_ 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
  17     ExtractorError,
  18     HEADRequest,
  19     smuggle_url,
  20     unescapeHTML,
  21     unified_strdate,
  22     url_basename,
  23 )
  24 from .brightcove import BrightcoveIE
  25 from .ooyala import OoyalaIE
  26
  27
  28 class GenericIE(InfoExtractor):
  29     IE_DESC = 'Generic downloader that works on some sites'
  30     _VALID_URL = r'.*'
  31     IE_NAME = 'generic'
  32     _TESTS = [
  33         {
  34             'url': 'http://www.hodiho.fr/2013/02/regis-plante-sa-jeep.html',
  35             'file': '13601338388002.mp4',
  36             'md5': '6e15c93721d7ec9e9ca3fdbf07982cfd',
  37             'info_dict': {
  38                 'uploader': 'www.hodiho.fr',
  39                 'title': 'R\u00e9gis plante sa Jeep',
  40             }
  41         },
  42         # bandcamp page with custom domain
  43         {
  44             'add_ie': ['Bandcamp'],
  45             'url': 'http://bronyrock.com/track/the-pony-mash',
  46             'file': '3235767654.mp3',
  47             'info_dict': {
  48                 'title': 'The Pony Mash',
  49                 'uploader': 'M_Pallante',
  50             },
  51             'skip': 'There is a limit of 200 free downloads / month for the test song',
  52         },
  53         # embedded brightcove video
  54         # it also tests brightcove videos that need to set the 'Referer' in the
  55         # http requests
  56         {
  57             'add_ie': ['Brightcove'],
  58             'url': 'http://www.bfmtv.com/video/bfmbusiness/cours-bourse/cours-bourse-l-analyse-technique-154522/',
  59             'info_dict': {
  60                 'id': '2765128793001',
  61                 'ext': 'mp4',
  62                 'title': 'Le cours de bourse : l’analyse technique',
  63                 'description': 'md5:7e9ad046e968cb2d1114004aba466fd9',
  64                 'uploader': 'BFM BUSINESS',
  65             },
  66             'params': {
  67                 'skip_download': True,
  68             },
  69         },
  70         {
  71             # https://github.com/rg3/youtube-dl/issues/2253
  72             'url': 'http://bcove.me/i6nfkrc3',
  73             'file': '3101154703001.mp4',
  74             'md5': '0ba9446db037002366bab3b3eb30c88c',
  75             'info_dict': {
  76                 'title': 'Still no power',
  77                 'uploader': 'thestar.com',
  78                 '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.',
  79             },
  80             'add_ie': ['Brightcove'],
  81         },
  82         # Direct link to a video
  83         {
  84             'url': 'http://media.w3.org/2010/05/sintel/trailer.mp4',
  85             'file': 'trailer.mp4',
  86             'md5': '67d406c2bcb6af27fa886f31aa934bbe',
  87             'info_dict': {
  88                 'id': 'trailer',
  89                 'title': 'trailer',
  90                 'upload_date': '20100513',
  91             }
  92         },
  93         # ooyala video
  94         {
  95             'url': 'http://www.rollingstone.com/music/videos/norwegian-dj-cashmere-cat-goes-spartan-on-with-me-premiere-20131219',
  96             'file': 'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ.mp4',
  97             'md5': '5644c6ca5d5782c1d0d350dad9bd840c',
  98             'info_dict': {
  99                 'id': 'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ',
 100                 'ext': 'mp4',
 101                 'title': '2cc213299525360.mov',  # that's what we get
 102             },
 103         },
 104     ]
 105
 106     def report_download_webpage(self, video_id):
 107         """Report webpage download."""
 108         if not self._downloader.params.get('test', False):
 109             self._downloader.report_warning('Falling back on generic information extractor.')
 110         super(GenericIE, self).report_download_webpage(video_id)
 111
 112     def report_following_redirect(self, new_url):
 113         """Report information extraction."""
 114         self._downloader.to_screen('[redirect] Following redirect to %s' % new_url)
 115
 116     def _send_head(self, url):
 117         """Check if it is a redirect, like url shorteners, in case return the new url."""
 118
 119         class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
 120             """
 121             Subclass the HTTPRedirectHandler to make it use our
 122             HEADRequest also on the redirected URL
 123             """
 124             def redirect_request(self, req, fp, code, msg, headers, newurl):
 125                 if code in (301, 302, 303, 307):
 126                     newurl = newurl.replace(' ', '%20')
 127                     newheaders = dict((k,v) for k,v in req.headers.items()
 128                                       if k.lower() not in ("content-length", "content-type"))
 129                     return HEADRequest(newurl,
 130                                        headers=newheaders,
 131                                        origin_req_host=req.get_origin_req_host(),
 132                                        unverifiable=True)
 133                 else:
 134                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 135
 136         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 137             """
 138             Fallback to GET if HEAD is not allowed (405 HTTP error)
 139             """
 140             def http_error_405(self, req, fp, code, msg, headers):
 141                 fp.read()
 142                 fp.close()
 143
 144                 newheaders = dict((k,v) for k,v in req.headers.items()
 145                                   if k.lower() not in ("content-length", "content-type"))
 146                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 147                                                  headers=newheaders,
 148                                                  origin_req_host=req.get_origin_req_host(),
 149                                                  unverifiable=True))
 150
 151         # Build our opener
 152         opener = compat_urllib_request.OpenerDirector()
 153         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 154                         HTTPMethodFallback, HEADRedirectHandler,
 155                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 156             opener.add_handler(handler())
 157
 158         response = opener.open(HEADRequest(url))
 159         if response is None:
 160             raise ExtractorError('Invalid URL protocol')
 161         return response
 162
 163     def _extract_rss(self, url, video_id, doc):
 164         playlist_title = doc.find('./channel/title').text
 165         playlist_desc_el = doc.find('./channel/description')
 166         playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text
 167
 168         entries = [{
 169             '_type': 'url',
 170             'url': e.find('link').text,
 171             'title': e.find('title').text,
 172         } for e in doc.findall('./channel/item')]
 173
 174         return {
 175             '_type': 'playlist',
 176             'id': url,
 177             'title': playlist_title,
 178             'description': playlist_desc,
 179             'entries': entries,
 180         }
 181
 182     def _real_extract(self, url):
 183         parsed_url = compat_urlparse.urlparse(url)
 184         if not parsed_url.scheme:
 185             default_search = self._downloader.params.get('default_search')
 186             if default_search is None:
 187                 default_search = 'auto'
 188
 189             if default_search == 'auto':
 190                 if '/' in url:
 191                     self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 192                     return self.url_result('http://' + url)
 193                 else:
 194                     return self.url_result('ytsearch:' + url)
 195             else:
 196                 assert ':' in default_search
 197                 return self.url_result(default_search + url)
 198         video_id = os.path.splitext(url.split('/')[-1])[0]
 199
 200         self.to_screen('%s: Requesting header' % video_id)
 201
 202         try:
 203             response = self._send_head(url)
 204
 205             # Check for redirect
 206             new_url = response.geturl()
 207             if url != new_url:
 208                 self.report_following_redirect(new_url)
 209                 return self.url_result(new_url)
 210
 211             # Check for direct link to a video
 212             content_type = response.headers.get('Content-Type', '')
 213             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 214             if m:
 215                 upload_date = response.headers.get('Last-Modified')
 216                 if upload_date:
 217                     upload_date = unified_strdate(upload_date)
 218                 return {
 219                     'id': video_id,
 220                     'title': os.path.splitext(url_basename(url))[0],
 221                     'formats': [{
 222                         'format_id': m.group('format_id'),
 223                         'url': url,
 224                         'vcodec': 'none' if m.group('type') == 'audio' else None
 225                     }],
 226                     'upload_date': upload_date,
 227                 }
 228
 229         except compat_urllib_error.HTTPError:
 230             # This may be a stupid server that doesn't like HEAD, our UA, or so
 231             pass
 232
 233         try:
 234             webpage = self._download_webpage(url, video_id)
 235         except ValueError:
 236             # since this is the last-resort InfoExtractor, if
 237             # this error is thrown, it'll be thrown here
 238             raise ExtractorError('Failed to download URL: %s' % url)
 239
 240         self.report_extraction(video_id)
 241
 242         # Is it an RSS feed?
 243         try:
 244             doc = xml.etree.ElementTree.fromstring(webpage)
 245             if doc.tag == 'rss':
 246                 return self._extract_rss(url, video_id, doc)
 247         except xml.etree.ElementTree.ParseError:
 248             pass
 249
 250         # it's tempting to parse this further, but you would
 251         # have to take into account all the variations like
 252         #   Video Title - Site Name
 253         #   Site Name | Video Title
 254         #   Video Title - Tagline | Site Name
 255         # and so on and so forth; it's just not practical
 256         video_title = self._html_search_regex(
 257             r'(?s)<title>(.*?)</title>', webpage, 'video title',
 258             default='video')
 259
 260         # video uploader is domain name
 261         video_uploader = self._search_regex(
 262             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 263
 264         # Look for BrightCove:
 265         bc_urls = BrightcoveIE._extract_brightcove_urls(webpage)
 266         if bc_urls:
 267             self.to_screen('Brightcove video detected.')
 268             entries = [{
 269                 '_type': 'url',
 270                 'url': smuggle_url(bc_url, {'Referer': url}),
 271                 'ie_key': 'Brightcove'
 272             } for bc_url in bc_urls]
 273
 274             return {
 275                 '_type': 'playlist',
 276                 'title': video_title,
 277                 'id': video_id,
 278                 'entries': entries,
 279             }
 280
 281         # Look for embedded (iframe) Vimeo player
 282         mobj = re.search(
 283             r'<iframe[^>]+?src="((?:https?:)?//player\.vimeo\.com/video/.+?)"', webpage)
 284         if mobj:
 285             player_url = unescapeHTML(mobj.group(1))
 286             surl = smuggle_url(player_url, {'Referer': url})
 287             return self.url_result(surl, 'Vimeo')
 288
 289         # Look for embedded (swf embed) Vimeo player
 290         mobj = re.search(
 291             r'<embed[^>]+?src="(https?://(?:www\.)?vimeo\.com/moogaloop\.swf.+?)"', webpage)
 292         if mobj:
 293             return self.url_result(mobj.group(1), 'Vimeo')
 294
 295         # Look for embedded YouTube player
 296         matches = re.findall(r'''(?x)
 297             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 298             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 299                 (?:embed|v)/.+?)
 300             \1''', webpage)
 301         if matches:
 302             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 303                      for tuppl in matches]
 304             return self.playlist_result(
 305                 urlrs, playlist_id=video_id, playlist_title=video_title)
 306
 307         # Look for embedded Dailymotion player
 308         matches = re.findall(
 309             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 310         if matches:
 311             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 312                      for tuppl in matches]
 313             return self.playlist_result(
 314                 urlrs, playlist_id=video_id, playlist_title=video_title)
 315
 316         # Look for embedded Wistia player
 317         match = re.search(
 318             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 319         if match:
 320             return {
 321                 '_type': 'url_transparent',
 322                 'url': unescapeHTML(match.group('url')),
 323                 'ie_key': 'Wistia',
 324                 'uploader': video_uploader,
 325                 'title': video_title,
 326                 'id': video_id,
 327             }
 328
 329         # Look for embedded blip.tv player
 330         mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
 331         if mobj:
 332             return self.url_result('http://blip.tv/a/a-'+mobj.group(1), 'BlipTV')
 333         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9]+)', webpage)
 334         if mobj:
 335             return self.url_result(mobj.group(1), 'BlipTV')
 336
 337         # Look for Bandcamp pages with custom domain
 338         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 339         if mobj is not None:
 340             burl = unescapeHTML(mobj.group(1))
 341             # Don't set the extractor because it can be a track url or an album
 342             return self.url_result(burl)
 343
 344         # Look for embedded Vevo player
 345         mobj = re.search(
 346             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 347         if mobj is not None:
 348             return self.url_result(mobj.group('url'))
 349
 350         # Look for Ooyala videos
 351         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 352         if mobj is not None:
 353             return OoyalaIE._build_url_result(mobj.group(1))
 354
 355         # Look for Aparat videos
 356         mobj = re.search(r'<iframe src="(http://www\.aparat\.com/video/[^"]+)"', webpage)
 357         if mobj is not None:
 358             return self.url_result(mobj.group(1), 'Aparat')
 359
 360         # Look for MPORA videos
 361         mobj = re.search(r'<iframe .*?src="(http://mpora\.(?:com|de)/videos/[^"]+)"', webpage)
 362         if mobj is not None:
 363             return self.url_result(mobj.group(1), 'Mpora')
 364
 365         # Look for embedded Novamov player
 366         mobj = re.search(
 367             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?novamov\.com/embed\.php.+?)\1', webpage)
 368         if mobj is not None:
 369             return self.url_result(mobj.group('url'), 'Novamov')
 370
 371         # Look for embedded Facebook player
 372         mobj = re.search(
 373             r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
 374         if mobj is not None:
 375             return self.url_result(mobj.group('url'), 'Facebook')
 376
 377         # Look for embedded Huffington Post player
 378         mobj = re.search(
 379             r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed\.live\.huffingtonpost\.com/.+?)\1', webpage)
 380         if mobj is not None:
 381             return self.url_result(mobj.group('url'), 'HuffPost')
 382
 383         # Start with something easy: JW Player in SWFObject
 384         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 385         if mobj is None:
 386             # Look for gorilla-vid style embedding
 387             mobj = re.search(r'(?s)(?:jw_plugins|JWPlayerOptions).*?file\s*:\s*["\'](.*?)["\']', webpage)
 388         if mobj is None:
 389             # Broaden the search a little bit
 390             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 391         if mobj is None:
 392             # Broaden the search a little bit: JWPlayer JS loader
 393             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
 394         if mobj is None:
 395             # Try to find twitter cards info
 396             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 397         if mobj is None:
 398             # We look for Open Graph info:
 399             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 400             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 401             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 402             if m_video_type is not None:
 403                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 404         if mobj is None:
 405             # HTML5 video
 406             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 407         if mobj is None:
 408             raise ExtractorError('Unsupported URL: %s' % url)
 409
 410         # It's possible that one of the regexes
 411         # matched, but returned an empty group:
 412         if mobj.group(1) is None:
 413             raise ExtractorError('Did not find a valid video URL at %s' % url)
 414
 415         video_url = mobj.group(1)
 416         video_url = compat_urlparse.urljoin(url, video_url)
 417         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 418
 419         # Sometimes, jwplayer extraction will result in a YouTube URL
 420         if YoutubeIE.suitable(video_url):
 421             return self.url_result(video_url, 'Youtube')
 422
 423         # here's a fun little line of code for you:
 424         video_id = os.path.splitext(video_id)[0]
 425
 426         return {
 427             'id': video_id,
 428             'url': video_url,
 429             'uploader': video_uploader,
 430             'title': video_title,
 431         }