This PR references issue gh-140797

It adds validation to re.Scanner.init that rejects lexicon patterns containing capturing groups. If a user-supplied pattern contains any capturing groups, Scanner now raises ValueError with a clear message advising the use of non-capturing groups (?:...) instead.
python · serhiy-storchaka · Nov 4, 2025 · Nov 3, 2025 · Nov 4, 2025 · Nov 4, 2025
commit 29db6ca11a3db0d04006bfe844c0c97bb179e60d
diff --git a/Lib/re/__init__.py b/Lib/re/__init__.py
@@ -397,9 +397,16 @@ def __init__(self, lexicon, flags=0):
         s = _parser.State()
         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"
+                   )
+
             gid = s.opengroup()
             p.append(_parser.SubPattern(s, [
-                (SUBPATTERN, (gid, 0, 0, _parser.parse(phrase, flags))),
+                (SUBPATTERN, (gid, 0, 0, sub_pattern)),
                 ]))
             s.closegroup(gid, p[-1])
         p = _parser.SubPattern(s, [(BRANCH, (None, p))])

diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py
@@ -1639,6 +1639,25 @@ 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
+
+        #Presence of Capturing group throws an error
+        lex = [("(a)b", None)]
+        with self.assertRaises(ValueError):
+            Scanner(lex)
+
+        #Presence of 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
@@ -0,0 +1,4 @@
+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.