[utils] Fix unescapeHTML for misformed string like "&a"" (#13935)

7 years ago · 95f3f7c20a
parent f5469da9e6
commit 95f3f7c20a
3 changed files with 8 additions and 1 deletions
--- a/6
+++ b/6
@ -1,3 +1,9 @@
+version <unreleased>
+
+Core
+* [utils] Fix unescapeHTML for misformed string like "&a&quot;" (#13935)
+
+
 version 2017.08.18

 Core
--- a/test/test_utils.py
+++ b/test/test_utils.py
@ -279,6 +279,7 @@ class TestUtil(unittest.TestCase):
        self.assertEqual(unescapeHTML('&#47;'), '/')
        self.assertEqual(unescapeHTML('&eacute;'), 'é')
        self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
+        self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
        # HTML5 entities
        self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')

--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@ -596,7 +596,7 @@ def unescapeHTML(s):
    assert type(s) == compat_str

    return re.sub(
-        r'&([^;]+;)', lambda m: _htmlentity_transform(m.group(1)), s)
+        r'&([^&;]+;)', lambda m: _htmlentity_transform(m.group(1)), s)


 def get_subprocess_encoding():