From 60c2d054ee2e34014b081f331ced46b102d2fdbc Mon Sep 17 00:00:00 2001
From: krasimir <krasimir@chalmers.se>
Date: Fri, 27 Jun 2008 10:33:46 +0000
Subject: [PATCH] the shell now supports cp1251 coding with se command. works
 with the word completion as well

---
 src/GF/Compile/Coding.hs |  7 +------
 src/GF/Text/CP1251.hs    | 18 ++++++++++++++++++
 src/GF/Text/Lexing.hs    | 16 +---------------
 src/GFI.hs               | 26 ++++++++++++++++----------
 4 files changed, 36 insertions(+), 31 deletions(-)
 create mode 100644 src/GF/Text/CP1251.hs

diff --git a/src/GF/Compile/Coding.hs b/src/GF/Compile/Coding.hs
index 9d271964f..a226265c5 100644
--- a/src/GF/Compile/Coding.hs
+++ b/src/GF/Compile/Coding.hs
@@ -3,6 +3,7 @@ module GF.Compile.Coding where
 import GF.Grammar.Grammar
 import GF.Grammar.Macros
 import GF.Text.UTF8
+import GF.Text.CP1251
 import GF.Infra.Modules
 import GF.Infra.Option
 import GF.Data.Operations
@@ -47,9 +48,3 @@ codeSourceModule co (id,moi) = case moi of
       PSeq p q -> PSeq (codp p) (codp q)
       PAlt p q -> PAlt (codp p) (codp q)
       _ -> p
-
----- from Text.Lexing
-decodeCP1251 = map convert where
-  convert c
-   | c >= '\192' && c <= '\255' = chr (ord c + 848)
-   | otherwise                  = c
diff --git a/src/GF/Text/CP1251.hs b/src/GF/Text/CP1251.hs
new file mode 100644
index 000000000..af5c829a2
--- /dev/null
+++ b/src/GF/Text/CP1251.hs
@@ -0,0 +1,18 @@
+module GF.Text.CP1251 where
+
+import Data.Char
+
+decodeCP1251 = map convert where
+  convert c
+   | c >= '\192' && c <= '\255' = chr (ord c + 848)
+   | c == '\168'                = chr 1025      -- cyrillic capital letter lo
+   | c == '\184'                = chr 1105      -- cyrillic small   letter lo
+   | otherwise                  = c
+
+encodeCP1251 = map convert where
+  convert c
+   | oc >= 1040 && oc <= 1103 = chr (oc - 848)
+   | oc == 1025               = chr 168         -- cyrillic capital letter lo
+   | oc == 1105               = chr 184         -- cyrillic small   letter lo
+   | otherwise                = c
+   where oc = ord c
diff --git a/src/GF/Text/Lexing.hs b/src/GF/Text/Lexing.hs
index 2c6b417b8..38018ff54 100644
--- a/src/GF/Text/Lexing.hs
+++ b/src/GF/Text/Lexing.hs
@@ -2,6 +2,7 @@ module GF.Text.Lexing (stringOp) where
 
 import GF.Text.Transliterations
 import GF.Text.UTF8
+import GF.Text.CP1251
 
 import Data.Char
 import Data.List (intersperse)
@@ -98,18 +99,3 @@ unlexMixed = concat . alternate False where
 isPunct = flip elem ".?!,:;"
 isParen = flip elem "()[]{}"
 isClosing = flip elem ")]}"
- 
-
--- might be in a file of its own: Windows Cyrillic, used in Bulgarian resource
-
-decodeCP1251 = map convert where
-  convert c
-   | c >= '\192' && c <= '\255' = chr (ord c + 848)
-   | otherwise                  = c
-
-encodeCP1251 = map convert where
-  convert c
-   | oc >= 1040 && oc <= 1103 = chr (oc - 848)
-   | otherwise                  = c
-   where oc = ord c
-
diff --git a/src/GFI.hs b/src/GFI.hs
index 18fc4143f..65daf5675 100644
--- a/src/GFI.hs
+++ b/src/GFI.hs
@@ -12,6 +12,7 @@ import GF.Infra.Option
 import GF.System.Readline
 
 import GF.Text.UTF8 ----
+import GF.Text.CP1251
 
 import PGF
 import PGF.Data
@@ -42,7 +43,7 @@ loop :: Options -> GFEnv -> IO GFEnv
 loop opts gfenv0 = do
   let env = commandenv gfenv0
   let sgr = sourcegrammar gfenv0
-  setCompletionFunction (Just (wordCompletion (commandenv gfenv0)))
+  setCompletionFunction (Just (wordCompletion gfenv0))
   s0 <- fetchCommand (prompt env)
   let gfenv = gfenv0 {history = s0 : history gfenv0}
   let loopNewCPU gfenv' = do 
@@ -111,7 +112,7 @@ loop opts gfenv0 = do
 
           "ph":_ -> 
             mapM_ (putStrLn . enc) (reverse (history gfenv0)) >> loopNewCPU gfenv
-          "se":c -> loopNewCPU $ gfenv {coding = s}
+          "se":c:_ -> loopNewCPU $ gfenv {coding = c}
 
   -- ordinary commands, working on CommandEnv
           _ -> do
@@ -169,16 +170,17 @@ emptyGFEnv :: GFEnv
 emptyGFEnv = GFEnv emptyGrammar (mkCommandEnv encodeUTF8 emptyPGF) [] 0 "utf8"
 
 encode env = case coding env of
-  "utf8" -> encodeUTF8
-  _ -> id
+  "utf8"   -> encodeUTF8
+  "cp1251" -> encodeCP1251
+  _        -> id
 
 decode env = case coding env of
-  "utf8" -> decodeUTF8
-  _ -> id
+  "utf8"   -> decodeUTF8
+  "cp1251" -> decodeCP1251
+  _        -> id
 
 
-
-wordCompletion cmdEnv line prefix p =
+wordCompletion gfenv line0 prefix0 p =
   case wc_type (take p line) of
     CmplCmd pref
       -> ret ' ' [name | name <- Map.keys (commands cmdEnv), isPrefixOf pref name]
@@ -188,7 +190,7 @@ wordCompletion cmdEnv line prefix p =
               Right state0 -> let ws     = words (take (length s - length prefix) s)
                                   state  = foldl nextState state0 ws
                                   compls = getCompletions state prefix
-                              in ret ' ' (Map.keys compls)
+                              in ret ' ' (map (encode gfenv) (Map.keys compls))
               Left  _      -> ret ' ' []
     CmplOpt (Just (Command n _ _)) pref
       -> case Map.lookup n (commands cmdEnv) of
@@ -206,7 +208,11 @@ wordCompletion cmdEnv line prefix p =
               Left  _   -> ret ' ' []
     _ -> ret ' ' []
   where
-    pgf = multigrammar cmdEnv
+    line   = decode gfenv line0
+    prefix = decode gfenv prefix0
+
+    pgf    = multigrammar cmdEnv
+    cmdEnv = commandenv gfenv
     optLang opts = valIdOpts "lang" (head (languages pgf)) opts
     optCat  opts = valIdOpts "cat"  (lookStartCat pgf)     opts