[1.11.X] Fixed CVE-2019-14232 -- Adjusted regex to avoid backtracking…

… issues when truncating HTML.

Thanks to Guido Vranken for initial report.

django/utils/text.py

@@ -27,8 +27,8 @@ def capfirst(x):
 
 
 # Set up regular expressions
-re_words = re.compile(r'<.*?>|((?:\w[-\w]*|&.*?;)+)', re.U | re.S)
-re_chars = re.compile(r'<.*?>|(.)', re.U | re.S)
+re_words = re.compile(r'<[^>]+?>|([^<>\s]+)', re.S)
+re_chars = re.compile(r'<[^>]+?>|(.)', re.S)
 re_tag = re.compile(r'<(/)?(\S+?)(?:(\s*/)|\s.*?)?>', re.S)
 re_newlines = re.compile(r'\r\n|\r')  # Used in normalize_newlines
 re_camel_case = re.compile(r'(((?<=[a-z])[A-Z])|([A-Z](?![A-Z]|$)))')

docs/releases/1.11.23.txt

@@ -5,3 +5,17 @@ Django 1.11.23 release notes
 *August 1, 2019*
 
 Django 1.11.23 fixes security issues in 1.11.22.
+
+CVE-2019-14232: Denial-of-service possibility in ``django.utils.text.Truncator``
+================================================================================
+
+If ``django.utils.text.Truncator``'s ``chars()`` and ``words()`` methods
+were passed the ``html=True`` argument, they were extremely slow to evaluate
+certain inputs due to a catastrophic backtracking vulnerability in a regular
+expression. The ``chars()`` and ``words()`` methods are used to implement the
+:tfilter:`truncatechars_html` and :tfilter:`truncatewords_html` template
+filters, which were thus vulnerable.
+
+The regular expressions used by ``Truncator`` have been simplified in order to
+avoid potential backtracking issues. As a consequence, trailing punctuation may
+now at times be included in the truncated output.

tests/template_tests/filter_tests/test_truncatewords_html.py

@@ -19,13 +19,13 @@ def test_truncate(self):
     def test_truncate2(self):
         self.assertEqual(
             truncatewords_html('<p>one <a href="#">two - three <br>four</a> five</p>', 4),
-            '<p>one <a href="#">two - three <br>four ...</a></p>',
+            '<p>one <a href="#">two - three ...</a></p>',
         )
 
     def test_truncate3(self):
         self.assertEqual(
             truncatewords_html('<p>one <a href="#">two - three <br>four</a> five</p>', 5),
-            '<p>one <a href="#">two - three <br>four</a> five</p>',
+            '<p>one <a href="#">two - three <br>four ...</a></p>',
         )
 
     def test_truncate4(self):

tests/utils_tests/test_text.py

@@ -88,6 +88,16 @@ def test_truncate_chars(self):
         # lazy strings are handled correctly
         self.assertEqual(text.Truncator(lazystr('The quick brown fox')).chars(12), 'The quick...')
 
+    def test_truncate_chars_html(self):
+        perf_test_values = [
+            (('</a' + '\t' * 50000) + '//>', None),
+            ('&' * 50000, '&' * 7 + '...'),
+            ('_X<<<<<<<<<<<>', None),
+        ]
+        for value, expected in perf_test_values:
+            truncator = text.Truncator(value)
+            self.assertEqual(expected if expected else value, truncator.chars(10, html=True))
+
     def test_truncate_words(self):
         truncator = text.Truncator('The quick brown fox jumped over the lazy dog.')
         self.assertEqual('The quick brown fox jumped over the lazy dog.', truncator.words(10))
@@ -137,11 +147,16 @@ def test_truncate_html_words(self):
         truncator = text.Truncator('<i>Buenos d&iacute;as! &#x00bf;C&oacute;mo est&aacute;?</i>')
         self.assertEqual('<i>Buenos d&iacute;as! &#x00bf;C&oacute;mo...</i>', truncator.words(3, '...', html=True))
         truncator = text.Truncator('<p>I &lt;3 python, what about you?</p>')
-        self.assertEqual('<p>I &lt;3 python...</p>', truncator.words(3, '...', html=True))
+        self.assertEqual('<p>I &lt;3 python,...</p>', truncator.words(3, '...', html=True))
 
-        re_tag_catastrophic_test = ('</a' + '\t' * 50000) + '//>'
-        truncator = text.Truncator(re_tag_catastrophic_test)
-        self.assertEqual(re_tag_catastrophic_test, truncator.words(500, html=True))
+        perf_test_values = [
+            ('</a' + '\t' * 50000) + '//>',
+            '&' * 50000,
+            '_X<<<<<<<<<<<>',
+        ]
+        for value in perf_test_values:
+            truncator = text.Truncator(value)
+            self.assertEqual(value, truncator.words(50, html=True))
 
     def test_wrap(self):
         digits = '1234 67 9'