modernized parser in EmbedAPI

2 files changed, 26 insertions, 87 deletions

diff --git a/src/GF/Embed/EmbedAPI.hs b/src/GF/Embed/EmbedAPI.hs
index a3a130a36..43e4f2546 100644
--- a/src/GF/Embed/EmbedAPI.hs
+++ b/src/GF/Embed/EmbedAPI.hs
@@ -15,7 +15,7 @@
 
 module GF.Embed.EmbedAPI where
 
-import GF.Compile.ShellState (ShellState,grammar2shellState,canModules,stateGrammarOfLang,abstract,grammar,firstStateGrammar,allLanguages,allCategories,stateOptions)
+import GF.Compile.ShellState (ShellState,grammar2shellState,canModules,stateGrammarOfLang,abstract,grammar,firstStateGrammar,allLanguages,allCategories,stateOptions,firstAbsCat)
 import GF.UseGrammar.Linear (linTree2string)
 import GF.UseGrammar.GetTree (string2tree)
 import GF.Embed.EmbedParsing (parseString)
@@ -53,8 +53,11 @@ file2grammar :: FilePath -> IO MultiGrammar
 linearize    :: MultiGrammar -> Language -> Tree -> String
 parse        :: MultiGrammar -> Language -> Category -> String -> [Tree]
 
-linearizeAll :: MultiGrammar             -> Tree -> [String]
-parseAll     :: MultiGrammar             -> Category -> String -> [[Tree]]
+linearizeAll     :: MultiGrammar -> Tree -> [String]
+linearizeAllLang :: MultiGrammar -> Tree -> [(Language,String)]
+
+parseAll     :: MultiGrammar -> Category -> String -> [[Tree]]
+parseAllLang :: MultiGrammar -> Category -> String -> [(Language,[Tree])]
 
 readTree     :: MultiGrammar -> String -> Tree
 showTree     ::                 Tree -> String
@@ -62,6 +65,8 @@ showTree     ::                 Tree -> String
 languages    :: MultiGrammar -> [Language]
 categories   :: MultiGrammar -> [Category]
 
+startCat :: MultiGrammar -> Category
+
 ---------------------------------------------------
 -- Implementation
 ---------------------------------------------------
@@ -84,15 +89,19 @@ linearize mgr lang =
 parse mgr lang cat = 
   map tree2exp . 
   errVal [] . 
-  parseString noOptions sgr cfcat
+  parseString (stateOptions sgr) sgr cfcat
  where
    sgr   = stateGrammarOfLang mgr (zIdent lang)
    cfcat = string2CFCat abs cat
    abs   = maybe (error "no abstract syntax") prIdent $ abstract mgr
 
-linearizeAll mgr t = [linearize mgr lang t | lang <- languages mgr]
+linearizeAll mgr = map snd . linearizeAllLang mgr
+linearizeAllLang mgr t = [(lang,linearize mgr lang t) | lang <- languages mgr]
 
-parseAll mgr cat s = [parse mgr lang cat s | lang <- languages mgr]
+parseAll mgr cat = map snd . parseAllLang mgr cat
+
+parseAllLang mgr cat s = 
+  [(lang,ts) | lang <- languages mgr, let ts = parse mgr lang cat s, not (null ts)]
 
 readTree mgr s = tree2exp $ string2tree (firstStateGrammar mgr) s
 
@@ -101,3 +110,5 @@ showTree t  = prt_ t
 languages mgr = [prt_ l | l <- allLanguages mgr]
 
 categories mgr = [prt_ c | (_,c) <- allCategories mgr]
+
+startCat = prt_ . snd . firstAbsCat noOptions . firstStateGrammar
diff --git a/src/GF/Embed/EmbedParsing.hs b/src/GF/Embed/EmbedParsing.hs
index 40378c491..43909f355 100644
--- a/src/GF/Embed/EmbedParsing.hs
+++ b/src/GF/Embed/EmbedParsing.hs
@@ -33,8 +33,7 @@ import GF.Infra.Option
 import GF.Compile.ShellState
 import GF.Embed.EmbedCustom
 import GF.CF.PPrCF (prCFTree)
-
-import qualified GF.OldParsing.ParseCF as PCFOld -- OBSOLETE
+import qualified GF.Parsing.GFC as New
 
 
 -- import qualified GF.Parsing.GFC as New
@@ -55,83 +54,12 @@ parseStringMsg os sg cat s = do
   return (ts,unlines ss)
 
 parseStringC :: Options -> StateGrammar -> CFCat -> String -> Check [Tree]
-parseStringC opts0 sg cat s
-
- | otherwise = do
-  let opts = unionOptions opts0 $ stateOptions sg
-      cf  = stateCF sg
-      gr  = stateGrammarST sg
-      cn  = cncId sg
-      tok = customOrDefault opts useTokenizer customTokenizer sg
-      parser = PCFOld.parse "ibn" (stateCF sg) cat -- customOrDefault opts useParser customParser sg cat
-  tokens2trms opts sg cn parser (tok s)
-
-tokens2trms :: Options ->StateGrammar ->Ident -> CFParser -> [CFTok] -> Check [Tree]
-tokens2trms opts sg cn parser toks = trees2trms opts sg cn toks trees info
-    where result = parser toks
-	  info   = snd result
-	  trees  = {- nub $ -} cfParseResults result -- peb 25/5-04: removed nub (O(n^2))
-
-trees2trms :: Options -> StateGrammar -> Ident -> [CFTok] -> [CFTree] -> String -> Check [Tree]
-trees2trms opts sg cn as ts0 info = do
-  ts  <- case () of
-    _ | null ts0 -> checkWarn "No success in cf parsing" >> return []
-    _ | raw      -> do
-      ts1 <- return (map cf2trm0 ts0) ----- should not need annot
-      checks [
-         mapM (checkErr . (annotate gr) . trExp) ts1 ---- complicated, often fails
-        ,checkWarn (unlines ("Raw CF trees:":(map prCFTree ts0))) >> return []
-        ]
-    _ -> do
-      let num = optIntOrN opts flagRawtrees 999999
-      let (ts01,rest) = splitAt num ts0
-      if null rest then return () 
-         else checkWarn ("Warning: only" +++ show num +++ "raw parses out of" +++ 
-                          show (length ts0) +++ 
-                          "considered; use -rawtrees=<Int> to see more"
-                     )
-      (ts1,ss) <- checkErr $ mapErrN 1 postParse ts01
-      if null ts1 then raise ss else return ()
-      ts2 <- mapM (checkErr . annotate gr . refreshMetas [] . trExp) ts1 ---- 
-      if forgive then return ts2 else do
-        let tsss = [(t, allLinsOfTree gr cn t) | t <- ts2]
-            ps = [t | (t,ss) <- tsss, 
-                      any (compatToks as) (map str2cftoks ss)]
-        if null ps 
-           then raise $ "Failure in morphology." ++
-                  if verb 
-                     then "\nPossible corrections: " +++++
-                          unlines (nub (map sstr (concatMap snd tsss)))
-                     else ""
-           else return ps
-
-  if verb 
-     then checkWarn $ " the token list" +++ show as ++++ unknown as +++++ info
-     else return ()
-
-  return $ optIntOrAll opts flagNumber $ nub ts
- where
-   gr  = stateGrammarST sg
-
-   raw     = oElem rawParse opts
-   verb    = oElem beVerbose opts
-   forgive = oElem forgiveParse opts
-
-   unknown ts = case filter noMatch [t | t@(TS _) <- ts] of
-     [] -> "where all words are known"
-     us -> "with the unknown tokens" +++ show us --- needs to be fixed for literals
-   terminals = map TS $ stateGrammarWords sg
-   noMatch t = all (not . compatTok t) terminals 
-     
-
---- too much type checking in building term info? return FullTerm to save work?
-
--- | raw parsing: so simple it is for a context-free CF grammar
-cf2trm0 :: CFTree -> C.Exp
-cf2trm0 (CFTree (fun, (_, trees))) = mkAppAtom (cffun2trm fun) (map cf2trm0 trees)
- where
-   cffun2trm (CFFun (fun,_)) = fun
-   mkApp = foldl C.EApp
-   mkAppAtom a = mkApp (C.EAtom a)
-
+parseStringC opts0 sg cat s = do
+  let opts      = unionOptions opts0 $ stateOptions sg
+      algorithm = "f" -- default algorithm: FCFG
+      strategy  = "bottomup" 
+      tokenizer = customOrDefault opts useTokenizer customTokenizer sg
+      toks = tokenizer s 
+  ts  <- checkErr $ New.parse algorithm strategy (pInfo sg) (absId sg) cat toks
+  checkErr $ allChecks $ map (annotate (stateGrammarST sg) . refreshMetas []) ts