diff --git a/Lib/re/_compiler.py b/Lib/re/_compiler.py index d0a4c55caf6e41..f5fd160ba00435 100644 --- a/Lib/re/_compiler.py +++ b/Lib/re/_compiler.py @@ -100,6 +100,13 @@ def _compile(code, pattern, flags): emit(ANY_ALL) else: emit(ANY) + elif op is POSSESSIVE_REPEAT: + # gh-106052: Possessive quantifiers do not work when the + # subpattern contains backtracking, i.e. "(?:ab?c)*+". + # Implement it as equivalent greedy qualifier in atomic group. + p = [(MAX_REPEAT, av)] + p = [(ATOMIC_GROUP, p)] + _compile(code, p, flags) elif op in REPEATING_CODES: if _simple(av[2]): emit(REPEATING_CODES[op][2]) diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py index e4d14356402db7..efdd6320aed692 100644 --- a/Lib/test/test_re.py +++ b/Lib/test/test_re.py @@ -2398,6 +2398,16 @@ def test_bug_gh91616(self): self.assertTrue(re.fullmatch(r'(?s:(?>.*?\.).*)\Z', "a.txt")) # reproducer self.assertTrue(re.fullmatch(r'(?s:(?=(?P.*?\.))(?P=g0).*)\Z', "a.txt")) + def test_bug_gh106052(self): + self.assertEqual(re.match("(?>(?:ab?c)+)", "aca").span(), (0, 2)) + self.assertEqual(re.match("(?:ab?c)++", "aca").span(), (0, 2)) + self.assertEqual(re.match("(?>(?:ab?c)*)", "aca").span(), (0, 2)) + self.assertEqual(re.match("(?:ab?c)*+", "aca").span(), (0, 2)) + self.assertEqual(re.match("(?>(?:ab?c)?)", "a").span(), (0, 0)) + self.assertEqual(re.match("(?:ab?c)?+", "a").span(), (0, 0)) + self.assertEqual(re.match("(?>(?:ab?c){1,3})", "aca").span(), (0, 2)) + self.assertEqual(re.match("(?:ab?c){1,3}+", "aca").span(), (0, 2)) + @unittest.skipIf(multiprocessing is None, 'test requires multiprocessing') def test_regression_gh94675(self): pattern = re.compile(r'(?<=[({}])(((//[^\n]*)?[\n])([\000-\040])*)*' @@ -2491,6 +2501,7 @@ def test_atomic_group(self): 17: SUCCESS ''') + @unittest.expectedFailure # gh-106052 def test_possesive_repeat_one(self): self.assertEqual(get_debug_out(r'a?+'), '''\ POSSESSIVE_REPEAT 0 1 @@ -2503,6 +2514,7 @@ def test_possesive_repeat_one(self): 12: SUCCESS ''') + @unittest.expectedFailure # gh-106052 def test_possesive_repeat(self): self.assertEqual(get_debug_out(r'(?:ab)?+'), '''\ POSSESSIVE_REPEAT 0 1 diff --git a/Misc/NEWS.d/next/Library/2023-07-07-14-52-31.gh-issue-106052.ak8nbs.rst b/Misc/NEWS.d/next/Library/2023-07-07-14-52-31.gh-issue-106052.ak8nbs.rst new file mode 100644 index 00000000000000..f2d4c2f7b18ec7 --- /dev/null +++ b/Misc/NEWS.d/next/Library/2023-07-07-14-52-31.gh-issue-106052.ak8nbs.rst @@ -0,0 +1,2 @@ +:mod:`re` module: fix the matching of possessive quantifiers in the case of +a subpattern containing backtracking.