X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/46113edab215c2211a604c06245c16d5d4e57dcf..bbf0d3ad085e9b9fdee0b06a32d303becf0df916:/youtube_dl/extractor/worldstarhiphop.py?ds=sidebyside diff --git a/youtube_dl/extractor/worldstarhiphop.py b/youtube_dl/extractor/worldstarhiphop.py index bda3870..09415b5 100644 --- a/youtube_dl/extractor/worldstarhiphop.py +++ b/youtube_dl/extractor/worldstarhiphop.py @@ -6,16 +6,24 @@ from .common import InfoExtractor class WorldStarHipHopIE(InfoExtractor): - _VALID_URL = r'https?://(?:www|m)\.worldstar(?:candy|hiphop)\.com/videos/video\.php\?v=(?P.*)' - _TEST = { - "url": "http://www.worldstarhiphop.com/videos/video.php?v=wshh6a7q1ny0G34ZwuIO", - "md5": "9d04de741161603bf7071bbf4e883186", - "info_dict": { - "id": "wshh6a7q1ny0G34ZwuIO", - "ext": "mp4", - "title": "KO Of The Week: MMA Fighter Gets Knocked Out By Swift Head Kick!" + _VALID_URL = r'https?://(?:www|m)\.worldstar(?:candy|hiphop)\.com/(?:videos|android)/video\.php\?v=(?P.*)' + _TESTS = [{ + 'url': 'http://www.worldstarhiphop.com/videos/video.php?v=wshh6a7q1ny0G34ZwuIO', + 'md5': '9d04de741161603bf7071bbf4e883186', + 'info_dict': { + 'id': 'wshh6a7q1ny0G34ZwuIO', + 'ext': 'mp4', + 'title': 'KO Of The Week: MMA Fighter Gets Knocked Out By Swift Head Kick!' } - } + }, { + 'url': 'http://m.worldstarhiphop.com/android/video.php?v=wshh6a7q1ny0G34ZwuIO', + 'md5': 'dc1c76c83ecc4190bb1eb143899b87d3', + 'info_dict': { + 'id': 'wshh6a7q1ny0G34ZwuIO', + 'ext': 'mp4', + 'title': 'KO Of The Week: MMA Fighter Gets Knocked Out By Swift Head Kick!' + } + }] def _real_extract(self, url): video_id = self._match_id(url) @@ -26,19 +34,22 @@ class WorldStarHipHopIE(InfoExtractor): return self.url_result('vevo:%s' % m_vevo_id.group(1), ie='Vevo') video_url = self._search_regex( - r'so\.addVariable\("file","(.*?)"\)', webpage, 'video URL') + [r'so\.addVariable\("file","(.*?)"\)', + r'
\s*]+href="([^"]+)">'], + webpage, 'video URL') if 'youtube' in video_url: return self.url_result(video_url, ie='Youtube') video_title = self._html_search_regex( - r'(?s)
\s*

(.*?)

', + [r'(?s)
\s*

(.*?)

', + r']+class="tc-sp-pinned-title">(.*)'], webpage, 'title') # Getting thumbnail and if not thumbnail sets correct title for WSHH candy video. thumbnail = self._html_search_regex( r'rel="image_src" href="(.*)" />', webpage, 'thumbnail', - fatal=False) + default=None) if not thumbnail: _title = r'candytitles.*>(.*)' mobj = re.search(_title, webpage) @@ -51,4 +62,3 @@ class WorldStarHipHopIE(InfoExtractor): 'title': video_title, 'thumbnail': thumbnail, } -