6 from .common import InfoExtractor
10 compat_urllib_request,
19 from .brightcove import BrightcoveIE
20 from .ooyala import OoyalaIE
23 class GenericIE(InfoExtractor):
24 IE_DESC = u'Generic downloader that works on some sites'
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',
33 u"uploader": u"www.hodiho.fr",
34 u"title": u"R\u00e9gis plante sa Jeep"
37 # embedded vimeo video
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',
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",
49 # bandcamp page with custom domain
51 u'add_ie': ['Bandcamp'],
52 u'url': u'http://bronyrock.com/track/the-pony-mash',
53 u'file': u'3235767654.mp3',
55 u'title': u'The Pony Mash',
56 u'uploader': u'M_Pallante',
58 u'skip': u'There is a limit of 200 free downloads / month for the test song',
60 # embedded brightcove video
61 # it also tests brightcove videos that need to set the 'Referer' in the
64 u'add_ie': ['Brightcove'],
65 u'url': u'http://www.bfmtv.com/video/bfmbusiness/cours-bourse/cours-bourse-l-analyse-technique-154522/',
67 u'id': u'2765128793001',
69 u'title': u'Le cours de bourse : l’analyse technique',
70 u'description': u'md5:7e9ad046e968cb2d1114004aba466fd9',
71 u'uploader': u'BFM BUSINESS',
74 u'skip_download': True,
77 # Direct link to a video
79 u'url': u'http://media.w3.org/2010/05/sintel/trailer.mp4',
80 u'file': u'trailer.mp4',
81 u'md5': u'67d406c2bcb6af27fa886f31aa934bbe',
85 u'upload_date': u'20100513',
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',
93 u'id': u'BwY2RxaTrTkslxOfcan0UCf0YqyvWysJ',
95 u'title': u'2cc213299525360.mov', #that's what we get
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)
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)
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):
116 class HEADRedirectHandler(compat_urllib_request.HTTPRedirectHandler):
118 Subclass the HTTPRedirectHandler to make it use our
119 HeadRequest also on the redirected URL
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,
128 origin_req_host=req.get_origin_req_host(),
131 raise compat_urllib_error.HTTPError(req.get_full_url(), code, msg, headers, fp)
133 class HTTPMethodFallback(compat_urllib_request.BaseHandler):
135 Fallback to GET if HEAD is not allowed (405 HTTP error)
137 def http_error_405(self, req, fp, code, msg, headers):
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(),
145 origin_req_host=req.get_origin_req_host(),
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())
155 response = opener.open(HeadRequest(url))
157 raise ExtractorError(u'Invalid URL protocol')
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]
168 response = self._send_head(url)
171 new_url = response.geturl()
173 self.report_following_redirect(new_url)
174 return self.url_result(new_url)
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)
180 upload_date = response.headers.get('Last-Modified')
182 upload_date = unified_strdate(upload_date)
185 'title': os.path.splitext(url_basename(url))[0],
187 'format_id': m.group('format_id'),
189 'vcodec': u'none' if m.group('type') == 'audio' else None
191 'upload_date': upload_date,
194 except compat_urllib_error.HTTPError:
195 # This may be a stupid server that doesn't like HEAD, our UA, or so
199 webpage = self._download_webpage(url, video_id)
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)
205 self.report_extraction(video_id)
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',
217 # video uploader is domain name
218 video_uploader = self._search_regex(
219 r'^(?:https?://)?([^/]*)/.*', url, u'video uploader')
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')
227 # Look for embedded Vimeo player
229 r'<iframe[^>]+?src="(https?://player.vimeo.com/video/.+?)"', webpage)
231 player_url = unescapeHTML(mobj.group(1))
232 surl = smuggle_url(player_url, {'Referer': url})
233 return self.url_result(surl, 'Vimeo')
235 # Look for embedded YouTube player
236 matches = re.findall(
237 r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?youtube\.com/embed/.+?)\1', webpage)
239 urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Youtube')
240 for tuppl in matches]
241 return self.playlist_result(
242 urlrs, playlist_id=video_id, playlist_title=video_title)
244 # Look for embedded Dailymotion player
245 matches = re.findall(
246 r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?dailymotion\.com/embed/video/.+?)\1', webpage)
248 urlrs = [self.url_result(unescapeHTML(tuppl[1]), 'Dailymotion')
249 for tuppl in matches]
250 return self.playlist_result(
251 urlrs, playlist_id=video_id, playlist_title=video_title)
253 # Look for embedded Wistia player
255 r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
258 '_type': 'url_transparent',
259 'url': unescapeHTML(match.group('url')),
261 'uploader': video_uploader,
262 'title': video_title,
266 # Look for embedded blip.tv player
267 mobj = re.search(r'<meta\s[^>]*https?://api.blip.tv/\w+/redirect/\w+/(\d+)', webpage)
269 return self.url_result('http://blip.tv/seo/-'+mobj.group(1), 'BlipTV')
270 mobj = re.search(r'<(?:iframe|embed|object)\s[^>]*https?://(?:\w+\.)?blip.tv/(?:play/|api\.swf#)([a-zA-Z0-9]+)', webpage)
272 player_url = 'http://blip.tv/play/%s.x?p=1' % mobj.group(1)
273 player_page = self._download_webpage(player_url, mobj.group(1))
274 blip_video_id = self._search_regex(r'data-episode-id="(\d+)', player_page, u'blip_video_id', fatal=False)
276 return self.url_result('http://blip.tv/seo/-'+blip_video_id, 'BlipTV')
278 # Look for Bandcamp pages with custom domain
279 mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage)
281 burl = unescapeHTML(mobj.group(1))
282 # Don't set the extractor because it can be a track url or an album
283 return self.url_result(burl)
285 # Look for embedded Vevo player
287 r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:cache\.)?vevo\.com/.+?)\1', webpage)
289 return self.url_result(mobj.group('url'))
291 # Look for Ooyala videos
292 mobj = re.search(r'player.ooyala.com/[^"?]+\?[^"]*?(?:embedCode|ec)=([^"&]+)', webpage)
294 return OoyalaIE._build_url_result(mobj.group(1))
296 # Start with something easy: JW Player in SWFObject
297 mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage)
299 # Broaden the search a little bit
300 mobj = re.search(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)', webpage)
302 # Broaden the search a little bit: JWPlayer JS loader
303 mobj = re.search(r'[^A-Za-z0-9]?file["\']?:\s*["\'](http[^\'"]*)', webpage)
305 # Try to find twitter cards info
306 mobj = re.search(r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)
308 # We look for Open Graph info:
309 # We have to match any number spaces between elements, some sites try to align them (eg.: statigr.am)
310 m_video_type = re.search(r'<meta.*?property="og:video:type".*?content="video/(.*?)"', webpage)
311 # We only look in og:video if the MIME type is a video, don't try if it's a Flash player:
312 if m_video_type is not None:
313 mobj = re.search(r'<meta.*?property="og:video".*?content="(.*?)"', webpage)
316 mobj = re.search(r'<video[^<]*(?:>.*?<source.*?)? src="([^"]+)"', webpage, flags=re.DOTALL)
318 raise ExtractorError(u'Unsupported URL: %s' % url)
320 # It's possible that one of the regexes
321 # matched, but returned an empty group:
322 if mobj.group(1) is None:
323 raise ExtractorError(u'Did not find a valid video URL at %s' % url)
325 video_url = mobj.group(1)
326 video_url = compat_urlparse.urljoin(url, video_url)
327 video_id = compat_urllib_parse.unquote(os.path.basename(video_url))
329 # here's a fun little line of code for you:
330 video_id = os.path.splitext(video_id)[0]
335 'uploader': video_uploader,
336 'title': video_title,