diff options
| author | Philipp Hagemeister <phihag@phihag.de> | 2013-02-01 17:29:50 +0100 |
|---|---|---|
| committer | Philipp Hagemeister <phihag@phihag.de> | 2013-02-01 17:29:50 +0100 |
| commit | 3b024e17afcfe12f4ea55e9a200b9cbd61ec3f99 (patch) | |
| tree | 04223a595f09b476ebbb971d163261bcb9983d2a /youtube_dl | |
| parent | ec71c13ab891566abff9010710afb915e8f22523 (diff) | |
Work around buggy HTML Parser in Python < 2.7.3 (Closes #662)
Diffstat (limited to 'youtube_dl')
| -rw-r--r-- | youtube_dl/utils.py | 6 |
1 files changed, 6 insertions, 0 deletions
diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 532e8c782..e6ce028d6 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -280,6 +280,12 @@ class AttrParser(compat_html_parser.HTMLParser): lines[-1] = lines[-1][:self.result[2][1]-self.result[1][1]] lines[-1] = lines[-1][:self.result[2][1]] return '\n'.join(lines).strip() +# Hack for https://github.com/rg3/youtube-dl/issues/662 +if sys.version_info < (2, 7, 3): + AttrParser.parse_endtag = (lambda self, i: + i + len("</scr'+'ipt>") + if self.rawdata[i:].startswith("</scr'+'ipt>") + else compat_html_parser.HTMLParser.parse_endtag(self, i)) def get_element_by_id(id, html): """Return the content of the tag with the specified ID in the passed HTML document""" |