X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fgeneric.py;h=e645d1bb32d26f1e93ba57c93bd4b211d362d2fa;hb=08f2a92c9c23cf460b00a290b5b3819c7972231b;hp=fbbc79a574ca03f1e483738a726f2fde0bf6b21d;hpb=f076b63821100af17379ed8b470e508846bb8432;p=youtube-dl diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index fbbc79a57..e645d1bb3 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -26,12 +26,15 @@ from ..utils import ( unsmuggle_url, UnsupportedError, url_basename, + xpath_text, ) from .brightcove import BrightcoveIE +from .nbc import NBCSportsVPlayerIE from .ooyala import OoyalaIE from .rutv import RUTVIE from .smotri import SmotriIE from .condenast import CondeNastIE +from .udn import UDNEmbedIE class GenericIE(InfoExtractor): @@ -473,6 +476,7 @@ class GenericIE(InfoExtractor): { 'url': 'http://discourse.ubuntu.com/t/unity-8-desktop-mode-windows-on-mir/1986', 'info_dict': { + 'id': '1986', 'title': 'Unity 8 desktop-mode windows on Mir! - Ubuntu Discourse', }, 'playlist_mincount': 2, @@ -524,6 +528,205 @@ class GenericIE(InfoExtractor): 'upload_date': '20150126', }, 'add_ie': ['Viddler'], + }, + # Libsyn embed + { + 'url': 'http://thedailyshow.cc.com/podcast/episodetwelve', + 'info_dict': { + 'id': '3377616', + 'ext': 'mp3', + 'title': "The Daily Show Podcast without Jon Stewart - Episode 12: Bassem Youssef: Egypt's Jon Stewart", + 'description': 'md5:601cb790edd05908957dae8aaa866465', + 'upload_date': '20150220', + }, + }, + # jwplayer YouTube + { + 'url': 'http://media.nationalarchives.gov.uk/index.php/webinar-using-discovery-national-archives-online-catalogue/', + 'info_dict': { + 'id': 'Mrj4DVp2zeA', + 'ext': 'mp4', + 'upload_date': '20150212', + 'uploader': 'The National Archives UK', + 'description': 'md5:a236581cd2449dd2df4f93412f3f01c6', + 'uploader_id': 'NationalArchives08', + 'title': 'Webinar: Using Discovery, The National Archives’ online catalogue', + }, + }, + # 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', + } + }, + # Zapiks embed + { + 'url': 'http://www.skipass.com/news/116090-bon-appetit-s5ep3-baqueira-mi-cor.html', + 'info_dict': { + 'id': '118046', + 'ext': 'mp4', + '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', + }, + }, + # Eagle.Platform embed (generic URL) + { + 'url': 'http://lenta.ru/news/2015/03/06/navalny/', + 'info_dict': { + 'id': '227304', + 'ext': 'mp4', + 'title': 'Навальный вышел на свободу', + 'description': 'md5:d97861ac9ae77377f3f20eaf9d04b4f5', + 'thumbnail': 're:^https?://.*\.jpg$', + 'duration': 87, + 'view_count': int, + 'age_limit': 0, + }, + }, + # ClipYou (Eagle.Platform) embed (custom URL) + { + 'url': 'http://muz-tv.ru/play/7129/', + 'info_dict': { + 'id': '12820', + 'ext': 'mp4', + 'title': "'O Sole Mio", + 'thumbnail': 're:^https?://.*\.jpg$', + 'duration': 216, + 'view_count': int, + }, + }, + # Pladform embed + { + 'url': 'http://muz-tv.ru/kinozal/view/7400/', + 'info_dict': { + 'id': '100183293', + 'ext': 'mp4', + '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/', + 'md5': '4c6f127a30736b59b3e2c19234ee2bf7', + 'info_dict': { + 'id': '518726732', + 'ext': 'mp4', + 'title': 'Facebook Creates "On This Day" | Crunch Report', + }, + }, + # 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', + } + }, + # 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 + { + 'url': 'http://www.riderfans.com/forum/showthread.php?121827-Freeman&s=e98fa1ea6dc08e886b1678d35212494a', + 'info_dict': { + 'id': 'ln7x1qSThw4k', + 'ext': 'flv', + '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, + } } ] @@ -536,11 +739,24 @@ class GenericIE(InfoExtractor): playlist_desc_el = doc.find('./channel/description') playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text - entries = [{ - '_type': 'url', - 'url': e.find('link').text, - 'title': e.find('title').text, - } for e in doc.findall('./channel/item')] + entries = [] + for it in doc.findall('./channel/item'): + next_url = xpath_text(it, 'link', fatal=False) + if not next_url: + enclosure_nodes = it.findall('./enclosure') + for e in enclosure_nodes: + next_url = e.attrib.get('url') + if next_url: + break + + if not next_url: + continue + + entries.append({ + '_type': 'url', + 'url': next_url, + 'title': it.find('title').text, + }) return { '_type': 'playlist', @@ -769,6 +985,13 @@ class GenericIE(InfoExtractor): 'entries': entries, } + # Look for embedded rtl.nl player + matches = re.findall( + r']+?src=(["\'])(?P(?:https?:)?//player\.vimeo\.com/video/.+?)\1', webpage) @@ -776,7 +999,6 @@ class GenericIE(InfoExtractor): 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) @@ -893,10 +1115,24 @@ class GenericIE(InfoExtractor): if mobj is not None: return self.url_result(mobj.group('url')) + # Look for NYTimes player + mobj = re.search( + r']+src=(["\'])(?P(?:https?:)?//graphics8\.nytimes\.com/bcvideo/[^/]+/iframe/embed\.html.+?)\1>', + webpage) + if mobj is not None: + return self.url_result(mobj.group('url')) + + # Look for Libsyn player + mobj = re.search( + r']+src=(["\'])(?P(?:https?:)?//html5-player\.libsyn\.com/embed/.+?)\1', webpage) + if mobj is not None: + return self.url_result(mobj.group('url')) + # 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')) @@ -1034,7 +1270,12 @@ class GenericIE(InfoExtractor): # Look for embedded sbs.com.au player mobj = re.search( - r']+?src=(["\'])(?Phttps?://(?:www\.)sbs\.com\.au/ondemand/video/single/.+?)\1', + r'''(?x) + (?: + ]+?src= + ) + (["\'])(?Phttps?://(?:www\.)?sbs\.com\.au/ondemand/video/.+?)\1''', webpage) if mobj is not None: return self.url_result(mobj.group('url'), 'SBS') @@ -1064,7 +1305,69 @@ class GenericIE(InfoExtractor): if mobj is not None: return self.url_result(mobj.group('url'), 'Livestream') + # Look for Zapiks embed + mobj = re.search( + r']+src="(?Phttps?://(?:www\.)?zapiks\.fr/index\.php\?.+?)"', webpage) + if mobj is not None: + return self.url_result(mobj.group('url'), 'Zapiks') + + # Look for Kaltura embeds + mobj = re.search( + r"(?s)kWidget\.(?:thumb)?[Ee]mbed\(\{.*?'wid'\s*:\s*'_?(?P[^']+)',.*?'entry_id'\s*:\s*'(?P[^']+)',", webpage) + if mobj is not None: + return self.url_result('kaltura:%(partner_id)s:%(id)s' % mobj.groupdict(), 'Kaltura') + + # Look for Eagle.Platform embeds + mobj = re.search( + r']+src="(?Phttps?://.+?\.media\.eagleplatform\.com/index/player\?.+?)"', webpage) + if mobj is not None: + return self.url_result(mobj.group('url'), 'EaglePlatform') + + # Look for ClipYou (uses Eagle.Platform) embeds + mobj = re.search( + r']+src="https?://(?Pmedia\.clipyou\.ru)/index/player\?.*\brecord_id=(?P\d+).*"', webpage) + if mobj is not None: + return self.url_result('eagleplatform:%(host)s:%(id)s' % mobj.groupdict(), 'EaglePlatform') + + # Look for Pladform embeds + mobj = re.search( + r']+src="(?Phttps?://out\.pladform\.ru/player\?.+?)"', webpage) + 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') + def check_video(vurl): + if YoutubeIE.suitable(vurl): + return True vpath = compat_urlparse.urlparse(vurl).path vext = determine_ext(vpath) return '.' in vpath and vext not in ('swf', 'png', 'jpg', 'srt', 'sbv', 'sub', 'vtt', 'ttml') @@ -1082,7 +1385,8 @@ class GenericIE(InfoExtractor): JWPlayerOptions| jwplayer\s*\(\s*["'][^'"]+["']\s*\)\s*\.setup ) - .*?file\s*:\s*["\'](.*?)["\']''', webpage)) + .*? + ['"]?file['"]?\s*:\s*["\'](.*?)["\']''', webpage)) if not found: # Broaden the search a little bit found = filter_video(re.findall(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)) @@ -1117,10 +1421,16 @@ class GenericIE(InfoExtractor): # HTML5 video found = re.findall(r'(?s).*?]*)?\s+src=["\'](.*?)["\']', webpage) if not found: + REDIRECT_REGEX = r'[0-9]{,2};\s*(?:URL|url)=\'?([^\'"]+)' found = re.search( r'(?i)