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

Co-authored-by: Łukasz Langa <lukasz@langa.pl>
This commit is contained in:
Pablo Galindo Salgado 2022-06-10 16:56:45 +01:00 committed by GitHub
parent 21a9a85ff4
commit 53a8b17895
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 36 additions and 3 deletions

View File

@ -471,7 +471,7 @@ or_pattern[pattern_ty]:
| patterns[asdl_pattern_seq*]='|'.closed_pattern+ { | patterns[asdl_pattern_seq*]='|'.closed_pattern+ {
asdl_seq_LEN(patterns) == 1 ? asdl_seq_GET(patterns, 0) : _PyAST_MatchOr(patterns, EXTRA) } 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 | literal_pattern
| capture_pattern | capture_pattern
| wildcard_pattern | wildcard_pattern
@ -558,7 +558,7 @@ maybe_star_pattern[pattern_ty]:
| star_pattern | star_pattern
| pattern | pattern
star_pattern[pattern_ty]: star_pattern[pattern_ty] (memo):
| '*' target=pattern_capture_target { | '*' target=pattern_capture_target {
_PyAST_MatchStar(target->v.Name.id, EXTRA) } _PyAST_MatchStar(target->v.Name.id, EXTRA) }
| '*' wildcard_pattern { | '*' wildcard_pattern {
@ -1312,4 +1312,4 @@ invalid_kvpair:
| a=expression !(':') { | a=expression !(':') {
RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, a->lineno, a->end_col_offset - 1, a->end_lineno, -1, "':' expected after dictionary key") } 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='*' 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 ':'") }

View File

@ -3151,6 +3151,27 @@ class TestTracing(unittest.TestCase):
self.assertListEqual(self._trace(f, "go x"), [1, 2, 3]) self.assertListEqual(self._trace(f, "go x"), [1, 2, 3])
self.assertListEqual(self._trace(f, "spam"), [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__": if __name__ == "__main__":
""" """

View File

@ -0,0 +1,2 @@
Fix some exponential backtrace case happening with deeply nested sequence
patterns in match statements. Patch by Pablo Galindo

10
Parser/parser.c generated
View File

@ -7945,6 +7945,10 @@ closed_pattern_rule(Parser *p)
return NULL; return NULL;
} }
pattern_ty _res = NULL; pattern_ty _res = NULL;
if (_PyPegen_is_memoized(p, closed_pattern_type, &_res)) {
p->level--;
return _res;
}
int _mark = p->mark; int _mark = p->mark;
{ // literal_pattern { // literal_pattern
if (p->error_indicator) { if (p->error_indicator) {
@ -8100,6 +8104,7 @@ closed_pattern_rule(Parser *p)
} }
_res = NULL; _res = NULL;
done: done:
_PyPegen_insert_memo(p, _mark, closed_pattern_type, _res);
p->level--; p->level--;
return _res; return _res;
} }
@ -9623,6 +9628,10 @@ star_pattern_rule(Parser *p)
return NULL; return NULL;
} }
pattern_ty _res = NULL; pattern_ty _res = NULL;
if (_PyPegen_is_memoized(p, star_pattern_type, &_res)) {
p->level--;
return _res;
}
int _mark = p->mark; int _mark = p->mark;
if (p->mark == p->fill && _PyPegen_fill_token(p) < 0) { if (p->mark == p->fill && _PyPegen_fill_token(p) < 0) {
p->error_indicator = 1; p->error_indicator = 1;
@ -9707,6 +9716,7 @@ star_pattern_rule(Parser *p)
} }
_res = NULL; _res = NULL;
done: done:
_PyPegen_insert_memo(p, _mark, star_pattern_type, _res);
p->level--; p->level--;
return _res; return _res;
} }