_ Git - youtube-dl/blob - youtube_dl/extractor/generic.py

   1 # encoding: utf-8
   2
   3 from __future__ import unicode_literals
   4
   5 import os
   6 import re
   7
   8 from .common import InfoExtractor
   9 from .youtube import YoutubeIE
  10 from ..utils import (
  11     compat_urllib_error,
  12     compat_urllib_parse,
  13     compat_urllib_request,
  14     compat_urlparse,
  15     compat_xml_parse_error,
  16
  17     ExtractorError,
  18     HEADRequest,
  19     parse_xml,
  20     smuggle_url,
  21     unescapeHTML,
  22     unified_strdate,
  23     url_basename,
  24 )
  25 from .brightcove import BrightcoveIE
  26 from .ooyala import OoyalaIE
  27 from .rutv import RUTVIE
  28
  29
  30 class GenericIE(InfoExtractor):
  31     IE_DESC = 'Generic downloader that works on some sites'
  32     _VALID_URL = r'.*'
  33     IE_NAME = 'generic'
  34     _TESTS = [
  35         {
  36             'url': 'http://www.hodiho.fr/2013/02/regis-plante-sa-jeep.html',
  37             'file': '13601338388002.mp4',
  38             'md5': '6e15c93721d7ec9e9ca3fdbf07982cfd',
  39             'info_dict': {
  40                 'uploader': 'www.hodiho.fr',
  41                 'title': 'R\u00e9gis plante sa Jeep',
  42             }
  43         },
  44         # bandcamp page with custom domain
  45         {
  46             'add_ie': ['Bandcamp'],
  47             'url': 'http://bronyrock.com/track/the-pony-mash',
  48             'file': '3235767654.mp3',
  49             'info_dict': {
  50                 'title': 'The Pony Mash',
  51                 'uploader': 'M_Pallante',
  52             },
  53             'skip': 'There is a limit of 200 free downloads / month for the test song',
  54         },
  55         # embedded brightcove video
  56         # it also tests brightcove videos that need to set the 'Referer' in the
  57         # http requests
  58         {
  59             'add_ie': ['Brightcove'],
  60             'url': 'http://www.bfmtv.com/video/bfmbusiness/cours-bourse/cours-bourse-l-analyse-technique-154522/',
  61             'info_dict': {
  62                 'id': '2765128793001',
  63                 'ext': 'mp4',
  64                 'title': 'Le cours de bourse : l’analyse technique',
  65                 'description': 'md5:7e9ad046e968cb2d1114004aba466fd9',
  66                 'uploader': 'BFM BUSINESS',
  67             },
  68             'params': {
  69                 'skip_download': True,
  70             },
  71         },
  72         {
  73             # https://github.com/rg3/youtube-dl/issues/2253
  74             'url': 'http://bcove.me/i6nfkrc3',
  75             'file': '3101154703001.mp4',
  76             'md5': '0ba9446db037002366bab3b3eb30c88c',
  77             'info_dict': {
  78                 'title': 'Still no power',
  79                 'uploader': 'thestar.com',
  80                 'description': 'Mississauga resident David Farmer is still out of power as a result of the ice storm a month ago. To keep the house warm, Farmer cuts wood from his property for a wood burning stove downstairs.',
  81             },
  82             'add_ie': ['Brightcove'],
  83         },
  84         # Direct link to a video
  85         {
  86             'url': 'http://media.w3.org/2010/05/sintel/trailer.mp4',
  87             'md5': '67d406c2bcb6af27fa886f31aa934bbe',
  88             'info_dict': {
  89                 'id': 'trailer',
  90                 'ext': 'mp4',
  91                 'title': 'trailer',
  92                 'upload_date': '20100513',
  93             }
  94         },
  95         # ooyala video
  96         {
  97             'url': 'http://www.rollingstone.com/music/videos/norwegian-dj-cashmere-cat-goes-spartan-on-with-me-premiere-20131219',
  98             'md5': '5644c6ca5d5782c1d0d350dad9bd840c',
  99             'info_dict': {
 100                 'id': 'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ',
 101                 'ext': 'mp4',
 102                 'title': '2cc213299525360.mov',  # that's what we get
 103             },
 104         },
 105         # google redirect
 106         {
 107             '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',
 108             'info_dict': {
 109                 'id': 'cmQHVoWB5FY',
 110                 'ext': 'mp4',
 111                 'upload_date': '20130224',
 112                 'uploader_id': 'TheVerge',
 113                 'description': 'Chris Ziegler takes a look at the Alcatel OneTouch Fire and the ZTE Open; two of the first Firefox OS handsets to be officially announced.',
 114                 'uploader': 'The Verge',
 115                 'title': 'First Firefox OS phones side-by-side',
 116             },
 117             'params': {
 118                 'skip_download': False,
 119             }
 120         },
 121         # embed.ly video
 122         {
 123             'url': 'http://www.tested.com/science/weird/460206-tested-grinding-coffee-2000-frames-second/',
 124             'info_dict': {
 125                 'id': '9ODmcdjQcHQ',
 126                 'ext': 'mp4',
 127                 'title': 'Tested: Grinding Coffee at 2000 Frames Per Second',
 128                 'upload_date': '20140225',
 129                 'description': 'md5:06a40fbf30b220468f1e0957c0f558ff',
 130                 'uploader': 'Tested',
 131                 'uploader_id': 'testedcom',
 132             },
 133             # No need to test YoutubeIE here
 134             'params': {
 135                 'skip_download': True,
 136             },
 137         },
 138         # funnyordie embed
 139         {
 140             'url': 'http://www.theguardian.com/world/2014/mar/11/obama-zach-galifianakis-between-two-ferns',
 141             'md5': '7cf780be104d40fea7bae52eed4a470e',
 142             'info_dict': {
 143                 'id': '18e820ec3f',
 144                 'ext': 'mp4',
 145                 'title': 'Between Two Ferns with Zach Galifianakis: President Barack Obama',
 146                 'description': 'Episode 18: President Barack Obama sits down with Zach Galifianakis for his most memorable interview yet.',
 147             },
 148         },
 149         # RUTV embed
 150         {
 151             'url': 'http://www.rg.ru/2014/03/15/reg-dfo/anklav-anons.html',
 152             'info_dict': {
 153                 'id': '776940',
 154                 'ext': 'mp4',
 155                 'title': 'Охотское море стало целиком российским',
 156                 'description': 'md5:5ed62483b14663e2a95ebbe115eb8f43',
 157             },
 158             'params': {
 159                 # m3u8 download
 160                 'skip_download': True,
 161             },
 162         }
 163     ]
 164
 165     def report_download_webpage(self, video_id):
 166         """Report webpage download."""
 167         if not self._downloader.params.get('test', False):
 168             self._downloader.report_warning('Falling back on generic information extractor.')
 169         super(GenericIE, self).report_download_webpage(video_id)
 170
 171     def report_following_redirect(self, new_url):
 172         """Report information extraction."""
 173         self._downloader.to_screen('[redirect] Following redirect to %s' % new_url)
 174
 175     def _send_head(self, url):
 176         """Check if it is a redirect, like url shorteners, in case return the new url."""
 177
 178         class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
 179             """
 180             Subclass the HTTPRedirectHandler to make it use our
 181             HEADRequest also on the redirected URL
 182             """
 183             def redirect_request(self, req, fp, code, msg, headers, newurl):
 184                 if code in (301, 302, 303, 307):
 185                     newurl = newurl.replace(' ', '%20')
 186                     newheaders = dict((k,v) for k,v in req.headers.items()
 187                                       if k.lower() not in ("content-length", "content-type"))
 188                     try:
 189                         # This function was deprecated in python 3.3 and removed in 3.4
 190                         origin_req_host = req.get_origin_req_host()
 191                     except AttributeError:
 192                         origin_req_host = req.origin_req_host
 193                     return HEADRequest(newurl,
 194                                        headers=newheaders,
 195                                        origin_req_host=origin_req_host,
 196                                        unverifiable=True)
 197                 else:
 198                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 199
 200         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 201             """
 202             Fallback to GET if HEAD is not allowed (405 HTTP error)
 203             """
 204             def http_error_405(self, req, fp, code, msg, headers):
 205                 fp.read()
 206                 fp.close()
 207
 208                 newheaders = dict((k,v) for k,v in req.headers.items()
 209                                   if k.lower() not in ("content-length", "content-type"))
 210                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 211                                                  headers=newheaders,
 212                                                  origin_req_host=req.get_origin_req_host(),
 213                                                  unverifiable=True))
 214
 215         # Build our opener
 216         opener = compat_urllib_request.OpenerDirector()
 217         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 218                         HTTPMethodFallback, HEADRedirectHandler,
 219                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 220             opener.add_handler(handler())
 221
 222         response = opener.open(HEADRequest(url))
 223         if response is None:
 224             raise ExtractorError('Invalid URL protocol')
 225         return response
 226
 227     def _extract_rss(self, url, video_id, doc):
 228         playlist_title = doc.find('./channel/title').text
 229         playlist_desc_el = doc.find('./channel/description')
 230         playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text
 231
 232         entries = [{
 233             '_type': 'url',
 234             'url': e.find('link').text,
 235             'title': e.find('title').text,
 236         } for e in doc.findall('./channel/item')]
 237
 238         return {
 239             '_type': 'playlist',
 240             'id': url,
 241             'title': playlist_title,
 242             'description': playlist_desc,
 243             'entries': entries,
 244         }
 245
 246     def _real_extract(self, url):
 247         parsed_url = compat_urlparse.urlparse(url)
 248         if not parsed_url.scheme:
 249             default_search = self._downloader.params.get('default_search')
 250             if default_search is None:
 251                 default_search = 'auto'
 252
 253             if default_search == 'auto':
 254                 if '/' in url:
 255                     self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 256                     return self.url_result('http://' + url)
 257                 else:
 258                     return self.url_result('ytsearch:' + url)
 259             else:
 260                 assert ':' in default_search
 261                 return self.url_result(default_search + url)
 262         video_id = os.path.splitext(url.rstrip('/').split('/')[-1])[0]
 263
 264         self.to_screen('%s: Requesting header' % video_id)
 265
 266         try:
 267             response = self._send_head(url)
 268
 269             # Check for redirect
 270             new_url = response.geturl()
 271             if url != new_url:
 272                 self.report_following_redirect(new_url)
 273                 return self.url_result(new_url)
 274
 275             # Check for direct link to a video
 276             content_type = response.headers.get('Content-Type', '')
 277             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 278             if m:
 279                 upload_date = response.headers.get('Last-Modified')
 280                 if upload_date:
 281                     upload_date = unified_strdate(upload_date)
 282                 return {
 283                     'id': video_id,
 284                     'title': os.path.splitext(url_basename(url))[0],
 285                     'formats': [{
 286                         'format_id': m.group('format_id'),
 287                         'url': url,
 288                         'vcodec': 'none' if m.group('type') == 'audio' else None
 289                     }],
 290                     'upload_date': upload_date,
 291                 }
 292
 293         except compat_urllib_error.HTTPError:
 294             # This may be a stupid server that doesn't like HEAD, our UA, or so
 295             pass
 296
 297         try:
 298             webpage = self._download_webpage(url, video_id)
 299         except ValueError:
 300             # since this is the last-resort InfoExtractor, if
 301             # this error is thrown, it'll be thrown here
 302             raise ExtractorError('Failed to download URL: %s' % url)
 303
 304         self.report_extraction(video_id)
 305
 306         # Is it an RSS feed?
 307         try:
 308             doc = parse_xml(webpage)
 309             if doc.tag == 'rss':
 310                 return self._extract_rss(url, video_id, doc)
 311         except compat_xml_parse_error:
 312             pass
 313
 314         # it's tempting to parse this further, but you would
 315         # have to take into account all the variations like
 316         #   Video Title - Site Name
 317         #   Site Name | Video Title
 318         #   Video Title - Tagline | Site Name
 319         # and so on and so forth; it's just not practical
 320         video_title = self._html_search_regex(
 321             r'(?s)<title>(.*?)</title>', webpage, 'video title',
 322             default='video')
 323
 324         # video uploader is domain name
 325         video_uploader = self._search_regex(
 326             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 327
 328         # Look for BrightCove:
 329         bc_urls = BrightcoveIE._extract_brightcove_urls(webpage)
 330         if bc_urls:
 331             self.to_screen('Brightcove video detected.')
 332             entries = [{
 333                 '_type': 'url',
 334                 'url': smuggle_url(bc_url, {'Referer': url}),
 335                 'ie_key': 'Brightcove'
 336             } for bc_url in bc_urls]
 337
 338             return {
 339                 '_type': 'playlist',
 340                 'title': video_title,
 341                 'id': video_id,
 342                 'entries': entries,
 343             }
 344
 345         # Look for embedded (iframe) Vimeo player
 346         mobj = re.search(
 347             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//player\.vimeo\.com/video/.+?)\1', webpage)
 348         if mobj:
 349             player_url = unescapeHTML(mobj.group('url'))
 350             surl = smuggle_url(player_url, {'Referer': url})
 351             return self.url_result(surl, 'Vimeo')
 352
 353         # Look for embedded (swf embed) Vimeo player
 354         mobj = re.search(
 355             r'<embed[^>]+?src="(https?://(?:www\.)?vimeo\.com/moogaloop\.swf.+?)"', webpage)
 356         if mobj:
 357             return self.url_result(mobj.group(1), 'Vimeo')
 358
 359         # Look for embedded YouTube player
 360         matches = re.findall(r'''(?x)
 361             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 362             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 363                 (?:embed|v)/.+?)
 364             \1''', webpage)
 365         if matches:
 366             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 367                      for tuppl in matches]
 368             return self.playlist_result(
 369                 urlrs, playlist_id=video_id, playlist_title=video_title)
 370
 371         # Look for embedded Dailymotion player
 372         matches = re.findall(
 373             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 374         if matches:
 375             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 376                      for tuppl in matches]
 377             return self.playlist_result(
 378                 urlrs, playlist_id=video_id, playlist_title=video_title)
 379
 380         # Look for embedded Wistia player
 381         match = re.search(
 382             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 383         if match:
 384             return {
 385                 '_type': 'url_transparent',
 386                 'url': unescapeHTML(match.group('url')),
 387                 'ie_key': 'Wistia',
 388                 'uploader': video_uploader,
 389                 'title': video_title,
 390                 'id': video_id,
 391             }
 392
 393         # Look for embedded blip.tv player
 394         mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
 395         if mobj:
 396             return self.url_result('http://blip.tv/a/a-'+mobj.group(1), 'BlipTV')
 397         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9]+)', webpage)
 398         if mobj:
 399             return self.url_result(mobj.group(1), 'BlipTV')
 400
 401         # Look for Bandcamp pages with custom domain
 402         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 403         if mobj is not None:
 404             burl = unescapeHTML(mobj.group(1))
 405             # Don't set the extractor because it can be a track url or an album
 406             return self.url_result(burl)
 407
 408         # Look for embedded Vevo player
 409         mobj = re.search(
 410             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 411         if mobj is not None:
 412             return self.url_result(mobj.group('url'))
 413
 414         # Look for Ooyala videos
 415         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 416         if mobj is not None:
 417             return OoyalaIE._build_url_result(mobj.group(1))
 418
 419         # Look for Aparat videos
 420         mobj = re.search(r'<iframe src="(http://www\.aparat\.com/video/[^"]+)"', webpage)
 421         if mobj is not None:
 422             return self.url_result(mobj.group(1), 'Aparat')
 423
 424         # Look for MPORA videos
 425         mobj = re.search(r'<iframe .*?src="(http://mpora\.(?:com|de)/videos/[^"]+)"', webpage)
 426         if mobj is not None:
 427             return self.url_result(mobj.group(1), 'Mpora')
 428
 429         # Look for embedded NovaMov player
 430         mobj = re.search(
 431             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?novamov\.com/embed\.php.+?)\1', webpage)
 432         if mobj is not None:
 433             return self.url_result(mobj.group('url'), 'NovaMov')
 434
 435         # Look for embedded NowVideo player
 436         mobj = re.search(
 437             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?nowvideo\.(?:ch|sx|eu)/embed\.php.+?)\1', webpage)
 438         if mobj is not None:
 439             return self.url_result(mobj.group('url'), 'NowVideo')
 440
 441         # Look for embedded Facebook player
 442         mobj = re.search(
 443             r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
 444         if mobj is not None:
 445             return self.url_result(mobj.group('url'), 'Facebook')
 446
 447         # Look for embedded VK player
 448         mobj = re.search(r'<iframe[^>]+?src=(["\'])(?P<url>https?://vk\.com/video_ext\.php.+?)\1', webpage)
 449         if mobj is not None:
 450             return self.url_result(mobj.group('url'), 'VK')
 451
 452         # Look for embedded Huffington Post player
 453         mobj = re.search(
 454             r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed\.live\.huffingtonpost\.com/.+?)\1', webpage)
 455         if mobj is not None:
 456             return self.url_result(mobj.group('url'), 'HuffPost')
 457
 458         # Look for embed.ly
 459         mobj = re.search(r'class=["\']embedly-card["\'][^>]href=["\'](?P<url>[^"\']+)', webpage)
 460         if mobj is not None:
 461             return self.url_result(mobj.group('url'))
 462         mobj = re.search(r'class=["\']embedly-embed["\'][^>]src=["\'][^"\']*url=(?P<url>[^&]+)', webpage)
 463         if mobj is not None:
 464             return self.url_result(compat_urllib_parse.unquote(mobj.group('url')))
 465
 466         # Look for funnyordie embed
 467         matches = re.findall(r'<iframe[^>]+?src="(https?://(?:www\.)?funnyordie\.com/embed/[^"]+)"', webpage)
 468         if matches:
 469             urlrs = [self.url_result(unescapeHTML(eurl), 'FunnyOrDie')
 470                      for eurl in matches]
 471             return self.playlist_result(
 472                 urlrs, playlist_id=video_id, playlist_title=video_title)
 473
 474         # Look for embedded RUTV player
 475         rutv_url = RUTVIE._extract_url(webpage)
 476         if rutv_url:
 477             return self.url_result(rutv_url, 'RUTV')
 478
 479         # Start with something easy: JW Player in SWFObject
 480         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 481         if mobj is None:
 482             # Look for gorilla-vid style embedding
 483             mobj = re.search(r'(?s)(?:jw_plugins|JWPlayerOptions).*?file\s*:\s*["\'](.*?)["\']', webpage)
 484         if mobj is None:
 485             # Broaden the search a little bit
 486             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 487         if mobj is None:
 488             # Broaden the search a little bit: JWPlayer JS loader
 489             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
 490
 491         # Look for embedded TED player
 492         mobj = re.search(
 493             r'<iframe[^>]+?src=(["\'])(?P<url>http://embed\.ted\.com/.+?)\1', webpage)
 494         if mobj is not None:
 495             return self.url_result(mobj.group('url'), 'TED')
 496
 497         if mobj is None:
 498             # Try to find twitter cards info
 499             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 500         if mobj is None:
 501             # We look for Open Graph info:
 502             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 503             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 504             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 505             if m_video_type is not None:
 506                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 507         if mobj is None:
 508             # HTML5 video
 509             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 510         if mobj is None:
 511             mobj = re.search(
 512                 r'(?i)<meta\s+(?=(?:[a-z-]+="[^"]+"\s+)*http-equiv="refresh")'
 513                 r'(?:[a-z-]+="[^"]+"\s+)*?content="[0-9]{,2};url=\'([^\']+)\'"',
 514                 webpage)
 515             if mobj:
 516                 new_url = mobj.group(1)
 517                 self.report_following_redirect(new_url)
 518                 return {
 519                     '_type': 'url',
 520                     'url': new_url,
 521                 }
 522         if mobj is None:
 523             raise ExtractorError('Unsupported URL: %s' % url)
 524
 525         # It's possible that one of the regexes
 526         # matched, but returned an empty group:
 527         if mobj.group(1) is None:
 528             raise ExtractorError('Did not find a valid video URL at %s' % url)
 529
 530         video_url = mobj.group(1)
 531         video_url = compat_urlparse.urljoin(url, video_url)
 532         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 533
 534         # Sometimes, jwplayer extraction will result in a YouTube URL
 535         if YoutubeIE.suitable(video_url):
 536             return self.url_result(video_url, 'Youtube')
 537
 538         # here's a fun little line of code for you:
 539         video_id = os.path.splitext(video_id)[0]
 540
 541         return {
 542             'id': video_id,
 543             'url': video_url,
 544             'uploader': video_uploader,
 545             'title': video_title,
 546         }