queryparse: collect literals into set and return it

2 files changed, 22 insertions, 16 deletions

diff --git a/queryparse.py b/queryparse.py
index 94d1451..9bab83e 100644
--- a/queryparse.py
+++ b/queryparse.py
@@ -69,45 +69,46 @@ e4 ::= TAG | "[" e1 "]";
 class ParseError(Exception):
 	pass
 
-def e4(l):
+def e4(lits, l):
 	typ,s = classify_head(l)
 
 	if typ == TokType.OPEN:
-		pt,l = e1(l[1:])
+		pt,l = e1(lits, l[1:])
 		typ,_ = classify_head(l)
 		if typ != TokType.CLOSE:
 			raise ParseError
 		return pt,l[1:]
 	elif typ in LitType:
+		lits.add((typ, s))
 		return Atom(typ, s), l[1:]
 	else:
 		raise ParseError
 
-def e3(l):
+def e3(lits, l):
 	typ,_ = classify_head(l)
 	if typ == TokType.NOT:
-		pt,l = e4(l[1:])
+		pt,l = e4(lits, l[1:])
 		return encapsulate(pt, TokType.NOT), l
-	return e4(l)
+	return e4(lits, l)
 
-def e2(l):
-	pt,l = e3(l)
+def e2(lits, l):
+	pt,l = e3(lits, l)
 	try:
 		while True:
-			pt2,l = e3(l)
+			pt2,l = e3(lits, l)
 			pt = encapsulate(pt, TokType.AND)
 			pt.push(pt2)
 	except ParseError:
 		pass
 	return pt,l
 
-def e1(l):
-	pt,l = e2(l)
+def e1(lits, l):
+	pt,l = e2(lits, l)
 	typ,_ = classify_head(l)
 
 	try:
 		while typ == TokType.OR:
-			pt2,l = e2(l[1:])
+			pt2,l = e2(lits, l[1:])
 			pt = encapsulate(pt, TokType.OR)
 			pt.push(pt2)
 			typ,_ = classify_head(l)
@@ -116,7 +117,8 @@ def e1(l):
 	return pt,l
 
 def parse(l):
-	x,l = e1(l)
+	lits = set()
+	x,l = e1(lits, l)
 	if l != []:
 		raise ParseError
-	return x
+	return x,lits
diff --git a/test/t_queryparse.py b/test/t_queryparse.py
index ac9cf24..df21639 100755
--- a/test/t_queryparse.py
+++ b/test/t_queryparse.py
@@ -5,12 +5,16 @@ import queryparse
 
 def f(x):
 	print(x)
-	print(queryparse.parse(x))
+	v,lits = queryparse.parse(x)
+	print(v)
+	print(lits)
+	print()
 
 f(["x", "OR", "y"])
-f(["x", "y", "OR", "a", "b"])
+f(["x", "y", "OR", "x", "b"])
 f(["x", "y", "z", "w"])
 f(["NOT", "x", "OR", "y"])
 f(["NOT", "[", "x", "y", "z", "]"])
-f(["x", "y", "NOT", "z", "[", "a", "OR", "b", "]"])
+f(["x", "y", "NOT", "z", "[", "a", "OR", "b", "]", "c"])
+f(["x", "y", "z", "OR", "NOT", "x", "a", "z"])
 f(["~~internal-*", "~-Application*", "NOT", "~~*xyz*"])