X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fgeneric.py;h=355067a509fc197c10a2085bad058d4459729bc9;hb=86b7c00adca578b36138b165b0add5978972917e;hp=f2c577f98c07f30f76d832d1bcaecaa5c0f0bc7a;hpb=73cf76a93fe48240bf82b1685b1403f05b793ebf;p=youtube-dl diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index f2c577f98..355067a50 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -22,6 +22,9 @@ from ..utils import ( HEADRequest, is_html, js_to_json, + KNOWN_EXTENSIONS, + merge_dicts, + mimetype2ext, orderedSet, sanitized_Request, smuggle_url, @@ -36,11 +39,15 @@ from .brightcove import ( BrightcoveLegacyIE, BrightcoveNewIE, ) +from .nexx import ( + NexxIE, + NexxEmbedIE, +) from .nbc import NBCSportsVPlayerIE from .ooyala import OoyalaIE from .rutv import RUTVIE from .tvc import TVCIE -from .sportbox import SportBoxEmbedIE +from .sportbox import SportBoxIE from .smotri import SmotriIE from .myvi import MyviIE from .condenast import CondeNastIE @@ -52,11 +59,13 @@ from .xhamster import XHamsterEmbedIE from .tnaflix import TNAFlixNetworkEmbedIE from .drtuber import DrTuberIE from .redtube import RedTubeIE +from .tube8 import Tube8IE +from .mofosex import MofosexEmbedIE +from .spankwire import SpankwireIE +from .youporn import YouPornIE from .vimeo import VimeoIE -from .dailymotion import ( - DailymotionIE, - DailymotionCloudIE, -) +from .dailymotion import DailymotionIE +from .dailymail import DailyMailIE from .onionstudios import OnionStudiosIE from .viewlift import ViewLiftEmbedIE from .mtv import MTVServicesEmbeddedIE @@ -71,11 +80,10 @@ from .instagram import InstagramIE from .liveleak import LiveLeakIE from .threeqsdn import ThreeQSDNIE from .theplatform import ThePlatformIE -from .vessel import VesselIE from .kaltura import KalturaIE from .eagleplatform import EaglePlatformIE from .facebook import FacebookIE -from .soundcloud import SoundcloudIE +from .soundcloud import SoundcloudEmbedIE from .tunein import TuneInBaseIE from .vbox7 import Vbox7IE from .dbtv import DBTVIE @@ -83,7 +91,6 @@ from .piksel import PikselIE from .videa import VideaIE from .twentymin import TwentyMinutenIE from .ustream import UstreamIE -from .openload import OpenloadIE from .videopress import VideoPressIE from .rutube import RutubeIE from .limelight import LimelightBaseIE @@ -92,6 +99,26 @@ from .washingtonpost import WashingtonPostIE from .wistia import WistiaIE from .mediaset import MediasetIE from .joj import JojIE +from .megaphone import MegaphoneIE +from .vzaar import VzaarIE +from .channel9 import Channel9IE +from .vshare import VShareIE +from .mediasite import MediasiteIE +from .springboardplatform import SpringboardPlatformIE +from .yapfiles import YapFilesIE +from .vice import ViceIE +from .xfileshare import XFileShareIE +from .cloudflarestream import CloudflareStreamIE +from .peertube import PeerTubeIE +from .teachable import TeachableIE +from .indavideo import IndavideoEmbedIE +from .apa import APAIE +from .foxnews import FoxNewsIE +from .viqeo import ViqeoIE +from .expressen import ExpressenIE +from .zype import ZypeIE +from .odnoklassniki import OdnoklassnikiIE +from .kinja import KinjaEmbedIE class GenericIE(InfoExtractor): @@ -176,6 +203,16 @@ class GenericIE(InfoExtractor): 'title': 'pdv_maddow_netcast_m4v-02-27-2015-201624', } }, + # RSS feed with enclosures and unsupported link URLs + { + 'url': 'http://www.hellointernet.fm/podcast?format=rss', + 'info_dict': { + 'id': 'http://www.hellointernet.fm/podcast?format=rss', + 'description': 'CGP Grey and Brady Haran talk about YouTube, life, work, whatever.', + 'title': 'Hello Internet', + }, + 'playlist_mincount': 100, + }, # SMIL from http://videolectures.net/promogram_igor_mekjavic_eng { 'url': 'http://videolectures.net/promogram_igor_mekjavic_eng/video/1/smil.xml', @@ -396,7 +433,7 @@ class GenericIE(InfoExtractor): }, }, { - # https://github.com/rg3/youtube-dl/issues/2253 + # https://github.com/ytdl-org/youtube-dl/issues/2253 'url': 'http://bcove.me/i6nfkrc3', 'md5': '0ba9446db037002366bab3b3eb30c88c', 'info_dict': { @@ -421,7 +458,7 @@ class GenericIE(InfoExtractor): }, }, { - # https://github.com/rg3/youtube-dl/issues/3541 + # https://github.com/ytdl-org/youtube-dl/issues/3541 'add_ie': ['BrightcoveLegacy'], 'url': 'http://www.kijk.nl/sbs6/leermijvrouwenkennen/videos/jqMiXKAYan2S/aflevering-1', 'info_dict': { @@ -569,6 +606,19 @@ class GenericIE(InfoExtractor): }, 'skip': 'movie expired', }, + # ooyala video embedded with http://player.ooyala.com/static/v4/production/latest/core.min.js + { + 'url': 'http://wnep.com/2017/07/22/steampunk-fest-comes-to-honesdale/', + 'info_dict': { + 'id': 'lwYWYxYzE6V5uJMjNGyKtwwiw9ZJD7t2', + 'ext': 'mp4', + 'title': 'Steampunk Fest Comes to Honesdale', + 'duration': 43.276, + }, + 'params': { + 'skip_download': True, + } + }, # embed.ly video { 'url': 'http://www.tested.com/science/weird/460206-tested-grinding-coffee-2000-frames-second/', @@ -760,6 +810,20 @@ class GenericIE(InfoExtractor): }, 'add_ie': ['Dailymotion'], }, + # DailyMail embed + { + 'url': 'http://www.bumm.sk/krimi/2017/07/05/biztonsagi-kamera-buktatta-le-az-agg-ferfit-utlegelo-apolot', + 'info_dict': { + 'id': '1495629', + 'ext': 'mp4', + 'title': 'Care worker punches elderly dementia patient in head 11 times', + 'description': 'md5:3a743dee84e57e48ec68bf67113199a5', + }, + 'add_ie': ['DailyMail'], + 'params': { + 'skip_download': True, + }, + }, # YouTube embed { 'url': 'http://www.badzine.de/ansicht/datum/2014/06/09/so-funktioniert-die-neue-englische-badminton-liga.html', @@ -858,7 +922,7 @@ class GenericIE(InfoExtractor): } }, # Multiple brightcove videos - # https://github.com/rg3/youtube-dl/issues/2283 + # https://github.com/ytdl-org/youtube-dl/issues/2283 { 'url': 'http://www.newyorker.com/online/blogs/newsdesk/2014/01/always-never-nuclear-command-and-control.html', 'info_dict': { @@ -1054,23 +1118,24 @@ class GenericIE(InfoExtractor): 'ext': 'mp4', 'upload_date': '20150212', 'uploader': 'The National Archives UK', - 'description': 'md5:a236581cd2449dd2df4f93412f3f01c6', + 'description': 'md5:8078af856dca76edc42910b61273dbbf', 'uploader_id': 'NationalArchives08', 'title': 'Webinar: Using Discovery, The National Archivesâ online catalogue', }, }, # jwplayer rtmp { - 'url': 'http://www.suffolk.edu/sjc/', + 'url': 'http://www.suffolk.edu/sjc/live.php', 'info_dict': { - 'id': 'sjclive', + 'id': 'live', 'ext': 'flv', 'title': 'Massachusetts Supreme Judicial Court Oral Arguments', 'uploader': 'www.suffolk.edu', }, 'params': { 'skip_download': True, - } + }, + 'skip': 'Only has video a few mornings per month, see http://www.suffolk.edu/sjc/', }, # Complex jwplayer { @@ -1079,6 +1144,7 @@ class GenericIE(InfoExtractor): 'id': 'videos', 'ext': 'mp4', 'title': 'king machine trailer 1', + 'description': 'Browse King Machine videos & audio for sweet media. Your eyes will thank you.', 'thumbnail': r're:^https?://.*\.jpg$', }, }, @@ -1096,13 +1162,55 @@ class GenericIE(InfoExtractor): 'skip_download': True, } }, + { + # JWPlatform iframe + 'url': 'https://www.mediaite.com/tv/dem-senator-claims-gary-cohn-faked-a-bad-connection-during-trump-call-to-get-him-off-the-phone/', + 'md5': 'ca00a040364b5b439230e7ebfd02c4e9', + 'info_dict': { + 'id': 'O0c5JcKT', + 'ext': 'mp4', + 'upload_date': '20171122', + 'timestamp': 1511366290, + 'title': 'Dem Senator Claims Gary Cohn Faked a Bad Connection During Trump Call to Get Him Off the Phone', + }, + 'add_ie': [JWPlatformIE.ie_key()], + }, + { + # Video.js embed, multiple formats + 'url': 'http://ortcam.com/solidworks-ÑÑок-6-наÑÑÑойка-ÑеÑÑежа_33f9b7351.html', + 'info_dict': { + 'id': 'yygqldloqIk', + 'ext': 'mp4', + 'title': 'SolidWorks. УÑок 6 ÐаÑÑÑойка ÑеÑÑежа', + 'description': 'md5:baf95267792646afdbf030e4d06b2ab3', + 'upload_date': '20130314', + 'uploader': 'PROÑÑое3D', + 'uploader_id': 'PROstoe3D', + }, + 'params': { + 'skip_download': True, + }, + }, + { + # Video.js embed, single format + 'url': 'https://www.vooplayer.com/v3/watch/watch.php?v=NzgwNTg=', + 'info_dict': { + 'id': 'watch', + 'ext': 'mp4', + 'title': 'Step 1 - Good Foundation', + 'description': 'md5:d1e7ff33a29fc3eb1673d6c270d344f4', + }, + 'params': { + 'skip_download': True, + }, + }, # rtl.nl embed { 'url': 'http://www.rtlnieuws.nl/nieuws/buitenland/aanslagen-kopenhagen', 'playlist_mincount': 5, 'info_dict': { 'id': 'aanslagen-kopenhagen', - 'title': 'Aanslagen Kopenhagen | RTL Nieuws', + 'title': 'Aanslagen Kopenhagen', } }, # Zapiks embed @@ -1135,7 +1243,7 @@ class GenericIE(InfoExtractor): 'title': '35871', 'timestamp': 1355743100, 'upload_date': '20121217', - 'uploader_id': 'batchUser', + 'uploader_id': 'cplapp@learn360.com', }, 'add_ie': ['Kaltura'], }, @@ -1186,23 +1294,38 @@ class GenericIE(InfoExtractor): }, 'add_ie': ['Kaltura'], }, - # EaglePlatform embed (generic URL) { - 'url': 'http://lenta.ru/news/2015/03/06/navalny/', - # Not checking MD5 as sometimes the direct HTTP link results in 404 and HLS is used + # Kaltura iframe embed, more sophisticated + 'url': 'http://www.cns.nyu.edu/~eero/math-tools/Videos/lecture-05sep2017.html', 'info_dict': { - 'id': '227304', + 'id': '1_9gzouybz', 'ext': 'mp4', - 'title': 'ÐавалÑнÑй вÑÑел на ÑвободÑ', - 'description': 'md5:d97861ac9ae77377f3f20eaf9d04b4f5', - 'thumbnail': r're:^https?://.*\.jpg$', - 'duration': 87, - 'view_count': int, - 'age_limit': 0, + 'title': 'lecture-05sep2017', + 'description': 'md5:40f347d91fd4ba047e511c5321064b49', + 'upload_date': '20170913', + 'uploader_id': 'eps2', + 'timestamp': 1505340777, }, 'params': { 'skip_download': True, }, + 'add_ie': ['Kaltura'], + }, + { + # meta twitter:player + 'url': 'http://thechive.com/2017/12/08/all-i-want-for-christmas-is-more-twerk/', + 'info_dict': { + 'id': '0_01b42zps', + 'ext': 'mp4', + 'title': 'Main Twerk (Video)', + 'upload_date': '20171208', + 'uploader_id': 'sebastian.salinas@thechive.com', + 'timestamp': 1512713057, + }, + 'params': { + 'skip_download': True, + }, + 'add_ie': ['Kaltura'], }, # referrer protected EaglePlatform embed { @@ -1234,6 +1357,7 @@ class GenericIE(InfoExtractor): 'params': { 'skip_download': True, }, + 'skip': 'This video is unavailable.', }, # Pladform embed { @@ -1247,6 +1371,7 @@ class GenericIE(InfoExtractor): 'duration': 694, 'age_limit': 0, }, + 'skip': 'HTTP Error 404: Not Found', }, # Playwire embed { @@ -1267,17 +1392,14 @@ class GenericIE(InfoExtractor): 'id': '518726732', 'ext': 'mp4', 'title': 'Facebook Creates "On This Day" | Crunch Report', + 'description': 'Amazon updates Fire TV line, Tesla\'s Model X spotted in the wild', + 'timestamp': 1427237531, + 'uploader': 'Crunch Report', + 'upload_date': '20150324', }, - }, - # SVT embed - { - 'url': 'http://www.svt.se/sport/ishockey/jagr-tacklar-giroux-under-intervjun', - 'info_dict': { - 'id': '2900353', - 'ext': 'flv', - 'title': 'Här trycker Jagr till Giroux (under SVT-intervjun)', - 'duration': 27, - 'age_limit': 0, + 'params': { + # m3u8 download + 'skip_download': True, }, }, # Crooks and Liars embed @@ -1318,16 +1440,20 @@ class GenericIE(InfoExtractor): 'upload_date': '20140107', 'timestamp': 1389118457, }, + 'skip': 'Invalid Page URL', }, # NBC News embed { 'url': 'http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html', 'md5': '1aa589c675898ae6d37a17913cf68d66', 'info_dict': { - 'id': '701714499682', + 'id': 'x_dtl_oa_LettermanliftPR_160608', 'ext': 'mp4', - 'title': 'PREVIEW: On Assignment: David Letterman', + 'title': 'David Letterman: A Preview', '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.', + 'upload_date': '20160609', + 'timestamp': 1465431544, + 'uploader': 'NBCU-NEWS', }, }, # UDN embed @@ -1344,21 +1470,7 @@ class GenericIE(InfoExtractor): # m3u8 download 'skip_download': True, }, - }, - # 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', - 'duration': 191.933, - }, - 'params': { - # m3u8 downloads - 'skip_download': True, - } + 'expected_warnings': ['Failed to parse JSON Expecting value'], }, # Brightcove URL in single quotes { @@ -1375,32 +1487,18 @@ class GenericIE(InfoExtractor): 'timestamp': 1432570283, }, }, - # Dailymotion Cloud video - { - 'url': 'http://replay.publicsenat.fr/vod/le-debat/florent-kolandjian,dominique-cena,axel-decourtye,laurence-abeille,bruno-parmentier/175910', - 'md5': 'dcaf23ad0c67a256f4278bce6e0bae38', - 'info_dict': { - 'id': 'x2uy8t3', - 'ext': 'mp4', - 'title': 'Sauvons les abeilles ! - Le débat', - 'description': 'md5:d9082128b1c5277987825d684939ca26', - 'thumbnail': r're:^https?://.*\.jpe?g$', - 'timestamp': 1434970506, - 'upload_date': '20150622', - 'uploader': 'Public Sénat', - 'uploader_id': 'xa9gza', - } - }, - # OnionStudios embed + # Kinja embed { 'url': 'http://www.clickhole.com/video/dont-understand-bitcoin-man-will-mumble-explanatio-2537', 'info_dict': { - 'id': '2855', + 'id': '106351', 'ext': 'mp4', 'title': 'Donât Understand Bitcoin? This Man Will Mumble An Explanation At You', + 'description': 'Migrated from OnionStudios', 'thumbnail': r're:^https?://.*\.jpe?g$', - 'uploader': 'ClickHole', - 'uploader_id': 'clickhole', + 'uploader': 'clickhole', + 'upload_date': '20150527', + 'timestamp': 1432744860, } }, # SnagFilms embed @@ -1485,14 +1583,27 @@ class GenericIE(InfoExtractor): # LiveLeak embed { 'url': 'http://www.wykop.pl/link/3088787/', - 'md5': 'ace83b9ed19b21f68e1b50e844fdf95d', + 'md5': '7619da8c820e835bef21a1efa2a0fc71', 'info_dict': { 'id': '874_1459135191', 'ext': 'mp4', 'title': 'Man shows poor quality of new apartment building', 'description': 'The wall is like a sand pile.', 'uploader': 'Lake8737', - } + }, + 'add_ie': [LiveLeakIE.ie_key()], + }, + # Another LiveLeak embed pattern (#13336) + { + 'url': 'https://milo.yiannopoulos.net/2017/06/concealed-carry-robbery/', + 'info_dict': { + 'id': '2eb_1496309988', + 'ext': 'mp4', + 'title': 'Thief robs place where everyone was armed', + 'description': 'md5:694d73ee79e535953cf2488562288eee', + 'uploader': 'brazilwtf', + }, + 'add_ie': [LiveLeakIE.ie_key()], }, # Duplicated embedded video URLs { @@ -1597,6 +1708,15 @@ class GenericIE(InfoExtractor): }, 'add_ie': ['Kaltura'], }, + { + # multiple kaltura embeds, nsfw + 'url': 'https://www.quartier-rouge.be/prive/femmes/kamila-avec-video-jaime-sadomie.html', + 'info_dict': { + 'id': 'kamila-avec-video-jaime-sadomie', + 'title': "Kamila avec vÃdeo âJ'aime sadomieâ", + }, + 'playlist_count': 8, + }, { # Non-standard Vimeo embed 'url': 'https://openclassrooms.com/courses/understanding-the-web', @@ -1735,6 +1855,21 @@ class GenericIE(InfoExtractor): }, 'playlist_mincount': 5, }, + { + # Limelight embed (LimelightPlayerUtil.embed) + 'url': 'https://tv5.ca/videos?v=xuu8qowr291ri', + 'info_dict': { + 'id': '95d035dc5c8a401588e9c0e6bd1e9c92', + 'ext': 'mp4', + 'title': '07448641', + 'timestamp': 1499890639, + 'upload_date': '20170712', + }, + 'params': { + 'skip_download': True, + }, + 'add_ie': ['LimelightMedia'], + }, { 'url': 'http://kron4.com/2017/04/28/standoff-with-walnut-creek-murder-suspect-ends-with-arrest/', 'info_dict': { @@ -1791,6 +1926,231 @@ class GenericIE(InfoExtractor): 'title': 'СÑÐ°Ñ Ðамин: «ÐÑ Ð½Ð°ÑÑÑили девÑÑвенноÑÑÑ ÐÑемлÑ»', }, }, + { + # vzaar embed + 'url': 'http://help.vzaar.com/article/165-embedding-video', + 'md5': '7e3919d9d2620b89e3e00bec7fe8c9d4', + 'info_dict': { + 'id': '8707641', + 'ext': 'mp4', + 'title': 'Building A Business Online: Principal Chairs Q & A', + }, + }, + { + # multiple HTML5 videos on one page + 'url': 'https://www.paragon-software.com/home/rk-free/keyscenarios.html', + 'info_dict': { + 'id': 'keyscenarios', + 'title': 'Rescue Kit 14 Free Edition - Getting started', + }, + 'playlist_count': 4, + }, + { + # vshare embed + 'url': 'https://youtube-dl-demo.neocities.org/vshare.html', + 'md5': '17b39f55b5497ae8b59f5fbce8e35886', + 'info_dict': { + 'id': '0f64ce6', + 'title': 'vl14062007715967', + 'ext': 'mp4', + } + }, + { + 'url': 'http://www.heidelberg-laureate-forum.org/blog/video/lecture-friday-september-23-2016-sir-c-antony-r-hoare/', + 'md5': 'aecd089f55b1cb5a59032cb049d3a356', + 'info_dict': { + 'id': '90227f51a80c4d8f86c345a7fa62bd9a1d', + 'ext': 'mp4', + 'title': 'Lecture: Friday, September 23, 2016 - Sir Tony Hoare', + 'description': 'md5:5a51db84a62def7b7054df2ade403c6c', + 'timestamp': 1474354800, + 'upload_date': '20160920', + } + }, + { + 'url': 'http://www.kidzworld.com/article/30935-trolls-the-beat-goes-on-interview-skylar-astin-and-amanda-leighton', + 'info_dict': { + 'id': '1731611', + 'ext': 'mp4', + 'title': 'Official Trailer | TROLLS: THE BEAT GOES ON!', + 'description': 'md5:eb5f23826a027ba95277d105f248b825', + 'timestamp': 1516100691, + 'upload_date': '20180116', + }, + 'params': { + 'skip_download': True, + }, + 'add_ie': [SpringboardPlatformIE.ie_key()], + }, + { + 'url': 'https://www.youtube.com/shared?ci=1nEzmT-M4fU', + 'info_dict': { + 'id': 'uPDB5I9wfp8', + 'ext': 'webm', + 'title': 'Pocoyo: 90 minutos de episódios completos Português para crianças - PARTE 3', + 'description': 'md5:d9e4d9346a2dfff4c7dc4c8cec0f546d', + 'upload_date': '20160219', + 'uploader': 'Pocoyo - Português (BR)', + 'uploader_id': 'PocoyoBrazil', + }, + 'add_ie': [YoutubeIE.ie_key()], + 'params': { + 'skip_download': True, + }, + }, + { + 'url': 'https://www.yapfiles.ru/show/1872528/690b05d3054d2dbe1e69523aa21bb3b1.mp4.html', + 'info_dict': { + 'id': 'vMDE4NzI1Mjgt690b', + 'ext': 'mp4', + 'title': 'ÐоÑÑÑа', + }, + 'add_ie': [YapFilesIE.ie_key()], + 'params': { + 'skip_download': True, + }, + }, + { + # CloudflareStream embed + 'url': 'https://www.cloudflare.com/products/cloudflare-stream/', + 'info_dict': { + 'id': '31c9291ab41fac05471db4e73aa11717', + 'ext': 'mp4', + 'title': '31c9291ab41fac05471db4e73aa11717', + }, + 'add_ie': [CloudflareStreamIE.ie_key()], + 'params': { + 'skip_download': True, + }, + }, + { + # PeerTube embed + 'url': 'https://joinpeertube.org/fr/home/', + 'info_dict': { + 'id': 'home', + 'title': 'Reprenez le contrôle de vos vidéos ! #JoinPeertube', + }, + 'playlist_count': 2, + }, + { + # Indavideo embed + 'url': 'https://streetkitchen.hu/receptek/igy_kell_otthon_hamburgert_sutni/', + 'info_dict': { + 'id': '1693903', + 'ext': 'mp4', + 'title': 'Ãgy kell otthon hamburgert sütni', + 'description': 'md5:f5a730ecf900a5c852e1e00540bbb0f7', + 'timestamp': 1426330212, + 'upload_date': '20150314', + 'uploader': 'StreetKitchen', + 'uploader_id': '546363', + }, + 'add_ie': [IndavideoEmbedIE.ie_key()], + 'params': { + 'skip_download': True, + }, + }, + { + # APA embed via JWPlatform embed + 'url': 'http://www.vol.at/blue-man-group/5593454', + 'info_dict': { + 'id': 'jjv85FdZ', + 'ext': 'mp4', + 'title': '"Blau ist mysteriös": Die Blue Man Group im Interview', + 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e', + 'thumbnail': r're:^https?://.*\.jpg$', + 'duration': 254, + 'timestamp': 1519211149, + 'upload_date': '20180221', + }, + 'params': { + 'skip_download': True, + }, + }, + { + 'url': 'http://share-videos.se/auto/video/83645793?uid=13', + 'md5': 'b68d276de422ab07ee1d49388103f457', + 'info_dict': { + 'id': '83645793', + 'title': 'Lock up and get excited', + 'ext': 'mp4' + }, + 'skip': 'TODO: fix nested playlists processing in tests', + }, + { + # Viqeo embeds + 'url': 'https://viqeo.tv/', + 'info_dict': { + 'id': 'viqeo', + 'title': 'All-new video platform', + }, + 'playlist_count': 6, + }, + { + # Squarespace video embed, 2019-08-28 + 'url': 'http://ootboxford.com', + 'info_dict': { + 'id': 'Tc7b_JGdZfw', + 'title': 'Out of the Blue, at Childish Things 10', + 'ext': 'mp4', + 'description': 'md5:a83d0026666cf5ee970f8bd1cfd69c7f', + 'uploader_id': 'helendouglashouse', + 'uploader': 'Helen & Douglas House', + 'upload_date': '20140328', + }, + 'params': { + 'skip_download': True, + }, + }, + { + # Zype embed + 'url': 'https://www.cookscountry.com/episode/554-smoky-barbecue-favorites', + 'info_dict': { + 'id': '5b400b834b32992a310622b9', + 'ext': 'mp4', + 'title': 'Smoky Barbecue Favorites', + 'thumbnail': r're:^https?://.*\.jpe?g', + 'description': 'md5:5ff01e76316bd8d46508af26dc86023b', + 'upload_date': '20170909', + 'timestamp': 1504915200, + }, + 'add_ie': [ZypeIE.ie_key()], + 'params': { + 'skip_download': True, + }, + }, + { + # videojs embed + 'url': 'https://video.sibnet.ru/shell.php?videoid=3422904', + 'info_dict': { + 'id': 'shell', + 'ext': 'mp4', + 'title': 'ÐоÑÑавÑик пиÑÑÑ ÑпÑоÑил ÑазÑеÑÐµÐ½Ð¸Ñ ÑÑгÑаÑÑ Ð½Ð° ÑоÑÑепиано', + 'description': 'md5:89209cdc587dab1e4a090453dbaa2cb1', + 'thumbnail': r're:^https?://.*\.jpg$', + }, + 'params': { + 'skip_download': True, + }, + 'expected_warnings': ['Failed to download MPD manifest'], + }, + { + # DailyMotion embed with DM.player + 'url': 'https://www.beinsports.com/us/copa-del-rey/video/the-locker-room-valencia-beat-barca-in-copa/1203804', + 'info_dict': { + 'id': 'k6aKkGHd9FJs4mtJN39', + 'ext': 'mp4', + 'title': 'The Locker Room: Valencia Beat Barca In Copa del Rey Final', + 'description': 'This video is private.', + 'uploader_id': 'x1jf30l', + 'uploader': 'beIN SPORTS USA', + 'upload_date': '20190528', + 'timestamp': 1559062971, + }, + 'params': { + 'skip_download': True, + }, + }, # { # # TODO: find another test # # http://schema.org/VideoObject @@ -1821,13 +2181,15 @@ class GenericIE(InfoExtractor): entries = [] for it in doc.findall('./channel/item'): - next_url = xpath_text(it, 'link', fatal=False) + next_url = None + enclosure_nodes = it.findall('./enclosure') + for e in enclosure_nodes: + next_url = e.attrib.get('url') + if next_url: + break + if not next_url: - enclosure_nodes = it.findall('./enclosure') - for e in enclosure_nodes: - next_url = e.attrib.get('url') - if next_url: - break + next_url = xpath_text(it, 'link', fatal=False) if not next_url: continue @@ -1885,10 +2247,7 @@ class GenericIE(InfoExtractor): def _real_extract(self, url): if url.startswith('//'): - return { - '_type': 'url', - 'url': self.http_scheme() + url, - } + return self.url_result(self.http_scheme() + url) parsed_url = compat_urlparse.urlparse(url) if not parsed_url.scheme: @@ -1897,7 +2256,7 @@ class GenericIE(InfoExtractor): default_search = 'fixup_error' if default_search in ('auto', 'auto_warning', 'fixup_error'): - if '/' in url: + if re.match(r'^[^\s/]+\.[^\s/]+/', url): self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http') return self.url_result('http://' + url) elif default_search != 'fixup_error': @@ -2037,10 +2396,14 @@ class GenericIE(InfoExtractor): self._sort_formats(smil['formats']) return smil elif doc.tag == '{http://xspf.org/ns/0/}playlist': - return self.playlist_result(self._parse_xspf(doc, video_id), video_id) + return self.playlist_result( + self._parse_xspf( + doc, video_id, xspf_url=url, + xspf_base_url=full_response.geturl()), + video_id) elif re.match(r'(?i)^(?:{[^}]+})?MPD$', doc.tag): info_dict['formats'] = self._parse_mpd_formats( - doc, video_id, + doc, mpd_base_url=full_response.geturl().rpartition('/')[0], mpd_url=url) self._sort_formats(info_dict['formats']) @@ -2058,10 +2421,16 @@ class GenericIE(InfoExtractor): return camtasia_res # Sometimes embedded video player is hidden behind percent encoding - # (e.g. https://github.com/rg3/youtube-dl/issues/2448) + # (e.g. https://github.com/ytdl-org/youtube-dl/issues/2448) # Unescaping the whole page allows to handle those cases in a generic way webpage = compat_urllib_parse_unquote(webpage) + # Unescape squarespace embeds to be detected by generic extractor, + # see https://github.com/ytdl-org/youtube-dl/issues/21294 + webpage = re.sub( + r'