Fixed grammars

examples/json_parser.py

@@ -24,11 +24,11 @@ json_grammar = r"""
     object : "{" [pair ("," pair)*] "}"
     pair   : string ":" value
 
-    number: FLOAT
+    number: SIGNED_NUMBER
     string : ESCAPED_STRING
 
     %import common.ESCAPED_STRING
-    %import common.FLOAT
+    %import common.SIGNED_NUMBER
     %import common.WS
 
     %ignore WS

lark/grammars/common.g

@@ -3,13 +3,18 @@
 //
 
 DIGIT: "0".."9"
+HEXDIGIT: "a".."f"|"A".."F"|DIGIT
 
 INT: DIGIT+
-DECIMAL: INT ("." INT)?
+SIGNED_INT: ["+"|"-"] INT
+DECIMAL: INT "." INT? | "." INT
 
 // float = /-?\d+(\.\d+)?([eE][+-]?\d+)?/
-FLOAT: "-"? DECIMAL (("e"|"E")("+"|"-")? INT)?
+_EXP: ("e"|"E") SIGNED_INT
+FLOAT: INT _EXP | DECIMAL _EXP?
 
+NUMBER: FLOAT | INT
+SIGNED_NUMBER: ["+"|"-"] NUMBER
 
 //
 // Strings

tests/test_parser.py

@@ -18,6 +18,7 @@ logging.basicConfig(level=logging.INFO)
 
 from lark.lark import Lark
 from lark.common import GrammarError, ParseError
+from lark.lexer import LexError
 
 __path__ = os.path.dirname(__file__)
 def _read(n, *args):
@@ -276,6 +277,20 @@ def _make_parser_test(PARSER):
             x = g.parse('Hello HelloWorld')
             self.assertSequenceEqual(x.children, ['HelloWorld'])
 
+        # def test_string_priority(self):
+        #     g = _Lark("""start: (A | /a?bb/)+
+        #                  A: "a"  """)
+        #     x = g.parse('abb')
+        #     self.assertEqual(len(x.children), 2)
+
+        #     # This parse raises an exception because the lexer will always try to consume
+        #     # "a" first and will never match the regular expression
+        #     # This behavior is subject to change!!
+        #     # Thie won't happen with ambiguity handling.
+        #     g = _Lark("""start: (A | /a?ab/)+
+        #                  A: "a"  """)
+        #     self.assertRaises(LexError, g.parse, 'aab')
+
         def test_undefined_rule(self):
             self.assertRaises(GrammarError, _Lark, """start: a""")