_ 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     ]
 121
 122     def report_download_webpage(self, video_id):
 123         """Report webpage download."""
 124         if not self._downloader.params.get('test', False):
 125             self._downloader.report_warning('Falling back on generic information extractor.')
 126         super(GenericIE, self).report_download_webpage(video_id)
 127
 128     def report_following_redirect(self, new_url):
 129         """Report information extraction."""
 130         self._downloader.to_screen('[redirect] Following redirect to %s' % new_url)
 131
 132     def _send_head(self, url):
 133         """Check if it is a redirect, like url shorteners, in case return the new url."""
 134
 135         class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
 136             """
 137             Subclass the HTTPRedirectHandler to make it use our
 138             HEADRequest also on the redirected URL
 139             """
 140             def redirect_request(self, req, fp, code, msg, headers, newurl):
 141                 if code in (301, 302, 303, 307):
 142                     newurl = newurl.replace(' ', '%20')
 143                     newheaders = dict((k,v) for k,v in req.headers.items()
 144                                       if k.lower() not in ("content-length", "content-type"))
 145                     return HEADRequest(newurl,
 146                                        headers=newheaders,
 147                                        origin_req_host=req.get_origin_req_host(),
 148                                        unverifiable=True)
 149                 else:
 150                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 151
 152         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 153             """
 154             Fallback to GET if HEAD is not allowed (405 HTTP error)
 155             """
 156             def http_error_405(self, req, fp, code, msg, headers):
 157                 fp.read()
 158                 fp.close()
 159
 160                 newheaders = dict((k,v) for k,v in req.headers.items()
 161                                   if k.lower() not in ("content-length", "content-type"))
 162                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 163                                                  headers=newheaders,
 164                                                  origin_req_host=req.get_origin_req_host(),
 165                                                  unverifiable=True))
 166
 167         # Build our opener
 168         opener = compat_urllib_request.OpenerDirector()
 169         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 170                         HTTPMethodFallback, HEADRedirectHandler,
 171                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 172             opener.add_handler(handler())
 173
 174         response = opener.open(HEADRequest(url))
 175         if response is None:
 176             raise ExtractorError('Invalid URL protocol')
 177         return response
 178
 179     def _extract_rss(self, url, video_id, doc):
 180         playlist_title = doc.find('./channel/title').text
 181         playlist_desc_el = doc.find('./channel/description')
 182         playlist_desc = None if playlist_desc_el is None else playlist_desc_el.text
 183
 184         entries = [{
 185             '_type': 'url',
 186             'url': e.find('link').text,
 187             'title': e.find('title').text,
 188         } for e in doc.findall('./channel/item')]
 189
 190         return {
 191             '_type': 'playlist',
 192             'id': url,
 193             'title': playlist_title,
 194             'description': playlist_desc,
 195             'entries': entries,
 196         }
 197
 198     def _real_extract(self, url):
 199         parsed_url = compat_urlparse.urlparse(url)
 200         if not parsed_url.scheme:
 201             default_search = self._downloader.params.get('default_search')
 202             if default_search is None:
 203                 default_search = 'auto'
 204
 205             if default_search == 'auto':
 206                 if '/' in url:
 207                     self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 208                     return self.url_result('http://' + url)
 209                 else:
 210                     return self.url_result('ytsearch:' + url)
 211             else:
 212                 assert ':' in default_search
 213                 return self.url_result(default_search + url)
 214         video_id = os.path.splitext(url.split('/')[-1])[0]
 215
 216         self.to_screen('%s: Requesting header' % video_id)
 217
 218         try:
 219             response = self._send_head(url)
 220
 221             # Check for redirect
 222             new_url = response.geturl()
 223             if url != new_url:
 224                 self.report_following_redirect(new_url)
 225                 return self.url_result(new_url)
 226
 227             # Check for direct link to a video
 228             content_type = response.headers.get('Content-Type', '')
 229             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 230             if m:
 231                 upload_date = response.headers.get('Last-Modified')
 232                 if upload_date:
 233                     upload_date = unified_strdate(upload_date)
 234                 return {
 235                     'id': video_id,
 236                     'title': os.path.splitext(url_basename(url))[0],
 237                     'formats': [{
 238                         'format_id': m.group('format_id'),
 239                         'url': url,
 240                         'vcodec': 'none' if m.group('type') == 'audio' else None
 241                     }],
 242                     'upload_date': upload_date,
 243                 }
 244
 245         except compat_urllib_error.HTTPError:
 246             # This may be a stupid server that doesn't like HEAD, our UA, or so
 247             pass
 248
 249         try:
 250             webpage = self._download_webpage(url, video_id)
 251         except ValueError:
 252             # since this is the last-resort InfoExtractor, if
 253             # this error is thrown, it'll be thrown here
 254             raise ExtractorError('Failed to download URL: %s' % url)
 255
 256         self.report_extraction(video_id)
 257
 258         # Is it an RSS feed?
 259         try:
 260             doc = xml.etree.ElementTree.fromstring(webpage.encode('utf-8'))
 261             if doc.tag == 'rss':
 262                 return self._extract_rss(url, video_id, doc)
 263         except compat_xml_parse_error:
 264             pass
 265
 266         # it's tempting to parse this further, but you would
 267         # have to take into account all the variations like
 268         #   Video Title - Site Name
 269         #   Site Name | Video Title
 270         #   Video Title - Tagline | Site Name
 271         # and so on and so forth; it's just not practical
 272         video_title = self._html_search_regex(
 273             r'(?s)<title>(.*?)</title>', webpage, 'video title',
 274             default='video')
 275
 276         # video uploader is domain name
 277         video_uploader = self._search_regex(
 278             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 279
 280         # Look for BrightCove:
 281         bc_urls = BrightcoveIE._extract_brightcove_urls(webpage)
 282         if bc_urls:
 283             self.to_screen('Brightcove video detected.')
 284             entries = [{
 285                 '_type': 'url',
 286                 'url': smuggle_url(bc_url, {'Referer': url}),
 287                 'ie_key': 'Brightcove'
 288             } for bc_url in bc_urls]
 289
 290             return {
 291                 '_type': 'playlist',
 292                 'title': video_title,
 293                 'id': video_id,
 294                 'entries': entries,
 295             }
 296
 297         # Look for embedded (iframe) Vimeo player
 298         mobj = re.search(
 299             r'<iframe[^>]+?src="((?:https?:)?//player\.vimeo\.com/video/.+?)"', webpage)
 300         if mobj:
 301             player_url = unescapeHTML(mobj.group(1))
 302             surl = smuggle_url(player_url, {'Referer': url})
 303             return self.url_result(surl, 'Vimeo')
 304
 305         # Look for embedded (swf embed) Vimeo player
 306         mobj = re.search(
 307             r'<embed[^>]+?src="(https?://(?:www\.)?vimeo\.com/moogaloop\.swf.+?)"', webpage)
 308         if mobj:
 309             return self.url_result(mobj.group(1), 'Vimeo')
 310
 311         # Look for embedded YouTube player
 312         matches = re.findall(r'''(?x)
 313             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 314             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 315                 (?:embed|v)/.+?)
 316             \1''', webpage)
 317         if matches:
 318             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 319                      for tuppl in matches]
 320             return self.playlist_result(
 321                 urlrs, playlist_id=video_id, playlist_title=video_title)
 322
 323         # Look for embedded Dailymotion player
 324         matches = re.findall(
 325             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 326         if matches:
 327             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 328                      for tuppl in matches]
 329             return self.playlist_result(
 330                 urlrs, playlist_id=video_id, playlist_title=video_title)
 331
 332         # Look for embedded Wistia player
 333         match = re.search(
 334             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 335         if match:
 336             return {
 337                 '_type': 'url_transparent',
 338                 'url': unescapeHTML(match.group('url')),
 339                 'ie_key': 'Wistia',
 340                 'uploader': video_uploader,
 341                 'title': video_title,
 342                 'id': video_id,
 343             }
 344
 345         # Look for embedded blip.tv player
 346         mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
 347         if mobj:
 348             return self.url_result('http://blip.tv/a/a-'+mobj.group(1), 'BlipTV')
 349         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9]+)', webpage)
 350         if mobj:
 351             return self.url_result(mobj.group(1), 'BlipTV')
 352
 353         # Look for Bandcamp pages with custom domain
 354         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 355         if mobj is not None:
 356             burl = unescapeHTML(mobj.group(1))
 357             # Don't set the extractor because it can be a track url or an album
 358             return self.url_result(burl)
 359
 360         # Look for embedded Vevo player
 361         mobj = re.search(
 362             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 363         if mobj is not None:
 364             return self.url_result(mobj.group('url'))
 365
 366         # Look for Ooyala videos
 367         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 368         if mobj is not None:
 369             return OoyalaIE._build_url_result(mobj.group(1))
 370
 371         # Look for Aparat videos
 372         mobj = re.search(r'<iframe src="(http://www\.aparat\.com/video/[^"]+)"', webpage)
 373         if mobj is not None:
 374             return self.url_result(mobj.group(1), 'Aparat')
 375
 376         # Look for MPORA videos
 377         mobj = re.search(r'<iframe .*?src="(http://mpora\.(?:com|de)/videos/[^"]+)"', webpage)
 378         if mobj is not None:
 379             return self.url_result(mobj.group(1), 'Mpora')
 380
 381         # Look for embedded NovaMov player
 382         mobj = re.search(
 383             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?novamov\.com/embed\.php.+?)\1', webpage)
 384         if mobj is not None:
 385             return self.url_result(mobj.group('url'), 'NovaMov')
 386
 387         # Look for embedded NowVideo player
 388         mobj = re.search(
 389             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?nowvideo\.(?:ch|sx|eu)/embed\.php.+?)\1', webpage)
 390         if mobj is not None:
 391             return self.url_result(mobj.group('url'), 'NowVideo')
 392
 393         # Look for embedded Facebook player
 394         mobj = re.search(
 395             r'<iframe[^>]+?src=(["\'])(?P<url>https://www\.facebook\.com/video/embed.+?)\1', webpage)
 396         if mobj is not None:
 397             return self.url_result(mobj.group('url'), 'Facebook')
 398
 399         # Look for embedded Huffington Post player
 400         mobj = re.search(
 401             r'<iframe[^>]+?src=(["\'])(?P<url>https?://embed\.live\.huffingtonpost\.com/.+?)\1', webpage)
 402         if mobj is not None:
 403             return self.url_result(mobj.group('url'), 'HuffPost')
 404
 405         # Start with something easy: JW Player in SWFObject
 406         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 407         if mobj is None:
 408             # Look for gorilla-vid style embedding
 409             mobj = re.search(r'(?s)(?:jw_plugins|JWPlayerOptions).*?file\s*:\s*["\'](.*?)["\']', webpage)
 410         if mobj is None:
 411             # Broaden the search a little bit
 412             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 413         if mobj is None:
 414             # Broaden the search a little bit: JWPlayer JS loader
 415             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
 416         if mobj is None:
 417             # Try to find twitter cards info
 418             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 419         if mobj is None:
 420             # We look for Open Graph info:
 421             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 422             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 423             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 424             if m_video_type is not None:
 425                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 426         if mobj is None:
 427             # HTML5 video
 428             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 429         if mobj is None:
 430             mobj = re.search(
 431                 r'(?i)<meta\s+(?=(?:[a-z-]+="[^"]+"\s+)*http-equiv="refresh")'
 432                 r'(?:[a-z-]+="[^"]+"\s+)*?content="[0-9]{,2};url=\'([^\']+)\'"',
 433                 webpage)
 434             if mobj:
 435                 new_url = mobj.group(1)
 436                 self.report_following_redirect(new_url)
 437                 return {
 438                     '_type': 'url',
 439                     'url': new_url,
 440                 }
 441         if mobj is None:
 442             raise ExtractorError('Unsupported URL: %s' % url)
 443
 444         # It's possible that one of the regexes
 445         # matched, but returned an empty group:
 446         if mobj.group(1) is None:
 447             raise ExtractorError('Did not find a valid video URL at %s' % url)
 448
 449         video_url = mobj.group(1)
 450         video_url = compat_urlparse.urljoin(url, video_url)
 451         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 452
 453         # Sometimes, jwplayer extraction will result in a YouTube URL
 454         if YoutubeIE.suitable(video_url):
 455             return self.url_result(video_url, 'Youtube')
 456
 457         # here's a fun little line of code for you:
 458         video_id = os.path.splitext(video_id)[0]
 459
 460         return {
 461             'id': video_id,
 462             'url': video_url,
 463             'uploader': video_uploader,
 464             'title': video_title,
 465         }