X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fgeneric.py;h=798c109c6f7dc739ff30de85ad4d64c2072a8e3d;hb=a2252385308898074f5006ed737aeb98bb8b0402;hp=b48ccfc97ba2ca1fae72704cf09d81c807166f08;hpb=5d39176f6de8bab1e019ead7cd497659f3fc1a94;p=youtube-dl diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index b48ccfc97..798c109c6 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -62,6 +62,8 @@ from .digiteka import DigitekaIE from .instagram import InstagramIE from .liveleak import LiveLeakIE from .threeqsdn import ThreeQSDNIE +from .theplatform import ThePlatformIE +from .vessel import VesselIE class GenericIE(InfoExtractor): @@ -717,15 +719,18 @@ class GenericIE(InfoExtractor): }, # Wistia embed { - 'url': 'http://education-portal.com/academy/lesson/north-american-exploration-failed-colonies-of-spain-france-england.html#lesson', - 'md5': '8788b683c777a5cf25621eaf286d0c23', + 'url': 'http://study.com/academy/lesson/north-american-exploration-failed-colonies-of-spain-france-england.html#lesson', + 'md5': '1953f3a698ab51cfc948ed3992a0b7ff', 'info_dict': { - 'id': '1cfaf6b7ea', + 'id': '6e2wtrbdaf', 'ext': 'mov', - 'title': 'md5:51364a8d3d009997ba99656004b5e20d', - 'duration': 643.0, - 'filesize': 182808282, - 'uploader': 'education-portal.com', + 'title': 'paywall_north-american-exploration-failed-colonies-of-spain-france-england', + 'description': 'a Paywall Videos video from Remilon', + 'duration': 644.072, + 'uploader': 'study.com', + 'timestamp': 1459678540, + 'upload_date': '20160403', + 'filesize': 24687186, }, }, { @@ -734,14 +739,30 @@ class GenericIE(InfoExtractor): 'info_dict': { 'id': 'uxjb0lwrcz', 'ext': 'mp4', - 'title': 'Conversation about Hexagonal Rails Part 1 - ThoughtWorks', + 'title': 'Conversation about Hexagonal Rails Part 1', 'description': 'a Martin Fowler video from ThoughtWorks', 'duration': 1715.0, 'uploader': 'thoughtworks.wistia.com', - 'upload_date': '20140603', 'timestamp': 1401832161, + 'upload_date': '20140603', }, }, + # Wistia standard embed (async) + { + 'url': 'https://www.getdrip.com/university/brennan-dunn-drip-workshop/', + 'info_dict': { + 'id': '807fafadvk', + 'ext': 'mp4', + 'title': 'Drip Brennan Dunn Workshop', + 'description': 'a JV Webinars video from getdrip-1', + 'duration': 4986.95, + 'timestamp': 1463607249, + 'upload_date': '20160518', + }, + 'params': { + 'skip_download': True, + } + }, # Soundcloud embed { 'url': 'http://nakedsecurity.sophos.com/2014/10/29/sscc-171-are-you-sure-that-1234-is-a-bad-password-podcast/', @@ -764,6 +785,19 @@ class GenericIE(InfoExtractor): 'title': 'Rosetta #CometLanding webcast HL 10', } }, + # Another Livestream embed, without 'new.' in URL + { + 'url': 'https://www.freespeech.org/', + 'info_dict': { + 'id': '123537347', + 'ext': 'mp4', + 'title': 're:^FSTV [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', + }, + 'params': { + # Live stream + 'skip_download': True, + }, + }, # LazyYT { 'url': 'http://discourse.ubuntu.com/t/unity-8-desktop-mode-windows-on-mir/1986', @@ -848,18 +882,6 @@ class GenericIE(InfoExtractor): 'title': 'EP3S5 - Bon Appétit - Baqueira Mi Corazon !', } }, - # Kaltura embed - { - 'url': 'http://www.monumentalnetwork.com/videos/john-carlson-postgame-2-25-15', - 'info_dict': { - 'id': '1_eergr3h1', - 'ext': 'mp4', - 'upload_date': '20150226', - 'uploader_id': 'MonumentalSports-Kaltura@perfectsensedigital.com', - 'timestamp': int, - 'title': 'John Carlson Postgame 2/25/15', - }, - }, # Kaltura embed (different embed code) { 'url': 'http://www.premierchristianradio.com/Shows/Saturday/Unbelievable/Conference-Videos/Os-Guinness-Is-It-Fools-Talk-Unbelievable-Conference-2014', @@ -885,6 +907,19 @@ class GenericIE(InfoExtractor): 'uploader_id': 'echojecka', }, }, + # Kaltura embed with single quotes + { + 'url': 'http://fod.infobase.com/p_ViewPlaylist.aspx?AssignmentID=NUN8ZY', + 'info_dict': { + 'id': '0_izeg5utt', + 'ext': 'mp4', + 'title': '35871', + 'timestamp': 1355743100, + 'upload_date': '20121217', + 'uploader_id': 'batchUser', + }, + 'add_ie': ['Kaltura'], + }, # Eagle.Platform embed (generic URL) { 'url': 'http://lenta.ru/news/2015/03/06/navalny/', @@ -997,16 +1032,31 @@ class GenericIE(InfoExtractor): 'timestamp': 1389118457, }, }, + # NBC News embed + { + 'url': 'http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html', + 'md5': '1aa589c675898ae6d37a17913cf68d66', + 'info_dict': { + 'id': '701714499682', + 'ext': 'mp4', + 'title': 'PREVIEW: On Assignment: David Letterman', + 'description': 'A preview of Tom Brokaw\'s interview with David Letterman as part of the On Assignment series powered by Dateline. Airs Sunday June 12 at 7/6c.', + }, + }, # UDN embed { - 'url': 'http://www.udn.com/news/story/7314/822787', + 'url': 'https://video.udn.com/news/300346', 'md5': 'fd2060e988c326991037b9aff9df21a6', 'info_dict': { 'id': '300346', 'ext': 'mp4', 'title': 'ä¸ä¸ä¸ç·å¸«è®æ§ å ¨æ ¡å¸«çåæº', 'thumbnail': 're:^https?://.*\.jpg$', - } + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, }, # Ooyala embed { @@ -1174,6 +1224,16 @@ class GenericIE(InfoExtractor): 'uploader': 'Lake8737', } }, + # Duplicated embedded video URLs + { + 'url': 'http://www.hudl.com/athlete/2538180/highlights/149298443', + 'info_dict': { + 'id': '149298443_480_16c25b74_2', + 'ext': 'mp4', + 'title': 'vs. Blue Orange Spring Game', + 'uploader': 'www.hudl.com', + }, + }, ] def report_following_redirect(self, new_url): @@ -1428,7 +1488,8 @@ class GenericIE(InfoExtractor): # Site Name | Video Title # Video Title - Tagline | Site Name # and so on and so forth; it's just not practical - video_title = self._html_search_regex( + video_title = self._og_search_title( + webpage, default=None) or self._html_search_regex( r'(?s)