2 from __future__ import unicode_literals
8 from .common import InfoExtractor
16 class RutubeIE(InfoExtractor):
18 IE_DESC = 'Rutube videos'
19 _VALID_URL = r'https?://rutube\.ru/video/(?P<long_id>\w+)'
22 'url': 'http://rutube.ru/video/3eac3b4561676c17df9132a9a1e62e3e/',
23 'file': '3eac3b4561676c17df9132a9a1e62e3e.mp4',
25 'title': 'Раненный кенгуру забежал в аптеку',
26 'uploader': 'NTDRussian',
27 'uploader_id': '29790',
30 # It requires ffmpeg (m3u8 download)
31 'skip_download': True,
35 def _get_api_response(self, short_id, subpath):
36 api_url = 'http://rutube.ru/api/play/%s/%s/?format=json' % (subpath, short_id)
37 response_json = self._download_webpage(api_url, short_id,
38 'Downloading %s json' % subpath)
39 return json.loads(response_json)
41 def _real_extract(self, url):
42 mobj = re.match(self._VALID_URL, url)
43 long_id = mobj.group('long_id')
44 webpage = self._download_webpage(url, long_id)
45 og_video = self._og_search_video_url(webpage)
46 short_id = compat_urlparse.urlparse(og_video).path[1:]
47 options = self._get_api_response(short_id, 'options')
48 trackinfo = self._get_api_response(short_id, 'trackinfo')
49 # Some videos don't have the author field
50 author = trackinfo.get('author') or {}
51 m3u8_url = trackinfo['video_balancer'].get('m3u8')
53 raise ExtractorError('Couldn\'t find m3u8 manifest url')
56 'id': trackinfo['id'],
57 'title': trackinfo['title'],
60 'thumbnail': options['thumbnail_url'],
61 'uploader': author.get('name'),
62 'uploader_id': compat_str(author['id']) if author else None,
66 class RutubeChannelIE(InfoExtractor):
67 IE_NAME = 'rutube:channel'
68 IE_DESC = 'Rutube channels'
69 _VALID_URL = r'http://rutube\.ru/tags/video/(?P<id>\d+)'
71 _PAGE_TEMPLATE = 'http://rutube.ru/api/tags/video/%s/?page=%s&format=json'
73 def _extract_videos(self, channel_id, channel_title=None):
75 for pagenum in itertools.count(1):
76 response_json = self._download_webpage(self._PAGE_TEMPLATE % (channel_id, pagenum),
77 channel_id, 'Downloading page %s' % pagenum)
78 page = json.loads(response_json)
79 if 'detail' in page and page['detail'] == 'Not found':
80 raise ExtractorError('Channel %s does not exist' % channel_id, expected=True)
81 results = page['results']
84 entries.extend(self.url_result(v['video_url'], 'Rutube') for v in results)
85 if page['has_next'] is False:
87 return self.playlist_result(entries, channel_id, channel_title)
89 def _real_extract(self, url):
90 mobj = re.match(self._VALID_URL, url)
91 channel_id = mobj.group('id')
92 return self._extract_videos(channel_id)
95 class RutubeMovieIE(RutubeChannelIE):
96 IE_NAME = 'rutube:movie'
97 IE_DESC = 'Rutube movies'
98 _VALID_URL = r'http://rutube\.ru/metainfo/tv/(?P<id>\d+)'
100 _MOVIE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/?format=json'
101 _PAGE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/video?page=%s&format=json'
103 def _real_extract(self, url):
104 mobj = re.match(self._VALID_URL, url)
105 movie_id = mobj.group('id')
106 movie_json = self._download_webpage(self._MOVIE_TEMPLATE % movie_id, movie_id,
107 'Downloading movie JSON')
108 movie = json.loads(movie_json)
109 if 'detail' in movie and movie['detail'] == 'Not found':
110 raise ExtractorError('Movie %s does not exist' % movie_id, expected=True)
111 movie_name = movie['name']
112 return self._extract_videos(movie_id, movie_name)