This PR references issue gh-140797

Abhi210 · Abhi210 · commit 12bf67bc69d7 · 2025-11-03T16:38:53.000+05:30
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.
diff --git a/Lib/re/__init__.py b/Lib/re/__init__.py
@@ -389,6 +389,17 @@ def _pickle(p):
 class Scanner:
     def __init__(self, lexicon, flags=0):
         from ._constants import BRANCH, SUBPATTERN
+
+
+        for phrase,action in lexicon:
+            _compiled_phrase= _compiler.compile(phrase,flags)
+
+            if _compiled_phrase.groups !=0:
+                raise ValueError(
+                    "re.Scanner lexicon patterns must not contain capturing groups;\n"
+                        "Please use non-capturing groups (?:...) instead"
+                )
+
         if isinstance(flags, RegexFlag):
             flags = flags.value
         self.lexicon = lexicon
diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py
@@ -1639,6 +1639,30 @@ 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-Za-z]+)*="
+        with self.assertRaises(ValueError):
+            Scanner([(pattern, None)])
+
+        #Testing the pattern given by the user
+        lexicon = [('(?P<group1>a)(?P<group2>b)', None)]
+        with self.assertRaises(ValueError):
+            Scanner(lexicon)
+
     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.
