X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fkaltura.py;h=41c1f3d96be288957cbaba2c21e3477a2969b24d;hb=3995d37da58ed071b54b7f81757cff4d534f5b9b;hp=c0ddad6f9629b054f3e0064a9b3d60dc0f41dcf0;hpb=8ab7e6c4cc93d998a39fda9733587b58f5252999;p=youtube-dl diff --git a/youtube_dl/extractor/kaltura.py b/youtube_dl/extractor/kaltura.py index c0ddad6f9..41c1f3d96 100644 --- a/youtube_dl/extractor/kaltura.py +++ b/youtube_dl/extractor/kaltura.py @@ -23,11 +23,11 @@ class KalturaIE(InfoExtractor): (?: kaltura:(?P\d+):(?P[0-9a-z_]+)| https?:// - (:?(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/ + (:?(?:www|cdnapi(?:sec)?)\.)?kaltura\.com(?::\d+)?/ (?: (?: # flash player - index\.php/kwidget| + index\.php/(?:kwidget|extwidget/preview)| # html5 player html5/html5lib/[^/]+/mwEmbedFrame\.php ) @@ -91,35 +91,54 @@ class KalturaIE(InfoExtractor): }], }, }, + 'skip': 'Gone. Maybe https://www.safaribooksonline.com/library/tutorials/introduction-to-python-anon/3469/', 'params': { 'skip_download': True, }, + }, + { + 'url': 'https://www.kaltura.com/index.php/extwidget/preview/partner_id/1770401/uiconf_id/37307382/entry_id/0_58u8kme7/embed/iframe?&flashvars[streamerType]=auto', + 'only_matching': True, + }, + { + 'url': 'https://www.kaltura.com:443/index.php/extwidget/preview/partner_id/1770401/uiconf_id/37307382/entry_id/0_58u8kme7/embed/iframe?&flashvars[streamerType]=auto', + 'only_matching': True, } ] @staticmethod def _extract_url(webpage): + # Embed codes: https://knowledge.kaltura.com/embedding-kaltura-media-players-your-site mobj = ( re.search( r"""(?xs) kWidget\.(?:thumb)?[Ee]mbed\( \{.*? - (?P['\"])wid(?P=q1)\s*:\s* - (?P['\"])_?(?P(?:(?!(?P=q2)).)+)(?P=q2),.*? - (?P['\"])entry_?[Ii]d(?P=q3)\s*:\s* - (?P['\"])(?P(?:(?!(?P=q4)).)+)(?P=q4)(?:,|\s*\}) + (?P['"])wid(?P=q1)\s*:\s* + (?P['"])_?(?P(?:(?!(?P=q2)).)+)(?P=q2),.*? + (?P['"])entry_?[Ii]d(?P=q3)\s*:\s* + (?P['"])(?P(?:(?!(?P=q4)).)+)(?P=q4)(?:,|\s*\}) """, webpage) or re.search( r'''(?xs) - (?P["\']) - (?:https?:)?//cdnapi(?:sec)?\.kaltura\.com/(?:(?!(?P=q1)).)*(?:p|partner_id)/(?P\d+)(?:(?!(?P=q1)).)* + (?P["']) + (?:https?:)?//cdnapi(?:sec)?\.kaltura\.com(?::\d+)?/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P\d+)(?:(?!(?P=q1)).)* (?P=q1).*? (?: entry_?[Ii]d| - (?P["\'])entry_?[Ii]d(?P=q2) + (?P["'])entry_?[Ii]d(?P=q2) )\s*:\s* - (?P["\'])(?P(?:(?!(?P=q3)).)+)(?P=q3) - ''', webpage)) + (?P["'])(?P(?:(?!(?P=q3)).)+)(?P=q3) + ''', webpage) or + re.search( + r'''(?xs) + ]+src=(?P["']) + (?:https?:)?//(?:www\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P\d+) + (?:(?!(?P=q1)).)* + [?&]entry_id=(?P(?:(?!(?P=q1))[^&])+) + (?P=q1) + ''', webpage) + ) if mobj: embed_info = mobj.groupdict() url = 'kaltura:%(partner_id)s:%(id)s' % embed_info @@ -209,6 +228,8 @@ class KalturaIE(InfoExtractor): partner_id = params['wid'][0][1:] elif 'p' in params: partner_id = params['p'][0] + elif 'partner_id' in params: + partner_id = params['partner_id'][0] else: raise ExtractorError('Invalid URL', expected=True) if 'entry_id' in params: @@ -266,9 +287,12 @@ class KalturaIE(InfoExtractor): # skip for now. if f.get('fileExt') == 'chun': continue - if not f.get('fileExt') and f.get('containerFormat') == 'qt': + if not f.get('fileExt'): # QT indicates QuickTime; some videos have broken fileExt - f['fileExt'] = 'mov' + if f.get('containerFormat') == 'qt': + f['fileExt'] = 'mov' + else: + f['fileExt'] = 'mp4' video_url = sign_url( '%s/flavorId/%s' % (data_url, f['id'])) # audio-only has no videoCodecId (e.g. kaltura:1926081:0_c03e1b5g @@ -319,6 +343,6 @@ class KalturaIE(InfoExtractor): 'thumbnail': info.get('thumbnailUrl'), 'duration': info.get('duration'), 'timestamp': info.get('createdAt'), - 'uploader_id': info.get('userId'), + 'uploader_id': info.get('userId') if info.get('userId') != 'None' else None, 'view_count': info.get('plays'), }