2 from __future__ import unicode_literals
7 from .common import InfoExtractor
11 compat_urllib_parse_urlparse,
21 class RutubeIE(InfoExtractor):
23 IE_DESC = 'Rutube videos'
24 _VALID_URL = r'https?://rutube\.ru/(?:video|(?:play/)?embed)/(?P<id>[\da-z]{32})'
27 'url': 'http://rutube.ru/video/3eac3b4561676c17df9132a9a1e62e3e/',
29 'id': '3eac3b4561676c17df9132a9a1e62e3e',
31 'title': 'Раненный кенгуру забежал в аптеку',
32 'description': 'http://www.ntdtv.ru ',
34 'uploader': 'NTDRussian',
35 'uploader_id': '29790',
36 'upload_date': '20131016',
40 # It requires ffmpeg (m3u8 download)
41 'skip_download': True,
44 'url': 'http://rutube.ru/play/embed/a10e53b86e8f349080f718582ce4c661',
45 'only_matching': True,
47 'url': 'http://rutube.ru/embed/a10e53b86e8f349080f718582ce4c661',
48 'only_matching': True,
50 'url': 'http://rutube.ru/video/3eac3b4561676c17df9132a9a1e62e3e/?pl_id=4252',
51 'only_matching': True,
55 def suitable(cls, url):
56 parts = compat_urllib_parse_urlparse(url)
57 params = compat_parse_qs(parts.query)
59 # see if URL without parameters is OK
60 res = super(RutubeIE, cls).suitable(url)
62 if params: # we only allow pl_id parameter in the url
63 res = res and 'pl_id' in params and len(params) == 1
68 def _extract_urls(webpage):
69 return [mobj.group('url') for mobj in re.finditer(
70 r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//rutube\.ru/embed/[\da-z]{32}.*?)\1',
73 def _real_extract(self, url):
74 video_id = self._match_id(url)
75 video = self._download_json(
76 'http://rutube.ru/api/video/%s/?format=json' % video_id,
77 video_id, 'Downloading video JSON')
79 # Some videos don't have the author field
80 author = video.get('author') or {}
82 options = self._download_json(
83 'http://rutube.ru/api/play/options/%s/?format=json' % video_id,
84 video_id, 'Downloading options JSON')
87 for format_id, format_url in options['video_balancer'].items():
88 ext = determine_ext(format_url)
90 formats.extend(self._extract_m3u8_formats(
91 format_url, video_id, 'mp4', m3u8_id=format_id, fatal=False))
93 formats.extend(self._extract_f4m_formats(
94 format_url, video_id, f4m_id=format_id, fatal=False))
98 'format_id': format_id,
100 self._sort_formats(formats)
104 'title': video['title'],
105 'description': video['description'],
106 'duration': video['duration'],
107 'view_count': video['hits'],
109 'thumbnail': video['thumbnail_url'],
110 'uploader': author.get('name'),
111 'uploader_id': compat_str(author['id']) if author else None,
112 'upload_date': unified_strdate(video['created_ts']),
113 'age_limit': 18 if video['is_adult'] else 0,
117 class RutubeEmbedIE(InfoExtractor):
118 IE_NAME = 'rutube:embed'
119 IE_DESC = 'Rutube embedded videos'
120 _VALID_URL = r'https?://rutube\.ru/(?:video|play)/embed/(?P<id>[0-9]+)'
123 'url': 'http://rutube.ru/video/embed/6722881?vk_puid37=&vk_puid38=',
125 'id': 'a10e53b86e8f349080f718582ce4c661',
127 'upload_date': '20131223',
128 'uploader_id': '297833',
129 'description': 'Видео группы ★http://vk.com/foxkidsreset★ музей Fox Kids и Jetix<br/><br/> восстановлено и сделано в шикоформате subziro89 http://vk.com/subziro89',
130 'uploader': 'subziro89 ILya',
131 'title': 'Мистический городок Эйри в Индиан 5 серия озвучка subziro89',
134 'skip_download': 'Requires ffmpeg',
137 'url': 'http://rutube.ru/play/embed/8083783',
138 'only_matching': True,
141 def _real_extract(self, url):
142 embed_id = self._match_id(url)
143 webpage = self._download_webpage(url, embed_id)
145 canonical_url = self._html_search_regex(
146 r'<link\s+rel="canonical"\s+href="([^"]+?)"', webpage,
148 return self.url_result(canonical_url, 'Rutube')
151 class RutubeChannelIE(InfoExtractor):
152 IE_NAME = 'rutube:channel'
153 IE_DESC = 'Rutube channels'
154 _VALID_URL = r'https?://rutube\.ru/tags/video/(?P<id>\d+)'
156 'url': 'http://rutube.ru/tags/video/1800/',
160 'playlist_mincount': 68,
163 _PAGE_TEMPLATE = 'http://rutube.ru/api/tags/video/%s/?page=%s&format=json'
165 def _extract_videos(self, channel_id, channel_title=None):
167 for pagenum in itertools.count(1):
168 page = self._download_json(
169 self._PAGE_TEMPLATE % (channel_id, pagenum),
170 channel_id, 'Downloading page %s' % pagenum)
171 results = page['results']
174 entries.extend(self.url_result(result['video_url'], 'Rutube') for result in results)
175 if not page['has_next']:
177 return self.playlist_result(entries, channel_id, channel_title)
179 def _real_extract(self, url):
180 mobj = re.match(self._VALID_URL, url)
181 channel_id = mobj.group('id')
182 return self._extract_videos(channel_id)
185 class RutubeMovieIE(RutubeChannelIE):
186 IE_NAME = 'rutube:movie'
187 IE_DESC = 'Rutube movies'
188 _VALID_URL = r'https?://rutube\.ru/metainfo/tv/(?P<id>\d+)'
191 _MOVIE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/?format=json'
192 _PAGE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/video?page=%s&format=json'
194 def _real_extract(self, url):
195 movie_id = self._match_id(url)
196 movie = self._download_json(
197 self._MOVIE_TEMPLATE % movie_id, movie_id,
198 'Downloading movie JSON')
199 movie_name = movie['name']
200 return self._extract_videos(movie_id, movie_name)
203 class RutubePersonIE(RutubeChannelIE):
204 IE_NAME = 'rutube:person'
205 IE_DESC = 'Rutube person videos'
206 _VALID_URL = r'https?://rutube\.ru/video/person/(?P<id>\d+)'
208 'url': 'http://rutube.ru/video/person/313878/',
212 'playlist_mincount': 37,
215 _PAGE_TEMPLATE = 'http://rutube.ru/api/video/person/%s/?page=%s&format=json'
218 class RutubePlaylistIE(InfoExtractor):
219 IE_NAME = 'rutube:playlist'
220 IE_DESC = 'Rutube playlists'
222 'url': 'https://rutube.ru/video/10b3a03fc01d5bbcc632a2f3514e8aab/?pl_id=4252&pl_type=source',
226 'playlist_count': 25,
229 _VALID_URL = r'https?://rutube\.ru/(?:video|(?:play/)?embed)/[\da-z]{32}/\?(?:.+)?pl_id=(?P<id>\d+)'
230 _PAGE_TEMPLATE = 'http://rutube.ru/api/playlist/source/%s/?page=%s'
234 params = compat_parse_qs(compat_urllib_parse_urlparse(url).query)
235 return params.get('pl_id') and int_or_none(params['pl_id'][0]) \
236 and params.get('pl_type')
238 def _real_extract(self, url):
239 playlist_id = self._match_id(url)
240 return self._extract_playlist(playlist_id)
242 def _extract_playlist(self, playlist_id):
244 for pagenum in itertools.count(1):
245 page_url = self._PAGE_TEMPLATE % (playlist_id, pagenum)
247 # download_json will sent an accept: application/xml header
248 page = self._download_json(page_url, playlist_id,
249 "Downloading metadata for page %s" % pagenum,
250 headers={'Accept': 'application/json'})
252 if not page['results']:
255 results = page['results']
256 for result in results:
257 entry = self.url_result(result.get('video_url'), 'Rutube')
258 category = try_get(result, lambda x: x['category']['name'])
260 'id': result.get('id'),
261 'uploader': try_get(result, lambda x: x['author']['name']),
262 'uploader_id': try_get(result, lambda x: x['author']['id']),
263 'upload_date': unified_strdate(result.get('created_ts')),
264 'title': result.get('title'),
265 'description': result.get('description'),
266 'thumbnail': result.get('thumbnail_url'),
267 'duration': int_or_none(result.get('duration')),
268 'category': [category] if category else None,
269 'age_limit': 18 if result.get('is_adult') else 0,
270 'view_count': int_or_none(result.get('hits')),
271 'is_live': result.get('is_livestream'),
272 'webpage_url': result.get('video_url'),
274 entries.append(entry)
276 if page['has_next'] is False:
279 return self.playlist_result(entries, playlist_id, page['name'])