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