From 59aff851a7b6aca4ba9fac8b6fc528e3b4f8fe15 Mon Sep 17 00:00:00 2001
From: aarne <aarne@cs.chalmers.se>
Date: Tue, 10 Jun 2008 16:00:42 +0000
Subject: [PATCH] preliminary version of morpho analysis

---
 src-3.0/GF/Command/Commands.hs | 34 ++++++++++++++++++++++---
 src-3.0/PGF/Morphology.hs      | 46 ++++++++++++++++++++++++++++++++++
 src-3.0/PGF/ShowLinearize.hs   |  1 +
 3 files changed, 78 insertions(+), 3 deletions(-)
 create mode 100644 src-3.0/PGF/Morphology.hs

diff --git a/src-3.0/GF/Command/Commands.hs b/src-3.0/GF/Command/Commands.hs
index 6aa19dfba..292e802b9 100644
--- a/src-3.0/GF/Command/Commands.hs
+++ b/src-3.0/GF/Command/Commands.hs
@@ -14,6 +14,7 @@ import PGF.CId
 import PGF.ShowLinearize
 import PGF.Macros
 import PGF.Data ----
+import PGF.Morphology
 import GF.Compile.Export
 import GF.Infra.UseIO
 import GF.Data.ErrM ----
@@ -60,9 +61,9 @@ commandHelp full (co,info) = unlines $ [
   "flags: " ++ unwords (flags info)
   ] else []
 
--- this list must be kept sorted by the command name!
+-- this list must no more be kept sorted by the command name
 allCommands :: PGF -> Map.Map String CommandInfo
-allCommands pgf = Map.fromAscList [
+allCommands pgf = Map.fromList [
   ("cc", emptyCommandInfo {
      longname = "compute_concrete",
      synopsis = "computes concrete syntax term using the source grammar",
@@ -144,15 +145,39 @@ allCommands pgf = Map.fromAscList [
      options = ["all","record","table","term"],
      flags = ["lang"]
      }),
+
+  ("ma", emptyCommandInfo {
+     longname = "morpho_analyse",
+     synopsis = "print the morphological analyses of all words in the string",
+     explanation = unlines [
+       "Prints all the analyses of space-separated words in the input string,",
+       "using the morphological analyser of the actual grammar (see command pf)"
+       ],
+     exec  = \opts ->  
+               return . fromString . unlines . 
+               map prMorphoAnalysis . concatMap (morphos opts) . 
+               concatMap words . toStrings
+     }),
+
   ("p", emptyCommandInfo {
      longname = "parse",
      synopsis = "parse a string to abstract syntax expression",
      explanation = "Shows all trees (expressions) returned for String by the actual\n"++
                    "grammar (overridden by the -lang flag), in the category S (overridden\n"++
                    "by the -cat flag).",
-     exec = \opts -> return . fromTrees . concatMap (par opts). toStrings,
+     exec = \opts -> return . fromTrees . concatMap (par opts) . toStrings,
      flags = ["cat","lang"]
      }),
+  ("pf", emptyCommandInfo {
+     longname = "print_fullform",
+     synopsis = "print the full-form lexicon of the actual grammar",
+     explanation = unlines [
+       "Prints all the strings in the actual grammar with their possible analyses"
+       ],
+     exec  = \opts _ -> 
+               return $ fromString $ concatMap 
+                  (prFullFormLexicon . buildMorpho pgf . mkCId) $ optLangs opts
+     }),
   ("pg", emptyCommandInfo {
      longname = "print_grammar",
      synopsis = "print the actual grammar with the given printer",
@@ -206,3 +231,6 @@ allCommands pgf = Map.fromAscList [
    prGrammar opts = case valIdOpts "printer" "" opts of
      "cats" -> unwords $ categories pgf
      v -> prPGF (read v) pgf (prCId (absname pgf))
+
+   morphos opts s = 
+     [lookupMorpho (buildMorpho pgf (mkCId la)) s | la <- optLangs opts]
diff --git a/src-3.0/PGF/Morphology.hs b/src-3.0/PGF/Morphology.hs
new file mode 100644
index 000000000..97def6b9a
--- /dev/null
+++ b/src-3.0/PGF/Morphology.hs
@@ -0,0 +1,46 @@
+module PGF.Morphology where
+
+import PGF.ShowLinearize
+import PGF.Data
+import PGF.CId
+
+import qualified Data.Map as Map
+import Data.List (intersperse)
+
+-- these 4 definitions depend on the datastructure used
+
+type Morpho = Map.Map String [(Lemma,Analysis)]
+
+lookupMorpho :: Morpho -> String -> [(Lemma,Analysis)]
+lookupMorpho mo s = maybe noAnalysis id $ Map.lookup s mo
+
+buildMorpho :: PGF -> CId -> Morpho
+buildMorpho pgf = Map.fromListWith (++) . collectWords pgf
+
+prFullFormLexicon :: Morpho -> String
+prFullFormLexicon mo = 
+  unlines [w ++ " : " ++ prMorphoAnalysis ts | (w,ts) <- Map.assocs mo]
+
+prMorphoAnalysis :: [(Lemma,Analysis)] -> String
+prMorphoAnalysis lps = unlines [l ++ " " ++ p | (l,p) <- lps]
+
+type Lemma = String
+type Analysis = String
+
+noAnalysis :: [(Lemma,Analysis)]
+noAnalysis = []
+
+collectWords :: PGF -> CId -> [(String, [(Lemma,Analysis)])]
+collectWords pgf lang = 
+    concatMap collOne 
+      [(f,c,0) | (f,(DTyp [] c _,_)) <- Map.toList $ funs $ abstract pgf] 
+  where
+    collOne (f,c,i) = 
+      fromRec f [prCId c] (recLinearize pgf lang (EApp f (replicate i (EMeta 888))))
+    fromRec f v r = case r of
+      RR  rs -> concat [fromRec f v t | (_,t) <- rs] 
+      RT  rs -> concat [fromRec f (p:v) t | (p,t) <- rs]
+      RFV rs -> concatMap (fromRec f v) rs
+      RS  s  -> [(s,[(prCId f,unwords (reverse v))])]
+      RCon c -> [] ---- inherent
+
diff --git a/src-3.0/PGF/ShowLinearize.hs b/src-3.0/PGF/ShowLinearize.hs
index a1c1e476a..98a0806ba 100644
--- a/src-3.0/PGF/ShowLinearize.hs
+++ b/src-3.0/PGF/ShowLinearize.hs
@@ -1,4 +1,5 @@
 module PGF.ShowLinearize (
+  Record (..), recLinearize, --- used in PGF.Morphology
   tableLinearize,
   recordLinearize,
   termLinearize,