_ 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
  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         # embedded vimeo video
  42         {
  43             'add_ie': ['Vimeo'],
  44             'url': 'http://skillsmatter.com/podcast/home/move-semanticsperfect-forwarding-and-rvalue-references',
  45             'file': '22444065.mp4',
  46             'md5': '2903896e23df39722c33f015af0666e2',
  47             'info_dict': {
  48                 'title': 'ACCU 2011: Move Semantics,Perfect Forwarding, and Rvalue references- Scott Meyers- 13/04/2011',
  49                 'uploader_id': 'skillsmatter',
  50                 'uploader': 'Skills Matter',
  51             }
  52         },
  53         # bandcamp page with custom domain
  54         {
  55             'add_ie': ['Bandcamp'],
  56             'url': 'http://bronyrock.com/track/the-pony-mash',
  57             'file': '3235767654.mp3',
  58             'info_dict': {
  59                 'title': 'The Pony Mash',
  60                 'uploader': 'M_Pallante',
  61             },
  62             'skip': 'There is a limit of 200 free downloads / month for the test song',
  63         },
  64         # embedded brightcove video
  65         # it also tests brightcove videos that need to set the 'Referer' in the
  66         # http requests
  67         {
  68             'add_ie': ['Brightcove'],
  69             'url': 'http://www.bfmtv.com/video/bfmbusiness/cours-bourse/cours-bourse-l-analyse-technique-154522/',
  70             'info_dict': {
  71                 'id': '2765128793001',
  72                 'ext': 'mp4',
  73                 'title': 'Le cours de bourse : l’analyse technique',
  74                 'description': 'md5:7e9ad046e968cb2d1114004aba466fd9',
  75                 'uploader': 'BFM BUSINESS',
  76             },
  77             'params': {
  78                 'skip_download': True,
  79             },
  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             self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 166             return self.url_result('http://' + url)
 167         video_id = os.path.splitext(url.split('/')[-1])[0]
 168
 169         self.to_screen('%s: Requesting header' % video_id)
 170
 171         try:
 172             response = self._send_head(url)
 173
 174             # Check for redirect
 175             new_url = response.geturl()
 176             if url != new_url:
 177                 self.report_following_redirect(new_url)
 178                 return self.url_result(new_url)
 179
 180             # Check for direct link to a video
 181             content_type = response.headers.get('Content-Type', '')
 182             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 183             if m:
 184                 upload_date = response.headers.get('Last-Modified')
 185                 if upload_date:
 186                     upload_date = unified_strdate(upload_date)
 187                 return {
 188                     'id': video_id,
 189                     'title': os.path.splitext(url_basename(url))[0],
 190                     'formats': [{
 191                         'format_id': m.group('format_id'),
 192                         'url': url,
 193                         'vcodec': 'none' if m.group('type') == 'audio' else None
 194                     }],
 195                     'upload_date': upload_date,
 196                 }
 197
 198         except compat_urllib_error.HTTPError:
 199             # This may be a stupid server that doesn't like HEAD, our UA, or so
 200             pass
 201
 202         try:
 203             webpage = self._download_webpage(url, video_id)
 204         except ValueError:
 205             # since this is the last-resort InfoExtractor, if
 206             # this error is thrown, it'll be thrown here
 207             raise ExtractorError('Failed to download URL: %s' % url)
 208
 209         self.report_extraction(video_id)
 210
 211         # it's tempting to parse this further, but you would
 212         # have to take into account all the variations like
 213         #   Video Title - Site Name
 214         #   Site Name | Video Title
 215         #   Video Title - Tagline | Site Name
 216         # and so on and so forth; it's just not practical
 217         video_title = self._html_search_regex(
 218             r'(?s)<title>(.*?)</title>', webpage, 'video title',
 219             default='video')
 220
 221         # video uploader is domain name
 222         video_uploader = self._search_regex(
 223             r'^(?:https?://)?([^/]*)/.*', url, 'video uploader')
 224
 225         # Look for BrightCove:
 226         bc_url = BrightcoveIE._extract_brightcove_url(webpage)
 227         if bc_url is not None:
 228             self.to_screen('Brightcove video detected.')
 229             surl = smuggle_url(bc_url, {'Referer': url})
 230             return self.url_result(surl, 'Brightcove')
 231
 232         # Look for embedded (iframe) Vimeo player
 233         mobj = re.search(
 234             r'<iframe[^>]+?src="((?:https?:)?//player.vimeo.com/video/.+?)"', webpage)
 235         if mobj:
 236             player_url = unescapeHTML(mobj.group(1))
 237             surl = smuggle_url(player_url, {'Referer': url})
 238             return self.url_result(surl, 'Vimeo')
 239
 240         # Look for embedded (swf embed) Vimeo player
 241         mobj = re.search(
 242             r'<embed[^>]+?src="(https?://(?:www\.)?vimeo.com/moogaloop.swf.+?)"', webpage)
 243         if mobj:
 244             return self.url_result(mobj.group(1), 'Vimeo')
 245
 246         # Look for embedded YouTube player
 247         matches = re.findall(r'''(?x)
 248             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 249             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 250                 (?:embed|v)/.+?)
 251             \1''', webpage)
 252         if matches:
 253             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 254                      for tuppl in matches]
 255             return self.playlist_result(
 256                 urlrs, playlist_id=video_id, playlist_title=video_title)
 257
 258         # Look for embedded Dailymotion player
 259         matches = re.findall(
 260             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 261         if matches:
 262             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 263                      for tuppl in matches]
 264             return self.playlist_result(
 265                 urlrs, playlist_id=video_id, playlist_title=video_title)
 266
 267         # Look for embedded Wistia player
 268         match = re.search(
 269             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 270         if match:
 271             return {
 272                 '_type': 'url_transparent',
 273                 'url': unescapeHTML(match.group('url')),
 274                 'ie_key': 'Wistia',
 275                 'uploader': video_uploader,
 276                 'title': video_title,
 277                 'id': video_id,
 278             }
 279
 280         # Look for embedded blip.tv player
 281         mobj = re.search(r'<meta\s[^>]*https?://api\.blip\.tv/\w+/redirect/\w+/(\d+)', webpage)
 282         if mobj:
 283             return self.url_result('http://blip.tv/a/a-'+mobj.group(1), 'BlipTV')
 284         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*(https?://(?:\w+\.)?blip\.tv/(?:play/|api\.swf#)[a-zA-Z0-9]+)', webpage)
 285         if mobj:
 286             return self.url_result(mobj.group(1), 'BlipTV')
 287
 288         # Look for Bandcamp pages with custom domain
 289         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 290         if mobj is not None:
 291             burl = unescapeHTML(mobj.group(1))
 292             # Don't set the extractor because it can be a track url or an album
 293             return self.url_result(burl)
 294
 295         # Look for embedded Vevo player
 296         mobj = re.search(
 297             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 298         if mobj is not None:
 299             return self.url_result(mobj.group('url'))
 300
 301         # Look for Ooyala videos
 302         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 303         if mobj is not None:
 304             return OoyalaIE._build_url_result(mobj.group(1))
 305
 306         # Look for Aparat videos
 307         mobj = re.search(r'<iframe src="(http://www\.aparat\.com/video/[^"]+)"', webpage)
 308         if mobj is not None:
 309             return self.url_result(mobj.group(1), 'Aparat')
 310
 311         # Look for MPORA videos
 312         mobj = re.search(r'<iframe .*?src="(http://mpora\.com/videos/[^"]+)"', webpage)
 313         if mobj is not None:
 314             return self.url_result(mobj.group(1), 'Mpora')
 315
 316         # Look for embedded Novamov player
 317         mobj = re.search(
 318             r'<iframe[^>]+?src=(["\'])(?P<url>http://(?:(?:embed|www)\.)?novamov\.com/embed\.php.+?)\1', webpage)
 319         if mobj is not None:
 320             return self.url_result(mobj.group('url'), 'Novamov')
 321
 322         # Start with something easy: JW Player in SWFObject
 323         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 324         if mobj is None:
 325             # Look for gorilla-vid style embedding
 326             mobj = re.search(r'(?s)jw_plugins.*?file:\s*["\'](.*?)["\']', webpage)
 327         if mobj is None:
 328             # Broaden the search a little bit
 329             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 330         if mobj is None:
 331             # Broaden the search a little bit: JWPlayer JS loader
 332             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http(?![^\'"]+\.[0-9]+[\'"])[^\'"]+)["\']', webpage)
 333         if mobj is None:
 334             # Try to find twitter cards info
 335             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 336         if mobj is None:
 337             # We look for Open Graph info:
 338             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 339             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 340             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 341             if m_video_type is not None:
 342                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 343         if mobj is None:
 344             # HTML5 video
 345             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 346         if mobj is None:
 347             raise ExtractorError('Unsupported URL: %s' % url)
 348
 349         # It's possible that one of the regexes
 350         # matched, but returned an empty group:
 351         if mobj.group(1) is None:
 352             raise ExtractorError('Did not find a valid video URL at %s' % url)
 353
 354         video_url = mobj.group(1)
 355         video_url = compat_urlparse.urljoin(url, video_url)
 356         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 357
 358         # Sometimes, jwplayer extraction will result in a YouTube URL
 359         if YoutubeIE.suitable(video_url):
 360             return self.url_result(video_url, 'Youtube')
 361
 362         # here's a fun little line of code for you:
 363         video_id = os.path.splitext(video_id)[0]
 364
 365         return {
 366             'id': video_id,
 367             'url': video_url,
 368             'uploader': video_uploader,
 369             'title': video_title,
 370         }