2 from __future__ import unicode_literals
7 from .common import InfoExtractor
11 compat_urllib_parse_urlparse,
21 class RutubeBaseIE(InfoExtractor):
22 def _extract_video(self, video, video_id=None, require_title=True):
23 title = video['title'] if require_title else video.get('title')
25 age_limit = video.get('is_adult')
26 if age_limit is not None:
27 age_limit = 18 if age_limit is True else 0
29 uploader_id = try_get(video, lambda x: x['author']['id'])
30 category = try_get(video, lambda x: x['category']['name'])
33 'id': video.get('id') or video_id,
35 'description': video.get('description'),
36 'thumbnail': video.get('thumbnail_url'),
37 'duration': int_or_none(video.get('duration')),
38 'uploader': try_get(video, lambda x: x['author']['name']),
39 'uploader_id': compat_str(uploader_id) if uploader_id else None,
40 'timestamp': unified_timestamp(video.get('created_ts')),
41 'category': [category] if category else None,
42 'age_limit': age_limit,
43 'view_count': int_or_none(video.get('hits')),
44 'comment_count': int_or_none(video.get('comments_count')),
45 'is_live': video.get('is_livestream'),
49 class RutubeIE(RutubeBaseIE):
51 IE_DESC = 'Rutube videos'
52 _VALID_URL = r'https?://rutube\.ru/(?:video|(?:play/)?embed)/(?P<id>[\da-z]{32})'
55 'url': 'http://rutube.ru/video/3eac3b4561676c17df9132a9a1e62e3e/',
56 'md5': '79938ade01294ef7e27574890d0d3769',
58 'id': '3eac3b4561676c17df9132a9a1e62e3e',
60 'title': 'Раненный кенгуру забежал в аптеку',
61 'description': 'http://www.ntdtv.ru ',
63 'uploader': 'NTDRussian',
64 'uploader_id': '29790',
65 'timestamp': 1381943602,
66 'upload_date': '20131016',
70 'url': 'http://rutube.ru/play/embed/a10e53b86e8f349080f718582ce4c661',
71 'only_matching': True,
73 'url': 'http://rutube.ru/embed/a10e53b86e8f349080f718582ce4c661',
74 'only_matching': True,
76 'url': 'http://rutube.ru/video/3eac3b4561676c17df9132a9a1e62e3e/?pl_id=4252',
77 'only_matching': True,
79 'url': 'https://rutube.ru/video/10b3a03fc01d5bbcc632a2f3514e8aab/?pl_type=source',
80 'only_matching': True,
84 def suitable(cls, url):
85 return False if RutubePlaylistIE.suitable(url) else super(RutubeIE, cls).suitable(url)
88 def _extract_urls(webpage):
89 return [mobj.group('url') for mobj in re.finditer(
90 r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//rutube\.ru/embed/[\da-z]{32}.*?)\1',
93 def _real_extract(self, url):
94 video_id = self._match_id(url)
96 video = self._download_json(
97 'http://rutube.ru/api/video/%s/?format=json' % video_id,
98 video_id, 'Downloading video JSON')
100 info = self._extract_video(video, video_id)
102 options = self._download_json(
103 'http://rutube.ru/api/play/options/%s/?format=json' % video_id,
104 video_id, 'Downloading options JSON')
107 for format_id, format_url in options['video_balancer'].items():
108 ext = determine_ext(format_url)
110 formats.extend(self._extract_m3u8_formats(
111 format_url, video_id, 'mp4', m3u8_id=format_id, fatal=False))
113 formats.extend(self._extract_f4m_formats(
114 format_url, video_id, f4m_id=format_id, fatal=False))
118 'format_id': format_id,
120 self._sort_formats(formats)
122 info['formats'] = formats
126 class RutubeEmbedIE(InfoExtractor):
127 IE_NAME = 'rutube:embed'
128 IE_DESC = 'Rutube embedded videos'
129 _VALID_URL = r'https?://rutube\.ru/(?:video|play)/embed/(?P<id>[0-9]+)'
132 'url': 'http://rutube.ru/video/embed/6722881?vk_puid37=&vk_puid38=',
134 'id': 'a10e53b86e8f349080f718582ce4c661',
136 'timestamp': 1387830582,
137 'upload_date': '20131223',
138 'uploader_id': '297833',
139 'description': 'Видео группы ★http://vk.com/foxkidsreset★ музей Fox Kids и Jetix<br/><br/> восстановлено и сделано в шикоформате subziro89 http://vk.com/subziro89',
140 'uploader': 'subziro89 ILya',
141 'title': 'Мистический городок Эйри в Индиан 5 серия озвучка subziro89',
144 'skip_download': True,
147 'url': 'http://rutube.ru/play/embed/8083783',
148 'only_matching': True,
151 def _real_extract(self, url):
152 embed_id = self._match_id(url)
153 webpage = self._download_webpage(url, embed_id)
155 canonical_url = self._html_search_regex(
156 r'<link\s+rel="canonical"\s+href="([^"]+?)"', webpage,
158 return self.url_result(canonical_url, RutubeIE.ie_key())
161 class RutubePlaylistBaseIE(RutubeBaseIE):
162 def _next_page_url(self, page_num, playlist_id, *args, **kwargs):
163 return self._PAGE_TEMPLATE % (playlist_id, page_num)
165 def _entries(self, playlist_id, *args, **kwargs):
167 for pagenum in itertools.count(1):
168 page = self._download_json(
169 next_page_url or self._next_page_url(
170 pagenum, playlist_id, *args, **kwargs),
171 playlist_id, 'Downloading page %s' % pagenum)
173 results = page.get('results')
174 if not results or not isinstance(results, list):
177 for result in results:
178 video_url = result.get('video_url')
179 if not video_url or not isinstance(video_url, compat_str):
181 entry = self._extract_video(result, require_title=False)
185 'ie_key': RutubeIE.ie_key(),
189 next_page_url = page.get('next')
190 if not next_page_url or not page.get('has_next'):
193 def _extract_playlist(self, playlist_id, *args, **kwargs):
194 return self.playlist_result(
195 self._entries(playlist_id, *args, **kwargs),
196 playlist_id, kwargs.get('playlist_name'))
198 def _real_extract(self, url):
199 return self._extract_playlist(self._match_id(url))
202 class RutubeChannelIE(RutubePlaylistBaseIE):
203 IE_NAME = 'rutube:channel'
204 IE_DESC = 'Rutube channels'
205 _VALID_URL = r'https?://rutube\.ru/tags/video/(?P<id>\d+)'
207 'url': 'http://rutube.ru/tags/video/1800/',
211 'playlist_mincount': 68,
214 _PAGE_TEMPLATE = 'http://rutube.ru/api/tags/video/%s/?page=%s&format=json'
217 class RutubeMovieIE(RutubePlaylistBaseIE):
218 IE_NAME = 'rutube:movie'
219 IE_DESC = 'Rutube movies'
220 _VALID_URL = r'https?://rutube\.ru/metainfo/tv/(?P<id>\d+)'
223 _MOVIE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/?format=json'
224 _PAGE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/video?page=%s&format=json'
226 def _real_extract(self, url):
227 movie_id = self._match_id(url)
228 movie = self._download_json(
229 self._MOVIE_TEMPLATE % movie_id, movie_id,
230 'Downloading movie JSON')
231 return self._extract_playlist(
232 movie_id, playlist_name=movie.get('name'))
235 class RutubePersonIE(RutubePlaylistBaseIE):
236 IE_NAME = 'rutube:person'
237 IE_DESC = 'Rutube person videos'
238 _VALID_URL = r'https?://rutube\.ru/video/person/(?P<id>\d+)'
240 'url': 'http://rutube.ru/video/person/313878/',
244 'playlist_mincount': 37,
247 _PAGE_TEMPLATE = 'http://rutube.ru/api/video/person/%s/?page=%s&format=json'
250 class RutubePlaylistIE(RutubePlaylistBaseIE):
251 IE_NAME = 'rutube:playlist'
252 IE_DESC = 'Rutube playlists'
253 _VALID_URL = r'https?://rutube\.ru/(?:video|(?:play/)?embed)/[\da-z]{32}/\?.*?\bpl_id=(?P<id>\d+)'
255 'url': 'https://rutube.ru/video/cecd58ed7d531fc0f3d795d51cee9026/?pl_id=3097&pl_type=tag',
259 'playlist_count': 27,
261 'url': 'https://rutube.ru/video/10b3a03fc01d5bbcc632a2f3514e8aab/?pl_id=4252&pl_type=source',
262 'only_matching': True,
265 _PAGE_TEMPLATE = 'http://rutube.ru/api/playlist/%s/%s/?page=%s&format=json'
269 params = compat_parse_qs(compat_urllib_parse_urlparse(url).query)
270 return params.get('pl_type', [None])[0] and int_or_none(params.get('pl_id', [None])[0])
272 def _next_page_url(self, page_num, playlist_id, item_kind):
273 return self._PAGE_TEMPLATE % (item_kind, playlist_id, page_num)
275 def _real_extract(self, url):
276 qs = compat_parse_qs(compat_urllib_parse_urlparse(url).query)
277 playlist_kind = qs['pl_type'][0]
278 playlist_id = qs['pl_id'][0]
279 return self._extract_playlist(playlist_id, item_kind=playlist_kind)