diff --git a/test/test_utils.py b/test/test_utils.py index 4cd818850..9e1808e6b 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -899,6 +899,7 @@ class TestUtil(unittest.TestCase): def test_clean_html(self): self.assertEqual(clean_html('a:\nb'), 'a: b') self.assertEqual(clean_html('a:\n "b"'), 'a: "b"') + self.assertEqual(clean_html('a
\xa0b'), 'a\nb') def test_intlist_to_bytes(self): self.assertEqual( diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 91e235ff2..41bc20544 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -421,8 +421,8 @@ def clean_html(html): # Newline vs
html = html.replace('\n', ' ') - html = re.sub(r'\s*<\s*br\s*/?\s*>\s*', '\n', html) - html = re.sub(r'<\s*/\s*p\s*>\s*<\s*p[^>]*>', '\n', html) + html = re.sub(r'(?u)\s*<\s*br\s*/?\s*>\s*', '\n', html) + html = re.sub(r'(?u)<\s*/\s*p\s*>\s*<\s*p[^>]*>', '\n', html) # Strip html tags html = re.sub('<.*?>', '', html) # Replace html entities