_ 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
  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     ]
 149
 150     def report_download_webpage(self, video_id):
 151         """Report webpage download."""
 152         if not self._downloader.params.get('test', False):
 153             self._downloader.report_warning('Falling back on generic information extractor.')
 154         super(GenericIE, self).report_download_webpage(video_id)
 155
 156     def report_following_redirect(self, new_url):
 157         """Report information extraction."""
 158         self._downloader.to_screen('[redirect] Following redirect to %s' % new_url)
 159
 160     def _send_head(self, url):
 161         """Check if it is a redirect, like url shorteners, in case return the new url."""
 162
 163         class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
 164             """
 165             Subclass the HTTPRedirectHandler to make it use our
 166             HEADRequest also on the redirected URL
 167             """
 168             def redirect_request(self, req, fp, code, msg, headers, newurl):
 169                 if code in (301, 302, 303, 307):
 170                     newurl = newurl.replace(' ', '%20')
 171                     newheaders = dict((k,v) for k,v in req.headers.items()
 172                                       if k.lower() not in ("content-length", "content-type"))
 173                     return HEADRequest(newurl,
 174                                        headers=newheaders,
 175                                        origin_req_host=req.get_origin_req_host(),
 176                                        unverifiable=True)
 177                 else:
 178                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 179
 180         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 181             """
 182             Fallback to GET if HEAD is not allowed (405 HTTP error)
 183             """
 184             def http_error_405(self, req, fp, code, msg, headers):
 185                 fp.read()
 186                 fp.close()
 187
 188                 newheaders = dict((k,v) for k,v in req.headers.items()
 189                                   if k.lower() not in ("content-length", "content-type"))
 190                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 191                                                  headers=newheaders,
 192                                                  origin_req_host=req.get_origin_req_host(),
 193                                                  unverifiable=True))
 194
 195         # Build our opener
 196         opener = compat_urllib_request.OpenerDirector()
 197         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 198                         HTTPMethodFallback, HEADRedirectHandler,
 199                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 200             opener.add_handler(handler())
 201
 202         response = opener.open(HEADRequest(url))
 203         if response is None:
 204             raise ExtractorError('Invalid URL protocol')
 205         return response
 206
 207     def _extract_rss(self, url, video_id, doc):
 208         playlist_title = doc.find('./channel/title').text
 209         playlist_desc_el = doc.find('./channel/description')
 210         playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text
 211
 212         entries = [{
 213             '_type': 'url',
 214             'url': e.find('link').text,
 215             'title': e.find('title').text,
 216         } for e in doc.findall('./channel/item')]
 217
 218         return {
 219             '_type': 'playlist',
 220             'id': url,
 221             'title': playlist_title,
 222             'description': playlist_desc,
 223             'entries': entries,
 224         }
 225
 226     def _real_extract(self, url):
 227         parsed_url = compat_urlparse.urlparse(url)
 228         if not parsed_url.scheme:
 229             default_search = self._downloader.params.get('default_search')
 230             if default_search is None:
 231                 default_search = 'auto'
 232
 233             if default_search == 'auto':
 234                 if '/' in url:
 235                     self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 236                     return self.url_result('http://' + url)
 237                 else:
 238                     return self.url_result('ytsearch:' + url)
 239             else:
 240                 assert ':' in default_search
 241                 return self.url_result(default_search + url)
 242         video_id = os.path.splitext(url.rstrip('/').split('/')[-1])[0]
 243
 244         self.to_screen('%s: Requesting header' % video_id)
 245
 246         try:
 247             response = self._send_head(url)
 248
 249             # Check for redirect
 250             new_url = response.geturl()
 251             if url != new_url:
 252                 self.report_following_redirect(new_url)
 253                 return self.url_result(new_url)
 254
 255             # Check for direct link to a video
 256             content_type = response.headers.get('Content-Type', '')
 257             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 258             if m:
 259                 upload_date = response.headers.get('Last-Modified')
 260                 if upload_date:
 261                     upload_date = unified_strdate(upload_date)
 262                 return {
 263                     'id': video_id,
 264                     'title': os.path.splitext(url_basename(url))[0],
 265                     'formats': [{
 266                         'format_id': m.group('format_id'),
 267                         'url': url,
 268                         'vcodec': 'none' if m.group('type') == 'audio' else None
 269                     }],
 270                     'upload_date': upload_date,
 271                 }
 272
 273         except compat_urllib_error.HTTPError:
 274             # This may be a stupid server that doesn't like HEAD, our UA, or so
 275             pass
 276
 277         try:
 278             webpage = self._download_webpage(url, video_id)
 279         except ValueError:
 280             # since this is the last-resort InfoExtractor, if
 281             # this error is thrown, it'll be thrown here
 282             raise ExtractorError('Failed to download URL: %s' % url)
 283
 284         self.report_extraction(video_id)
 285
 286         # Is it an RSS feed?
 287         try:
 288             doc = parse_xml(webpage)
 289             if doc.tag == 'rss':
 290                 return self._extract_rss(url, video_id, doc)
 291         except compat_xml_parse_error:
 292             pass
 293
 294         # it's tempting to parse this further, but you would
 295         # have to take into account all the variations like
 296         #   Video Title - Site Name
 297         #   Site Name | Video Title
 298         #   Video Title - Tagline | Site Name
 299         # and so on and so forth; it's just not practical
 300         video_title = self._html_search_regex(
 301             r'(?s)<title>(.*?)</title>', webpage, 'video title',
 302             default='video')
 303
 304         # video uploader is domain name
 305         video_uploader = self._search_regex(
 306             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 307
 308         # Look for BrightCove:
 309         bc_urls = BrightcoveIE._extract_brightcove_urls(webpage)
 310         if bc_urls:
 311             self.to_screen('Brightcove video detected.')
 312             entries = [{
 313                 '_type': 'url',
 314                 'url': smuggle_url(bc_url, {'Referer': url}),
 315                 'ie_key': 'Brightcove'
 316             } for bc_url in bc_urls]
 317
 318             return {
 319                 '_type': 'playlist',
 320                 'title': video_title,
 321                 'id': video_id,
 322                 'entries': entries,
 323             }
 324
 325         # Look for embedded (iframe) Vimeo player
 326         mobj = re.search(
 327             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//player\.vimeo\.com/video/.+?)\1', webpage)
 328         if mobj:
 329             player_url = unescapeHTML(mobj.group('url'))
 330             surl = smuggle_url(player_url, {'Referer': url})
 331             return self.url_result(surl, 'Vimeo')
 332
 333         # Look for embedded (swf embed) Vimeo player
 334         mobj = re.search(
 335             r'<embed[^>]+?src="(https?://(?:www\.)?vimeo\.com/moogaloop\.swf.+?)"', webpage)
 336         if mobj:
 337             return self.url_result(mobj.group(1), 'Vimeo')
 338
 339         # Look for embedded YouTube player
 340         matches = re.findall(r'''(?x)
 341             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 342             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 343                 (?:embed|v)/.+?)
 344             \1''', webpage)
 345         if matches:
 346             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 347                      for tuppl in matches]
 348             return self.playlist_result(
 349                 urlrs, playlist_id=video_id, playlist_title=video_title)
 350
 351         # Look for embedded Dailymotion player
 352         matches = re.findall(
 353             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 354         if matches:
 355             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 356                      for tuppl in matches]
 357             return self.playlist_result(
 358                 urlrs, playlist_id=video_id, playlist_title=video_title)
 359
 360         # Look for embedded Wistia player
 361         match = re.search(
 362             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 363         if match:
 364             return {
 365                 '_type': 'url_transparent',
 366                 'url': unescapeHTML(match.group('url')),
 367                 'ie_key': 'Wistia',
 368                 'uploader': video_uploader,
 369                 'title': video_title,
 370                 'id': video_id,
 371             }
 372
 373         # Look for embedded blip.tv player
 374         mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
 375         if mobj:
 376             return self.url_result('http://blip.tv/a/a-'+mobj.group(1), 'BlipTV')
 377         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9]+)', webpage)
 378         if mobj:
 379             return self.url_result(mobj.group(1), 'BlipTV')
 380
 381         # Look for Bandcamp pages with custom domain
 382         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 383         if mobj is not None:
 384             burl = unescapeHTML(mobj.group(1))
 385             # Don't set the extractor because it can be a track url or an album
 386             return self.url_result(burl)
 387
 388         # Look for embedded Vevo player
 389         mobj = re.search(
 390             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 391         if mobj is not None:
 392             return self.url_result(mobj.group('url'))
 393
 394         # Look for Ooyala videos
 395         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 396         if mobj is not None:
 397             return OoyalaIE._build_url_result(mobj.group(1))
 398
 399         # Look for Aparat videos
 400         mobj = re.search(r'<iframe src="(http://www\.aparat\.com/video/[^"]+)"', webpage)
 401         if mobj is not None:
 402             return self.url_result(mobj.group(1), 'Aparat')
 403
 404         # Look for MPORA videos
 405         mobj = re.search(r'<iframe .*?src="(http://mpora\.(?:com|de)/videos/[^"]+)"', webpage)
 406         if mobj is not None:
 407             return self.url_result(mobj.group(1), 'Mpora')
 408
 409         # Look for embedded NovaMov player
 410         mobj = re.search(
 411             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?novamov\.com/embed\.php.+?)\1', webpage)
 412         if mobj is not None:
 413             return self.url_result(mobj.group('url'), 'NovaMov')
 414
 415         # Look for embedded NowVideo player
 416         mobj = re.search(
 417             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?nowvideo\.(?:ch|sx|eu)/embed\.php.+?)\1', webpage)
 418         if mobj is not None:
 419             return self.url_result(mobj.group('url'), 'NowVideo')
 420
 421         # Look for embedded Facebook player
 422         mobj = re.search(
 423             r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
 424         if mobj is not None:
 425             return self.url_result(mobj.group('url'), 'Facebook')
 426
 427         # Look for embedded VK player
 428         mobj = re.search(r'<iframe[^>]+?src=(["\'])(?P<url>https?://vk\.com/video_ext\.php.+?)\1', webpage)
 429         if mobj is not None:
 430             return self.url_result(mobj.group('url'), 'VK')
 431
 432         # Look for embedded Huffington Post player
 433         mobj = re.search(
 434             r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed\.live\.huffingtonpost\.com/.+?)\1', webpage)
 435         if mobj is not None:
 436             return self.url_result(mobj.group('url'), 'HuffPost')
 437
 438         # Look for embed.ly
 439         mobj = re.search(r'class=["\']embedly-card["\'][^>]href=["\'](?P<url>[^"\']+)', webpage)
 440         if mobj is not None:
 441             return self.url_result(mobj.group('url'))
 442         mobj = re.search(r'class=["\']embedly-embed["\'][^>]src=["\'][^"\']*url=(?P<url>[^&]+)', webpage)
 443         if mobj is not None:
 444             return self.url_result(compat_urllib_parse.unquote(mobj.group('url')))
 445
 446         # Look for funnyordie embed
 447         matches = re.findall(r'<iframe[^>]+?src="(https?://(?:www\.)?funnyordie\.com/embed/[^"]+)"', webpage)
 448         if matches:
 449             urlrs = [self.url_result(unescapeHTML(eurl), 'FunnyOrDie')
 450                      for eurl in matches]
 451             return self.playlist_result(
 452                 urlrs, playlist_id=video_id, playlist_title=video_title)
 453
 454         # Start with something easy: JW Player in SWFObject
 455         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 456         if mobj is None:
 457             # Look for gorilla-vid style embedding
 458             mobj = re.search(r'(?s)(?:jw_plugins|JWPlayerOptions).*?file\s*:\s*["\'](.*?)["\']', webpage)
 459         if mobj is None:
 460             # Broaden the search a little bit
 461             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 462         if mobj is None:
 463             # Broaden the search a little bit: JWPlayer JS loader
 464             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
 465         if mobj is None:
 466             # Try to find twitter cards info
 467             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 468         if mobj is None:
 469             # We look for Open Graph info:
 470             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 471             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 472             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 473             if m_video_type is not None:
 474                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 475         if mobj is None:
 476             # HTML5 video
 477             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 478         if mobj is None:
 479             mobj = re.search(
 480                 r'(?i)<meta\s+(?=(?:[a-z-]+="[^"]+"\s+)*http-equiv="refresh")'
 481                 r'(?:[a-z-]+="[^"]+"\s+)*?content="[0-9]{,2};url=\'([^\']+)\'"',
 482                 webpage)
 483             if mobj:
 484                 new_url = mobj.group(1)
 485                 self.report_following_redirect(new_url)
 486                 return {
 487                     '_type': 'url',
 488                     'url': new_url,
 489                 }
 490         if mobj is None:
 491             raise ExtractorError('Unsupported URL: %s' % url)
 492
 493         # It's possible that one of the regexes
 494         # matched, but returned an empty group:
 495         if mobj.group(1) is None:
 496             raise ExtractorError('Did not find a valid video URL at %s' % url)
 497
 498         video_url = mobj.group(1)
 499         video_url = compat_urlparse.urljoin(url, video_url)
 500         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 501
 502         # Sometimes, jwplayer extraction will result in a YouTube URL
 503         if YoutubeIE.suitable(video_url):
 504             return self.url_result(video_url, 'Youtube')
 505
 506         # here's a fun little line of code for you:
 507         video_id = os.path.splitext(video_id)[0]
 508
 509         return {
 510             'id': video_id,
 511             'url': video_url,
 512             'uploader': video_uploader,
 513             'title': video_title,
 514         }