Modified m_urls regex and video_url

Some videos have a leading slash, some do not
This commit is contained in:
pishposhmcgee 2013-07-29 15:45:20 -05:00
parent caeefc29eb
commit a3124ba49f

View file

@ -35,12 +35,12 @@ def _real_extract(self, url):
self.report_extraction(video_id) self.report_extraction(video_id)
video_info = json.loads(info_json) video_info = json.loads(info_json)
m_urls = list(re.finditer(r'<video src="(?P<ext>.*?):(?P<url>.*?)"', links_webpage)) m_urls = list(re.finditer(r'<video src="(?P<ext>.*?):/?(?P<url>.*?)"', links_webpage))
if m_urls is None or len(m_urls) == 0: if m_urls is None or len(m_urls) == 0:
raise ExtractorError(u'Unable to extract video url') raise ExtractorError(u'Unable to extract video url')
# They are sorted from worst to best quality # They are sorted from worst to best quality
m_url = m_urls[-1] m_url = m_urls[-1]
video_url = base_url + m_url.group('url') video_url = base_url + '/' + m_url.group('url')
ext = m_url.group('ext') ext = m_url.group('ext')
return {'url': video_url, return {'url': video_url,