (CSel v sub_sel) ctype pt ts = do
-  vs <- getAllParamValues pt
-  case lookup v (zip vs ts) of
-    Just t  -> convertTerm opts sub_sel ctype t
-    Nothing -> ppbug ( "convertTbl:" <+> ("missing value" <+> v $$
-                                          "among" <+> vcat vs))
-convertTbl opts _ ctype _ _ = bug ("convertTbl: "++show ctype)
-
-
-goB :: Branch (Value SeqId) -> Path -> [SeqId] -> BacktrackM Env [SeqId]
-goB (Case nr path bs) rpath ss = do (value,b) <- member bs
-                                    restrictArg nr path value
-                                    goB b rpath ss
-goB (Variant bs)      rpath ss = do b <- member bs
-                                    goB b rpath ss
-goB (Return  v)       rpath ss = goV v rpath ss
-
-goV :: Value SeqId -> Path -> [SeqId] -> BacktrackM Env [SeqId]
-goV (CRec xs)    rpath ss = foldM (\ss (lbl,b) -> goB b (CProj lbl rpath) ss) ss (reverse xs)
-goV (CTbl _ xs)  rpath ss = foldM (\ss (trm,b) -> goB b (CSel  trm rpath) ss) ss (reverse xs)
-goV (CStr seqid) rpath ss = return (seqid : ss)
-goV (CPar t)     rpath ss = restrictHead (reversePath rpath) t >> return ss
-
-
-----------------------------------------------------------------------
--- SeqSet
-
-type SeqSet   = Map.Map Sequence SeqId
-
-addSequencesB :: SeqSet -> Branch (Value [Symbol]) -> (SeqSet, Branch (Value SeqId))
-addSequencesB seqs (Case nr path bs) = let !(seqs1,bs1) = mapAccumL' (\seqs (trm,b) -> let !(seqs',b') = addSequencesB seqs b
-                                                                                       in (seqs',(trm,b'))) seqs bs
-                                       in (seqs1,Case nr path bs1)
-addSequencesB seqs (Variant bs)      = let !(seqs1,bs1) = mapAccumL' addSequencesB seqs bs
-                                       in (seqs1,Variant bs1)
-addSequencesB seqs (Return  v)       = let !(seqs1,v1) = addSequencesV seqs v
-                                       in (seqs1,Return v1)
-
-addSequencesV :: SeqSet -> Value [Symbol] -> (SeqSet, Value SeqId)
-addSequencesV seqs (CRec vs)  = let !(seqs1,vs1) = mapAccumL' (\seqs (lbl,b) -> let !(seqs',b') = addSequencesB seqs b
-                                                                                in (seqs',(lbl,b'))) seqs vs
-                                in (seqs1,CRec vs1)
-addSequencesV seqs (CTbl pt vs)=let !(seqs1,vs1) = mapAccumL' (\seqs (trm,b) -> let !(seqs',b') = addSequencesB seqs b
-                                                                                in (seqs',(trm,b'))) seqs vs
-                                in (seqs1,CTbl pt vs1)
-addSequencesV seqs (CStr lin) = let !(seqs1,seqid) = addSequence seqs lin
-                                in (seqs1,CStr seqid)
-addSequencesV seqs (CPar i)   = (seqs,CPar i)
-
--- a strict version of Data.List.mapAccumL
-mapAccumL' f s []     = (s,[])
-mapAccumL' f s (x:xs) = (s'',y:ys)
-                        where !(s', y ) = f s x
-                              !(s'',ys) = mapAccumL' f s' xs
-
-addSequence :: SeqSet -> [Symbol] -> (SeqSet,SeqId)
-addSequence seqs lst =
-  case Map.lookup seq seqs of
-    Just id -> (seqs,id)
-    Nothing -> let !last_seq = Map.size seqs
-               in (Map.insert seq last_seq seqs, last_seq)
-  where
-    seq = mkArray lst
-
-
-------------------------------------------------------------
--- eval a term to ground terms
-
-evalTerm :: Path -> Term -> CnvMonad Term
-evalTerm CNil (QC f)       = return (QC f)
-evalTerm CNil (App x y)    = do x <- evalTerm CNil x
-                                y <- evalTerm CNil y
-                                return (App x y)
-evalTerm path (Vr x)       = choices (getVarIndex x) path
-evalTerm path (R rs)       =
-    case path of
-      CProj lbl path -> evalTerm path (projectRec lbl rs)
-      CNil           -> R `fmap` mapM (\(lbl,(_,t)) -> assign lbl `fmap` evalTerm path t) rs
-evalTerm path (P term lbl) = evalTerm (CProj lbl path) term
-evalTerm path (V pt ts)    =
-   case path of
-     CNil          -> V pt `fmap` mapM (evalTerm path) ts
-     CSel trm path ->
-         do vs <- getAllParamValues pt
-            case lookup trm (zip vs ts) of
-              Just t  -> evalTerm path t
-              Nothing -> ppbug $ "evalTerm: missing value:"<+>trm
-                                 $$ "among:" <+>fsep (map (ppU 10) vs)
-evalTerm path (S term sel) = do v <- evalTerm CNil sel
-                                evalTerm (CSel v path) term
-evalTerm path (FV terms)   = variants terms >>= evalTerm path
-evalTerm path (EInt n)     = return (EInt n)
-evalTerm path t            = ppbug ("evalTerm" <+> parens t)
---evalTerm path t            = ppbug (text "evalTerm" <+> sep [parens (text (show path)),parens (text (show t))])
-
-getVarIndex x = maybe err id $ getArgIndex x
-  where err = bug ("getVarIndex "++show x)
-
-----------------------------------------------------------------------
--- GrammarEnv
-
-data PMCFGEnv = PMCFGEnv !ProdSet !FunSet
-type ProdSet  = Set.Set Production
-type FunSet   = Map.Map (UArray LIndex SeqId) FunId
-
-emptyPMCFGEnv =
-  PMCFGEnv Set.empty Map.empty
-
-addFunction :: PMCFGEnv -> FId -> UArray LIndex SeqId -> [[FId]] -> PMCFGEnv
-addFunction (PMCFGEnv prodSet funSet) !fid fun args =
-  case Map.lookup fun funSet of
-    Just !funid -> PMCFGEnv (Set.insert (Production fid funid args) prodSet)
-                            funSet
-    Nothing     -> let !funid = Map.size funSet
-                   in PMCFGEnv (Set.insert (Production fid funid args) prodSet)
-                               (Map.insert fun funid funSet)
-
-getPMCFG :: PMCFGEnv -> PMCFG
-getPMCFG (PMCFGEnv prodSet funSet) =
-  PMCFG (optimize prodSet) (mkSetArray funSet)
-  where
-    optimize ps = Map.foldrWithKey ff [] (Map.fromListWith (++) [((fid,funid),[args]) | (Production fid funid args) <- Set.toList ps])
-      where
-        ff :: (FId,FunId) -> [[[FId]]] -> [Production] -> [Production]
-        ff (fid,funid) xs prods
-          | product (map IntSet.size ys) == count
-                      = (Production fid funid (map IntSet.toList ys)) : prods
-          | otherwise = map (Production fid funid) xs ++ prods
-          where
-            count = sum (map (product . map length) xs)
-            ys    = foldl (zipWith (foldr IntSet.insert)) (repeat IntSet.empty) xs
-
-------------------------------------------------------------
--- updating the MCF rule
-
-restrictArg :: LIndex -> Path -> Term -> BacktrackM Env ()
-restrictArg nr path index = do
-  (head, args) <- get
-  args <- updateNthM (restrictProtoFCat path index) nr args
-  put (head, args)
-
-restrictHead :: Path -> Term -> BacktrackM Env ()
-restrictHead path term = do
-  (head, args) <- get
-  head <- restrictProtoFCat path term head
-  put (head, args)
-
-restrictProtoFCat :: (Functor m, MonadPlus m) => Path -> Term -> ProtoFCat -> m ProtoFCat
-restrictProtoFCat path v (PFCat cat f schema) = do
-  schema <- addConstraint path v schema
-  return (PFCat cat f schema)
-  where
-    addConstraint (CProj lbl path) v (CRec rs)     = fmap CRec      $ update lbl (addConstraint path v) rs
-    addConstraint (CSel  trm path) v (CTbl pt cs)  = fmap (CTbl pt) $ update trm (addConstraint path v) cs
-    addConstraint CNil             v (CPar (m,vs)) = case lookup v vs of
-                                                       Just index -> return (CPar (m,[(v,index)]))
-                                                       Nothing    -> mzero
-    addConstraint CNil             v (CStr _)      = bug "restrictProtoFCat: string path"
-    
-    update k0 f [] = return []
-    update k0 f (x@(k,Identity v):xs)
-      | k0 == k    = do v <- f v
-                        return ((k,Identity v):xs)
-      | otherwise  = do xs <- update k0 f xs
-                        return (x:xs)
-
-mkArray    lst = listArray (0,length lst-1) lst
-mkSetArray map = array (0,Map.size map-1) [(v,k) | (k,v) <- Map.toList map]
-
-bug msg = ppbug msg
-ppbug msg = error completeMsg
- where
-  originalMsg = render $ hang "Internal error in GeneratePMCFG:" 4 msg
-  completeMsg =
-    case render msg of -- the error message for pattern matching a runtime string
-      "descend (CStr 0,CNil,CProj (LIdent (Id {rawId2utf8 = \"s\"})) CNil)"
-        -> unlines [originalMsg -- add more helpful output
-            ,""
-            ,"1) Check that you are not trying to pattern match a /runtime string/."
-            ,"   These are illegal:"
-            ,"     lin Test foo = case foo.s of {"
-            ,"                       \"str\" => … } ;   <- explicit matching argument of a lin"
-            ,"     lin Test foo = opThatMatches foo   <- calling an oper that pattern matches"
-            ,""
-            ,"2) Not about pattern matching? Submit a bug report and we update the error message."
-            ,"     https://github.com/GrammaticalFramework/gf-core/issues"
-            ]
-      _ -> originalMsg -- any other message: just print it as is
-
-ppU = ppTerm Unqualified
+pgfCncCat = error "TODO: pgfCncCat"

src/compiler/GF/Compile/GetGrammar.hs

@@ -50,20 +50,13 @@ getSourceModule opts file0 =
        Right (i,mi0) ->
          do liftIO $ removeTemp tmp
             let mi =mi0 {mflags=mflags mi0 `addOptions` opts, msrc=file0}
-                optCoding' = renameEncoding `fmap` flag optEncoding (mflags mi0)
-            case (optCoding,optCoding') of
-              {-
-              (Nothing,Nothing) ->
-                  unless (BS.all isAscii raw) $
-                    ePutStrLn $ file0++":\n    Warning: default encoding has changed from Latin-1 to UTF-8"
-              -}
-              (_,Just coding') -> 
-                  when (coding/=coding') $
+            case renameEncoding `fmap` flag optEncoding (mflags mi0) of
+              Just coding' -> 
+                when (coding/=coding') $
                   raise $ "Encoding mismatch: "++coding++" /= "++coding'
                 where coding = maybe defaultEncoding renameEncoding optCoding
               _ -> return ()
-          --liftIO $ transcodeModule' (i,mi) -- old lexer
-            return (i,mi) -- new lexer
+            return (i,mi)
 
 getBNFCRules :: Options -> FilePath -> IOE [BNFCRule]
 getBNFCRules opts fpath = do

src/compiler/GF/Compile/GrammarToCanonical.hs

@@ -6,32 +6,38 @@ module GF.Compile.GrammarToCanonical(
        ) where
 import Data.List(nub,partition)
 import qualified Data.Map as M
+import Data.Maybe(fromMaybe)
 import qualified Data.Set as S
 import GF.Data.ErrM
 import GF.Text.Pretty
-import GF.Grammar.Grammar
+import GF.Grammar.Grammar as G
 import GF.Grammar.Lookup(lookupOrigInfo,allOrigInfos,allParamValues)
-import GF.Grammar.Macros(typeForm,collectOp,collectPattOp,mkAbs,mkApp,term2patt)
+import GF.Grammar.Macros(typeForm,collectOp,collectPattOp,composSafeOp,mkAbs,mkApp,term2patt,sortRec)
 import GF.Grammar.Lockfield(isLockLabel)
 import GF.Grammar.Predef(cPredef,cInts)
-import GF.Compile.Compute.Predef(predef)
-import GF.Compile.Compute.Value(Predefined(..))
-import GF.Infra.Ident(ModuleName(..),Ident,prefixIdent,showIdent,isWildIdent)
-import GF.Infra.Option(optionsPGF)
-import PGF.Internal(Literal(..))
-import GF.Compile.Compute.ConcreteNew(normalForm,resourceValues)
+-- import GF.Compile.Compute.Value(Predefined(..))
+import GF.Infra.Ident(ModuleName(..),Ident,ident2raw,rawIdentS,showIdent,isWildIdent)
+import GF.Infra.Option(Options,optionsPGF)
+import GF.Infra.CheckM
+import PGF2(Literal(..))
+import GF.Compile.Compute.Concrete(normalForm)
 import GF.Grammar.Canonical as C
-import Debug.Trace
+import System.FilePath ((</>), (<.>))
+import qualified Debug.Trace as T
+
 
 -- | Generate Canonical code for the named abstract syntax and all associated
 -- concrete syntaxes
-grammar2canonical opts absname gr =
-   Grammar (abstract2canonical absname gr)
-           (map snd (concretes2canonical opts absname gr))
+grammar2canonical :: Options -> ModuleName -> G.Grammar -> Check C.Grammar
+grammar2canonical opts absname gr = do
+  abs <- abstract2canonical absname gr
+  cncs <- concretes2canonical opts absname gr
+  return (Grammar abs (map snd cncs))
 
 -- | Generate Canonical code for the named abstract syntax
+abstract2canonical :: ModuleName -> G.Grammar -> Check Abstract
 abstract2canonical absname gr =
-    Abstract (modId absname) (convFlags gr absname) cats funs
+  return (Abstract (modId absname) (convFlags gr absname) cats funs)
   where
     cats = [CatDef (gId c) (convCtx ctx) | ((_,c),AbsCat ctx) <- adefs]
 
@@ -44,6 +50,7 @@ abstract2canonical absname gr =
     convHypo (bt,name,t) =
       case typeForm t of
         ([],(_,cat),[]) -> gId cat -- !!
+        tf -> error ("abstract2canonical convHypo: " ++ show tf)
 
     convType t =
       case typeForm t of
@@ -54,28 +61,26 @@ abstract2canonical absname gr =
 
     convHypo' (bt,name,t) = TypeBinding (gId name) (convType t)
 
-
 -- | Generate Canonical code for the all concrete syntaxes associated with
 -- the named abstract syntax in given the grammar.
+concretes2canonical :: Options -> ModuleName -> G.Grammar -> Check [(FilePath, Concrete)]
 concretes2canonical opts absname gr =
-  [(cncname,concrete2canonical gr cenv absname cnc cncmod)
-     | let cenv = resourceValues opts gr,
-       cnc<-allConcretes gr absname,
-       let cncname = "canonical/"++render cnc ++ ".gf" :: FilePath
-           Ok cncmod = lookupModule gr cnc
-  ]
+  sequence
+    [fmap ((,) cncname) (concrete2canonical gr absname cnc cncmod)
+        | cnc<-allConcretes gr absname,
+          let cncname = "canonical" </> render cnc <.> "gf"
+              Ok cncmod = lookupModule gr cnc
+        ]
 
 -- | Generate Canonical GF for the given concrete module.
-concrete2canonical gr cenv absname cnc modinfo =
-  Concrete (modId cnc) (modId absname) (convFlags gr cnc)
-      (neededParamTypes S.empty (params defs))
-      [lincat|(_,Left lincat)<-defs]
-      [lin|(_,Right lin)<-defs]
+concrete2canonical :: G.Grammar -> ModuleName -> ModuleName -> ModuleInfo -> Check Concrete
+concrete2canonical gr absname cnc modinfo = do
+  defs <- fmap concat $ mapM (toCanonical gr absname) (M.toList (jments modinfo))
+  return (Concrete (modId cnc) (modId absname) (convFlags gr cnc)
+                   (neededParamTypes S.empty (params defs))
+                   [lincat | (_,Left lincat) <- defs]
+                   [lin | (_,Right lin) <- defs])
   where
-    defs = concatMap (toCanonical gr absname cenv) . 
-           M.toList $
-           jments modinfo
-
     params = S.toList . S.unions . map fst
 
     neededParamTypes have [] = []
@@ -85,35 +90,30 @@ concrete2canonical gr cenv absname cnc modinfo =
         else let ((got,need),def) = paramType gr q
              in def++neededParamTypes (S.union got have) (S.toList need++qs)
 
-toCanonical gr absname cenv (name,jment) =
+-- toCanonical :: G.Grammar -> ModuleName -> (Ident, Info) -> [(S.Set QIdent, Either LincatDef LinDef)]
+toCanonical gr absname (name,jment) =
   case jment of
-    CncCat (Just (L loc typ)) _ _ pprn _ ->
-        [(pts,Left (LincatDef (gId name) (convType ntyp)))]
-      where
-        pts = paramTypes gr ntyp
-        ntyp = nf loc typ
-    CncFun (Just r@(cat,ctx,lincat)) (Just (L loc def)) pprn _ ->
-        [(tts,Right (LinDef (gId name) (map gId args) (convert gr e')))]
-      where
-        tts = tableTypes gr [e']
-
-        e' = unAbs (length params) $
-             nf loc (mkAbs params (mkApp def (map Vr args)))
-        params = [(b,x)|(b,x,_)<-ctx]
-        args = map snd params
-
+    CncCat (Just (L loc typ)) _ _ pprn _ -> do
+      ntyp <- normalForm gr typ
+      let pts = paramTypes gr ntyp
+      return [(pts,Left (LincatDef (gId name) (convType ntyp)))]
+    CncFun (Just r@(_,cat,ctx,lincat)) (Just (L loc def)) pprn _ -> do
+      let params = [(b,x)|(b,x,_)<-ctx]
+          args   = map snd params
+      e0 <- normalForm gr (mkAbs params (mkApp def (map Vr args)))
+      let e   = cleanupRecordFields lincat (unAbs (length params) e0)
+          tts = tableTypes gr [e]
+      return [(tts,Right (LinDef (gId name) (map gId args) (convert gr e)))]
     AnyInd _ m  -> case lookupOrigInfo gr (m,name) of
-                     Ok (m,jment) -> toCanonical gr absname cenv (name,jment)
-                     _ -> []
-    _ -> []
+                     Ok (m,jment) -> toCanonical gr absname (name,jment)
+                     _ -> return []
+    _ -> return []
   where
-    nf loc = normalForm cenv (L loc name)
---  aId n = prefixIdent "A." (gId n)
-
     unAbs 0 t = t
     unAbs n (Abs _ _ t) = unAbs (n-1) t
     unAbs _ t = t
 
+tableTypes :: G.Grammar -> [Term] -> S.Set QIdent
 tableTypes gr ts = S.unions (map tabtys ts)
   where
     tabtys t =
@@ -122,6 +122,7 @@ tableTypes gr ts = S.unions (map tabtys ts)
         T (TTyped t) cs -> S.union (paramTypes gr t) (tableTypes gr (map snd cs))
         _ -> collectOp tabtys t
 
+paramTypes :: G.Grammar -> G.Type -> S.Set QIdent
 paramTypes gr t =
   case t of
     RecType fs -> S.unions (map (paramTypes gr.snd) fs)
@@ -140,11 +141,26 @@ paramTypes gr t =
                  Ok (_,ResParam {}) -> S.singleton q
                  _ -> ignore
 
-    ignore = trace ("Ignore: "++show t) S.empty
+    ignore = T.trace ("Ignore: " ++ show t) S.empty
 
+-- | Filter out record fields from definitions which don't appear in lincat.
+cleanupRecordFields :: G.Type -> Term -> Term
+cleanupRecordFields (RecType ls) (R as) =
+  let defnFields = M.fromList ls
+  in R
+      [ (lbl, (mty, t'))
+      | (lbl, (mty, t)) <- as
+      , M.member lbl defnFields
+      , let Just ty = M.lookup lbl defnFields
+      , let t' = cleanupRecordFields ty t
+      ]
+cleanupRecordFields ty t@(FV _) = composSafeOp (cleanupRecordFields ty) t
+cleanupRecordFields _ t = t
 
+convert :: G.Grammar -> Term -> LinValue
 convert gr = convert' gr []
 
+convert' :: G.Grammar -> [Ident] -> Term -> LinValue
 convert' gr vs = ppT
   where
     ppT0 = convert' gr vs
@@ -162,24 +178,24 @@ convert' gr vs = ppT
         S t p -> selection (ppT t) (ppT p)
         C t1 t2 -> concatValue (ppT t1) (ppT t2)
         App f a -> ap (ppT f) (ppT a)
-        R r -> RecordValue (fields r)
+        R r -> RecordValue (fields (sortRec r))
         P t l -> projection (ppT t) (lblId l)
         Vr x -> VarValue (gId x)
         Cn x -> VarValue (gId x) -- hmm
         Con c -> ParamConstant (Param (gId c) [])
         Sort k -> VarValue (gId k)
-        EInt n -> LiteralValue (IntConstant n)
-        Q (m,n) -> if m==cPredef then ppPredef n else VarValue ((gQId m n))
-        QC (m,n) -> ParamConstant (Param ((gQId m n)) [])
-        K s -> LiteralValue (StrConstant s)
-        Empty -> LiteralValue (StrConstant "")
+        EInt n -> LiteralValue (LInt n)
+        Q (m,n) -> if m==cPredef then ppPredef n else VarValue (gQId m n)
+        QC (m,n) -> ParamConstant (Param (gQId m n) [])
+        K s -> LiteralValue (LStr s)
+        Empty -> LiteralValue (LStr "")
         FV ts -> VariantValue (map ppT ts)
         Alts t' vs -> alts vs (ppT t')
-        _ -> error $ "convert' "++show t
+        _ -> error $ "convert' ppT: " ++ show t
 
     ppCase (p,t) = TableRow (ppP p) (ppTv (patVars p++vs) t)
 
-    ppPredef n =
+    ppPredef n = error "TODO: ppPredef" {-
       case predef n of
         Ok BIND       -> p "BIND"
         Ok SOFT_BIND  -> p "SOFT_BIND"
@@ -188,12 +204,12 @@ convert' gr vs = ppT
         Ok ALL_CAPIT  -> p "ALL_CAPIT"
         _ -> VarValue (gQId cPredef n) -- hmm
       where
-       p = PredefValue . PredefId
-      
+       p = PredefValue . PredefId . rawIdentS
+-}
     ppP p =
       case p of
         PC c ps -> ParamPattern (Param (gId c) (map ppP ps))
-        PP (m,c) ps -> ParamPattern (Param ((gQId m c)) (map ppP ps))
+        PP (m,c) ps -> ParamPattern (Param (gQId m c) (map ppP ps))
         PR r -> RecordPattern (fields r) {-
         PW -> WildPattern
         PV x -> VarP x
@@ -202,6 +218,7 @@ convert' gr vs = ppT
         PFloat x -> Lit (show x)
         PT _ p -> ppP p
         PAs x p -> AsP x (ppP p) -}
+        _ -> error $ "convert' ppP: " ++ show p
       where
         fields = map field . filter (not.isLockLabel.fst)
         field (l,p) = RecordRow (lblId l) (ppP p)
@@ -217,13 +234,13 @@ convert' gr vs = ppT
         pre (K s) = [s]
         pre Empty = [""]  -- Empty == K ""
         pre (Strs ts) = concatMap pre ts
-        pre (EPatt p) = pat p
-        pre t = error $ "pre "++show t
+        pre (EPatt _ _ p) = pat p
+        pre t = error $ "convert' alts pre: " ++ show t
 
         pat (PString s) = [s]
         pat (PAlt p1 p2) = pat p1++pat p2
-        pat (PSeq p1 p2) = [s1++s2 | s1<-pat p1, s2<-pat p2]
-        pat p = error $ "pat "++show p
+        pat (PSeq _ _ p1 _ _ p2) = [s1++s2 | s1<-pat p1, s2<-pat p2]
+        pat p = error $ "convert' alts pat: "++show p
 
     fields = map field . filter (not.isLockLabel.fst)
     field (l,(_,t)) = RecordRow (lblId l) (ppT t)
@@ -236,28 +253,32 @@ convert' gr vs = ppT
                  ParamConstant (Param p (ps++[a]))
                _ -> error $ "convert' ap: "++render (ppA f <+> ppA a)
 
+concatValue :: LinValue -> LinValue -> LinValue
 concatValue v1 v2 =
   case (v1,v2) of
-    (LiteralValue (StrConstant ""),_) -> v2
-    (_,LiteralValue (StrConstant "")) -> v1
+    (LiteralValue (LStr ""),_) -> v2
+    (_,LiteralValue (LStr "")) -> v1
     _ -> ConcatValue v1 v2
 
 -- | Smart constructor for projections
-projection r l = maybe (Projection r l) id (proj r l)
+projection :: LinValue -> LabelId -> LinValue
+projection r l = fromMaybe (Projection r l) (proj r l)
 
+proj :: LinValue -> LabelId -> Maybe LinValue
 proj r l =
   case r of
-    RecordValue r -> case [v|RecordRow l' v<-r,l'==l] of
+    RecordValue r -> case [v | RecordRow l' v <- r, l'==l] of
                           [v] -> Just v
                           _ -> Nothing
     _ -> Nothing
 
 -- | Smart constructor for selections
+selection :: LinValue -> LinValue -> LinValue
 selection t v =
   -- Note: impossible cases can become possible after grammar transformation
   case t of
     TableValue tt r ->
-        case nub [rv|TableRow _ rv<-keep] of
+        case nub [rv | TableRow _ rv <- keep] of
           [rv] -> rv
           _ -> Selection (TableValue tt r') v
       where
@@ -276,13 +297,16 @@ selection t v =
         (keep,discard) = partition (mightMatchRow v) r
     _ -> Selection t v
 
+impossible :: LinValue -> LinValue
 impossible = CommentedValue "impossible"
 
+mightMatchRow :: LinValue -> TableRow rhs -> Bool
 mightMatchRow v (TableRow p _) =
   case p of
     WildPattern -> True
     _ -> mightMatch v p
 
+mightMatch :: LinValue -> LinPattern -> Bool
 mightMatch v p =
   case v of
     ConcatValue _ _ -> False
@@ -294,16 +318,18 @@ mightMatch v p =
     RecordValue rv ->
       case p of
         RecordPattern rp ->
-          and [maybe False (flip mightMatch p) (proj v l) | RecordRow l p<-rp]
+          and [maybe False (`mightMatch` p) (proj v l) | RecordRow l p<-rp]
         _ -> False
     _ -> True
 
+patVars :: Patt -> [Ident]
 patVars p =
   case p of
     PV x -> [x]
     PAs x p -> x:patVars p
     _ -> collectPattOp patVars p
 
+convType :: Term -> LinType
 convType = ppT
   where
     ppT t =
@@ -315,9 +341,9 @@ convType = ppT
         Sort k -> convSort k
 --      EInt n -> tcon0 (identS ("({-"++show n++"-})")) -- type level numeric literal
         FV (t:ts) -> ppT t -- !!
-        QC (m,n) -> ParamType (ParamTypeId ((gQId m n)))
-        Q (m,n) -> ParamType (ParamTypeId ((gQId m n)))
-        _ -> error $ "Missing case in convType for: "++show t
+        QC (m,n) -> ParamType (ParamTypeId (gQId m n))
+        Q (m,n) -> ParamType (ParamTypeId (gQId m n))
+        _ -> error $ "convType ppT: " ++ show t
 
     convFields = map convField . filter (not.isLockLabel.fst)
     convField (l,r) = RecordRow (lblId l) (ppT r)
@@ -326,15 +352,20 @@ convType = ppT
                    "Float" -> FloatType
                    "Int" -> IntType
                    "Str" -> StrType
-                   _ -> error ("convSort "++show k)
+                   _ -> error $ "convType convSort: " ++ show k
 
+toParamType :: Term -> ParamType
 toParamType t = case convType t of
                   ParamType pt -> pt
-                  _ -> error ("toParamType "++show t)
+                  _ -> error $ "toParamType: " ++ show t
 
+toParamId :: Term -> ParamId
 toParamId t = case toParamType t of
                    ParamTypeId p -> p
 
+paramType :: G.Grammar
+             -> (ModuleName, Ident)
+             -> ((S.Set (ModuleName, Ident), S.Set QIdent), [ParamDef])
 paramType gr q@(_,n) =
     case lookupOrigInfo gr q of
       Ok (m,ResParam (Just (L _ ps)) _)
@@ -342,7 +373,7 @@ paramType gr q@(_,n) =
          ((S.singleton (m,n),argTypes ps),
           [ParamDef name (map (param m) ps)]
          )
-       where name = (gQId m n)
+       where name = gQId m n
       Ok (m,ResOper  _ (Just (L _ t)))
         | m==cPredef && n==cInts ->
            ((S.empty,S.empty),[]) {-
@@ -350,40 +381,44 @@ paramType gr q@(_,n) =
             [Type (ConAp ((gQId m n)) [identS "n"]) (TId (identS "Int"))])-}
         | otherwise ->
            ((S.singleton (m,n),paramTypes gr t),
-            [ParamAliasDef ((gQId m n)) (convType t)])
+            [ParamAliasDef (gQId m n) (convType t)])
       _ -> ((S.empty,S.empty),[])
   where
-    param m (n,ctx) = Param ((gQId m n)) [toParamId t|(_,_,t)<-ctx]
+    param m (n,ctx) = Param (gQId m n) [toParamId t|(_,_,t)<-ctx]
     argTypes = S.unions . map argTypes1
     argTypes1 (n,ctx) = S.unions [paramTypes gr t|(_,_,t)<-ctx]
 
-lblId = LabelId . render -- hmm
-modId (MN m) = ModId (showIdent m)
+lblId :: Label -> C.LabelId
+lblId (LIdent ri) = LabelId ri
+lblId (LVar i) = LabelId (rawIdentS (show i)) -- hmm
 
-class FromIdent i where gId :: Ident -> i
+modId :: ModuleName -> C.ModId
+modId (MN m) = ModId (ident2raw m)
+
+class FromIdent i where
+  gId :: Ident -> i
 
 instance FromIdent VarId where
-  gId i = if isWildIdent i then Anonymous else VarId (showIdent i)
+  gId i = if isWildIdent i then Anonymous else VarId (ident2raw i)
 
-instance FromIdent C.FunId where gId = C.FunId . showIdent
-instance FromIdent CatId where gId = CatId . showIdent
+instance FromIdent C.FunId where gId = C.FunId . ident2raw
+instance FromIdent CatId where gId = CatId . ident2raw
 instance FromIdent ParamId where gId = ParamId . unqual
 instance FromIdent VarValueId where gId = VarValueId . unqual
 
-class FromIdent i => QualIdent i where gQId :: ModuleName -> Ident -> i
+class FromIdent i => QualIdent i where
+  gQId :: ModuleName -> Ident -> i
 
-instance QualIdent ParamId    where gQId m n = ParamId (qual m n)
+instance QualIdent ParamId where gQId m n = ParamId (qual m n)
 instance QualIdent VarValueId where gQId m n = VarValueId (qual m n)
 
-qual m n = Qual (modId m) (showIdent n)
-unqual n = Unqual (showIdent n)
+qual :: ModuleName -> Ident -> QualId
+qual m n = Qual (modId m) (ident2raw n)
 
+unqual :: Ident -> QualId
+unqual n = Unqual (ident2raw n)
+
+convFlags :: G.Grammar -> ModuleName -> Flags
 convFlags gr mn =
-  Flags [(n,convLit v) |
+  Flags [(rawIdentS n,v) |
          (n,v)<-err (const []) (optionsPGF.mflags) (lookupModule gr mn)]
-  where
-    convLit l =
-      case l of
-        LStr s -> Str s
-        LInt i -> C.Int i
-        LFlt d -> Flt d

src/compiler/GF/Compile/GrammarToPGF.hs

@@ -1,17 +1,14 @@
-{-# LANGUAGE BangPatterns, FlexibleContexts #-}
-module GF.Compile.GrammarToPGF (mkCanon2pgf) where
+{-# LANGUAGE BangPatterns, FlexibleContexts, MagicHash #-}
+module GF.Compile.GrammarToPGF (grammar2PGF) where
 
---import GF.Compile.Export
 import GF.Compile.GeneratePMCFG
 import GF.Compile.GenerateBC
+import GF.Compile.OptimizePGF
 
-import PGF(CId,mkCId,utf8CId)
-import PGF.Internal(fidInt,fidFloat,fidString,fidVar)
-import PGF.Internal(updateProductionIndices)
-import qualified PGF.Internal as C
-import qualified PGF.Internal as D
+import PGF2 hiding (mkType)
+import PGF2.Transactions
 import GF.Grammar.Predef
-import GF.Grammar.Grammar
+import GF.Grammar.Grammar hiding (Production)
 import qualified GF.Grammar.Lookup as Look
 import qualified GF.Grammar as A
 import qualified GF.Grammar.Macros as GM
@@ -22,111 +19,162 @@ import GF.Infra.UseIO (IOE)
 import GF.Data.Operations
 
 import Data.List
+import Data.Char
 import qualified Data.Set as Set
 import qualified Data.Map as Map
 import qualified Data.IntMap as IntMap
 import Data.Array.IArray
+import Data.Maybe(fromMaybe)
+import System.FilePath
+import System.Directory
 
+import GHC.Prim
+import GHC.Base(getTag)
 
-mkCanon2pgf :: Options -> SourceGrammar -> ModuleName -> IOE D.PGF
-mkCanon2pgf opts gr am = do
-  (an,abs) <- mkAbstr am
-  cncs     <- mapM mkConcr (allConcretes gr am)
-  return $ updateProductionIndices (D.PGF Map.empty an abs (Map.fromList cncs))
+grammar2PGF :: Options -> SourceGrammar -> ModuleName -> Map.Map PGF2.Fun Double -> IO PGF
+grammar2PGF opts gr am probs = do
+  gr <- mkAbstr am probs
+  return gr {-do
+  cnc_infos <- getConcreteInfos gr am
+  return $
+    build (let gflags   = if flag optSplitPGF opts 
+                            then [("split", LStr "true")]
+                            else []
+               (an,abs) = mkAbstr am probs
+               cncs     = map (mkConcr opts abs) cnc_infos
+           in newPGF gflags an abs cncs)-}
   where
-    cenv = resourceValues opts gr
+    aflags = err (const noOptions) mflags (lookupModule gr am)
 
-    mkAbstr am = return (mi2i am, D.Abstr flags funs cats)
+    mkAbstr :: ModuleName -> Map.Map PGF2.Fun Double -> IO PGF
+    mkAbstr am probs = do
+      let abs_name = mi2i am
+      mb_ngf_path <-
+         if snd (flag optLinkTargets opts)
+           then do let fname = maybe id (</>)
+                                     (flag optOutputDir opts)
+                                     (fromMaybe abs_name (flag optName opts)<.>"ngf")
+                   exists <- doesFileExist fname
+                   if exists
+                     then removeFile fname
+                     else return ()
+                   putStr ("(Boot image "++fname++") ")
+                   return (Just fname)
+           else do return Nothing
+      gr <- newNGF abs_name mb_ngf_path
+      modifyPGF gr $ do
+        sequence_ [setAbstractFlag name value | (name,value) <- flags]
+        sequence_ [createCategory c ctxt p | (c,ctxt,p) <- cats]
+        sequence_ [createFunction f ty arity p | (f,ty,arity,_,p) <- funs]
       where
-        aflags = err (const noOptions) mflags (lookupModule gr am)
-
         adefs =
             [((cPredefAbs,c), AbsCat (Just (L NoLoc []))) | c <- [cFloat,cInt,cString]] ++ 
             Look.allOrigInfos gr am
 
-        flags = Map.fromList [(mkCId f,x) | (f,x) <- optionsPGF aflags]
+        flags = optionsPGF aflags
 
-        funs = Map.fromList [(i2i f, (mkType [] ty, arity, mkDef gr arity mdef, 0)) | 
+        toLogProb = realToFrac . negate . log
+
+        cats = [(c', snd (mkContext [] cont), toLogProb (fromMaybe 0 (Map.lookup c' probs))) |
+                                   ((m,c),AbsCat (Just (L _ cont))) <- adefs, let c' = i2i c]
+
+        funs = [(f', mkType [] ty, arity, bcode, toLogProb (fromMaybe 0 (Map.lookup f' funs_probs))) |
                                    ((m,f),AbsFun (Just (L _ ty)) ma mdef _) <- adefs,
-                                   let arity = mkArity ma mdef ty]
-
-        cats = Map.fromList [(i2i c, (snd (mkContext [] cont),catfuns c, 0)) |
-                                   ((m,c),AbsCat (Just (L _ cont))) <- adefs]
-
-        catfuns cat =
-              [(0,i2i f) | ((m,f),AbsFun (Just (L _ ty)) _ _ (Just True)) <- adefs, snd (GM.valCat ty) == cat]
-
-    mkConcr cm = do
+                                   let arity = mkArity ma mdef ty,
+                                   let bcode = mkDef gr arity mdef,
+                                   let f' = i2i f]
+                                   
+        funs_probs = (Map.fromList . concat . Map.elems . fmap pad . Map.fromListWith (++))
+                        [(i2i cat,[(i2i f,Map.lookup f' probs)]) | ((m,f),AbsFun (Just (L _ ty)) _ _ _) <- adefs,
+                                                                   let (_,(_,cat),_) = GM.typeForm ty,
+                                                                   let f' = i2i f]
+          where
+            pad :: [(a,Maybe Double)] -> [(a,Double)]
+            pad pfs = [(f,fromMaybe deflt mb_p) | (f,mb_p) <- pfs]
+              where
+                deflt = case length [f | (f,Nothing) <- pfs] of
+                          0 -> 0
+                          n -> max 0 ((1 - sum [d | (f,Just d) <- pfs]) / fromIntegral n)
+{-
+    mkConcr opts abs (cm,ex_seqs,cdefs) =
       let cflags = err (const noOptions) mflags (lookupModule gr cm)
           ciCmp | flag optCaseSensitive cflags = compare
-                | otherwise                    = C.compareCaseInsensitve
+                | otherwise                    = compareCaseInsensitive
 
-      (ex_seqs,cdefs) <- addMissingPMCFGs
-                            Map.empty 
-                            ([((cPredefAbs,c), CncCat (Just (L NoLoc GM.defLinType)) Nothing Nothing Nothing Nothing) | c <- [cInt,cFloat,cString]] ++
-                             Look.allOrigInfos gr cm)
+          flags = optionsPGF aflags
 
-      let flags = Map.fromList [(mkCId f,x) | (f,x) <- optionsPGF cflags]
-
-          seqs = (mkArray . C.sortNubBy ciCmp . concat) $
-                     (Map.keys ex_seqs : [maybe [] elems (mseqs mi) | (m,mi) <- allExtends gr cm])
-
-          ex_seqs_arr = mkMapArray ex_seqs :: Array SeqId Sequence
+          seqs = (mkSetArray . Set.fromList . concat) $
+                       (elems (ex_seqs :: Array SeqId [Symbol]) : [maybe [] elems (mseqs mi) | (m,mi) <- allExtends gr cm])
 
           !(!fid_cnt1,!cnccats) = genCncCats gr am cm cdefs
+          cnccat_ranges = Map.fromList (map (\(cid,s,e,_) -> (cid,(s,e))) cnccats)
           !(!fid_cnt2,!productions,!lindefs,!linrefs,!cncfuns)
-                                = genCncFuns gr am cm ex_seqs_arr ciCmp seqs cdefs fid_cnt1 cnccats
+                                = genCncFuns gr am cm ex_seqs ciCmp seqs cdefs fid_cnt1 cnccat_ranges
 
           printnames = genPrintNames cdefs
-      return (mi2i cm, D.Concr flags
-                              printnames
-                              cncfuns
-                              lindefs
-                              linrefs
-                              seqs
-                              productions
-                              IntMap.empty
-                              Map.empty
-                              cnccats
-                              IntMap.empty
-                              fid_cnt2)
+
+          startCat = (fromMaybe "S" (flag optStartCat aflags))
+
+          (lindefs',linrefs',productions',cncfuns',sequences',cnccats') =
+               (if flag optOptimizePGF opts then optimizePGF startCat else id)
+                    (lindefs,linrefs,productions,cncfuns,elems seqs,cnccats)
+
+      in (mi2i cm, newConcr abs
+                            flags
+                            printnames
+                            lindefs'
+                            linrefs'
+                            productions'
+                            cncfuns'
+                            sequences'
+                            cnccats'
+                            fid_cnt2)
+
+    getConcreteInfos gr am = mapM flatten (allConcretes gr am)
       where
+        flatten cm = do
+          (seqs,infos) <- addMissingPMCFGs cm Map.empty
+                                           (lit_infos ++ Look.allOrigInfos gr cm)
+          return (cm,mkMapArray seqs :: Array SeqId [Symbol],infos)
+
+        lit_infos = [((cPredefAbs,c), CncCat (Just (L NoLoc GM.defLinType)) Nothing Nothing Nothing Nothing) | c <- [cInt,cFloat,cString]]
+
         -- if some module was compiled with -no-pmcfg, then
         -- we have to create the PMCFG code just before linking
-        addMissingPMCFGs seqs []                  = return (seqs,[])
-        addMissingPMCFGs seqs (((m,id), info):is) = do
-          (seqs,info) <- addPMCFG opts gr cenv Nothing am cm seqs id info
-          (seqs,is  ) <- addMissingPMCFGs seqs is
-          return (seqs, ((m,id), info) : is)
+        addMissingPMCFGs cm seqs []                  = return (seqs,[])
+        addMissingPMCFGs cm seqs (((m,id), info):is) = do
+          (seqs,info)  <- addPMCFG opts gr cenv Nothing am cm seqs id info
+          (seqs,infos) <- addMissingPMCFGs cm seqs is
+          return (seqs, ((m,id), info) : infos)
+-}
+i2i :: Ident -> String
+i2i = showIdent
 
-i2i :: Ident -> CId
-i2i = utf8CId . ident2utf8
-
-mi2i :: ModuleName -> CId
+mi2i :: ModuleName -> String
 mi2i (MN i) = i2i i
 
-mkType :: [Ident] -> A.Type -> C.Type
+mkType :: [Ident] -> A.Type -> PGF2.Type
 mkType scope t =
   case GM.typeForm t of
     (hyps,(_,cat),args) -> let (scope',hyps') = mkContext scope hyps
-                           in C.DTyp hyps' (i2i cat) (map (mkExp scope') args)
+                           in DTyp hyps' (i2i cat) (map (mkExp scope') args)
 
-mkExp :: [Ident] -> A.Term -> C.Expr
-mkExp scope t = 
+mkExp :: [Ident] -> A.Term -> Expr
+mkExp scope t =
   case t of
-    Q (_,c)  -> C.EFun (i2i c)
-    QC (_,c) -> C.EFun (i2i c)
+    Q (_,c)  -> EFun (i2i c)
+    QC (_,c) -> EFun (i2i c)
     Vr x     -> case lookup x (zip scope [0..]) of
-                  Just i  -> C.EVar  i
-                  Nothing -> C.EMeta 0
-    Abs b x t-> C.EAbs b (i2i x) (mkExp (x:scope) t)
-    App t1 t2-> C.EApp (mkExp scope t1) (mkExp scope t2)
-    EInt i   -> C.ELit (C.LInt (fromIntegral i))
-    EFloat f -> C.ELit (C.LFlt f)
-    K s      -> C.ELit (C.LStr s)
-    Meta i   -> C.EMeta i
-    _        -> C.EMeta 0
-
+                  Just i  -> EVar  i
+                  Nothing -> EMeta 0
+    Abs b x t-> EAbs b (i2i x) (mkExp (x:scope) t)
+    App t1 t2-> EApp (mkExp scope t1) (mkExp scope t2)
+    EInt i   -> ELit (LInt (fromIntegral i))
+    EFloat f -> ELit (LFlt f)
+    K s      -> ELit (LStr s)
+    Meta i   -> EMeta i
+    _        -> EMeta 0
+{-
 mkPatt scope p = 
   case p of
     A.PP (_,c) ps->let (scope',ps') = mapAccumL mkPatt scope ps
@@ -141,67 +189,65 @@ mkPatt scope p =
     A.PImplArg p-> let (scope',p') = mkPatt scope p
                    in (scope',C.PImplArg p')
     A.PTilde t  -> (  scope,C.PTilde (mkExp scope t))
+-}
 
-mkContext :: [Ident] -> A.Context -> ([Ident],[C.Hypo])
+mkContext :: [Ident] -> A.Context -> ([Ident],[PGF2.Hypo])
 mkContext scope hyps = mapAccumL (\scope (bt,x,ty) -> let ty' = mkType scope ty
                                                       in if x == identW
                                                            then (  scope,(bt,i2i x,ty'))
                                                            else (x:scope,(bt,i2i x,ty'))) scope hyps 
 
-mkDef gr arity (Just eqs) = Just ([C.Equ ps' (mkExp scope' e) | L _ (ps,e) <- eqs, let (scope',ps') = mapAccumL mkPatt [] ps]
-                                 ,generateByteCode gr arity eqs
-                                 )
-mkDef gr arity Nothing    = Nothing
+mkDef gr arity (Just eqs) = generateByteCode gr arity eqs
+mkDef gr arity Nothing    = []
 
 mkArity (Just a) _        ty = a   -- known arity, i.e. defined function
 mkArity Nothing  (Just _) ty = 0   -- defined function with no arity - must be an axiom
 mkArity Nothing  _        ty = let (ctxt, _, _) = GM.typeForm ty  -- constructor
                                in length ctxt
-
-genCncCats gr am cm cdefs =
-  let (index,cats) = mkCncCats 0 cdefs
-  in (index, Map.fromList cats)
+{-
+genCncCats gr am cm cdefs = mkCncCats 0 cdefs
   where
     mkCncCats index []                                                = (index,[])
     mkCncCats index (((m,id),CncCat (Just (L _ lincat)) _ _ _ _):cdefs) 
       | id == cInt    = 
-            let cc            = pgfCncCat gr lincat fidInt
+            let cc            = pgfCncCat gr (i2i id) lincat fidInt
                 (index',cats) = mkCncCats index cdefs
-            in (index', (i2i id,cc) : cats)
+            in (index', cc : cats)
       | id == cFloat  = 
-            let cc            = pgfCncCat gr lincat fidFloat
+            let cc            = pgfCncCat gr (i2i id) lincat fidFloat
                 (index',cats) = mkCncCats index cdefs
-            in (index', (i2i id,cc) : cats)
+            in (index', cc : cats)
       | id == cString = 
-            let cc            = pgfCncCat gr lincat fidString
+            let cc            = pgfCncCat gr (i2i id) lincat fidString
                 (index',cats) = mkCncCats index cdefs
-            in (index', (i2i id,cc) : cats)
+            in (index', cc : cats)
       | otherwise     =
-            let cc@(C.CncCat _s e _) = pgfCncCat gr lincat index
-                (index',cats)        = mkCncCats (e+1) cdefs
-            in (index', (i2i id,cc) : cats)
-    mkCncCats index (_                      :cdefs) = mkCncCats index cdefs
+            let cc@(_, _s, e, _) = pgfCncCat gr (i2i id) lincat index
+                (index',cats)    = mkCncCats (e+1) cdefs
+            in (index', cc : cats)
+    mkCncCats index (_                                          :cdefs) = mkCncCats index cdefs
 
 genCncFuns :: Grammar
            -> ModuleName
            -> ModuleName
-           -> Array SeqId Sequence
-           -> (Sequence -> Sequence -> Ordering)
-           -> Array SeqId Sequence
+           -> Array SeqId [Symbol]
+           -> ([Symbol] -> [Symbol] -> Ordering)
+           -> Array SeqId [Symbol]
            -> [(QIdent, Info)]
            -> FId
-           -> Map.Map CId D.CncCat
+           -> Map.Map PGF2.Cat (Int,Int)
            -> (FId,
-               IntMap.IntMap (Set.Set D.Production),
-               IntMap.IntMap [FunId],
-               IntMap.IntMap [FunId],
-               Array FunId D.CncFun)
-genCncFuns gr am cm ex_seqs ciCmp seqs cdefs fid_cnt cnccats =
+               [(FId, [Production])],
+               [(FId, [FunId])],
+               [(FId, [FunId])],
+               [(PGF2.Fun,[SeqId])])
+genCncFuns gr am cm ex_seqs ciCmp seqs cdefs fid_cnt cnccat_ranges =
   let (fid_cnt1,funs_cnt1,funs1,lindefs,linrefs) = mkCncCats cdefs fid_cnt  0 [] IntMap.empty IntMap.empty
-      (fid_cnt2,funs_cnt2,funs2,prods)           = mkCncFuns cdefs fid_cnt1 funs_cnt1 funs1 lindefs Map.empty IntMap.empty
-  in (fid_cnt2,prods,lindefs,linrefs,array (0,funs_cnt2-1) funs2)
+      (fid_cnt2,funs_cnt2,funs2,prods0)          = mkCncFuns cdefs fid_cnt1 funs_cnt1 funs1 lindefs Map.empty IntMap.empty
+      prods                                      = [(fid,Set.toList prodSet) | (fid,prodSet) <- IntMap.toList prods0]
+  in (fid_cnt2,prods,IntMap.toList lindefs,IntMap.toList linrefs,reverse funs2)
   where
-    mkCncCats []                                                        fid_cnt funs_cnt funs lindefs linrefs =
+    mkCncCats []                                                          fid_cnt funs_cnt funs lindefs linrefs =
       (fid_cnt,funs_cnt,funs,lindefs,linrefs)
     mkCncCats (((m,id),CncCat _ _ _ _ (Just (PMCFG prods0 funs0))):cdefs) fid_cnt funs_cnt funs lindefs linrefs =
       let !funs_cnt' = let (s_funid, e_funid) = bounds funs0
@@ -210,17 +256,16 @@ genCncFuns gr am cm ex_seqs ciCmp seqs cdefs fid_cnt cnccats =
           linrefs'   = foldl' (toLinRef (am,id) funs_cnt) linrefs prods0
           funs'      = foldl' (toCncFun funs_cnt (m,mkLinDefId id)) funs (assocs funs0)
       in mkCncCats cdefs fid_cnt funs_cnt' funs' lindefs' linrefs'
-    mkCncCats (_                                                :cdefs) fid_cnt funs_cnt funs lindefs linrefs =
+    mkCncCats (_                                                  :cdefs) fid_cnt funs_cnt funs lindefs linrefs =
       mkCncCats cdefs fid_cnt funs_cnt funs lindefs linrefs
 
     mkCncFuns []                                                        fid_cnt funs_cnt funs lindefs crc prods =
       (fid_cnt,funs_cnt,funs,prods)
     mkCncFuns (((m,id),CncFun _ _ _ (Just (PMCFG prods0 funs0))):cdefs) fid_cnt funs_cnt funs lindefs crc prods =
-      let ---Ok ty_C        = fmap GM.typeForm (Look.lookupFunType gr am id)
-          ty_C           = err error (\x -> x) $ fmap GM.typeForm (Look.lookupFunType gr am id)
+      let ty_C           = err error (\x -> x) $ fmap GM.typeForm (Look.lookupFunType gr am id)
           !funs_cnt'     = let (s_funid, e_funid) = bounds funs0
                            in funs_cnt+(e_funid-s_funid+1)
-          !(fid_cnt',crc',prods') 
+          !(fid_cnt',crc',prods')
                          = foldl' (toProd lindefs ty_C funs_cnt)
                                   (fid_cnt,crc,prods) prods0
           funs'          = foldl' (toCncFun funs_cnt (m,id)) funs (assocs funs0)
@@ -228,23 +273,23 @@ genCncFuns gr am cm ex_seqs ciCmp seqs cdefs fid_cnt cnccats =
     mkCncFuns (_                                                :cdefs) fid_cnt funs_cnt funs lindefs crc prods = 
       mkCncFuns cdefs fid_cnt funs_cnt funs lindefs crc prods
 
-    toProd lindefs (ctxt_C,res_C,_) offs st (Production fid0 funid0 args0) =
-      let !((fid_cnt,crc,prods),args) = mapAccumL mkArg st (zip ctxt_C args0) 
-          set0    = Set.fromList (map (C.PApply (offs+funid0)) (sequence args))
+    toProd lindefs (ctxt_C,res_C,_) offs st (A.Production fid0 funid0 args0) =
+      let !((fid_cnt,crc,prods),args) = mapAccumL mkArg st (zip ctxt_C args0)
+          set0    = Set.fromList (map (PApply (offs+funid0)) (sequence args))
           fid     = mkFId res_C fid0
           !prods' = case IntMap.lookup fid prods of
                      Just set -> IntMap.insert fid (Set.union set0 set) prods
                      Nothing  -> IntMap.insert fid set0 prods
       in (fid_cnt,crc,prods')
       where
-        mkArg st@(fid_cnt,crc,prods) ((_,_,ty),fid0s ) =
+        mkArg st@(fid_cnt,crc,prods) ((_,_,ty),fid0s) =
           case fid0s of
-            [fid0] -> (st,map (flip C.PArg (mkFId arg_C fid0)) ctxt)
+            [fid0] -> (st,map (flip PArg (mkFId arg_C fid0)) ctxt)
             fid0s  -> case Map.lookup fids crc of
-                        Just fid -> (st,map (flip C.PArg fid) ctxt)
+                        Just fid -> (st,map (flip PArg fid) ctxt)
                         Nothing  -> let !crc'   = Map.insert fids fid_cnt crc
-                                        !prods' = IntMap.insert fid_cnt (Set.fromList (map C.PCoerce fids)) prods
-                                    in ((fid_cnt+1,crc',prods'),map (flip C.PArg fid_cnt) ctxt)
+                                        !prods' = IntMap.insert fid_cnt (Set.fromList (map PCoerce fids)) prods
+                                    in ((fid_cnt+1,crc',prods'),map (flip PArg fid_cnt) ctxt)
           where
             (hargs_C,arg_C) = GM.catSkeleton ty
             ctxt = mapM (mkCtxt lindefs) hargs_C
@@ -252,14 +297,14 @@ genCncFuns gr am cm ex_seqs ciCmp seqs cdefs fid_cnt cnccats =
 
     mkLinDefId id = prefixIdent "lindef " id
 
-    toLinDef res offs lindefs (Production fid0 funid0 args) =
+    toLinDef res offs lindefs (A.Production fid0 funid0 args) =
       if args == [[fidVar]]
         then IntMap.insertWith (++) fid [offs+funid0] lindefs
         else lindefs
       where
         fid = mkFId res fid0
 
-    toLinRef res offs linrefs (Production fid0 funid0 [fargs]) =
+    toLinRef res offs linrefs (A.Production fid0 funid0 [fargs]) =
       if fid0 == fidVar
         then foldr (\fid -> IntMap.insertWith (++) fid [offs+funid0]) linrefs fids
         else linrefs
@@ -267,20 +312,20 @@ genCncFuns gr am cm ex_seqs ciCmp seqs cdefs fid_cnt cnccats =
         fids = map (mkFId res) fargs
 
     mkFId (_,cat) fid0 =
-      case Map.lookup (i2i cat) cnccats of
-        Just (C.CncCat s e _) -> s+fid0
-        Nothing               -> error ("GrammarToPGF.mkFId: missing category "++showIdent cat)
+      case Map.lookup (i2i cat) cnccat_ranges of
+        Just (s,e) -> s+fid0
+        Nothing    -> error ("GrammarToPGF.mkFId: missing category "++showIdent cat)
 
     mkCtxt lindefs (_,cat) =
-      case Map.lookup (i2i cat) cnccats of
-        Just (C.CncCat s e _) -> [(C.fidVar,fid) | fid <- [s..e], Just _ <- [IntMap.lookup fid lindefs]]
-        Nothing               -> error "GrammarToPGF.mkCtxt failed"
+      case Map.lookup (i2i cat) cnccat_ranges of
+        Just (s,e) -> [(fid,fid) | fid <- [s..e], Just _ <- [IntMap.lookup fid lindefs]]
+        Nothing    -> error "GrammarToPGF.mkCtxt failed"
 
     toCncFun offs (m,id) funs (funid0,lins0) =
       let mseqs = case lookupModule gr m of
                     Ok (ModInfo{mseqs=Just mseqs}) -> mseqs
                     _                              -> ex_seqs
-      in (offs+funid0,C.CncFun (i2i id) (amap (newIndex mseqs) lins0)):funs                                          
+      in (i2i id, map (newIndex mseqs) (elems lins0)):funs
       where
         newIndex mseqs i = binSearch (mseqs ! i) seqs (bounds seqs)
 
@@ -293,8 +338,9 @@ genCncFuns gr am cm ex_seqs ciCmp seqs cdefs fid_cnt cnccats =
           where
             k = (i+j) `div` 2
 
+
 genPrintNames cdefs =
-  Map.fromAscList [(i2i id, name) | ((m,id),info) <- cdefs, name <- prn info]
+  [(i2i id, name) | ((m,id),info) <- cdefs, name <- prn info]
   where
     prn (CncFun _ _   (Just (L _ tr)) _) = [flatten tr]
     prn (CncCat _ _ _ (Just (L _ tr)) _) = [flatten tr]
@@ -306,3 +352,119 @@ genPrintNames cdefs =
 
 mkArray    lst = listArray (0,length lst-1) lst
 mkMapArray map = array (0,Map.size map-1) [(v,k) | (k,v) <- Map.toList map]
+mkSetArray set = listArray (0,Set.size set-1) (Set.toList set)
+
+-- The following is a version of Data.List.sortBy which together
+-- with the sorting also eliminates duplicate values 
+sortNubBy cmp = mergeAll . sequences
+  where
+    sequences (a:b:xs) =
+      case cmp a b of
+        GT -> descending b [a]  xs
+        EQ -> sequences (b:xs)
+        LT -> ascending  b (a:) xs
+    sequences xs = [xs]
+
+    descending a as []     = [a:as]
+    descending a as (b:bs) =
+      case cmp a b of
+        GT -> descending b (a:as) bs
+        EQ -> descending a as bs
+        LT -> (a:as) : sequences (b:bs)
+
+    ascending a as []     = let !x = as [a]
+                            in [x]
+    ascending a as (b:bs) =
+      case cmp a b of
+        GT -> let !x = as [a]
+              in x : sequences (b:bs)
+        EQ -> ascending a as bs
+        LT -> ascending b (\ys -> as (a:ys)) bs
+
+    mergeAll [x] = x
+    mergeAll xs  = mergeAll (mergePairs xs)
+
+    mergePairs (a:b:xs) = let !x = merge a b
+                          in x : mergePairs xs
+    mergePairs xs       = xs
+
+    merge as@(a:as') bs@(b:bs') =
+      case cmp a b of
+        GT -> b:merge as  bs'
+        EQ -> a:merge as' bs'
+        LT -> a:merge as' bs
+    merge [] bs         = bs
+    merge as []         = as
+
+-- The following function does case-insensitive comparison of sequences.
+-- This is used to allow case-insensitive parsing, while
+-- the linearizer still has access to the original cases.
+
+compareCaseInsensitive []     []     = EQ
+compareCaseInsensitive []     _      = LT
+compareCaseInsensitive _      []     = GT
+compareCaseInsensitive (x:xs) (y:ys) =
+  case compareSym x y of
+    EQ -> compareCaseInsensitive xs ys
+    x  -> x
+  where
+    compareSym s1 s2 =
+      case s1 of
+        SymCat d1 r1
+          -> case s2 of
+               SymCat d2 r2
+                 -> case compare d1 d2 of
+                      EQ -> r1 `compare` r2
+                      x  -> x
+               _ -> LT
+        SymLit d1 r1
+          -> case s2 of
+               SymCat {} -> GT
+               SymLit d2 r2
+                 -> case compare d1 d2 of
+                      EQ -> r1 `compare` r2
+                      x  -> x
+               _ -> LT
+        SymVar d1 r1
+          -> if tagToEnum# (getTag s2 ># 2#)
+               then LT
+               else case s2 of
+                      SymVar d2 r2
+                        -> case compare d1 d2 of
+                             EQ -> r1 `compare` r2
+                             x  -> x
+                      _ -> GT
+        SymKS t1
+          -> if tagToEnum# (getTag s2 ># 3#)
+               then LT
+               else case s2 of
+                      SymKS t2 -> t1 `compareToken` t2
+                      _          -> GT
+        SymKP a1 b1
+          -> if tagToEnum# (getTag s2 ># 4#)
+               then LT
+               else case s2 of
+                      SymKP a2 b2
+                        -> case compare a1 a2 of
+                             EQ -> b1 `compare` b2
+                             x  -> x
+                      _ -> GT
+        _ -> let t1 = getTag s1
+                 t2 = getTag s2
+             in if tagToEnum# (t1 <# t2) 
+                  then LT
+                  else if tagToEnum# (t1 ==# t2)
+                         then EQ
+                         else GT
+  
+    compareToken []     []     = EQ
+    compareToken []     _      = LT
+    compareToken _      []     = GT
+    compareToken (x:xs) (y:ys)
+      | x == y    = compareToken xs ys
+      | otherwise = case compare (toLower x) (toLower y) of
+                      EQ -> case compareToken xs ys of
+                              EQ -> compare x y
+                              x  -> x
+                      x  -> x
+-}

src/compiler/GF/Compile/Optimize.hs

@@ -1,232 +0,0 @@
-{-# LANGUAGE PatternGuards #-}
-----------------------------------------------------------------------
--- |
--- Module      : Optimize
--- Maintainer  : AR
--- Stability   : (stable)
--- Portability : (portable)
---
--- > CVS $Date: 2005/09/16 13:56:13 $ 
--- > CVS $Author: aarne $
--- > CVS $Revision: 1.18 $
---
--- Top-level partial evaluation for GF source modules.
------------------------------------------------------------------------------
-
-module GF.Compile.Optimize (optimizeModule) where
-
-import GF.Grammar.Grammar
-import GF.Infra.Ident
-import GF.Grammar.Printer
-import GF.Grammar.Macros
-import GF.Grammar.Lookup
-import GF.Grammar.Predef
-import GF.Compile.Compute.ConcreteNew(GlobalEnv,normalForm,resourceValues)
-import GF.Data.Operations
-import GF.Infra.Option
-
-import Control.Monad
-import qualified Data.Set as Set
-import qualified Data.Map as Map
-import GF.Text.Pretty
-import Debug.Trace
-
--- | partial evaluation of concrete syntax. AR 6\/2001 -- 16\/5\/2003 -- 5\/2\/2005.
-
-optimizeModule :: Options -> SourceGrammar -> SourceModule -> Err SourceModule
-optimizeModule opts sgr m@(name,mi)
-  | mstatus mi == MSComplete = do
-      ids <- topoSortJments m
-      mi <- foldM updateEvalInfo mi ids
-      return (name,mi)
-  | otherwise = return m
- where
-   oopts = opts `addOptions` mflags mi
-
-   resenv = resourceValues oopts sgr
-
-   updateEvalInfo mi (i,info) = do
-     info <- evalInfo oopts resenv sgr (name,mi) i info
-     return (mi{jments=Map.insert i info (jments mi)})
-
-evalInfo :: Options -> GlobalEnv -> SourceGrammar -> SourceModule -> Ident -> Info -> Err Info
-evalInfo opts resenv sgr m c info = do
-
- (if verbAtLeast opts Verbose then trace (" " ++ showIdent c) else id) return ()
-
- errIn ("optimizing " ++ showIdent c) $ case info of
-
-  CncCat ptyp pde pre ppr mpmcfg -> do
-    pde' <- case (ptyp,pde) of
-      (Just (L _ typ), Just (L loc de)) -> do
-        de <- partEval opts gr ([(Explicit, varStr, typeStr)], typ) de
-        return (Just (L loc (factor param c 0 de)))
-      (Just (L loc typ), Nothing) -> do
-        de <- mkLinDefault gr typ
-        de <- partEval opts gr ([(Explicit, varStr, typeStr)], typ) de
-        return (Just (L loc (factor param c 0 de)))
-      _ -> return pde   -- indirection
-
-    pre' <- case (ptyp,pre) of
-      (Just (L _ typ), Just (L loc re)) -> do
-        re <- partEval opts gr ([(Explicit, varStr, typ)], typeStr) re
-        return (Just (L loc (factor param c 0 re)))
-      (Just (L loc typ), Nothing) -> do
-        re <- mkLinReference gr typ
-        re <- partEval opts gr ([(Explicit, varStr, typ)], typeStr) re
-        return (Just (L loc (factor param c 0 re)))
-      _ -> return pre   -- indirection
-
-    let ppr' = fmap (evalPrintname resenv c) ppr
-
-    return (CncCat ptyp pde' pre' ppr' mpmcfg)
-
-  CncFun (mt@(Just (_,cont,val))) pde ppr mpmcfg -> --trace (prt c) $
-       eIn ("linearization in type" <+> mkProd cont val [] $$ "of function") $ do
-    pde' <- case pde of
-      Just (L loc de) -> do de <- partEval opts gr (cont,val) de
-                            return (Just (L loc (factor param c 0 de)))
-      Nothing -> return pde
-    let ppr' = fmap (evalPrintname resenv c) ppr
-    return $ CncFun mt pde' ppr' mpmcfg -- only cat in type actually needed
-{-
-  ResOper pty pde 
-    | not new && OptExpand `Set.member` optim -> do
-         pde' <- case pde of
-                   Just (L loc de) -> do de <- computeConcrete gr de
-                                         return (Just (L loc (factor param c 0 de)))
-                   Nothing -> return Nothing
-         return $ ResOper pty pde'
--}
-  _ ->  return info
- where
--- new = flag optNewComp opts -- computations moved to GF.Compile.GeneratePMCFG
-
-   gr = prependModule sgr m
-   optim = flag optOptimizations opts
-   param = OptParametrize `Set.member` optim
-   eIn cat = errIn (render ("Error optimizing" <+> cat <+> c <+> ':'))
-
--- | the main function for compiling linearizations
-partEval :: Options -> SourceGrammar -> (Context,Type) -> Term -> Err Term
-partEval opts = {-if flag optNewComp opts
-                then-} partEvalNew opts
-                {-else partEvalOld opts-}
-
-partEvalNew opts gr (context, val) trm =
-    errIn (render ("partial evaluation" <+> ppTerm Qualified 0 trm)) $
-    checkPredefError trm
-{-
-partEvalOld opts gr (context, val) trm = errIn (render (text "partial evaluation" <+> ppTerm Qualified 0 trm)) $ do
-  let vars  = map (\(bt,x,t) -> x) context
-      args  = map Vr vars
-      subst = [(v, Vr v) | v <- vars]
-      trm1 = mkApp trm args
-  trm2 <- computeTerm gr subst trm1
-  trm3 <- if rightType trm2
-          then computeTerm gr subst trm2 -- compute twice??
-          else recordExpand val trm2 >>= computeTerm gr subst
-  trm4 <- checkPredefError trm3
-  return $ mkAbs [(Explicit,v) | v <- vars] trm4
-  where
-    -- don't eta expand records of right length (correct by type checking)
-    rightType (R rs) = case val of
-                         RecType ts -> length rs == length ts
-                         _          -> False
-    rightType _      = False
-
-
--- here we must be careful not to reduce
---   variants {{s = "Auto" ; g = N} ; {s = "Wagen" ; g = M}}
---   {s  = variants {"Auto" ; "Wagen"} ; g  = variants {N ; M}} ;
-
-recordExpand :: Type -> Term -> Err Term
-recordExpand typ trm = case typ of
-  RecType tys -> case trm of
-    FV rs -> return $ FV [R [assign lab (P r lab) | (lab,_) <- tys] | r <- rs]
-    _ -> return $ R [assign lab (P trm lab) | (lab,_) <- tys]
-  _ -> return trm
-
--}
--- | auxiliaries for compiling the resource
-
-mkLinDefault :: SourceGrammar -> Type -> Err Term
-mkLinDefault gr typ = liftM (Abs Explicit varStr) $ mkDefField typ
- where
-   mkDefField typ = case typ of
-     Table p t  -> do
-       t' <- mkDefField t
-       let T _ cs = mkWildCases t'
-       return $ T (TWild p) cs
-     Sort s | s == cStr -> return $ Vr varStr
-     QC p           -> do vs <- lookupParamValues gr p
-                          case vs of
-                            v:_ -> return v
-                            _   -> Bad (render ("no parameter values given to type" <+> ppQIdent Qualified p))
-     RecType r  -> do
-       let (ls,ts) = unzip r
-       ts <- mapM mkDefField ts
-       return $ R (zipWith assign ls ts)
-     _ | Just _ <- isTypeInts typ -> return $ EInt 0 -- exists in all as first val
-     _ -> Bad (render ("linearization type field cannot be" <+> typ))
-
-mkLinReference :: SourceGrammar -> Type -> Err Term
-mkLinReference gr typ = 
- liftM (Abs Explicit varStr) $ 
-   case mkDefField typ (Vr varStr) of
-     Bad "no string" -> return Empty
-     x               -> x
- where
-   mkDefField ty trm = 
-     case ty of
-       Table pty ty -> do ps <- allParamValues gr pty
-                          case ps of
-                            []     -> Bad "no string"
-                            (p:ps) -> mkDefField ty (S trm p)
-       Sort s | s == cStr -> return trm
-       QC p       -> Bad "no string"
-       RecType [] -> Bad "no string"
-       RecType rs -> do
-         msum (map (\(l,ty) -> mkDefField ty (P trm l)) (sortRec rs))
-           `mplus` Bad "no string"
-       _ | Just _ <- isTypeInts typ -> Bad "no string"
-       _ -> Bad (render ("linearization type field cannot be" <+> typ))
-
-evalPrintname :: GlobalEnv -> Ident -> L Term -> L Term
-evalPrintname resenv c (L loc pr) = L loc (normalForm resenv (L loc c) pr)
-
--- do even more: factor parametric branches
-
-factor :: Bool -> Ident -> Int -> Term -> Term
-factor param c i t =
-  case t of
-    T (TComp ty) cs -> factors ty [(p, factor param c (i+1) v) | (p, v) <- cs]
-    _               -> composSafeOp (factor param c i) t
-  where
-    factors ty pvs0  
-                 | not param = V ty (map snd pvs0)
-    factors ty []            = V ty []
-    factors ty pvs0@[(p,v)]  = V ty [v]
-    factors ty pvs0@(pv:pvs) =
-      let t  = mkFun pv
-          ts = map mkFun pvs
-      in if all (==t) ts
-           then T (TTyped ty) (mkCases t)
-           else V ty (map snd pvs0)
-
-    --- we hope this will be fresh and don't check... in GFC would be safe
-    qvar = identS ("q_" ++ showIdent c ++ "__" ++ show i)
-
-    mkFun (patt, val) = replace (patt2term patt) (Vr qvar) val
-    mkCases t = [(PV qvar, t)]
-
---  we need to replace subterms
-replace :: Term -> Term -> Term -> Term
-replace old new trm =
-  case trm of
-    -- these are the important cases, since they can correspond to patterns  
-    QC _     | trm == old -> new
-    App _ _  | trm == old -> new
-    R _      | trm == old -> new
-    App x y               -> App (replace old new x) (replace old new y)
-    _                     -> composSafeOp (replace old new) trm

src/compiler/GF/Compile/OptimizePGF.hs

@@ -0,0 +1,191 @@
+{-# LANGUAGE BangPatterns #-}
+module GF.Compile.OptimizePGF(optimizePGF) where
+
+import PGF2(Cat,Fun)
+import PGF2.Transactions
+import Data.Array.ST
+import Data.Array.Unboxed
+import qualified Data.Map as Map
+import qualified Data.Set as Set
+import qualified Data.IntSet as IntSet
+import qualified Data.IntMap as IntMap
+import qualified Data.List as List
+import Control.Monad.ST
+
+type ConcrData = ()
+{-([(FId,[FunId])],          -- ^ Lindefs
+                  [(FId,[FunId])],          -- ^ Linrefs
+                  [(FId,[Production])],     -- ^ Productions
+                  [(Fun,[SeqId])],          -- ^ Concrete functions   (must be sorted by Fun)
+                  [[Symbol]],               -- ^ Sequences            (must be sorted)
+                  [(Cat,FId,FId,[String])]) -- ^ Concrete categories
+-}
+optimizePGF :: Cat -> ConcrData -> ConcrData
+optimizePGF startCat = error "TODO: optimizePGF" {- topDownFilter startCat . bottomUpFilter
+
+catString = "String"
+catInt    = "Int"
+catFloat  = "Float"
+catVar    = "__gfVar"
+
+topDownFilter :: Cat -> ConcrData -> ConcrData
+topDownFilter startCat (lindefs,linrefs,prods,cncfuns,sequences,cnccats) =
+  let env0             = (Map.empty,Map.empty)
+      (env1,lindefs')  = List.mapAccumL (\env (fid,funids) -> let (env',funids') = List.mapAccumL (optimizeFun fid [PArg [] fidVar]) env funids in (env',(fid,funids')))
+                                        env0
+                                        lindefs
+      (env2,linrefs')  = List.mapAccumL (\env (fid,funids) -> let (env',funids') = List.mapAccumL (optimizeFun fidVar [PArg [] fid]) env funids in (env',(fid,funids')))
+                                        env1
+                                        linrefs
+      (env3,prods')    = List.mapAccumL (\env (fid,set)    -> let (env',set') = List.mapAccumL (optimizeProd fid) env set in (env',(fid,set')))
+                                        env2
+                                        prods
+      cnccats' = map filterCatLabels cnccats
+      (sequences',cncfuns') = env3
+  in (lindefs',linrefs',prods',mkSetArray cncfuns',mkSetArray  sequences',cnccats')
+  where
+    cncfuns_array   = listArray (0,length cncfuns-1)   cncfuns   :: Array FunId (Fun, [SeqId])
+    sequences_array = listArray (0,length sequences-1) sequences :: Array SeqId [Symbol]
+    prods_map  = IntMap.fromList prods
+    fid2catMap = IntMap.fromList ((fidVar,catVar) :  [(fid,cat) | (cat,start,end,lbls) <- cnccats,
+                                                                  fid <- [start..end]])
+
+    fid2cat fid =
+      case IntMap.lookup fid fid2catMap of
+        Just cat -> cat
+        Nothing  -> case [fid | Just set <- [IntMap.lookup fid prods_map], PCoerce fid <- set] of
+                      (fid:_) -> fid2cat fid
+                      _       -> error "unknown forest id"
+
+    starts =
+      [(startCat,lbl) | (cat,_,_,lbls) <- cnccats, cat==startCat, lbl <- [0..length lbls-1]]
+
+    allRelations =
+      Map.unionsWith Set.union
+                     [rel fid prod | (fid,set) <- prods, prod <- set]
+      where
+        rel fid (PApply funid args) = Map.fromList [((fid2cat fid,lbl),deps args seqid) | (lbl,seqid) <- zip [0..] lin]
+          where
+            (_,lin) = cncfuns_array ! funid
+        rel fid _                   = Map.empty
+
+        deps args seqid = Set.fromList [let PArg _ fid = args !! r in (fid2cat fid,d) | SymCat r d <- seq]
+          where
+            seq = sequences_array ! seqid
+
+    -- here we create a mapping from a category to an array of indices.
+    -- An element of the array is equal to -1 if the corresponding index
+    -- is not going to be used in the optimized grammar, or the new index
+    -- if it will be used
+    closure :: Map.Map Cat [Int]
+    closure = runST $ do
+      set <- initSet
+      addLitCat catString set
+      addLitCat catInt    set
+      addLitCat catFloat  set
+      addLitCat catVar    set
+      closureSet set starts
+      doneSet set
+      where
+        initSet :: ST s (Map.Map Cat (STUArray s Int Int))
+        initSet =
+          fmap Map.fromList $ sequence
+                        [fmap ((,) cat) (newArray (0,length lbls-1) (-1))
+                                             | (cat,_,_,lbls) <- cnccats]
+
+        addLitCat cat set =
+          case Map.lookup cat set of
+            Just indices -> writeArray indices 0 0
+            Nothing      -> return ()
+
+        closureSet set []                 = return ()
+        closureSet set (x@(cat,index):xs) =
+          case Map.lookup cat set of
+            Just indices -> do v <- readArray indices index
+                               writeArray indices index 0
+                               if v < 0
+                                 then case Map.lookup x allRelations of
+                                        Just ys -> closureSet set (Set.toList ys++xs)
+                                        Nothing -> closureSet set xs
+                                 else closureSet set xs
+            Nothing      -> error "unknown cat"
+
+        doneSet :: Map.Map Cat (STUArray s Int Int) -> ST s (Map.Map Cat [Int])
+        doneSet set =
+          fmap Map.fromAscList $ mapM done (Map.toAscList set)
+          where
+            done (cat,indices) = do
+              indices <- fmap (reindex 0) (getElems indices)
+              return (cat,indices)
+
+            reindex k []     = []
+            reindex k (v:vs)
+              | v < 0        = v : reindex k     vs
+              | otherwise    = k : reindex (k+1) vs
+
+    optimizeProd res env (PApply funid args) =
+      let (env',funid') = optimizeFun res args env funid
+      in (env', PApply funid' args)
+    optimizeProd res env prod = (env,prod)
+
+    optimizeFun res args (seqs,funs) funid =
+      let (seqs',lin')   = List.mapAccumL addUnique seqs [map updateSymbol (sequences_array ! seqid) | 
+                                                                   (idx,seqid) <- zip (indicesOf res) lin, idx >= 0]
+          (funs',funid') = addUnique funs (fun, lin')
+      in ((seqs',funs'), funid')
+      where
+        (fun,lin) = cncfuns_array ! funid
+
+        indicesOf fid
+          | fid < 0   = [0]
+          | otherwise =
+              case Map.lookup (fid2cat fid) closure of
+                Just indices -> indices
+                Nothing      -> error "unknown category"
+
+        addUnique seqs seq =
+          case Map.lookup seq seqs of
+            Just seqid -> (seqs,seqid)
+            Nothing    -> let seqid = Map.size seqs
+                          in (Map.insert seq seqid seqs, seqid)
+
+        updateSymbol (SymCat r d) = let PArg _ fid = args !! r in SymCat r (indicesOf fid !! d)
+        updateSymbol s            = s
+
+    filterCatLabels (cat,start,end,lbls) =
+      case Map.lookup cat closure of
+        Just indices -> let lbls' = [lbl | (idx,lbl) <- zip indices lbls, idx >= 0]
+                        in (cat,start,end,lbls')
+        Nothing      -> error ("unknown category")
+
+    mkSetArray map = sortSnd (Map.toList map)
+      where
+        sortSnd = List.map fst . List.sortBy (\(_,i) (_,j) -> compare i j)
+
+
+bottomUpFilter :: ConcrData -> ConcrData
+bottomUpFilter (lindefs,linrefs,prods,cncfuns,sequences,cnccats) =
+  (lindefs,linrefs,filterProductions IntMap.empty IntSet.empty prods,cncfuns,sequences,cnccats)
+
+filterProductions prods0 hoc0 prods
+  | prods0 == prods1 = IntMap.toList prods0
+  | otherwise        = filterProductions prods1 hoc1 prods
+  where
+    (prods1,hoc1) = foldl foldProdSet (IntMap.empty,IntSet.empty) prods
+
+    foldProdSet (!prods,!hoc) (fid,set)
+      | null set1 = (prods,hoc)
+      | otherwise = (IntMap.insert fid set1 prods,hoc1)
+      where
+        set1 = filter filterRule set
+        hoc1 = foldl accumHOC hoc set1
+
+    filterRule (PApply funid args) = all (\(PArg _ fid) -> isLive fid) args
+    filterRule (PCoerce fid)       = isLive fid
+    filterRule _                   = True
+
+    isLive fid = isPredefFId fid || IntMap.member fid prods0 || IntSet.member fid hoc0
+
+    accumHOC hoc (PApply funid args) = List.foldl' (\hoc (PArg hypos _) -> List.foldl' (\hoc fid -> IntSet.insert fid hoc) hoc (map snd hypos)) hoc args
+    accumHOC hoc _                   = hoc
+-}

src/compiler/GF/Compile/PGFtoHaskell.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/06/17 12:39:07 $ 
+-- > CVS $Date: 2005/06/17 12:39:07 $
 -- > CVS $Author: bringert $
 -- > CVS $Revision: 1.8 $
 --
@@ -16,13 +16,14 @@
 
 module GF.Compile.PGFtoHaskell (grammar2haskell) where
 
-import PGF(showCId)
-import PGF.Internal
+import PGF2
+import PGF2.Internal
 
 import GF.Data.Operations
 import GF.Infra.Option
 
-import Data.List --(isPrefixOf, find, intersperse)
+import Data.List(isPrefixOf,find,intercalate,intersperse,groupBy,sortBy)
+import Data.Maybe(mapMaybe)
 import qualified Data.Map as Map
 
 type Prefix = String -> String
@@ -34,11 +35,12 @@ grammar2haskell :: Options
                 -> PGF
                 -> String
 grammar2haskell opts name gr = foldr (++++) [] $
-  pragmas ++ haskPreamble gadt name derivingClause extraImports ++
+  pragmas ++ haskPreamble gadt name derivingClause (extraImports ++ pgfImports) ++
   [types, gfinstances gId lexical gr'] ++ compos
     where gr' = hSkeleton gr
           gadt = haskellOption opts HaskellGADT
           dataExt = haskellOption opts HaskellData
+          pgf2 = haskellOption opts HaskellPGF2
           lexical cat = haskellOption opts HaskellLexical && isLexicalCat opts cat
           gId | haskellOption opts HaskellNoPrefix = rmForbiddenChars
               | otherwise = ("G"++) . rmForbiddenChars
@@ -50,21 +52,23 @@ grammar2haskell opts name gr = foldr (++++) [] $
           derivingClause
                  | dataExt = "deriving (Show,Data)"
                  | otherwise = "deriving Show"
-          extraImports | gadt = ["import Control.Monad.Identity",
-                                 "import Data.Monoid"]
+          extraImports | gadt = ["import Control.Monad.Identity", "import Data.Monoid"]
                        | dataExt = ["import Data.Data"]
                        | otherwise = []
+          pgfImports | pgf2 = ["import PGF2 hiding (Tree)", "", "showCId :: CId -> String", "showCId = id"]
+                     | otherwise = ["import PGF hiding (Tree)"]
           types | gadt = datatypesGADT gId lexical gr'
                 | otherwise = datatypes gId derivingClause lexical gr'
           compos | gadt = prCompos gId lexical gr' ++ composClass
                  | otherwise = []
 
-haskPreamble gadt name derivingClause extraImports =
+haskPreamble :: Bool -> String -> String -> [String] -> [String]
+haskPreamble gadt name derivingClause imports =
  [
   "module " ++ name ++ " where",
   ""
- ] ++ extraImports ++ [
-  "import PGF hiding (Tree)",
+ ] ++ imports ++ [
+  "",
   "----------------------------------------------------",
   "-- automatic translation from GF to Haskell",
   "----------------------------------------------------",
@@ -85,10 +89,11 @@ haskPreamble gadt name derivingClause extraImports =
   ""
  ]
 
+predefInst :: Bool -> String -> String -> String -> String -> String -> String
 predefInst gadt derivingClause gtyp typ destr consr =
   (if gadt
     then []
-    else ("newtype" +++ gtyp +++ "=" +++ gtyp +++ typ +++ derivingClause ++ "\n\n")
+    else "newtype" +++ gtyp +++ "=" +++ gtyp +++ typ +++ derivingClause ++ "\n\n"
     )
   ++
   "instance Gf" +++ gtyp +++ "where" ++++
@@ -103,10 +108,10 @@ type OIdent = String
 type HSkeleton = [(OIdent, [(OIdent, [OIdent])])]
 
 datatypes :: Prefix -> DerivingClause -> (OIdent -> Bool) -> (String,HSkeleton) -> String
-datatypes gId derivingClause lexical = (foldr (+++++) "") . (filter (/="")) . (map (hDatatype gId derivingClause lexical)) . snd
+datatypes gId derivingClause lexical = foldr (+++++) "" . filter (/="") . map (hDatatype gId derivingClause lexical) . snd
 
 gfinstances :: Prefix -> (OIdent -> Bool) -> (String,HSkeleton) -> String
-gfinstances gId lexical (m,g) = (foldr (+++++) "") $ (filter (/="")) $ (map (gfInstance gId lexical m)) g
+gfinstances gId lexical (m,g) = foldr (+++++) "" $ filter (/="") $ map (gfInstance gId lexical m) g
 
 
 hDatatype  :: Prefix -> DerivingClause -> (OIdent -> Bool) -> (OIdent, [(OIdent, [OIdent])]) -> String
@@ -131,16 +136,17 @@ nonLexicalRules True rules = [r | r@(f,t) <- rules, not (null t)]
 lexicalConstructor :: OIdent -> String
 lexicalConstructor cat = "Lex" ++ cat
 
+predefTypeSkel :: HSkeleton
 predefTypeSkel = [(c,[]) | c <- ["String", "Int", "Float"]]
 
 -- GADT version of data types
 datatypesGADT :: Prefix -> (OIdent -> Bool) -> (String,HSkeleton) -> String
-datatypesGADT gId lexical (_,skel) = unlines $ 
+datatypesGADT gId lexical (_,skel) = unlines $
     concatMap (hCatTypeGADT gId) (skel ++ predefTypeSkel) ++
-    [ 
-      "", 
+    [
+      "",
       "data Tree :: * -> * where"
-    ] ++ 
+    ] ++
     concatMap (map ("  "++) . hDatatypeGADT gId lexical) skel ++
     [
       "  GString :: String -> Tree GString_",
@@ -164,23 +170,23 @@ hCatTypeGADT gId (cat,rules)
        "data"+++gId cat++"_"]
 
 hDatatypeGADT :: Prefix -> (OIdent -> Bool) -> (OIdent, [(OIdent, [OIdent])]) -> [String]
-hDatatypeGADT gId lexical (cat, rules) 
+hDatatypeGADT gId lexical (cat, rules)
     | isListCat (cat,rules) = [gId cat+++"::"+++"["++gId (elemCat cat)++"]" +++ "->" +++ t]
     | otherwise =
-        [ gId f +++ "::" +++ concatMap (\a -> gId a +++ "-> ") args ++ t 
+        [ gId f +++ "::" +++ concatMap (\a -> gId a +++ "-> ") args ++ t
           | (f,args) <- nonLexicalRules (lexical cat) rules ]
         ++ if lexical cat then [lexicalConstructor cat +++ ":: String ->"+++ t] else []
   where t = "Tree" +++ gId cat ++ "_"
 
 hEqGADT :: Prefix -> (OIdent -> Bool) -> (OIdent, [(OIdent, [OIdent])]) -> [String]
 hEqGADT gId lexical (cat, rules)
-  | isListCat (cat,rules) = let r = listr cat in ["(" ++ patt "x" r ++ "," ++ patt "y" r ++ ") -> " ++ listeqs] 
+  | isListCat (cat,rules) = let r = listr cat in ["(" ++ patt "x" r ++ "," ++ patt "y" r ++ ") -> " ++ listeqs]
   | otherwise = ["(" ++ patt "x" r ++ "," ++ patt "y" r ++ ") -> " ++ eqs r | r <- nonLexicalRules (lexical cat) rules]
           ++ if lexical cat then ["(" ++ lexicalConstructor cat +++ "x" ++ "," ++ lexicalConstructor cat +++ "y" ++ ") -> x == y"] else []
 
  where
    patt s (f,xs) = unwords (gId f : mkSVars s (length xs))
-   eqs (_,xs) = unwords ("and" : "[" : intersperse "," [x ++ " == " ++ y | 
+   eqs (_,xs) = unwords ("and" : "[" : intersperse "," [x ++ " == " ++ y |
      (x,y) <- zip (mkSVars "x" (length xs)) (mkSVars "y" (length xs)) ] ++ ["]"])
    listr c = (c,["foo"]) -- foo just for length = 1
    listeqs = "and [x == y | (x,y) <- zip x1 y1]"
@@ -189,25 +195,26 @@ prCompos :: Prefix -> (OIdent -> Bool) -> (String,HSkeleton) -> [String]
 prCompos gId lexical (_,catrules) =
     ["instance Compos Tree where",
      "  compos r a f t = case t of"]
-    ++ 
+    ++
     ["    " ++ prComposCons (gId f) xs | (c,rs) <- catrules, not (isListCat (c,rs)),
-                                         (f,xs) <- rs, not (null xs)] 
-    ++ 
+                                         (f,xs) <- rs, not (null xs)]
+    ++
     ["    " ++ prComposCons (gId c) ["x1"] | (c,rs) <- catrules, isListCat (c,rs)]
-    ++ 
+    ++
     ["    _ -> r t"]
   where
-    prComposCons f xs = let vs = mkVars (length xs) in 
+    prComposCons f xs = let vs = mkVars (length xs) in
                         f +++ unwords vs +++ "->" +++ rhs f (zip vs xs)
     rhs f vcs = "r" +++ f +++ unwords (map (prRec f) vcs)
-    prRec f (v,c) 
+    prRec f (v,c)
       | isList f  = "`a` foldr (a . a (r (:)) . f) (r [])" +++ v
       | otherwise = "`a`" +++ "f" +++ v
-    isList f = (gId "List") `isPrefixOf` f
+    isList f = gId "List" `isPrefixOf` f
 
 gfInstance :: Prefix -> (OIdent -> Bool) -> String -> (OIdent, [(OIdent, [OIdent])]) -> String
 gfInstance gId lexical m crs = hInstance gId lexical m crs ++++ fInstance gId lexical m crs
 
+hInstance :: (String -> String) -> (String -> Bool) -> String -> (String, [(OIdent, [OIdent])]) -> String
 ----hInstance m ("Cn",_) = "" --- seems to belong to an old applic. AR 18/5/2004
 hInstance gId _ m (cat,[]) = unlines [
   "instance Show" +++ gId cat,
@@ -216,15 +223,15 @@ hInstance gId _ m (cat,[]) = unlines [
   "  gf _ = undefined",
   "  fg _ = undefined"
   ]
-hInstance gId lexical m (cat,rules) 
+hInstance gId lexical m (cat,rules)
  | isListCat (cat,rules) =
   "instance Gf" +++ gId cat +++ "where" ++++
-     "  gf (" ++ gId cat +++ "[" ++ concat (intersperse "," baseVars) ++ "])" 
+     "  gf (" ++ gId cat +++ "[" ++ intercalate "," baseVars ++ "])"
            +++ "=" +++ mkRHS ("Base"++ec) baseVars ++++
-     "  gf (" ++ gId cat +++ "(x:xs)) = " 
-           ++ mkRHS ("Cons"++ec) ["x",prParenth (gId cat+++"xs")] 
+     "  gf (" ++ gId cat +++ "(x:xs)) = "
+           ++ mkRHS ("Cons"++ec) ["x",prParenth (gId cat+++"xs")]
 -- no show for GADTs
---     ++++ " gf (" ++ gId cat +++ "xs) = error (\"Bad " ++ cat ++ " value: \" ++ show xs)" 
+--     ++++ " gf (" ++ gId cat +++ "xs) = error (\"Bad " ++ cat ++ " value: \" ++ show xs)"
  | otherwise =
   "instance Gf" +++ gId cat +++ "where\n" ++
   unlines ([mkInst f xx | (f,xx) <- nonLexicalRules (lexical cat) rules]
@@ -233,23 +240,26 @@ hInstance gId lexical m (cat,rules)
    ec = elemCat cat
    baseVars = mkVars (baseSize (cat,rules))
    mkInst f xx = let xx' = mkVars (length xx) in "  gf " ++
-     (if length xx == 0 then gId f else prParenth (gId f +++ foldr1 (+++) xx')) +++
+     (if null xx then gId f else prParenth (gId f +++ foldr1 (+++) xx')) +++
      "=" +++ mkRHS f xx'
-   mkRHS f vars = "mkApp (mkCId \"" ++ f ++ "\")" +++ 
-		   "[" ++ prTList ", " ["gf" +++ x | x <- vars] ++ "]"
+   mkRHS f vars = "mkApp (mkCId \"" ++ f ++ "\")" +++
+       "[" ++ prTList ", " ["gf" +++ x | x <- vars] ++ "]"
 
+mkVars :: Int -> [String]
 mkVars = mkSVars "x"
+
+mkSVars :: String -> Int -> [String]
 mkSVars s n = [s ++ show i | i <- [1..n]]
 
 ----fInstance m ("Cn",_) = "" ---
 fInstance _ _ m (cat,[]) = ""
 fInstance gId lexical m (cat,rules) =
   "  fg t =" ++++
-  (if isList 
+  (if isList
     then "    " ++ gId cat ++ " (fgs t) where\n     fgs t = case unApp t of"
     else "    case unApp t of") ++++
   unlines [mkInst f xx | (f,xx) <- nonLexicalRules (lexical cat) rules] ++++
-  (if lexical cat then "      Just (i,[]) -> " ++ lexicalConstructor cat +++ "(showCId i)" else "") ++++
+  (if lexical cat then "      Just (i,[]) -> " ++ lexicalConstructor cat +++ "i" else "") ++++
   "      _ -> error (\"no" +++ cat ++ " \" ++ show t)"
    where
     isList = isListCat (cat,rules)
@@ -257,31 +267,35 @@ fInstance gId lexical m (cat,rules) =
      "      Just (i," ++
      "[" ++ prTList "," xx' ++ "])" +++
      "| i == mkCId \"" ++ f ++ "\" ->" +++ mkRHS f xx'
-       where xx' = ["x" ++ show i | (_,i) <- zip xx [1..]]
-	     mkRHS f vars 
-		 | isList =
-		     if "Base" `isPrefixOf` f 
-                       then "[" ++ prTList ", " [ "fg" +++ x | x <- vars ] ++ "]"
-		       else "fg" +++ (vars !! 0) +++ ":" +++ "fgs" +++ (vars !! 1)
-		 | otherwise = 
-		     gId f +++  
-		     prTList " " [prParenth ("fg" +++ x) | x <- vars]
+       where
+         xx' = ["x" ++ show i | (_,i) <- zip xx [1..]]
+         mkRHS f vars
+           | isList =
+               if "Base" `isPrefixOf` f
+                             then "[" ++ prTList ", " [ "fg" +++ x | x <- vars ] ++ "]"
+                 else "fg" +++ (vars !! 0) +++ ":" +++ "fgs" +++ (vars !! 1)
+           | otherwise =
+               gId f +++
+               prTList " " [prParenth ("fg" +++ x) | x <- vars]
 
 --type HSkeleton = [(OIdent, [(OIdent, [OIdent])])]
 hSkeleton :: PGF -> (String,HSkeleton)
 hSkeleton gr = 
-  (showCId (absname gr), 
+  (abstractName gr,
    let fs = 
-         [(showCId c, [(showCId f, map showCId cs) | (f, (cs,_)) <- fs]) | 
-                                        fs@((_, (_,c)):_) <- fns]
-   in fs ++ [(sc, []) | c <- cts, let sc = showCId c, notElem sc (["Int", "Float", "String"] ++ map fst fs)]
+         [(c, [(f, cs) | (f, cs,_) <- fs]) | 
+                                        fs@((_, _,c):_) <- fns]
+   in fs ++ [(c, []) | c <- cts, notElem c (["Int", "Float", "String"] ++ map fst fs)]
   )
  where
-   cts = Map.keys (cats (abstract gr)) 
-   fns = groupBy valtypg (sortBy valtyps (map jty (Map.assocs (funs (abstract gr)))))
-   valtyps (_, (_,x)) (_, (_,y)) = compare x y
-   valtypg (_, (_,x)) (_, (_,y)) = x == y
-   jty (f,(ty,_,_,_)) = (f,catSkeleton ty)
+   cts = categories gr
+   fns = groupBy valtypg (sortBy valtyps (mapMaybe jty (functions gr)))
+   valtyps (_,_,x) (_,_,y) = compare x y
+   valtypg (_,_,x) (_,_,y) = x == y
+   jty f = case functionType gr f of
+             Just ty -> let (hypos,valcat,_) = unType ty
+                        in Just (f,[argcat | (_,_,ty) <- hypos, let (_,argcat,_) = unType ty],valcat)
+             Nothing -> Nothing
 {-
 updateSkeleton :: OIdent -> HSkeleton -> (OIdent, [OIdent]) -> HSkeleton
 updateSkeleton cat skel rule =
@@ -291,9 +305,10 @@ updateSkeleton cat skel rule =
 -}
 isListCat :: (OIdent, [(OIdent, [OIdent])]) -> Bool
 isListCat (cat,rules) = "List" `isPrefixOf` cat && length rules == 2
-		    && ("Base"++c) `elem` fs && ("Cons"++c) `elem` fs
-    where c = elemCat cat
-	  fs = map fst rules
+        && ("Base"++c) `elem` fs && ("Cons"++c) `elem` fs
+    where
+      c = elemCat cat
+      fs = map fst rules
 
 -- | Gets the element category of a list category.
 elemCat :: OIdent -> OIdent
@@ -310,7 +325,7 @@ baseSize (_,rules) = length bs
     where Just (_,bs) = find (("Base" `isPrefixOf`) . fst) rules
 
 composClass :: [String]
-composClass = 
+composClass =
     [
      "",
      "class Compos t where",
@@ -337,4 +352,3 @@ composClass =
      "",
      "newtype C b a = C { unC :: b }"
     ]
-

src/compiler/GF/Compile/PGFtoJS.hs

@@ -1,105 +0,0 @@
-module GF.Compile.PGFtoJS (pgf2js) where
-
-import PGF(showCId)
-import PGF.Internal as M
-import qualified GF.JavaScript.AbsJS as JS
-import qualified GF.JavaScript.PrintJS as JS
-
---import GF.Data.ErrM
---import GF.Infra.Option
-
---import Control.Monad (mplus)
---import Data.Array.Unboxed (UArray)
-import qualified Data.Array.IArray as Array
---import Data.Maybe (fromMaybe)
-import Data.Map (Map)
-import qualified Data.Set as Set
-import qualified Data.Map as Map
-import qualified Data.IntMap as IntMap
-
-pgf2js :: PGF -> String
-pgf2js pgf =
-  JS.printTree $ JS.Program [JS.ElStmt $ JS.SDeclOrExpr $ JS.Decl [JS.DInit (JS.Ident n) grammar]]
- where
-   n  = showCId $ absname pgf
-   as = abstract pgf
-   cs = Map.assocs (concretes pgf)
-   start = showCId $ M.lookStartCat pgf
-   grammar = new "GFGrammar" [js_abstract, js_concrete]
-   js_abstract = abstract2js start as
-   js_concrete = JS.EObj $ map concrete2js cs
-
-abstract2js :: String -> Abstr -> JS.Expr
-abstract2js start ds = new "GFAbstract" [JS.EStr start, JS.EObj $ map absdef2js (Map.assocs (funs ds))]
-
-absdef2js :: (CId,(Type,Int,Maybe ([Equation],[[M.Instr]]),Double)) -> JS.Property
-absdef2js (f,(typ,_,_,_)) =
-  let (args,cat) = M.catSkeleton typ in 
-    JS.Prop (JS.IdentPropName (JS.Ident (showCId f))) (new "Type" [JS.EArray [JS.EStr (showCId x) | x <- args], JS.EStr (showCId cat)])
-
-lit2js (LStr s) = JS.EStr s
-lit2js (LInt n) = JS.EInt n
-lit2js (LFlt d) = JS.EDbl d
-
-concrete2js :: (CId,Concr) -> JS.Property
-concrete2js (c,cnc) =
-  JS.Prop l (new "GFConcrete" [mapToJSObj (lit2js) $ cflags cnc,
-                               JS.EObj $ [JS.Prop (JS.IntPropName cat) (JS.EArray (map frule2js (Set.toList set))) | (cat,set) <- IntMap.toList (productions cnc)],
-                               JS.EArray $ (map ffun2js (Array.elems (cncfuns cnc))),
-                               JS.EArray $ (map seq2js (Array.elems (sequences cnc))),
-                               JS.EObj $ map cats (Map.assocs (cnccats cnc)),
-                               JS.EInt (totalCats cnc)])
-  where 
-   l  = JS.IdentPropName (JS.Ident (showCId c))
-{-
-   litslins = [JS.Prop (JS.StringPropName    "Int") (JS.EFun [children] [JS.SReturn $ new "Arr" [JS.EIndex (JS.EVar children) (JS.EInt 0)]]), 
-               JS.Prop (JS.StringPropName  "Float") (JS.EFun [children] [JS.SReturn $ new "Arr" [JS.EIndex (JS.EVar children) (JS.EInt 0)]]),
-               JS.Prop (JS.StringPropName "String") (JS.EFun [children] [JS.SReturn $ new "Arr" [JS.EIndex (JS.EVar children) (JS.EInt 0)]])]
--}
-   cats (c,CncCat start end _) = JS.Prop (JS.IdentPropName (JS.Ident (showCId c))) (JS.EObj [JS.Prop (JS.IdentPropName (JS.Ident "s")) (JS.EInt start)
-                                                                                            ,JS.Prop (JS.IdentPropName (JS.Ident "e")) (JS.EInt end)])
-{-
-mkStr :: String -> JS.Expr
-mkStr s = new "Str" [JS.EStr s]
-
-mkSeq :: [JS.Expr] -> JS.Expr
-mkSeq [x] = x
-mkSeq xs = new "Seq" xs
-
-argIdent :: Integer -> JS.Ident
-argIdent n = JS.Ident ("x" ++ show n)
--}
-children :: JS.Ident
-children = JS.Ident "cs"
-
-frule2js :: Production -> JS.Expr
-frule2js (PApply funid args) = new "Apply"  [JS.EInt funid, JS.EArray (map farg2js args)]
-frule2js (PCoerce arg)       = new "Coerce" [JS.EInt arg]
-
-farg2js (PArg hypos fid) = new "PArg" (map (JS.EInt . snd) hypos ++ [JS.EInt fid])
-
-ffun2js (CncFun f lins) = new "CncFun" [JS.EStr (showCId f), JS.EArray (map JS.EInt (Array.elems lins))]
-
-seq2js :: Array.Array DotPos Symbol -> JS.Expr
-seq2js seq = JS.EArray [sym2js s | s <- Array.elems seq]
-
-sym2js :: Symbol -> JS.Expr
-sym2js (SymCat n l)    = new "SymCat" [JS.EInt n, JS.EInt l]
-sym2js (SymLit n l)    = new "SymLit" [JS.EInt n, JS.EInt l]
-sym2js (SymVar n l)    = new "SymVar" [JS.EInt n, JS.EInt l]
-sym2js (SymKS t)       = new "SymKS"  [JS.EStr t]
-sym2js (SymKP ts alts) = new "SymKP"  [JS.EArray (map sym2js ts), JS.EArray (map alt2js alts)]
-sym2js SymBIND         = new "SymKS"  [JS.EStr "&+"]
-sym2js SymSOFT_BIND    = new "SymKS"  [JS.EStr "&+"]
-sym2js SymSOFT_SPACE   = new "SymKS"  [JS.EStr "&+"]
-sym2js SymCAPIT        = new "SymKS"  [JS.EStr "&|"]
-sym2js SymALL_CAPIT    = new "SymKS"  [JS.EStr "&|"]
-sym2js SymNE           = new "SymNE"  []
-
-alt2js (ps,ts) = new "Alt" [JS.EArray (map sym2js ps), JS.EArray (map JS.EStr ts)]
-
-new :: String -> [JS.Expr] -> JS.Expr
-new f xs = JS.ENew (JS.Ident f) xs
-
-mapToJSObj :: (a -> JS.Expr) -> Map CId a -> JS.Expr
-mapToJSObj f m = JS.EObj [ JS.Prop (JS.IdentPropName (JS.Ident (showCId k))) (f v) | (k,v) <- Map.toList m ]

src/compiler/GF/Compile/PGFtoJSON.hs

@@ -1,156 +1,111 @@
 module GF.Compile.PGFtoJSON (pgf2json) where
 
-import PGF (showCId)
-import qualified PGF.Internal as M
-import PGF.Internal (
-  Abstr,
-  CId,
-  CncCat(..),
-  CncFun(..),
-  Concr,
-  DotPos,
-  Equation(..),
-  Literal(..),
-  PArg(..),
-  PGF,
-  Production(..),
-  Symbol(..),
-  Type,
-  absname,
-  abstract,
-  cflags,
-  cnccats,
-  cncfuns,
-  concretes,
-  funs,
-  productions,
-  sequences,
-  totalCats
-  )
-
-import qualified Text.JSON as JSON
-import Text.JSON (JSValue(..))
-
-import qualified Data.Array.IArray as Array
-import Data.Map (Map)
-import qualified Data.Set as Set
+import PGF2
+import PGF2.Internal
+import Text.JSON
 import qualified Data.Map as Map
-import qualified Data.IntMap as IntMap
 
 pgf2json :: PGF -> String
-pgf2json pgf =
-  JSON.encode $ JSON.makeObj
-    [ ("abstract", json_abstract)
-    , ("concretes", json_concretes)
-    ]
- where
-   n  = showCId $ absname pgf
-   as = abstract pgf
-   cs = Map.assocs (concretes pgf)
-   start = showCId $ M.lookStartCat pgf
-   json_abstract = abstract2json n start as
-   json_concretes = JSON.makeObj $ map concrete2json cs
-
-abstract2json :: String -> String -> Abstr -> JSValue
-abstract2json name start ds =
-  JSON.makeObj
-    [ ("name", mkJSStr name)
-    , ("startcat", mkJSStr start)
-    , ("funs", JSON.makeObj $ map absdef2json (Map.assocs (funs ds)))
+pgf2json pgf = error "TODO: pgf2json"
+{-  encode $ makeObj
+    [ ("abstract", abstract2json pgf)
+    , ("concretes", makeObj $ map concrete2json
+                                (Map.toList (languages pgf)))
     ]
 
-absdef2json :: (CId,(Type,Int,Maybe ([Equation],[[M.Instr]]),Double)) -> (String,JSValue)
-absdef2json (f,(typ,_,_,_)) = (showCId f,sig)
+abstract2json :: PGF -> JSValue
+abstract2json pgf =
+  makeObj
+    [ ("name", showJSON (abstractName pgf))
+    , ("startcat", showJSON (showType [] (startCat pgf)))
+    , ("funs", makeObj $ map (absdef2json pgf) (functions pgf))
+    ]
+
+absdef2json :: PGF -> Fun -> (String,JSValue)
+absdef2json pgf f = (f,sig)
   where
-    (args,cat) = M.catSkeleton typ
-    sig = JSON.makeObj
-      [ ("args", JSArray $ map (mkJSStr.showCId) args)
-      , ("cat", mkJSStr $ showCId cat)
+    Just (hypos,cat,_) = fmap unType (functionType pgf f)
+    sig = makeObj
+      [ ("args", showJSON $ map (\(_,_,ty) -> showType [] ty) hypos)
+      , ("cat", showJSON cat)
       ]
 
 lit2json :: Literal -> JSValue
-lit2json (LStr s) = mkJSStr s
-lit2json (LInt n) = mkJSInt n
-lit2json (LFlt d) = JSRational True (toRational d)
+lit2json (LStr s) = showJSON s
+lit2json (LInt n) = showJSON n
+lit2json (LFlt d) = showJSON d
 
-concrete2json :: (CId,Concr) -> (String,JSValue)
-concrete2json (c,cnc) = (showCId c,obj)
+concrete2json :: (ConcName,Concr) -> (String,JSValue)
+concrete2json (c,cnc) = (c,obj)
   where
-    obj = JSON.makeObj
-      [ ("flags", JSON.makeObj [ (showCId k, lit2json v) | (k,v) <- Map.toList (cflags cnc) ])
-      , ("productions", JSON.makeObj [ (show cat, JSArray (map frule2json (Set.toList set))) | (cat,set) <- IntMap.toList (productions cnc)])
-      , ("functions", JSArray (map ffun2json (Array.elems (cncfuns cnc))))
-      , ("sequences", JSArray (map seq2json (Array.elems (sequences cnc))))
-      , ("categories", JSON.makeObj $ map cats2json (Map.assocs (cnccats cnc)))
-      , ("totalfids", mkJSInt (totalCats cnc))
+    obj = makeObj
+      [ ("flags", makeObj [(k, lit2json v) | (k,v) <- concrFlags cnc])
+      , ("productions", makeObj [(show fid, showJSON (map frule2json (concrProductions cnc fid))) | (_,start,end,_) <- concrCategories cnc, fid <- [start..end]])
+      , ("functions", showJSON [ffun2json funid (concrFunction cnc funid) | funid <- [0..concrTotalFuns cnc-1]])
+      , ("sequences", showJSON [seq2json seqid (concrSequence cnc seqid) | seqid <- [0..concrTotalSeqs cnc-1]])
+      , ("categories", makeObj $ map cat2json (concrCategories cnc))
+      , ("totalfids", showJSON (concrTotalCats cnc))
       ]
 
-cats2json :: (CId, CncCat) -> (String,JSValue)
-cats2json (c,CncCat start end _) = (showCId c, ixs)
+cat2json :: (Cat,FId,FId,[String]) -> (String,JSValue)
+cat2json (cat,start,end,_) = (cat, ixs)
   where
-    ixs = JSON.makeObj
-      [ ("start", mkJSInt start)
-      , ("end", mkJSInt end)
+    ixs = makeObj
+      [ ("start", showJSON start)
+      , ("end",   showJSON end)
       ]
 
 frule2json :: Production -> JSValue
 frule2json (PApply fid args) =
-  JSON.makeObj
-    [ ("type", mkJSStr "Apply")
-    , ("fid", mkJSInt fid)
-    , ("args", JSArray (map farg2json args))
+  makeObj
+    [ ("type", showJSON "Apply")
+    , ("fid",  showJSON fid)
+    , ("args", showJSON (map farg2json args))
     ]
 frule2json (PCoerce arg) =
-  JSON.makeObj
-    [ ("type", mkJSStr "Coerce")
-    , ("arg", mkJSInt arg)
+  makeObj
+    [ ("type", showJSON "Coerce")
+    , ("arg",  showJSON arg)
     ]
 
 farg2json :: PArg -> JSValue
 farg2json (PArg hypos fid) =
-  JSON.makeObj
-    [ ("type", mkJSStr "PArg")
-    , ("hypos", JSArray $ map (mkJSInt . snd) hypos)
-    , ("fid", mkJSInt fid)
+  makeObj
+    [ ("type", showJSON "PArg")
+    , ("hypos", JSArray $ map (showJSON . snd) hypos)
+    , ("fid", showJSON fid)
     ]
 
-ffun2json :: CncFun -> JSValue
-ffun2json (CncFun f lins) =
-  JSON.makeObj
-    [ ("name", mkJSStr $ showCId f)
-    , ("lins", JSArray (map mkJSInt (Array.elems lins)))
+ffun2json :: FunId -> (Fun,[SeqId]) -> JSValue
+ffun2json funid (fun,seqids) =
+  makeObj
+    [ ("name", showJSON fun)
+    , ("lins", showJSON seqids)
     ]
 
-seq2json :: Array.Array DotPos Symbol -> JSValue
-seq2json seq = JSArray [sym2json s | s <- Array.elems seq]
+seq2json :: SeqId -> [Symbol] -> JSValue
+seq2json seqid seq = showJSON [sym2json sym | sym <- seq]
 
 sym2json :: Symbol -> JSValue
-sym2json (SymCat n l)    = new "SymCat" [mkJSInt n, mkJSInt l]
-sym2json (SymLit n l)    = new "SymLit" [mkJSInt n, mkJSInt l]
-sym2json (SymVar n l)    = new "SymVar" [mkJSInt n, mkJSInt l]
-sym2json (SymKS t)       = new "SymKS"  [mkJSStr t]
+sym2json (SymCat n l)    = new "SymCat" [showJSON n, showJSON l]
+sym2json (SymLit n l)    = new "SymLit" [showJSON n, showJSON l]
+sym2json (SymVar n l)    = new "SymVar" [showJSON n, showJSON l]
+sym2json (SymKS t)       = new "SymKS"  [showJSON t]
 sym2json (SymKP ts alts) = new "SymKP"  [JSArray (map sym2json ts), JSArray (map alt2json alts)]
-sym2json SymBIND         = new "SymKS"  [mkJSStr "&+"]
-sym2json SymSOFT_BIND    = new "SymKS"  [mkJSStr "&+"]
-sym2json SymSOFT_SPACE   = new "SymKS"  [mkJSStr "&+"]
-sym2json SymCAPIT        = new "SymKS"  [mkJSStr "&|"]
-sym2json SymALL_CAPIT    = new "SymKS"  [mkJSStr "&|"]
+sym2json SymBIND         = new "SymKS"  [showJSON "&+"]
+sym2json SymSOFT_BIND    = new "SymKS"  [showJSON "&+"]
+sym2json SymSOFT_SPACE   = new "SymKS"  [showJSON "&+"]
+sym2json SymCAPIT        = new "SymKS"  [showJSON "&|"]
+sym2json SymALL_CAPIT    = new "SymKS"  [showJSON "&|"]
 sym2json SymNE           = new "SymNE"  []
 
 alt2json :: ([Symbol],[String]) -> JSValue
-alt2json (ps,ts) = new "Alt" [JSArray (map sym2json ps), JSArray (map mkJSStr ts)]
+alt2json (ps,ts) = new "Alt" [showJSON (map sym2json ps), showJSON ts]
 
 new :: String -> [JSValue] -> JSValue
 new f xs =
-  JSON.makeObj
-    [ ("type", mkJSStr f)
-    , ("args", JSArray xs)
+  makeObj
+    [ ("type", showJSON f)
+    , ("args", showJSON xs)
     ]
-
--- | Make JSON value from string
-mkJSStr :: String -> JSValue
-mkJSStr = JSString . JSON.toJSString
-
--- | Make JSON value from integer
-mkJSInt :: Integral a => a -> JSValue
-mkJSInt = JSRational False . toRational
+-}

src/compiler/GF/Compile/PGFtoJava.hs

@@ -1,6 +1,6 @@
 module GF.Compile.PGFtoJava (grammar2java) where
 
-import PGF
+import PGF2
 import Data.Maybe(maybe)
 import Data.List(intercalate)
 import GF.Infra.Option
@@ -24,9 +24,8 @@ javaPreamble name =
  ]
  
 javaMethod gr fun =
-  "  public static Expr "++name++"("++arg_decls++") { return new Expr("++show name++args++"); }"
+  "  public static Expr "++fun++"("++arg_decls++") { return new Expr("++show fun++args++"); }"
   where
-    name  = showCId fun
     arity = maybe 0 getArrity (functionType gr fun)
     vars  = ['e':show i | i <- [1..arity]]
     

src/compiler/GF/Compile/PGFtoProlog.hs

@@ -1,262 +0,0 @@
-----------------------------------------------------------------------
--- |
--- Module      : PGFtoProlog
--- Maintainer  : Peter Ljunglöf
---
--- exports a GF grammar into a Prolog module
------------------------------------------------------------------------------
-
-module GF.Compile.PGFtoProlog (grammar2prolog) where
-
-import PGF(mkCId,wildCId,showCId)
-import PGF.Internal
---import PGF.Macros
-
-import GF.Data.Operations
-
-import qualified Data.Array.IArray as Array
-import qualified Data.Set as Set
-import qualified Data.Map as Map
-import qualified Data.IntMap as IntMap
-import Data.Char (isAlphaNum, isAscii, isAsciiLower, isAsciiUpper, ord)
-import Data.List (isPrefixOf, mapAccumL)
-
-grammar2prolog :: PGF -> String
-grammar2prolog pgf
-    = ("%% This file was automatically generated by GF" +++++
-       ":- style_check(-singleton)." +++++
-       plFacts wildCId "abstract" 1 "(?AbstractName)"
-                   [[plp name]] ++++
-       plFacts wildCId "concrete" 2 "(?AbstractName, ?ConcreteName)" 
-                   [[plp name, plp cncname] | 
-                    cncname <- Map.keys (concretes pgf)] ++++
-       plFacts wildCId "flag" 2 "(?Flag, ?Value): global flags" 
-                   [[plp f, plp v] | 
-                    (f, v) <- Map.assocs (gflags pgf)] ++++
-       plAbstract name (abstract pgf) ++++
-       unlines (map plConcrete (Map.assocs (concretes pgf)))
-      )
-    where name = absname pgf
-
-----------------------------------------------------------------------
--- abstract syntax
-
-plAbstract :: CId -> Abstr -> String
-plAbstract name abs 
-    = (plHeader "Abstract syntax" ++++
-       plFacts name "flag" 2 "(?Flag, ?Value): flags for abstract syntax"
-                   [[plp f, plp v] | 
-                    (f, v) <- Map.assocs (aflags abs)] ++++
-       plFacts name "cat" 2 "(?Type, ?[X:Type,...])"
-                   [[plType cat args, plHypos hypos'] |
-                    (cat, (hypos,_,_)) <- Map.assocs (cats abs),
-                    let ((_, subst), hypos') = mapAccumL alphaConvertHypo emptyEnv hypos,
-                        let args = reverse [EFun x | (_,x) <- subst]] ++++
-       plFacts name "fun" 3 "(?Fun, ?Type, ?[X:Type,...])"
-                   [[plp fun, plType cat args, plHypos hypos] |
-                    (fun, (typ, _, _, _)) <- Map.assocs (funs abs),
-                    let (_, DTyp hypos cat args) = alphaConvert emptyEnv typ] ++++
-       plFacts name "def" 2 "(?Fun, ?Expr)" 
-                   [[plp fun, plp expr] |
-                    (fun, (_, _, Just (eqs,_), _)) <- Map.assocs (funs abs),
-                    let (_, expr) = alphaConvert emptyEnv eqs]
-      )
-    where plType cat args = plTerm (plp cat) (map plp args)
-          plHypos hypos = plList [plOper ":" (plp x) (plp ty) | (_, x, ty) <- hypos]
-
-----------------------------------------------------------------------
--- concrete syntax
-
-plConcrete :: (CId, Concr) -> String
-plConcrete (name, cnc) 
-    = (plHeader ("Concrete syntax: " ++ plp name) ++++
-       plFacts name "flag" 2 "(?Flag, ?Value): flags for concrete syntax"
-                   [[plp f, plp v] | 
-                    (f, v) <- Map.assocs (cflags cnc)] ++++
-       plFacts name "printname" 2 "(?AbsFun/AbsCat, ?Atom)"
-                   [[plp f, plp n] | 
-                    (f, n) <- Map.assocs (printnames cnc)] ++++
-       plFacts name "lindef" 2 "(?CncCat, ?CncFun)"
-                   [[plCat cat, plFun fun] | 
-                    (cat, funs) <- IntMap.assocs (lindefs cnc),
-                    fun <- funs] ++++
-       plFacts name "prod" 3 "(?CncCat, ?CncFun, ?[CncCat])"
-                   [[plCat cat, fun, plTerm "c" (map plCat args)] |
-                    (cat, set) <- IntMap.toList (productions cnc),
-                    (fun, args) <- map plProduction (Set.toList set)] ++++
-       plFacts name "cncfun" 3 "(?CncFun, ?[Seq,...], ?AbsFun)"
-                   [[plFun fun, plTerm "s" (map plSeq (Array.elems lins)), plp absfun] |
-                    (fun, CncFun absfun lins) <- Array.assocs (cncfuns cnc)] ++++
-       plFacts name "seq" 2 "(?Seq, ?[Term])"
-                   [[plSeq seq, plp (Array.elems symbols)] |
-                    (seq, symbols) <- Array.assocs (sequences cnc)] ++++
-       plFacts name "cnccat" 2 "(?AbsCat, ?[CnCCat])" 
-                   [[plp cat, plList (map plCat [start..end])] |
-                    (cat, CncCat start end _) <- Map.assocs (cnccats cnc)]
-      )
-    where plProduction (PCoerce arg)       = ("-", [arg])
-          plProduction (PApply funid args) = (plFun funid, [fid | PArg hypos fid <- args])
-
-----------------------------------------------------------------------
--- prolog-printing pgf datatypes
-
-instance PLPrint Type where
-    plp (DTyp hypos cat args) 
-        | null hypos = result
-        | otherwise  = plOper " -> " plHypos result
-        where result = plTerm (plp cat) (map plp args)
-              plHypos = plList [plOper ":" (plp x) (plp ty) | (_,x,ty) <- hypos]
-
-instance PLPrint Expr where
-    plp (EFun x)    = plp x
-    plp (EAbs _ x e)= plOper "^" (plp x) (plp e)
-    plp (EApp e e') = plOper " * " (plp e) (plp e')
-    plp (ELit lit)  = plp lit
-    plp (EMeta n)   = "Meta_" ++ show n
-
-instance PLPrint Patt where
-    plp (PVar x)    = plp x
-    plp (PApp f ps) = plOper " * " (plp f) (plp ps)
-    plp (PLit lit)  = plp lit
-
-instance PLPrint Equation where
-    plp (Equ patterns result) = plOper ":" (plp patterns) (plp result)
-
-instance PLPrint CId where
-    plp cid | isLogicalVariable str || cid == wildCId = plVar str
-            | otherwise = plAtom str
-        where str = showCId cid
-
-instance PLPrint Literal where
-    plp (LStr s) = plp s
-    plp (LInt n) = plp (show n)
-    plp (LFlt f) = plp (show f)
-
-instance PLPrint Symbol where
-    plp (SymCat n l)    = plOper ":" (show n) (show l)
-    plp (SymLit n l)    = plTerm "lit" [show n, show l]
-    plp (SymVar n l)    = plTerm "var" [show n, show l]
-    plp (SymKS t)       = plAtom t
-    plp (SymKP ts alts) = plTerm "pre" [plList (map plp ts), plList (map plAlt alts)]
-        where plAlt (ps,ts) = plOper "/" (plList (map plp ps)) (plList (map plAtom ts))
-
-class PLPrint a where
-    plp :: a -> String
-    plps :: [a] -> String
-    plps = plList . map plp
-
-instance PLPrint Char where
-    plp  c = plAtom [c]
-    plps s = plAtom s
-
-instance PLPrint a => PLPrint [a] where
-    plp = plps
-
-----------------------------------------------------------------------
--- other prolog-printing functions
-
-plCat :: Int -> String
-plCat n = plAtom ('c' : show n)
-
-plFun :: Int -> String
-plFun n = plAtom ('f' : show n)
-
-plSeq :: Int -> String
-plSeq n = plAtom ('s' : show n)
-
-plHeader :: String -> String
-plHeader hdr = "%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%\n%% " ++ hdr ++ "\n"
-
-plFacts :: CId -> String -> Int -> String -> [[String]] -> String
-plFacts mod pred arity comment facts = "%% " ++ pred ++ comment ++++ clauses
-    where clauses = (if facts == [] then ":- dynamic " ++ pred ++ "/" ++ show arity ++ ".\n"
-                     else unlines [mod' ++ plTerm pred args ++ "." | args <- facts])
-          mod' = if mod == wildCId then "" else plp mod ++ ": "
-
-plTerm :: String -> [String] -> String
-plTerm fun args = plAtom fun ++ prParenth (prTList ", " args)
-
-plList :: [String] -> String
-plList xs = prBracket (prTList "," xs)
-
-plOper :: String -> String -> String -> String
-plOper op a b = prParenth (a ++ op ++ b)
-
-plVar :: String -> String
-plVar = varPrefix  . concatMap changeNonAlphaNum 
-    where varPrefix var@(c:_) | isAsciiUpper c || c=='_' = var
-                              | otherwise = "_" ++ var
-          changeNonAlphaNum c | isAlphaNumUnderscore c = [c]
-                              | otherwise = "_" ++ show (ord c) ++ "_"
-
-plAtom :: String -> String
-plAtom "" = "''"
-plAtom atom@(c:cs) | isAsciiLower c && all isAlphaNumUnderscore cs 
-                     || c == '\'' && cs /= "" && last cs == '\''   = atom
-                   | otherwise = "'" ++ changeQuote atom ++ "'"
-    where changeQuote ('\'':cs) = '\\' : '\'' : changeQuote cs
-          changeQuote ('\\':cs) = '\\' : '\\' : changeQuote cs
-          changeQuote (c:cs) = c : changeQuote cs
-          changeQuote "" = ""
-
-isAlphaNumUnderscore :: Char -> Bool
-isAlphaNumUnderscore c = (isAscii c && isAlphaNum c) || c == '_'
-
-----------------------------------------------------------------------
--- prolog variables 
-
-createLogicalVariable :: Int -> CId
-createLogicalVariable n = mkCId (logicalVariablePrefix ++ show n)
-
-isLogicalVariable :: String -> Bool
-isLogicalVariable = isPrefixOf logicalVariablePrefix 
-
-logicalVariablePrefix :: String 
-logicalVariablePrefix = "X"
-
-----------------------------------------------------------------------
--- alpha convert variables to (unique) logical variables
--- * this is needed if we want to translate variables to Prolog variables
--- * used for abstract syntax, not concrete
--- * not (yet?) used for variables bound in pattern equations
-
-type ConvertEnv = (Int, [(CId,CId)])
-
-emptyEnv :: ConvertEnv
-emptyEnv = (0, [])
-
-class AlphaConvert a where
-    alphaConvert :: ConvertEnv -> a -> (ConvertEnv, a)
-
-instance AlphaConvert a => AlphaConvert [a] where
-    alphaConvert env [] = (env, [])
-    alphaConvert env (a:as) = (env'', a':as')
-        where (env',  a')  = alphaConvert env  a
-              (env'', as') = alphaConvert env' as
-
-instance AlphaConvert Type where
-    alphaConvert env@(_,subst) (DTyp hypos cat args) 
-        = ((ctr,subst), DTyp hypos' cat args')
-        where (env',   hypos') = mapAccumL alphaConvertHypo env hypos
-              ((ctr,_), args') = alphaConvert env' args
-
-alphaConvertHypo env (b,x,typ) = ((ctr+1,(x,x'):subst), (b,x',typ'))
-    where ((ctr,subst), typ') = alphaConvert env typ
-          x' = createLogicalVariable ctr
-
-instance AlphaConvert Expr where
-    alphaConvert (ctr,subst) (EAbs b x e) = ((ctr',subst), EAbs b x' e')
-        where ((ctr',_), e') = alphaConvert (ctr+1,(x,x'):subst) e
-              x' = createLogicalVariable ctr
-    alphaConvert env (EApp e1 e2) = (env'', EApp e1' e2')
-        where (env',  e1') = alphaConvert env  e1
-              (env'', e2') = alphaConvert env' e2
-    alphaConvert env expr@(EFun i) = (env, maybe expr EFun (lookup i (snd env)))
-    alphaConvert env expr = (env, expr)
-
--- pattern variables are not alpha converted
--- (but they probably should be...)
-instance AlphaConvert Equation where
-    alphaConvert env@(_,subst) (Equ patterns result)
-        = ((ctr,subst), Equ patterns result')
-        where ((ctr,_), result') = alphaConvert env result

src/compiler/GF/Compile/PGFtoPython.hs

@@ -1,122 +0,0 @@
-----------------------------------------------------------------------
--- |
--- Module      : PGFtoPython
--- Maintainer  : Peter Ljunglöf
---
--- exports a GF grammar into a Python module
------------------------------------------------------------------------------
-
-{-# LANGUAGE FlexibleContexts #-}
-module GF.Compile.PGFtoPython (pgf2python) where
-
-import PGF(showCId)
-import PGF.Internal as M
-
-import GF.Data.Operations
-
-import qualified Data.Array.IArray as Array
-import qualified Data.Set as Set
-import qualified Data.Map as Map
-import qualified Data.IntMap as IntMap
---import Data.List (intersperse)
-
-pgf2python :: PGF -> String
-pgf2python pgf = ("# -*- coding: utf-8 -*-" ++++
-                  "# This file was automatically generated by GF" +++++
-                  showCId name +++ "=" +++ 
-                  pyDict 1 pyStr id [
-                              ("flags", pyDict 2 pyCId pyLiteral (Map.assocs (gflags pgf))),
-                              ("abstract", pyDict 2 pyStr id [
-                                              ("name", pyCId name), 
-                                              ("start", pyCId start), 
-                                              ("flags", pyDict 3 pyCId pyLiteral (Map.assocs (aflags abs))),
-                                              ("funs", pyDict 3 pyCId pyAbsdef (Map.assocs (funs abs)))
-                                             ]),
-                              ("concretes", pyDict 2 pyCId pyConcrete (Map.assocs cncs))
-                             ] ++ "\n")
-    where
-      name  = absname pgf
-      start = M.lookStartCat pgf
-      abs = abstract pgf
-      cncs = concretes pgf
-
-pyAbsdef :: (Type, Int, Maybe ([Equation], [[M.Instr]]), Double) -> String
-pyAbsdef (typ, _, _, _) = pyTuple 0 id [pyCId cat, pyList 0 pyCId args]
-    where (args, cat) = M.catSkeleton typ 
-
-pyLiteral :: Literal -> String
-pyLiteral (LStr s) = pyStr s
-pyLiteral (LInt n) = show n
-pyLiteral (LFlt d) = show d
-
-pyConcrete :: Concr -> String
-pyConcrete cnc = pyDict 3 pyStr id [
-                  ("flags", pyDict 0 pyCId pyLiteral (Map.assocs (cflags cnc))),
-                  ("printnames", pyDict 4 pyCId pyStr (Map.assocs (printnames cnc))),
-                  ("lindefs", pyDict 4 pyCat (pyList 0 pyFun) (IntMap.assocs (lindefs cnc))),
-                  ("productions", pyDict 4 pyCat pyProds (IntMap.assocs (productions cnc))),
-                  ("cncfuns", pyDict 4 pyFun pyCncFun (Array.assocs (cncfuns cnc))),
-                  ("sequences",  pyDict 4 pySeq pySymbols (Array.assocs (sequences cnc))),
-                  ("cnccats", pyDict 4 pyCId pyCncCat (Map.assocs (cnccats cnc))),
-                  ("size",  show (totalCats cnc))
-                 ]
-    where pyProds prods = pyList 5 pyProduction (Set.toList prods)
-          pyCncCat (CncCat start end _) = pyList 0 pyCat [start..end]
-          pyCncFun (CncFun f lins) = pyTuple 0 id [pyList 0 pySeq (Array.elems lins), pyCId f]
-          pySymbols syms = pyList 0 pySymbol (Array.elems syms)
-
-pyProduction :: Production -> String
-pyProduction (PCoerce arg)       = pyTuple 0 id [pyStr "", pyList 0 pyCat [arg]]
-pyProduction (PApply funid args) = pyTuple 0 id [pyFun funid, pyList 0 pyPArg args]
-    where pyPArg (PArg [] fid) = pyCat fid
-          pyPArg (PArg hypos fid) = pyTuple 0 pyCat (fid : map snd hypos)
-
-pySymbol :: Symbol -> String
-pySymbol (SymCat n l)    = pyTuple 0 show [n, l]
-pySymbol (SymLit n l)    = pyDict 0 pyStr id [("lit", pyTuple 0 show [n, l])]
-pySymbol (SymVar n l)    = pyDict 0 pyStr id [("var", pyTuple 0 show [n, l])]
-pySymbol (SymKS t)       = pyStr t
-pySymbol (SymKP ts alts) = pyDict 0 pyStr id [("pre", pyList 0 pySymbol ts), ("alts", pyList 0 alt2py alts)]
-    where alt2py (ps,ts) = pyTuple 0 (pyList 0 pyStr) [map pySymbol ps, ts]
-pySymbol SymBIND         = pyStr "&+"
-pySymbol SymSOFT_BIND    = pyStr "&+"
-pySymbol SymSOFT_SPACE   = pyStr "&+"
-pySymbol SymCAPIT        = pyStr "&|"
-pySymbol SymALL_CAPIT    = pyStr "&|"
-pySymbol SymNE           = pyDict 0 pyStr id [("nonExist", pyTuple 0 id [])]
-
-----------------------------------------------------------------------
--- python helpers 
-
-pyDict :: Int -> (k -> String) -> (v -> String) -> [(k, v)] -> String
-pyDict n pk pv [] = "{}"
-pyDict n pk pv kvlist = prCurly (pyIndent n ++ prTList ("," ++ pyIndent n) (map pyKV kvlist) ++ pyIndent n)
-    where pyKV (k, v) = pk k ++ ":" ++ pv v
-
-pyList :: Int -> (v -> String) -> [v] -> String
-pyList n pv [] = "[]"
-pyList n pv xs = prBracket (pyIndent n ++ prTList ("," ++ pyIndent n) (map pv xs) ++ pyIndent n)
-
-pyTuple :: Int -> (v -> String) -> [v] -> String
-pyTuple n pv [] = "()"
-pyTuple n pv [x] = prParenth (pyIndent n ++ pv x ++ "," ++ pyIndent n)
-pyTuple n pv xs = prParenth (pyIndent n ++ prTList ("," ++ pyIndent n) (map pv xs) ++ pyIndent n)
-
-pyCat :: Int -> String
-pyCat n = pyStr ('C' : show n)
-
-pyFun :: Int -> String
-pyFun n = pyStr ('F' : show n)
-
-pySeq :: Int -> String
-pySeq n = pyStr ('S' : show n)
-
-pyStr :: String -> String
-pyStr s = 'u' : prQuotedString s
-
-pyCId :: CId -> String
-pyCId = pyStr . showCId
-
-pyIndent :: Int -> String
-pyIndent n | n > 0 = "\n" ++ replicate n ' '
-           | otherwise = ""

src/compiler/GF/Compile/Rename.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/05/30 18:39:44 $ 
+-- > CVS $Date: 2005/05/30 18:39:44 $
 -- > CVS $Author: aarne $
 -- > CVS $Revision: 1.19 $
 --
@@ -23,9 +23,9 @@
 -----------------------------------------------------------------------------
 
 module GF.Compile.Rename (
-	       renameSourceTerm,
-	       renameModule
-	      ) where
+     renameSourceTerm,
+     renameModule
+    ) where
 
 import GF.Infra.Ident
 import GF.Infra.CheckM
@@ -39,6 +39,7 @@ import GF.Data.Operations
 
 import Control.Monad
 import Data.List (nub,(\\))
+import qualified Data.List as L
 import qualified Data.Map as Map
 import Data.Maybe(mapMaybe)
 import GF.Text.Pretty
@@ -67,7 +68,7 @@ renameIdentTerm env = accumulateError (renameIdentTerm' env)
 
 -- Fails immediately on error, makes it possible to try other possibilities
 renameIdentTerm' :: Status -> Term -> Check Term
-renameIdentTerm' env@(act,imps) t0 = 
+renameIdentTerm' env@(act,imps) t0 =
   case t0 of
     Vr c -> ident predefAbs c
     Cn c -> ident (\_ s -> checkError s) c
@@ -84,8 +85,8 @@ renameIdentTerm' env@(act,imps) t0 =
     _ -> return t0
   where
     opens   = [st  | (OSimple _,st) <- imps]
-    qualifs = [(m, st) | (OQualif m _, st) <- imps] ++ 
-              [(m, st) | (OQualif _ m, st) <- imps] ++ 
+    qualifs = [(m, st) | (OQualif m _, st) <- imps] ++
+              [(m, st) | (OQualif _ m, st) <- imps] ++
               [(m, st) | (OSimple m, st) <- imps] -- qualif is always possible
 
     -- this facility is mainly for BWC with GF1: you need not import PredefAbs
@@ -93,7 +94,7 @@ renameIdentTerm' env@(act,imps) t0 =
       | isPredefCat c = return (Q (cPredefAbs,c))
       | otherwise     = checkError s
 
-    ident alt c = 
+    ident alt c =
       case Map.lookup c act of
         Just f -> return (f c)
         _      -> case mapMaybe (Map.lookup c) opens of
@@ -105,13 +106,32 @@ renameIdentTerm' env@(act,imps) t0 =
                               ts@(t:_) -> do checkWarn ("atomic term" <+> ppTerm Qualified 0 t0 $$
                                                         "conflict" <+> hsep (punctuate ',' (map (ppTerm Qualified 0) ts)) $$
                                                         "given" <+> fsep (punctuate ',' (map fst qualifs)))
-                                             return t
+                                             return (bestTerm ts) -- Heuristic for resource grammar. Returns t for all others.
+       where
+        -- Hotfix for https://github.com/GrammaticalFramework/gf-core/issues/56
+        -- Real bug is probably somewhere deeper in recognising excluded functions. /IL 2020-06-06
+        notFromCommonModule :: Term -> Bool
+        notFromCommonModule term =
+          let t = render $ ppTerm Qualified 0 term :: String
+           in not $ any (\moduleName -> moduleName `L.isPrefixOf` t)
+                        ["CommonX", "ConstructX", "ExtendFunctor"
+                        ,"MarkHTMLX", "ParamX", "TenseX", "TextX"]
+
+        -- If one of the terms comes from the common modules,
+        -- we choose the other one, because that's defined in the grammar.
+        bestTerm :: [Term] -> Term
+        bestTerm [] = error "constant not found" -- not reached: bestTerm is only called for case ts@(t:_)
+        bestTerm ts@(t:_) =
+          let notCommon = [t | t <- ts, notFromCommonModule t]
+           in case notCommon of
+                []    -> t -- All terms are from common modules, return first of original list
+                (u:_) -> u -- ≥1 terms are not from common modules, return first of those
 
 info2status :: Maybe ModuleName -> Ident -> Info -> StatusInfo
 info2status mq c i = case i of
   AbsFun _ _ Nothing _ -> maybe Con (curry QC) mq
-  ResValue _  -> maybe Con (curry QC) mq
-  ResParam _ _  -> maybe Con (curry QC) mq
+  ResValue _ _ -> maybe Con (curry QC) mq
+  ResParam _ _ -> maybe Con (curry QC) mq
   AnyInd True m -> maybe Con (const (curry QC m)) mq
   AnyInd False m -> maybe Cn (const (curry Q m)) mq
   _           -> maybe Cn (curry Q) mq
@@ -137,7 +157,7 @@ modInfo2status (o,mo) = (o,tree2status o (jments mo))
 self2status :: ModuleName -> ModuleInfo -> StatusMap
 self2status c m = Map.mapWithKey (info2status (Just c)) (jments m)
 
-  
+
 renameInfo :: FilePath -> Status -> Module -> Ident -> Info -> Check Info
 renameInfo cwd status (m,mi) i info =
   case info of
@@ -148,9 +168,9 @@ renameInfo cwd status (m,mi) i info =
     ResParam (Just pp) m -> do
       pp' <- renLoc (mapM (renParam status)) pp
       return (ResParam (Just pp') m)
-    ResValue t -> do
+    ResValue t i -> do
       t <- renLoc (renameTerm status []) t
-      return (ResValue t)
+      return (ResValue t i)
     CncCat mcat mdef mref mpr mpmcfg -> liftM5 CncCat (renTerm mcat) (renTerm mdef) (renTerm mref) (renTerm mpr) (return mpmcfg)
     CncFun mty mtr mpr mpmcfg -> liftM3 (CncFun mty)         (renTerm mtr) (renTerm mpr) (return mpmcfg)
     _ -> return info
@@ -188,7 +208,7 @@ renameTerm env vars = ren vars where
     Abs b x t    -> liftM  (Abs b x) (ren (x:vs) t)
     Prod bt x a b -> liftM2 (Prod bt x) (ren vs a) (ren (x:vs) b)
     Typed a b  -> liftM2 Typed (ren vs a) (ren vs b)
-    Vr x      
+    Vr x
       | elem x vs -> return trm
       | otherwise -> renid trm
     Cn _   -> renid trm
@@ -199,7 +219,7 @@ renameTerm env vars = ren vars where
       i' <- case i of
         TTyped ty -> liftM TTyped $ ren vs ty -- the only annotation in source
         _ -> return i
-      liftM (T i') $ mapM (renCase vs) cs  
+      liftM (T i') $ mapM (renCase vs) cs
 
     Let (x,(m,a)) b -> do
       m' <- case m of
@@ -209,7 +229,7 @@ renameTerm env vars = ren vars where
       b' <- ren (x:vs) b
       return $ Let (x,(m',a')) b'
 
-    P t@(Vr r) l                                               -- Here we have $r.l$ and this is ambiguous it could be either 
+    P t@(Vr r) l                                               -- Here we have $r.l$ and this is ambiguous it could be either
                                                                -- record projection from variable or constant $r$ or qualified expression with module $r$
       | elem r vs -> return trm                                -- try var proj first ..
       | otherwise -> checks [ renid' (Q (MN r,label2ident l))      -- .. and qualified expression second.
@@ -217,9 +237,9 @@ renameTerm env vars = ren vars where
                             , checkError ("unknown qualified constant" <+> trm)
                             ]
 
-    EPatt p -> do
+    EPatt minp maxp p -> do
       (p',_) <- renpatt p
-      return $ EPatt p'
+      return $ EPatt minp maxp p'
 
     _ -> composOp (ren vs) trm
 
@@ -286,14 +306,14 @@ renamePattern env patt =
         (q',ws) <- renp q
         return (PAlt p' q', vs ++ ws)
 
-      PSeq p q -> do
+      PSeq minp maxp p minq maxq q -> do
         (p',vs) <- renp p
         (q',ws) <- renp q
-        return (PSeq p' q', vs ++ ws)
+        return (PSeq minp maxp p' minq maxq q', vs ++ ws)
 
-      PRep p -> do
+      PRep minp maxp p -> do
         (p',vs) <- renp p
-        return (PRep p', vs)
+        return (PRep minp maxp p', vs)
 
       PNeg p -> do
         (p',vs) <- renp p
@@ -311,7 +331,7 @@ renamePattern env patt =
 renameContext :: Status -> Context -> Check Context
 renameContext b = renc [] where
   renc vs cont = case cont of
-    (bt,x,t) : xts 
+    (bt,x,t) : xts
       | isWildIdent x -> do
           t'   <- ren vs t
           xts' <- renc vs xts

src/compiler/GF/Compile/Tags.hs

@@ -31,7 +31,7 @@ getLocalTags x (m,mi) =
     getLocations (AbsFun mb_type _ mb_eqs _)    = maybe (ltype "fun")        mb_type ++
                                                   maybe (list (loc "def"))   mb_eqs  
     getLocations (ResParam mb_params _)         = maybe (loc "param")        mb_params
-    getLocations (ResValue mb_type)             = ltype "param-value"          mb_type
+    getLocations (ResValue mb_type _)           = ltype "param-value"          mb_type
     getLocations (ResOper  mb_type mb_def)      = maybe (ltype "oper-type")  mb_type ++
                                                   maybe (loc "oper-def")     mb_def
     getLocations (ResOverload _ defs)           = list (\(x,y) -> ltype "overload-type" x ++ 

src/compiler/GF/Compile/ToAPI.hs

@@ -2,8 +2,7 @@ module GF.Compile.ToAPI
  (stringToAPI,exprToAPI)
   where
 
-import PGF.Internal
-import PGF(showCId)
+import PGF2
 import Data.Maybe
 --import System.IO
 --import Control.Monad
@@ -47,12 +46,12 @@ exprToFunc :: Expr -> APIfunc
 exprToFunc expr = 
    case unApp expr of
       Just (cid,l) -> 
-         case Map.lookup (showCId cid) syntaxFuncs of 
+         case Map.lookup cid syntaxFuncs of 
             Just sig -> mkAPI True (fst sig,expr)
             _        -> case l of 
-                          [] -> BasicFunc (showCId cid) 
+                          [] -> BasicFunc cid 
                           _  -> let es = map exprToFunc l 
-                                   in  AppFunc (showCId cid) es
+                                   in  AppFunc cid es
       _ -> BasicFunc (showExpr [] expr)
 
         
@@ -69,8 +68,8 @@ mkAPI opt (ty,expr) =
   where
      rephraseSentence ty expr = 
        case unApp expr of 
-         Just (cid,es) -> if isPrefixOf "Use" (showCId cid) then 
-                                                             let newCat = drop 3 (showCId cid)
+         Just (cid,es) -> if isPrefixOf "Use" cid then 
+                                                             let newCat = drop 3 cid
                                                                  afClause = mkAPI True (newCat, es !! 2)
                                                                  afPol = mkAPI True ("Pol",es !! 1)                                         
                                                                  lTense = mkAPI True ("Temp", head es)                                        
@@ -98,9 +97,9 @@ mkAPI opt (ty,expr) =
 computeAPI :: (String,Expr) ->  APIfunc
 computeAPI (ty,expr) =
    case (unApp expr) of 
-     Just (cid,[]) ->  getSimpCat (showCId cid) ty
+     Just (cid,[]) ->  getSimpCat cid ty
      Just (cid,es) -> 
-        let p = specFunction (showCId cid) es  
+        let p = specFunction cid es  
           in if isJust p then fromJust p
               else case Map.lookup (show cid) syntaxFuncs of
                     Nothing -> exprToFunc expr    
@@ -147,23 +146,23 @@ optimize expr = optimizeNP expr
 optimizeNP expr = 
    case unApp expr of
      Just (cid,es) -> 
-         if showCId cid == "MassNP" then let afs = nounAsCN (head es)
-                                           in AppFunc "mkNP" [afs]
-           else if showCId cid == "DetCN" then let quants = quantAsDet (head es)
-                                                   ns     = nounAsCN (head $ tail es)
-                                                  in AppFunc "mkNP" (quants ++ [ns])
+         if cid == "MassNP" then let afs = nounAsCN (head es)
+                                 in AppFunc "mkNP" [afs]
+           else if cid == "DetCN" then let quants = quantAsDet (head es)
+                                           ns     = nounAsCN (head $ tail es)
+                                       in AppFunc "mkNP" (quants ++ [ns])
                  else mkAPI False ("NP",expr)
      _             -> error $ "incorrect expression " ++ (showExpr [] expr)
     where 
      nounAsCN expr = 
       case unApp expr of 
-       Just (cid,es) -> if showCId cid == "UseN" then (mkAPI False) ("N",head es)
+       Just (cid,es) -> if cid == "UseN" then (mkAPI False) ("N",head es)
            else (mkAPI False) ("CN",expr)
        _ -> error $ "incorrect expression "++ (showExpr [] expr)
 
      quantAsDet expr =
        case unApp expr of
-        Just (cid,es) -> if showCId cid == "DetQuant" then map (mkAPI False) [("Quant", head es),("Num",head $ tail es)]
+        Just (cid,es) -> if cid == "DetQuant" then map (mkAPI False) [("Quant", head es),("Num",head $ tail es)]
                            else [mkAPI False ("Det",expr)]
                                   
         _             -> error $ "incorrect expression "++ (showExpr [] expr)

src/compiler/GF/Compile/TypeCheck/Abstract.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/09/15 16:22:02 $ 
+-- > CVS $Date: 2005/09/15 16:22:02 $
 -- > CVS $Author: aarne $
 -- > CVS $Revision: 1.16 $
 --
@@ -13,11 +13,11 @@
 -----------------------------------------------------------------------------
 
 module GF.Compile.TypeCheck.Abstract (-- * top-level type checking functions; TC should not be called directly.
-		  checkContext,
-		  checkTyp,
-		  checkDef,
-		  checkConstrs,
-		 ) where
+    checkContext,
+    checkTyp,
+    checkDef,
+    checkConstrs,
+   ) where
 
 import GF.Data.Operations
 
@@ -33,8 +33,8 @@ import GF.Text.Pretty
 --import Control.Monad (foldM, liftM, liftM2)
 
 -- | invariant way of creating TCEnv from context
-initTCEnv gamma = 
-  (length gamma,[(x,VGen i x) | ((x,_),i) <- zip gamma [0..]], gamma) 
+initTCEnv gamma =
+  (length gamma,[(x,VGen i x) | ((x,_),i) <- zip gamma [0..]], gamma)
 
 -- interface to TC type checker
 

src/compiler/GF/Compile/TypeCheck/Concrete.hs

@@ -1,6 +1,7 @@
 {-# LANGUAGE PatternGuards #-}
-module GF.Compile.TypeCheck.Concrete( {-checkLType, inferLType, computeLType, ppType-} ) where
-{-
+module GF.Compile.TypeCheck.Concrete( checkLType, inferLType, computeLType, ppType ) where
+import Prelude hiding ((<>)) -- GHC 8.4.1 clash with Text.PrettyPrint
+
 import GF.Infra.CheckM
 import GF.Data.Operations
 
@@ -12,6 +13,7 @@ import GF.Grammar.Lockfield (isLockLabel, lockRecType, unlockRecord)
 import GF.Compile.TypeCheck.Primitives
 
 import Data.List
+import Data.Maybe(fromMaybe)
 import Control.Monad
 import GF.Text.Pretty
 
@@ -22,10 +24,16 @@ computeLType gr g0 t = comp (reverse [(b,x, Vr x) | (b,x,_) <- g0] ++ g0) t
     _ | Just _ <- isTypeInts ty -> return ty ---- shouldn't be needed
       | isPredefConstant ty     -> return ty ---- shouldn't be needed
 
-    Q (m,ident) -> checkIn (text "module" <+> ppIdent m) $ do
+    Q (m,ident) -> checkIn ("module" <+> m) $ do
       ty' <- lookupResDef gr (m,ident)
       if ty' == ty then return ty else comp g ty' --- is this necessary to test?
 
+    AdHocOverload ts -> do
+     over <- getOverload gr g (Just typeType) t
+     case over of
+       Just (tr,_) -> return tr
+       _ -> checkError ("unresolved overloading of constants" <+> ppTerm Qualified 0 t)
+
     Vr ident  -> checkLookup ident g -- never needed to compute!
 
     App f a -> do
@@ -62,7 +70,6 @@ computeLType gr g0 t = comp (reverse [(b,x, Vr x) | (b,x,_) <- g0] ++ g0) t
       lockRecType c t' ---- locking to be removed AR 20/6/2009
 
     _ | ty == typeTok -> return typeStr
-    _ | isPredefConstant ty -> return ty
 
     _ -> composOp (comp g) ty
 
@@ -73,26 +80,26 @@ inferLType gr g trm = case trm of
 
    Q (m,ident) | isPredef m -> termWith trm $ case typPredefined ident of
                                                 Just ty -> return ty
-                                                Nothing -> checkError (text "unknown in Predef:" <+> ppIdent ident)
+                                                Nothing -> checkError ("unknown in Predef:" <+> ident)
 
    Q ident -> checks [
      termWith trm $ lookupResType gr ident >>= computeLType gr g
      ,
      lookupResDef gr ident >>= inferLType gr g
      ,
-     checkError (text "cannot infer type of constant" <+> ppTerm Unqualified 0 trm)
+     checkError ("cannot infer type of constant" <+> ppTerm Unqualified 0 trm)
      ]
 
    QC (m,ident) | isPredef m -> termWith trm $ case typPredefined ident of
                                                  Just ty -> return ty
-                                                 Nothing -> checkError (text "unknown in Predef:" <+> ppIdent ident)
+                                                 Nothing -> checkError ("unknown in Predef:" <+> ident)
 
    QC ident -> checks [
        termWith trm $ lookupResType gr ident >>= computeLType gr g
        ,
        lookupResDef gr ident >>= inferLType gr g
        ,
-       checkError (text "cannot infer type of canonical constant" <+> ppTerm Unqualified 0 trm)
+       checkError ("cannot infer type of canonical constant" <+> ppTerm Unqualified 0 trm)
        ]
 
    Vr ident -> termWith trm $ checkLookup ident g
@@ -100,7 +107,12 @@ inferLType gr g trm = case trm of
    Typed e t -> do
      t' <- computeLType gr g t
      checkLType gr g e t'
-     return (e,t')
+
+   AdHocOverload ts -> do
+     over <- getOverload gr g Nothing trm
+     case over of
+       Just trty -> return trty
+       _ -> checkError ("unresolved overloading of constants" <+> ppTerm Qualified 0 trm)
 
    App f a -> do
      over <- getOverload gr g Nothing trm
@@ -110,13 +122,17 @@ inferLType gr g trm = case trm of
          (f',fty) <- inferLType gr g f
          fty' <- computeLType gr g fty
          case fty' of
-           Prod bt z arg val -> do 
+           Prod bt z arg val -> do
              a' <- justCheck g a arg
-             ty <- if isWildIdent z 
+             ty <- if isWildIdent z
                       then return val
                       else substituteLType [(bt,z,a')] val
-             return (App f' a',ty) 
-           _ -> checkError (text "A function type is expected for" <+> ppTerm Unqualified 0 f <+> text "instead of type" <+> ppType fty)
+             return (App f' a',ty)
+           _ ->
+             let term = ppTerm Unqualified 0 f
+                 funName = pp . head . words .render $ term
+              in checkError ("A function type is expected for" <+> term <+> "instead of type" <+> ppType fty $$
+                             "\n   ** Maybe you gave too many arguments to" <+> funName <+> "\n")
 
    S f x -> do
      (f', fty) <- inferLType gr g f
@@ -124,7 +140,7 @@ inferLType gr g trm = case trm of
        Table arg val -> do
          x'<- justCheck g x arg
          return (S f' x', val)
-       _ -> checkError (text "table lintype expected for the table in" $$ nest 2 (ppTerm Unqualified 0 trm))
+       _ -> checkError ("table lintype expected for the table in" $$ nest 2 (ppTerm Unqualified 0 trm))
 
    P t i -> do
      (t',ty) <- inferLType gr g t   --- ??
@@ -132,16 +148,16 @@ inferLType gr g trm = case trm of
      let tr2 = P t' i
      termWith tr2 $ case ty' of
        RecType ts -> case lookup i ts of
-                       Nothing -> checkError (text "unknown label" <+> ppLabel i <+> text "in" $$ nest 2 (ppTerm Unqualified 0 ty'))
+                       Nothing -> checkError ("unknown label" <+> i <+> "in" $$ nest 2 (ppTerm Unqualified 0 ty'))
                        Just x  -> return x
-       _          -> checkError (text "record type expected for:" <+> ppTerm Unqualified 0 t $$
-                                 text " instead of the inferred:" <+> ppTerm Unqualified 0 ty')
+       _          -> checkError ("record type expected for:" <+> ppTerm Unqualified 0 t $$
+                                 " instead of the inferred:" <+> ppTerm Unqualified 0 ty')
 
    R r -> do
      let (ls,fs) = unzip r
      fsts <- mapM inferM fs
      let ts = [ty | (Just ty,_) <- fsts]
-     checkCond (text "cannot infer type of record" $$ nest 2 (ppTerm Unqualified 0 trm)) (length ts == length fsts)
+     checkCond ("cannot infer type of record" $$ nest 2 (ppTerm Unqualified 0 trm)) (length ts == length fsts)
      return $ (R (zip ls fsts), RecType (zip ls ts))
 
    T (TTyped arg) pts -> do
@@ -152,10 +168,10 @@ inferLType gr g trm = case trm of
      checkLType gr g trm (Table arg val)
    T ti pts -> do  -- tries to guess: good in oper type inference
      let pts' = [pt | pt@(p,_) <- pts, isConstPatt p]
-     case pts' of 
-       [] -> checkError (text "cannot infer table type of" <+> ppTerm Unqualified 0 trm)
-----       PInt k : _ -> return $ Ints $ max [i | PInt i <- pts'] 
-       _  -> do 
+     case pts' of
+       [] -> checkError ("cannot infer table type of" <+> ppTerm Unqualified 0 trm)
+----       PInt k : _ -> return $ Ints $ max [i | PInt i <- pts']
+       _  -> do
          (arg,val) <- checks $ map (inferCase Nothing) pts'
          checkLType gr g trm (Table arg val)
    V arg pts -> do
@@ -166,9 +182,9 @@ inferLType gr g trm = case trm of
    K s  -> do
      if elem ' ' s
         then do
-          let ss = foldr C Empty (map K (words s))  
+          let ss = foldr C Empty (map K (words s))
           ----- removed irritating warning AR 24/5/2008
-          ----- checkWarn ("token \"" ++ s ++ 
+          ----- checkWarn ("token \"" ++ s ++
           -----              "\" converted to token list" ++ prt ss)
           return (ss, typeStr)
         else return (trm, typeStr)
@@ -179,50 +195,56 @@ inferLType gr g trm = case trm of
 
    Empty -> return (trm, typeStr)
 
-   C s1 s2 -> 
+   C s1 s2 ->
      check2 (flip (justCheck g) typeStr) C s1 s2 typeStr
 
-   Glue s1 s2 -> 
+   Glue s1 s2 ->
      check2 (flip (justCheck g) typeStr) Glue s1 s2 typeStr ---- typeTok
 
 ---- hack from Rename.identRenameTerm, to live with files with naming conflicts 18/6/2007
    Strs (Cn c : ts) | c == cConflict -> do
-     checkWarn (text "unresolved constant, could be any of" <+> hcat (map (ppTerm Unqualified 0) ts))
+     checkWarn ("unresolved constant, could be any of" <+> hcat (map (ppTerm Unqualified 0) ts))
      inferLType gr g (head ts)
 
    Strs ts -> do
-     ts' <- mapM (\t -> justCheck g t typeStr) ts 
+     ts' <- mapM (\t -> justCheck g t typeStr) ts
      return (Strs ts', typeStrs)
 
    Alts t aa -> do
      t'  <- justCheck g t typeStr
      aa' <- flip mapM aa (\ (c,v) -> do
-        c' <- justCheck g c typeStr 
+        c' <- justCheck g c typeStr
         v' <- checks $ map (justCheck g v) [typeStrs, EPattType typeStr]
         return (c',v'))
      return (Alts t' aa', typeStr)
 
    RecType r -> do
      let (ls,ts) = unzip r
-     ts' <- mapM (flip (justCheck g) typeType) ts 
+     ts' <- mapM (flip (justCheck g) typeType) ts
      return (RecType (zip ls ts'), typeType)
 
    ExtR r s -> do
-     (r',rT) <- inferLType gr g r 
+
+---     over <- getOverload gr g Nothing r
+---     let r1 = maybe r fst over
+     let r1 = r ---
+
+     (r',rT) <- inferLType gr g r1
      rT'     <- computeLType gr g rT
+
      (s',sT) <- inferLType gr g s
      sT'     <- computeLType gr g sT
 
      let trm' = ExtR r' s'
-     ---- trm'    <- plusRecord r' s'
      case (rT', sT') of
        (RecType rs, RecType ss) -> do
-         rt <- plusRecType rT' sT'
+         let rt = RecType ([field | field@(l,_) <- rs, notElem l (map fst ss)] ++ ss) -- select types of later fields
          checkLType gr g trm' rt ---- return (trm', rt)
-       _ | rT' == typeType && sT' == typeType -> return (trm', typeType)
-       _ -> checkError (text "records or record types expected in" <+> ppTerm Unqualified 0 trm)
+       _ | rT' == typeType && sT' == typeType -> do
+         return (trm', typeType)
+       _ -> checkError ("records or record types expected in" <+> ppTerm Unqualified 0 trm)
 
-   Sort _     -> 
+   Sort _     ->
      termWith trm $ return typeType
 
    Prod bt x a b -> do
@@ -231,7 +253,7 @@ inferLType gr g trm = case trm of
      return (Prod bt x a' b', typeType)
 
    Table p t  -> do
-     p' <- justCheck g p typeType --- check p partype! 
+     p' <- justCheck g p typeType --- check p partype!
      t' <- justCheck g t typeType
      return $ (Table p' t', typeType)
 
@@ -243,16 +265,17 @@ inferLType gr g trm = case trm of
    EPattType ty -> do
      ty' <- justCheck g ty typeType
      return (EPattType ty',typeType)
-   EPatt p -> do
+   EPatt _ _ p -> do
      ty <- inferPatt p
-     return (trm, EPattType ty)
+     let (minp,maxp,p') = measurePatt gr p
+     return (EPatt minp maxp p', EPattType ty)
 
    ELin c trm -> do
      (trm',ty) <- inferLType gr g trm
      ty' <- lockRecType c ty ---- lookup c; remove lock AR 20/6/2009
-     return $ (ELin c trm', ty') 
+     return $ (ELin c trm', ty')
 
-   _ -> checkError (text "cannot infer lintype of" <+> ppTerm Unqualified 0 trm)
+   _ -> checkError ("cannot infer lintype of" <+> ppTerm Unqualified 0 trm)
 
  where
    isPredef m = elem m [cPredef,cPredefAbs]
@@ -269,7 +292,7 @@ inferLType gr g trm = case trm of
    inferCase mty (patt,term) = do
      arg  <- maybe (inferPatt patt) return mty
      cont <- pattContext gr g arg patt
-     (_,val) <- inferLType gr (reverse cont ++ g) term
+     (term',val) <- inferLType gr (reverse cont ++ g) term
      return (arg,val)
    isConstPatt p = case p of
      PC _ ps -> True --- all isConstPatt ps
@@ -281,9 +304,9 @@ inferLType gr g trm = case trm of
      PFloat _ -> True
      PChar -> True
      PChars _ -> True
-     PSeq p q -> isConstPatt p && isConstPatt q
+     PSeq _ _ p _ _ q -> isConstPatt p && isConstPatt q
      PAlt p q -> isConstPatt p && isConstPatt q
-     PRep p -> isConstPatt p
+     PRep _ _ p -> isConstPatt p
      PNeg p -> isConstPatt p
      PAs _ p -> isConstPatt p
      _ -> False
@@ -293,12 +316,43 @@ inferLType gr g trm = case trm of
      PAs _ p  -> inferPatt p
      PNeg p   -> inferPatt p
      PAlt p q -> checks [inferPatt p, inferPatt q]
-     PSeq _ _ -> return $ typeStr
-     PRep _   -> return $ typeStr
+     PSeq _ _ _ _ _ _ -> return $ typeStr
+     PRep _ _ _ -> return $ typeStr
      PChar    -> return $ typeStr
      PChars _ -> return $ typeStr
      _ -> inferLType gr g (patt2term p) >>= return . snd
 
+measurePatt gr p =
+  case p of
+    PM q       -> case lookupResDef gr q of
+                    Ok t    -> case t of
+                                 EPatt minp maxp _ -> (minp,maxp,p)
+                                 _ -> error "Expected pattern macro"
+                    Bad msg -> error msg
+    PR ass     -> let p' = PR (map (\(lbl,p) -> let (_,_,p') = measurePatt gr p in (lbl,p')) ass)
+                  in (0,Nothing,p')
+    PString s  -> let len=length s
+                  in (len,Just len,p)
+    PT t p     -> let (min,max,p') = measurePatt gr p
+                  in (min,max,PT t p')
+    PAs x p    -> let (min,max,p') = measurePatt gr p
+                  in (min,max,PAs x p')
+    PImplArg p -> let (min,max,p') = measurePatt gr p
+                  in (min,max,PImplArg p')
+    PNeg p     -> let (_,_,p') = measurePatt gr p
+                  in (0,Nothing,PNeg p')
+    PAlt p1 p2 -> let (min1,max1,p1') = measurePatt gr p1
+                      (min2,max2,p2') = measurePatt gr p2
+                  in (min min1 min2,liftM2 max max1 max2,PAlt p1' p2')
+    PSeq _ _ p1 _ _ p2
+               -> let (min1,max1,p1') = measurePatt gr p1
+                      (min2,max2,p2') = measurePatt gr p2
+                  in (min1+min2,liftM2 (+) max1 max2,PSeq min1 max1 p1' min2 max2 p2')
+    PRep _ _ p -> let (minp,maxp,p') = measurePatt gr p
+                  in (0,Nothing,PRep minp maxp p')
+    PChar      -> (1,Just 1,p)
+    PChars _   -> (1,Just 1,p)
+    _          -> (0,Nothing,p)
 
 -- type inference: Nothing, type checking: Just t
 -- the latter permits matching with value type
@@ -310,15 +364,28 @@ getOverload gr g mt ot = case appForm ot of
          v <- matchOverload f typs ttys
          return $ Just v
        _ -> return Nothing
+     (AdHocOverload cs@(f:_), ts) -> do     --- the function name f is only used in error messages
+       let typs = concatMap collectOverloads cs
+       ttys <- mapM (inferLType gr g) ts
+       v <- matchOverload f typs ttys
+       return $ Just v
      _ -> return Nothing
+
  where
+   collectOverloads tr@(Q c) = case lookupOverload gr c of
+     Ok typs -> typs
+     _ -> case lookupResType gr c of
+       Ok ty -> let (args,val) = typeFormCnc ty in [(map (\(b,x,t) -> t) args,(val,tr))]
+       _ -> []
+   collectOverloads _ = [] --- constructors QC
+
    matchOverload f typs ttys = do
      let (tts,tys) = unzip ttys
      let vfs = lookupOverloadInstance tys typs
      let matches = [vf | vf@((_,v,_),_) <- vfs, matchVal mt v]
      let showTypes ty = hsep (map ppType ty)
 
-    
+
      let (stys,styps) = (showTypes tys, [showTypes ty | (ty,_) <- typs])
 
      -- to avoid strange error msg e.g. in case of unmatch record extension, show whole types if needed AR 28/1/2013
@@ -329,50 +396,57 @@ getOverload gr g mt ot = case appForm ot of
      case ([vf | (vf,True) <- matches],[vf | (vf,False) <- matches]) of
        ([(_,val,fun)],_) -> return (mkApp fun tts, val)
        ([],[(pre,val,fun)]) -> do
-         checkWarn $  text "ignoring lock fields in resolving" <+> ppTerm Unqualified 0 ot $$
-                      text "for" $$
+         checkWarn $  "ignoring lock fields in resolving" <+> ppTerm Unqualified 0 ot $$
+                      "for" $$
                       nest 2 (showTypes tys) $$
-                      text "using" $$
+                      "using" $$
                       nest 2 (showTypes pre)
          return (mkApp fun tts, val)
        ([],[]) -> do
-         checkError $ text "no overload instance of" <+> ppTerm Unqualified 0 f $$
-                      text "for" $$
+         checkError $ "no overload instance of" <+> ppTerm Qualified 0 f $$
+                      maybe empty (\x -> "with value type" <+> ppType x) mt $$
+                      "for argument list" $$
                       nest 2 stysError $$
-                      text "among" $$
-                      nest 2 (vcat stypsError) $$
-                      maybe empty (\x -> text "with value type" <+> ppType x) mt
+                      "among alternatives" $$
+                      nest 2 (vcat stypsError)
+
 
        (vfs1,vfs2) -> case (noProds vfs1,noProds vfs2) of
          ([(val,fun)],_) -> do
            return (mkApp fun tts, val)
          ([],[(val,fun)]) -> do
-           checkWarn (text "ignoring lock fields in resolving" <+> ppTerm Unqualified 0 ot) 
+           checkWarn ("ignoring lock fields in resolving" <+> ppTerm Unqualified 0 ot)
            return (mkApp fun tts, val)
 
 ----- unsafely exclude irritating warning AR 24/5/2008
------           checkWarn $ "overloading of" +++ prt f +++ 
+-----           checkWarn $ "overloading of" +++ prt f +++
 -----             "resolved by excluding partial applications:" ++++
 -----             unlines [prtType env ty | (ty,_) <- vfs', not (noProd ty)]
 
-
-         _ -> checkError $ text "ambiguous overloading of" <+> ppTerm Unqualified 0 f <+>
-                           text "for" <+> hsep (map ppType tys) $$ 
-                           text "with alternatives" $$
-                           nest 2 (vcat [ppType ty | (_,ty,_) <- if null vfs1 then vfs2 else vfs2])
+--- now forgiving ambiguity with a warning AR 1/2/2014
+--  This gives ad hoc overloading the same behaviour as the choice of the first match in renaming did before.
+--  But it also gives a chance to ambiguous overloadings that were banned before.
+         (nps1,nps2) -> do
+              checkWarn $  "ambiguous overloading of" <+> ppTerm Unqualified 0 f <+>
+                  ----     "with argument types" <+> hsep (map (ppTerm Qualified 0) tys) $$
+                           "resolved by selecting the first of the alternatives" $$
+                           nest 2 (vcat [ppTerm Qualified 0 fun | (_,ty,fun) <- vfs1 ++ if null vfs1 then vfs2 else []])
+              case [(mkApp fun tts,val) | (val,fun) <- nps1 ++ nps2] of
+                 [] -> checkError $ "no alternatives left when resolving" <+> ppTerm Unqualified 0 f
+                 h:_ -> return h
 
    matchVal mt v = elem mt [Nothing,Just v,Just (unlocked v)]
 
    unlocked v = case v of
-     RecType fs -> RecType $ filter (not . isLockLabel . fst) fs
+     RecType fs -> RecType $ filter (not . isLockLabel . fst) (sortRec fs)
      _ -> v
    ---- TODO: accept subtypes
    ---- TODO: use a trie
-   lookupOverloadInstance tys typs = 
-     [((pre,mkFunType rest val, t),isExact) | 
+   lookupOverloadInstance tys typs =
+     [((pre,mkFunType rest val, t),isExact) |
        let lt = length tys,
        (ty,(val,t)) <- typs, length ty >= lt,
-       let (pre,rest) = splitAt lt ty, 
+       let (pre,rest) = splitAt lt ty,
        let isExact = pre == tys,
        isExact || map unlocked pre == map unlocked tys
      ]
@@ -385,20 +459,21 @@ getOverload gr g mt ot = case appForm ot of
 
 checkLType :: SourceGrammar -> Context -> Term -> Type -> Check (Term, Type)
 checkLType gr g trm typ0 = do
-
   typ <- computeLType gr g typ0
 
   case trm of
 
     Abs bt x c -> do
       case typ of
-        Prod bt' z a b -> do 
+        Prod bt' z a b -> do
           (c',b') <- if isWildIdent z
                        then checkLType gr ((bt,x,a):g) c b
-                       else do b' <- checkIn (text "abs") $ substituteLType [(bt',z,Vr x)] b
+                       else do b' <- checkIn (pp "abs") $ substituteLType [(bt',z,Vr x)] b
                                checkLType gr ((bt,x,a):g) c b'
-          return $ (Abs bt x c', Prod bt' x a b')
-        _ -> checkError $ text "function type expected instead of" <+> ppType typ
+          return $ (Abs bt x c', Prod bt' z a b')
+        _ -> checkError $ "function type expected instead of" <+> ppType typ $$
+                          "\n   ** Double-check that the type signature of the operation" $$
+                          "matches the number of arguments given to it.\n"
 
     App f a -> do
        over <- getOverload gr g (Just typ) trm
@@ -408,6 +483,12 @@ checkLType gr g trm typ0 = do
           (trm',ty') <- inferLType gr g trm
           termWith trm' $ checkEqLType gr g typ ty' trm'
 
+    AdHocOverload ts -> do
+      over <- getOverload gr g Nothing trm
+      case over of
+        Just trty -> return trty
+        _ -> checkError ("unresolved overloading of constants" <+> ppTerm Qualified 0 trm)
+
     Q _ -> do
        over <- getOverload gr g (Just typ) trm
        case over of
@@ -417,21 +498,21 @@ checkLType gr g trm typ0 = do
           termWith trm' $ checkEqLType gr g typ ty' trm'
 
     T _ [] ->
-      checkError (text "found empty table in type" <+> ppTerm Unqualified 0 typ)
-    T _ cs -> case typ of 
-      Table arg val -> do 
+      checkError ("found empty table in type" <+> ppTerm Unqualified 0 typ)
+    T _ cs -> case typ of
+      Table arg val -> do
         case allParamValues gr arg of
           Ok vs -> do
             let ps0 = map fst cs
             ps <- testOvershadow ps0 vs
-            if null ps 
-              then return () 
-              else checkWarn (text "patterns never reached:" $$
+            if null ps
+              then return ()
+              else checkWarn ("patterns never reached:" $$
                               nest 2 (vcat (map (ppPatt Unqualified 0) ps)))
           _ -> return () -- happens with variable types
         cs' <- mapM (checkCase arg val) cs
         return (T (TTyped arg) cs', typ)
-      _ -> checkError $ text "table type expected for table instead of" $$ nest 2 (ppType typ)
+      _ -> checkError $ "table type expected for table instead of" $$ nest 2 (ppType typ)
     V arg0 vs ->
       case typ of
         Table arg1 val ->
@@ -439,51 +520,54 @@ checkLType gr g trm typ0 = do
               vs1 <- allParamValues gr arg1
               if length vs1 == length vs
                  then return ()
-                 else checkError $ text "wrong number of values in table" <+> ppTerm Unqualified 0 trm
+                 else checkError $ "wrong number of values in table" <+> ppTerm Unqualified 0 trm
               vs' <- map fst `fmap` sequence [checkLType gr g v val|v<-vs]
               return (V arg' vs',typ)
 
     R r -> case typ of --- why needed? because inference may be too difficult
        RecType rr -> do
-         let (ls,_) = unzip rr        -- labels of expected type
+       --let (ls,_) = unzip rr        -- labels of expected type
          fsts <- mapM (checkM r) rr   -- check that they are found in the record
          return $ (R fsts, typ)       -- normalize record
 
-       _ -> checkError (text "record type expected in type checking instead of" $$ nest 2 (ppTerm Unqualified 0 typ))
+       _ -> checkError ("record type expected in type checking instead of" $$ nest 2 (ppTerm Unqualified 0 typ))
 
     ExtR r s -> case typ of
        _ | typ == typeType -> do
          trm' <- computeLType gr g trm
          case trm' of
-           RecType _ -> termWith trm $ return typeType
-           ExtR (Vr _) (RecType _) -> termWith trm $ return typeType 
+           RecType _ -> termWith trm' $ return typeType
+           ExtR (Vr _) (RecType _) -> termWith trm' $ return typeType
                                       -- ext t = t ** ...
-           _ -> checkError (text "invalid record type extension" <+> nest 2 (ppTerm Unqualified 0 trm))
+           _ -> checkError ("invalid record type extension" <+> nest 2 (ppTerm Unqualified 0 trm))
 
        RecType rr -> do
-         (r',ty,s') <- checks [
-            do (r',ty) <- inferLType gr g r
-               return (r',ty,s)
-           ,
-            do (s',ty) <- inferLType gr g s
-               return (s',ty,r)
-           ]
 
-         case ty of
-           RecType rr1 -> do
-                let (rr0,rr2) = recParts rr rr1
-                r2 <- justCheck g r' rr0
-                s2 <- justCheck g s' rr2
-                return $ (ExtR r2 s2, typ) 
-           _ -> checkError (text "record type expected in extension of" <+> ppTerm Unqualified 0 r $$
-                            text "but found" <+> ppTerm Unqualified 0 ty)
+         ll2 <- case s of
+           R ss -> return $ map fst ss
+           _ -> do
+             (s',typ2) <- inferLType gr g s
+             case typ2 of
+               RecType ss -> return $ map fst ss
+               _ ->  checkError ("cannot get labels from" $$ nest 2 (ppTerm Unqualified 0 typ2))
+         let ll1 = [l | (l,_) <- rr, notElem l ll2]
+
+---         over <- getOverload gr g Nothing r --- this would solve #66 but fail ParadigmsAra. AR 6/7/2020
+---         let r1 = maybe r fst over
+         let r1 = r ---
+
+         (r',_) <- checkLType gr g r1 (RecType [field | field@(l,_) <- rr, elem l ll1])
+         (s',_) <- checkLType gr g s  (RecType [field | field@(l,_) <- rr, elem l ll2])
+
+         let rec = R ([(l,(Nothing,P r' l)) | l <- ll1] ++ [(l,(Nothing,P s' l)) | l <- ll2])
+         return (rec, typ)
 
        ExtR ty ex -> do
          r' <- justCheck g r ty
          s' <- justCheck g s ex
          return $ (ExtR r' s', typ) --- is this all? it assumes the same division in trm and typ
 
-       _ -> checkError (text "record extension not meaningful for" <+> ppTerm Unqualified 0 typ)
+       _ -> checkError ("record extension not meaningful for" <+> ppTerm Unqualified 0 typ)
 
     FV vs -> do
       ttys <- mapM (flip (checkLType gr g) typ) vs
@@ -498,7 +582,7 @@ checkLType gr g trm typ0 = do
           (arg',val) <- checkLType gr g arg p
           checkEqLType gr g typ t trm
           return (S tab' arg', t)
-        _ -> checkError (text "table type expected for applied table instead of" <+> ppType ty')
+        _ -> checkError ("table type expected for applied table instead of" <+> ppType ty')
      , do
       (arg',ty) <- inferLType gr g arg
       ty'       <- computeLType gr g ty
@@ -507,7 +591,8 @@ checkLType gr g trm typ0 = do
       ]
     Let (x,(mty,def)) body -> case mty of
       Just ty -> do
-        (def',ty') <- checkLType gr g def ty
+        (ty0,_)    <- checkLType gr g ty typeType
+        (def',ty') <- checkLType gr g def ty0
         body' <- justCheck ((Explicit,x,ty'):g) body typ
         return (Let (x,(Just ty',def')) body', typ)
       _ -> do
@@ -523,10 +608,10 @@ checkLType gr g trm typ0 = do
       termWith trm' $ checkEqLType gr g typ ty' trm'
  where
    justCheck g ty te = checkLType gr g ty te >>= return . fst
-
-   recParts rr t = (RecType rr1,RecType rr2) where 
-     (rr1,rr2) = partition (flip elem (map fst t) . fst) rr 
-
+{-
+   recParts rr t = (RecType rr1,RecType rr2) where
+     (rr1,rr2) = partition (flip elem (map fst t) . fst) rr
+-}
    checkM rms (l,ty) = case lookup l rms of
      Just (Just ty0,t) -> do
        checkEqLType gr g ty ty0 t
@@ -535,17 +620,18 @@ checkLType gr g trm typ0 = do
      Just (_,t) -> do
        (t',ty') <- checkLType gr g t ty
        return (l,(Just ty',t'))
-     _ -> checkError $ 
-            if isLockLabel l 
+     _ -> checkError $
+            if isLockLabel l
               then let cat = drop 5 (showIdent (label2ident l))
-                   in ppTerm Unqualified 0 (R rms) <+> text "is not in the lincat of" <+> text cat <> 
-                      text "; try wrapping it with lin" <+> text cat
-              else text "cannot find value for label" <+> ppLabel l <+> text "in" <+> ppTerm Unqualified 0 (R rms)
+                   in ppTerm Unqualified 0 (R rms) <+> "is not in the lincat of" <+> cat <>
+                      "; try wrapping it with lin" <+> cat
+              else "cannot find value for label" <+> l <+> "in" <+> ppTerm Unqualified 0 (R rms)
 
    checkCase arg val (p,t) = do
      cont <- pattContext gr g arg p
      t' <- justCheck (reverse cont ++ g) t val
-     return (p,t')
+     let (_,_,p') = measurePatt gr p
+     return (p',t')
 
 pattContext :: SourceGrammar -> Context -> Type -> Patt -> Check Context
 pattContext env g typ p = case p of
@@ -553,7 +639,7 @@ pattContext env g typ p = case p of
   PP (q,c) ps | q /= cPredef -> do ---- why this /=? AR 6/1/2006
     t <- lookupResType env (q,c)
     let (cont,v) = typeFormCnc t
-    checkCond (text "wrong number of arguments for constructor in" <+> ppPatt Unqualified 0 p) 
+    checkCond ("wrong number of arguments for constructor in" <+> ppPatt Unqualified 0 p)
               (length cont == length ps)
     checkEqLType env g typ v (patt2term p)
     mapM (\((_,_,ty),p) -> pattContext env g ty p) (zip cont ps) >>= return . concat
@@ -564,7 +650,7 @@ pattContext env g typ p = case p of
         let pts = [(ty,tr) | (l,tr) <- r, Just ty <- [lookup l t]]
         ----- checkWarn $ prt p ++++ show pts ----- debug
         mapM (uncurry (pattContext env g)) pts >>= return . concat
-      _ -> checkError (text "record type expected for pattern instead of" <+> ppTerm Unqualified 0 typ')
+      _ -> checkError ("record type expected for pattern instead of" <+> ppTerm Unqualified 0 typ')
   PT t p' -> do
     checkEqLType env g typ t (patt2term p')
     pattContext env g typ p'
@@ -577,24 +663,24 @@ pattContext env g typ p = case p of
     g1 <- pattContext env g typ p'
     g2 <- pattContext env g typ q
     let pts = nub ([x | pt@(_,x,_) <- g1, notElem pt g2] ++ [x | pt@(_,x,_) <- g2, notElem pt g1])
-    checkCond 
-      (text "incompatible bindings of" <+>
-       fsep (map ppIdent pts) <+> 
-       text "in pattern alterantives" <+> ppPatt Unqualified 0 p) (null pts) 
+    checkCond
+      ("incompatible bindings of" <+>
+       fsep pts <+>
+       "in pattern alterantives" <+> ppPatt Unqualified 0 p) (null pts)
     return g1 -- must be g1 == g2
-  PSeq p q -> do
+  PSeq _ _ p _ _ q -> do
     g1 <- pattContext env g typ p
     g2 <- pattContext env g typ q
     return $ g1 ++ g2
-  PRep p' -> noBind typeStr p'
+  PRep _ _ p' -> noBind typeStr p'
   PNeg p' -> noBind typ p'
 
   _ -> return [] ---- check types!
- where 
+ where
    noBind typ p' = do
     co <- pattContext env g typ p'
     if not (null co)
-      then checkWarn (text "no variable bound inside pattern" <+> ppPatt Unqualified 0 p) 
+      then checkWarn ("no variable bound inside pattern" <+> ppPatt Unqualified 0 p)
            >> return []
       else return []
 
@@ -603,9 +689,31 @@ checkEqLType gr g t u trm = do
   (b,t',u',s) <- checkIfEqLType gr g t u trm
   case b of
     True  -> return t'
-    False -> checkError $ text s <+> text "type of" <+> ppTerm Unqualified 0 trm $$
-                          text "expected:" <+> ppType t $$
-                          text "inferred:" <+> ppType u
+    False ->
+      let inferredType = ppTerm Qualified 0 u
+          expectedType = ppTerm Qualified 0 t
+          term = ppTerm Unqualified 0 trm
+          funName = pp . head . words .render $ term
+          helpfulMsg =
+            case (arrows inferredType, arrows expectedType) of
+              (0,0) -> pp "" -- None of the types is a function
+              _ -> "\n   **" <+>
+                  if expectedType `isLessApplied` inferredType
+                    then "Maybe you gave too few arguments to" <+> funName
+                    else pp "Double-check that type signature and number of arguments match."
+      in checkError $ s <+> "type of" <+> term $$
+                            "expected:" <+> expectedType $$ -- ppqType t u $$
+                            "inferred:" <+> inferredType $$ -- ppqType u t
+                            helpfulMsg
+  where
+    -- count the number of arrows in the prettyprinted term
+    arrows :: Doc -> Int
+    arrows = length . filter (=="->") . words . render
+
+    -- If prettyprinted type t has fewer arrows then prettyprinted type u,
+    -- then t is "less applied", and we can print out more helpful error msg.
+    isLessApplied :: Doc -> Doc -> Bool
+    isLessApplied t u = arrows t < arrows u
 
 checkIfEqLType :: SourceGrammar -> Context -> Type -> Type -> Term -> Check (Bool,Type,Type,String)
 checkIfEqLType gr g t u trm = do
@@ -617,60 +725,62 @@ checkIfEqLType gr g t u trm = do
     --- better: use a flag to forgive? (AR 31/1/2006)
     _ -> case missingLock [] t' u' of
       Ok lo -> do
-        checkWarn $ text "missing lock field" <+> fsep (map ppLabel lo)
+        checkWarn $ "missing lock field" <+> fsep lo
         return (True,t',u',[])
       Bad s -> return (False,t',u',s)
 
   where
 
-   -- t is a subtype of u 
+   -- check that u is a subtype of t
    --- quick hack version of TC.eqVal
-   alpha g t u = case (t,u) of  
+   alpha g t u = case (t,u) of
 
      -- error (the empty type!) is subtype of any other type
      (_,u) | u == typeError -> True
 
      -- contravariance
-     (Prod _ x a b, Prod _ y c d) -> alpha g c a && alpha ((x,y):g) b d 
-                              
+     (Prod _ x a b, Prod _ y c d) -> alpha g c a && alpha ((x,y):g) b d
+
      -- record subtyping
-     (RecType rs, RecType ts) -> all (\ (l,a) -> 
-                                   any (\ (k,b) -> alpha g a b && l == k) ts) rs
+     (RecType rs, RecType ts) -> all (\ (l,a) ->
+                                   any (\ (k,b) -> l == k && alpha g a b) ts) rs
      (ExtR r s, ExtR r' s') -> alpha g r r' && alpha g s s'
      (ExtR r s, t) -> alpha g r t || alpha g s t
 
      -- the following say that Ints n is a subset of Int and of Ints m >= n
-     (t,u) | Just m <- isTypeInts t, Just n <- isTypeInts t -> m >= n
+     -- But why does it also allow Int as a subtype of Ints m? /TH 2014-04-04
+     (t,u) | Just m <- isTypeInts t, Just n <- isTypeInts u -> m >= n
            | Just _ <- isTypeInts t, u == typeInt           -> True ---- check size!
            | t == typeInt,           Just _ <- isTypeInts u -> True ---- why this ???? AR 11/12/2005
 
      ---- this should be made in Rename
-     (Q  (m,a), Q  (n,b)) | a == b -> elem m (allExtendsPlus gr n) 
+     (Q  (m,a), Q  (n,b)) | a == b -> elem m (allExtendsPlus gr n)
                                    || elem n (allExtendsPlus gr m)
                                    || m == n --- for Predef
-     (QC (m,a), QC (n,b)) | a == b -> elem m (allExtendsPlus gr n) 
+     (QC (m,a), QC (n,b)) | a == b -> elem m (allExtendsPlus gr n)
                                    || elem n (allExtendsPlus gr m)
-     (QC (m,a), Q  (n,b)) | a == b -> elem m (allExtendsPlus gr n) 
+     (QC (m,a), Q  (n,b)) | a == b -> elem m (allExtendsPlus gr n)
                                    || elem n (allExtendsPlus gr m)
-     (Q  (m,a), QC (n,b)) | a == b -> elem m (allExtendsPlus gr n) 
+     (Q  (m,a), QC (n,b)) | a == b -> elem m (allExtendsPlus gr n)
                                    || elem n (allExtendsPlus gr m)
 
-     (Table a b,  Table c d)  -> alpha g a c && alpha g b d
+     -- contravariance
+     (Table a b,  Table c d)  -> alpha g c a && alpha g b d
      (Vr x,       Vr y)       -> x == y || elem (x,y) g || elem (y,x) g
-     _                        -> t == u 
+     _                        -> t == u
      --- the following should be one-way coercions only. AR 4/1/2001
                                   || elem t sTypes && elem u sTypes
-                                  || (t == typeType && u == typePType) 
-                                  || (u == typeType && t == typePType) 
+                                  || (t == typeType && u == typePType)
+                                  || (u == typeType && t == typePType)
 
-   missingLock g t u = case (t,u) of  
-     (RecType rs, RecType ts) -> 
-       let 
-         ls = [l | (l,a) <- rs, 
+   missingLock g t u = case (t,u) of
+     (RecType rs, RecType ts) ->
+       let
+         ls = [l | (l,a) <- rs,
                    not (any (\ (k,b) -> alpha g a b && l == k) ts)]
          (locks,others) = partition isLockLabel ls
        in case others of
-         _:_ -> Bad $ render (text "missing record fields:" <+> fsep (punctuate comma (map ppLabel others)))
+         _:_ -> Bad $ render ("missing record fields:" <+> fsep (punctuate ',' (others)))
          _ -> return locks
      -- contravariance
      (Prod _ x a b, Prod _ y c d) -> do
@@ -696,7 +806,7 @@ termWith t ct = do
   return (t,ty)
 
 -- | compositional check\/infer of binary operations
-check2 :: (Term -> Check Term) -> (Term -> Term -> Term) -> 
+check2 :: (Term -> Check Term) -> (Term -> Term -> Term) ->
           Term -> Term -> Type -> Check (Term,Type)
 check2 chk con a b t = do
   a' <- chk a
@@ -708,14 +818,18 @@ ppType :: Type -> Doc
 ppType ty =
   case ty of
     RecType fs   -> case filter isLockLabel $ map fst fs of
-                      [lock] -> text (drop 5 (showIdent (label2ident lock)))
+                      [lock] -> pp (drop 5 (showIdent (label2ident lock)))
                       _      -> ppTerm Unqualified 0 ty
-    Prod _ x a b -> ppType a <+> text "->" <+> ppType b
+    Prod _ x a b -> ppType a <+> "->" <+> ppType b
     _            -> ppTerm Unqualified 0 ty
-
+{-
+ppqType :: Type -> Type -> Doc
+ppqType t u = case (ppType t, ppType u) of
+  (pt,pu) | render pt == render pu -> ppTerm Qualified 0 t
+  (pt,_) -> pt
+-}
 checkLookup :: Ident -> Context -> Check Type
 checkLookup x g =
   case [ty | (b,y,ty) <- g, x == y] of
-    []     -> checkError (text "unknown variable" <+> ppIdent x)
+    []     -> checkError ("unknown variable" <+> x)
     (ty:_) -> return ty
--}

src/compiler/GF/Compile/TypeCheck/ConcreteNew.hs

@@ -10,8 +10,7 @@ import GF.Grammar hiding (Env, VGen, VApp, VRecType)
 import GF.Grammar.Lookup
 import GF.Grammar.Predef
 import GF.Grammar.Lockfield
-import GF.Compile.Compute.ConcreteNew
-import GF.Compile.Compute.Predef(predef,predefName)
+import GF.Compile.Compute.Concrete
 import GF.Infra.CheckM
 import GF.Data.Operations
 import Control.Applicative(Applicative(..))
@@ -22,20 +21,20 @@ import qualified Data.IntMap as IntMap
 import Data.Maybe(fromMaybe,isNothing)
 import qualified Control.Monad.Fail as Fail
 
-checkLType :: GlobalEnv -> Term -> Type -> Check (Term, Type)
-checkLType ge t ty = runTcM $ do
+checkLType :: Grammar -> Term -> Type -> Check (Term, Type)
+checkLType ge t ty = error "TODO: checkLType" {- runTcM $ do
   vty <- liftErr (eval ge [] ty)
   (t,_) <- tcRho ge [] t (Just vty)
   t <- zonkTerm t
-  return (t,ty)
+  return (t,ty) -}
 
-inferLType :: GlobalEnv -> Term -> Check (Term, Type)
-inferLType ge t = runTcM $ do
+inferLType :: Grammar -> Term -> Check (Term, Type)
+inferLType ge t = error "TODO: inferLType" {- runTcM $ do
   (t,ty) <- inferSigma ge [] t
   t  <- zonkTerm t
   ty <- zonkTerm =<< tc_value2term (geLoc ge) [] ty
-  return (t,ty)
-
+  return (t,ty) -}
+{-
 inferSigma :: GlobalEnv -> Scope -> Term -> TcM (Term,Sigma)
 inferSigma ge scope t = do                                      -- GEN1
   (t,ty) <- tcRho ge scope t Nothing
@@ -133,7 +132,7 @@ tcRho ge scope t@(RecType rs) (Just ty) = do
                          [] -> unifyVar ge scope i env vs vtypePType
                          _  -> return ()
     ty              -> do ty <- zonkTerm =<< tc_value2term (geLoc ge) (scopeVars scope) ty
-                          tcError ("The record type" <+> ppTerm Unqualified 0 t $$ 
+                          tcError ("The record type" <+> ppTerm Unqualified 0 t $$
                                    "cannot be of type" <+> ppTerm Unqualified 0 ty)
   (rs,mb_ty) <- tcRecTypeFields ge scope rs (Just ty')
   return (f (RecType rs),ty)
@@ -187,7 +186,7 @@ tcRho ge scope (R rs) (Just ty) = do
   case ty' of
     (VRecType ltys) -> do lttys <- checkRecFields ge scope rs ltys
                           rs <- mapM (\(l,t,ty) -> tc_value2term (geLoc ge) (scopeVars scope) ty >>= \ty -> return (l, (Just ty, t))) lttys
-                          return ((f . R)  rs, 
+                          return ((f . R)  rs,
                                   VRecType [(l, ty) | (l,t,ty) <- lttys]
                                  )
     ty              -> do lttys <- inferRecFields ge scope rs
@@ -277,11 +276,11 @@ tcApp ge scope (App fun arg) =                                  -- APP2
         varg <- liftErr (eval ge (scopeEnv scope) arg)
         return (App fun arg, res_ty varg)
 tcApp ge scope (Q id) =                                          -- VAR (global)
-  mkTcA (lookupOverloadTypes (geGrammar ge) id) `bindTcA` \(t,ty) -> 
+  mkTcA (lookupOverloadTypes (geGrammar ge) id) `bindTcA` \(t,ty) ->
      do ty <- liftErr (eval ge [] ty)
         return (t,ty)
 tcApp ge scope (QC id) =                                         -- VAR (global)
-  mkTcA (lookupOverloadTypes (geGrammar ge) id) `bindTcA` \(t,ty) -> 
+  mkTcA (lookupOverloadTypes (geGrammar ge) id) `bindTcA` \(t,ty) ->
      do ty <- liftErr (eval ge [] ty)
         return (t,ty)
 tcApp ge scope t =
@@ -318,7 +317,7 @@ tcPatt ge scope (PString s) ty0 = do
 tcPatt ge scope PChar ty0 = do
   unify ge scope ty0 vtypeStr
   return scope
-tcPatt ge scope (PSeq p1 p2) ty0 = do
+tcPatt ge scope (PSeq _ _ p1 _ _ p2) ty0 = do
   unify ge scope ty0 vtypeStr
   scope <- tcPatt ge scope p1 vtypeStr
   scope <- tcPatt ge scope p2 vtypeStr
@@ -350,7 +349,7 @@ tcPatt ge scope (PM q) ty0 = do
     Bad err -> tcError (pp err)
 tcPatt ge scope p ty = unimplemented ("tcPatt "++show p)
 
-inferRecFields ge scope rs = 
+inferRecFields ge scope rs =
   mapM (\(l,r) -> tcRecField ge scope l r Nothing) rs
 
 checkRecFields ge scope []          ltys
@@ -368,7 +367,7 @@ checkRecFields ge scope ((l,t):lts) ltys =
   where
     takeIt l1 []  = (Nothing, [])
     takeIt l1 (lty@(l2,ty):ltys)
-      | l1 == l2  = (Just ty,ltys) 
+      | l1 == l2  = (Just ty,ltys)
       | otherwise = let (mb_ty,ltys') = takeIt l1 ltys
                     in (mb_ty,lty:ltys')
 
@@ -390,13 +389,13 @@ tcRecTypeFields ge scope ((l,ty):rs) mb_ty = do
                | s == cPType -> return mb_ty
              VMeta _ _ _     -> return mb_ty
              _               -> do sort <- zonkTerm =<< tc_value2term (geLoc ge) (scopeVars scope) sort
-                                   tcError ("The record type field" <+> l <+> ':' <+> ppTerm Unqualified 0 ty $$ 
+                                   tcError ("The record type field" <+> l <+> ':' <+> ppTerm Unqualified 0 ty $$
                                             "cannot be of type" <+> ppTerm Unqualified 0 sort)
   (rs,mb_ty) <- tcRecTypeFields ge scope rs mb_ty
   return ((l,ty):rs,mb_ty)
 
 -- | Invariant: if the third argument is (Just rho),
--- 	            then rho is in weak-prenex form
+--              then rho is in weak-prenex form
 instSigma :: GlobalEnv -> Scope -> Term -> Sigma -> Maybe Rho -> TcM (Term, Rho)
 instSigma ge scope t ty1 Nothing    = return (t,ty1)           -- INST1
 instSigma ge scope t ty1 (Just ty2) = do                       -- INST2
@@ -444,11 +443,11 @@ subsCheckRho ge scope t (VApp p1 _) (VApp p2 _)                         -- Rule
   | predefName p1 == cInts && predefName p2 == cInt = return t
 subsCheckRho ge scope t (VApp p1 [VInt i]) (VApp p2 [VInt j])           -- Rule INT2
   | predefName p1 == cInts && predefName p2 == cInts =
-      if i <= j 
+      if i <= j
         then return t
         else tcError ("Ints" <+> i <+> "is not a subtype of" <+> "Ints" <+> j)
 subsCheckRho ge scope t ty1@(VRecType rs1) ty2@(VRecType rs2) = do      -- Rule REC
-  let mkAccess scope t = 
+  let mkAccess scope t =
         case t of
           ExtR t1 t2 -> do (scope,mkProj1,mkWrap1) <- mkAccess scope t1
                            (scope,mkProj2,mkWrap2) <- mkAccess scope t2
@@ -557,7 +556,7 @@ unify ge scope v (VMeta i env vs) = unifyVar ge scope i env vs v
 unify ge scope v1 v2 = do
   t1 <- zonkTerm =<< tc_value2term (geLoc ge) (scopeVars scope) v1
   t2 <- zonkTerm =<< tc_value2term (geLoc ge) (scopeVars scope) v2
-  tcError ("Cannot unify terms:" <+> (ppTerm Unqualified 0 t1 $$ 
+  tcError ("Cannot unify terms:" <+> (ppTerm Unqualified 0 t1 $$
                                       ppTerm Unqualified 0 t2))
 
 -- | Invariant: tv1 is a flexible type variable
@@ -568,9 +567,9 @@ unifyVar ge scope i env vs ty2 = do            -- Check whether i is bound
     Bound ty1       -> do v <- liftErr (eval ge env ty1)
                           unify ge scope (vapply (geLoc ge) v vs) ty2
     Unbound scope' _ -> case value2term (geLoc ge) (scopeVars scope') ty2 of
-                          Left i     -> let (v,_) = reverse scope !! i
-                                        in tcError ("Variable" <+> pp v <+> "has escaped")
-                          Right ty2' -> do ms2 <- getMetaVars (geLoc ge) [(scope,ty2)]
+                          -- Left i     -> let (v,_) = reverse scope !! i
+                          --               in tcError ("Variable" <+> pp v <+> "has escaped")
+                                ty2' -> do ms2 <- getMetaVars (geLoc ge) [(scope,ty2)]
                                            if i `elem` ms2
                                              then tcError ("Occurs check for" <+> ppMeta i <+> "in:" $$
                                                            nest 2 (ppTerm Unqualified 0 ty2'))
@@ -609,7 +608,7 @@ quantify ge scope t tvs ty0 = do
   ty <- tc_value2term (geLoc ge) (scopeVars scope) ty0
   let used_bndrs = nub (bndrs ty)  -- Avoid quantified type variables in use
       new_bndrs  = take (length tvs) (allBinders \\ used_bndrs)
-  mapM_ bind (tvs `zip` new_bndrs)   -- 'bind' is just a cunning way 
+  mapM_ bind (tvs `zip` new_bndrs)   -- 'bind' is just a cunning way
   ty <- zonkTerm ty                  -- of doing the substitution
   vty <- liftErr (eval ge [] (foldr (\v ty -> Prod Implicit v typeType ty) ty new_bndrs))
   return (foldr (Abs Implicit) t new_bndrs,vty)
@@ -619,7 +618,7 @@ quantify ge scope t tvs ty0 = do
     bndrs (Prod _ x t1 t2) = [x] ++ bndrs t1  ++ bndrs t2
     bndrs _                = []
 
-allBinders :: [Ident]    -- a,b,..z, a1, b1,... z1, a2, b2,... 
+allBinders :: [Ident]    -- a,b,..z, a1, b1,... z1, a2, b2,...
 allBinders = [ identS [x]          | x <- ['a'..'z'] ] ++
              [ identS (x : show i) | i <- [1 :: Integer ..], x <- ['a'..'z']]
 
@@ -631,8 +630,8 @@ allBinders = [ identS [x]          | x <- ['a'..'z'] ] ++
 type Scope = [(Ident,Value)]
 
 type Sigma = Value
-type Rho   = Value	-- No top-level ForAll
-type Tau   = Value	-- No ForAlls anywhere
+type Rho   = Value -- No top-level ForAll
+type Tau   = Value -- No ForAlls anywhere
 
 data MetaValue
   = Unbound Scope Sigma
@@ -688,12 +687,12 @@ runTcM f = case unTcM f IntMap.empty [] of
              TcFail (msg:msgs) -> do checkWarnings msgs; checkError msg
 
 newMeta :: Scope -> Sigma -> TcM MetaId
-newMeta scope ty = TcM (\ms msgs -> 
+newMeta scope ty = TcM (\ms msgs ->
   let i = IntMap.size ms
   in TcOk i (IntMap.insert i (Unbound scope ty) ms) msgs)
 
 getMeta :: MetaId -> TcM MetaValue
-getMeta i = TcM (\ms msgs -> 
+getMeta i = TcM (\ms msgs ->
   case IntMap.lookup i ms of
     Just mv -> TcOk mv ms msgs
     Nothing -> TcFail (("Unknown metavariable" <+> ppMeta i) : msgs))
@@ -702,7 +701,7 @@ setMeta :: MetaId -> MetaValue -> TcM ()
 setMeta i mv = TcM (\ms msgs -> TcOk () (IntMap.insert i mv ms) msgs)
 
 newVar :: Scope -> Ident
-newVar scope = head [x | i <- [1..], 
+newVar scope = head [x | i <- [1..],
                          let x = identS ('v':show i),
                          isFree scope x]
   where
@@ -721,11 +720,11 @@ getMetaVars loc sc_tys = do
   return (foldr go [] tys)
   where
     -- Get the MetaIds from a term; no duplicates in result
-    go (Vr tv)    acc = acc 
+    go (Vr tv)    acc = acc
     go (App x y)  acc = go x (go y acc)
     go (Meta i)   acc
-	  | i `elem` acc  = acc
-	  | otherwise	  = i : acc
+      | i `elem` acc  = acc
+      | otherwise     = i : acc
     go (Q _)      acc = acc
     go (QC _)     acc = acc
     go (Sort _)   acc = acc
@@ -741,10 +740,10 @@ getFreeVars loc sc_tys = do
   tys <- mapM (\(scope,ty) -> zonkTerm =<< tc_value2term loc (scopeVars scope) ty) sc_tys
   return (foldr (go []) [] tys)
   where
-    go bound (Vr tv)            acc 
-	  | tv `elem` bound             = acc
-	  | tv `elem` acc               = acc
-	  | otherwise                   = tv : acc
+    go bound (Vr tv)            acc
+      | tv `elem` bound             = acc
+      | tv `elem` acc               = acc
+      | otherwise                   = tv : acc
     go bound (App x y)          acc = go bound x (go bound y acc)
     go bound (Meta _)           acc = acc
     go bound (Q _)              acc = acc
@@ -765,13 +764,13 @@ zonkTerm (Meta i) = do
 zonkTerm t = composOp zonkTerm t
 
 tc_value2term loc xs v =
-  case value2term loc xs v of
-    Left i  -> tcError ("Variable #" <+> pp i <+> "has escaped")
-    Right t -> return t
+  return $ value2term loc xs v
+    -- Old value2term error message:
+    -- Left i  -> tcError ("Variable #" <+> pp i <+> "has escaped")
 
 
 
-data TcA x a 
+data TcA x a
   = TcSingle   (MetaStore -> [Message] -> TcResult a)
   | TcMany [x] (MetaStore -> [Message] -> [(a,MetaStore,[Message])])
 
@@ -800,3 +799,4 @@ runTcA g f = TcM (\ms msgs -> case f of
                                                  [(x,ms,msgs)] -> TcOk x ms msgs
                                                  rs            -> unTcM (g xs) ms msgs
                                 TcSingle f                     -> f ms msgs)
+-}

src/compiler/GF/Compile/TypeCheck/Primitives.hs

@@ -8,7 +8,7 @@ typPredefined :: Ident -> Maybe Type
 typPredefined f = case Map.lookup f primitives of
                     Just (ResOper (Just (L _ ty)) _) -> Just ty
                     Just (ResParam _ _)              -> Just typePType
-                    Just (ResValue (L _ ty))         -> Just ty
+                    Just (ResValue (L _ ty) _)       -> Just ty
                     _                                -> Nothing
 
 primitives = Map.fromList
@@ -16,9 +16,9 @@ primitives = Map.fromList
   , (cInt      , ResOper (Just (noLoc typePType)) Nothing)
   , (cFloat    , ResOper (Just (noLoc typePType)) Nothing)
   , (cInts     , fun [typeInt] typePType)
-  , (cPBool    , ResParam (Just (noLoc [(cPTrue,[]),(cPFalse,[])])) (Just [QC (cPredef,cPTrue), QC (cPredef,cPFalse)]))
-  , (cPTrue    , ResValue (noLoc typePBool))
-  , (cPFalse   , ResValue (noLoc typePBool))
+  , (cPBool    , ResParam (Just (noLoc [(cPTrue,[]),(cPFalse,[])])) (Just ([QC (cPredef,cPTrue), QC (cPredef,cPFalse)],2)))
+  , (cPTrue    , ResValue (noLoc typePBool) 0)
+  , (cPFalse   , ResValue (noLoc typePBool) 1)
   , (cError    , fun [typeStr] typeError)  -- non-can. of empty set
   , (cLength   , fun [typeTok] typeInt)
   , (cDrop     , fun [typeInt,typeTok] typeTok)

src/compiler/GF/Compile/TypeCheck/RConcrete.hs

@@ -1,801 +0,0 @@
-{-# LANGUAGE PatternGuards #-}
-module GF.Compile.TypeCheck.RConcrete( checkLType, inferLType, computeLType, ppType ) where
-import Prelude hiding ((<>)) -- GHC 8.4.1 clash with Text.PrettyPrint
-
-import GF.Infra.CheckM
-import GF.Data.Operations
-
-import GF.Grammar
-import GF.Grammar.Lookup
-import GF.Grammar.Predef
-import GF.Grammar.PatternMatch
-import GF.Grammar.Lockfield (isLockLabel, lockRecType, unlockRecord)
-import GF.Compile.TypeCheck.Primitives
-
-import Data.List
-import Control.Monad
-import GF.Text.Pretty
-
-computeLType :: SourceGrammar -> Context -> Type -> Check Type
-computeLType gr g0 t = comp (reverse [(b,x, Vr x) | (b,x,_) <- g0] ++ g0) t
- where
-  comp g ty = case ty of
-    _ | Just _ <- isTypeInts ty -> return ty ---- shouldn't be needed
-      | isPredefConstant ty     -> return ty ---- shouldn't be needed
-
-    Q (m,ident) -> checkIn ("module" <+> m) $ do
-      ty' <- lookupResDef gr (m,ident)
-      if ty' == ty then return ty else comp g ty' --- is this necessary to test?
-
-    AdHocOverload ts -> do
-     over <- getOverload gr g (Just typeType) t
-     case over of
-       Just (tr,_) -> return tr
-       _ -> checkError ("unresolved overloading of constants" <+> ppTerm Qualified 0 t) 
-
-    Vr ident  -> checkLookup ident g -- never needed to compute!
-
-    App f a -> do
-      f' <- comp g f
-      a' <- comp g a
-      case f' of
-        Abs b x t -> comp ((b,x,a'):g) t
-        _ -> return $ App f' a'
-
-    Prod bt x a b -> do
-      a' <- comp g a
-      b' <- comp ((bt,x,Vr x) : g) b
-      return $ Prod bt x a' b'
-
-    Abs bt x b -> do
-      b' <- comp ((bt,x,Vr x):g) b
-      return $ Abs bt x b'
-
-    Let (x,(_,a)) b -> comp ((Explicit,x,a):g) b
-
-    ExtR r s -> do
-      r' <- comp g r
-      s' <- comp g s
-      case (r',s') of
-        (RecType rs, RecType ss) -> plusRecType r' s' >>= comp g
-        _ -> return $ ExtR r' s'
-
-    RecType fs -> do
-      let fs' = sortRec fs
-      liftM RecType $ mapPairsM (comp g) fs'
-
-    ELincat c t -> do
-      t' <- comp g t
-      lockRecType c t' ---- locking to be removed AR 20/6/2009
-
-    _ | ty == typeTok -> return typeStr
-    _ | isPredefConstant ty -> return ty
-
-    _ -> composOp (comp g) ty
-
--- the underlying algorithms
-
-inferLType :: SourceGrammar -> Context -> Term -> Check (Term, Type)
-inferLType gr g trm = case trm of
-
-   Q (m,ident) | isPredef m -> termWith trm $ case typPredefined ident of
-                                                Just ty -> return ty
-                                                Nothing -> checkError ("unknown in Predef:" <+> ident)
-
-   Q ident -> checks [
-     termWith trm $ lookupResType gr ident >>= computeLType gr g
-     ,
-     lookupResDef gr ident >>= inferLType gr g
-     ,
-     checkError ("cannot infer type of constant" <+> ppTerm Unqualified 0 trm)
-     ]
-
-   QC (m,ident) | isPredef m -> termWith trm $ case typPredefined ident of
-                                                 Just ty -> return ty
-                                                 Nothing -> checkError ("unknown in Predef:" <+> ident)
-
-   QC ident -> checks [
-       termWith trm $ lookupResType gr ident >>= computeLType gr g
-       ,
-       lookupResDef gr ident >>= inferLType gr g
-       ,
-       checkError ("cannot infer type of canonical constant" <+> ppTerm Unqualified 0 trm)
-       ]
-
-   Vr ident -> termWith trm $ checkLookup ident g
-
-   Typed e t -> do
-     t' <- computeLType gr g t
-     checkLType gr g e t'
-
-   AdHocOverload ts -> do
-     over <- getOverload gr g Nothing trm
-     case over of
-       Just trty -> return trty
-       _ -> checkError ("unresolved overloading of constants" <+> ppTerm Qualified 0 trm) 
-
-   App f a -> do
-     over <- getOverload gr g Nothing trm
-     case over of
-       Just trty -> return trty
-       _ -> do
-         (f',fty) <- inferLType gr g f
-         fty' <- computeLType gr g fty
-         case fty' of
-           Prod bt z arg val -> do 
-             a' <- justCheck g a arg
-             ty <- if isWildIdent z 
-                      then return val
-                      else substituteLType [(bt,z,a')] val
-             return (App f' a',ty)
-           _ ->
-             let term = ppTerm Unqualified 0 f
-                 funName = pp . head . words .render $ term
-              in checkError ("A function type is expected for" <+> term <+> "instead of type" <+> ppType fty $$
-                             "\n   ** Maybe you gave too many arguments to" <+> funName <+> "\n")
-
-   S f x -> do
-     (f', fty) <- inferLType gr g f
-     case fty of
-       Table arg val -> do
-         x'<- justCheck g x arg
-         return (S f' x', val)
-       _ -> checkError ("table lintype expected for the table in" $$ nest 2 (ppTerm Unqualified 0 trm))
-
-   P t i -> do
-     (t',ty) <- inferLType gr g t   --- ??
-     ty' <- computeLType gr g ty
-     let tr2 = P t' i
-     termWith tr2 $ case ty' of
-       RecType ts -> case lookup i ts of
-                       Nothing -> checkError ("unknown label" <+> i <+> "in" $$ nest 2 (ppTerm Unqualified 0 ty'))
-                       Just x  -> return x
-       _          -> checkError ("record type expected for:" <+> ppTerm Unqualified 0 t $$
-                                 " instead of the inferred:" <+> ppTerm Unqualified 0 ty')
-
-   R r -> do
-     let (ls,fs) = unzip r
-     fsts <- mapM inferM fs
-     let ts = [ty | (Just ty,_) <- fsts]
-     checkCond ("cannot infer type of record" $$ nest 2 (ppTerm Unqualified 0 trm)) (length ts == length fsts)
-     return $ (R (zip ls fsts), RecType (zip ls ts))
-
-   T (TTyped arg) pts -> do
-     (_,val) <- checks $ map (inferCase (Just arg)) pts
-     checkLType gr g trm (Table arg val)
-   T (TComp arg) pts -> do
-     (_,val) <- checks $ map (inferCase (Just arg)) pts
-     checkLType gr g trm (Table arg val)
-   T ti pts -> do  -- tries to guess: good in oper type inference
-     let pts' = [pt | pt@(p,_) <- pts, isConstPatt p]
-     case pts' of 
-       [] -> checkError ("cannot infer table type of" <+> ppTerm Unqualified 0 trm)
-----       PInt k : _ -> return $ Ints $ max [i | PInt i <- pts'] 
-       _  -> do 
-         (arg,val) <- checks $ map (inferCase Nothing) pts'
-         checkLType gr g trm (Table arg val)
-   V arg pts -> do
-     (_,val) <- checks $ map (inferLType gr g) pts
---   return (trm, Table arg val) -- old, caused issue 68
-     checkLType gr g trm (Table arg val)
-
-   K s  -> do
-     if elem ' ' s
-        then do
-          let ss = foldr C Empty (map K (words s))  
-          ----- removed irritating warning AR 24/5/2008
-          ----- checkWarn ("token \"" ++ s ++ 
-          -----              "\" converted to token list" ++ prt ss)
-          return (ss, typeStr)
-        else return (trm, typeStr)
-
-   EInt i -> return (trm, typeInt)
-
-   EFloat i -> return (trm, typeFloat)
-
-   Empty -> return (trm, typeStr)
-
-   C s1 s2 -> 
-     check2 (flip (justCheck g) typeStr) C s1 s2 typeStr
-
-   Glue s1 s2 -> 
-     check2 (flip (justCheck g) typeStr) Glue s1 s2 typeStr ---- typeTok
-
----- hack from Rename.identRenameTerm, to live with files with naming conflicts 18/6/2007
-   Strs (Cn c : ts) | c == cConflict -> do
-     checkWarn ("unresolved constant, could be any of" <+> hcat (map (ppTerm Unqualified 0) ts))
-     inferLType gr g (head ts)
-
-   Strs ts -> do
-     ts' <- mapM (\t -> justCheck g t typeStr) ts 
-     return (Strs ts', typeStrs)
-
-   Alts t aa -> do
-     t'  <- justCheck g t typeStr
-     aa' <- flip mapM aa (\ (c,v) -> do
-        c' <- justCheck g c typeStr 
-        v' <- checks $ map (justCheck g v) [typeStrs, EPattType typeStr]
-        return (c',v'))
-     return (Alts t' aa', typeStr)
-
-   RecType r -> do
-     let (ls,ts) = unzip r
-     ts' <- mapM (flip (justCheck g) typeType) ts 
-     return (RecType (zip ls ts'), typeType)
-
-   ExtR r s -> do
-   
----     over <- getOverload gr g Nothing r
----     let r1 = maybe r fst over
-     let r1 = r --- 
-
-     (r',rT) <- inferLType gr g r1
-     rT'     <- computeLType gr g rT
-     
-     (s',sT) <- inferLType gr g s
-     sT'     <- computeLType gr g sT
-
-     let trm' = ExtR r' s'
-     case (rT', sT') of
-       (RecType rs, RecType ss) -> do
-         let rt = RecType ([field | field@(l,_) <- rs, notElem l (map fst ss)] ++ ss) -- select types of later fields 
-         checkLType gr g trm' rt ---- return (trm', rt)
-       _ | rT' == typeType && sT' == typeType -> do
-         return (trm', typeType)
-       _ -> checkError ("records or record types expected in" <+> ppTerm Unqualified 0 trm)
-
-   Sort _     -> 
-     termWith trm $ return typeType
-
-   Prod bt x a b -> do
-     a' <- justCheck g a typeType
-     b' <- justCheck ((bt,x,a'):g) b typeType
-     return (Prod bt x a' b', typeType)
-
-   Table p t  -> do
-     p' <- justCheck g p typeType --- check p partype! 
-     t' <- justCheck g t typeType
-     return $ (Table p' t', typeType)
-
-   FV vs -> do
-     (_,ty) <- checks $ map (inferLType gr g) vs
----     checkIfComplexVariantType trm ty
-     checkLType gr g trm ty
-
-   EPattType ty -> do
-     ty' <- justCheck g ty typeType
-     return (EPattType ty',typeType)
-   EPatt p -> do
-     ty <- inferPatt p
-     return (trm, EPattType ty)
-
-   ELin c trm -> do
-     (trm',ty) <- inferLType gr g trm
-     ty' <- lockRecType c ty ---- lookup c; remove lock AR 20/6/2009
-     return $ (ELin c trm', ty') 
-
-   _ -> checkError ("cannot infer lintype of" <+> ppTerm Unqualified 0 trm)
-
- where
-   isPredef m = elem m [cPredef,cPredefAbs]
-
-   justCheck g ty te = checkLType gr g ty te >>= return . fst
-
-   -- for record fields, which may be typed
-   inferM (mty, t) = do
-     (t', ty') <- case mty of
-       Just ty -> checkLType gr g t ty
-       _ -> inferLType gr g t
-     return (Just ty',t')
-
-   inferCase mty (patt,term) = do
-     arg  <- maybe (inferPatt patt) return mty
-     cont <- pattContext gr g arg patt
-     (_,val) <- inferLType gr (reverse cont ++ g) term
-     return (arg,val)
-   isConstPatt p = case p of
-     PC _ ps -> True --- all isConstPatt ps
-     PP _ ps -> True --- all isConstPatt ps
-     PR ps -> all (isConstPatt . snd) ps
-     PT _ p -> isConstPatt p
-     PString _ -> True
-     PInt _ -> True
-     PFloat _ -> True
-     PChar -> True
-     PChars _ -> True
-     PSeq p q -> isConstPatt p && isConstPatt q
-     PAlt p q -> isConstPatt p && isConstPatt q
-     PRep p -> isConstPatt p
-     PNeg p -> isConstPatt p
-     PAs _ p -> isConstPatt p
-     _ -> False
-
-   inferPatt p = case p of
-     PP (q,c) ps | q /= cPredef -> liftM valTypeCnc (lookupResType gr (q,c))
-     PAs _ p  -> inferPatt p
-     PNeg p   -> inferPatt p
-     PAlt p q -> checks [inferPatt p, inferPatt q]
-     PSeq _ _ -> return $ typeStr
-     PRep _   -> return $ typeStr
-     PChar    -> return $ typeStr
-     PChars _ -> return $ typeStr
-     _ -> inferLType gr g (patt2term p) >>= return . snd
-
--- type inference: Nothing, type checking: Just t
--- the latter permits matching with value type
-getOverload :: SourceGrammar -> Context -> Maybe Type -> Term -> Check (Maybe (Term,Type))
-getOverload gr g mt ot = case appForm ot of
-     (f@(Q c), ts) -> case lookupOverload gr c of
-       Ok typs -> do
-         ttys <- mapM (inferLType gr g) ts
-         v <- matchOverload f typs ttys
-         return $ Just v
-       _ -> return Nothing
-     (AdHocOverload cs@(f:_), ts) -> do     --- the function name f is only used in error messages
-       let typs = concatMap collectOverloads cs
-       ttys <- mapM (inferLType gr g) ts
-       v <- matchOverload f typs ttys
-       return $ Just v
-     _ -> return Nothing
-     
- where
-   collectOverloads tr@(Q c) = case lookupOverload gr c of 
-     Ok typs -> typs
-     _ -> case lookupResType gr c of
-       Ok ty -> let (args,val) = typeFormCnc ty in [(map (\(b,x,t) -> t) args,(val,tr))]
-       _ -> []
-   collectOverloads _ = [] --- constructors QC
-
-   matchOverload f typs ttys = do
-     let (tts,tys) = unzip ttys
-     let vfs = lookupOverloadInstance tys typs
-     let matches = [vf | vf@((_,v,_),_) <- vfs, matchVal mt v]
-     let showTypes ty = hsep (map ppType ty)
-
-    
-     let (stys,styps) = (showTypes tys, [showTypes ty | (ty,_) <- typs])
-
-     -- to avoid strange error msg e.g. in case of unmatch record extension, show whole types if needed AR 28/1/2013
-     let (stysError,stypsError) = if elem (render stys) (map render styps)
-            then (hsep (map (ppTerm Unqualified 0) tys), [hsep (map (ppTerm Unqualified 0) ty) | (ty,_) <- typs])
-            else (stys,styps)
-
-     case ([vf | (vf,True) <- matches],[vf | (vf,False) <- matches]) of
-       ([(_,val,fun)],_) -> return (mkApp fun tts, val)
-       ([],[(pre,val,fun)]) -> do
-         checkWarn $  "ignoring lock fields in resolving" <+> ppTerm Unqualified 0 ot $$
-                      "for" $$
-                      nest 2 (showTypes tys) $$
-                      "using" $$
-                      nest 2 (showTypes pre)
-         return (mkApp fun tts, val)
-       ([],[]) -> do
-         checkError $ "no overload instance of" <+> ppTerm Qualified 0 f $$
-                      maybe empty (\x -> "with value type" <+> ppType x) mt $$
-                      "for argument list" $$
-                      nest 2 stysError $$
-                      "among alternatives" $$
-                      nest 2 (vcat stypsError) 
-
-
-       (vfs1,vfs2) -> case (noProds vfs1,noProds vfs2) of
-         ([(val,fun)],_) -> do
-           return (mkApp fun tts, val)
-         ([],[(val,fun)]) -> do
-           checkWarn ("ignoring lock fields in resolving" <+> ppTerm Unqualified 0 ot) 
-           return (mkApp fun tts, val)
-
------ unsafely exclude irritating warning AR 24/5/2008
------           checkWarn $ "overloading of" +++ prt f +++ 
------             "resolved by excluding partial applications:" ++++
------             unlines [prtType env ty | (ty,_) <- vfs', not (noProd ty)]
-
---- now forgiving ambiguity with a warning AR 1/2/2014
---  This gives ad hoc overloading the same behaviour as the choice of the first match in renaming did before. 
---  But it also gives a chance to ambiguous overloadings that were banned before. 
-         (nps1,nps2) -> do 
-              checkWarn $  "ambiguous overloading of" <+> ppTerm Unqualified 0 f <+>
-                  ----     "with argument types" <+> hsep (map (ppTerm Qualified 0) tys) $$ 
-                           "resolved by selecting the first of the alternatives" $$
-                           nest 2 (vcat [ppTerm Qualified 0 fun | (_,ty,fun) <- vfs1 ++ if null vfs1 then vfs2 else []])
-              case [(mkApp fun tts,val) | (val,fun) <- nps1 ++ nps2] of
-                 [] -> checkError $ "no alternatives left when resolving" <+> ppTerm Unqualified 0 f
-                 h:_ -> return h
-
-   matchVal mt v = elem mt [Nothing,Just v,Just (unlocked v)]
-
-   unlocked v = case v of
-     RecType fs -> RecType $ filter (not . isLockLabel . fst) (sortRec fs)
-     _ -> v
-   ---- TODO: accept subtypes
-   ---- TODO: use a trie
-   lookupOverloadInstance tys typs = 
-     [((pre,mkFunType rest val, t),isExact) | 
-       let lt = length tys,
-       (ty,(val,t)) <- typs, length ty >= lt,
-       let (pre,rest) = splitAt lt ty, 
-       let isExact = pre == tys,
-       isExact || map unlocked pre == map unlocked tys
-     ]
-
-   noProds vfs = [(v,f) | (_,v,f) <- vfs, noProd v]
-
-   noProd ty = case ty of
-     Prod _ _ _ _ -> False
-     _ -> True
-
-checkLType :: SourceGrammar -> Context -> Term -> Type -> Check (Term, Type)
-checkLType gr g trm typ0 = do
-  typ <- computeLType gr g typ0
-
-  case trm of
-
-    Abs bt x c -> do
-      case typ of
-        Prod bt' z a b -> do 
-          (c',b') <- if isWildIdent z
-                       then checkLType gr ((bt,x,a):g) c b
-                       else do b' <- checkIn (pp "abs") $ substituteLType [(bt',z,Vr x)] b
-                               checkLType gr ((bt,x,a):g) c b'
-          return $ (Abs bt x c', Prod bt' z a b')
-        _ -> checkError $ "function type expected instead of" <+> ppType typ $$
-                          "\n   ** Double-check that the type signature of the operation" $$
-                          "matches the number of arguments given to it.\n"
-
-    App f a -> do
-       over <- getOverload gr g (Just typ) trm
-       case over of
-         Just trty -> return trty
-         _ -> do
-          (trm',ty') <- inferLType gr g trm
-          termWith trm' $ checkEqLType gr g typ ty' trm'
-
-    AdHocOverload ts -> do
-      over <- getOverload gr g Nothing trm
-      case over of
-        Just trty -> return trty
-        _ -> checkError ("unresolved overloading of constants" <+> ppTerm Qualified 0 trm) 
-
-    Q _ -> do
-       over <- getOverload gr g (Just typ) trm
-       case over of
-         Just trty -> return trty
-         _ -> do
-          (trm',ty') <- inferLType gr g trm
-          termWith trm' $ checkEqLType gr g typ ty' trm'
-
-    T _ [] ->
-      checkError ("found empty table in type" <+> ppTerm Unqualified 0 typ)
-    T _ cs -> case typ of 
-      Table arg val -> do 
-        case allParamValues gr arg of
-          Ok vs -> do
-            let ps0 = map fst cs
-            ps <- testOvershadow ps0 vs
-            if null ps 
-              then return () 
-              else checkWarn ("patterns never reached:" $$
-                              nest 2 (vcat (map (ppPatt Unqualified 0) ps)))
-          _ -> return () -- happens with variable types
-        cs' <- mapM (checkCase arg val) cs
-        return (T (TTyped arg) cs', typ)
-      _ -> checkError $ "table type expected for table instead of" $$ nest 2 (ppType typ)
-    V arg0 vs ->
-      case typ of
-        Table arg1 val ->
-           do arg' <- checkEqLType gr g arg0 arg1 trm
-              vs1 <- allParamValues gr arg1
-              if length vs1 == length vs
-                 then return ()
-                 else checkError $ "wrong number of values in table" <+> ppTerm Unqualified 0 trm
-              vs' <- map fst `fmap` sequence [checkLType gr g v val|v<-vs]
-              return (V arg' vs',typ)
-
-    R r -> case typ of --- why needed? because inference may be too difficult
-       RecType rr -> do
-       --let (ls,_) = unzip rr        -- labels of expected type
-         fsts <- mapM (checkM r) rr   -- check that they are found in the record
-         return $ (R fsts, typ)       -- normalize record
-
-       _ -> checkError ("record type expected in type checking instead of" $$ nest 2 (ppTerm Unqualified 0 typ))
-
-    ExtR r s -> case typ of
-       _ | typ == typeType -> do
-         trm' <- computeLType gr g trm
-         case trm' of
-           RecType _ -> termWith trm' $ return typeType
-           ExtR (Vr _) (RecType _) -> termWith trm' $ return typeType 
-                                      -- ext t = t ** ...
-           _ -> checkError ("invalid record type extension" <+> nest 2 (ppTerm Unqualified 0 trm))
-
-       RecType rr -> do
-
-         ll2 <- case s of
-           R ss -> return $ map fst ss
-           _ -> do
-             (s',typ2) <- inferLType gr g s
-             case typ2 of
-               RecType ss -> return $ map fst ss
-               _ ->  checkError ("cannot get labels from" $$ nest 2 (ppTerm Unqualified 0 typ2))
-         let ll1 = [l | (l,_) <- rr, notElem l ll2]
-
----         over <- getOverload gr g Nothing r --- this would solve #66 but fail ParadigmsAra. AR 6/7/2020
----         let r1 = maybe r fst over
-         let r1 = r ---
-         
-         (r',_) <- checkLType gr g r1 (RecType [field | field@(l,_) <- rr, elem l ll1])
-         (s',_) <- checkLType gr g s  (RecType [field | field@(l,_) <- rr, elem l ll2])
-
-         let rec = R ([(l,(Nothing,P r' l)) | l <- ll1] ++ [(l,(Nothing,P s' l)) | l <- ll2])
-         return (rec, typ)
-
-       ExtR ty ex -> do
-         r' <- justCheck g r ty
-         s' <- justCheck g s ex
-         return $ (ExtR r' s', typ) --- is this all? it assumes the same division in trm and typ
-
-       _ -> checkError ("record extension not meaningful for" <+> ppTerm Unqualified 0 typ)
-
-    FV vs -> do
-      ttys <- mapM (flip (checkLType gr g) typ) vs
----      checkIfComplexVariantType trm typ
-      return (FV (map fst ttys), typ) --- typ' ?
-
-    S tab arg -> checks [ do
-      (tab',ty) <- inferLType gr g tab
-      ty'       <- computeLType gr g ty
-      case ty' of
-        Table p t -> do
-          (arg',val) <- checkLType gr g arg p
-          checkEqLType gr g typ t trm
-          return (S tab' arg', t)
-        _ -> checkError ("table type expected for applied table instead of" <+> ppType ty')
-     , do
-      (arg',ty) <- inferLType gr g arg
-      ty'       <- computeLType gr g ty
-      (tab',_)  <- checkLType gr g tab (Table ty' typ)
-      return (S tab' arg', typ)
-      ]
-    Let (x,(mty,def)) body -> case mty of
-      Just ty -> do
-        (ty0,_)    <- checkLType gr g ty typeType
-        (def',ty') <- checkLType gr g def ty0
-        body' <- justCheck ((Explicit,x,ty'):g) body typ
-        return (Let (x,(Just ty',def')) body', typ)
-      _ -> do
-        (def',ty) <- inferLType gr g def  -- tries to infer type of local constant
-        checkLType gr g (Let (x,(Just ty,def')) body) typ
-
-    ELin c tr -> do
-      tr1 <- unlockRecord c tr
-      checkLType gr g tr1 typ
-
-    _ -> do
-      (trm',ty') <- inferLType gr g trm
-      termWith trm' $ checkEqLType gr g typ ty' trm'
- where
-   justCheck g ty te = checkLType gr g ty te >>= return . fst
-{-
-   recParts rr t = (RecType rr1,RecType rr2) where 
-     (rr1,rr2) = partition (flip elem (map fst t) . fst) rr 
--}
-   checkM rms (l,ty) = case lookup l rms of
-     Just (Just ty0,t) -> do
-       checkEqLType gr g ty ty0 t
-       (t',ty') <- checkLType gr g t ty
-       return (l,(Just ty',t'))
-     Just (_,t) -> do
-       (t',ty') <- checkLType gr g t ty
-       return (l,(Just ty',t'))
-     _ -> checkError $ 
-            if isLockLabel l 
-              then let cat = drop 5 (showIdent (label2ident l))
-                   in ppTerm Unqualified 0 (R rms) <+> "is not in the lincat of" <+> cat <> 
-                      "; try wrapping it with lin" <+> cat
-              else "cannot find value for label" <+> l <+> "in" <+> ppTerm Unqualified 0 (R rms)
-
-   checkCase arg val (p,t) = do
-     cont <- pattContext gr g arg p
-     t' <- justCheck (reverse cont ++ g) t val
-     return (p,t')
-
-pattContext :: SourceGrammar -> Context -> Type -> Patt -> Check Context
-pattContext env g typ p = case p of
-  PV x -> return [(Explicit,x,typ)]
-  PP (q,c) ps | q /= cPredef -> do ---- why this /=? AR 6/1/2006
-    t <- lookupResType env (q,c)
-    let (cont,v) = typeFormCnc t
-    checkCond ("wrong number of arguments for constructor in" <+> ppPatt Unqualified 0 p) 
-              (length cont == length ps)
-    checkEqLType env g typ v (patt2term p)
-    mapM (\((_,_,ty),p) -> pattContext env g ty p) (zip cont ps) >>= return . concat
-  PR r -> do
-    typ' <- computeLType env g typ
-    case typ' of
-      RecType t -> do
-        let pts = [(ty,tr) | (l,tr) <- r, Just ty <- [lookup l t]]
-        ----- checkWarn $ prt p ++++ show pts ----- debug
-        mapM (uncurry (pattContext env g)) pts >>= return . concat
-      _ -> checkError ("record type expected for pattern instead of" <+> ppTerm Unqualified 0 typ')
-  PT t p' -> do
-    checkEqLType env g typ t (patt2term p')
-    pattContext env g typ p'
-
-  PAs x p -> do
-    g' <- pattContext env g typ p
-    return ((Explicit,x,typ):g')
-
-  PAlt p' q -> do
-    g1 <- pattContext env g typ p'
-    g2 <- pattContext env g typ q
-    let pts = nub ([x | pt@(_,x,_) <- g1, notElem pt g2] ++ [x | pt@(_,x,_) <- g2, notElem pt g1])
-    checkCond 
-      ("incompatible bindings of" <+>
-       fsep pts <+> 
-       "in pattern alterantives" <+> ppPatt Unqualified 0 p) (null pts) 
-    return g1 -- must be g1 == g2
-  PSeq p q -> do
-    g1 <- pattContext env g typ p
-    g2 <- pattContext env g typ q
-    return $ g1 ++ g2
-  PRep p' -> noBind typeStr p'
-  PNeg p' -> noBind typ p'
-
-  _ -> return [] ---- check types!
- where 
-   noBind typ p' = do
-    co <- pattContext env g typ p'
-    if not (null co)
-      then checkWarn ("no variable bound inside pattern" <+> ppPatt Unqualified 0 p) 
-           >> return []
-      else return []
-
-checkEqLType :: SourceGrammar -> Context -> Type -> Type -> Term -> Check Type
-checkEqLType gr g t u trm = do
-  (b,t',u',s) <- checkIfEqLType gr g t u trm
-  case b of
-    True  -> return t'
-    False ->
-      let inferredType = ppTerm Qualified 0 u
-          expectedType = ppTerm Qualified 0 t
-          term = ppTerm Unqualified 0 trm
-          funName = pp . head . words .render $ term
-          helpfulMsg =
-            case (arrows inferredType, arrows expectedType) of
-              (0,0) -> pp "" -- None of the types is a function
-              _ -> "\n   **" <+> 
-                  if expectedType `isLessApplied` inferredType
-                    then "Maybe you gave too few arguments to" <+> funName
-                    else pp "Double-check that type signature and number of arguments match."
-      in checkError $ s <+> "type of" <+> term $$
-                            "expected:" <+> expectedType $$ -- ppqType t u $$
-                            "inferred:" <+> inferredType $$ -- ppqType u t
-                            helpfulMsg
-  where
-    -- count the number of arrows in the prettyprinted term
-    arrows :: Doc -> Int
-    arrows = length . filter (=="->") . words . render
-
-    -- If prettyprinted type t has fewer arrows then prettyprinted type u,
-    -- then t is "less applied", and we can print out more helpful error msg.
-    isLessApplied :: Doc -> Doc -> Bool
-    isLessApplied t u = arrows t < arrows u
-
-checkIfEqLType :: SourceGrammar -> Context -> Type -> Type -> Term -> Check (Bool,Type,Type,String)
-checkIfEqLType gr g t u trm = do
-  t' <- computeLType gr g t
-  u' <- computeLType gr g u
-  case t' == u' || alpha [] t' u' of
-    True -> return (True,t',u',[])
-    -- forgive missing lock fields by only generating a warning.
-    --- better: use a flag to forgive? (AR 31/1/2006)
-    _ -> case missingLock [] t' u' of
-      Ok lo -> do
-        checkWarn $ "missing lock field" <+> fsep lo
-        return (True,t',u',[])
-      Bad s -> return (False,t',u',s)
-
-  where
-
-   -- check that u is a subtype of t
-   --- quick hack version of TC.eqVal
-   alpha g t u = case (t,u) of  
-
-     -- error (the empty type!) is subtype of any other type
-     (_,u) | u == typeError -> True
-
-     -- contravariance
-     (Prod _ x a b, Prod _ y c d) -> alpha g c a && alpha ((x,y):g) b d 
-                              
-     -- record subtyping
-     (RecType rs, RecType ts) -> all (\ (l,a) -> 
-                                   any (\ (k,b) -> l == k && alpha g a b) ts) rs
-     (ExtR r s, ExtR r' s') -> alpha g r r' && alpha g s s'
-     (ExtR r s, t) -> alpha g r t || alpha g s t
-
-     -- the following say that Ints n is a subset of Int and of Ints m >= n
-     -- But why does it also allow Int as a subtype of Ints m? /TH 2014-04-04
-     (t,u) | Just m <- isTypeInts t, Just n <- isTypeInts u -> m >= n
-           | Just _ <- isTypeInts t, u == typeInt           -> True ---- check size!
-           | t == typeInt,           Just _ <- isTypeInts u -> True ---- why this ???? AR 11/12/2005
-
-     ---- this should be made in Rename
-     (Q  (m,a), Q  (n,b)) | a == b -> elem m (allExtendsPlus gr n) 
-                                   || elem n (allExtendsPlus gr m)
-                                   || m == n --- for Predef
-     (QC (m,a), QC (n,b)) | a == b -> elem m (allExtendsPlus gr n) 
-                                   || elem n (allExtendsPlus gr m)
-     (QC (m,a), Q  (n,b)) | a == b -> elem m (allExtendsPlus gr n) 
-                                   || elem n (allExtendsPlus gr m)
-     (Q  (m,a), QC (n,b)) | a == b -> elem m (allExtendsPlus gr n) 
-                                   || elem n (allExtendsPlus gr m)
-
-     -- contravariance
-     (Table a b,  Table c d)  -> alpha g c a && alpha g b d
-     (Vr x,       Vr y)       -> x == y || elem (x,y) g || elem (y,x) g
-     _                        -> t == u 
-     --- the following should be one-way coercions only. AR 4/1/2001
-                                  || elem t sTypes && elem u sTypes
-                                  || (t == typeType && u == typePType) 
-                                  || (u == typeType && t == typePType) 
-
-   missingLock g t u = case (t,u) of  
-     (RecType rs, RecType ts) -> 
-       let 
-         ls = [l | (l,a) <- rs, 
-                   not (any (\ (k,b) -> alpha g a b && l == k) ts)]
-         (locks,others) = partition isLockLabel ls
-       in case others of
-         _:_ -> Bad $ render ("missing record fields:" <+> fsep (punctuate ',' (others)))
-         _ -> return locks
-     -- contravariance
-     (Prod _ x a b, Prod _ y c d) -> do
-        ls1 <- missingLock g c a
-        ls2 <- missingLock g b d
-        return $ ls1 ++ ls2
-
-     _ -> Bad ""
-
-   sTypes = [typeStr, typeTok, typeString]
-
--- auxiliaries
-
--- | light-weight substitution for dep. types
-substituteLType :: Context -> Type -> Check Type
-substituteLType g t = case t of
-  Vr x -> return $ maybe t id $ lookup x [(x,t) | (_,x,t) <- g]
-  _ -> composOp (substituteLType g) t
-
-termWith :: Term -> Check Type -> Check (Term, Type)
-termWith t ct = do
-  ty <- ct
-  return (t,ty)
-
--- | compositional check\/infer of binary operations
-check2 :: (Term -> Check Term) -> (Term -> Term -> Term) -> 
-          Term -> Term -> Type -> Check (Term,Type)
-check2 chk con a b t = do
-  a' <- chk a
-  b' <- chk b
-  return (con a' b', t)
-
--- printing a type with a lock field lock_C as C
-ppType :: Type -> Doc
-ppType ty =
-  case ty of
-    RecType fs   -> case filter isLockLabel $ map fst fs of
-                      [lock] -> pp (drop 5 (showIdent (label2ident lock)))
-                      _      -> ppTerm Unqualified 0 ty
-    Prod _ x a b -> ppType a <+> "->" <+> ppType b
-    _            -> ppTerm Unqualified 0 ty
-{-
-ppqType :: Type -> Type -> Doc
-ppqType t u = case (ppType t, ppType u) of
-  (pt,pu) | render pt == render pu -> ppTerm Qualified 0 t
-  (pt,_) -> pt
--}
-checkLookup :: Ident -> Context -> Check Type
-checkLookup x g =
-  case [ty | (b,y,ty) <- g, x == y] of
-    []     -> checkError ("unknown variable" <+> x)
-    (ty:_) -> return ty

src/compiler/GF/Compile/TypeCheck/TC.hs

@@ -5,21 +5,22 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/10/02 20:50:19 $ 
+-- > CVS $Date: 2005/10/02 20:50:19 $
 -- > CVS $Author: aarne $
 -- > CVS $Revision: 1.11 $
 --
 -- Thierry Coquand's type checking algorithm that creates a trace
 -----------------------------------------------------------------------------
 
-module GF.Compile.TypeCheck.TC (AExp(..),
-	   Theory,
-	   checkExp,
-	   inferExp,
-	   checkBranch,
-	   eqVal,
-	   whnf
-	  ) where
+module GF.Compile.TypeCheck.TC (
+    AExp(..),
+    Theory,
+    checkExp,
+    inferExp,
+    checkBranch,
+    eqVal,
+    whnf
+  ) where
 
 import GF.Data.Operations
 import GF.Grammar
@@ -31,17 +32,17 @@ import Data.Maybe
 import GF.Text.Pretty
 
 data AExp =
-     AVr   Ident Val 
+     AVr   Ident Val
    | ACn   QIdent Val
-   | AType 
-   | AInt  Int 
+   | AType
+   | AInt  Integer
    | AFloat Double
    | AStr  String
    | AMeta MetaId Val
    | ALet  (Ident,(Val,AExp)) AExp
-   | AApp  AExp AExp Val 
-   | AAbs  Ident Val AExp 
-   | AProd Ident AExp AExp 
+   | AApp  AExp AExp Val
+   | AAbs  Ident Val AExp
+   | AProd Ident AExp AExp
 -- -- | AEqs  [([Exp],AExp)] --- not used
    | ARecType [ALabelling]
    | AR    [AAssign]
@@ -50,7 +51,7 @@ data AExp =
    | AData Val
   deriving (Eq,Show)
 
-type ALabelling = (Label, AExp) 
+type ALabelling = (Label, AExp)
 type AAssign = (Label, (Val, AExp))
 
 type Theory = QIdent -> Err Val
@@ -71,7 +72,7 @@ whnf :: Val -> Err Val
 whnf v = ---- errIn ("whnf" +++ prt v) $ ---- debug
          case v of
   VApp u w -> do
-    u' <- whnf u 
+    u' <- whnf u
     w' <- whnf w
     app u' w'
   VClos env e -> eval env e
@@ -81,9 +82,9 @@ app :: Val -> Val -> Err Val
 app u v = case u of
   VClos env (Abs _ x e) -> eval ((x,v):env) e
   _ -> return $ VApp u v
-  
+
 eval :: Env -> Term -> Err Val
-eval env e = ---- errIn ("eval" +++ prt e +++ "in" +++ prEnv env) $ 
+eval env e = ---- errIn ("eval" +++ prt e +++ "in" +++ prEnv env) $
              case e of
   Vr x    -> lookupVar env x
   Q c   -> return $ VCn c
@@ -95,23 +96,23 @@ eval env e = ---- errIn ("eval" +++ prt e +++ "in" +++ prEnv env) $
   _ -> return $ VClos env e
 
 eqVal :: Int -> Val -> Val -> Err [(Val,Val)]
-eqVal k u1 u2 = ---- errIn (prt u1 +++ "<>" +++ prBracket (show k) +++ prt u2) $ 
+eqVal k u1 u2 = ---- errIn (prt u1 +++ "<>" +++ prBracket (show k) +++ prt u2) $
                 do
   w1 <- whnf u1
-  w2 <- whnf u2 
+  w2 <- whnf u2
   let v = VGen k
   case (w1,w2) of
     (VApp f1 a1, VApp f2 a2) -> liftM2 (++) (eqVal k f1 f2) (eqVal k a1 a2)
     (VClos env1 (Abs _ x1 e1), VClos env2 (Abs _ x2 e2)) ->
       eqVal (k+1) (VClos ((x1,v x1):env1) e1) (VClos ((x2,v x1):env2) e2)
     (VClos env1 (Prod _ x1 a1 e1), VClos env2 (Prod _ x2 a2 e2)) ->
-      liftM2 (++) 
+      liftM2 (++)
         (eqVal k     (VClos            env1  a1) (VClos            env2  a2))
         (eqVal (k+1) (VClos ((x1,v x1):env1) e1) (VClos ((x2,v x1):env2) e2))
     (VGen i _, VGen j _) -> return [(w1,w2) | i /= j]
-    (VCn (_, i), VCn (_,j)) -> return [(w1,w2) | i /= j] 
+    (VCn (_, i), VCn (_,j)) -> return [(w1,w2) | i /= j]
     --- thus ignore qualifications; valid because inheritance cannot
-    --- be qualified. Simplifies annotation. AR 17/3/2005 
+    --- be qualified. Simplifies annotation. AR 17/3/2005
     _ -> return [(w1,w2) | w1 /= w2]
 -- invariant: constraints are in whnf
 
@@ -127,10 +128,10 @@ checkExp th tenv@(k,rho,gamma) e ty = do
 
     Abs _ x t -> case typ of
       VClos env (Prod _ y a b) -> do
-	a' <- whnf $ VClos env a ---
-	(t',cs) <- checkExp th 
-                           (k+1,(x,v x):rho, (x,a'):gamma) t (VClos ((y,v x):env) b)
-	return (AAbs x a' t', cs)
+                           a' <- whnf $ VClos env a ---
+                           (t',cs) <- checkExp th
+                                          (k+1,(x,v x):rho, (x,a'):gamma) t (VClos ((y,v x):env) b)
+                           return (AAbs x a' t', cs)
       _ -> Bad (render ("function type expected for" <+> ppTerm Unqualified 0 e <+> "instead of" <+> ppValue Unqualified 0 typ))
 
     Let (x, (mb_typ, e1)) e2 -> do
@@ -150,7 +151,7 @@ checkExp th tenv@(k,rho,gamma) e ty = do
       (b',csb) <- checkType th (k+1, (x,v x):rho, (x,VClos rho a):gamma) b
       return (AProd x a' b', csa ++ csb)
 
-    R xs -> 
+    R xs ->
       case typ of
         VRecType ys -> do case [l | (l,_) <- ys, isNothing (lookup l xs)] of
                             [] -> return ()
@@ -174,7 +175,7 @@ checkInferExp th tenv@(k,_,_) e typ = do
   (e',w,cs1) <- inferExp th tenv e
   cs2 <- eqVal k w typ
   return (e',cs1 ++ cs2)
-      
+
 inferExp :: Theory -> TCEnv -> Term -> Err (AExp, Val, [(Val,Val)])
 inferExp th tenv@(k,rho,gamma) e = case e of
    Vr x -> mkAnnot (AVr x) $ noConstr $ lookupVar gamma x
@@ -200,13 +201,13 @@ inferExp th tenv@(k,rho,gamma) e = case e of
     (e2,val2,cs2) <- inferExp th (k,rho,(x,val1):gamma) e2
     return (ALet (x,(val1,e1)) e2, val2, cs1++cs2)
    App f t -> do
-    (f',w,csf) <- inferExp th tenv f 
+    (f',w,csf) <- inferExp th tenv f
     typ <- whnf w
     case typ of
       VClos env (Prod _ x a b) -> do
         (a',csa) <- checkExp th tenv t (VClos env a)
-	b' <- whnf $ VClos ((x,VClos rho t):env) b
-	return $ (AApp f' a' b', b', csf ++ csa)
+        b' <- whnf $ VClos ((x,VClos rho t):env) b
+        return $ (AApp f' a' b', b', csf ++ csa)
       _ -> Bad (render ("Prod expected for function" <+> ppTerm Unqualified 0 f <+> "instead of" <+> ppValue Unqualified 0 typ))
    _ -> Bad (render ("cannot infer type of expression" <+> ppTerm Unqualified 0 e))
 
@@ -232,9 +233,9 @@ checkAssign th tenv@(k,rho,gamma) typs (lbl,(Nothing,exp)) = do
                    return ((lbl,(val,aexp)),cs)
 
 checkBranch :: Theory -> TCEnv -> Equation -> Val -> Err (([Term],AExp),[(Val,Val)])
-checkBranch th tenv b@(ps,t) ty = errIn ("branch" +++ show b) $ 
-                                  chB tenv' ps' ty 
- where 
+checkBranch th tenv b@(ps,t) ty = errIn ("branch" +++ show b) $
+                                  chB tenv' ps' ty
+ where
 
   (ps',_,rho2,k') = ps2ts k ps
   tenv' = (k, rho2++rho, gamma) ---- k' ?
@@ -245,11 +246,11 @@ checkBranch th tenv b@(ps,t) ty = errIn ("branch" +++ show b) $
       typ <- whnf ty
       case typ of
         VClos env (Prod _ y a b) -> do
-	  a' <- whnf $ VClos env a
+          a' <- whnf $ VClos env a
           (p', sigma, binds, cs1) <- checkP tenv p y a'
           let tenv' = (length binds, sigma ++ rho, binds ++ gamma)
           ((ps',exp),cs2) <- chB tenv' ps2 (VClos ((y,p'):env) b)
-	  return ((p:ps',exp), cs1 ++ cs2) -- don't change the patt
+          return ((p:ps',exp), cs1 ++ cs2) -- don't change the patt
         _ -> Bad (render ("Product expected for definiens" <+> ppTerm Unqualified 0 t <+> "instead of" <+> ppValue Unqualified 0 typ))
     [] -> do
       (e,cs) <- checkExp th tenv t ty
@@ -259,15 +260,15 @@ checkBranch th tenv b@(ps,t) ty = errIn ("branch" +++ show b) $
      let sigma = [(x, VGen i x) | ((x,_),i) <- zip delta [k..]]
      return (VClos sigma t, sigma, delta, cs)
 
-  ps2ts k = foldr p2t ([],0,[],k) 
+  ps2ts k = foldr p2t ([],0,[],k)
   p2t p (ps,i,g,k) = case p of
-     PW      -> (Meta i : ps, i+1,g,k) 
+     PW      -> (Meta i : ps, i+1,g,k)
      PV x    -> (Vr x   : ps, i, upd x k g,k+1)
      PAs x p -> p2t p (ps,i,g,k)
      PString s -> (K s : ps, i, g, k)
      PInt n -> (EInt n : ps, i, g, k)
      PFloat n -> (EFloat n : ps, i, g, k)
-     PP c xs -> (mkApp (Q c) xss : ps, j, g',k') 
+     PP c xs -> (mkApp (Q c) xss : ps, j, g',k')
                     where (xss,j,g',k') = foldr p2t ([],i,g,k) xs
      PImplArg p -> p2t p (ps,i,g,k)
      PTilde t -> (t : ps, i, g, k)
@@ -307,8 +308,8 @@ checkPatt th tenv exp val = do
        case typ of
          VClos env (Prod _ x a b) -> do
            (a',_,csa) <- checkExpP tenv t (VClos env a)
-	   b' <- whnf $ VClos ((x,VClos rho t):env) b
-	   return $ (AApp f' a' b', b', csf ++ csa)
+           b' <- whnf $ VClos ((x,VClos rho t):env) b
+           return $ (AApp f' a' b', b', csf ++ csa)
          _ -> Bad (render ("Prod expected for function" <+> ppTerm Unqualified 0 f <+> "instead of" <+> ppValue Unqualified 0 typ))
      _ -> Bad (render ("cannot typecheck pattern" <+> ppTerm Unqualified 0 exp))
 
@@ -321,4 +322,3 @@ mkAnnot :: (Val -> AExp) -> Err (Val,[(Val,Val)]) -> Err (AExp,Val,[(Val,Val)])
 mkAnnot a ti = do
   (v,cs) <- ti
   return (a v, v, cs)
-

src/compiler/GF/Compile/Update.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/05/30 18:39:44 $ 
+-- > CVS $Date: 2005/05/30 18:39:44 $
 -- > CVS $Author: aarne $
 -- > CVS $Revision: 1.8 $
 --
@@ -34,14 +34,14 @@ buildAnyTree :: Fail.MonadFail m => ModuleName -> [(Ident,Info)] -> m (Map.Map I
 buildAnyTree m = go Map.empty
   where
     go map []         = return map
-    go map ((c,j):is) = do
+    go map ((c,j):is) =
       case Map.lookup c map of
         Just i  -> case unifyAnyInfo m i j of
-		     Ok k  -> go (Map.insert c k map) is
-		     Bad _ -> fail $ render ("conflicting information in module"<+>m $$
-		                             nest 4 (ppJudgement Qualified (c,i)) $$
-		                             "and" $+$
-		                             nest 4 (ppJudgement Qualified (c,j)))
+          Ok k  -> go (Map.insert c k map) is
+          Bad _ -> fail $ render ("conflicting information in module"<+>m $$
+                                  nest 4 (ppJudgement Qualified (c,i)) $$
+                                  "and" $+$
+                                  nest 4 (ppJudgement Qualified (c,j)))
         Nothing -> go (Map.insert c j map) is
 
 extendModule :: FilePath -> SourceGrammar -> SourceModule -> Check SourceModule
@@ -51,14 +51,14 @@ extendModule cwd gr (name,m)
   ---- Should be replaced by real control. AR 4/2/2005
   | mstatus m == MSIncomplete && isModCnc m = return (name,m)
   | otherwise                               = checkInModule cwd m NoLoc empty $ do
-                                                 m' <- foldM extOne m (mextend m) 
+                                                 m' <- foldM extOne m (mextend m)
                                                  return (name,m')
  where
    extOne mo (n,cond) = do
      m0 <- lookupModule gr n
 
      -- test that the module types match, and find out if the old is complete
-     unless (sameMType (mtype m) (mtype mo)) 
+     unless (sameMType (mtype m) (mtype mo))
             (checkError ("illegal extension type to module" <+> name))
 
      let isCompl = isCompleteModule m0
@@ -67,7 +67,7 @@ extendModule cwd gr (name,m)
      js1 <- extendMod gr isCompl ((n,m0), isInherited cond) name (jments mo)
 
      -- if incomplete, throw away extension information
-     return $ 
+     return $
           if isCompl
             then mo {jments = js1}
             else mo {mextend= filter ((/=n) . fst) (mextend mo)
@@ -75,10 +75,10 @@ extendModule cwd gr (name,m)
                     ,jments = js1
                     }
 
--- | rebuilding instance + interface, and "with" modules, prior to renaming. 
+-- | rebuilding instance + interface, and "with" modules, prior to renaming.
 -- AR 24/10/2003
 rebuildModule :: FilePath -> SourceGrammar -> SourceModule -> Check SourceModule
-rebuildModule cwd gr mo@(i,mi@(ModInfo mt stat fs_ me mw ops_ med_ msrc_ env_ js_)) =
+rebuildModule cwd gr mo@(i,mi@(ModInfo mt stat fs_ me mw ops_ med_ msrc_ js_)) =
   checkInModule cwd mi NoLoc empty $ do
 
 ----  deps <- moduleDeps ms
@@ -88,8 +88,8 @@ rebuildModule cwd gr mo@(i,mi@(ModInfo mt stat fs_ me mw ops_ med_ msrc_ env_ js
 
     -- add the information given in interface into an instance module
     Nothing -> do
-      unless (null is || mstatus mi == MSIncomplete) 
-             (checkError ("module" <+> i <+> 
+      unless (null is || mstatus mi == MSIncomplete)
+             (checkError ("module" <+> i <+>
                           "has open interfaces and must therefore be declared incomplete"))
       case mt of
         MTInstance (i0,mincl) -> do
@@ -113,9 +113,9 @@ rebuildModule cwd gr mo@(i,mi@(ModInfo mt stat fs_ me mw ops_ med_ msrc_ env_ js
       let stat' = if all (flip elem infs) is
                     then MSComplete
                     else MSIncomplete
-      unless (stat' == MSComplete || stat == MSIncomplete) 
+      unless (stat' == MSComplete || stat == MSIncomplete)
              (checkError ("module" <+> i <+> "remains incomplete"))
-      ModInfo mt0 _ fs me' _ ops0 _ fpath _ js <- lookupModule gr ext
+      ModInfo mt0 _ fs me' _ ops0 _ fpath js <- lookupModule gr ext
       let ops1 = nub $
                    ops_ ++ -- N.B. js has been name-resolved already
                    [OQualif i j | (i,j) <- ops] ++
@@ -131,7 +131,7 @@ rebuildModule cwd gr mo@(i,mi@(ModInfo mt stat fs_ me mw ops_ med_ msrc_ env_ js
                                     js
       let js1 = Map.union js0 js_
       let med1= nub (ext : infs ++ insts ++ med_)
-      return $ ModInfo mt0 stat' fs1 me Nothing ops1 med1 msrc_ env_ js1
+      return $ ModInfo mt0 stat' fs1 me Nothing ops1 med1 msrc_ js1
 
   return (i,mi')
 
@@ -141,24 +141,24 @@ rebuildModule cwd gr mo@(i,mi@(ModInfo mt stat fs_ me mw ops_ med_ msrc_ env_ js
 extendMod :: Grammar ->
              Bool -> (Module,Ident -> Bool) -> ModuleName ->
              Map.Map Ident Info -> Check (Map.Map Ident Info)
-extendMod gr isCompl ((name,mi),cond) base new = foldM try new $ Map.toList (jments mi) 
+extendMod gr isCompl ((name,mi),cond) base new = foldM try new $ Map.toList (jments mi)
   where
     try new (c,i0)
       | not (cond c) = return new
       | otherwise    = case Map.lookup c new of
                          Just j -> case unifyAnyInfo name i j of
-		                     Ok k  -> return $ Map.insert c k new
-		                     Bad _ -> do (base,j) <- case j of 
-		                                               AnyInd _ m -> lookupOrigInfo gr (m,c)
-		                                               _          -> return (base,j)
-		                                 (name,i) <- case i of 
+                                     Ok k  -> return $ Map.insert c k new
+                                     Bad _ -> do (base,j) <- case j of
+                                                               AnyInd _ m -> lookupOrigInfo gr (m,c)
+                                                               _          -> return (base,j)
+                                                 (name,i) <- case i of
                                                                AnyInd _ m -> lookupOrigInfo gr (m,c)
                                                                _          -> return (name,i)
-		                                 checkError ("cannot unify the information" $$ 
-		                                             nest 4 (ppJudgement Qualified (c,i)) $$
-		                                             "in module" <+> name <+> "with" $$
-		                                             nest 4 (ppJudgement Qualified (c,j)) $$
-		                                             "in module" <+> base)
+                                                 checkError ("cannot unify the information" $$
+                                                             nest 4 (ppJudgement Qualified (c,i)) $$
+                                                             "in module" <+> name <+> "with" $$
+                                                             nest 4 (ppJudgement Qualified (c,j)) $$
+                                                             "in module" <+> base)
                          Nothing-> if isCompl
                                      then return $ Map.insert c (indirInfo name i) new
                                      else return $ Map.insert c i new
@@ -166,11 +166,11 @@ extendMod gr isCompl ((name,mi),cond) base new = foldM try new $ Map.toList (jme
         i = globalizeLoc (msrc mi) i0
 
     indirInfo :: ModuleName -> Info -> Info
-    indirInfo n info = AnyInd b n' where 
+    indirInfo n info = AnyInd b n' where
       (b,n') = case info of
-        ResValue _ -> (True,n)
+        ResValue _ _ -> (True,n)
         ResParam _ _ -> (True,n)
-        AbsFun _ _ Nothing _ -> (True,n) 
+        AbsFun _ _ Nothing _ -> (True,n)
         AnyInd b k -> (b,k)
         _ -> (False,n) ---- canonical in Abs
 
@@ -179,7 +179,7 @@ globalizeLoc fpath i =
     AbsCat mc             -> AbsCat (fmap gl mc)
     AbsFun mt ma md moper -> AbsFun (fmap gl mt) ma (fmap (fmap gl) md) moper
     ResParam mt mv        -> ResParam (fmap gl mt) mv
-    ResValue t            -> ResValue (gl t)
+    ResValue t i          -> ResValue (gl t) i
     ResOper mt m          -> ResOper (fmap gl mt) (fmap gl m)
     ResOverload ms os     -> ResOverload ms (map (\(x,y) -> (gl x,gl y)) os)
     CncCat mc md mr mp mpmcfg-> CncCat (fmap gl mc) (fmap gl md) (fmap gl mr) (fmap gl mp) mpmcfg
@@ -194,24 +194,24 @@ globalizeLoc fpath i =
 
 unifyAnyInfo :: ModuleName -> Info -> Info -> Err Info
 unifyAnyInfo m i j = case (i,j) of
-  (AbsCat mc1, AbsCat mc2) -> 
+  (AbsCat mc1, AbsCat mc2) ->
     liftM AbsCat (unifyMaybeL mc1 mc2)
-  (AbsFun mt1 ma1 md1 moper1, AbsFun mt2 ma2 md2 moper2) -> 
+  (AbsFun mt1 ma1 md1 moper1, AbsFun mt2 ma2 md2 moper2) ->
     liftM4 AbsFun (unifyMaybeL mt1 mt2) (unifAbsArrity ma1 ma2) (unifAbsDefs md1 md2) (unifyMaybe moper1 moper2) -- adding defs
 
   (ResParam mt1 mv1, ResParam mt2 mv2) ->
     liftM2 ResParam (unifyMaybeL mt1 mt2) (unifyMaybe mv1 mv2)
-  (ResValue (L l1 t1), ResValue (L l2 t2)) 
-      | t1==t2    -> return (ResValue (L l1 t1))
-      | otherwise -> fail ""
+  (ResValue (L l1 t1) i1, ResValue (L l2 t2) i2)
+      | t1==t2 && i1 == i2 -> return (ResValue (L l1 t1) i1)
+      | otherwise          -> fail ""
   (_, ResOverload ms t) | elem m ms ->
     return $ ResOverload ms t
-  (ResOper mt1 m1, ResOper mt2 m2) -> 
+  (ResOper mt1 m1, ResOper mt2 m2) ->
     liftM2 ResOper (unifyMaybeL mt1 mt2) (unifyMaybeL m1 m2)
 
-  (CncCat mc1 md1 mr1 mp1 mpmcfg1, CncCat mc2 md2 mr2 mp2 mpmcfg2) -> 
+  (CncCat mc1 md1 mr1 mp1 mpmcfg1, CncCat mc2 md2 mr2 mp2 mpmcfg2) ->
     liftM5 CncCat (unifyMaybeL mc1 mc2) (unifyMaybeL md1 md2) (unifyMaybeL mr1 mr2) (unifyMaybeL mp1 mp2)  (unifyMaybe mpmcfg1 mpmcfg2)
-  (CncFun m mt1 md1 mpmcfg1, CncFun _ mt2 md2 mpmcfg2) -> 
+  (CncFun m mt1 md1 mpmcfg1, CncFun _ mt2 md2 mpmcfg2) ->
     liftM3 (CncFun m) (unifyMaybeL mt1 mt2) (unifyMaybeL md1 md2) (unifyMaybe mpmcfg1 mpmcfg2)
 
   (AnyInd b1 m1, AnyInd b2 m2) -> do

src/compiler/GF/CompileInParallel.hs

@@ -1,6 +1,6 @@
 -- | Parallel grammar compilation
 module GF.CompileInParallel(parallelBatchCompile) where
-import Prelude hiding (catch,(<>)) -- GHC 8.4.1 clash with Text.PrettyPrint
+import Prelude hiding (catch,(<>))
 import Control.Monad(join,ap,when,unless)
 import Control.Applicative
 import GF.Infra.Concurrency
@@ -36,11 +36,8 @@ import qualified Control.Monad.Fail as Fail
 parallelBatchCompile jobs opts rootfiles0 =
   do setJobs jobs
      rootfiles <- mapM canonical rootfiles0
-     lib_dirs1 <- getLibraryDirectory opts
-     lib_dirs2 <- mapM canonical lib_dirs1
-     let lib_dir = head lib_dirs2
-     when (length lib_dirs2 >1) $ ePutStrLn ("GF_LIB_PATH defines more than one directory; using the first, " ++ show lib_dir)
-     filepaths <- mapM (getPathFromFile [lib_dir] opts) rootfiles
+     lib_dir  <- canonical =<< getLibraryDirectory opts
+     filepaths <- mapM (getPathFromFile lib_dir opts) rootfiles
      let groups = groupFiles lib_dir filepaths
          n = length groups
      when (n>1) $ ePutStrLn "Grammar mixes present and alltenses, dividing modules into two groups"

src/compiler/GF/CompileOne.hs

@@ -8,7 +8,6 @@ module GF.CompileOne(-- ** Compiling a single module
 import GF.Compile.GetGrammar(getSourceModule)
 import GF.Compile.Rename(renameModule)
 import GF.Compile.CheckGrammar(checkModule)
-import GF.Compile.Optimize(optimizeModule)
 import GF.Compile.SubExOpt(subexpModule,unsubexpModule)
 import GF.Compile.GeneratePMCFG(generatePMCFG)
 import GF.Compile.Update(extendModule,rebuildModule)
@@ -107,10 +106,9 @@ compileSourceModule opts cwd mb_gfFile gr =
 
     -- Apply to complete modules when not generating tags
     backend mo3 =
-      do mo4 <- runPassE Optimize "optimizing" $ optimizeModule opts gr mo3
-         if isModCnc (snd mo4) && flag optPMCFG opts
-          then runPassI "generating PMCFG" $ generatePMCFG opts gr mb_gfFile mo4
-          else runPassI "" $ return mo4
+      do if isModCnc (snd mo3) && flag optPMCFG opts
+          then runPassI "generating PMCFG" $ fmap fst $ runCheck' opts (generatePMCFG opts cwd gr mo3)
+          else runPassI "" $ return mo3
 
     ifComplete yes mo@(_,mi) =
       if isCompleteModule mi then yes mo else return mo
@@ -128,7 +126,6 @@ compileSourceModule opts cwd mb_gfFile gr =
 
     -- * Running a compiler pass, with impedance matching
     runPass = runPass' fst fst snd (liftErr . runCheck' opts)
-    runPassE = runPass2e liftErr id
     runPassI = runPass2e id id Canon
     runPass2e lift dump = runPass' id dump (const "") lift
 

src/compiler/GF/Compiler.hs

@@ -1,8 +1,7 @@
-module GF.Compiler (mainGFC, linkGrammars, writePGF, writeOutputs) where
+module GF.Compiler (mainGFC, linkGrammars, writeGrammar, writeOutputs) where
 
-import PGF
-import PGF.Internal(concretes,optimizePGF,unionPGF)
-import PGF.Internal(putSplitAbs,encodeFile,runPut)
+import PGF2
+import PGF2.Internal(unionPGF,writeConcr)
 import GF.Compile as S(batchCompile,link,srcAbsName)
 import GF.CompileInParallel as P(parallelBatchCompile)
 import GF.Compile.Export
@@ -16,6 +15,7 @@ import GF.Grammar.CFG
 --import GF.Infra.Ident(showIdent)
 import GF.Infra.UseIO
 import GF.Infra.Option
+import GF.Infra.CheckM
 import GF.Data.ErrM
 import GF.System.Directory
 import GF.Text.Pretty(render,render80)
@@ -68,22 +68,25 @@ compileSourceFiles opts fs =
       where
         ofmts = flag optOutputFormats opts
 
-    cnc2haskell (cnc,gr) =
-      do mapM_ writeExport $ concretes2haskell opts (srcAbsName gr cnc) gr
+    cnc2haskell (cnc,gr) = do
+      (res,_) <- runCheck (concretes2haskell opts (srcAbsName gr cnc) gr)
+      mapM_ writeExport res
 
-    abs2canonical (cnc,gr) =
-        writeExport ("canonical/"++render absname++".gf",render80 canAbs)
+    abs2canonical (cnc,gr) = do
+      (canAbs,_) <- runCheck (abstract2canonical absname gr)
+      writeExport ("canonical/"++render absname++".gf",render80 canAbs)
       where
         absname = srcAbsName gr cnc
-        canAbs = abstract2canonical absname gr
 
-    cnc2canonical (cnc,gr) =
-      mapM_ (writeExport.fmap render80) $
-            concretes2canonical opts (srcAbsName gr cnc) gr
+    cnc2canonical (cnc,gr) = do
+      (res,_) <- runCheck (concretes2canonical opts (srcAbsName gr cnc) gr)
+      mapM_ (writeExport.fmap render80) res
 
-    grammar2json (cnc,gr) = encodeJSON (render absname ++ ".json") gr_canon
-      where absname = srcAbsName gr cnc
-            gr_canon = grammar2canonical opts absname gr
+    grammar2json (cnc,gr) = do
+      (gr_canon,_) <- runCheck (grammar2canonical opts absname gr)
+      return (encodeJSON (render absname ++ ".json") gr_canon)
+      where
+        absname = srcAbsName gr cnc
 
     writeExport (path,s) = writing opts path $ writeUTF8File path s
 
@@ -92,7 +95,7 @@ compileSourceFiles opts fs =
 -- in the 'Options') from the output of 'parallelBatchCompile'.
 -- If a @.pgf@ file by the same name already exists and it is newer than the
 -- source grammar files (as indicated by the 'UTCTime' argument), it is not
--- recreated. Calls 'writePGF' and 'writeOutputs'.
+-- recreated. Calls 'writeGrammar' and 'writeOutputs'.
 linkGrammars opts (t_src,~cnc_grs@(~(cnc,gr):_)) =
     do let abs = render (srcAbsName gr cnc)
            pgfFile = outputPath opts (grammarName' opts abs<.>"pgf")
@@ -102,10 +105,8 @@ linkGrammars opts (t_src,~cnc_grs@(~(cnc,gr):_)) =
        if t_pgf >= Just t_src
          then putIfVerb opts $ pgfFile ++ " is up-to-date."
          else do pgfs <- mapM (link opts) cnc_grs
-                 let pgf0 = foldl1 unionPGF pgfs
-                 probs <- maybe (return . defaultProbabilities) readProbabilitiesFromFile (flag optProbsFile opts) pgf0
-                 let pgf = setProbabilities probs pgf0
-                 writePGF opts pgf
+                 let pgf = foldl1 (\one two -> fromMaybe two (unionPGF one two)) pgfs
+                 writeGrammar opts pgf
                  writeOutputs opts pgf
 
 compileCFFiles :: Options -> [FilePath] -> IOE ()
@@ -115,12 +116,11 @@ compileCFFiles opts fs = do
   startCat <- case rules of
                 (Rule cat _ _ : _) -> return cat
                 _                  -> fail "empty CFG"
-  let pgf = cf2pgf (last fs) (mkCFG startCat Set.empty rules)
+  probs <- liftIO (maybe (return Map.empty) readProbabilitiesFromFile (flag optProbsFile opts))
+  let pgf = cf2pgf opts (last fs) (mkCFG startCat Set.empty rules) probs
   unless (flag optStopAfterPhase opts == Compile) $
-     do probs <- liftIO (maybe (return . defaultProbabilities) readProbabilitiesFromFile (flag optProbsFile opts) pgf)
-        let pgf' = setProbabilities probs $ if flag optOptimizePGF opts then optimizePGF pgf else pgf
-        writePGF opts pgf'
-        writeOutputs opts pgf'
+     do writeGrammar opts pgf
+        writeOutputs opts pgf
 
 unionPGFFiles :: Options -> [FilePath] -> IOE ()
 unionPGFFiles opts fs =
@@ -138,14 +138,11 @@ unionPGFFiles opts fs =
 
     doIt =
       do pgfs <- mapM readPGFVerbose fs
-         let pgf0 = foldl1 unionPGF pgfs
-             pgf1 = if flag optOptimizePGF opts then optimizePGF pgf0 else pgf0
-         probs <- liftIO (maybe (return . defaultProbabilities) readProbabilitiesFromFile (flag optProbsFile opts) pgf1)
-         let pgf  = setProbabilities probs pgf1
-             pgfFile = outputPath opts (grammarName opts pgf <.> "pgf")
+         let pgf = foldl1 (\one two -> fromMaybe two (unionPGF one two)) pgfs
+         let pgfFile = outputPath opts (grammarName opts pgf <.> "pgf")
          if pgfFile `elem` fs
            then putStrLnE $ "Refusing to overwrite " ++ pgfFile
-           else writePGF opts pgf
+           else writeGrammar opts pgf
          writeOutputs opts pgf
 
     readPGFVerbose f =
@@ -162,21 +159,24 @@ writeOutputs opts pgf = do
 -- | Write the result of compiling a grammar (e.g. with 'compileToPGF' or
 -- 'link') to a @.pgf@ file.
 -- A split PGF file is output if the @-split-pgf@ option is used.
-writePGF :: Options -> PGF -> IOE ()
-writePGF opts pgf =
-    if flag optSplitPGF opts then writeSplitPGF else writeNormalPGF
+writeGrammar :: Options -> PGF -> IOE ()
+writeGrammar opts pgf =
+  if fst (flag optLinkTargets opts)
+    then if flag optSplitPGF opts
+           then writeSplitPGF
+           else writeNormalPGF
+    else return ()
   where
     writeNormalPGF =
        do let outfile = outputPath opts (grammarName opts pgf <.> "pgf")
-          writing opts outfile $ encodeFile outfile pgf
+          writing opts outfile (writePGF outfile pgf)
 
     writeSplitPGF =
       do let outfile = outputPath opts (grammarName opts pgf <.> "pgf")
-         writing opts outfile $ BSL.writeFile outfile (runPut (putSplitAbs pgf))
-                                --encodeFile_ outfile (putSplitAbs pgf)
-         forM_ (Map.toList (concretes pgf)) $ \cnc -> do
-           let outfile = outputPath opts (showCId (fst cnc) <.> "pgf_c")
-           writing opts outfile $ encodeFile outfile cnc
+         writing opts outfile $ writePGF outfile pgf
+         forM_ (Map.toList (languages pgf)) $ \(concrname,concr) -> do
+           let outfile = outputPath opts (concrname <.> "pgf_c")
+           writing opts outfile (writeConcr outfile concr)
 
 
 writeOutput :: Options -> FilePath-> String -> IOE ()
@@ -186,7 +186,7 @@ writeOutput opts file str = writing opts path $ writeUTF8File path str
 -- * Useful helper functions
 
 grammarName :: Options -> PGF -> String
-grammarName opts pgf = grammarName' opts (showCId (abstractName pgf))
+grammarName opts pgf = grammarName' opts (abstractName pgf)
 grammarName' opts abs = fromMaybe abs (flag optName opts)
 
 outputJustPGF opts = null (flag optOutputFormats opts) && not (flag optSplitPGF opts)

src/compiler/GF/Data/BacktrackM.hs

@@ -16,18 +16,18 @@
 {-# LANGUAGE CPP #-}
 module GF.Data.BacktrackM (
                     -- * the backtracking state monad
-		    BacktrackM,
-		    -- * monad specific utilities
-		    member,
-		    cut,
-		    -- * running the monad
-		    foldBM,          runBM,
-		    foldSolutions,   solutions,
-		    foldFinalStates, finalStates,
-		    
-		    -- * reexport the 'MonadState' class
-		    module Control.Monad.State.Class,
-		  ) where
+                    BacktrackM,
+                    -- * monad specific utilities
+                    member,
+                    cut,
+                    -- * running the monad
+                    foldBM,          runBM,
+                    foldSolutions,   solutions,
+                    foldFinalStates, finalStates,
+
+                    -- * reexport the 'MonadState' class
+                    module Control.Monad.State.Class,
+                  ) where
 
 import Data.List
 import Control.Applicative
@@ -70,7 +70,7 @@ instance Applicative (BacktrackM s) where
 instance Monad (BacktrackM s) where
     return a   = BM (\c s b -> c a s b)
     BM m >>= k = BM (\c s b -> m (\a s b -> unBM (k a) c s b) s b)
-	where unBM (BM m) = m
+        where unBM (BM m) = m
 
 #if !(MIN_VERSION_base(4,13,0))
     fail = Fail.fail

src/compiler/GF/Data/Graph.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/11/10 16:43:44 $ 
+-- > CVS $Date: 2005/11/10 16:43:44 $
 -- > CVS $Author: bringert $
 -- > CVS $Revision: 1.2 $
 --
@@ -34,7 +34,7 @@ import Data.Set (Set)
 import qualified Data.Set as Set
 
 data Graph n a b = Graph [n] ![Node n a] ![Edge n b]
-		 deriving (Eq,Show)
+    deriving (Eq,Show)
 
 type Node n a = (n,a)
 type Edge n b = (n,n,b)
@@ -63,7 +63,7 @@ emap f (Graph c ns es) = Graph c ns [(x,y,f l) | (x,y,l) <- es]
 
 -- | Add a node to the graph.
 newNode :: a               -- ^ Node label
-        -> Graph n a b 
+        -> Graph n a b
         -> (Graph n a b,n) -- ^ Node graph and name of new node
 newNode l (Graph (c:cs) ns es) = (Graph cs ((c,l):ns) es, c)
 
@@ -83,7 +83,7 @@ newEdges es g = foldl' (flip newEdge) g es
 -- lazy version:
 -- newEdges es' (Graph c ns es) = Graph c ns (es'++es)
 
-insertEdgeWith :: Eq n => 
+insertEdgeWith :: Eq n =>
                   (b -> b -> b) -> Edge n b -> Graph n a b -> Graph n a b
 insertEdgeWith f e@(x,y,l) (Graph c ns es) = Graph c ns (h es)
   where h [] = [e]
@@ -97,7 +97,7 @@ removeNode n = removeNodes (Set.singleton n)
 -- | Remove a set of nodes and all edges to and from those nodes.
 removeNodes :: Ord n => Set n -> Graph n a b -> Graph n a b
 removeNodes xs (Graph c ns es) = Graph c ns' es'
-  where 
+  where
   keepNode n = not (Set.member n xs)
   ns' = [ x | x@(n,_) <- ns, keepNode n ]
   es' = [ e | e@(f,t,_) <- es, keepNode f && keepNode t ]
@@ -105,7 +105,7 @@ removeNodes xs (Graph c ns es) = Graph c ns' es'
 -- | Get a map of node names to info about each node.
 nodeInfo :: Ord n => Graph n a b -> NodeInfo n a b
 nodeInfo g = Map.fromList [ (n, (x, fn inc n, fn out n)) | (n,x) <- nodes g ]
-  where 
+  where
   inc = groupEdgesBy edgeTo g
   out = groupEdgesBy edgeFrom g
   fn m n = fromMaybe [] (Map.lookup n m)
@@ -148,16 +148,16 @@ reverseGraph :: Graph n a b -> Graph n a b
 reverseGraph (Graph c ns es) = Graph c ns [ (t,f,l) | (f,t,l) <- es ]
 
 -- | Add the nodes from the second graph to the first graph.
---   The nodes in the second graph will be renamed using the name 
+--   The nodes in the second graph will be renamed using the name
 --   supply in the first graph.
 --   This function is more efficient when the second graph
 --   is smaller than the first.
-mergeGraphs :: Ord m => Graph n a b -> Graph m a b 
+mergeGraphs :: Ord m => Graph n a b -> Graph m a b
             -> (Graph n a b, m -> n) -- ^ The new graph and a function translating
                                       --  the old names of nodes in the second graph
                                       --  to names in the new graph.
 mergeGraphs (Graph c ns1 es1) g2 = (Graph c' (ns2++ns1) (es2++es1), newName)
-  where 
+  where
   (xs,c') = splitAt (length (nodes g2)) c
   newNames = Map.fromList (zip (map fst (nodes g2)) xs)
   newName n = fromJust $ Map.lookup n newNames
@@ -170,7 +170,7 @@ renameNodes :: (n -> m) -- ^ renaming function
             -> Graph n a b -> Graph m a b
 renameNodes newName c (Graph _ ns es) = Graph c ns' es'
     where ns' = map' (\ (n,x) -> (newName n,x)) ns
-	  es' = map' (\ (f,t,l) -> (newName f, newName t, l)) es
+          es' = map' (\ (f,t,l) -> (newName f, newName t, l)) es
 
 -- | A strict 'map'
 map' :: (a -> b) -> [a] -> [b]

src/compiler/GF/Data/Graphviz.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/09/15 18:10:44 $ 
+-- > CVS $Date: 2005/09/15 18:10:44 $
 -- > CVS $Author: bringert $
 -- > CVS $Revision: 1.2 $
 --
@@ -13,14 +13,14 @@
 -----------------------------------------------------------------------------
 
 module GF.Data.Graphviz (
-				  Graph(..), GraphType(..), 
-				  Node(..), Edge(..),
-				  Attr,
-                                  addSubGraphs,
-                                  setName,
-                                  setAttr,
-				  prGraphviz
-			) where
+                          Graph(..), GraphType(..),
+                          Node(..), Edge(..),
+                          Attr,
+                          addSubGraphs,
+                          setName,
+                          setAttr,
+                          prGraphviz
+                        ) where
 
 import Data.Char
 
@@ -70,14 +70,14 @@ prGraphviz g@(Graph t i _ _ _ _) =
     graphtype t ++ " " ++ maybe "" esc i ++ " {\n" ++ prGraph g ++ "}\n"
 
 prSubGraph :: Graph -> String
-prSubGraph g@(Graph _ i _ _ _ _) = 
+prSubGraph g@(Graph _ i _ _ _ _) =
     "subgraph" ++ " " ++ maybe "" esc i ++ " {\n" ++ prGraph g ++ "}"
 
 prGraph :: Graph -> String
-prGraph (Graph t id at ns es ss) = 
+prGraph (Graph t id at ns es ss) =
     unlines $ map (++";") (map prAttr at
-		           ++ map prNode ns 
-		           ++ map (prEdge t) es
+                           ++ map prNode ns
+                           ++ map (prEdge t) es
                            ++ map prSubGraph ss)
 
 graphtype :: GraphType -> String
@@ -96,7 +96,7 @@ edgeop Undirected = "--"
 
 prAttrList :: [Attr] -> String
 prAttrList [] = ""
-prAttrList at =	"[" ++ join "," (map prAttr at) ++ "]"
+prAttrList at = "[" ++ join "," (map prAttr at) ++ "]"
 
 prAttr :: Attr -> String
 prAttr (n,v) = esc n ++ " = " ++ esc v

src/compiler/GF/Data/Operations.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/11/11 16:12:41 $ 
+-- > CVS $Date: 2005/11/11 16:12:41 $
 -- > CVS $Author: bringert $
 -- > CVS $Revision: 1.22 $
 --
@@ -15,34 +15,34 @@
 -----------------------------------------------------------------------------
 
 module GF.Data.Operations (
-		   -- ** The Error monad
-		   Err(..), err, maybeErr, testErr, fromErr, errIn, 
-		   lookupErr,
+     -- ** The Error monad
+     Err(..), err, maybeErr, testErr, fromErr, errIn,
+     lookupErr,
 
-		   -- ** Error monad class
-		   ErrorMonad(..), checks, --doUntil, allChecks, checkAgain,
-                   liftErr,
-		   
-		   -- ** Checking
-		   checkUnique, unifyMaybeBy, unifyMaybe,
+     -- ** Error monad class
+     ErrorMonad(..), checks, --doUntil, allChecks, checkAgain,
+                 liftErr,
 
-           -- ** Monadic operations on lists and pairs
-		   mapPairsM, pairM,
- 
-		   -- ** Printing
-		   indent, (+++), (++-), (++++), (+++-), (+++++),
-		   prUpper, prReplicate, prTList, prQuotedString, prParenth, prCurly, 
-		   prBracket, prArgList, prSemicList, prCurlyList, restoreEscapes,
-		   numberedParagraphs, prConjList, prIfEmpty, wrapLines,
+     -- ** Checking
+     checkUnique, unifyMaybeBy, unifyMaybe,
 
-		   -- ** Topological sorting
-		   topoTest, topoTest2,
+         -- ** Monadic operations on lists and pairs
+     mapPairsM, pairM,
 
-		   -- ** Misc
-		   readIntArg,
-		   iterFix,  chunks,
-                
-		  ) where
+     -- ** Printing
+     indent, (+++), (++-), (++++), (+++-), (+++++),
+     prUpper, prReplicate, prTList, prQuotedString, prParenth, prCurly,
+     prBracket, prArgList, prSemicList, prCurlyList, restoreEscapes,
+     numberedParagraphs, prConjList, prIfEmpty, wrapLines,
+
+     -- ** Topological sorting
+     topoTest, topoTest2,
+
+     -- ** Misc
+     readIntArg,
+     iterFix,  chunks,
+
+    ) where
 
 import Data.Char (isSpace, toUpper, isSpace, isDigit)
 import Data.List (nub, partition, (\\))
@@ -107,7 +107,7 @@ indent i s = replicate i ' ' ++ s
 (+++), (++-), (++++), (+++-), (+++++) :: String -> String -> String
 a +++ b   = a ++ " "    ++ b
 
-a ++- ""  = a 
+a ++- ""  = a
 a ++- b   = a +++ b
 
 a ++++ b  = a ++ "\n"   ++ b
@@ -145,20 +145,20 @@ prCurly   s = "{" ++ s ++ "}"
 prBracket s = "[" ++ s ++ "]"
 
 prArgList, prSemicList, prCurlyList :: [String] -> String
-prArgList   = prParenth . prTList "," 
+prArgList   = prParenth . prTList ","
 prSemicList = prTList " ; "
 prCurlyList = prCurly . prSemicList
 
 restoreEscapes :: String -> String
-restoreEscapes s = 
-  case s of 
+restoreEscapes s =
+  case s of
     []       -> []
     '"' : t  -> '\\' : '"'  : restoreEscapes t
     '\\': t  -> '\\' : '\\' : restoreEscapes t
     c   : t  -> c : restoreEscapes t
 
 numberedParagraphs :: [[String]] -> [String]
-numberedParagraphs t = case t of 
+numberedParagraphs t = case t of
   []   -> []
   p:[] -> p
   _    -> concat [(show n ++ ".") : s | (n,s) <- zip [1..] t]
@@ -204,12 +204,12 @@ topoTest2 g0 = maybe (Right cycles) Left (tsort g)
         ([],[]) -> Just []
         ([],_) -> Nothing
         (ns,rest) -> (leaves:) `fmap` tsort [(n,es \\ leaves) | (n,es)<-rest]
-	  where leaves = map fst ns
+            where leaves = map fst ns
 
 
 -- | Fix point iterator (for computing e.g. transitive closures or reachability)
 iterFix :: Eq a => ([a] -> [a]) -> [a] -> [a]
-iterFix more start = iter start start 
+iterFix more start = iter start start
   where
     iter old new = if (null new')
                       then old
@@ -241,7 +241,7 @@ liftErr e = err raise return e
 {-
 instance ErrorMonad (STM s) where
   raise msg = STM (\s -> raise msg)
-  handle (STM f) g = STM (\s -> (f s) 
+  handle (STM f) g = STM (\s -> (f s)
                                 `handle` (\e -> let STM g' = (g e) in
                                                     g' s))
 

src/compiler/GF/Data/Relation.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/10/26 17:13:13 $ 
+-- > CVS $Date: 2005/10/26 17:13:13 $
 -- > CVS $Author: bringert $
 -- > CVS $Revision: 1.1 $
 --
@@ -83,7 +83,7 @@ transitiveClosure r = fix (Map.map growSet) r
   where growSet ys = foldl Set.union ys (map (allRelated r) $ Set.toList ys)
 
 reflexiveClosure_ :: Ord a => [a] -- ^ The set over which the relation is defined.
-		 -> Rel a -> Rel a
+                  -> Rel a -> Rel a
 reflexiveClosure_ u r = relates [(x,x) | x <- u] r
 
 -- | Uses 'domain'
@@ -104,7 +104,7 @@ reflexiveElements :: Ord a => Rel a -> Set a
 reflexiveElements r = Set.fromList [ x | (x,ys) <- Map.toList r, x `Set.member` ys ]
 
 -- | Keep the related pairs for which the predicate is true.
-filterRel :: Ord a => (a -> a -> Bool) -> Rel a -> Rel a 
+filterRel :: Ord a => (a -> a -> Bool) -> Rel a -> Rel a
 filterRel p = fst . purgeEmpty . Map.mapWithKey (Set.filter . p)
 
 -- | Remove keys that map to no elements.
@@ -112,16 +112,16 @@ purgeEmpty :: Ord a => Rel a -> (Rel a, Set a)
 purgeEmpty r = let (r',r'') = Map.partition (not . Set.null) r
                 in (r', Map.keysSet r'')
 
--- | Get the equivalence classes from an equivalence relation. 
+-- | Get the equivalence classes from an equivalence relation.
 equivalenceClasses :: Ord a => Rel a -> [Set a]
 equivalenceClasses r = equivalenceClasses_ (Map.keys r) r
  where equivalenceClasses_ [] _ = []
        equivalenceClasses_ (x:xs) r = ys:equivalenceClasses_ zs r
-	   where ys = allRelated r x
-                 zs = [x' | x' <- xs, not (x' `Set.member` ys)]
+          where ys = allRelated r x
+                zs = [x' | x' <- xs, not (x' `Set.member` ys)]
 
 isTransitive :: Ord a => Rel a -> Bool
-isTransitive r = and [z `Set.member` ys | (x,ys) <- Map.toList r, 
+isTransitive r = and [z `Set.member` ys | (x,ys) <- Map.toList r,
                       y <- Set.toList ys, z <- Set.toList (allRelated r y)]
 
 isReflexive :: Ord a => Rel a -> Bool
@@ -181,7 +181,7 @@ remove x r = let (mss,r') = Map.updateLookupWithKey (\_ _ -> Nothing) x r
                    Nothing      -> (r', Set.empty, Set.empty)
                    -- remove element from all incoming and outgoing sets
                    -- of other elements
-                   Just (is,os) -> 
+                   Just (is,os) ->
                        let r''  = foldr (\i -> Map.adjust (\ (is',os') -> (is', Set.delete x os')) i) r'  $ Set.toList is
                            r''' = foldr (\o -> Map.adjust (\ (is',os') -> (Set.delete x is', os')) o) r'' $ Set.toList os
                         in (r''', is, os)
@@ -190,4 +190,4 @@ incoming :: Ord a => a -> Rel' a -> Set a
 incoming x r = maybe Set.empty fst $ Map.lookup x r
 
 --outgoing :: Ord a => a -> Rel' a -> Set a
---outgoing x r = maybe Set.empty snd $ Map.lookup x r
+--outgoing x r = maybe Set.empty snd $ Map.lookup x r

src/compiler/GF/Data/Utilities.hs

@@ -4,7 +4,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/10/26 18:47:16 $ 
+-- > CVS $Date: 2005/10/26 18:47:16 $
 -- > CVS $Author: bringert $
 -- > CVS $Revision: 1.6 $
 --
@@ -33,7 +33,7 @@ longerThan    n = not . notLongerThan n
 lookupList :: Eq a => a -> [(a, b)] -> [b]
 lookupList a [] = []
 lookupList a (p:ps) | a == fst p = snd p : lookupList a ps
-		    | otherwise  =         lookupList a ps
+                    | otherwise  =         lookupList a ps
 
 split :: [a] -> ([a], [a])
 split (x : y : as) = (x:xs, y:ys)
@@ -48,8 +48,8 @@ splitBy p (a : as) = if p a then (a:xs, ys) else (xs, a:ys)
 foldMerge :: (a -> a -> a) -> a -> [a] -> a
 foldMerge merge zero = fm
     where fm [] = zero
-	  fm [a] = a
-	  fm abs = let (as, bs) = split abs in fm as `merge` fm bs
+          fm [a] = a
+          fm abs = let (as, bs) = split abs in fm as `merge` fm bs
 
 select :: [a] -> [(a, [a])]
 select [] = []
@@ -68,7 +68,7 @@ safeInit :: [a] -> [a]
 safeInit [] = []
 safeInit xs = init xs
 
--- | Sorts and then groups elements given an ordering of the 
+-- | Sorts and then groups elements given an ordering of the
 --   elements.
 sortGroupBy :: (a -> a -> Ordering) -> [a] -> [[a]]
 sortGroupBy f = groupBy (compareEq f) . sortBy f

src/compiler/GF/Grammar/Analyse.hs

@@ -29,7 +29,7 @@ stripInfo i = case i of
   AbsCat _ -> i
   AbsFun mt mi me mb -> AbsFun mt mi Nothing mb
   ResParam mp mt -> ResParam mp Nothing
-  ResValue lt -> i ----
+  ResValue lt _ -> i ----
   ResOper mt md -> ResOper mt Nothing
   ResOverload is fs -> ResOverload is [(lty, L loc (EInt 0)) | (lty,L loc _) <- fs]
   CncCat mty mte _ mtf mpmcfg -> CncCat mty Nothing Nothing Nothing Nothing
@@ -108,7 +108,7 @@ sizeInfo i = case i of
     sum [sum (map (sizeTerm . patt2term) ps) + sizeTerm t | Just es <- [me], L _ (ps,t) <- es]
   ResParam mp mt -> 
     1 + sum [1 + sum [1 + sizeTerm ty | (_,_,ty) <- co] | Just (L _ ps) <- [mp], (_,co) <- ps]
-  ResValue lt -> 0
+  ResValue _ _ -> 0
   ResOper mt md -> 1 + msize mt + msize md
   ResOverload is fs -> 1 + sum [sizeTerm ty + sizeTerm tr | (L _ ty, L _ tr) <- fs]
   CncCat mty _ _ _ _ -> 1 + msize mty   -- ignoring lindef, linref and printname

src/compiler/GF/Grammar/BNFC.hs

@@ -15,7 +15,6 @@
 module GF.Grammar.BNFC(BNFCRule(..), BNFCSymbol, Symbol(..), CFTerm(..), bnfc2cf) where
 
 import GF.Grammar.CFG
-import PGF (Token, mkCId)
 import Data.List (partition)
 
 type IsList = Bool
@@ -26,7 +25,7 @@ data BNFCRule = BNFCRule {
                  ruleName :: CFTerm }
               | BNFCCoercions {
                  coerCat :: Cat,
-                 coerNum :: Int }
+                 coerNum :: Integer }
               | BNFCTerminator {
                  termNonEmpty :: Bool,
                  termCat :: Cat,
@@ -64,12 +63,12 @@ transformRules sepMap (BNFCCoercions c num) = rules ++ [lastRule]
                 lastRule = Rule (c',[0]) ss rn
                   where c' = c ++ show num
                         ss = [Terminal "(", NonTerminal (c,[0]), Terminal ")"]
-                        rn = CFObj (mkCId $ "coercion_" ++ c) []
+                        rn = CFObj ("coercion_" ++ c) []
 
 fRules c n = Rule (c',[0]) ss rn
    where c' = if n == 0 then c else c ++ show n
          ss = [NonTerminal (c ++ show (n+1),[0])]
-         rn = CFObj (mkCId $ "coercion_" ++ c') []
+         rn = CFObj ("coercion_" ++ c') []
 
 transformSymb :: SepMap -> BNFCSymbol -> (String, ParamCFSymbol)
 transformSymb sepMap s = case s of
@@ -94,7 +93,7 @@ createListRules' ne isSep symb c = ruleBase : ruleCons
                        then [NonTerminal (c,[0]) | ne]
                        else [NonTerminal (c,[0]) | ne] ++
                             [Terminal symb | symb /= "" && ne]
-                rn   = CFObj (mkCId $ "Base" ++ c) []
+                rn   = CFObj ("Base" ++ c) []
         ruleCons
           | isSep && symb /= "" && not ne = [Rule ("List" ++ c,[1]) smbs0 rn
                                             ,Rule ("List" ++ c,[1]) smbs1 rn]
@@ -107,4 +106,4 @@ createListRules' ne isSep symb c = ruleBase : ruleCons
                 smbs = [NonTerminal (c,[0])] ++
                        [Terminal symb | symb /= ""] ++
                        [NonTerminal ("List" ++ c,[0])]
-                rn   = CFObj (mkCId $ "Cons" ++ c) []
+                rn   = CFObj ("Cons" ++ c) []

src/compiler/GF/Grammar/Binary.hs

@@ -10,9 +10,9 @@
 module GF.Grammar.Binary(VersionTagged(..),decodeModuleHeader,decodeModule,encodeModule) where
 
 import Prelude hiding (catch)
+import Control.Monad
 import Control.Exception(catch,ErrorCall(..),throwIO)
-
-import PGF.Internal(Binary(..),Word8,putWord8,getWord8,encodeFile,decodeFile)
+import Data.Binary
 import qualified Data.Map as Map(empty)
 import qualified Data.ByteString.Char8 as BS
 
@@ -22,11 +22,11 @@ import GF.Infra.Option
 import GF.Infra.UseIO(MonadIO(..))
 import GF.Grammar.Grammar
 
-import PGF() -- Binary instances
-import PGF.Internal(Literal(..))
+import PGF2(Literal(..))
+import PGF2.Transactions(Symbol(..))
 
 -- Please change this every time when the GFO format is changed
-gfoVersion = "GF04"
+gfoVersion = "GF05"
 
 instance Binary Ident where
   put id = put (ident2utf8 id)
@@ -44,9 +44,9 @@ instance Binary Grammar where
   get = fmap mGrammar get
 
 instance Binary ModuleInfo where
-  put mi = do put (mtype mi,mstatus mi,mflags mi,mextend mi,mwith mi,mopens mi,mexdeps mi,msrc mi,mseqs mi,jments mi)
-  get    = do (mtype,mstatus,mflags,mextend,mwith,mopens,med,msrc,mseqs,jments) <- get
-              return (ModInfo mtype mstatus mflags mextend mwith mopens med msrc mseqs jments)
+  put mi = do put (mtype mi,mstatus mi,mflags mi,mextend mi,mwith mi,mopens mi,mexdeps mi,msrc mi,jments mi)
+  get    = do (mtype,mstatus,mflags,mextend,mwith,mopens,med,msrc,jments) <- get
+              return (ModInfo mtype mstatus mflags mextend mwith mopens med msrc jments)
 
 instance Binary ModuleType where
   put MTAbstract       = putWord8 0
@@ -103,24 +103,19 @@ instance Binary Options where
              toString (LInt n) = show n
              toString (LFlt d) = show d
 
-instance Binary Production where
-  put (Production res funid args) = put (res,funid,args)
-  get = do res   <- get
-           funid <- get
-           args  <- get
-           return (Production res funid args)
+instance Binary PMCFGCat where
+  put (PMCFGCat r rs) = put (r,rs)
+  get = get >>= \(r,rs) -> return (PMCFGCat r rs)
 
-instance Binary PMCFG where
-  put (PMCFG prods funs) = put (prods,funs)
-  get = do prods <- get
-           funs  <- get
-           return (PMCFG prods funs)
+instance Binary PMCFGRule where
+  put (PMCFGRule res args rules) = put (res,args,rules)
+  get = get >>= \(res,args,rules) -> return (PMCFGRule res args rules)
 
 instance Binary Info where
   put (AbsCat x)       = putWord8 0 >> put x
   put (AbsFun w x y z) = putWord8 1 >> put (w,x,y,z)
   put (ResParam x y)   = putWord8 2 >> put (x,y)
-  put (ResValue x)     = putWord8 3 >> put x
+  put (ResValue x y)   = putWord8 3 >> put (x,y)
   put (ResOper x y)    = putWord8 4 >> put (x,y)
   put (ResOverload x y)= putWord8 5 >> put (x,y)
   put (CncCat v w x y z)=putWord8 6 >> put (v,w,x,y,z)
@@ -131,7 +126,7 @@ instance Binary Info where
              0 -> get >>= \x         -> return (AbsCat x)
              1 -> get >>= \(w,x,y,z) -> return (AbsFun w x y z)
              2 -> get >>= \(x,y)     -> return (ResParam x y)
-             3 -> get >>= \x         -> return (ResValue x)
+             3 -> get >>= \(x,y)     -> return (ResValue x y)
              4 -> get >>= \(x,y)     -> return (ResOper x y)
              5 -> get >>= \(x,y)     -> return (ResOverload x y)
              6 -> get >>= \(v,w,x,y,z)->return (CncCat v w x y z)
@@ -182,14 +177,13 @@ instance Binary Term where
   put (QC x)        = putWord8 25 >> put x
   put (C x y)       = putWord8 26 >> put (x,y)
   put (Glue x y)    = putWord8 27 >> put (x,y)
-  put (EPatt x)     = putWord8 28 >> put x
+  put (EPatt x y z) = putWord8 28 >> put (x,y,z)
   put (EPattType x) = putWord8 29 >> put x
   put (ELincat x y) = putWord8 30 >> put (x,y)
   put (ELin x y)    = putWord8 31 >> put (x,y)
   put (FV x)        = putWord8 32 >> put x
   put (Alts x y)    = putWord8 33 >> put (x,y)
   put (Strs x)      = putWord8 34 >> put x
-  put (Error x)     = putWord8 35 >> put x
 
   get = do tag <- getWord8
            case tag of
@@ -221,14 +215,13 @@ instance Binary Term where
              25 -> get >>= \x       -> return (QC x)
              26 -> get >>= \(x,y)   -> return (C x y)
              27 -> get >>= \(x,y)   -> return (Glue x y)
-             28 -> get >>= \x       -> return (EPatt x)
+             28 -> get >>= \(x,y,z) -> return (EPatt x y z)
              29 -> get >>= \x       -> return (EPattType x)
              30 -> get >>= \(x,y)   -> return (ELincat x y)
              31 -> get >>= \(x,y)   -> return (ELin x y)
              32 -> get >>= \x       -> return (FV x)
              33 -> get >>= \(x,y)   -> return (Alts x y)
              34 -> get >>= \x       -> return (Strs x)
-             35 -> get >>= \x       -> return (Error x)
              _  -> decodingError
 
 instance Binary Patt where
@@ -244,8 +237,8 @@ instance Binary Patt where
   put (PAs  x y)   = putWord8 10 >> put (x,y)
   put (PNeg x)     = putWord8 11 >> put x
   put (PAlt x y)   = putWord8 12 >> put (x,y)
-  put (PSeq x y)   = putWord8 13 >> put (x,y)
-  put (PRep x)     = putWord8 14 >> put x
+  put (PSeq minx maxx x miny maxy y)   = putWord8 13 >> put (minx,maxx,x,miny,maxy,y)
+  put (PRep minx maxx x)= putWord8 14 >> put (minx,maxx,x)
   put (PChar)      = putWord8 15
   put (PChars x)   = putWord8 16 >> put x
   put (PMacro x)   = putWord8 17 >> put x
@@ -266,8 +259,8 @@ instance Binary Patt where
              10 -> get >>= \(x,y)   -> return (PAs  x y)
              11 -> get >>= \x       -> return (PNeg x)
              12 -> get >>= \(x,y)   -> return (PAlt x y)
-             13 -> get >>= \(x,y)   -> return (PSeq x y)
-             14 -> get >>= \x       -> return (PRep x)
+             13 -> get >>= \(minx,maxx,x,miny,maxy,y) -> return (PSeq minx maxx x miny maxy y)
+             14 -> get >>= \(minx,maxx,x)-> return (PRep minx maxx x)
              15 ->                     return (PChar)
              16 -> get >>= \x       -> return (PChars x)
              17 -> get >>= \x       -> return (PMacro x)
@@ -298,6 +291,53 @@ instance Binary Label where
              1 -> fmap LVar   get
              _ -> decodingError
 
+instance Binary BindType where
+  put Explicit = putWord8 0
+  put Implicit = putWord8 1
+  get = do tag <- getWord8
+           case tag of
+             0 -> return Explicit
+             1 -> return Implicit
+             _ -> decodingError
+
+instance Binary Literal where
+  put (LStr s) = putWord8 0 >> put s
+  put (LInt i) = putWord8 1 >> put i
+  put (LFlt d) = putWord8 2 >> put d
+  get = do tag <- getWord8
+           case tag of
+             0 -> liftM  LStr get
+             1 -> liftM  LInt get
+             2 -> liftM  LFlt get
+             _ -> decodingError
+
+instance Binary Symbol where
+  put (SymCat d r rs)    = putWord8 0 >> put (d,r,rs)
+  put (SymLit n l)       = putWord8 1 >> put (n,l)
+  put (SymVar n l)       = putWord8 2 >> put (n,l)
+  put (SymKS ts)         = putWord8 3 >> put ts
+  put (SymKP d vs)       = putWord8 4 >> put (d,vs)
+  put SymBIND            = putWord8 5
+  put SymSOFT_BIND       = putWord8 6
+  put SymNE              = putWord8 7
+  put SymSOFT_SPACE      = putWord8 8
+  put SymCAPIT           = putWord8 9
+  put SymALL_CAPIT       = putWord8 10
+  get = do tag <- getWord8
+           case tag of
+             0 -> liftM3 SymCat get get get
+             1 -> liftM2 SymLit get get
+             2 -> liftM2 SymVar get get
+             3 -> liftM  SymKS  get
+             4 -> liftM2 (\d vs -> SymKP d vs) get get
+             5 -> return SymBIND
+             6 -> return SymSOFT_BIND
+             7 -> return SymNE
+             8 -> return SymSOFT_SPACE
+             9 -> return SymCAPIT
+             10-> return SymALL_CAPIT
+             _ -> decodingError
+
 --putGFOVersion = mapM_ (putWord8 . fromIntegral . ord) gfoVersion
 --getGFOVersion = replicateM (length gfoVersion) (fmap (chr . fromIntegral) getWord8)
 --putGFOVersion = put gfoVersion
@@ -332,7 +372,7 @@ decodeModuleHeader :: MonadIO io => FilePath -> io (VersionTagged Module)
 decodeModuleHeader = liftIO . fmap (fmap conv) . decodeFile'
   where
     conv (m,mtype,mstatus,mflags,mextend,mwith,mopens,med,msrc) =
-        (m,ModInfo mtype mstatus mflags mextend mwith mopens med msrc Nothing Map.empty)
+        (m,ModInfo mtype mstatus mflags mextend mwith mopens med msrc Map.empty)
 
 encodeModule :: MonadIO io => FilePath -> SourceModule -> io ()
 encodeModule fpath mo = liftIO $ encodeFile fpath (Tagged mo)

src/compiler/GF/Grammar/CFG.hs

@@ -4,10 +4,11 @@
 --
 -- Context-free grammar representation and manipulation.
 ----------------------------------------------------------------------
-module GF.Grammar.CFG where
+module GF.Grammar.CFG(Cat,Token, module GF.Grammar.CFG) where
 
 import GF.Data.Utilities
-import PGF
+import PGF2(Fun,Cat)
+import PGF2.Transactions(Token)
 import GF.Data.Relation
 
 import Data.Map (Map)
@@ -20,8 +21,6 @@ import qualified Data.Set as Set
 -- * Types
 --
 
-type Cat = String
-
 data Symbol c t = NonTerminal c | Terminal t
   deriving (Eq, Ord, Show)
 
@@ -39,12 +38,12 @@ data Grammar c t = Grammar {
   deriving (Eq, Ord, Show)
 
 data CFTerm
-    = CFObj CId [CFTerm] -- ^ an abstract syntax function with arguments
+    = CFObj Fun [CFTerm] -- ^ an abstract syntax function with arguments
     | CFAbs Int CFTerm -- ^ A lambda abstraction. The Int is the variable id.
     | CFApp CFTerm CFTerm -- ^ Application
     | CFRes Int -- ^ The result of the n:th (0-based) non-terminal
     | CFVar Int -- ^ A lambda-bound variable
-    | CFMeta CId -- ^ A metavariable
+    | CFMeta Fun -- ^ A metavariable
   deriving (Eq, Ord, Show)
 
 type CFSymbol = Symbol  Cat Token
@@ -232,7 +231,7 @@ uniqueFuns = snd . mapAccumL uniqueFun Set.empty
     uniqueFun funs (Rule cat items (CFObj fun args)) = (Set.insert fun' funs,Rule cat items (CFObj fun' args))
       where
         fun' = head [fun'|suffix<-"":map show ([2..]::[Int]),
-                          let fun'=mkCId (showCId fun++suffix),
+                          let fun'=fun++suffix,
                           not (fun' `Set.member` funs)]
 
 -- | Gets all rules in a CFG.
@@ -310,12 +309,12 @@ prProductions prods =
 prCFTerm :: CFTerm -> String
 prCFTerm = pr 0
   where
-    pr p (CFObj f args) = paren p (showCId f ++ " (" ++ concat (intersperse "," (map (pr 0) args)) ++ ")")
+    pr p (CFObj f args) = paren p (f ++ " (" ++ concat (intersperse "," (map (pr 0) args)) ++ ")")
     pr p (CFAbs i t) = paren p ("\\x" ++ show i ++ ". " ++ pr 0 t)
     pr p (CFApp t1 t2) = paren p (pr 1 t1 ++ "(" ++ pr 0 t2 ++ ")")
     pr _ (CFRes i) = "$" ++ show i
     pr _ (CFVar i) = "x" ++ show i
-    pr _ (CFMeta c) = "?" ++ showCId c
+    pr _ (CFMeta c) = "?" ++ c
     paren 0 x = x
     paren 1 x = "(" ++ x ++ ")"
 
@@ -323,12 +322,12 @@ prCFTerm = pr 0
 -- * CFRule Utilities
 --
 
-ruleFun :: Rule c t -> CId
+ruleFun :: Rule c t -> Fun
 ruleFun (Rule _ _ t) = f t
   where f (CFObj n _) = n
         f (CFApp _ x) = f x
         f (CFAbs _ x) = f x
-        f _ = mkCId ""
+        f _ = ""
 
 -- | Check if any of the categories used on the right-hand side
 --   are in the given list of categories.
@@ -336,7 +335,7 @@ anyUsedBy :: Eq c => [c] -> Rule c t -> Bool
 anyUsedBy cs (Rule _ ss _) = any (`elem` cs) (filterCats ss)
 
 mkCFTerm :: String -> CFTerm
-mkCFTerm n = CFObj (mkCId n) []
+mkCFTerm n = CFObj n []
 
 ruleIsNonRecursive :: Ord c => Set c -> Rule c t -> Bool
 ruleIsNonRecursive cs = noCatsInSet cs . ruleRhs

src/compiler/GF/Grammar/Canonical.hs

@@ -9,8 +9,11 @@
 
 {-# LANGUAGE DeriveTraversable #-}
 module GF.Grammar.Canonical where
+
 import Prelude hiding ((<>))
 import GF.Text.Pretty
+import GF.Infra.Ident (RawIdent)
+import PGF(Literal(..))
 
 -- | A Complete grammar
 data Grammar = Grammar Abstract [Concrete] deriving Show
@@ -44,12 +47,12 @@ data LincatDef = LincatDef CatId LinType  deriving Show
 data LinDef    = LinDef FunId [VarId] LinValue  deriving Show
 
 -- | Linearization type, RHS of @lincat@
-data LinType = FloatType 
-             | IntType 
+data LinType = FloatType
+             | IntType
              | ParamType ParamType
              | RecordType [RecordRowType]
-             | StrType 
-             | TableType LinType LinType 
+             | StrType
+             | TableType LinType LinType
              | TupleType [LinType]
               deriving (Eq,Ord,Show)
 
@@ -57,9 +60,9 @@ newtype ParamType = ParamTypeId ParamId deriving (Eq,Ord,Show)
 
 -- | Linearization value, RHS of @lin@
 data LinValue = ConcatValue LinValue LinValue
-              | LiteralValue LinLiteral
+              | LiteralValue Literal
               | ErrorValue String
-              | ParamConstant ParamValue 
+              | ParamConstant ParamValue
               | PredefValue PredefId
               | RecordValue [RecordRowValue]
               | TableValue LinType [TableRowValue]
@@ -73,11 +76,6 @@ data LinValue = ConcatValue LinValue LinValue
               | CommentedValue String LinValue
               deriving (Eq,Ord,Show)
 
-data LinLiteral = FloatConstant Float 
-                | IntConstant Int 
-                | StrConstant String 
-                deriving (Eq,Ord,Show)
-
 data LinPattern = ParamPattern ParamPattern
                 | RecordPattern [RecordRow LinPattern]
                 | TuplePattern [LinPattern]
@@ -106,7 +104,7 @@ newtype PredefId = PredefId Id        deriving (Eq,Ord,Show)
 newtype LabelId  = LabelId Id         deriving (Eq,Ord,Show)
 data VarValueId  = VarValueId QualId  deriving (Eq,Ord,Show)
 
--- | Name of param type or param value 
+-- | Name of param type or param value
 newtype ParamId = ParamId QualId  deriving (Eq,Ord,Show)
 
 --------------------------------------------------------------------------------
@@ -119,14 +117,13 @@ newtype FunId = FunId Id  deriving (Eq,Show)
 
 data VarId = Anonymous | VarId Id  deriving Show
 
-newtype Flags = Flags [(FlagName,FlagValue)] deriving Show
+newtype Flags = Flags [(FlagName,Literal)] deriving Show
 type FlagName = Id
-data FlagValue = Str String | Int Int | Flt Double deriving Show
 
 
 -- *** Identifiers
 
-type Id = String
+type Id = RawIdent
 data QualId = Qual ModId Id | Unqual Id  deriving (Eq,Ord,Show)
 
 --------------------------------------------------------------------------------
@@ -242,14 +239,14 @@ instance PPA LinValue where
              VarValue v -> pp v
              _ -> parens lv
 
-instance Pretty LinLiteral where pp = ppA
+instance Pretty Literal where pp = ppA
 
-instance PPA LinLiteral where
+instance PPA Literal where
   ppA l = case l of
-            FloatConstant f -> pp f
-            IntConstant n -> pp n
-            StrConstant s -> doubleQuotes s -- hmm
-  
+            LFlt f -> pp f
+            LInt n -> pp n
+            LStr s -> doubleQuotes s -- hmm
+
 instance RhsSeparator LinValue where rhsSep _ = pp "="
 
 instance Pretty LinPattern where
@@ -264,8 +261,7 @@ instance PPA LinPattern where
       ParamPattern pv -> ppA pv
       RecordPattern r -> block r
       TuplePattern ps -> "<"<>punctuate "," ps<>">"
-      WildPattern     -> pp "_"                
-      _ -> parens p
+      WildPattern     -> pp "_"
 
 instance RhsSeparator LinPattern where rhsSep _ = pp "="
 
@@ -298,11 +294,6 @@ instance Pretty Flags where
     where
       ppFlag (name,value) = name <+> "=" <+> value <>";"
 
-instance Pretty FlagValue where
-  pp (Str s) = pp s
-  pp (Int i) = pp i
-  pp (Flt d) = pp d
-
 --------------------------------------------------------------------------------
 -- | Pretty print atomically (i.e. wrap it in parentheses if necessary)
 class Pretty a => PPA a where ppA :: a -> Doc

src/compiler/GF/Grammar/CanonicalJSON.hs

@@ -7,7 +7,8 @@ import Control.Applicative ((<|>))
 import Data.Ratio (denominator, numerator)
 import GF.Grammar.Canonical
 import Control.Monad (guard)
-
+import GF.Infra.Ident (RawIdent,showRawIdent,rawIdentS)
+import PGF(Literal(..))
 
 encodeJSON :: FilePath -> Grammar -> IO ()
 encodeJSON fpath g = writeFile fpath (encode g)
@@ -29,7 +30,7 @@ instance JSON Grammar where
 -- ** Abstract Syntax
 
 instance JSON Abstract where
-  showJSON (Abstract absid flags cats funs) 
+  showJSON (Abstract absid flags cats funs)
     = makeObj [("abs", showJSON absid),
                ("flags", showJSON flags),
                ("cats", showJSON cats),
@@ -81,7 +82,7 @@ instance JSON TypeBinding where
 -- ** Concrete syntax
 
 instance JSON Concrete where
-  showJSON (Concrete cncid absid flags params lincats lins) 
+  showJSON (Concrete cncid absid flags params lincats lins)
     = makeObj [("cnc", showJSON cncid),
                ("abs", showJSON absid),
                ("flags", showJSON flags),
@@ -170,13 +171,13 @@ instance JSON LinValue where
     <|>        do vs <- readJSON o :: Result [LinValue]
                   return (foldr1 ConcatValue vs)
 
-instance JSON LinLiteral where
+instance JSON Literal where
   -- basic values (Str, Float, Int) are encoded as JSON strings/numbers:
-  showJSON (StrConstant   s) = showJSON s
-  showJSON (FloatConstant f) = showJSON f
-  showJSON (IntConstant   n) = showJSON n
+  showJSON (LStr s) = showJSON s
+  showJSON (LFlt f) = showJSON f
+  showJSON (LInt n) = showJSON n
 
-  readJSON = readBasicJSON StrConstant IntConstant FloatConstant
+  readJSON = readBasicJSON LStr LInt LFlt
 
 instance JSON LinPattern where
   -- wildcards and patterns without arguments are encoded as strings:
@@ -204,12 +205,12 @@ instance JSON a => JSON (RecordRow a) where
   -- record rows and lists of record rows are both encoded as JSON records (i.e., objects)
   showJSON  row  = showJSONs [row]
   showJSONs rows = makeObj (map toRow rows)
-    where toRow (RecordRow (LabelId lbl) val) = (lbl, showJSON val)
+    where toRow (RecordRow (LabelId lbl) val) = (showRawIdent lbl, showJSON val)
 
   readJSON  obj = head <$> readJSONs obj
   readJSONs obj = mapM fromRow (assocsJSObject obj)
     where fromRow (lbl, jsvalue) = do value <- readJSON jsvalue
-                                      return (RecordRow (LabelId lbl) value)
+                                      return (RecordRow (LabelId (rawIdentS lbl)) value)
 
 instance JSON rhs => JSON (TableRow rhs) where
   showJSON (TableRow l v) = makeObj [(".pattern", showJSON l), (".value", showJSON v)]
@@ -219,19 +220,19 @@ instance JSON rhs => JSON (TableRow rhs) where
 
 -- *** Identifiers in Concrete Syntax
 
-instance JSON PredefId   where showJSON (PredefId    s) = showJSON s ; readJSON = fmap PredefId    . readJSON 
-instance JSON LabelId    where showJSON (LabelId     s) = showJSON s ; readJSON = fmap LabelId     . readJSON 
-instance JSON VarValueId where showJSON (VarValueId  s) = showJSON s ; readJSON = fmap VarValueId  . readJSON 
-instance JSON ParamId    where showJSON (ParamId     s) = showJSON s ; readJSON = fmap ParamId     . readJSON 
-instance JSON ParamType  where showJSON (ParamTypeId s) = showJSON s ; readJSON = fmap ParamTypeId . readJSON 
+instance JSON PredefId   where showJSON (PredefId    s) = showJSON s ; readJSON = fmap PredefId    . readJSON
+instance JSON LabelId    where showJSON (LabelId     s) = showJSON s ; readJSON = fmap LabelId     . readJSON
+instance JSON VarValueId where showJSON (VarValueId  s) = showJSON s ; readJSON = fmap VarValueId  . readJSON
+instance JSON ParamId    where showJSON (ParamId     s) = showJSON s ; readJSON = fmap ParamId     . readJSON
+instance JSON ParamType  where showJSON (ParamTypeId s) = showJSON s ; readJSON = fmap ParamTypeId . readJSON
 
 
 --------------------------------------------------------------------------------
 -- ** Used in both Abstract and Concrete Syntax
 
-instance JSON ModId where showJSON (ModId s) = showJSON s ; readJSON = fmap ModId . readJSON 
-instance JSON CatId where showJSON (CatId s) = showJSON s ; readJSON = fmap CatId . readJSON 
-instance JSON FunId where showJSON (FunId s) = showJSON s ; readJSON = fmap FunId . readJSON 
+instance JSON ModId where showJSON (ModId s) = showJSON s ; readJSON = fmap ModId . readJSON
+instance JSON CatId where showJSON (CatId s) = showJSON s ; readJSON = fmap CatId . readJSON
+instance JSON FunId where showJSON (FunId s) = showJSON s ; readJSON = fmap FunId . readJSON
 
 instance JSON VarId where
   -- the anonymous variable is the underscore:
@@ -242,29 +243,24 @@ instance JSON VarId where
     <|>        VarId <$> readJSON o
 
 instance JSON QualId where
-  showJSON (Qual (ModId m) n) = showJSON (m++"."++n)
+  showJSON (Qual (ModId m) n) = showJSON (showRawIdent m++"."++showRawIdent n)
   showJSON (Unqual n) = showJSON n
 
   readJSON o = do qualid <- readJSON o
                   let (mod, id) = span (/= '.') qualid
-                  return $ if null mod then Unqual id else Qual (ModId mod) id
+                  return $ if null mod then Unqual (rawIdentS id) else Qual (ModId (rawIdentS mod)) (rawIdentS id)
+
+instance JSON RawIdent where
+  showJSON i = showJSON $ showRawIdent i
+  readJSON o = rawIdentS <$> readJSON o
 
 instance JSON Flags where
   -- flags are encoded directly as JSON records (i.e., objects):
-  showJSON (Flags fs) = makeObj [(f, showJSON v) | (f, v) <- fs]
+  showJSON (Flags fs) = makeObj [(showRawIdent f, showJSON v) | (f, v) <- fs]
 
   readJSON obj = Flags <$> mapM fromRow (assocsJSObject obj)
     where fromRow (lbl, jsvalue) = do value <- readJSON jsvalue
-                                      return (lbl, value)
-
-instance JSON FlagValue where
-  -- flag values are encoded as basic JSON types:
-  showJSON (Str s) = showJSON s
-  showJSON (Int i) = showJSON i
-  showJSON (Flt f) = showJSON f
-
-  readJSON = readBasicJSON Str Int Flt
-
+                                      return (rawIdentS lbl, value)
 
 --------------------------------------------------------------------------------
 -- ** Convenience functions

src/compiler/GF/Grammar/EBNF.hs

@@ -16,7 +16,6 @@ module GF.Grammar.EBNF (EBNF, ERule, ERHS(..), ebnf2cf) where
 
 import GF.Data.Operations
 import GF.Grammar.CFG
-import PGF (mkCId)
 
 type EBNF = [ERule]
 type ERule = (ECat, ERHS)
@@ -40,7 +39,7 @@ ebnf2cf :: EBNF -> [ParamCFRule]
 ebnf2cf ebnf = 
   [Rule cat items (mkCFF i cat) | (i,(cat,items)) <- zip [0..] (normEBNF ebnf)]
   where
-    mkCFF i (c,_) = CFObj (mkCId ("Mk" ++ c ++ "_" ++ show i)) []
+    mkCFF i (c,_) = CFObj ("Mk" ++ c ++ "_" ++ show i) []
 
 normEBNF :: EBNF -> [CFJustRule]
 normEBNF erules = let

src/compiler/GF/Grammar/Grammar.hs

@@ -64,7 +64,7 @@ module GF.Grammar.Grammar (
         Location(..), L(..), unLoc, noLoc, ppLocation, ppL,
 
         -- ** PMCFG        
-        PMCFG(..), Production(..), FId, FunId, SeqId, LIndex, Sequence
+        PMCFGCat(..), PMCFGRule(..)
         ) where
 
 import GF.Infra.Ident
@@ -73,7 +73,8 @@ import GF.Infra.Location
 
 import GF.Data.Operations
 
-import PGF.Internal (FId, FunId, SeqId, LIndex, Sequence, BindType(..))
+import PGF2(BindType(..))
+import PGF2.Transactions(Symbol,LIndex,LParam)
 
 import Data.Array.IArray(Array)
 import Data.Array.Unboxed(UArray)
@@ -99,7 +100,6 @@ data ModuleInfo = ModInfo {
     mopens  :: [OpenSpec],
     mexdeps :: [ModuleName],
     msrc    :: FilePath,
-    mseqs   :: Maybe (Array SeqId Sequence),
     jments  :: Map.Map Ident Info
   }
 
@@ -304,14 +304,11 @@ allConcreteModules gr =
   [i | (i, m) <- modules gr, MTConcrete _ <- [mtype m], isCompleteModule m]
 
 
-data Production = Production {-# UNPACK #-} !FId
-                             {-# UNPACK #-} !FunId
-                             [[FId]]
-                  deriving (Eq,Ord,Show)
+data PMCFGCat  = PMCFGCat LIndex [(LIndex,LParam)]
+                 deriving (Eq,Show)
 
-data PMCFG = PMCFG [Production]
-                   (Array FunId (UArray LIndex SeqId)) 
-             deriving (Eq,Show)
+data PMCFGRule = PMCFGRule PMCFGCat [PMCFGCat] [[Symbol]]
+                 deriving (Eq,Show)
 
 -- | the constructors are judgements in 
 --
@@ -328,15 +325,18 @@ data Info =
  | AbsFun   (Maybe (L Type)) (Maybe Int) (Maybe [L Equation]) (Maybe Bool) -- ^ (/ABS/) type, arrity and definition of a function
 
 -- judgements in resource
- | ResParam (Maybe (L [Param])) (Maybe [Term])   -- ^ (/RES/) the second parameter is list of all possible values
- | ResValue (L Type)                             -- ^ (/RES/) to mark parameter constructors for lookup
- | ResOper  (Maybe (L Type)) (Maybe (L Term))    -- ^ (/RES/)
+ | ResParam (Maybe (L [Param])) (Maybe ([Term],Int)) -- ^ (/RES/) The second argument is list of all possible values
+                                                     -- and its precomputed length.
+ | ResValue (L Type) Int                             -- ^ (/RES/) to mark parameter constructors for lookup.
+                                                     -- The second argument is the offset into the list of all values
+                                                     -- where that constructor appears first.
+ | ResOper  (Maybe (L Type)) (Maybe (L Term))        -- ^ (/RES/)
 
- | ResOverload [ModuleName] [(L Type,L Term)]         -- ^ (/RES/) idents: modules inherited
+ | ResOverload [ModuleName] [(L Type,L Term)]        -- ^ (/RES/) idents: modules inherited
 
 -- judgements in concrete syntax
- | CncCat  (Maybe (L Type))             (Maybe (L Term)) (Maybe (L Term)) (Maybe (L Term)) (Maybe PMCFG) -- ^ (/CNC/) lindef ini'zed, 
- | CncFun  (Maybe (Ident,Context,Type)) (Maybe (L Term))                  (Maybe (L Term)) (Maybe PMCFG) -- ^ (/CNC/) type info added at 'TC'
+ | CncCat  (Maybe (L Type))                     (Maybe (L Term)) (Maybe (L Term)) (Maybe (L Term)) (Maybe [PMCFGRule]) -- ^ (/CNC/) lindef ini'zed, 
+ | CncFun  (Maybe ([Ident],Ident,Context,Type)) (Maybe (L Term))                  (Maybe (L Term)) (Maybe [PMCFGRule]) -- ^ (/CNC/) type info added at 'TC'
 
 -- indirection to module Ident
  | AnyInd Bool ModuleName                        -- ^ (/INDIR/) the 'Bool' says if canonical
@@ -353,7 +353,7 @@ data Term =
  | Cn Ident                      -- ^ constant
  | Con Ident                     -- ^ constructor
  | Sort Ident                    -- ^ basic type
- | EInt Int                      -- ^ integer literal
+ | EInt Integer                  -- ^ integer literal
  | EFloat Double                 -- ^ floating point literal
  | K String                      -- ^ string literal or token: @\"foo\"@
  | Empty                         -- ^ the empty string @[]@
@@ -385,7 +385,7 @@ data Term =
  | C Term Term                   -- ^ concatenation: @s ++ t@
  | Glue Term Term                -- ^ agglutination: @s + t@
 
- | EPatt Patt                    -- ^ pattern (in macro definition): # p
+ | EPatt Int (Maybe Int) Patt    -- ^ pattern (in macro definition): # p
  | EPattType Term                -- ^ pattern type: pattern T
 
  | ELincat Ident Term            -- ^ boxed linearization type of Ident
@@ -397,7 +397,7 @@ data Term =
 
  | Alts Term [(Term, Term)]      -- ^ alternatives by prefix: @pre {t ; s\/c ; ...}@
  | Strs [Term]                   -- ^ conditioning prefix strings: @strs {s ; ...}@
- | Error String                  -- ^ error values returned by Predef.error
+ | TSymCat Int LIndex [(LIndex,Ident)]
   deriving (Show, Eq, Ord)
 
 -- | Patterns
@@ -408,7 +408,7 @@ data Patt =
  | PW                     -- ^ wild card pattern: @_@
  | PR [(Label,Patt)]      -- ^ record pattern: @{r = p ; ...}@  -- only concrete
  | PString String         -- ^ string literal pattern: @\"foo\"@  -- only abstract
- | PInt    Int            -- ^ integer literal pattern: @12@    -- only abstract
+ | PInt    Integer        -- ^ integer literal pattern: @12@    -- only abstract
  | PFloat Double          -- ^ float literal pattern: @1.2@    -- only abstract
  | PT Type Patt           -- ^ type-annotated pattern
 
@@ -420,18 +420,23 @@ data Patt =
  -- regular expression patterns
  | PNeg Patt              -- ^ negated pattern: -p
  | PAlt Patt Patt         -- ^ disjunctive pattern: p1 | p2
- | PSeq Patt Patt         -- ^ sequence of token parts: p + q
- | PMSeq MPatt MPatt      -- ^ sequence of token parts: p + q
- | PRep Patt              -- ^ repetition of token part: p*
+ | PSeq Int (Maybe Int) Patt Int (Maybe Int) Patt
+                          -- ^ sequence of token parts: p + q
+                          -- In the constructor PSeq minp maxp p minq maxq q,
+                          -- minp/maxp and minq/maxq are the minimal/maximal
+                          -- length of a matching string for p/q.
+ | PRep Int (Maybe Int) Patt
+                          -- ^ repetition of token part: p*
+                          -- In the constructor PRep minp maxp p,
+                          -- minp/maxp is the minimal/maximal length of
+                          -- a matching string for p.
+ 
  | PChar                  -- ^ string of length one: ?
  | PChars [Char]          -- ^ character list: ["aeiou"]
  | PMacro Ident           -- #p
  | PM QIdent              -- #m.p
   deriving (Show, Eq, Ord)
 
--- | Measured pattern (paired with the min & max matching length)
-type MPatt = ((Int,Int),Patt)
-
 -- | to guide computation and type checking of tables
 data TInfo = 
    TRaw         -- ^ received from parser; can be anything
@@ -448,7 +453,7 @@ data Label =
 
 type MetaId = Int
 
-type Hypo     = (BindType,Ident,Term)   -- (x:A)  (_:A)  A  ({x}:A)
+type Hypo     = (BindType,Ident,Type)   -- (x:A)  (_:A)  A  ({x}:A)
 type Context  = [Hypo]                  -- (x:A)(y:B)   (x,y:A)   (_,_:A)
 type Equation = ([Patt],Term) 
 

src/compiler/GF/Grammar/Lexer.x

@@ -130,7 +130,7 @@ data Token
  | T_separator
  | T_nonempty
  | T_String  String          -- string literals
- | T_Integer Int             -- integer literals
+ | T_Integer Integer         -- integer literals
  | T_Double  Double          -- double precision float literals
  | T_Ident   Ident
  | T_EOF
@@ -267,7 +267,7 @@ type AlexInput2 = (AlexInput,AlexInput)
 
 data ParseResult a
   = POk AlexInput2 a
-  | PFailed Posn	-- The position of the error
+  | PFailed Posn        -- The position of the error
             String      -- The error message
 
 newtype P a = P { unP :: AlexInput2 -> ParseResult a }

src/compiler/GF/Grammar/Lookup.hs

@@ -6,7 +6,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/10/27 13:21:53 $ 
+-- > CVS $Date: 2005/10/27 13:21:53 $
 -- > CVS $Author: aarne $
 -- > CVS $Revision: 1.15 $
 --
@@ -20,17 +20,17 @@ module GF.Grammar.Lookup (
            lookupOrigInfo,
            allOrigInfos,
            lookupResDef, lookupResDefLoc,
-           lookupResType, 
+           lookupResType,
            lookupOverload,
            lookupOverloadTypes,
-           lookupParamValues, 
+           lookupParamValues,
            allParamValues,
-           lookupAbsDef, 
-           lookupLincat, 
+           lookupAbsDef,
+           lookupLincat,
            lookupFunType,
            lookupCatContext,
            allOpers, allOpersTo
-	      ) where
+          ) where
 
 import GF.Data.Operations
 import GF.Infra.Ident
@@ -69,7 +69,7 @@ lookupResDef gr x = fmap unLoc (lookupResDefLoc gr x)
 lookupResDefLoc gr (m,c)
   | isPredefCat c = fmap noLoc (lock c defLinType)
   | otherwise     = look m c
-  where 
+  where
     look m c = do
       info <- lookupQIdentInfo gr (m,c)
       case info of
@@ -77,13 +77,13 @@ lookupResDefLoc gr (m,c)
         ResOper _ Nothing  -> return (noLoc (Q (m,c)))
         CncCat (Just (L l ty)) _ _ _ _ -> fmap (L l) (lock c ty)
         CncCat _ _ _ _ _         -> fmap noLoc (lock c defLinType)
-      
-        CncFun (Just (cat,_,_)) (Just (L l tr)) _ _ -> fmap (L l) (unlock cat tr)
-        CncFun _                (Just ltr) _ _ -> return ltr
+
+        CncFun (Just (_,cat,_,_)) (Just (L l tr)) _ _ -> fmap (L l) (unlock cat tr)
+        CncFun _                  (Just ltr) _ _ -> return ltr
 
         AnyInd _ n        -> look n c
         ResParam _ _      -> return (noLoc (QC (m,c)))
-        ResValue _        -> return (noLoc (QC (m,c)))
+        ResValue _ _      -> return (noLoc (QC (m,c)))
         _   -> raise $ render (c <+> "is not defined in resource" <+> m)
 
 lookupResType :: ErrorMonad m => Grammar -> QIdent -> m Type
@@ -94,12 +94,12 @@ lookupResType gr (m,c) = do
 
     -- used in reused concrete
     CncCat _ _ _ _ _ -> return typeType
-    CncFun (Just (cat,cont,val)) _ _ _ -> do
-          val' <- lock cat val 
+    CncFun (Just (_,cat,cont,val)) _ _ _ -> do
+          val' <- lock cat val
           return $ mkProd cont val' []
     AnyInd _ n        -> lookupResType gr (n,c)
-    ResParam _ _      -> return typePType
-    ResValue (L _ t)  -> return t
+    ResParam _ _    -> return typePType
+    ResValue (L _ t) _ -> return t
     _   -> raise $ render (c <+> "has no type defined in resource" <+> m)
 
 lookupOverloadTypes :: ErrorMonad m => Grammar -> QIdent -> m [(Term,Type)]
@@ -110,11 +110,11 @@ lookupOverloadTypes gr id@(m,c) = do
 
     -- used in reused concrete
     CncCat _ _ _ _ _ -> ret typeType
-    CncFun (Just (cat,cont,val)) _ _ _ -> do
-          val' <- lock cat val 
+    CncFun (Just (_,cat,cont,val)) _ _ _ -> do
+          val' <- lock cat val
           ret $ mkProd cont val' []
-    ResParam _ _      -> ret typePType
-    ResValue (L _ t)  -> ret t
+    ResParam _ _    -> ret typePType
+    ResValue (L _ t) _ -> ret t
     ResOverload os tysts -> do
             tss <- mapM (\x -> lookupOverloadTypes gr (x,c)) os
             return $ [(tr,ty) | (L _ ty,L _ tr) <- tysts] ++
@@ -130,8 +130,8 @@ lookupOverload gr (m,c) = do
     case info of
       ResOverload os tysts -> do
             tss <- mapM (\x -> lookupOverload gr (x,c)) os
-            return $ [let (args,val) = typeFormCnc ty in (map (\(b,x,t) -> t) args,(val,tr)) | 
-                      (L _ ty,L _ tr) <- tysts] ++ 
+            return $ [let (args,val) = typeFormCnc ty in (map (\(b,x,t) -> t) args,(val,tr)) |
+                      (L _ ty,L _ tr) <- tysts] ++
                      concat tss
 
       AnyInd _ n  -> lookupOverload gr (n,c)
@@ -154,8 +154,8 @@ lookupParamValues :: ErrorMonad m => Grammar -> QIdent -> m [Term]
 lookupParamValues gr c = do
   (_,info) <- lookupOrigInfo gr c
   case info of
-    ResParam _ (Just pvs) -> return pvs
-    _                     -> raise $ render (ppQIdent Qualified c <+> "has no parameter values defined")
+    ResParam _ (Just (pvs,_)) -> return pvs
+    _                         -> raise $ render (ppQIdent Qualified c <+> "has no parameter values defined")
 
 allParamValues :: ErrorMonad m => Grammar -> Type -> m [Term]
 allParamValues cnc ptyp =
@@ -216,7 +216,7 @@ lookupCatContext gr m c = do
 -- notice that it only gives the modules that are reachable and the opers that are included
 
 allOpers :: Grammar -> [(QIdent,Type,Location)]
-allOpers gr = 
+allOpers gr =
   [((m,op),typ,loc) |
       (m,mi)  <- maybe [] (allExtends gr) (greatestResource gr),
       (op,info)  <- Map.toList (jments mi),
@@ -226,9 +226,9 @@ allOpers gr =
     typesIn info = case info of
       AbsFun  (Just ltyp) _ _ _ -> [ltyp]
       ResOper (Just ltyp) _     -> [ltyp]
-      ResValue ltyp             -> [ltyp]
+      ResValue ltyp _           -> [ltyp]
       ResOverload _ tytrs       -> [ltyp | (ltyp,_) <- tytrs]
-      CncFun  (Just (i,ctx,typ)) _ _ _ ->
+      CncFun  (Just (_,i,ctx,typ)) _ _ _ ->
                                    [L NoLoc (mkProdSimple ctx (lock' i typ))]
       _                         -> []
 

src/compiler/GF/Grammar/Macros.hs

@@ -5,7 +5,7 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/11/11 16:38:00 $ 
+-- > CVS $Date: 2005/11/11 16:38:00 $
 -- > CVS $Author: bringert $
 -- > CVS $Revision: 1.24 $
 --
@@ -51,14 +51,14 @@ typeForm t =
     _      -> error (render ("no normal form of type" <+> ppTerm Unqualified 0 t))
 
 typeFormCnc :: Type -> (Context, Type)
-typeFormCnc t = 
+typeFormCnc t =
   case t of
     Prod b x a t -> let (x', v) = typeFormCnc t
                     in ((b,x,a):x',v)
     _            -> ([],t)
 
 valCat :: Type -> Cat
-valCat typ = 
+valCat typ =
   let (_,cat,_) = typeForm typ
   in cat
 
@@ -99,7 +99,7 @@ isHigherOrderType t = fromErr True $ do  -- pessimistic choice
 contextOfType :: Monad m => Type -> m Context
 contextOfType typ = case typ of
   Prod b x a t -> liftM ((b,x,a):) $ contextOfType t
-  _            -> return [] 
+  _            -> return []
 
 termForm :: Monad m => Term -> m ([(BindType,Ident)], Term, [Term])
 termForm t = case t of
@@ -108,8 +108,8 @@ termForm t = case t of
         return ((b,x):x', fun, args)
    App c a ->
      do (_,fun, args) <- termForm c
-        return ([],fun,args ++ [a]) 
-   _       -> 
+        return ([],fun,args ++ [a])
+   _       ->
      return ([],t,[])
 
 termFormCnc :: Term -> ([(BindType,Ident)], Term)
@@ -216,7 +216,7 @@ typeTok   = Sort cTok
 typeStrs  = Sort cStrs
 
 typeString, typeFloat, typeInt :: Type
-typeInts :: Int -> Type
+typeInts :: Integer -> Type
 typePBool :: Type
 typeError :: Type
 
@@ -227,7 +227,7 @@ typeInts i = App (cnPredef cInts) (EInt i)
 typePBool = cnPredef cPBool
 typeError = cnPredef cErrorType
 
-isTypeInts :: Type -> Maybe Int
+isTypeInts :: Type -> Maybe Integer
 isTypeInts (App c (EInt i)) | c == cnPredef cInts = Just i
 isTypeInts _                                      = Nothing
 
@@ -238,12 +238,6 @@ isPredefConstant t = case t of
   Q (mod,_) | mod == cPredef || mod == cPredefAbs -> True
   _                                               -> False
 
-checkPredefError :: Fail.MonadFail m => Term -> m Term
-checkPredefError t =
-    case t of
-      Error s -> fail ("Error: "++s)
-      _ -> return t
-
 cnPredef :: Ident -> Term
 cnPredef f = Q (cPredef,f)
 
@@ -254,7 +248,7 @@ mkTable :: [Term] -> Term -> Term
 mkTable tt t = foldr Table t tt
 
 mkCTable :: [(BindType,Ident)] -> Term -> Term
-mkCTable ids v = foldr ccase v ids where 
+mkCTable ids v = foldr ccase v ids where
   ccase (_,x) t = T TRaw [(PV x,t)]
 
 mkHypo :: Term -> Hypo
@@ -287,7 +281,7 @@ plusRecType t1 t2 = case (t1, t2) of
     filter (`elem` (map fst r1)) (map fst r2) of
       [] -> return (RecType (r1 ++ r2))
       ls -> raise $ render ("clashing labels" <+> hsep ls)
-  _ -> raise $ render ("cannot add record types" <+> ppTerm Unqualified 0 t1 <+> "and" <+> ppTerm Unqualified 0 t2) 
+  _ -> raise $ render ("cannot add record types" <+> ppTerm Unqualified 0 t1 <+> "and" <+> ppTerm Unqualified 0 t2)
 
 --plusRecord :: Term -> Term -> Err Term
 plusRecord t1 t2 =
@@ -304,7 +298,7 @@ defLinType = RecType [(theLinLabel,  typeStr)]
 
 -- | refreshing variables
 mkFreshVar :: [Ident] -> Ident
-mkFreshVar olds = varX (maxVarIndex olds + 1) 
+mkFreshVar olds = varX (maxVarIndex olds + 1)
 
 -- | trying to preserve a given symbol
 mkFreshVarX :: [Ident] -> Ident -> Ident
@@ -313,7 +307,7 @@ mkFreshVarX olds x = if (elem x olds) then (varX (maxVarIndex olds + 1)) else x
 maxVarIndex :: [Ident] -> Int
 maxVarIndex = maximum . ((-1):) . map varIndex
 
-mkFreshVars :: Int -> [Ident] -> [Ident] 
+mkFreshVars :: Int -> [Ident] -> [Ident]
 mkFreshVars n olds = [varX (maxVarIndex olds + i) | i <- [1..n]]
 
 -- | quick hack for refining with var in editor
@@ -324,7 +318,7 @@ freshAsTerm s = Vr (varX (readIntArg s))
 string2term :: String -> Term
 string2term = K
 
-int2term :: Int -> Term
+int2term :: Integer -> Term
 int2term = EInt
 
 float2term :: Double -> Term
@@ -384,7 +378,7 @@ term2patt trm = case termForm trm of
     return (PNeg a')
   Ok ([], Cn id, [a]) | id == cRep -> do
     a' <- term2patt a
-    return (PRep a')
+    return (PRep 0 Nothing a')
   Ok ([], Cn id, []) | id == cRep -> do
     return PChar
   Ok ([], Cn id,[K s]) | id == cChars  -> do
@@ -392,7 +386,7 @@ term2patt trm = case termForm trm of
   Ok ([], Cn id, [a,b]) | id == cSeq -> do
     a' <- term2patt a
     b' <- term2patt b
-    return (PSeq a' b')
+    return (PSeq 0 Nothing a' 0 Nothing b')
   Ok ([], Cn id, [a,b]) | id == cAlt -> do
     a' <- term2patt a
     b' <- term2patt b
@@ -413,18 +407,18 @@ patt2term pt = case pt of
   PC c pp   -> mkApp (Con c) (map patt2term pp)
   PP c pp   -> mkApp (QC  c) (map patt2term pp)
 
-  PR r      -> R [assign l (patt2term p) | (l,p) <- r] 
+  PR r      -> R [assign l (patt2term p) | (l,p) <- r]
   PT _ p    -> patt2term p
   PInt i    -> EInt i
   PFloat i  -> EFloat i
-  PString s -> K s 
+  PString s -> K s
 
   PAs x p   -> appCons cAs    [Vr x, patt2term p]            --- an encoding
   PChar     -> appCons cChar  []                             --- an encoding
   PChars s  -> appCons cChars [K s]                          --- an encoding
-  PSeq a b  -> appCons cSeq   [(patt2term a), (patt2term b)] --- an encoding
+  PSeq _ _ a _ _ b  -> appCons cSeq   [(patt2term a), (patt2term b)] --- an encoding
   PAlt a b  -> appCons cAlt   [(patt2term a), (patt2term b)] --- an encoding
-  PRep a    -> appCons cRep   [(patt2term a)]                --- an encoding
+  PRep _ _ a-> appCons cRep   [(patt2term a)]                --- an encoding
   PNeg a    -> appCons cNeg   [(patt2term a)]                --- an encoding
 
 
@@ -436,7 +430,7 @@ composSafeOp op = runIdentity . composOp (return . op)
 
 -- | to define compositional term functions
 composOp :: Monad m => (Term -> m Term) -> Term -> m Term
-composOp co trm = 
+composOp co trm =
  case trm of
    App c a          -> liftM2 App (co c) (co a)
    Abs b x t        -> liftM (Abs b x) (co t)
@@ -475,9 +469,8 @@ composPattOp op patt =
     PImplArg p      -> liftM  PImplArg (op p)
     PNeg p          -> liftM  PNeg (op p)
     PAlt p1 p2      -> liftM2 PAlt (op p1) (op p2)
-    PSeq p1 p2      -> liftM2 PSeq (op p1) (op p2)
-    PMSeq (_,p1) (_,p2) -> liftM2 PSeq (op p1) (op p2) -- information loss
-    PRep p          -> liftM  PRep (op p)
+    PSeq _ _ p1 _ _ p2  -> liftM2 (\p1 p2 -> PSeq 0 Nothing p1 0 Nothing p2) (op p1) (op p2)
+    PRep _ _ p      -> liftM  (PRep 0 Nothing) (op p)
     _               -> return patt -- covers cases without subpatterns
 
 collectOp :: Monoid m => (Term -> m) -> Term -> m
@@ -514,9 +507,8 @@ collectPattOp op patt =
     PImplArg p      -> op p
     PNeg p          -> op p
     PAlt p1 p2      -> op p1++op p2
-    PSeq p1 p2      -> op p1++op p2
-    PMSeq (_,p1) (_,p2) -> op p1++op p2
-    PRep p          -> op p
+    PSeq _ _ p1 _ _ p2 -> op p1++op p2
+    PRep _ _ p      -> op p
     _               -> []     -- covers cases without subpatterns
 
 
@@ -552,13 +544,13 @@ strsFromTerm t = case t of
     v0 <- mapM (strsFromTerm . fst) vs
     c0 <- mapM (strsFromTerm . snd) vs
   --let vs' = zip v0 c0
-    return [strTok (str2strings def) vars | 
+    return [strTok (str2strings def) vars |
               def  <- d0,
-              vars <- [[(str2strings v, map sstr c) | (v,c) <- zip vv c0] | 
+              vars <- [[(str2strings v, map sstr c) | (v,c) <- zip vv c0] |
                                                           vv <- sequence v0]
            ]
   FV ts -> mapM strsFromTerm ts >>= return . concat
-  Strs ts -> mapM strsFromTerm ts >>= return . concat  
+  Strs ts -> mapM strsFromTerm ts >>= return . concat
   _ -> raise (render ("cannot get Str from term" <+> ppTerm Unqualified 0 t))
 
 getTableType :: TInfo -> Err Type
@@ -590,11 +582,11 @@ noExist = FV []
 defaultLinType :: Type
 defaultLinType = mkRecType linLabel [typeStr]
 
--- normalize records and record types; put s first
+-- | normalize records and record types; put s first
 
 sortRec :: [(Label,a)] -> [(Label,a)]
 sortRec = sortBy ordLabel where
-  ordLabel (r1,_) (r2,_) = 
+  ordLabel (r1,_) (r2,_) =
     case (showIdent (label2ident r1), showIdent (label2ident r2)) of
       ("s",_) -> LT
       (_,"s") -> GT
@@ -605,7 +597,7 @@ sortRec = sortBy ordLabel where
 -- | dependency check, detecting circularities and returning topo-sorted list
 
 allDependencies :: (ModuleName -> Bool) -> Map.Map Ident Info -> [(Ident,[Ident])]
-allDependencies ism b = 
+allDependencies ism b =
   [(f, nub (concatMap opty (pts i))) | (f,i) <- Map.toList b]
   where
     opersIn t = case t of

src/compiler/GF/Grammar/Parser.y

@@ -25,7 +25,6 @@ import GF.Compile.Update (buildAnyTree)
 import Data.List(intersperse)
 import Data.Char(isAlphaNum)
 import qualified Data.Map as Map
-import PGF(mkCId)
 
 }
 
@@ -133,14 +132,14 @@ ModDef
                                           (opens,jments,opts) = case content of { Just c -> c; Nothing -> ([],[],noOptions) }
                                       jments <- mapM (checkInfoType mtype) jments
                                       defs <- buildAnyTree id jments
-                                      return (id, ModInfo mtype mstat opts extends with opens [] "" Nothing defs)  }
+                                      return (id, ModInfo mtype mstat opts extends with opens [] "" defs)  }
 
 ModHeader :: { SourceModule }
 ModHeader
   : ComplMod ModType '=' ModHeaderBody { let { mstat = $1 ;
                                                (mtype,id) = $2 ;
                                                (extends,with,opens) = $4 }
-                                         in (id, ModInfo mtype mstat noOptions extends with opens [] "" Nothing Map.empty) }
+                                         in (id, ModInfo mtype mstat noOptions extends with opens [] "" Map.empty) }
 
 ComplMod :: { ModuleStatus }
 ComplMod 
@@ -268,7 +267,7 @@ DataDef
 ParamDef :: { [(Ident,Info)] }
 ParamDef
   : Posn LhsIdent '=' ListParConstr Posn { ($2, ResParam (Just (mkL $1 $5 [param | L loc param <- $4])) Nothing) :
-                                        [(f, ResValue (L loc (mkProdSimple co (Cn $2)))) | L loc (f,co) <- $4] }
+                                           [(f, ResValue (L loc (mkProdSimple co (Cn $2))) 0) | L loc (f,co) <- $4] }
   | Posn LhsIdent                   Posn { [($2, ResParam Nothing Nothing)] }
 
 OperDef :: { [(Ident,Info)] }
@@ -445,7 +444,7 @@ Exp4
   | 'pre' '{' String ';' ListAltern '}' { Alts (K $3) $5 }
   | 'pre' '{' Ident ';' ListAltern '}' { Alts (Vr $3) $5 }
   | 'strs' '{' ListExp '}'           { Strs $3       }
-  | '#' Patt3                        { EPatt $2      }
+  | '#' Patt3                        { EPatt 0 Nothing $2 }
   | 'pattern' Exp5                   { EPattType $2  }
   | 'lincat' Ident Exp5              { ELincat $2 $3 }
   | 'lin' Ident Exp5                 { ELin $2 $3 }
@@ -486,14 +485,14 @@ Exps
 Patt :: { Patt }
 Patt
   : Patt '|' Patt1            { PAlt $1 $3 } 
-  | Patt '+' Patt1            { PSeq $1 $3 }
+  | Patt '+' Patt1            { PSeq 0 Nothing $1 0 Nothing $3 }
   | Patt1                     { $1         }
 
 Patt1 :: { Patt }
 Patt1
   : Ident ListPatt            { PC $1 $2 } 
   | ModuleName '.' Ident ListPatt  { PP ($1,$3) $4 }
-  | Patt3 '*'                 { PRep $1 }
+  | Patt3 '*'                 { PRep 0 Nothing $1 }
   | Patt2                     { $1 }
 
 Patt2 :: { Patt }
@@ -625,7 +624,7 @@ ListCFRule
 
 CFRule :: { [BNFCRule] }
 CFRule
-  : Ident '.' Ident '::=' ListCFSymbol ';' { [BNFCRule (showIdent $3) $5 (CFObj (mkCId (showIdent $1)) [])]
+  : Ident '.' Ident '::=' ListCFSymbol ';' { [BNFCRule (showIdent $3) $5 (CFObj (showIdent $1) [])]
                                            }
   | Ident '::=' ListCFRHS ';'              { let { cat = showIdent $1;
                                                    mkFun cat its =
@@ -638,7 +637,7 @@ CFRule
                                                        Terminal    c     -> filter isAlphaNum c;
                                                        NonTerminal (t,_) -> t
                                                      }
-                                             } in map (\rhs -> BNFCRule cat rhs (CFObj (mkCId (mkFun cat rhs)) [])) $3
+                                             } in map (\rhs -> BNFCRule cat rhs (CFObj (mkFun cat rhs) [])) $3
                                            }
   | 'coercions' Ident Integer ';'          { [BNFCCoercions (showIdent $2) $3]}
   | 'terminator' NonEmpty Ident String ';' { [BNFCTerminator $2 (showIdent $3) $4] }
@@ -775,7 +774,7 @@ checkInfoType mt jment@(id,info) =
     CncCat pty pd pr ppn _->ifConcrete mt (locPerh pty ++ locPerh pd ++ locPerh pr ++ locPerh ppn)
     CncFun _   pd ppn _  -> ifConcrete mt (locPerh pd ++ locPerh ppn)
     ResParam pparam _    -> ifResource mt (locPerh pparam)
-    ResValue ty          -> ifResource mt (locL ty)
+    ResValue ty _        -> ifResource mt (locL ty)
     ResOper  pty pt      -> ifOper mt pty pt
     ResOverload _ xs     -> ifResource mt (concat [[loc1,loc2] | (L loc1 _,L loc2 _) <- xs])
   where

src/compiler/GF/Grammar/PatternMatch.hs

@@ -5,18 +5,18 @@
 -- Stability   : (stable)
 -- Portability : (portable)
 --
--- > CVS $Date: 2005/10/12 12:38:29 $ 
+-- > CVS $Date: 2005/10/12 12:38:29 $
 -- > CVS $Author: aarne $
 -- > CVS $Revision: 1.7 $
 --
 -- pattern matching for both concrete and abstract syntax. AR -- 16\/6\/2003
 -----------------------------------------------------------------------------
 
-module GF.Grammar.PatternMatch (matchPattern,
-		     testOvershadow, 
-		     findMatch,
-                     measurePatt
-		    ) where
+module GF.Grammar.PatternMatch (
+                                matchPattern,
+                                testOvershadow,
+                                findMatch
+                               ) where
 
 import GF.Data.Operations
 import GF.Grammar.Grammar
@@ -24,13 +24,13 @@ import GF.Infra.Ident
 import GF.Grammar.Macros
 --import GF.Grammar.Printer
 
---import Data.List
+import Data.Maybe(fromMaybe)
 import Control.Monad
 import GF.Text.Pretty
 --import Debug.Trace
 
 matchPattern :: ErrorMonad m => [(Patt,rhs)] -> Term -> m (rhs, Substitution)
-matchPattern pts term = 
+matchPattern pts term =
   if not (isInConstantForm term)
     then raise (render ("variables occur in" <+> pp term))
   else do
@@ -61,15 +61,15 @@ testOvershadow pts vs = do
 findMatch :: ErrorMonad m => [([Patt],rhs)] -> [Term] -> m (rhs, Substitution)
 findMatch cases terms = case cases of
    [] -> raise (render ("no applicable case for" <+> hsep (punctuate ',' terms)))
-   (patts,_):_ | length patts /= length terms -> 
-       raise (render ("wrong number of args for patterns :" <+> hsep patts <+> 
+   (patts,_):_ | length patts /= length terms ->
+       raise (render ("wrong number of args for patterns :" <+> hsep patts <+>
                     "cannot take" <+> hsep terms))
    (patts,val):cc -> case mapM tryMatch (zip patts terms) of
        Ok substs -> return (val, concat substs)
        _         -> findMatch cc terms
 
 tryMatch :: (Patt, Term) -> Err [(Ident, Term)]
-tryMatch (p,t) = do 
+tryMatch (p,t) = do
   t' <- termForm t
   trym p t'
  where
@@ -83,26 +83,26 @@ tryMatch (p,t) = do
       (PString s, ([],K i,[])) | s==i -> return []
       (PInt s, ([],EInt i,[])) | s==i -> return []
       (PFloat s,([],EFloat i,[])) | s==i -> return [] --- rounding?
-      (PC p pp, ([], Con f, tt)) | 
+      (PC p pp, ([], Con f, tt)) |
             p `eqStrIdent` f && length pp == length tt ->
          do matches <- mapM tryMatch (zip pp tt)
             return (concat matches)
 
-      (PP (q,p) pp, ([], QC (r,f), tt)) | 
+      (PP (q,p) pp, ([], QC (r,f), tt)) |
             -- q `eqStrIdent` r &&  --- not for inherited AR 10/10/2005
             p `eqStrIdent` f && length pp == length tt ->
          do matches <- mapM tryMatch (zip pp tt)
             return (concat matches)
       ---- hack for AppPredef bug
-      (PP (q,p) pp, ([], Q (r,f), tt)) | 
-            -- q `eqStrIdent` r && --- 
+      (PP (q,p) pp, ([], Q (r,f), tt)) |
+            -- q `eqStrIdent` r && ---
             p `eqStrIdent` f && length pp == length tt ->
          do matches <- mapM tryMatch (zip pp tt)
             return (concat matches)
 
       (PR r, ([],R r',[])) |
             all (`elem` map fst r') (map fst r) ->
-         do matches <- mapM tryMatch 
+         do matches <- mapM tryMatch
                             [(p,snd a) | (l,p) <- r, let Just a = lookup l r']
             return (concat matches)
       (PT _ p',_) -> trym p' t'
@@ -121,11 +121,10 @@ tryMatch (p,t) = do
         Bad _ -> return []
         _ -> raise (render ("no match with negative pattern" <+> p))
 
-      (PSeq p1 p2, ([],K s, [])) -> matchPSeq p1 p2 s
-      (PMSeq mp1 mp2, ([],K s, [])) -> matchPMSeq mp1 mp2 s
+      (PSeq min1 max1 p1 min2 max2 p2, ([],K s, [])) -> matchPSeq min1 max1 p1 min2 max2 p2 s
 
-      (PRep p1, ([],K s, [])) -> checks [
-         trym (foldr (const (PSeq p1)) (PString "") 
+      (PRep _ _ p1, ([],K s, [])) -> checks [
+         trym (foldr (const (PSeq 0 Nothing p1 0 Nothing)) (PString "")
            [1..n]) t' | n <- [0 .. length s]
         ] >>
         return []
@@ -139,50 +138,14 @@ tryMatch (p,t) = do
   words2term [w]    = K w
   words2term (w:ws) = C (K w) (words2term ws)
 
-
-matchPMSeq (m1,p1) (m2,p2) s = matchPSeq' m1 p1 m2 p2 s
---matchPSeq p1 p2 s = matchPSeq' (0,maxBound::Int) p1 (0,maxBound::Int) p2 s
-matchPSeq p1 p2 s = matchPSeq' (lengthBounds p1) p1 (lengthBounds p2) p2 s
-
-matchPSeq' b1@(min1,max1) p1 b2@(min2,max2) p2 s =
+matchPSeq min1 max1 p1 min2 max2 p2 s =
   do let n = length s
-         lo = min1 `max` (n-max2)
-         hi = (n-min2) `min` max1
+         lo = min1 `max` (n-fromMaybe n max2)
+         hi = (n-min2) `min` (fromMaybe n max1)
          cuts = [splitAt i s | i <- [lo..hi]]
      matches <- checks [mapM tryMatch [(p1,K s1),(p2,K s2)] | (s1,s2) <- cuts]
      return (concat matches)
 
--- | Estimate the minimal length of the string that a pattern will match
-minLength = matchLength 0 id (+) min -- safe underestimate
-
--- | Estimate the maximal length of the string that a pattern will match
-maxLength =
-    maybe maxBound id . matchLength Nothing Just (liftM2 (+)) (liftM2 max)
-        -- safe overestimate
-
-matchLength unknown known seq alt = len
-  where
-    len p =
-      case p of
-        PString s  -> known (length s)
-        PSeq p1 p2 -> seq (len p1) (len p2)
-        PAlt p1 p2 -> alt (len p1) (len p2)
-        PChar      -> known 1
-        PChars _   -> known 1
-        PAs x p'   -> len p'
-        PT t p'    -> len p'
-        _          -> unknown
-
-lengthBounds p = (minLength p,maxLength p)
-
-mPatt p = (lengthBounds p,measurePatt p)
-
-measurePatt p =
-  case p of
-    PSeq p1 p2 -> PMSeq (mPatt p1) (mPatt p2)
-    _ -> composSafePattOp measurePatt p
-
-
 isInConstantForm :: Term -> Bool
 isInConstantForm trm = case trm of
     Cn _     -> True

src/compiler/GF/Grammar/Predef.hs

@@ -61,7 +61,6 @@ cRead = identS "read"
 cToStr = identS "toStr"
 cMapStr = identS "mapStr"
 cError = identS "error"
-cTrace = identS "trace"
 
 -- * Hacks: dummy identifiers used in various places.
 -- Not very nice!