Fix a catastrophic backtracking bug in JavaLexer (#1594)

* JavaLexer: Demonstrate a catastrophic backtracking bug * JavaLexer: Fix a catastrophic backtracking bug Closes #1586
pygments · Nov 9, 2020 · fb40b71 · fb40b71
1 parent afa9e3b
commit fb40b71
Show file tree

Hide file tree

Showing 2 changed files with 32 additions and 2 deletions.
diff --git a/pygments/lexers/jvm.py b/pygments/lexers/jvm.py
@@ -65,7 +65,7 @@ class JavaLexer(RegexLexer):
              'var'),
             (r'(import(?:\s+static)?)(\s+)', bygroups(Keyword.Namespace, Text),
              'import'),
-            (r'"(\\\\|\\"|[^"])*"', String),
+            (r'"', String, 'string'),
             (r"'\\.'|'[^\\]'|'\\u[0-9a-fA-F]{4}'", String.Char),
             (r'(\.)((?:[^\W\d]|\$)[\w$]*)', bygroups(Punctuation,
                                                      Name.Attribute)),
@@ -96,6 +96,13 @@ class JavaLexer(RegexLexer):
         'import': [
             (r'[\w.]+\*?', Name.Namespace, '#pop')
         ],
+        'string': [
+            (r'[^\\"]+', String),
+            (r'\\\\', String),  # Escaped backslash
+            (r'\\"', String),  # Escaped quote
+            (r'\\', String),  # Bare backslash
+            (r'"', String, '#pop'),  # Closing quote
+        ],
     }
 
 

diff --git 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'