From fb40b7113e6c71f29c110cc9c32077a0883b732b Mon Sep 17 00:00:00 2001 From: Kurt McKee Date: Mon, 9 Nov 2020 00:26:21 -0600 Subject: Fix a catastrophic backtracking bug in JavaLexer (#1594) * JavaLexer: Demonstrate a catastrophic backtracking bug * JavaLexer: Fix a catastrophic backtracking bug Closes #1586 --- tests/test_java.py | 25 ++++++++++++++++++++++++- 1 file changed, 24 insertions(+), 1 deletion(-) (limited to 'tests/test_java.py') diff --git a/tests/test_java.py b/tests/test_java.py index 3baec0ad..f7b16bd7 100644 --- a/tests/test_java.py +++ b/tests/test_java.py @@ -7,9 +7,11 @@ :license: BSD, see LICENSE for details. """ +import time + import pytest -from pygments.token import Text, Name, Punctuation, Keyword, Number +from pygments.token import Keyword, Name, Number, Punctuation, String, Text from pygments.lexers import JavaLexer @@ -76,3 +78,24 @@ def test_numeric_literals(lexer): (Text, '\n') ] assert list(lexer.get_tokens(fragment)) == tokens + + +@pytest.mark.parametrize( + 'text', + ( + '""', '"abc"', '"ひらがな"', '"123"', + '"\\\\"', '"\\t"' '"\\""', + ), +) +def test_string_literals_positive_match(lexer, text): + """Test positive matches for string literals.""" + tokens = list(lexer.get_tokens_unprocessed(text)) + assert all([token is String for _, token, _ in tokens]) + assert ''.join([value for _, _, value in tokens]) == text + + +def test_string_literals_backtracking(lexer): + """Test catastrophic backtracking for string literals.""" + start_time = time.time() + list(lexer.get_tokens_unprocessed('"' + '\\' * 100)) + assert time.time() - start_time < 1, 'possible backtracking bug' -- cgit v1.2.1