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