_ 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                     return HEADRequest(newurl,
 189                                        headers=newheaders,
 190                                        origin_req_host=req.get_origin_req_host(),
 191                                        unverifiable=True)
 192                 else:
 193                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 194
 195         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 196             """
 197             Fallback to GET if HEAD is not allowed (405 HTTP error)
 198             """
 199             def http_error_405(self, req, fp, code, msg, headers):
 200                 fp.read()
 201                 fp.close()
 202
 203                 newheaders = dict((k,v) for k,v in req.headers.items()
 204                                   if k.lower() not in ("content-length", "content-type"))
 205                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 206                                                  headers=newheaders,
 207                                                  origin_req_host=req.get_origin_req_host(),
 208                                                  unverifiable=True))
 209
 210         # Build our opener
 211         opener = compat_urllib_request.OpenerDirector()
 212         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 213                         HTTPMethodFallback, HEADRedirectHandler,
 214                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 215             opener.add_handler(handler())
 216
 217         response = opener.open(HEADRequest(url))
 218         if response is None:
 219             raise ExtractorError('Invalid URL protocol')
 220         return response
 221
 222     def _extract_rss(self, url, video_id, doc):
 223         playlist_title = doc.find('./channel/title').text
 224         playlist_desc_el = doc.find('./channel/description')
 225         playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text
 226
 227         entries = [{
 228             '_type': 'url',
 229             'url': e.find('link').text,
 230             'title': e.find('title').text,
 231         } for e in doc.findall('./channel/item')]
 232
 233         return {
 234             '_type': 'playlist',
 235             'id': url,
 236             'title': playlist_title,
 237             'description': playlist_desc,
 238             'entries': entries,
 239         }
 240
 241     def _real_extract(self, url):
 242         parsed_url = compat_urlparse.urlparse(url)
 243         if not parsed_url.scheme:
 244             default_search = self._downloader.params.get('default_search')
 245             if default_search is None:
 246                 default_search = 'auto'
 247
 248             if default_search == 'auto':
 249                 if '/' in url:
 250                     self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 251                     return self.url_result('http://' + url)
 252                 else:
 253                     return self.url_result('ytsearch:' + url)
 254             else:
 255                 assert ':' in default_search
 256                 return self.url_result(default_search + url)
 257         video_id = os.path.splitext(url.rstrip('/').split('/')[-1])[0]
 258
 259         self.to_screen('%s: Requesting header' % video_id)
 260
 261         try:
 262             response = self._send_head(url)
 263
 264             # Check for redirect
 265             new_url = response.geturl()
 266             if url != new_url:
 267                 self.report_following_redirect(new_url)
 268                 return self.url_result(new_url)
 269
 270             # Check for direct link to a video
 271             content_type = response.headers.get('Content-Type', '')
 272             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 273             if m:
 274                 upload_date = response.headers.get('Last-Modified')
 275                 if upload_date:
 276                     upload_date = unified_strdate(upload_date)
 277                 return {
 278                     'id': video_id,
 279                     'title': os.path.splitext(url_basename(url))[0],
 280                     'formats': [{
 281                         'format_id': m.group('format_id'),
 282                         'url': url,
 283                         'vcodec': 'none' if m.group('type') == 'audio' else None
 284                     }],
 285                     'upload_date': upload_date,
 286                 }
 287
 288         except compat_urllib_error.HTTPError:
 289             # This may be a stupid server that doesn't like HEAD, our UA, or so
 290             pass
 291
 292         try:
 293             webpage = self._download_webpage(url, video_id)
 294         except ValueError:
 295             # since this is the last-resort InfoExtractor, if
 296             # this error is thrown, it'll be thrown here
 297             raise ExtractorError('Failed to download URL: %s' % url)
 298
 299         self.report_extraction(video_id)
 300
 301         # Is it an RSS feed?
 302         try:
 303             doc = parse_xml(webpage)
 304             if doc.tag == 'rss':
 305                 return self._extract_rss(url, video_id, doc)
 306         except compat_xml_parse_error:
 307             pass
 308
 309         # it's tempting to parse this further, but you would
 310         # have to take into account all the variations like
 311         #   Video Title - Site Name
 312         #   Site Name | Video Title
 313         #   Video Title - Tagline | Site Name
 314         # and so on and so forth; it's just not practical
 315         video_title = self._html_search_regex(
 316             r'(?s)<title>(.*?)</title>', webpage, 'video title',
 317             default='video')
 318
 319         # video uploader is domain name
 320         video_uploader = self._search_regex(
 321             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 322
 323         # Look for BrightCove:
 324         bc_urls = BrightcoveIE._extract_brightcove_urls(webpage)
 325         if bc_urls:
 326             self.to_screen('Brightcove video detected.')
 327             entries = [{
 328                 '_type': 'url',
 329                 'url': smuggle_url(bc_url, {'Referer': url}),
 330                 'ie_key': 'Brightcove'
 331             } for bc_url in bc_urls]
 332
 333             return {
 334                 '_type': 'playlist',
 335                 'title': video_title,
 336                 'id': video_id,
 337                 'entries': entries,
 338             }
 339
 340         # Look for embedded (iframe) Vimeo player
 341         mobj = re.search(
 342             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//player\.vimeo\.com/video/.+?)\1', webpage)
 343         if mobj:
 344             player_url = unescapeHTML(mobj.group('url'))
 345             surl = smuggle_url(player_url, {'Referer': url})
 346             return self.url_result(surl, 'Vimeo')
 347
 348         # Look for embedded (swf embed) Vimeo player
 349         mobj = re.search(
 350             r'<embed[^>]+?src="(https?://(?:www\.)?vimeo\.com/moogaloop\.swf.+?)"', webpage)
 351         if mobj:
 352             return self.url_result(mobj.group(1), 'Vimeo')
 353
 354         # Look for embedded YouTube player
 355         matches = re.findall(r'''(?x)
 356             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 357             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 358                 (?:embed|v)/.+?)
 359             \1''', webpage)
 360         if matches:
 361             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 362                      for tuppl in matches]
 363             return self.playlist_result(
 364                 urlrs, playlist_id=video_id, playlist_title=video_title)
 365
 366         # Look for embedded Dailymotion player
 367         matches = re.findall(
 368             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 369         if matches:
 370             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 371                      for tuppl in matches]
 372             return self.playlist_result(
 373                 urlrs, playlist_id=video_id, playlist_title=video_title)
 374
 375         # Look for embedded Wistia player
 376         match = re.search(
 377             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 378         if match:
 379             return {
 380                 '_type': 'url_transparent',
 381                 'url': unescapeHTML(match.group('url')),
 382                 'ie_key': 'Wistia',
 383                 'uploader': video_uploader,
 384                 'title': video_title,
 385                 'id': video_id,
 386             }
 387
 388         # Look for embedded blip.tv player
 389         mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
 390         if mobj:
 391             return self.url_result('http://blip.tv/a/a-'+mobj.group(1), 'BlipTV')
 392         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9]+)', webpage)
 393         if mobj:
 394             return self.url_result(mobj.group(1), 'BlipTV')
 395
 396         # Look for Bandcamp pages with custom domain
 397         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 398         if mobj is not None:
 399             burl = unescapeHTML(mobj.group(1))
 400             # Don't set the extractor because it can be a track url or an album
 401             return self.url_result(burl)
 402
 403         # Look for embedded Vevo player
 404         mobj = re.search(
 405             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 406         if mobj is not None:
 407             return self.url_result(mobj.group('url'))
 408
 409         # Look for Ooyala videos
 410         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 411         if mobj is not None:
 412             return OoyalaIE._build_url_result(mobj.group(1))
 413
 414         # Look for Aparat videos
 415         mobj = re.search(r'<iframe src="(http://www\.aparat\.com/video/[^"]+)"', webpage)
 416         if mobj is not None:
 417             return self.url_result(mobj.group(1), 'Aparat')
 418
 419         # Look for MPORA videos
 420         mobj = re.search(r'<iframe .*?src="(http://mpora\.(?:com|de)/videos/[^"]+)"', webpage)
 421         if mobj is not None:
 422             return self.url_result(mobj.group(1), 'Mpora')
 423
 424         # Look for embedded NovaMov player
 425         mobj = re.search(
 426             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?novamov\.com/embed\.php.+?)\1', webpage)
 427         if mobj is not None:
 428             return self.url_result(mobj.group('url'), 'NovaMov')
 429
 430         # Look for embedded NowVideo player
 431         mobj = re.search(
 432             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?nowvideo\.(?:ch|sx|eu)/embed\.php.+?)\1', webpage)
 433         if mobj is not None:
 434             return self.url_result(mobj.group('url'), 'NowVideo')
 435
 436         # Look for embedded Facebook player
 437         mobj = re.search(
 438             r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
 439         if mobj is not None:
 440             return self.url_result(mobj.group('url'), 'Facebook')
 441
 442         # Look for embedded VK player
 443         mobj = re.search(r'<iframe[^>]+?src=(["\'])(?P<url>https?://vk\.com/video_ext\.php.+?)\1', webpage)
 444         if mobj is not None:
 445             return self.url_result(mobj.group('url'), 'VK')
 446
 447         # Look for embedded Huffington Post player
 448         mobj = re.search(
 449             r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed\.live\.huffingtonpost\.com/.+?)\1', webpage)
 450         if mobj is not None:
 451             return self.url_result(mobj.group('url'), 'HuffPost')
 452
 453         # Look for embed.ly
 454         mobj = re.search(r'class=["\']embedly-card["\'][^>]href=["\'](?P<url>[^"\']+)', webpage)
 455         if mobj is not None:
 456             return self.url_result(mobj.group('url'))
 457         mobj = re.search(r'class=["\']embedly-embed["\'][^>]src=["\'][^"\']*url=(?P<url>[^&]+)', webpage)
 458         if mobj is not None:
 459             return self.url_result(compat_urllib_parse.unquote(mobj.group('url')))
 460
 461         # Look for funnyordie embed
 462         matches = re.findall(r'<iframe[^>]+?src="(https?://(?:www\.)?funnyordie\.com/embed/[^"]+)"', webpage)
 463         if matches:
 464             urlrs = [self.url_result(unescapeHTML(eurl), 'FunnyOrDie')
 465                      for eurl in matches]
 466             return self.playlist_result(
 467                 urlrs, playlist_id=video_id, playlist_title=video_title)
 468
 469         # Look for embedded RUTV player
 470         rutv_url = RUTVIE._extract_url(webpage)
 471         if rutv_url:
 472             return self.url_result(rutv_url, 'RUTV')
 473
 474         # Start with something easy: JW Player in SWFObject
 475         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 476         if mobj is None:
 477             # Look for gorilla-vid style embedding
 478             mobj = re.search(r'(?s)(?:jw_plugins|JWPlayerOptions).*?file\s*:\s*["\'](.*?)["\']', webpage)
 479         if mobj is None:
 480             # Broaden the search a little bit
 481             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 482         if mobj is None:
 483             # Broaden the search a little bit: JWPlayer JS loader
 484             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
 485         if mobj is None:
 486             # Try to find twitter cards info
 487             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 488         if mobj is None:
 489             # We look for Open Graph info:
 490             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 491             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 492             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 493             if m_video_type is not None:
 494                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 495         if mobj is None:
 496             # HTML5 video
 497             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 498         if mobj is None:
 499             mobj = re.search(
 500                 r'(?i)<meta\s+(?=(?:[a-z-]+="[^"]+"\s+)*http-equiv="refresh")'
 501                 r'(?:[a-z-]+="[^"]+"\s+)*?content="[0-9]{,2};url=\'([^\']+)\'"',
 502                 webpage)
 503             if mobj:
 504                 new_url = mobj.group(1)
 505                 self.report_following_redirect(new_url)
 506                 return {
 507                     '_type': 'url',
 508                     'url': new_url,
 509                 }
 510         if mobj is None:
 511             raise ExtractorError('Unsupported URL: %s' % url)
 512
 513         # It's possible that one of the regexes
 514         # matched, but returned an empty group:
 515         if mobj.group(1) is None:
 516             raise ExtractorError('Did not find a valid video URL at %s' % url)
 517
 518         video_url = mobj.group(1)
 519         video_url = compat_urlparse.urljoin(url, video_url)
 520         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 521
 522         # Sometimes, jwplayer extraction will result in a YouTube URL
 523         if YoutubeIE.suitable(video_url):
 524             return self.url_result(video_url, 'Youtube')
 525
 526         # here's a fun little line of code for you:
 527         video_id = os.path.splitext(video_id)[0]
 528
 529         return {
 530             'id': video_id,
 531             'url': video_url,
 532             'uploader': video_uploader,
 533             'title': video_title,
 534         }