printing GSyntax with GFCC trees

4 files changed, 254 insertions, 35 deletions

diff --git a/src/GF/Canon/GFCC/FCFGParsing.hs b/src/GF/Canon/GFCC/FCFGParsing.hs
index dfedc6622..eeea653d2 100644
--- a/src/GF/Canon/GFCC/FCFGParsing.hs
+++ b/src/GF/Canon/GFCC/FCFGParsing.hs
@@ -118,13 +118,11 @@ cnv_forests2 (FFloat  x)   = FFloat  x
 
 tree2term :: SyntaxTree Fun -> Exp
 tree2term (TNode f ts) = Tr (AC f) (map tree2term ts)
-{- ----
-tree2term (TString  s) = Macros.string2term s
-tree2term (TInt     n) = Macros.int2term    n
-tree2term (TFloat   f) = Macros.float2term  f
-tree2term (TMeta)      = Macros.mkMeta 0
--}
 
+tree2term (TString  s) = Tr (AS s) []
+tree2term (TInt     n) = Tr (AI n) []
+tree2term (TFloat   f) = Tr (AF f) []
+tree2term (TMeta)      = Tr AM []
 
 ----------------------------------------------------------------------
 -- conversion and unification of forests
diff --git a/src/GF/Canon/GFCC/GFCCAPI.hs b/src/GF/Canon/GFCC/GFCCAPI.hs
index 211f9f67b..9c3978442 100644
--- a/src/GF/Canon/GFCC/GFCCAPI.hs
+++ b/src/GF/Canon/GFCC/GFCCAPI.hs
@@ -22,7 +22,8 @@ import GF.Canon.GFCC.ParGFCC
 import GF.Canon.GFCC.PrintGFCC
 import GF.Canon.GFCC.ErrM
 import GF.Canon.GFCC.FCFGParsing
-import GF.Conversion.SimpleToFCFG (convertGrammarCId,FCat(..))
+import qualified GF.Canon.GFCC.GenGFCC as G
+import GF.Conversion.SimpleToFCFG (convertGrammarCId,FCat(..)) ----
 
 --import GF.Data.Operations
 --import GF.Infra.UseIO
@@ -52,8 +53,11 @@ parse        :: MultiGrammar -> Language -> Category -> String -> [Tree]
 linearizeAll     :: MultiGrammar -> Tree -> [String]
 linearizeAllLang :: MultiGrammar -> Tree -> [(Language,String)]
 
---parseAll     :: MultiGrammar -> Category -> String -> [[Tree]]
---parseAllLang :: MultiGrammar -> Category -> String -> [(Language,[Tree])]
+parseAll     :: MultiGrammar -> Category -> String -> [[Tree]]
+parseAllLang :: MultiGrammar -> Category -> String -> [(Language,[Tree])]
+
+generateAll    :: MultiGrammar -> Category -> [Tree]
+generateRandom :: MultiGrammar -> Category -> IO [Tree]
 
 readTree   :: MultiGrammar -> String -> Tree
 showTree   ::                 Tree -> String
@@ -80,26 +84,20 @@ linearize mgr lang = GF.Canon.GFCC.DataGFCC.linearize (gfcc mgr) (CId lang)
 parse mgr lang cat s = 
   err error id $ parserLang (gfcc mgr) (CId lang) (CId cat) (words s)
 
-{-
-  map tree2exp . 
-  errVal [] . 
-  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 = map snd . linearizeAllLang mgr
 linearizeAllLang mgr t = 
   [(lang,linearThis mgr lang t) | 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)]
--}
+
+generateRandom mgr cat = do
+  gen <- newStdGen
+  return $ G.generateRandom gen (gfcc mgr) (CId cat)
+
+generateAll mgr cat = G.generate (gfcc mgr) (CId cat)
 
 readTree _ = err (const exp0) id . (pExp . myLexer)
 
diff --git a/src/GF/Canon/GFCC/GFCCToHaskell.hs b/src/GF/Canon/GFCC/GFCCToHaskell.hs
new file mode 100644
index 000000000..890c1a76f
--- /dev/null
+++ b/src/GF/Canon/GFCC/GFCCToHaskell.hs
@@ -0,0 +1,212 @@
+----------------------------------------------------------------------
+-- |
+-- Module      : GrammarToHaskell
+-- Maintainer  : Aarne Ranta
+-- Stability   : (stable)
+-- Portability : (portable)
+--
+-- > CVS $Date: 2005/06/17 12:39:07 $ 
+-- > CVS $Author: bringert $
+-- > CVS $Revision: 1.8 $
+--
+-- to write a GF abstract grammar into a Haskell module with translations from
+-- data objects into GF trees. Example: GSyntax for Agda.
+-- AR 11/11/1999 -- 7/12/2000 -- 18/5/2004
+-----------------------------------------------------------------------------
+
+module GF.Canon.GFCC.GFCCToHaskell (grammar2haskell, grammar2haskellGADT) where
+
+import GF.Canon.GFCC.AbsGFCC
+import GF.Canon.GFCC.DataGFCC
+import GF.Data.Operations
+
+import Data.List --(isPrefixOf, find, intersperse)
+import qualified Data.Map as Map
+
+-- | the main function
+grammar2haskell :: GFCC -> String
+grammar2haskell gr = foldr (++++) [] $  
+  haskPreamble ++ [datatypes gr', gfinstances gr', fginstances gr']
+    where gr' = hSkeleton gr
+
+grammar2haskellGADT :: GFCC -> String
+grammar2haskellGADT gr = foldr (++++) [] $  
+  ["{-# OPTIONS_GHC -fglasgow-exts #-}"] ++ 
+  haskPreamble ++ [datatypesGADT gr', gfinstances gr', fginstances gr']
+    where gr' = hSkeleton gr
+
+-- | by this you can prefix all identifiers with stg; the default is 'G'
+gId :: OIdent -> OIdent 
+gId i = 'G':i
+
+haskPreamble =
+ [
+  "module GSyntax where",
+  "",
+  "import GF.Canon.GFCC.AbsGFCC",
+  "import GF.Canon.GFCC.DataGFCC",
+  "import GF.Data.Operations",
+  "----------------------------------------------------",
+  "-- automatic translation from GF to Haskell",
+  "----------------------------------------------------",
+  "", 
+  "class Gf a where gf :: a -> Exp",
+  "class Fg a where fg :: Exp -> a",
+  "",
+  predefInst "GString" "String" "Tr (AS s) []",
+  "",
+  predefInst "GInt" "Integer" "Tr (AI s) []",
+  "",
+  predefInst "GFloat" "Double" "Tr (AF s) []",
+  "",
+  "----------------------------------------------------",
+  "-- below this line machine-generated",
+  "----------------------------------------------------",
+  ""
+ ]
+
+predefInst gtyp typ patt = 
+  "newtype" +++ gtyp +++ "=" +++ gtyp +++ typ +++ " deriving Show" +++++
+  "instance Gf" +++ gtyp +++ "where" ++++
+  "  gf (" ++ gtyp +++ "s) =" +++ patt +++++
+  "instance Fg" +++ gtyp +++ "where" ++++
+  "  fg t =" ++++
+  "    case t of" ++++
+  "     " +++ patt +++ " ->" +++ gtyp +++ "s" ++++
+  "      _ -> error (\"no" +++ gtyp +++ "\" ++ show t)"
+
+type OIdent = String
+
+type HSkeleton = [(OIdent, [(OIdent, [OIdent])])]
+
+datatypes, gfinstances, fginstances :: (String,HSkeleton) -> String
+datatypes   = (foldr (+++++) "") . (filter (/="")) . (map hDatatype) . snd
+gfinstances (m,g) = (foldr (+++++) "") $ (filter (/="")) $ (map (hInstance m)) g
+fginstances (m,g) = (foldr (+++++) "") $ (filter (/="")) $ (map (fInstance m)) g
+
+hDatatype :: (OIdent, [(OIdent, [OIdent])]) -> String
+hInstance, fInstance :: String -> (OIdent, [(OIdent, [OIdent])]) -> String
+
+hDatatype ("Cn",_) = "" ---
+hDatatype (cat,[]) = ""
+hDatatype (cat,rules) | isListCat (cat,rules) = 
+ "newtype" +++ gId cat +++ "=" +++ gId cat +++ "[" ++ gId (elemCat cat) ++ "]" 
+  +++ "deriving Show"
+hDatatype (cat,rules) =
+ "data" +++ gId cat +++ "=" ++
+ (if length rules == 1 then "" else "\n  ") +++
+ foldr1 (\x y -> x ++ "\n |" +++ y) 
+        [gId f +++ foldr (+++) "" (map gId xx) | (f,xx) <- rules] ++++
+ "  deriving Show"
+
+-- GADT version of data types
+datatypesGADT :: (String,HSkeleton) -> String
+datatypesGADT (_,skel) = 
+    unlines (concatMap hCatTypeGADT skel)
+    +++++
+    "data Tree :: * -> * where" ++++ unlines (concatMap (map ("  "++) . hDatatypeGADT) skel)
+
+hCatTypeGADT :: (OIdent, [(OIdent, [OIdent])]) -> [String]
+hCatTypeGADT (cat,rules)
+    = ["type"+++gId cat+++"="+++"Tree"+++gId cat++"_",
+       "data"+++gId cat++"_"]
+
+hDatatypeGADT :: (OIdent, [(OIdent, [OIdent])]) -> [String]
+hDatatypeGADT (cat, rules) 
+    | isListCat (cat,rules) = [gId cat+++"::"+++"["++gId (elemCat cat)++"]" +++ "->" +++ t]
+    | otherwise =
+        [ gId f +++ "::" +++ concatMap (\a -> gId a +++ "-> ") args ++ t | (f,args) <- rules ]
+  where t = "Tree" +++ gId cat ++ "_"
+
+
+----hInstance m ("Cn",_) = "" --- seems to belong to an old applic. AR 18/5/2004
+hInstance m (cat,[]) = "" 
+hInstance m (cat,rules) 
+ | isListCat (cat,rules) =
+  "instance Gf" +++ gId cat +++ "where" ++++
+     " gf (" ++ gId cat +++ "[" ++ concat (intersperse "," baseVars) ++ "])" 
+           +++ "=" +++ mkRHS ("Base"++ec) baseVars ++++
+     " 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)" 
+ | otherwise =
+  "instance Gf" +++ gId cat +++ "where" ++
+  (if length rules == 1 then "" else "\n") +++
+  foldr1 (\x y -> x ++ "\n" +++ y) [mkInst f xx | (f,xx) <- rules]
+ where
+   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')) +++
+     "=" +++ mkRHS f xx'
+   mkVars n = ["x" ++ show i | i <- [1..n]]
+   mkRHS f vars = "Tr (AC (CId \"" ++ f ++ "\"))" +++ 
+		   "[" ++ prTList ", " ["gf" +++ x | x <- vars] ++ "]"
+
+
+----fInstance m ("Cn",_) = "" ---
+fInstance m (cat,[]) = ""
+fInstance m (cat,rules) =
+  "instance Fg" +++ gId cat +++ "where" ++++
+  " fg t =" ++++
+  "  case t of" ++++
+  foldr1 (\x y -> x ++ "\n" ++ y) [mkInst f xx | (f,xx) <- rules] ++++
+  "    _ -> error (\"no" +++ cat ++ " \" ++ show t)"
+   where
+    mkInst f xx =
+     "    Tr (AC (CId \"" ++ f ++ "\")) " ++
+     "[" ++ prTList "," xx' ++ "]" +++
+     "->" +++ mkRHS f xx'
+       where xx' = ["x" ++ show i | (_,i) <- zip xx [1..]]
+	     mkRHS f vars 
+		 | isListCat (cat,rules) =
+		     if "Base" `isPrefixOf` f then
+			gId cat +++ "[" ++ prTList ", " [ "fg" +++ x | x <- vars ] ++ "]"
+		      else
+                       let (i,t) = (init vars,last vars)
+                        in "let" +++ gId cat +++ "xs = fg " ++ t +++ "in" +++ 
+                          gId cat +++ prParenth (prTList ":" (["fg"+++v | v <- i] ++ ["xs"]))
+		 | otherwise = 
+		     gId f +++  
+		     prTList " " [prParenth ("fg" +++ x) | x <- vars]
+
+
+--type HSkeleton = [(OIdent, [(OIdent, [OIdent])])]
+hSkeleton :: GFCC -> (String,HSkeleton)
+hSkeleton gr = 
+  (pr (absname gr), 
+   [(pr c, [(pr f, map pr cs) | (f, Typ cs _) <- fs]) | 
+                                        fs@((_, Typ _ c):_) <- fs]
+  )
+ where
+   fs = groupBy valtypg (sortBy valtyps (Map.assocs (funs (abstract gr))))
+   valtyps (_, Typ _ x) (_, Typ _ y) = compare x y
+   valtypg (_, Typ _ x) (_, Typ _ y) = x == y
+   pr (CId c) = c
+
+updateSkeleton :: OIdent -> HSkeleton -> (OIdent, [OIdent]) -> HSkeleton
+updateSkeleton cat skel rule =
+ case skel of
+   (cat0,rules):rr | cat0 == cat -> (cat0, rule:rules) : rr
+   (cat0,rules):rr               -> (cat0, rules) : updateSkeleton cat rr 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
+
+-- | Gets the element category of a list category.
+elemCat :: OIdent -> OIdent
+elemCat = drop 4
+
+isBaseFun :: OIdent -> Bool
+isBaseFun f = "Base" `isPrefixOf` f
+
+isConsFun :: OIdent -> Bool
+isConsFun f = "Cons" `isPrefixOf` f
+
+baseSize :: (OIdent, [(OIdent, [OIdent])]) -> Int
+baseSize (_,rules) = length bs
+    where Just (_,bs) = find (("Base" `isPrefixOf`) . fst) rules
diff --git a/src/GF/Canon/GFCC/Shell.hs b/src/GF/Canon/GFCC/Shell.hs
index 2bee4a300..5285b89a8 100644
--- a/src/GF/Canon/GFCC/Shell.hs
+++ b/src/GF/Canon/GFCC/Shell.hs
@@ -5,7 +5,7 @@ import qualified GF.Canon.GFCC.GenGFCC as G ---
 import GF.Canon.GFCC.AbsGFCC (CId(CId)) ---
 import System.Random (newStdGen)
 import System (getArgs)
-
+import Data.Char (isDigit)
 
 -- Simple translation application built on GFCC. AR 7/9/2006 -- 19/9/2007
 
@@ -13,29 +13,37 @@ main :: IO ()
 main = do
   file:_  <- getArgs
   grammar <- file2grammar file
-  putStrLn $ "languages:  " ++ unwords (languages grammar)
-  putStrLn $ "categories: " ++ unwords (categories grammar)
+  printHelp grammar
   loop grammar
 
 loop :: MultiGrammar -> IO ()
 loop grammar = do
   s <- getLine
-  if s == "quit" then return () else do
+  if s == "q" then return () else do
     treat grammar s
     loop grammar
 
+printHelp grammar = do
+  putStrLn $ "languages:  " ++ unwords (languages grammar)
+  putStrLn $ "categories: " ++ unwords (categories grammar)
+  putStrLn commands
+
+
+commands = unlines [
+  "Commands:",
+  "  (gt | gtt | gr | grt) Cat Num - generate all or random",
+  "  p Lang Cat String             - parse (unquoted) string",
+  "  l Tree                        - linearize in all languages",
+  "  h                             - help",
+  "  q                             - quit"
+  ]
+
 treat :: MultiGrammar -> String -> IO ()
 treat mgr s = case words s of
-  "gt":cat:n:_ -> do
-    mapM_ prlinonly $ take (read n) $ G.generate grammar (CId cat)
-  "gtt":cat:n:_ -> do
-    mapM_ prlin $ take (read n) $ G.generate grammar (CId cat)
-  "gr":cat:n:_ -> do
-    gen <- newStdGen
-    mapM_ prlinonly $ take (read n) $ G.generateRandom gen grammar (CId cat)
-  "grt":cat:n:_ -> do
-    gen <- newStdGen
-    mapM_ prlin $ take (read n) $ G.generateRandom gen grammar (CId cat)
+  "gt" :cat:n:_ -> mapM_ prlinonly $ take (read1 n) $ generateAll mgr cat
+  "gtt":cat:n:_ -> mapM_ prlin $ generateAll mgr cat
+  "gr" :cat:n:_ -> generateRandom mgr cat >>= mapM_ prlinonly . take (read1 n) 
+  "grt":cat:n:_ -> generateRandom mgr cat >>= mapM_ prlin . take (read1 n) 
   "p":lang:cat:ws -> do
     let ts = parse mgr lang cat $ unwords ws
     mapM_ (putStrLn . showTree) ts 
@@ -43,6 +51,7 @@ treat mgr s = case words s of
     case G.parse (read n) grammar (CId cat) ws of
       t:_ -> prlin t
       _ -> putStrLn "no parse found" 
+  "h":_ -> printHelp mgr
   _ -> lins $ readTree mgr s
  where
   grammar = gfcc mgr
@@ -60,4 +69,6 @@ treat mgr s = case words s of
     putStrLn $ showTree t
     prlinonly t
   prlinonly t = mapM_ (lin t) $ langs
+  read1 s = if all isDigit s then read s else 1
+