_ Git - youtube-dl/blob - youtube_dl/extractor/generic.py

   1 # encoding: utf-8
   2
   3 import os
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     compat_urllib_error,
   9     compat_urllib_parse,
  10     compat_urllib_request,
  11     compat_urlparse,
  12
  13     ExtractorError,
  14     smuggle_url,
  15     unescapeHTML,
  16     unified_strdate,
  17     url_basename,
  18 )
  19 from .brightcove import BrightcoveIE
  20 from .ooyala import OoyalaIE
  21
  22
  23 class GenericIE(InfoExtractor):
  24     IE_DESC = u'Generic downloader that works on some sites'
  25     _VALID_URL = r'.*'
  26     IE_NAME = u'generic'
  27     _TESTS = [
  28         {
  29             u'url': u'http://www.hodiho.fr/2013/02/regis-plante-sa-jeep.html',
  30             u'file': u'13601338388002.mp4',
  31             u'md5': u'6e15c93721d7ec9e9ca3fdbf07982cfd',
  32             u'info_dict': {
  33                 u"uploader": u"www.hodiho.fr",
  34                 u"title": u"R\u00e9gis plante sa Jeep"
  35             }
  36         },
  37         # embedded vimeo video
  38         {
  39             u'add_ie': ['Vimeo'],
  40             u'url': u'http://skillsmatter.com/podcast/home/move-semanticsperfect-forwarding-and-rvalue-references',
  41             u'file': u'22444065.mp4',
  42             u'md5': u'2903896e23df39722c33f015af0666e2',
  43             u'info_dict': {
  44                 u'title': u'ACCU 2011: Move Semantics,Perfect Forwarding, and Rvalue references- Scott Meyers- 13/04/2011',
  45                 u"uploader_id": u"skillsmatter",
  46                 u"uploader": u"Skills Matter",
  47             }
  48         },
  49         # bandcamp page with custom domain
  50         {
  51             u'add_ie': ['Bandcamp'],
  52             u'url': u'http://bronyrock.com/track/the-pony-mash',
  53             u'file': u'3235767654.mp3',
  54             u'info_dict': {
  55                 u'title': u'The Pony Mash',
  56                 u'uploader': u'M_Pallante',
  57             },
  58             u'skip': u'There is a limit of 200 free downloads / month for the test song',
  59         },
  60         # embedded brightcove video
  61         # it also tests brightcove videos that need to set the 'Referer' in the
  62         # http requests
  63         {
  64             u'add_ie': ['Brightcove'],
  65             u'url': u'http://www.bfmtv.com/video/bfmbusiness/cours-bourse/cours-bourse-l-analyse-technique-154522/',
  66             u'info_dict': {
  67                 u'id': u'2765128793001',
  68                 u'ext': u'mp4',
  69                 u'title': u'Le cours de bourse : l’analyse technique',
  70                 u'description': u'md5:7e9ad046e968cb2d1114004aba466fd9',
  71                 u'uploader': u'BFM BUSINESS',
  72             },
  73             u'params': {
  74                 u'skip_download': True,
  75             },
  76         },
  77         # Direct link to a video
  78         {
  79             u'url': u'http://media.w3.org/2010/05/sintel/trailer.mp4',
  80             u'file': u'trailer.mp4',
  81             u'md5': u'67d406c2bcb6af27fa886f31aa934bbe',
  82             u'info_dict': {
  83                 u'id': u'trailer',
  84                 u'title': u'trailer',
  85                 u'upload_date': u'20100513',
  86             }
  87         },
  88         # ooyala video
  89         {
  90             u'url': u'http://www.rollingstone.com/music/videos/norwegian-dj-cashmere-cat-goes-spartan-on-with-me-premiere-20131219',
  91             u'md5': u'5644c6ca5d5782c1d0d350dad9bd840c',
  92             u'info_dict': {
  93                 u'id': u'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ',
  94                 u'ext': u'mp4',
  95                 u'title': u'2cc213299525360.mov', #that's what we get
  96             },
  97         },
  98     ]
  99
 100     def report_download_webpage(self, video_id):
 101         """Report webpage download."""
 102         if not self._downloader.params.get('test', False):
 103             self._downloader.report_warning(u'Falling back on generic information extractor.')
 104         super(GenericIE, self).report_download_webpage(video_id)
 105
 106     def report_following_redirect(self, new_url):
 107         """Report information extraction."""
 108         self._downloader.to_screen(u'[redirect] Following redirect to %s' % new_url)
 109
 110     def _send_head(self, url):
 111         """Check if it is a redirect, like url shorteners, in case return the new url."""
 112         class HeadRequest(compat_urllib_request.Request):
 113             def get_method(self):
 114                 return "HEAD"
 115
 116         class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
 117             """
 118             Subclass the HTTPRedirectHandler to make it use our
 119             HeadRequest also on the redirected URL
 120             """
 121             def redirect_request(self, req, fp, code, msg, headers, newurl):
 122                 if code in (301, 302, 303, 307):
 123                     newurl = newurl.replace(' ', '%20')
 124                     newheaders = dict((k,v) for k,v in req.headers.items()
 125                                       if k.lower() not in ("content-length", "content-type"))
 126                     return HeadRequest(newurl,
 127                                        headers=newheaders,
 128                                        origin_req_host=req.get_origin_req_host(),
 129                                        unverifiable=True)
 130                 else:
 131                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 132
 133         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 134             """
 135             Fallback to GET if HEAD is not allowed (405 HTTP error)
 136             """
 137             def http_error_405(self, req, fp, code, msg, headers):
 138                 fp.read()
 139                 fp.close()
 140
 141                 newheaders = dict((k,v) for k,v in req.headers.items()
 142                                   if k.lower() not in ("content-length", "content-type"))
 143                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 144                                                  headers=newheaders,
 145                                                  origin_req_host=req.get_origin_req_host(),
 146                                                  unverifiable=True))
 147
 148         # Build our opener
 149         opener = compat_urllib_request.OpenerDirector()
 150         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 151                         HTTPMethodFallback, HEADRedirectHandler,
 152                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 153             opener.add_handler(handler())
 154
 155         response = opener.open(HeadRequest(url))
 156         if response is None:
 157             raise ExtractorError(u'Invalid URL protocol')
 158         return response
 159
 160     def _real_extract(self, url):
 161         parsed_url = compat_urlparse.urlparse(url)
 162         if not parsed_url.scheme:
 163             self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 164             return self.url_result('http://' + url)
 165         video_id = os.path.splitext(url.split('/')[-1])[0]
 166
 167         try:
 168             response = self._send_head(url)
 169
 170             # Check for redirect
 171             new_url = response.geturl()
 172             if url != new_url:
 173                 self.report_following_redirect(new_url)
 174                 return self.url_result(new_url)
 175
 176             # Check for direct link to a video
 177             content_type = response.headers.get('Content-Type', '')
 178             m = re.match(r'^(?P<type>audio|video|application(?=/ogg$))/(?P<format_id>.+)$', content_type)
 179             if m:
 180                 upload_date = response.headers.get('Last-Modified')
 181                 if upload_date:
 182                     upload_date = unified_strdate(upload_date)
 183                 return {
 184                     'id': video_id,
 185                     'title': os.path.splitext(url_basename(url))[0],
 186                     'formats': [{
 187                         'format_id': m.group('format_id'),
 188                         'url': url,
 189                         'vcodec': u'none' if m.group('type') == 'audio' else None
 190                     }],
 191                     'upload_date': upload_date,
 192                 }
 193
 194         except compat_urllib_error.HTTPError:
 195             # This may be a stupid server that doesn't like HEAD, our UA, or so
 196             pass
 197
 198         try:
 199             webpage = self._download_webpage(url, video_id)
 200         except ValueError:
 201             # since this is the last-resort InfoExtractor, if
 202             # this error is thrown, it'll be thrown here
 203             raise ExtractorError(u'Failed to download URL: %s' % url)
 204
 205         self.report_extraction(video_id)
 206
 207         # it's tempting to parse this further, but you would
 208         # have to take into account all the variations like
 209         #   Video Title - Site Name
 210         #   Site Name | Video Title
 211         #   Video Title - Tagline | Site Name
 212         # and so on and so forth; it's just not practical
 213         video_title = self._html_search_regex(
 214             r'(?s)<title>(.*?)</title>', webpage, u'video title',
 215             default=u'video')
 216
 217         # video uploader is domain name
 218         video_uploader = self._search_regex(
 219             r'^(?:https?://)?([^/]*)/.*', url, u'video uploader')
 220
 221         # Look for BrightCove:
 222         bc_url = BrightcoveIE._extract_brightcove_url(webpage)
 223         if bc_url is not None:
 224             self.to_screen(u'Brightcove video detected.')
 225             return self.url_result(bc_url, 'Brightcove')
 226
 227         # Look for embedded Vimeo player
 228         mobj = re.search(
 229             r'<iframe[^>]+?src="(https?://player.vimeo.com/video/.+?)"', webpage)
 230         if mobj:
 231             player_url = unescapeHTML(mobj.group(1))
 232             surl = smuggle_url(player_url, {'Referer': url})
 233             return self.url_result(surl, 'Vimeo')
 234
 235         # Look for embedded YouTube player
 236         matches = re.findall(r'''(?x)
 237             (?:<iframe[^>]+?src=|embedSWF\(\s*)
 238             (["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/
 239                 (?:embed|v)/.+?)
 240             \1''', webpage)
 241         if matches:
 242             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 243                      for tuppl in matches]
 244             return self.playlist_result(
 245                 urlrs, playlist_id=video_id, playlist_title=video_title)
 246
 247         # Look for embedded Dailymotion player
 248         matches = re.findall(
 249             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
 250         if matches:
 251             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
 252                      for tuppl in matches]
 253             return self.playlist_result(
 254                 urlrs, playlist_id=video_id, playlist_title=video_title)
 255
 256         # Look for embedded Wistia player
 257         match = re.search(
 258             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
 259         if match:
 260             return {
 261                 '_type': 'url_transparent',
 262                 'url': unescapeHTML(match.group('url')),
 263                 'ie_key': 'Wistia',
 264                 'uploader': video_uploader,
 265                 'title': video_title,
 266                 'id': video_id,
 267             }
 268
 269         # Look for embedded blip.tv player
 270         mobj = re.search(r'<meta\s[^>]*https?://api.blip.tv/\w+/redirect/\w+/(\d+)', webpage)
 271         if mobj:
 272             return self.url_result('http://blip.tv/seo/-'+mobj.group(1), 'BlipTV')
 273         mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*https?://(?:\w+\.)?blip.tv/(?:play/|api\.swf#)([a-zA-Z0-9]+)', webpage)
 274         if mobj:
 275             player_url = 'http://blip.tv/play/%s.x?p=1' % mobj.group(1)
 276             player_page = self._download_webpage(player_url, mobj.group(1))
 277             blip_video_id = self._search_regex(r'data-episode-id="(\d+)', player_page, u'blip_video_id', fatal=False)
 278             if blip_video_id:
 279                 return self.url_result('http://blip.tv/seo/-'+blip_video_id, 'BlipTV')
 280
 281         # Look for Bandcamp pages with custom domain
 282         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 283         if mobj is not None:
 284             burl = unescapeHTML(mobj.group(1))
 285             # Don't set the extractor because it can be a track url or an album
 286             return self.url_result(burl)
 287
 288         # Look for embedded Vevo player
 289         mobj = re.search(
 290             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
 291         if mobj is not None:
 292             return self.url_result(mobj.group('url'))
 293
 294         # Look for Ooyala videos
 295         mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
 296         if mobj is not None:
 297             return OoyalaIE._build_url_result(mobj.group(1))
 298
 299         # Start with something easy: JW Player in SWFObject
 300         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 301         if mobj is None:
 302             # Broaden the search a little bit
 303             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 304         if mobj is None:
 305             # Broaden the search a little bit: JWPlayer JS loader
 306             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http[^\'"]*)', webpage)
 307         if mobj is None:
 308             # Try to find twitter cards info
 309             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 310         if mobj is None:
 311             # We look for Open Graph info:
 312             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 313             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 314             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 315             if m_video_type is not None:
 316                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 317         if mobj is None:
 318             # HTML5 video
 319             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 320         if mobj is None:
 321             raise ExtractorError(u'Unsupported URL: %s' % url)
 322
 323         # It's possible that one of the regexes
 324         # matched, but returned an empty group:
 325         if mobj.group(1) is None:
 326             raise ExtractorError(u'Did not find a valid video URL at %s' % url)
 327
 328         video_url = mobj.group(1)
 329         video_url = compat_urlparse.urljoin(url, video_url)
 330         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 331
 332         # here's a fun little line of code for you:
 333         video_id = os.path.splitext(video_id)[0]
 334
 335         return {
 336             'id':       video_id,
 337             'url':      video_url,
 338             'uploader': video_uploader,
 339             'title':    video_title,
 340         }