gh-140797 forbid capturing groups in re.Scanner lexicon patterns- Rev-7

python · serhiy-storchaka · Nov 4, 2025 · Nov 3, 2025 · Nov 4, 2025 · Nov 4, 2025
commit feaee4e3e82dd7727ea8173178e0513329fd7bd7
diff --git a/Lib/re/__init__.py b/Lib/re/__init__.py
@@ -398,12 +398,8 @@ def __init__(self, lexicon, flags=0):
         s.flags = flags
         for phrase, action in lexicon:
             sub_pattern = _parser.parse(phrase, flags)
-            if sub_pattern.state.groups != 1:  # <- 1 means always has \0
-                raise ValueError(
-                        "re.Scanner lexicon patterns must not contain capturing groups;\n"
-                            "Please use non-capturing groups (?:...) instead"
-                   )
-
+            if sub_pattern.state.groups != 1:
+                raise ValueError("Cannot use capturing groups in re.Scanner")
             gid = s.opengroup()
             p.append(_parser.SubPattern(s, [
                 (SUBPATTERN, (gid, 0, 0, sub_pattern)),

diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py
@@ -1639,25 +1639,24 @@ def s_int(scanner, token): return int(token)
                          (['sum', 'op=', 3, 'op*', 'foo', 'op+', 312.5,
                            'op+', 'bar'], ''))
 
-    def test_bug_140797(self):
-        #bug 140797: remove capturing groups compilation form re.Scanner
+    def test_bug_gh140797(self):
+        # gh140797: capturing groups is not allowed in re.Scanner
 
-        #Presence of Capturing group throws an error
-        lex = [("(a)b", None)]
-        with self.assertRaises(ValueError):
-            Scanner(lex)
+        msg = "Cannot use capturing groups in re.Scanner"
+        # Capturing group throws an error
+        with self.assertRaisesRegex(ValueError, msg):
+            Scanner([("(a)b", None)])
 
-        #Presence of non-capturing groups should pass normally
+        # Named Group
+        with self.assertRaisesRegex(ValueError, msg):
+            Scanner([("(?P<name>a)", None)])
+
+        # Non-capturing groups should pass normally
         s = Scanner([("(?:a)b", lambda scanner, token: token)])
         result, rem = s.scan("ab")
         self.assertEqual(result,['ab'])
         self.assertEqual(rem,'')
 
-        #Testing a very complex capturing group
-        pattern= "(?P<name>a)"
-        with self.assertRaises(ValueError):
-            Scanner([(pattern, None)])
-
     def test_bug_448951(self):
         # bug 448951 (similar to 429357, but with single char match)
         # (Also test greedy matches.)

diff --git a/Misc/NEWS.d/next/Library/2025-11-03-16-23-54.gh-issue-140797.DuFEeR.rst b/Misc/NEWS.d/next/Library/2025-11-03-16-23-54.gh-issue-140797.DuFEeR.rst
@@ -1,4 +1,2 @@
-The re.Scanner class now forbids regular expressions containing capturing
-groups in its lexicon patterns. Patterns using capturing groups could
-previously lead to crashes with segmentation fault. Use non-capturing groups
-(?:...) instead.
+The undocumented :class:`!re.Scanner` class now forbids regular expressions containing capturing groups in its lexicon patterns. Patterns using capturing groups could
+previously lead to crashes with segmentation fault. Use non-capturing groups (?:...) instead.