_ 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 import xml.etree.ElementTree
   8
   9 from .common import InfoExtractor
  10 from .youtube import YoutubeIE
  11 from ..utils import (
  12     compat_urllib_error,
  13     compat_urllib_parse,
  14     compat_urllib_request,
  15     compat_urlparse,
  16     compat_xml_parse_error,
  17
  18     ExtractorError,
  19     HEADRequest,
  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             },
 127             # No need to test YoutubeIE here
 128             'params': {
 129                 'skip_download': True,
 130             },
 131         },
 132     ]
 133
 134     def report_download_webpage(self, video_id):
 135         """Report webpage download."""
 136         if not self._downloader.params.get('test', False):
 137             self._downloader.report_warning('Falling back on generic information extractor.')
 138         super(GenericIE, self).report_download_webpage(video_id)
 139
 140     def report_following_redirect(self, new_url):
 141         """Report information extraction."""
 142         self._downloader.to_screen('[redirect] Following redirect to %s' % new_url)
 143
 144     def _send_head(self, url):
 145         """Check if it is a redirect, like url shorteners, in case return the new url."""
 146
 147         class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
 148             """
 149             Subclass the HTTPRedirectHandler to make it use our
 150             HEADRequest also on the redirected URL
 151             """
 152             def redirect_request(self, req, fp, code, msg, headers, newurl):
 153                 if code in (301, 302, 303, 307):
 154                     newurl = newurl.replace(' ', '%20')
 155                     newheaders = dict((k,v) for k,v in req.headers.items()
 156                                       if k.lower() not in ("content-length", "content-type"))
 157                     return HEADRequest(newurl,
 158                                        headers=newheaders,
 159                                        origin_req_host=req.get_origin_req_host(),
 160                                        unverifiable=True)
 161                 else:
 162                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 163
 164         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 165             """
 166             Fallback to GET if HEAD is not allowed (405 HTTP error)
 167             """
 168             def http_error_405(self, req, fp, code, msg, headers):
 169                 fp.read()
 170                 fp.close()
 171
 172                 newheaders = dict((k,v) for k,v in req.headers.items()
 173                                   if k.lower() not in ("content-length", "content-type"))
 174                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 175                                                  headers=newheaders,
 176                                                  origin_req_host=req.get_origin_req_host(),
 177                                                  unverifiable=True))
 178
 179         # Build our opener
 180         opener = compat_urllib_request.OpenerDirector()
 181         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 182                         HTTPMethodFallback, HEADRedirectHandler,
 183                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 184             opener.add_handler(handler())
 185
 186         response = opener.open(HEADRequest(url))
 187         if response is None:
 188             raise ExtractorError('Invalid URL protocol')
 189         return response
 190
 191     def _extract_rss(self, url, video_id, doc):
 192         playlist_title = doc.find('./channel/title').text
 193         playlist_desc_el = doc.find('./channel/description')
 194         playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text
 195
 196         entries = [{
 197             '_type': 'url',
 198             'url': e.find('link').text,
 199             'title': e.find('title').text,
 200         } for e in doc.findall('./channel/item')]
 201
 202         return {
 203             '_type': 'playlist',
 204             'id': url,
 205             'title': playlist_title,
 206             'description': playlist_desc,
 207             'entries': entries,
 208         }
 209
 210     def _real_extract(self, url):
 211         parsed_url = compat_urlparse.urlparse(url)
 212         if not parsed_url.scheme:
 213             default_search = self._downloader.params.get('default_search')
 214             if default_search is None:
 215                 default_search = 'auto'
 216
 217             if default_search == 'auto':
 218                 if '/' in url:
 219                     self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 220                     return self.url_result('http://' + url)
 221                 else:
 222                     return self.url_result('ytsearch:' + url)
 223             else:
 224                 assert ':' in default_search
 225                 return self.url_result(default_search + url)
 226         video_id = os.path.splitext(url.split('/')[-1])[0]
 227
 228         self.to_screen('%s: Requesting header' % video_id)
 229
 230         try:
 231             response = self._send_head(url)
 232
 233             # Check for redirect
 234             new_url = response.geturl()
 235             if url != new_url:
 236                 self.report_following_redirect(new_url)
 237                 return self.url_result(new_url)
 238
 239             # Check for direct link to a video
 240             content_type = response.headers.get('Content-Type', '')
 241             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 242             if m:
 243                 upload_date = response.headers.get('Last-Modified')
 244                 if upload_date:
 245                     upload_date = unified_strdate(upload_date)
 246                 return {
 247                     'id': video_id,
 248                     'title': os.path.splitext(url_basename(url))[0],
 249                     'formats': [{
 250                         'format_id': m.group('format_id'),
 251                         'url': url,
 252                         'vcodec': 'none' if m.group('type') == 'audio' else None
 253                     }],
 254                     'upload_date': upload_date,
 255                 }
 256
 257         except compat_urllib_error.HTTPError:
 258             # This may be a stupid server that doesn't like HEAD, our UA, or so
 259             pass
 260
 261         try:
 262             webpage = self._download_webpage(url, video_id)
 263         except ValueError:
 264             # since this is the last-resort InfoExtractor, if
 265             # this error is thrown, it'll be thrown here
 266             raise ExtractorError('Failed to download URL: %s' % url)
 267
 268         self.report_extraction(video_id)
 269
 270         # Is it an RSS feed?
 271         try:
 272             doc = xml.etree.ElementTree.fromstring(webpage.encode('utf-8'))
 273             if doc.tag == 'rss':
 274                 return self._extract_rss(url, video_id, doc)
 275         except compat_xml_parse_error:
 276             pass
 277
 278         # it's tempting to parse this further, but you would
 279         # have to take into account all the variations like
 280         #   Video Title - Site Name
 281         #   Site Name | Video Title
 282         #   Video Title - Tagline | Site Name
 283         # and so on and so forth; it's just not practical
 284         video_title = self._html_search_regex(
 285             r'(?s)<title>(.*?)</title>', webpage, 'video title',
 286             default='video')
 287
 288         # video uploader is domain name
 289         video_uploader = self._search_regex(
 290             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 291
 292         # Look for BrightCove:
 293         bc_urls = BrightcoveIE._extract_brightcove_urls(webpage)
 294         if bc_urls:
 295             self.to_screen('Brightcove video detected.')
 296             entries = [{
 297                 '_type': 'url',
 298                 'url': smuggle_url(bc_url, {'Referer': url}),
 299                 'ie_key': 'Brightcove'
 300             } for bc_url in bc_urls]
 301
 302             return {
 303                 '_type': 'playlist',
 304                 'title': video_title,
 305                 'id': video_id,
 306                 'entries': entries,
 307             }
 308
 309         # Look for embedded (iframe) Vimeo player
 310         mobj = re.search(
 311             r'<iframe[^>]+?src="((?:https?:)?//player\.vimeo\.com/video/.+?)"', webpage)
 312         if mobj:
 313             player_url = unescapeHTML(mobj.group(1))
 314             surl = smuggle_url(player_url, {'Referer': url})
 315             return self.url_result(surl, 'Vimeo')
 316
 317         # Look for embedded (swf embed) Vimeo player
 318         mobj = re.search(
 319             r'<embed[^>]+?src="(https?://(?:www\.)?vimeo\.com/moogaloop\.swf.+?)"', webpage)
 320         if mobj:
 321             return self.url_result(mobj.group(1), 'Vimeo')
 322
 323         # Look for embedded YouTube player
 324         matches = re.findall(r'''(?x)
 325             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 326             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 327                 (?:embed|v)/.+?)
 328             \1''', webpage)
 329         if matches:
 330             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 331                      for tuppl in matches]
 332             return self.playlist_result(
 333                 urlrs, playlist_id=video_id, playlist_title=video_title)
 334
 335         # Look for embedded Dailymotion player
 336         matches = re.findall(
 337             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 338         if matches:
 339             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 340                      for tuppl in matches]
 341             return self.playlist_result(
 342                 urlrs, playlist_id=video_id, playlist_title=video_title)
 343
 344         # Look for embedded Wistia player
 345         match = re.search(
 346             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 347         if match:
 348             return {
 349                 '_type': 'url_transparent',
 350                 'url': unescapeHTML(match.group('url')),
 351                 'ie_key': 'Wistia',
 352                 'uploader': video_uploader,
 353                 'title': video_title,
 354                 'id': video_id,
 355             }
 356
 357         # Look for embedded blip.tv player
 358         mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
 359         if mobj:
 360             return self.url_result('http://blip.tv/a/a-'+mobj.group(1), 'BlipTV')
 361         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9]+)', webpage)
 362         if mobj:
 363             return self.url_result(mobj.group(1), 'BlipTV')
 364
 365         # Look for Bandcamp pages with custom domain
 366         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 367         if mobj is not None:
 368             burl = unescapeHTML(mobj.group(1))
 369             # Don't set the extractor because it can be a track url or an album
 370             return self.url_result(burl)
 371
 372         # Look for embedded Vevo player
 373         mobj = re.search(
 374             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 375         if mobj is not None:
 376             return self.url_result(mobj.group('url'))
 377
 378         # Look for Ooyala videos
 379         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 380         if mobj is not None:
 381             return OoyalaIE._build_url_result(mobj.group(1))
 382
 383         # Look for Aparat videos
 384         mobj = re.search(r'<iframe src="(http://www\.aparat\.com/video/[^"]+)"', webpage)
 385         if mobj is not None:
 386             return self.url_result(mobj.group(1), 'Aparat')
 387
 388         # Look for MPORA videos
 389         mobj = re.search(r'<iframe .*?src="(http://mpora\.(?:com|de)/videos/[^"]+)"', webpage)
 390         if mobj is not None:
 391             return self.url_result(mobj.group(1), 'Mpora')
 392
 393         # Look for embedded NovaMov player
 394         mobj = re.search(
 395             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?novamov\.com/embed\.php.+?)\1', webpage)
 396         if mobj is not None:
 397             return self.url_result(mobj.group('url'), 'NovaMov')
 398
 399         # Look for embedded NowVideo player
 400         mobj = re.search(
 401             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?nowvideo\.(?:ch|sx|eu)/embed\.php.+?)\1', webpage)
 402         if mobj is not None:
 403             return self.url_result(mobj.group('url'), 'NowVideo')
 404
 405         # Look for embedded Facebook player
 406         mobj = re.search(
 407             r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
 408         if mobj is not None:
 409             return self.url_result(mobj.group('url'), 'Facebook')
 410
 411         # Look for embedded VK player
 412         mobj = re.search(r'<iframe[^>]+?src=(["\'])(?P<url>https?://vk\.com/video_ext\.php.+?)\1', webpage)
 413         if mobj is not None:
 414             return self.url_result(mobj.group('url'), 'VK')
 415
 416         # Look for embedded Huffington Post player
 417         mobj = re.search(
 418             r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed\.live\.huffingtonpost\.com/.+?)\1', webpage)
 419         if mobj is not None:
 420             return self.url_result(mobj.group('url'), 'HuffPost')
 421
 422         # Look for embed.ly
 423         mobj = re.search(r'class=["\']embedly-card["\'][^>]href=["\'](?P<url>[^"\']+)', webpage)
 424         if mobj is not None:
 425             return self.url_result(mobj.group('url'))
 426         mobj = re.search(r'class=["\']embedly-embed["\'][^>]src=["\'][^"\']*url=(?P<url>[^&]+)', webpage)
 427         if mobj is not None:
 428             return self.url_result(compat_urllib_parse.unquote(mobj.group('url')))
 429
 430         # Start with something easy: JW Player in SWFObject
 431         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 432         if mobj is None:
 433             # Look for gorilla-vid style embedding
 434             mobj = re.search(r'(?s)(?:jw_plugins|JWPlayerOptions).*?file\s*:\s*["\'](.*?)["\']', webpage)
 435         if mobj is None:
 436             # Broaden the search a little bit
 437             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 438         if mobj is None:
 439             # Broaden the search a little bit: JWPlayer JS loader
 440             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
 441         if mobj is None:
 442             # Try to find twitter cards info
 443             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 444         if mobj is None:
 445             # We look for Open Graph info:
 446             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 447             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 448             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 449             if m_video_type is not None:
 450                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 451         if mobj is None:
 452             # HTML5 video
 453             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 454         if mobj is None:
 455             mobj = re.search(
 456                 r'(?i)<meta\s+(?=(?:[a-z-]+="[^"]+"\s+)*http-equiv="refresh")'
 457                 r'(?:[a-z-]+="[^"]+"\s+)*?content="[0-9]{,2};url=\'([^\']+)\'"',
 458                 webpage)
 459             if mobj:
 460                 new_url = mobj.group(1)
 461                 self.report_following_redirect(new_url)
 462                 return {
 463                     '_type': 'url',
 464                     'url': new_url,
 465                 }
 466         if mobj is None:
 467             raise ExtractorError('Unsupported URL: %s' % url)
 468
 469         # It's possible that one of the regexes
 470         # matched, but returned an empty group:
 471         if mobj.group(1) is None:
 472             raise ExtractorError('Did not find a valid video URL at %s' % url)
 473
 474         video_url = mobj.group(1)
 475         video_url = compat_urlparse.urljoin(url, video_url)
 476         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 477
 478         # Sometimes, jwplayer extraction will result in a YouTube URL
 479         if YoutubeIE.suitable(video_url):
 480             return self.url_result(video_url, 'Youtube')
 481
 482         # here's a fun little line of code for you:
 483         video_id = os.path.splitext(video_id)[0]
 484
 485         return {
 486             'id': video_id,
 487             'url': video_url,
 488             'uploader': video_uploader,
 489             'title': video_title,
 490         }