4 from .common import InfoExtractor
11 class DailymotionIE(InfoExtractor):
12 """Information Extractor for Dailymotion"""
14 _VALID_URL = r'(?i)(?:https?://)?(?:www\.)?dailymotion\.[a-z]{2,3}/video/([^/]+)'
15 IE_NAME = u'dailymotion'
17 u'url': u'http://www.dailymotion.com/video/x33vw9_tutoriel-de-youtubeur-dl-des-video_tech',
18 u'file': u'x33vw9.mp4',
19 u'md5': u'392c4b85a60a90dc4792da41ce3144eb',
21 u"uploader": u"Alex and Van .",
22 u"title": u"Tutoriel de Youtubeur\"DL DES VIDEO DE YOUTUBE\""
26 def _real_extract(self, url):
27 # Extract id and simplified title from URL
28 mobj = re.match(self._VALID_URL, url)
30 video_id = mobj.group(1).split('_')[0].split('?')[0]
32 video_extension = 'mp4'
34 # Retrieve video webpage to extract further information
35 request = compat_urllib_request.Request(url)
36 request.add_header('Cookie', 'family_filter=off')
37 webpage = self._download_webpage(request, video_id)
39 # Extract URL, uploader and title from webpage
40 self.report_extraction(video_id)
42 video_uploader = self._search_regex([r'(?im)<span class="owner[^\"]+?">[^<]+?<a [^>]+?>([^<]+?)</a>',
43 # Looking for official user
44 r'<(?:span|a) .*?rel="author".*?>([^<]+?)</'],
45 webpage, 'video uploader')
47 video_upload_date = None
48 mobj = re.search(r'<div class="[^"]*uploaded_cont[^"]*" title="[^"]*">([0-9]{2})-([0-9]{2})-([0-9]{4})</div>', webpage)
50 video_upload_date = mobj.group(3) + mobj.group(2) + mobj.group(1)
52 embed_url = 'http://www.dailymotion.com/embed/video/%s' % video_id
53 embed_page = self._download_webpage(embed_url, video_id,
54 u'Downloading embed page')
55 info = self._search_regex(r'var info = ({.*?}),', embed_page, 'video info')
56 info = json.loads(info)
58 # TODO: support choosing qualities
60 for key in ['stream_h264_hd1080_url','stream_h264_hd_url',
61 'stream_h264_hq_url','stream_h264_url',
62 'stream_h264_ld_url']:
63 if info.get(key):#key in info and info[key]:
65 self.to_screen(u'Using %s' % key)
68 raise ExtractorError(u'Unable to extract video URL')
69 video_url = info[max_quality]
74 'uploader': video_uploader,
75 'upload_date': video_upload_date,
76 'title': self._og_search_title(webpage),
77 'ext': video_extension,
78 'thumbnail': info['thumbnail_url']