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 )
  17 from .brightcove import BrightcoveIE
  18
  19
  20 class GenericIE(InfoExtractor):
  21     IE_DESC = u'Generic downloader that works on some sites'
  22     _VALID_URL = r'.*'
  23     IE_NAME = u'generic'
  24     _TESTS = [
  25         {
  26             u'url': u'http://www.hodiho.fr/2013/02/regis-plante-sa-jeep.html',
  27             u'file': u'13601338388002.mp4',
  28             u'md5': u'6e15c93721d7ec9e9ca3fdbf07982cfd',
  29             u'info_dict': {
  30                 u"uploader": u"www.hodiho.fr",
  31                 u"title": u"R\u00e9gis plante sa Jeep"
  32             }
  33         },
  34         # embedded vimeo video
  35         {
  36             u'add_ie': ['Vimeo'],
  37             u'url': u'http://skillsmatter.com/podcast/home/move-semanticsperfect-forwarding-and-rvalue-references',
  38             u'file': u'22444065.mp4',
  39             u'md5': u'2903896e23df39722c33f015af0666e2',
  40             u'info_dict': {
  41                 u'title': u'ACCU 2011: Move Semantics,Perfect Forwarding, and Rvalue references- Scott Meyers- 13/04/2011',
  42                 u"uploader_id": u"skillsmatter",
  43                 u"uploader": u"Skills Matter",
  44             }
  45         },
  46         # bandcamp page with custom domain
  47         {
  48             u'add_ie': ['Bandcamp'],
  49             u'url': u'http://bronyrock.com/track/the-pony-mash',
  50             u'file': u'3235767654.mp3',
  51             u'info_dict': {
  52                 u'title': u'The Pony Mash',
  53                 u'uploader': u'M_Pallante',
  54             },
  55             u'skip': u'There is a limit of 200 free downloads / month for the test song',
  56         },
  57         # embedded brightcove video
  58         # it also tests brightcove videos that need to set the 'Referer' in the
  59         # http requests
  60         {
  61             u'add_ie': ['Brightcove'],
  62             u'url': u'http://www.bfmtv.com/video/bfmbusiness/cours-bourse/cours-bourse-l-analyse-technique-154522/',
  63             u'info_dict': {
  64                 u'id': u'2765128793001',
  65                 u'ext': u'mp4',
  66                 u'title': u'Le cours de bourse : l’analyse technique',
  67                 u'description': u'md5:7e9ad046e968cb2d1114004aba466fd9',
  68                 u'uploader': u'BFM BUSINESS',
  69             },
  70             u'params': {
  71                 u'skip_download': True,
  72             },
  73         },
  74     ]
  75
  76     def report_download_webpage(self, video_id):
  77         """Report webpage download."""
  78         if not self._downloader.params.get('test', False):
  79             self._downloader.report_warning(u'Falling back on generic information extractor.')
  80         super(GenericIE, self).report_download_webpage(video_id)
  81
  82     def report_following_redirect(self, new_url):
  83         """Report information extraction."""
  84         self._downloader.to_screen(u'[redirect] Following redirect to %s' % new_url)
  85
  86     def _test_redirect(self, url):
  87         """Check if it is a redirect, like url shorteners, in case return the new url."""
  88         class HeadRequest(compat_urllib_request.Request):
  89             def get_method(self):
  90                 return "HEAD"
  91
  92         class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
  93             """
  94             Subclass the HTTPRedirectHandler to make it use our
  95             HeadRequest also on the redirected URL
  96             """
  97             def redirect_request(self, req, fp, code, msg, headers, newurl):
  98                 if code in (301, 302, 303, 307):
  99                     newurl = newurl.replace(' ', '%20')
 100                     newheaders = dict((k,v) for k,v in req.headers.items()
 101                                       if k.lower() not in ("content-length", "content-type"))
 102                     return HeadRequest(newurl,
 103                                        headers=newheaders,
 104                                        origin_req_host=req.get_origin_req_host(),
 105                                        unverifiable=True)
 106                 else:
 107                     raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
 108
 109         class HTTPMethodFallback(compat_urllib_request.BaseHandler):
 110             """
 111             Fallback to GET if HEAD is not allowed (405 HTTP error)
 112             """
 113             def http_error_405(self, req, fp, code, msg, headers):
 114                 fp.read()
 115                 fp.close()
 116
 117                 newheaders = dict((k,v) for k,v in req.headers.items()
 118                                   if k.lower() not in ("content-length", "content-type"))
 119                 return self.parent.open(compat_urllib_request.Request(req.get_full_url(),
 120                                                  headers=newheaders,
 121                                                  origin_req_host=req.get_origin_req_host(),
 122                                                  unverifiable=True))
 123
 124         # Build our opener
 125         opener = compat_urllib_request.OpenerDirector()
 126         for handler in [compat_urllib_request.HTTPHandler, compat_urllib_request.HTTPDefaultErrorHandler,
 127                         HTTPMethodFallback, HEADRedirectHandler,
 128                         compat_urllib_request.HTTPErrorProcessor, compat_urllib_request.HTTPSHandler]:
 129             opener.add_handler(handler())
 130
 131         response = opener.open(HeadRequest(url))
 132         if response is None:
 133             raise ExtractorError(u'Invalid URL protocol')
 134         new_url = response.geturl()
 135
 136         if url == new_url:
 137             return False
 138
 139         self.report_following_redirect(new_url)
 140         return new_url
 141
 142     def _real_extract(self, url):
 143         parsed_url = compat_urlparse.urlparse(url)
 144         if not parsed_url.scheme:
 145             self._downloader.report_warning('The url doesn\'t specify the protocol, trying with http')
 146             return self.url_result('http://' + url)
 147
 148         try:
 149             new_url = self._test_redirect(url)
 150             if new_url:
 151                 return [self.url_result(new_url)]
 152         except compat_urllib_error.HTTPError:
 153             # This may be a stupid server that doesn't like HEAD, our UA, or so
 154             pass
 155
 156         video_id = url.split('/')[-1]
 157         try:
 158             webpage = self._download_webpage(url, video_id)
 159         except ValueError:
 160             # since this is the last-resort InfoExtractor, if
 161             # this error is thrown, it'll be thrown here
 162             raise ExtractorError(u'Failed to download URL: %s' % url)
 163
 164         self.report_extraction(video_id)
 165
 166         # it's tempting to parse this further, but you would
 167         # have to take into account all the variations like
 168         #   Video Title - Site Name
 169         #   Site Name | Video Title
 170         #   Video Title - Tagline | Site Name
 171         # and so on and so forth; it's just not practical
 172         video_title = self._html_search_regex(r'<title>(.*)</title>',
 173             webpage, u'video title', default=u'video', flags=re.DOTALL)
 174
 175         # Look for BrightCove:
 176         bc_url = BrightcoveIE._extract_brightcove_url(webpage)
 177         if bc_url is not None:
 178             self.to_screen(u'Brightcove video detected.')
 179             return self.url_result(bc_url, 'Brightcove')
 180
 181         # Look for embedded Vimeo player
 182         mobj = re.search(
 183             r'<iframe[^>]+?src="(https?://player.vimeo.com/video/.+?)"', webpage)
 184         if mobj:
 185             player_url = unescapeHTML(mobj.group(1))
 186             surl = smuggle_url(player_url, {'Referer': url})
 187             return self.url_result(surl, 'Vimeo')
 188
 189         # Look for embedded YouTube player
 190         matches = re.findall(
 191             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube.com/embed/.+?)\1', webpage)
 192         if matches:
 193             urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
 194                      for tuppl in matches]
 195             return self.playlist_result(
 196                 urlrs, playlist_id=video_id, playlist_title=video_title)
 197
 198         # Look for Bandcamp pages with custom domain
 199         mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
 200         if mobj is not None:
 201             burl = unescapeHTML(mobj.group(1))
 202             return self.url_result(burl, 'Bandcamp')
 203
 204         # Start with something easy: JW Player in SWFObject
 205         mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
 206         if mobj is None:
 207             # Broaden the search a little bit
 208             mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
 209         if mobj is None:
 210             # Broaden the search a little bit: JWPlayer JS loader
 211             mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http[^\'"&]*)', webpage)
 212         if mobj is None:
 213             # Try to find twitter cards info
 214             mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
 215         if mobj is None:
 216             # We look for Open Graph info:
 217             # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
 218             m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
 219             # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
 220             if m_video_type is not None:
 221                 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
 222         if mobj is None:
 223             # HTML5 video
 224             mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
 225         if mobj is None:
 226             raise ExtractorError(u'Unsupported URL: %s' % url)
 227
 228         # It's possible that one of the regexes
 229         # matched, but returned an empty group:
 230         if mobj.group(1) is None:
 231             raise ExtractorError(u'Did not find a valid video URL at %s' % url)
 232
 233         video_url = mobj.group(1)
 234         video_url = compat_urlparse.urljoin(url, video_url)
 235         video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
 236
 237         # here's a fun little line of code for you:
 238         video_extension = os.path.splitext(video_id)[1][1:]
 239         video_id = os.path.splitext(video_id)[0]
 240
 241         # video uploader is domain name
 242         video_uploader = self._search_regex(r'(?:https?://)?([^/]*)/.*',
 243             url, u'video uploader')
 244
 245         return [{
 246             'id':       video_id,
 247             'url':      video_url,
 248             'uploader': video_uploader,
 249             'upload_date':  None,
 250             'title':    video_title,
 251             'ext':      video_extension,
 252         }]