X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fgeneric.py;h=5b1da47e3d4bb0678a7e6433d024a6cadad55bed;hb=eedd20ef9637660d5585b1b7d221d7f2d31630ac;hp=2ff002643c9e4404b3427f4f309c187186ccc37a;hpb=d5c418f29fcce3d62bba7f6228b76b69b5b731ce;p=youtube-dl diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index 2ff002643..5b1da47e3 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -9,6 +9,8 @@ from .common import InfoExtractor from .youtube import YoutubeIE from ..compat import ( compat_urllib_parse, + compat_urllib_parse_unquote, + compat_urllib_request, compat_urlparse, compat_xml_parse_error, ) @@ -32,8 +34,20 @@ from .brightcove import BrightcoveIE from .nbc import NBCSportsVPlayerIE from .ooyala import OoyalaIE from .rutv import RUTVIE +from .tvc import TVCIE +from .sportbox import SportBoxEmbedIE from .smotri import SmotriIE from .condenast import CondeNastIE +from .udn import UDNEmbedIE +from .senateisvp import SenateISVPIE +from .bliptv import BlipTVIE +from .svt import SVTIE +from .pornhub import PornHubIE +from .xhamster import XHamsterEmbedIE +from .vimeo import VimeoIE +from .dailymotion import DailymotionCloudIE +from .onionstudios import OnionStudiosIE +from .snagfilms import SnagFilmsEmbedIE class GenericIE(InfoExtractor): @@ -41,6 +55,97 @@ class GenericIE(InfoExtractor): _VALID_URL = r'.*' IE_NAME = 'generic' _TESTS = [ + # Direct link to a video + { + 'url': 'http://media.w3.org/2010/05/sintel/trailer.mp4', + 'md5': '67d406c2bcb6af27fa886f31aa934bbe', + 'info_dict': { + 'id': 'trailer', + 'ext': 'mp4', + 'title': 'trailer', + 'upload_date': '20100513', + } + }, + # Direct link to media delivered compressed (until Accept-Encoding is *) + { + 'url': 'http://calimero.tk/muzik/FictionJunction-Parallel_Hearts.flac', + 'md5': '128c42e68b13950268b648275386fc74', + 'info_dict': { + 'id': 'FictionJunction-Parallel_Hearts', + 'ext': 'flac', + 'title': 'FictionJunction-Parallel_Hearts', + 'upload_date': '20140522', + }, + 'expected_warnings': [ + 'URL could be a direct video link, returning it as such.' + ] + }, + # Direct download with broken HEAD + { + 'url': 'http://ai-radio.org:8000/radio.opus', + 'info_dict': { + 'id': 'radio', + 'ext': 'opus', + 'title': 'radio', + }, + 'params': { + 'skip_download': True, # infinite live stream + }, + 'expected_warnings': [ + r'501.*Not Implemented' + ], + }, + # Direct link with incorrect MIME type + { + 'url': 'http://ftp.nluug.nl/video/nluug/2014-11-20_nj14/zaal-2/5_Lennart_Poettering_-_Systemd.webm', + 'md5': '4ccbebe5f36706d85221f204d7eb5913', + 'info_dict': { + 'url': 'http://ftp.nluug.nl/video/nluug/2014-11-20_nj14/zaal-2/5_Lennart_Poettering_-_Systemd.webm', + 'id': '5_Lennart_Poettering_-_Systemd', + 'ext': 'webm', + 'title': '5_Lennart_Poettering_-_Systemd', + 'upload_date': '20141120', + }, + 'expected_warnings': [ + 'URL could be a direct video link, returning it as such.' + ] + }, + # RSS feed + { + 'url': 'http://phihag.de/2014/youtube-dl/rss2.xml', + 'info_dict': { + 'id': 'http://phihag.de/2014/youtube-dl/rss2.xml', + 'title': 'Zero Punctuation', + 'description': 're:.*groundbreaking video review series.*' + }, + 'playlist_mincount': 11, + }, + # RSS feed with enclosure + { + 'url': 'http://podcastfeeds.nbcnews.com/audio/podcast/MSNBC-MADDOW-NETCAST-M4V.xml', + 'info_dict': { + 'id': 'pdv_maddow_netcast_m4v-02-27-2015-201624', + 'ext': 'm4v', + 'upload_date': '20150228', + 'title': 'pdv_maddow_netcast_m4v-02-27-2015-201624', + } + }, + # google redirect + { + 'url': 'http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&ved=0CCUQtwIwAA&url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DcmQHVoWB5FY&ei=F-sNU-LLCaXk4QT52ICQBQ&usg=AFQjCNEw4hL29zgOohLXvpJ-Bdh2bils1Q&bvm=bv.61965928,d.bGE', + 'info_dict': { + 'id': 'cmQHVoWB5FY', + 'ext': 'mp4', + 'upload_date': '20130224', + 'uploader_id': 'TheVerge', + 'description': 're:^Chris Ziegler takes a look at the\.*', + 'uploader': 'The Verge', + 'title': 'First Firefox OS phones side-by-side', + }, + 'params': { + 'skip_download': False, + } + }, { 'url': 'http://www.hodiho.fr/2013/02/regis-plante-sa-jeep.html', 'md5': '85b90ccc9d73b4acd9138d3af4c27f89', @@ -120,17 +225,6 @@ class GenericIE(InfoExtractor): 'skip_download': True, # m3u8 download }, }, - # Direct link to a video - { - 'url': 'http://media.w3.org/2010/05/sintel/trailer.mp4', - 'md5': '67d406c2bcb6af27fa886f31aa934bbe', - 'info_dict': { - 'id': 'trailer', - 'ext': 'mp4', - 'title': 'trailer', - 'upload_date': '20100513', - } - }, # ooyala video { 'url': 'http://www.rollingstone.com/music/videos/norwegian-dj-cashmere-cat-goes-spartan-on-with-me-premiere-20131219', @@ -155,22 +249,6 @@ class GenericIE(InfoExtractor): }, 'add_ie': ['Ooyala'], }, - # google redirect - { - 'url': 'http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&ved=0CCUQtwIwAA&url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DcmQHVoWB5FY&ei=F-sNU-LLCaXk4QT52ICQBQ&usg=AFQjCNEw4hL29zgOohLXvpJ-Bdh2bils1Q&bvm=bv.61965928,d.bGE', - 'info_dict': { - 'id': 'cmQHVoWB5FY', - 'ext': 'mp4', - 'upload_date': '20130224', - 'uploader_id': 'TheVerge', - 'description': 're:^Chris Ziegler takes a look at the\.*', - 'uploader': 'The Verge', - 'title': 'First Firefox OS phones side-by-side', - }, - 'params': { - 'skip_download': False, - } - }, # embed.ly video { 'url': 'http://www.tested.com/science/weird/460206-tested-grinding-coffee-2000-frames-second/', @@ -220,6 +298,55 @@ class GenericIE(InfoExtractor): 'skip_download': True, }, }, + # TVC embed + { + 'url': 'http://sch1298sz.mskobr.ru/dou_edu/karamel_ki/filial_galleries/video/iframe_src_http_tvc_ru_video_iframe_id_55304_isplay_false_acc_video_id_channel_brand_id_11_show_episodes_episode_id_32307_frameb/', + 'info_dict': { + 'id': '55304', + 'ext': 'mp4', + 'title': 'Дошкольное воспитание', + }, + }, + # SportBox embed + { + 'url': 'http://www.vestifinance.ru/articles/25753', + 'info_dict': { + 'id': '25753', + 'title': 'Вести Экономика ― Прямые трансляции с Форума-выставки "Госзаказ-2013"', + }, + 'playlist': [{ + 'info_dict': { + 'id': '370908', + 'title': 'Госзаказ. День 3', + 'ext': 'mp4', + } + }, { + 'info_dict': { + 'id': '370905', + 'title': 'Госзаказ. День 2', + 'ext': 'mp4', + } + }, { + 'info_dict': { + 'id': '370902', + 'title': 'Госзаказ. День 1', + 'ext': 'mp4', + } + }], + 'params': { + # m3u8 download + 'skip_download': True, + }, + }, + # XHamster embed + { + 'url': 'http://www.numisc.com/forum/showthread.php?11696-FM15-which-pumiscer-was-this-%28-vid-%29-%28-alfa-as-fuck-srx-%29&s=711f5db534502e22260dec8c5e2d66d8', + 'info_dict': { + 'id': 'showthread', + 'title': '[NSFL] [FM15] which pumiscer was this ( vid ) ( alfa as fuck srx )', + }, + 'playlist_mincount': 7, + }, # Embedded TED video { 'url': 'http://en.support.wordpress.com/videos/ted-talks/', @@ -371,16 +498,6 @@ class GenericIE(InfoExtractor): 'title': 'Busty Blonde Siri Tit Fuck While Wank at HandjobHub.com', } }, - # RSS feed - { - 'url': 'http://phihag.de/2014/youtube-dl/rss2.xml', - 'info_dict': { - 'id': 'http://phihag.de/2014/youtube-dl/rss2.xml', - 'title': 'Zero Punctuation', - 'description': 're:.*groundbreaking video review series.*' - }, - 'playlist_mincount': 11, - }, # Multiple brightcove videos # https://github.com/rg3/youtube-dl/issues/2283 { @@ -434,21 +551,6 @@ class GenericIE(InfoExtractor): 'uploader': 'thoughtworks.wistia.com', }, }, - # Direct download with broken HEAD - { - 'url': 'http://ai-radio.org:8000/radio.opus', - 'info_dict': { - 'id': 'radio', - 'ext': 'opus', - 'title': 'radio', - }, - 'params': { - 'skip_download': True, # infinite live stream - }, - 'expected_warnings': [ - r'501.*Not Implemented' - ], - }, # Soundcloud embed { 'url': 'http://nakedsecurity.sophos.com/2014/10/29/sscc-171-are-you-sure-that-1234-is-a-bad-password-podcast/', @@ -480,21 +582,6 @@ class GenericIE(InfoExtractor): }, 'playlist_mincount': 2, }, - # Direct link with incorrect MIME type - { - 'url': 'http://ftp.nluug.nl/video/nluug/2014-11-20_nj14/zaal-2/5_Lennart_Poettering_-_Systemd.webm', - 'md5': '4ccbebe5f36706d85221f204d7eb5913', - 'info_dict': { - 'url': 'http://ftp.nluug.nl/video/nluug/2014-11-20_nj14/zaal-2/5_Lennart_Poettering_-_Systemd.webm', - 'id': '5_Lennart_Poettering_-_Systemd', - 'ext': 'webm', - 'title': '5_Lennart_Poettering_-_Systemd', - 'upload_date': '20141120', - }, - 'expected_warnings': [ - 'URL could be a direct video link, returning it as such.' - ] - }, # Cinchcast embed { 'url': 'http://undergroundwellness.com/podcasts/306-5-steps-to-permanent-gut-healing/', @@ -614,13 +701,24 @@ class GenericIE(InfoExtractor): 'info_dict': { 'id': '100183293', 'ext': 'mp4', - 'title': 'Тайны перевала Дятлова • Тайна перевала Дятлова 1 серия 2 часть', + 'title': 'Тайны перевала Дятлова • 1 серия 2 часть', 'description': 'Документальный сериал-расследование одной из самых жутких тайн ХХ века', 'thumbnail': 're:^https?://.*\.jpg$', 'duration': 694, 'age_limit': 0, }, }, + # Playwire embed + { + 'url': 'http://www.cinemablend.com/new/First-Joe-Dirt-2-Trailer-Teaser-Stupid-Greatness-70874.html', + 'info_dict': { + 'id': '3519514', + 'ext': 'mp4', + 'title': 'Joe Dirt 2 Beautiful Loser Teaser Trailer', + 'thumbnail': 're:^https?://.*\.png$', + 'duration': 45.115, + }, + }, # 5min embed { 'url': 'http://techcrunch.com/video/facebook-creates-on-this-day-crunch-report/518726732/', @@ -631,15 +729,42 @@ class GenericIE(InfoExtractor): 'title': 'Facebook Creates "On This Day" | Crunch Report', }, }, - # RSS feed with enclosure + # SVT embed { - 'url': 'http://podcastfeeds.nbcnews.com/audio/podcast/MSNBC-MADDOW-NETCAST-M4V.xml', + 'url': 'http://www.svt.se/sport/ishockey/jagr-tacklar-giroux-under-intervjun', 'info_dict': { - 'id': 'pdv_maddow_netcast_m4v-02-27-2015-201624', - 'ext': 'm4v', - 'upload_date': '20150228', - 'title': 'pdv_maddow_netcast_m4v-02-27-2015-201624', - } + 'id': '2900353', + 'ext': 'flv', + 'title': 'Här trycker Jagr till Giroux (under SVT-intervjun)', + 'duration': 27, + 'age_limit': 0, + }, + }, + # Crooks and Liars embed + { + 'url': 'http://crooksandliars.com/2015/04/fox-friends-says-protecting-atheists', + 'info_dict': { + 'id': '8RUoRhRi', + 'ext': 'mp4', + 'title': "Fox & Friends Says Protecting Atheists From Discrimination Is Anti-Christian!", + 'description': 'md5:e1a46ad1650e3a5ec7196d432799127f', + 'timestamp': 1428207000, + 'upload_date': '20150405', + 'uploader': 'Heather', + }, + }, + # Crooks and Liars external embed + { + 'url': 'http://theothermccain.com/2010/02/02/video-proves-that-bill-kristol-has-been-watching-glenn-beck/comment-page-1/', + 'info_dict': { + 'id': 'MTE3MjUtMzQ2MzA', + 'ext': 'mp4', + 'title': 'md5:5e3662a81a4014d24c250d76d41a08d5', + 'description': 'md5:9b8e9542d6c3c5de42d6451b7d780cec', + 'timestamp': 1265032391, + 'upload_date': '20100201', + 'uploader': 'Heather', + }, }, # NBC Sports vplayer embed { @@ -650,6 +775,92 @@ class GenericIE(InfoExtractor): 'title': "PFT Live: New leader in the 'new-look' defense", 'description': 'md5:65a19b4bbfb3b0c0c5768bed1dfad74e', }, + }, + # UDN embed + { + 'url': 'http://www.udn.com/news/story/7314/822787', + 'md5': 'fd2060e988c326991037b9aff9df21a6', + 'info_dict': { + 'id': '300346', + 'ext': 'mp4', + 'title': '中一中男師變性 全校師生力挺', + 'thumbnail': 're:^https?://.*\.jpg$', + } + }, + # Ooyala embed + { + 'url': 'http://www.businessinsider.com/excel-index-match-vlookup-video-how-to-2015-2?IR=T', + 'info_dict': { + 'id': '50YnY4czr4ms1vJ7yz3xzq0excz_pUMs', + 'ext': 'mp4', + 'description': 'VIDEO: Index/Match versus VLOOKUP.', + 'title': 'This is what separates the Excel masters from the wannabes', + }, + 'params': { + # m3u8 downloads + 'skip_download': True, + } + }, + # Contains a SMIL manifest + { + 'url': 'http://www.telewebion.com/fa/1263668/%D9%82%D8%B1%D8%B9%D9%87%E2%80%8C%DA%A9%D8%B4%DB%8C-%D9%84%DB%8C%DA%AF-%D9%82%D9%87%D8%B1%D9%85%D8%A7%D9%86%D8%A7%D9%86-%D8%A7%D8%B1%D9%88%D9%BE%D8%A7/%2B-%D9%81%D9%88%D8%AA%D8%A8%D8%A7%D9%84.html', + 'info_dict': { + 'id': 'file', + 'ext': 'flv', + 'title': '+ Football: Lottery Champions League Europe', + 'uploader': 'www.telewebion.com', + }, + 'params': { + # rtmpe downloads + 'skip_download': True, + } + }, + # Brightcove URL in single quotes + { + 'url': 'http://www.sportsnet.ca/baseball/mlb/sn-presents-russell-martin-world-citizen/', + 'md5': '4ae374f1f8b91c889c4b9203c8c752af', + 'info_dict': { + 'id': '4255764656001', + 'ext': 'mp4', + 'title': 'SN Presents: Russell Martin, World Citizen', + 'description': 'To understand why he was the Toronto Blue Jays’ top off-season priority is to appreciate his background and upbringing in Montreal, where he first developed his baseball skills. Written and narrated by Stephen Brunt.', + 'uploader': 'Rogers Sportsnet', + }, + }, + # Dailymotion Cloud video + { + 'url': 'http://replay.publicsenat.fr/vod/le-debat/florent-kolandjian,dominique-cena,axel-decourtye,laurence-abeille,bruno-parmentier/175910', + 'md5': '49444254273501a64675a7e68c502681', + 'info_dict': { + 'id': '5585de919473990de4bee11b', + 'ext': 'mp4', + 'title': 'Le débat', + 'thumbnail': 're:^https?://.*\.jpe?g$', + } + }, + # OnionStudios embed + { + 'url': 'http://www.clickhole.com/video/dont-understand-bitcoin-man-will-mumble-explanatio-2537', + 'info_dict': { + 'id': '2855', + 'ext': 'mp4', + 'title': 'Don’t Understand Bitcoin? This Man Will Mumble An Explanation At You', + 'thumbnail': 're:^https?://.*\.jpe?g$', + 'uploader': 'ClickHole', + 'uploader_id': 'clickhole', + } + }, + # AdobeTVVideo embed + { + 'url': 'https://helpx.adobe.com/acrobat/how-to/new-experience-acrobat-dc.html?set=acrobat--get-started--essential-beginners', + 'md5': '43662b577c018ad707a63766462b1e87', + 'info_dict': { + 'id': '2456', + 'ext': 'mp4', + 'title': 'New experience with Acrobat DC', + 'description': 'New experience with Acrobat DC', + 'duration': 248.667, + }, } ] @@ -771,7 +982,7 @@ class GenericIE(InfoExtractor): force_videoid = smuggled_data['force_videoid'] video_id = force_videoid else: - video_id = os.path.splitext(url.rstrip('/').split('/')[-1])[0] + video_id = compat_urllib_parse_unquote(os.path.splitext(url.rstrip('/').split('/')[-1])[0]) self.to_screen('%s: Requesting header' % video_id) @@ -793,7 +1004,9 @@ class GenericIE(InfoExtractor): full_response = None if head_response is False: - full_response = self._request_webpage(url, video_id) + request = compat_urllib_request.Request(url) + request.add_header('Accept-Encoding', '*') + full_response = self._request_webpage(request, video_id) head_response = full_response # Check for direct link to a video @@ -804,7 +1017,7 @@ class GenericIE(InfoExtractor): head_response.headers.get('Last-Modified')) return { 'id': video_id, - 'title': os.path.splitext(url_basename(url))[0], + 'title': compat_urllib_parse_unquote(os.path.splitext(url_basename(url))[0]), 'direct': True, 'formats': [{ 'format_id': m.group('format_id'), @@ -815,10 +1028,22 @@ class GenericIE(InfoExtractor): } if not self._downloader.params.get('test', False) and not is_intentional: - self._downloader.report_warning('Falling back on generic information extractor.') + force = self._downloader.params.get('force_generic_extractor', False) + self._downloader.report_warning( + '%s on generic information extractor.' % ('Forcing' if force else 'Falling back')) if not full_response: - full_response = self._request_webpage(url, video_id) + request = compat_urllib_request.Request(url) + # Some webservers may serve compressed content of rather big size (e.g. gzipped flac) + # making it impossible to download only chunk of the file (yet we need only 512kB to + # test whether it's HTML or not). According to youtube-dl default Accept-Encoding + # that will always result in downloading the whole file that is not desirable. + # Therefore for extraction pass we have to override Accept-Encoding to any in order + # to accept raw bytes and being able to download only a chunk. + # It may probably better to solve this by checking Content-Type for application/octet-stream + # after HEAD request finishes, but not sure if we can rely on this. + request.add_header('Accept-Encoding', '*') + full_response = self._request_webpage(request, video_id) # Maybe it's a direct link to a video? # Be careful not to download the whole thing! @@ -830,7 +1055,7 @@ class GenericIE(InfoExtractor): head_response.headers.get('Last-Modified')) return { 'id': video_id, - 'title': os.path.splitext(url_basename(url))[0], + 'title': compat_urllib_parse_unquote(os.path.splitext(url_basename(url))[0]), 'direct': True, 'url': url, 'upload_date': upload_date, @@ -910,23 +1135,14 @@ class GenericIE(InfoExtractor): # Look for embedded rtl.nl player matches = re.findall( - r']+?src="((?:https?:)?//(?:www\.)?rtl\.nl/system/videoplayer/[^"]+(?:video_)?embed[^"]+)"', webpage) if matches: return _playlist_from_matches(matches, ie='RtlNl') - # Look for embedded (iframe) Vimeo player - mobj = re.search( - r']+?src=(["\'])(?P(?:https?:)?//player\.vimeo\.com/video/.+?)\1', webpage) - if mobj: - player_url = unescapeHTML(mobj.group('url')) - surl = smuggle_url(player_url, {'Referer': url}) - return self.url_result(surl) - # Look for embedded (swf embed) Vimeo player - mobj = re.search( - r']+?src="((?:https?:)?//(?:www\.)?vimeo\.com/moogaloop\.swf.+?)"', webpage) - if mobj: - return self.url_result(mobj.group(1)) + vimeo_url = VimeoIE._extract_vimeo_url(url, webpage) + if vimeo_url is not None: + return self.url_result(vimeo_url) # Look for embedded YouTube player matches = re.findall(r'''(?x) @@ -995,12 +1211,14 @@ class GenericIE(InfoExtractor): } # Look for embedded blip.tv player - mobj = re.search(r']*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage) - if mobj: - return self.url_result('http://blip.tv/a/a-' + mobj.group(1), 'BlipTV') - mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9_]+)', webpage) - if mobj: - return self.url_result(mobj.group(1), 'BlipTV') + bliptv_url = BlipTVIE._extract_url(webpage) + if bliptv_url: + return self.url_result(bliptv_url, 'BlipTV') + + # Look for SVT player + svt_url = SVTIE._extract_url(webpage) + if svt_url: + return self.url_result(svt_url, 'SVT') # Look for embedded condenast player matches = re.findall( @@ -1054,7 +1272,8 @@ class GenericIE(InfoExtractor): # Look for Ooyala videos mobj = (re.search(r'player\.ooyala\.com/[^"?]+\?[^"]*?(?:embedCode|ec)=(?P[^"&]+)', webpage) or re.search(r'OO\.Player\.create\([\'"].*?[\'"],\s*[\'"](?P.{32})[\'"]', webpage) or - re.search(r'SBN\.VideoLinkset\.ooyala\([\'"](?P.{32})[\'"]\)', webpage)) + re.search(r'SBN\.VideoLinkset\.ooyala\([\'"](?P.{32})[\'"]\)', webpage) or + re.search(r'data-ooyala-video-id\s*=\s*[\'"](?P.{32})[\'"]', webpage)) if mobj is not None: return OoyalaIE._build_url_result(mobj.group('ec')) @@ -1135,6 +1354,32 @@ class GenericIE(InfoExtractor): if rutv_url: return self.url_result(rutv_url, 'RUTV') + # Look for embedded TVC player + tvc_url = TVCIE._extract_url(webpage) + if tvc_url: + return self.url_result(tvc_url, 'TVC') + + # Look for embedded SportBox player + sportbox_urls = SportBoxEmbedIE._extract_urls(webpage) + if sportbox_urls: + return _playlist_from_matches(sportbox_urls, ie='SportBoxEmbed') + + # Look for embedded PornHub player + pornhub_url = PornHubIE._extract_url(webpage) + if pornhub_url: + return self.url_result(pornhub_url, 'PornHub') + + # Look for embedded XHamster player + xhamster_urls = XHamsterEmbedIE._extract_urls(webpage) + if xhamster_urls: + return _playlist_from_matches(xhamster_urls, ie='XHamsterEmbed') + + # Look for embedded Tvigle player + mobj = re.search( + r']+?src=(["\'])(?P(?:https?:)?//cloud\.tvigle\.ru/video/.+?)\1', webpage) + if mobj is not None: + return self.url_result(mobj.group('url'), 'Tvigle') + # Look for embedded TED player mobj = re.search( r']+?src=(["\'])(?Phttps?://embed(?:-ssl)?\.ted\.com/.+?)\1', webpage) @@ -1212,6 +1457,10 @@ class GenericIE(InfoExtractor): mobj = re.search( r']+?src=(["\'])(?Phttps?://m(?:lb)?\.mlb\.com/shared/video/embed/embed\.html\?.+?)\1', webpage) + if not mobj: + mobj = re.search( + r'data-video-link=["\'](?Phttp://m.mlb.com/video/[^"\']+)', + webpage) if mobj is not None: return self.url_result(mobj.group('url'), 'MLB') @@ -1257,17 +1506,65 @@ class GenericIE(InfoExtractor): if mobj is not None: return self.url_result(mobj.group('url'), 'Pladform') + # Look for Playwire embeds + mobj = re.search( + r']+data-config=(["\'])(?P(?:https?:)?//config\.playwire\.com/.+?)\1', webpage) + if mobj is not None: + return self.url_result(mobj.group('url')) + # Look for 5min embeds mobj = re.search( r']+property="og:video"[^>]+content="https?://embed\.5min\.com/(?P[0-9]+)/?', webpage) if mobj is not None: return self.url_result('5min:%s' % mobj.group('id'), 'FiveMin') + # Look for Crooks and Liars embeds + mobj = re.search( + r'<(?:iframe[^>]+src|param[^>]+value)=(["\'])(?P(?:https?:)?//embed\.crooksandliars\.com/(?:embed|v)/.+?)\1', webpage) + if mobj is not None: + return self.url_result(mobj.group('url')) + # Look for NBC Sports VPlayer embeds nbc_sports_url = NBCSportsVPlayerIE._extract_url(webpage) if nbc_sports_url: return self.url_result(nbc_sports_url, 'NBCSportsVPlayer') + # Look for UDN embeds + mobj = re.search( + r']+src="(?P%s)"' % UDNEmbedIE._VALID_URL, webpage) + if mobj is not None: + return self.url_result( + compat_urlparse.urljoin(url, mobj.group('url')), 'UDNEmbed') + + # Look for Senate ISVP iframe + senate_isvp_url = SenateISVPIE._search_iframe_url(webpage) + if senate_isvp_url: + return self.url_result(senate_isvp_url, 'SenateISVP') + + # Look for Dailymotion Cloud videos + dmcloud_url = DailymotionCloudIE._extract_dmcloud_url(webpage) + if dmcloud_url: + return self.url_result(dmcloud_url, 'DailymotionCloud') + + # Look for OnionStudios embeds + onionstudios_url = OnionStudiosIE._extract_url(webpage) + if onionstudios_url: + return self.url_result(onionstudios_url) + + # Look for SnagFilms embeds + snagfilms_url = SnagFilmsEmbedIE._extract_url(webpage) + if snagfilms_url: + return self.url_result(snagfilms_url) + + # Look for AdobeTVVideo embeds + mobj = re.search( + r']+src=[\'"]((?:https?:)?//video\.tv\.adobe\.com/v/\d+[^"]+)[\'"]', + webpage) + if mobj is not None: + return self.url_result( + self._proto_relative_url(unescapeHTML(mobj.group(1))), + 'AdobeTVVideo') + def check_video(vurl): if YoutubeIE.suitable(vurl): return True @@ -1335,7 +1632,7 @@ class GenericIE(InfoExtractor): if refresh_header: found = re.search(REDIRECT_REGEX, refresh_header) if found: - new_url = found.group(1) + new_url = compat_urlparse.urljoin(url, found.group(1)) self.report_following_redirect(new_url) return { '_type': 'url', @@ -1357,13 +1654,22 @@ class GenericIE(InfoExtractor): # here's a fun little line of code for you: video_id = os.path.splitext(video_id)[0] - entries.append({ - 'id': video_id, - 'url': video_url, - 'uploader': video_uploader, - 'title': video_title, - 'age_limit': age_limit, - }) + if determine_ext(video_url) == 'smil': + entries.append({ + 'id': video_id, + 'formats': self._extract_smil_formats(video_url, video_id), + 'uploader': video_uploader, + 'title': video_title, + 'age_limit': age_limit, + }) + else: + entries.append({ + 'id': video_id, + 'url': video_url, + 'uploader': video_uploader, + 'title': video_title, + 'age_limit': age_limit, + }) if len(entries) == 1: return entries[0]