Merge pull request #107 from boxed/incorrect-exponent-parser

Fixed incorrect tokenization case "d*e-1". Fixes #85

baron/grouper.py

@@ -96,7 +96,7 @@ def group_generator(sequence):
         if re.match(r'^\d+\.?[eE]$', current) and match_on_next(r'^\d+$', iterator):
             current += next(iterator)
-        if re.match(r'^\d*\.?\d*[eE]$', current) and match_on_next(r'^[-+]$', iterator) and iterator.show_next(2) and re.match(r'^\d+$', iterator.show_next(2)):
+        if re.match(r'^\d*\.?\d*[eE]$', current) and not re.match('[eE]', current) and match_on_next(r'^[-+]$', iterator) and iterator.show_next(2) and re.match(r'^\d+$', iterator.show_next(2)):
             current += next(iterator)
             current += next(iterator)

tests/test_dumper.py

@@ -534,3 +534,7 @@ def test_try_import_after_colon():
 def test_single_object():
     assert baron.dumps({"type": "name", "value": "a"}) == "a"
+
+
+def test_crash_issue_85():
+    check_dumps('d*e-1\n')
\ No newline at end of file

tests/test_tokenizer.py

@@ -80,6 +80,22 @@ def test_float_exponant():
     match('.5678E-10', 'FLOAT_EXPONANT')
+def test_floating_point_parser_bug_85():
+    from baron import parse
+    assert parse('d*e-1') == [
+        {'first': {'first': {'type': 'name', 'value': 'd'},
+                   'first_formatting': [],
+                   'second': {'type': 'name', 'value': 'e'},
+                   'second_formatting': [],
+                   'type': 'binary_operator',
+                   'value': '*'},
+         'first_formatting': [],
+         'second': {'section': 'number', 'type': 'int', 'value': '1'},
+         'second_formatting': [],
+         'type': 'binary_operator',
+         'value': '-'}]
+
+
 def test_left_parenthesis():
     match('(', 'LEFT_PARENTHESIS')