Added parser completions to py-bindings.

3 files changed, 102 insertions, 16 deletions

diff --git a/contrib/py-bindings/PyGF.hsc b/contrib/py-bindings/PyGF.hsc
index d9f249eab..ddd3920b6 100644
--- a/contrib/py-bindings/PyGF.hsc
+++ b/contrib/py-bindings/PyGF.hsc
@@ -1,7 +1,9 @@
 {-# LANGUAGE ForeignFunctionInterface #-} 
--- GF Python bindings                                                                                                                                        -- Jordi Saludes, upc.edu 2010
+--
+-- GF Python bindings
 -- Jordi Saludes, upc.edu 2010
 --
+
 module PyGF where
 
 import PGF
@@ -9,6 +11,8 @@ import Foreign
 import CString
 import Foreign.C.Types
 import Control.Monad
+import Data.Map (keys, (!))
+import Data.Char (isSpace)
 
 #include "pygf.h"
 
@@ -73,6 +77,7 @@ instance Storable Tree where
       deRefStablePtr sp
 -}
 
+
 foreign export ccall gf_freePGF :: Ptr PGF -> IO ()
 foreign export ccall gf_freeType :: Ptr Type -> IO ()
 foreign export ccall gf_freeLanguage :: Ptr Language -> IO ()
@@ -141,6 +146,18 @@ listToPy mk ls = do
           poke pl l
           pyl << pl
        
+
+listToPyStrings :: [String] -> IO (Ptr ())
+listToPyStrings ss = do
+     pyls <- pyList
+     mapM_ (mpoke pyls) ss
+     return pyls
+  where mpoke pyl s = do
+  	      cs <- newCString s
+	      pcs <- pyString cs
+	      pyl << pcs
+
+
 -- foreign export ccall "gf_freeArray" free :: Ptr a -> IO ()
 
               
@@ -256,6 +273,29 @@ gf_functiontype ppgf pcid = do
         _      -> return nullPtr
 
 
+foreign export ccall gf_completions :: Ptr PGF -> Ptr Language -> Ptr Type -> CString -> IO (Ptr ())
+gf_completions ppgf plang pcat ctoks = do
+	pgf  <- peek ppgf
+	lang <- peek plang
+	cat  <- peek pcat
+	toks <- peekCString ctoks
+	let (rpre,rs) = break isSpace (reverse toks)
+	    pre = reverse rpre
+	    ws = words (reverse rs)
+	    state0 = initState pgf lang cat
+	    completions =
+	    		case loop state0 ws of
+	    		     Nothing    -> []
+			     Just state -> keys $ getCompletions state pre
+	listToPyStrings completions
+   where
+        loop ps []     = Just ps
+        loop ps (w:ws) =
+            case nextState ps (simpleParseInput w) of
+                Left _   -> Nothing
+                Right ps -> loop ps ws
+ 	
+	
 foreign import ccall "newLang" pyLang :: IO (Ptr Language) 
 foreign import ccall "newPGF" pyPGF :: IO (Ptr PGF) 
 foreign import ccall "newTree" pyTree :: IO (Ptr Tree) 
@@ -263,4 +303,5 @@ foreign import ccall "newgfType" pyType :: IO (Ptr Type)
 foreign import ccall "newCId" pyCId :: IO (Ptr CId) 
 foreign import ccall "newExpr" pyExpr :: IO (Ptr Expr) 
 foreign import ccall "newList" pyList :: IO (Ptr ()) 
+foreign import ccall "newString" pyString :: CString -> IO (Ptr ())
 foreign import ccall "append" (<<) :: Ptr () -> Ptr a -> IO ()
diff --git a/contrib/py-bindings/gfmodule.c b/contrib/py-bindings/gfmodule.c
index bd4693d2e..dc8f583a7 100644
--- a/contrib/py-bindings/gfmodule.c
+++ b/contrib/py-bindings/gfmodule.c
@@ -1,5 +1,5 @@
 // GF Python bindings
-// Jordi Saludes, upc.edu 2010
+// Jordi Saludes, upc.edu 2010, 2011
 //
 
 #include <Python.h>
@@ -33,6 +33,7 @@ NEWGF(Tree,GF_Tree,TreeType,"gf.tree","gf tree")
 DEALLOCFN(CId_dealloc, CId, gf_freeCId, "freeCId")
 
 
+
 /* PGF methods, constructor and destructor */
 
 DEALLOCFN(PGF_dealloc, PGF, gf_freePGF, "freePGF")
@@ -123,17 +124,16 @@ parse(PGF *self, PyObject *args, PyObject *kws)
 	Lang *lang;
 	gfType *cat = NULL;
 	char *lexed;
-	static char *kwlist[] = {"lexed", "lang", "cat", NULL};
-	if (!PyArg_ParseTupleAndKeywords(args, kws, "sO|O", kwlist,
-				       &lexed, &lang, &cat))
-    	return NULL;
+	static char *kwlist[] = {"lang", "lexed", "cat", NULL};
+	if (!PyArg_ParseTupleAndKeywords(args, kws, "Os|O", kwlist, &lang, &lexed, &cat))
+	  return NULL;
 	if (!checkType(self, &PGFType)) return NULL;
 	if (!checkType(lang, &LangType)) return NULL;
 	if (cat) {
 	  if (!checkType(cat, &gfTypeType)) return NULL;
 	} else { 
-	  cat = gf_startCat(self);		
-	} 
+	  cat = gf_startCat(self);
+	}
 	return gf_parse(self, lang, cat, lexed);
 }
 
@@ -154,9 +154,27 @@ readPGF(PyObject *self, PyObject *args)
 }
 
 
+static PyObject*
+completions(PGF *self, PyObject *args, PyObject *kws)
+{ char   *tokens;
+  Lang   *lang;
+  gfType *cat = NULL;
+  static char *kwlist[] = {"lang", "tokens", "category", NULL};
+  if (!PyArg_ParseTupleAndKeywords(args, kws, "Os|O", kwlist, &lang, &tokens, &cat))
+    return NULL;
+  if (!checkType(self, &PGFType)) return NULL;
+  if (!checkType(lang, &LangType)) return NULL;
+  if (cat) {
+    if (!checkType(cat, &gfTypeType)) return NULL;
+  } else {
+    cat = gf_startCat(self);
+  }
+  return gf_completions(self, lang, cat, tokens);
+}
+
 
 static PyMethodDef pgf_methods[] = {
-  {"parse", (PyCFunction)parse, METH_VARARGS|METH_KEYWORDS,"Parse a string."},
+  {"parse", (PyCFunction)parse, METH_VARARGS|METH_KEYWORDS, "Parse a string."},
   {"lin", (PyCFunction)linearize, METH_VARARGS,"Linearize tree."},
   {"lang_code", (PyCFunction)languageCode, METH_VARARGS,"Get the language code."},
   {"print_name", (PyCFunction)printName, METH_VARARGS,"Get the print name for a id."},
@@ -166,6 +184,7 @@ static PyMethodDef pgf_methods[] = {
   {"functions", (PyCFunction)gf_functions, METH_NOARGS,"Get all functions."},
   {"abstract", (PyCFunction)abstractName, METH_NOARGS,"Get the module abstract name."},
   {"languages", (PyCFunction)gf_languages, METH_NOARGS,"Get the module languages."},
+  {"complete", (PyCFunction)completions, METH_VARARGS|METH_KEYWORDS, "Get completions for tokens."},
   {NULL, NULL, 0, NULL}  /* Sentinel */
 };
 
@@ -281,12 +300,17 @@ initgf(void)
 	if (PyType_Ready(&t) < 0) return;
 	
         READYTYPE(CIdType, CId_repr, CId_dealloc)
+	
 	PGFType.tp_methods = pgf_methods;
 	READYTYPE(PGFType, pgf_repr, PGF_dealloc)
+	
 	READYTYPE(LangType, lang_repr, Lang_dealloc)
+	
 	READYTYPE(gfTypeType, gfType_repr, gfType_dealloc)
-        ExprType.tp_methods = expr_methods;
+        
+	ExprType.tp_methods = expr_methods;
 	READYTYPE(ExprType, expr_repr, expr_dealloc)
+	
 	TreeType.tp_methods = expr_methods; // Tree == Expr ?
 	READYTYPE(TreeType, tree_repr, Tree_dealloc)	
 
@@ -311,4 +335,5 @@ PyModule_AddObject(m, "gf", (PyObject *)&t);
 /* List utilities to be imported by FFI */
 
 inline PyObject* newList() { return  PyList_New(0); }
+inline PyObject* newString(const char *s) { return  PyString_FromString(s); }
 inline void append(PyObject* l, PyObject* ob) { PyList_Append(l, ob); }
diff --git a/contrib/py-bindings/test.py b/contrib/py-bindings/test.py
index ebfa0a0c8..6ec3f77cc 100644
--- a/contrib/py-bindings/test.py
+++ b/contrib/py-bindings/test.py
@@ -7,7 +7,6 @@
 import gf
 import unittest
 
-
 samples = [
 	(['Odd', ['Number', 89]],
 	 {'eng': "is 89 odd",
@@ -93,7 +92,7 @@ class TestParsing(unittest.TestCase):
 		l = gf.read_language(self.lang)
 		for abs,cnc in self.lexed:
 			rabs = exp2str(abs)
-			ps = pgf.parse(cnc['eng'], l)
+			ps = pgf.parse(l, cnc['eng'])
 			self.failUnless(ps)
 			pt = rmprefix(ps[0])
 			self.assertEqual(pt,rabs)
@@ -108,7 +107,7 @@ class TestLinearize(unittest.TestCase):
 	def test_Linearize(self):
 		l = self.lang
 		for abs,cnc in self.samples:
-			ts = self.pgf.parse(cnc['eng'], l)
+			ts = self.pgf.parse(l, cnc['eng'])
 			self.assertEqual(cnc['eng'],self.pgf.lin(l,ts[0]))
 
 class TestTranslate(unittest.TestCase):
@@ -122,7 +121,7 @@ class TestTranslate(unittest.TestCase):
 			for i,l in self.langs:
 				for j,m in self.langs:
 					if i==j: continue
-					parsed = self.pgf.parse(cnc[i],l)
+					parsed = self.pgf.parse(l, cnc[i])
 					assert len(parsed) == 1
 					lin = self.pgf.lin(m,parsed[0])
 					self.assertEqual(lin,cnc[j])
@@ -146,7 +145,7 @@ class TestUnapplyExpr(unittest.TestCase):
 		lg = 'eng'
 		lang = self.langs[lg]
 		for abs,cnc in self.samples:
-			parsed = self.pgf.parse(cnc[lg],lang)
+			parsed = self.pgf.parse(lang, cnc[lg])
 			uparsed = self.deep_unapp(parsed[0])
 			self.assertEqual(abs,uparsed)
 
@@ -154,7 +153,7 @@ class TestUnapplyExpr(unittest.TestCase):
 		lg = 'eng'
 		lang = self.langs[lg]
 		cnc = self.samples[0][1]
-		parsed = self.pgf.parse(cnc[lg],lang)
+		parsed = self.pgf.parse(lang, cnc[lg])
 		exp = parsed[0]
 		for t in 'Question Object Int'.split():
 			self.assertEqual(`exp.infer(self.pgf)`, t)
@@ -163,5 +162,26 @@ class TestUnapplyExpr(unittest.TestCase):
 				exp = uexp[1]
 
 
+class TestComplete(unittest.TestCase):
+    def setUp(self):
+        self.lexed = samples
+	self.pgf = gf.read_pgf('Query.pgf')
+	self.langs = dict([(lang2iso(l),l) for l in self.pgf.languages()])
+
+    def test_complete(self):
+        for (_,d) in self.lexed:
+            for l,text in d.items():
+                lang = self.langs[l]
+		for k in range(len(text)):
+                    if text[k].isdigit() or text[k-1].isdigit(): # No completion for integer literals
+                         continue
+		    comps = self.pgf.complete(lang,text[:k])
+		    self.assertNotEqual(comps, [],
+					msg="while completing '%s^%s'" % (text[:k],text[k:]))
+
+		    self.assertTrue(any(w in text for w in comps),
+				    msg="None of %s is in '%s'" % (comps,text))
+					
+
 if __name__ == '__main__':
 	unittest.main()