_ 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             'file': 'trailer.mp4',
  87             'md5': '67d406c2bcb6af27fa886f31aa934bbe',
  88             'info_dict': {
  89                 'id': 'trailer',
  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             'file': 'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ.mp4',
  98             'md5': '5644c6ca5d5782c1d0d350dad9bd840c',
  99             'info_dict': {
 100                 'id': 'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ',
 101                 'ext': 'mp4',
 102                 'title': '2cc213299525360.mov',  # that's what we get
 103             },
 104         },
 105     ]
 106
 107     def report_download_webpage(self, video_id):
 108         """Report webpage download."""
 109         if not self._downloader.params.get('test', False):
 110             self._downloader.report_warning('Falling back on generic information extractor.')
 111         super(GenericIE, self).report_download_webpage(video_id)
 112
 113     def report_following_redirect(self, new_url):
 114         """Report information extraction."""
 115         self._downloader.to_screen('[redirect] Following redirect to %s' % new_url)
 116
 117     def _send_head(self, url):
 118         """Check if it is a redirect, like url shorteners, in case return the new url."""
 119
 120         class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
 121             """
 122             Subclass the HTTPRedirectHandler to make it use our
 123             HEADRequest also on the redirected URL
 124             """
 125             def redirect_request(self, req, fp, code, msg, headers, newurl):
 126                 if code in (301, 302, 303, 307):
 127                     newurl = newurl.replace(' ', '%20')
 128                     newheaders = dict((k,v) for k,v in req.headers.items()
 129                                       if k.lower() not in ("content-length", "content-type"))
 130                     return HEADRequest(newurl,
 131                                        headers=newheaders,
 132                                        origin_req_host=req.get_origin_req_host(),
 133                                        unverifiable=True)
 134                 else:
 135                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 136
 137         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 138             """
 139             Fallback to GET if HEAD is not allowed (405 HTTP error)
 140             """
 141             def http_error_405(self, req, fp, code, msg, headers):
 142                 fp.read()
 143                 fp.close()
 144
 145                 newheaders = dict((k,v) for k,v in req.headers.items()
 146                                   if k.lower() not in ("content-length", "content-type"))
 147                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 148                                                  headers=newheaders,
 149                                                  origin_req_host=req.get_origin_req_host(),
 150                                                  unverifiable=True))
 151
 152         # Build our opener
 153         opener = compat_urllib_request.OpenerDirector()
 154         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 155                         HTTPMethodFallback, HEADRedirectHandler,
 156                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 157             opener.add_handler(handler())
 158
 159         response = opener.open(HEADRequest(url))
 160         if response is None:
 161             raise ExtractorError('Invalid URL protocol')
 162         return response
 163
 164     def _extract_rss(self, url, video_id, doc):
 165         playlist_title = doc.find('./channel/title').text
 166         playlist_desc_el = doc.find('./channel/description')
 167         playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text
 168
 169         entries = [{
 170             '_type': 'url',
 171             'url': e.find('link').text,
 172             'title': e.find('title').text,
 173         } for e in doc.findall('./channel/item')]
 174
 175         return {
 176             '_type': 'playlist',
 177             'id': url,
 178             'title': playlist_title,
 179             'description': playlist_desc,
 180             'entries': entries,
 181         }
 182
 183     def _real_extract(self, url):
 184         parsed_url = compat_urlparse.urlparse(url)
 185         if not parsed_url.scheme:
 186             default_search = self._downloader.params.get('default_search')
 187             if default_search is None:
 188                 default_search = 'auto'
 189
 190             if default_search == 'auto':
 191                 if '/' in url:
 192                     self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 193                     return self.url_result('http://' + url)
 194                 else:
 195                     return self.url_result('ytsearch:' + url)
 196             else:
 197                 assert ':' in default_search
 198                 return self.url_result(default_search + url)
 199         video_id = os.path.splitext(url.split('/')[-1])[0]
 200
 201         self.to_screen('%s: Requesting header' % video_id)
 202
 203         try:
 204             response = self._send_head(url)
 205
 206             # Check for redirect
 207             new_url = response.geturl()
 208             if url != new_url:
 209                 self.report_following_redirect(new_url)
 210                 return self.url_result(new_url)
 211
 212             # Check for direct link to a video
 213             content_type = response.headers.get('Content-Type', '')
 214             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 215             if m:
 216                 upload_date = response.headers.get('Last-Modified')
 217                 if upload_date:
 218                     upload_date = unified_strdate(upload_date)
 219                 return {
 220                     'id': video_id,
 221                     'title': os.path.splitext(url_basename(url))[0],
 222                     'formats': [{
 223                         'format_id': m.group('format_id'),
 224                         'url': url,
 225                         'vcodec': 'none' if m.group('type') == 'audio' else None
 226                     }],
 227                     'upload_date': upload_date,
 228                 }
 229
 230         except compat_urllib_error.HTTPError:
 231             # This may be a stupid server that doesn't like HEAD, our UA, or so
 232             pass
 233
 234         try:
 235             webpage = self._download_webpage(url, video_id)
 236         except ValueError:
 237             # since this is the last-resort InfoExtractor, if
 238             # this error is thrown, it'll be thrown here
 239             raise ExtractorError('Failed to download URL: %s' % url)
 240
 241         self.report_extraction(video_id)
 242
 243         # Is it an RSS feed?
 244         try:
 245             doc = xml.etree.ElementTree.fromstring(webpage.encode('utf-8'))
 246             if doc.tag == 'rss':
 247                 return self._extract_rss(url, video_id, doc)
 248         except compat_xml_parse_error:
 249             pass
 250
 251         # it's tempting to parse this further, but you would
 252         # have to take into account all the variations like
 253         #   Video Title - Site Name
 254         #   Site Name | Video Title
 255         #   Video Title - Tagline | Site Name
 256         # and so on and so forth; it's just not practical
 257         video_title = self._html_search_regex(
 258             r'(?s)<title>(.*?)</title>', webpage, 'video title',
 259             default='video')
 260
 261         # video uploader is domain name
 262         video_uploader = self._search_regex(
 263             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 264
 265         # Look for BrightCove:
 266         bc_urls = BrightcoveIE._extract_brightcove_urls(webpage)
 267         if bc_urls:
 268             self.to_screen('Brightcove video detected.')
 269             entries = [{
 270                 '_type': 'url',
 271                 'url': smuggle_url(bc_url, {'Referer': url}),
 272                 'ie_key': 'Brightcove'
 273             } for bc_url in bc_urls]
 274
 275             return {
 276                 '_type': 'playlist',
 277                 'title': video_title,
 278                 'id': video_id,
 279                 'entries': entries,
 280             }
 281
 282         # Look for embedded (iframe) Vimeo player
 283         mobj = re.search(
 284             r'<iframe[^>]+?src="((?:https?:)?//player\.vimeo\.com/video/.+?)"', webpage)
 285         if mobj:
 286             player_url = unescapeHTML(mobj.group(1))
 287             surl = smuggle_url(player_url, {'Referer': url})
 288             return self.url_result(surl, 'Vimeo')
 289
 290         # Look for embedded (swf embed) Vimeo player
 291         mobj = re.search(
 292             r'<embed[^>]+?src="(https?://(?:www\.)?vimeo\.com/moogaloop\.swf.+?)"', webpage)
 293         if mobj:
 294             return self.url_result(mobj.group(1), 'Vimeo')
 295
 296         # Look for embedded YouTube player
 297         matches = re.findall(r'''(?x)
 298             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 299             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 300                 (?:embed|v)/.+?)
 301             \1''', webpage)
 302         if matches:
 303             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 304                      for tuppl in matches]
 305             return self.playlist_result(
 306                 urlrs, playlist_id=video_id, playlist_title=video_title)
 307
 308         # Look for embedded Dailymotion player
 309         matches = re.findall(
 310             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 311         if matches:
 312             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 313                      for tuppl in matches]
 314             return self.playlist_result(
 315                 urlrs, playlist_id=video_id, playlist_title=video_title)
 316
 317         # Look for embedded Wistia player
 318         match = re.search(
 319             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 320         if match:
 321             return {
 322                 '_type': 'url_transparent',
 323                 'url': unescapeHTML(match.group('url')),
 324                 'ie_key': 'Wistia',
 325                 'uploader': video_uploader,
 326                 'title': video_title,
 327                 'id': video_id,
 328             }
 329
 330         # Look for embedded blip.tv player
 331         mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
 332         if mobj:
 333             return self.url_result('http://blip.tv/a/a-'+mobj.group(1), 'BlipTV')
 334         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9]+)', webpage)
 335         if mobj:
 336             return self.url_result(mobj.group(1), 'BlipTV')
 337
 338         # Look for Bandcamp pages with custom domain
 339         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 340         if mobj is not None:
 341             burl = unescapeHTML(mobj.group(1))
 342             # Don't set the extractor because it can be a track url or an album
 343             return self.url_result(burl)
 344
 345         # Look for embedded Vevo player
 346         mobj = re.search(
 347             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 348         if mobj is not None:
 349             return self.url_result(mobj.group('url'))
 350
 351         # Look for Ooyala videos
 352         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 353         if mobj is not None:
 354             return OoyalaIE._build_url_result(mobj.group(1))
 355
 356         # Look for Aparat videos
 357         mobj = re.search(r'<iframe src="(http://www\.aparat\.com/video/[^"]+)"', webpage)
 358         if mobj is not None:
 359             return self.url_result(mobj.group(1), 'Aparat')
 360
 361         # Look for MPORA videos
 362         mobj = re.search(r'<iframe .*?src="(http://mpora\.(?:com|de)/videos/[^"]+)"', webpage)
 363         if mobj is not None:
 364             return self.url_result(mobj.group(1), 'Mpora')
 365
 366         # Look for embedded NovaMov player
 367         mobj = re.search(
 368             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?novamov\.com/embed\.php.+?)\1', webpage)
 369         if mobj is not None:
 370             return self.url_result(mobj.group('url'), 'NovaMov')
 371
 372         # Look for embedded NowVideo player
 373         mobj = re.search(
 374             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?nowvideo\.(?:ch|sx|eu)/embed\.php.+?)\1', webpage)
 375         if mobj is not None:
 376             return self.url_result(mobj.group('url'), 'NowVideo')
 377
 378         # Look for embedded Facebook player
 379         mobj = re.search(
 380             r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
 381         if mobj is not None:
 382             return self.url_result(mobj.group('url'), 'Facebook')
 383
 384         # Look for embedded Huffington Post player
 385         mobj = re.search(
 386             r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed\.live\.huffingtonpost\.com/.+?)\1', webpage)
 387         if mobj is not None:
 388             return self.url_result(mobj.group('url'), 'HuffPost')
 389
 390         # Start with something easy: JW Player in SWFObject
 391         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 392         if mobj is None:
 393             # Look for gorilla-vid style embedding
 394             mobj = re.search(r'(?s)(?:jw_plugins|JWPlayerOptions).*?file\s*:\s*["\'](.*?)["\']', webpage)
 395         if mobj is None:
 396             # Broaden the search a little bit
 397             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 398         if mobj is None:
 399             # Broaden the search a little bit: JWPlayer JS loader
 400             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
 401         if mobj is None:
 402             # Try to find twitter cards info
 403             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 404         if mobj is None:
 405             # We look for Open Graph info:
 406             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 407             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 408             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 409             if m_video_type is not None:
 410                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 411         if mobj is None:
 412             # HTML5 video
 413             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 414         if mobj is None:
 415             raise ExtractorError('Unsupported URL: %s' % url)
 416
 417         # It's possible that one of the regexes
 418         # matched, but returned an empty group:
 419         if mobj.group(1) is None:
 420             raise ExtractorError('Did not find a valid video URL at %s' % url)
 421
 422         video_url = mobj.group(1)
 423         video_url = compat_urlparse.urljoin(url, video_url)
 424         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 425
 426         # Sometimes, jwplayer extraction will result in a YouTube URL
 427         if YoutubeIE.suitable(video_url):
 428             return self.url_result(video_url, 'Youtube')
 429
 430         # here's a fun little line of code for you:
 431         video_id = os.path.splitext(video_id)[0]
 432
 433         return {
 434             'id': video_id,
 435             'url': video_url,
 436             'uploader': video_uploader,
 437             'title': video_title,
 438         }