2 from __future__ import unicode_literals
8 from .common import InfoExtractor
9 from ..compat import compat_urllib_parse_unquote
18 class MailRuIE(InfoExtractor):
20 IE_DESC = 'Видео@Mail.Ru'
23 (?:(?:www|m)\.)?my\.mail\.ru/+
25 video/.*\#video=/?(?P<idv1>(?:[^/]+/){3}\d+)|
26 (?:(?P<idv2prefix>(?:[^/]+/+){2})video/(?P<idv2suffix>[^/]+/\d+))\.html|
27 (?:video/embed|\+/video/meta)/(?P<metaid>\d+)
32 'url': 'http://my.mail.ru/video/top#video=/mail/sonypicturesrus/75/76',
33 'md5': 'dea205f03120046894db4ebb6159879a',
37 'title': 'Новый Человек-Паук. Высокое напряжение. Восстание Электро',
38 'timestamp': 1393235077,
39 'upload_date': '20140224',
40 'uploader': 'sonypicturesrus',
41 'uploader_id': 'sonypicturesrus@mail.ru',
44 'skip': 'Not accessible from Travis CI server',
47 'url': 'http://my.mail.ru/corp/hitech/video/news_hi-tech_mail_ru/1263.html',
48 'md5': '00a91a58c3402204dcced523777b475f',
50 'id': '46843144_1263',
52 'title': 'Samsung Galaxy S5 Hammer Smash Fail Battery Explosion',
53 'timestamp': 1397039888,
54 'upload_date': '20140409',
56 'uploader_id': 'hitech@corp.mail.ru',
59 'skip': 'Not accessible from Travis CI server',
62 # only available via metaUrl API
63 'url': 'http://my.mail.ru/mail/720pizle/video/_myvideo/502.html',
64 'md5': '3b26d2491c6949d031a32b96bd97c096',
69 'timestamp': 1449094163,
70 'upload_date': '20151202',
71 'uploader': '720pizle@mail.ru',
72 'uploader_id': '720pizle@mail.ru',
75 'skip': 'Not accessible from Travis CI server',
78 'url': 'http://m.my.mail.ru/mail/3sktvtr/video/_myvideo/138.html',
79 'only_matching': True,
82 'url': 'https://my.mail.ru/video/embed/7949340477499637815',
83 'only_matching': True,
86 'url': 'http://my.mail.ru/+/video/meta/7949340477499637815',
87 'only_matching': True,
90 'url': 'https://my.mail.ru//list/sinyutin10/video/_myvideo/4.html',
91 'only_matching': True,
94 'url': 'https://my.mail.ru//list//sinyutin10/video/_myvideo/4.html',
95 'only_matching': True,
99 def _real_extract(self, url):
100 mobj = re.match(self._VALID_URL, url)
101 meta_id = mobj.group('metaid')
105 meta_url = 'https://my.mail.ru/+/video/meta/%s' % meta_id
107 video_id = mobj.group('idv1')
109 video_id = mobj.group('idv2prefix') + mobj.group('idv2suffix')
110 webpage = self._download_webpage(url, video_id)
111 page_config = self._parse_json(self._search_regex(
112 r'(?s)<script[^>]+class="sp-video__page-config"[^>]*>(.+?)</script>',
113 webpage, 'page config', default='{}'), video_id, fatal=False)
115 meta_url = page_config.get('metaUrl') or page_config.get('video', {}).get('metaUrl')
121 video_data = self._download_json(
122 meta_url, video_id or meta_id, 'Downloading video meta JSON',
125 # Fallback old approach
127 video_data = self._download_json(
128 'http://api.video.mail.ru/videos/%s.json?new=1' % video_id,
129 video_id, 'Downloading video JSON')
132 for f in video_data['videos']:
133 video_url = f.get('url')
136 format_id = f.get('key')
137 height = int_or_none(self._search_regex(
138 r'^(\d+)[pP]$', format_id, 'height', default=None)) if format_id else None
141 'format_id': format_id,
144 self._sort_formats(formats)
146 meta_data = video_data['meta']
147 title = remove_end(meta_data['title'], '.mp4')
149 author = video_data.get('author')
150 uploader = author.get('name')
151 uploader_id = author.get('id') or author.get('email')
152 view_count = int_or_none(video_data.get('viewsCount') or video_data.get('views_count'))
154 acc_id = meta_data.get('accId')
155 item_id = meta_data.get('itemId')
156 content_id = '%s_%s' % (acc_id, item_id) if acc_id and item_id else video_id
158 thumbnail = meta_data.get('poster')
159 duration = int_or_none(meta_data.get('duration'))
160 timestamp = int_or_none(meta_data.get('timestamp'))
165 'thumbnail': thumbnail,
166 'timestamp': timestamp,
167 'uploader': uploader,
168 'uploader_id': uploader_id,
169 'duration': duration,
170 'view_count': view_count,
175 class MailRuMusicSearchBaseIE(InfoExtractor):
176 def _search(self, query, url, audio_id, limit=100, offset=0):
177 search = self._download_json(
178 'https://my.mail.ru/cgi-bin/my/ajax', audio_id,
179 'Downloading songs JSON page %d' % (offset // limit + 1),
182 'X-Requested-With': 'XMLHttpRequest',
186 'func_name': 'music.search',
191 'arg_search_params': json.dumps({
198 'arg_offset': offset,
200 return next(e for e in search if isinstance(e, dict))
203 def _extract_track(t, fatal=True):
204 audio_url = t['URL'] if fatal else t.get('URL')
208 audio_id = t['File'] if fatal else t.get('File')
212 thumbnail = t.get('AlbumCoverURL') or t.get('FiledAlbumCover')
213 uploader = t.get('OwnerName') or t.get('OwnerName_Text_HTML')
214 uploader_id = t.get('UploaderID')
215 duration = int_or_none(t.get('DurationInSeconds')) or parse_duration(
216 t.get('Duration') or t.get('DurationStr'))
217 view_count = int_or_none(t.get('PlayCount') or t.get('PlayCount_hr'))
219 track = t.get('Name') or t.get('Name_Text_HTML')
220 artist = t.get('Author') or t.get('Author_Text_HTML')
223 title = '%s - %s' % (artist, track) if artist else track
228 'extractor_key': MailRuMusicIE.ie_key(),
231 'thumbnail': thumbnail,
232 'uploader': uploader,
233 'uploader_id': uploader_id,
234 'duration': duration,
235 'view_count': view_count,
237 'abr': int_or_none(t.get('BitRate')),
240 'album': t.get('Album'),
245 class MailRuMusicIE(MailRuMusicSearchBaseIE):
246 IE_NAME = 'mailru:music'
247 IE_DESC = 'Музыка@Mail.Ru'
248 _VALID_URL = r'https?://my\.mail\.ru/+music/+songs/+[^/?#&]+-(?P<id>[\da-f]+)'
250 'url': 'https://my.mail.ru/music/songs/%D0%BC8%D0%BB8%D1%82%D1%85-l-a-h-luciferian-aesthetics-of-herrschaft-single-2017-4e31f7125d0dfaef505d947642366893',
251 'md5': '0f8c22ef8c5d665b13ac709e63025610',
253 'id': '4e31f7125d0dfaef505d947642366893',
255 'title': 'L.A.H. (Luciferian Aesthetics of Herrschaft) single, 2017 - М8Л8ТХ',
256 'uploader': 'Игорь Мудрый',
257 'uploader_id': '1459196328',
262 'track': 'L.A.H. (Luciferian Aesthetics of Herrschaft) single, 2017',
267 def _real_extract(self, url):
268 audio_id = self._match_id(url)
270 webpage = self._download_webpage(url, audio_id)
272 title = self._og_search_title(webpage)
273 music_data = self._search(title, url, audio_id)['MusicData']
274 t = next(t for t in music_data if t.get('File') == audio_id)
276 info = self._extract_track(t)
277 info['title'] = title
281 class MailRuMusicSearchIE(MailRuMusicSearchBaseIE):
282 IE_NAME = 'mailru:music:search'
283 IE_DESC = 'Музыка@Mail.Ru'
284 _VALID_URL = r'https?://my\.mail\.ru/+music/+search/+(?P<id>[^/?#&]+)'
286 'url': 'https://my.mail.ru/music/search/black%20shadow',
288 'id': 'black shadow',
290 'playlist_mincount': 532,
293 def _real_extract(self, url):
294 query = compat_urllib_parse_unquote(self._match_id(url))
301 for _ in itertools.count(1):
302 search = self._search(query, url, query, LIMIT, offset)
304 music_data = search.get('MusicData')
305 if not music_data or not isinstance(music_data, list):
309 track = self._extract_track(t, fatal=False)
311 entries.append(track)
314 search, lambda x: x['Results']['music']['Total'], int)
316 if total is not None:
322 return self.playlist_result(entries, query)