gh-93671: Avoid exponential backtracking in deeply nested sequence patterns in match statements (GH-93680)

miss-islington · pablogsal · web-flow · commit f9d0240db809 · 2022-06-10T09:21:04.000-07:00
Co-authored-by: Łukasz Langa <lukasz@langa.pl> (cherry picked from commit 53a8b17) Co-authored-by: Pablo Galindo Salgado <Pablogsal@gmail.com>
diff --git a/Grammar/python.gram b/Grammar/python.gram
@@ -471,7 +471,7 @@ or_pattern[pattern_ty]:
     | patterns[asdl_pattern_seq*]='|'.closed_pattern+ {
         asdl_seq_LEN(patterns) == 1 ? asdl_seq_GET(patterns, 0) : _PyAST_MatchOr(patterns, EXTRA) }
 
-closed_pattern[pattern_ty]:
+closed_pattern[pattern_ty] (memo):
     | literal_pattern
     | capture_pattern
     | wildcard_pattern
@@ -558,7 +558,7 @@ maybe_star_pattern[pattern_ty]:
     | star_pattern
     | pattern
 
-star_pattern[pattern_ty]:
+star_pattern[pattern_ty] (memo):
     | '*' target=pattern_capture_target {
         _PyAST_MatchStar(target->v.Name.id, EXTRA) }
     | '*' wildcard_pattern {
@@ -1312,4 +1312,4 @@ invalid_kvpair:
     | a=expression !(':') {
         RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, a->lineno, a->end_col_offset - 1, a->end_lineno, -1, "':' expected after dictionary key") }
     | expression ':' a='*' bitwise_or { RAISE_SYNTAX_ERROR_STARTING_FROM(a, "cannot use a starred expression in a dictionary value") }
-    | expression a=':' {RAISE_SYNTAX_ERROR_KNOWN_LOCATION(a, "expression expected after dictionary key and ':'") }
+    | expression a=':' {RAISE_SYNTAX_ERROR_KNOWN_LOCATION(a, "expression expected after dictionary key and ':'") }
diff --git a/Lib/test/test_patma.py b/Lib/test/test_patma.py
@@ -3151,6 +3151,27 @@ def f(command):             # 0
         self.assertListEqual(self._trace(f, "go x"), [1, 2, 3])
         self.assertListEqual(self._trace(f, "spam"), [1, 2, 3])
 
+    def test_parser_deeply_nested_patterns(self):
+        # Deeply nested patterns can cause exponential backtracking when parsing.
+        # See gh-93671 for more information.
+
+        levels = 100
+
+        patterns = [
+            "A" + "(" * levels + ")" * levels,
+            "{1:" * levels + "1" + "}" * levels,
+            "[" * levels + "1" + "]" * levels,
+        ]
+
+        for pattern in patterns:
+            with self.subTest(pattern):
+                code = inspect.cleandoc("""
+                    match None:
+                        case {}:
+                            pass
+                """.format(pattern))
+                compile(code, "<string>", "exec")
+
 
 if __name__ == "__main__":
     """
diff --git a/Misc/NEWS.d/next/Core and Builtins/2022-06-10-12-03-17.gh-issue-93671.idkQqG.rst b/Misc/NEWS.d/next/Core and Builtins/2022-06-10-12-03-17.gh-issue-93671.idkQqG.rst
@@ -0,0 +1,2 @@
+Fix some exponential backtrace case happening with deeply nested sequence
+patterns in match statements. Patch by Pablo Galindo
diff --git a/Parser/parser.c b/Parser/parser.c

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	`+Fix some exponential backtrace case happening with deeply nested sequence`
	`2`	`+patterns in match statements. Patch by Pablo Galindo`