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
  28
  29 class GenericIE(InfoExtractor):
  30     IE_DESC = 'Generic downloader that works on some sites'
  31     _VALID_URL = r'.*'
  32     IE_NAME = 'generic'
  33     _TESTS = [
  34         {
  35             'url': 'http://www.hodiho.fr/2013/02/regis-plante-sa-jeep.html',
  36             'file': '13601338388002.mp4',
  37             'md5': '6e15c93721d7ec9e9ca3fdbf07982cfd',
  38             'info_dict': {
  39                 'uploader': 'www.hodiho.fr',
  40                 'title': 'R\u00e9gis plante sa Jeep',
  41             }
  42         },
  43         # bandcamp page with custom domain
  44         {
  45             'add_ie': ['Bandcamp'],
  46             'url': 'http://bronyrock.com/track/the-pony-mash',
  47             'file': '3235767654.mp3',
  48             'info_dict': {
  49                 'title': 'The Pony Mash',
  50                 'uploader': 'M_Pallante',
  51             },
  52             'skip': 'There is a limit of 200 free downloads / month for the test song',
  53         },
  54         # embedded brightcove video
  55         # it also tests brightcove videos that need to set the 'Referer' in the
  56         # http requests
  57         {
  58             'add_ie': ['Brightcove'],
  59             'url': 'http://www.bfmtv.com/video/bfmbusiness/cours-bourse/cours-bourse-l-analyse-technique-154522/',
  60             'info_dict': {
  61                 'id': '2765128793001',
  62                 'ext': 'mp4',
  63                 'title': 'Le cours de bourse : l’analyse technique',
  64                 'description': 'md5:7e9ad046e968cb2d1114004aba466fd9',
  65                 'uploader': 'BFM BUSINESS',
  66             },
  67             'params': {
  68                 'skip_download': True,
  69             },
  70         },
  71         {
  72             # https://github.com/rg3/youtube-dl/issues/2253
  73             'url': 'http://bcove.me/i6nfkrc3',
  74             'file': '3101154703001.mp4',
  75             'md5': '0ba9446db037002366bab3b3eb30c88c',
  76             'info_dict': {
  77                 'title': 'Still no power',
  78                 'uploader': 'thestar.com',
  79                 '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.',
  80             },
  81             'add_ie': ['Brightcove'],
  82         },
  83         # Direct link to a video
  84         {
  85             'url': 'http://media.w3.org/2010/05/sintel/trailer.mp4',
  86             'md5': '67d406c2bcb6af27fa886f31aa934bbe',
  87             'info_dict': {
  88                 'id': 'trailer',
  89                 'ext': 'mp4',
  90                 'title': 'trailer',
  91                 'upload_date': '20100513',
  92             }
  93         },
  94         # ooyala video
  95         {
  96             'url': 'http://www.rollingstone.com/music/videos/norwegian-dj-cashmere-cat-goes-spartan-on-with-me-premiere-20131219',
  97             'md5': '5644c6ca5d5782c1d0d350dad9bd840c',
  98             'info_dict': {
  99                 'id': 'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ',
 100                 'ext': 'mp4',
 101                 'title': '2cc213299525360.mov',  # that's what we get
 102             },
 103         },
 104         # google redirect
 105         {
 106             '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',
 107             'info_dict': {
 108                 'id': 'cmQHVoWB5FY',
 109                 'ext': 'mp4',
 110                 'upload_date': '20130224',
 111                 'uploader_id': 'TheVerge',
 112                 '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.',
 113                 'uploader': 'The Verge',
 114                 'title': 'First Firefox OS phones side-by-side',
 115             },
 116             'params': {
 117                 'skip_download': False,
 118             }
 119         },
 120         # embed.ly video
 121         {
 122             'url': 'http://www.tested.com/science/weird/460206-tested-grinding-coffee-2000-frames-second/',
 123             'info_dict': {
 124                 'id': '9ODmcdjQcHQ',
 125                 'ext': 'mp4',
 126                 'title': 'Tested: Grinding Coffee at 2000 Frames Per Second',
 127                 'upload_date': '20140225',
 128                 'description': 'md5:06a40fbf30b220468f1e0957c0f558ff',
 129                 'uploader': 'Tested',
 130                 'uploader_id': 'testedcom',
 131             },
 132             # No need to test YoutubeIE here
 133             'params': {
 134                 'skip_download': True,
 135             },
 136         },
 137         # funnyordie embed
 138         {
 139             'url': 'http://www.theguardian.com/world/2014/mar/11/obama-zach-galifianakis-between-two-ferns',
 140             'md5': '7cf780be104d40fea7bae52eed4a470e',
 141             'info_dict': {
 142                 'id': '18e820ec3f',
 143                 'ext': 'mp4',
 144                 'title': 'Between Two Ferns with Zach Galifianakis: President Barack Obama',
 145                 'description': 'Episode 18: President Barack Obama sits down with Zach Galifianakis for his most memorable interview yet.',
 146             }
 147         },
 148         # nowvideo embed hidden behind percent encoding
 149         {
 150             'url': 'http://www.waoanime.tv/the-super-dimension-fortress-macross-episode-1/',
 151             'md5': '2baf4ddd70f697d94b1c18cf796d5107',
 152             'info_dict': {
 153                 'id': '06e53103ca9aa',
 154                 'ext': 'flv',
 155                 'title': 'Macross Episode 001  Watch Macross Episode 001 onl',
 156                 'description': 'No description',
 157             },
 158         }
 159     ]
 160
 161     def report_download_webpage(self, video_id):
 162         """Report webpage download."""
 163         if not self._downloader.params.get('test', False):
 164             self._downloader.report_warning('Falling back on generic information extractor.')
 165         super(GenericIE, self).report_download_webpage(video_id)
 166
 167     def report_following_redirect(self, new_url):
 168         """Report information extraction."""
 169         self._downloader.to_screen('[redirect] Following redirect to %s' % new_url)
 170
 171     def _send_head(self, url):
 172         """Check if it is a redirect, like url shorteners, in case return the new url."""
 173
 174         class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
 175             """
 176             Subclass the HTTPRedirectHandler to make it use our
 177             HEADRequest also on the redirected URL
 178             """
 179             def redirect_request(self, req, fp, code, msg, headers, newurl):
 180                 if code in (301, 302, 303, 307):
 181                     newurl = newurl.replace(' ', '%20')
 182                     newheaders = dict((k,v) for k,v in req.headers.items()
 183                                       if k.lower() not in ("content-length", "content-type"))
 184                     return HEADRequest(newurl,
 185                                        headers=newheaders,
 186                                        origin_req_host=req.get_origin_req_host(),
 187                                        unverifiable=True)
 188                 else:
 189                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 190
 191         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 192             """
 193             Fallback to GET if HEAD is not allowed (405 HTTP error)
 194             """
 195             def http_error_405(self, req, fp, code, msg, headers):
 196                 fp.read()
 197                 fp.close()
 198
 199                 newheaders = dict((k,v) for k,v in req.headers.items()
 200                                   if k.lower() not in ("content-length", "content-type"))
 201                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 202                                                  headers=newheaders,
 203                                                  origin_req_host=req.get_origin_req_host(),
 204                                                  unverifiable=True))
 205
 206         # Build our opener
 207         opener = compat_urllib_request.OpenerDirector()
 208         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 209                         HTTPMethodFallback, HEADRedirectHandler,
 210                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 211             opener.add_handler(handler())
 212
 213         response = opener.open(HEADRequest(url))
 214         if response is None:
 215             raise ExtractorError('Invalid URL protocol')
 216         return response
 217
 218     def _extract_rss(self, url, video_id, doc):
 219         playlist_title = doc.find('./channel/title').text
 220         playlist_desc_el = doc.find('./channel/description')
 221         playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text
 222
 223         entries = [{
 224             '_type': 'url',
 225             'url': e.find('link').text,
 226             'title': e.find('title').text,
 227         } for e in doc.findall('./channel/item')]
 228
 229         return {
 230             '_type': 'playlist',
 231             'id': url,
 232             'title': playlist_title,
 233             'description': playlist_desc,
 234             'entries': entries,
 235         }
 236
 237     def _real_extract(self, url):
 238         parsed_url = compat_urlparse.urlparse(url)
 239         if not parsed_url.scheme:
 240             default_search = self._downloader.params.get('default_search')
 241             if default_search is None:
 242                 default_search = 'auto'
 243
 244             if default_search == 'auto':
 245                 if '/' in url:
 246                     self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 247                     return self.url_result('http://' + url)
 248                 else:
 249                     return self.url_result('ytsearch:' + url)
 250             else:
 251                 assert ':' in default_search
 252                 return self.url_result(default_search + url)
 253         video_id = os.path.splitext(url.rstrip('/').split('/')[-1])[0]
 254
 255         self.to_screen('%s: Requesting header' % video_id)
 256
 257         try:
 258             response = self._send_head(url)
 259
 260             # Check for redirect
 261             new_url = response.geturl()
 262             if url != new_url:
 263                 self.report_following_redirect(new_url)
 264                 return self.url_result(new_url)
 265
 266             # Check for direct link to a video
 267             content_type = response.headers.get('Content-Type', '')
 268             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 269             if m:
 270                 upload_date = response.headers.get('Last-Modified')
 271                 if upload_date:
 272                     upload_date = unified_strdate(upload_date)
 273                 return {
 274                     'id': video_id,
 275                     'title': os.path.splitext(url_basename(url))[0],
 276                     'formats': [{
 277                         'format_id': m.group('format_id'),
 278                         'url': url,
 279                         'vcodec': 'none' if m.group('type') == 'audio' else None
 280                     }],
 281                     'upload_date': upload_date,
 282                 }
 283
 284         except compat_urllib_error.HTTPError:
 285             # This may be a stupid server that doesn't like HEAD, our UA, or so
 286             pass
 287
 288         try:
 289             webpage = self._download_webpage(url, video_id)
 290         except ValueError:
 291             # since this is the last-resort InfoExtractor, if
 292             # this error is thrown, it'll be thrown here
 293             raise ExtractorError('Failed to download URL: %s' % url)
 294
 295         self.report_extraction(video_id)
 296
 297         # Is it an RSS feed?
 298         try:
 299             doc = parse_xml(webpage)
 300             if doc.tag == 'rss':
 301                 return self._extract_rss(url, video_id, doc)
 302         except compat_xml_parse_error:
 303             pass
 304
 305         # Sometimes embedded video player is hidden behind percent encoding
 306         # (e.g. https://github.com/rg3/youtube-dl/issues/2448)
 307         # Unescaping the whole page allows to handle those cases in a generic way
 308         webpage = compat_urllib_parse.unquote(webpage)
 309
 310         # it's tempting to parse this further, but you would
 311         # have to take into account all the variations like
 312         #   Video Title - Site Name
 313         #   Site Name | Video Title
 314         #   Video Title - Tagline | Site Name
 315         # and so on and so forth; it's just not practical
 316         video_title = self._html_search_regex(
 317             r'(?s)<title>(.*?)</title>', webpage, 'video title',
 318             default='video')
 319
 320         # video uploader is domain name
 321         video_uploader = self._search_regex(
 322             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 323
 324         # Look for BrightCove:
 325         bc_urls = BrightcoveIE._extract_brightcove_urls(webpage)
 326         if bc_urls:
 327             self.to_screen('Brightcove video detected.')
 328             entries = [{
 329                 '_type': 'url',
 330                 'url': smuggle_url(bc_url, {'Referer': url}),
 331                 'ie_key': 'Brightcove'
 332             } for bc_url in bc_urls]
 333
 334             return {
 335                 '_type': 'playlist',
 336                 'title': video_title,
 337                 'id': video_id,
 338                 'entries': entries,
 339             }
 340
 341         # Look for embedded (iframe) Vimeo player
 342         mobj = re.search(
 343             r'<iframe[^>]+?src="((?:https?:)?//player\.vimeo\.com/video/.+?)"', webpage)
 344         if mobj:
 345             player_url = unescapeHTML(mobj.group(1))
 346             surl = smuggle_url(player_url, {'Referer': url})
 347             return self.url_result(surl, 'Vimeo')
 348
 349         # Look for embedded (swf embed) Vimeo player
 350         mobj = re.search(
 351             r'<embed[^>]+?src="(https?://(?:www\.)?vimeo\.com/moogaloop\.swf.+?)"', webpage)
 352         if mobj:
 353             return self.url_result(mobj.group(1), 'Vimeo')
 354
 355         # Look for embedded YouTube player
 356         matches = re.findall(r'''(?x)
 357             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 358             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 359                 (?:embed|v)/.+?)
 360             \1''', webpage)
 361         if matches:
 362             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 363                      for tuppl in matches]
 364             return self.playlist_result(
 365                 urlrs, playlist_id=video_id, playlist_title=video_title)
 366
 367         # Look for embedded Dailymotion player
 368         matches = re.findall(
 369             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 370         if matches:
 371             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 372                      for tuppl in matches]
 373             return self.playlist_result(
 374                 urlrs, playlist_id=video_id, playlist_title=video_title)
 375
 376         # Look for embedded Wistia player
 377         match = re.search(
 378             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 379         if match:
 380             return {
 381                 '_type': 'url_transparent',
 382                 'url': unescapeHTML(match.group('url')),
 383                 'ie_key': 'Wistia',
 384                 'uploader': video_uploader,
 385                 'title': video_title,
 386                 'id': video_id,
 387             }
 388
 389         # Look for embedded blip.tv player
 390         mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
 391         if mobj:
 392             return self.url_result('http://blip.tv/a/a-'+mobj.group(1), 'BlipTV')
 393         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9]+)', webpage)
 394         if mobj:
 395             return self.url_result(mobj.group(1), 'BlipTV')
 396
 397         # Look for Bandcamp pages with custom domain
 398         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 399         if mobj is not None:
 400             burl = unescapeHTML(mobj.group(1))
 401             # Don't set the extractor because it can be a track url or an album
 402             return self.url_result(burl)
 403
 404         # Look for embedded Vevo player
 405         mobj = re.search(
 406             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 407         if mobj is not None:
 408             return self.url_result(mobj.group('url'))
 409
 410         # Look for Ooyala videos
 411         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 412         if mobj is not None:
 413             return OoyalaIE._build_url_result(mobj.group(1))
 414
 415         # Look for Aparat videos
 416         mobj = re.search(r'<iframe src="(http://www\.aparat\.com/video/[^"]+)"', webpage)
 417         if mobj is not None:
 418             return self.url_result(mobj.group(1), 'Aparat')
 419
 420         # Look for MPORA videos
 421         mobj = re.search(r'<iframe .*?src="(http://mpora\.(?:com|de)/videos/[^"]+)"', webpage)
 422         if mobj is not None:
 423             return self.url_result(mobj.group(1), 'Mpora')
 424
 425         # Look for embedded NovaMov player
 426         mobj = re.search(
 427             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?novamov\.com/embed\.php.+?)\1', webpage)
 428         if mobj is not None:
 429             return self.url_result(mobj.group('url'), 'NovaMov')
 430
 431         # Look for embedded NowVideo player
 432         mobj = re.search(
 433             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?nowvideo\.(?:ch|sx|eu)/embed\.php.+?)\1', webpage)
 434         if mobj is not None:
 435             return self.url_result(mobj.group('url'), 'NowVideo')
 436
 437         # Look for embedded Facebook player
 438         mobj = re.search(
 439             r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
 440         if mobj is not None:
 441             return self.url_result(mobj.group('url'), 'Facebook')
 442
 443         # Look for embedded VK player
 444         mobj = re.search(r'<iframe[^>]+?src=(["\'])(?P<url>https?://vk\.com/video_ext\.php.+?)\1', webpage)
 445         if mobj is not None:
 446             return self.url_result(mobj.group('url'), 'VK')
 447
 448         # Look for embedded Huffington Post player
 449         mobj = re.search(
 450             r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed\.live\.huffingtonpost\.com/.+?)\1', webpage)
 451         if mobj is not None:
 452             return self.url_result(mobj.group('url'), 'HuffPost')
 453
 454         # Look for embed.ly
 455         mobj = re.search(r'class=["\']embedly-card["\'][^>]href=["\'](?P<url>[^"\']+)', webpage)
 456         if mobj is not None:
 457             return self.url_result(mobj.group('url'))
 458         mobj = re.search(r'class=["\']embedly-embed["\'][^>]src=["\'][^"\']*url=(?P<url>[^&]+)', webpage)
 459         if mobj is not None:
 460             return self.url_result(compat_urllib_parse.unquote(mobj.group('url')))
 461
 462         # Look for funnyordie embed
 463         matches = re.findall(r'<iframe[^>]+?src="(https?://(?:www\.)?funnyordie\.com/embed/[^"]+)"', webpage)
 464         if matches:
 465             urlrs = [self.url_result(unescapeHTML(eurl), 'FunnyOrDie')
 466                      for eurl in matches]
 467             return self.playlist_result(
 468                 urlrs, playlist_id=video_id, playlist_title=video_title)
 469
 470         # Start with something easy: JW Player in SWFObject
 471         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 472         if mobj is None:
 473             # Look for gorilla-vid style embedding
 474             mobj = re.search(r'(?s)(?:jw_plugins|JWPlayerOptions).*?file\s*:\s*["\'](.*?)["\']', webpage)
 475         if mobj is None:
 476             # Broaden the search a little bit
 477             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 478         if mobj is None:
 479             # Broaden the search a little bit: JWPlayer JS loader
 480             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
 481         if mobj is None:
 482             # Try to find twitter cards info
 483             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 484         if mobj is None:
 485             # We look for Open Graph info:
 486             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 487             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 488             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 489             if m_video_type is not None:
 490                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 491         if mobj is None:
 492             # HTML5 video
 493             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 494         if mobj is None:
 495             mobj = re.search(
 496                 r'(?i)<meta\s+(?=(?:[a-z-]+="[^"]+"\s+)*http-equiv="refresh")'
 497                 r'(?:[a-z-]+="[^"]+"\s+)*?content="[0-9]{,2};url=\'([^\']+)\'"',
 498                 webpage)
 499             if mobj:
 500                 new_url = mobj.group(1)
 501                 self.report_following_redirect(new_url)
 502                 return {
 503                     '_type': 'url',
 504                     'url': new_url,
 505                 }
 506         if mobj is None:
 507             raise ExtractorError('Unsupported URL: %s' % url)
 508
 509         # It's possible that one of the regexes
 510         # matched, but returned an empty group:
 511         if mobj.group(1) is None:
 512             raise ExtractorError('Did not find a valid video URL at %s' % url)
 513
 514         video_url = mobj.group(1)
 515         video_url = compat_urlparse.urljoin(url, video_url)
 516         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 517
 518         # Sometimes, jwplayer extraction will result in a YouTube URL
 519         if YoutubeIE.suitable(video_url):
 520             return self.url_result(video_url, 'Youtube')
 521
 522         # here's a fun little line of code for you:
 523         video_id = os.path.splitext(video_id)[0]
 524
 525         return {
 526             'id': video_id,
 527             'url': video_url,
 528             'uploader': video_uploader,
 529             'title': video_title,
 530         }