2 from __future__ import unicode_literals
4 from hashlib import md5
5 from datetime import datetime
9 from .common import InfoExtractor
10 from ..compat import (
11 compat_urllib_request,
16 class NetEaseMusicBaseIE(InfoExtractor):
17 _FORMATS = ['bMusic', 'mMusic', 'hMusic']
18 _NETEASE_SALT = '3go8&$8*3*3h0k(2)2'
19 _API_BASE = 'http://music.163.com/api/'
22 def _encrypt(cls, dfsid):
23 salt_bytes = bytearray(str(cls._NETEASE_SALT))
24 string_bytes = bytearray(str(dfsid))
25 salt_len = len(salt_bytes)
26 for i in xrange(len(string_bytes)):
27 string_bytes[i] = string_bytes[i] ^ salt_bytes[i % salt_len]
29 m.update(string_bytes)
30 result = m.digest().encode('base64')[:-1]
31 return result.replace('/', '_').replace('+', '-')
34 def extract_formats(cls, info):
36 for song_format in cls._FORMATS:
37 details = info.get(song_format)
41 'url': 'http://m1.music.126.net/%s/%s.%s' %
42 (cls._encrypt(details['dfsId']), details['dfsId'],
43 details['extension']),
44 'ext': details['extension'],
45 'abr': details['bitrate'] / 1000,
46 'preference': details['bitrate'],
47 'format_id': song_format,
48 'filesize': details['size'],
54 def convert_milliseconds(cls, ms):
55 return int(round(ms/1000.0))
57 def query_api(self, endpoint, video_id, note):
58 req = compat_urllib_request.Request('%s%s' % (self._API_BASE, endpoint))
59 req.add_header('Referer', self._API_BASE)
60 return self._download_json(req, video_id, note)
63 class NetEaseMusicIE(NetEaseMusicBaseIE):
64 IE_NAME = 'netease:song'
65 _VALID_URL = r'https?://music\.163\.com/(#/)?song\?id=(?P<id>[0-9]+)'
67 'url': 'http://music.163.com/#/song?id=32102397',
68 'md5': 'f2e97280e6345c74ba9d5677dd5dcb45',
72 'title': 'Bad Blood (feat. Kendrick Lamar)',
73 'creator': 'Taylor Swift / Kendrick Lamar',
74 'upload_date': '20150517',
75 'timestamp': 1431878400,
76 'description': 'md5:a10a54589c2860300d02e1de821eb2ef',
79 'note': 'No lyrics translation.',
80 'url': 'http://music.163.com/#/song?id=29822014',
86 'upload_date': '20141225',
87 'timestamp': 1419523200,
88 'description': 'md5:a4d8d89f44656af206b7b2555c0bce6c',
92 'url': 'http://music.163.com/song?id=17241424',
97 'creator': 'Dustin O\'Halloran',
98 'upload_date': '20080211',
99 'timestamp': 1202745600,
103 def _process_lyrics(self, lyrics_info):
104 original = lyrics_info.get('lrc', {}).get('lyric')
105 translated = lyrics_info.get('tlyric', {}).get('lyric')
110 lyrics_expr = r'(\[[0-9]{2}:[0-9]{2}\.[0-9]{2,}\])([^\n]+)'
111 original_ts_texts = re.findall(lyrics_expr, original)
112 translation_ts_dict = {
113 time_stamp: text for time_stamp, text in re.findall(lyrics_expr, translated)
117 '%s%s / %s' % (time_stamp, text, translation_ts_dict.get(time_stamp, ''))
118 for time_stamp, text in original_ts_texts
122 def _real_extract(self, url):
123 song_id = self._match_id(url)
127 'ids': '[%s]' % song_id
129 info = self.query_api(
130 'song/detail?' + compat_urllib_parse.urlencode(params),
131 song_id, 'Downloading song info')['songs'][0]
133 formats = self.extract_formats(info)
134 self._sort_formats(formats)
136 lyrics_info = self.query_api(
137 'song/lyric?id=%s&lv=-1&tv=-1' % song_id,
138 song_id, 'Downloading lyrics data')
139 lyrics = self._process_lyrics(lyrics_info)
142 if info.get('alias'):
143 alt_title = '/'.join(info.get('alias'))
147 'title': info['name'],
148 'alt_title': alt_title,
149 'creator': ' / '.join([artist['name'] for artist in info.get('artists', [])]),
150 'timestamp': self.convert_milliseconds(info.get('album', {}).get('publishTime')),
151 'thumbnail': info.get('album', {}).get('picUrl'),
152 'duration': self.convert_milliseconds(info.get('duration', 0)),
153 'description': lyrics,
158 class NetEaseMusicAlbumIE(NetEaseMusicBaseIE):
159 IE_NAME = 'netease:album'
160 _VALID_URL = r'https?://music\.163\.com/(#/)?album\?id=(?P<id>[0-9]+)'
162 'url': 'http://music.163.com/#/album?id=220780',
167 'playlist_count': 23,
170 def _real_extract(self, url):
171 album_id = self._match_id(url)
173 info = self.query_api(
174 'album/%s?id=%s' % (album_id, album_id),
175 album_id, 'Downloading album data')['album']
178 desc = info.get('description')
180 self.url_result('http://music.163.com/#/song?id=%s' % song['id'],
181 'NetEaseMusic', song['id'])
182 for song in info['songs']
184 return self.playlist_result(entries, album_id, name, desc)
187 class NetEaseMusicSingerIE(NetEaseMusicBaseIE):
188 IE_NAME = 'netease:singer'
189 _VALID_URL = r'https?://music\.163\.com/(#/)?artist\?id=(?P<id>[0-9]+)'
191 'note': 'Singer has aliases.',
192 'url': 'http://music.163.com/#/artist?id=10559',
195 'title': '张惠妹 - aMEI;阿密特',
197 'playlist_count': 50,
199 'note': 'Singer has translated name.',
200 'url': 'http://music.163.com/#/artist?id=124098',
203 'title': '李昇基 - 이승기',
205 'playlist_count': 50,
208 def _real_extract(self, url):
209 singer_id = self._match_id(url)
211 info = self.query_api(
212 'artist/%s?id=%s' % (singer_id, singer_id),
213 singer_id, 'Downloading singer data')
215 name = info['artist']['name']
216 if info['artist']['trans']:
217 name = '%s - %s' % (name, info['artist']['trans'])
218 if info['artist']['alias']:
219 name = '%s - %s' % (name, ";".join(info['artist']['alias']))
222 self.url_result('http://music.163.com/#/song?id=%s' % song['id'],
223 'NetEaseMusic', song['id'])
224 for song in info['hotSongs']
226 return self.playlist_result(entries, singer_id, name)
229 class NetEaseMusicListIE(NetEaseMusicBaseIE):
230 IE_NAME = 'netease:playlist'
231 _VALID_URL = r'https?://music\.163\.com/(#/)?(playlist|discover/toplist)\?id=(?P<id>[0-9]+)'
233 'url': 'http://music.163.com/#/playlist?id=79177352',
236 'title': 'Billboard 2007 Top 100',
237 'description': 'md5:12fd0819cab2965b9583ace0f8b7b022'
239 'playlist_count': 99,
241 'note': 'Toplist/Charts sample',
242 'url': 'http://music.163.com/#/discover/toplist?id=3733003',
245 'title': 're:韩国Melon排行榜周榜 [0-9]{4}-[0-9]{2}-[0-9]{2}',
246 'description': 'md5:73ec782a612711cadc7872d9c1e134fc',
248 'playlist_count': 50,
251 def _real_extract(self, url):
252 list_id = self._match_id(url)
254 info = self.query_api(
255 'playlist/detail?id=%s&lv=-1&tv=-1' % list_id,
256 list_id, 'Downloading playlist data')['result']
259 desc = info.get('description')
261 if info.get('specialType') == 10: # is a chart/toplist
262 datestamp = datetime.fromtimestamp(
263 self.convert_milliseconds(info['updateTime'])).strftime('%Y-%m-%d')
264 name = '%s %s' % (name, datestamp)
267 self.url_result('http://music.163.com/#/song?id=%s' % song['id'],
268 'NetEaseMusic', song['id'])
269 for song in info['tracks']
271 return self.playlist_result(entries, list_id, name, desc)
274 class NetEaseMusicMvIE(NetEaseMusicBaseIE):
275 IE_NAME = 'netease:mv'
276 _VALID_URL = r'https?://music\.163\.com/(#/)?mv\?id=(?P<id>[0-9]+)'
278 'url': 'http://music.163.com/#/mv?id=415350',
282 'title': '이럴거면 그러지말지',
283 'description': '白雅言自作曲唱甜蜜爱情',
285 'upload_date': '20150520',
289 def _real_extract(self, url):
290 mv_id = self._match_id(url)
292 info = self.query_api(
293 'mv/detail?id=%s&type=mp4' % mv_id,
294 mv_id, 'Downloading mv info')['data']
297 {'url': mv_url, 'ext': 'mp4', 'format_id': '%sp' % brs, 'preference': int(brs)}
298 for brs, mv_url in info['brs'].items()
300 self._sort_formats(formats)
304 'title': info['name'],
305 'description': info.get('desc') or info.get('briefDesc'),
306 'creator': info['artistName'],
307 'upload_date': info['publishTime'].replace('-', ''),
309 'thumbnail': info.get('cover'),
310 'duration': self.convert_milliseconds(info.get('duration', 0)),
314 class NetEaseMusicProgramIE(NetEaseMusicBaseIE):
315 IE_NAME = 'netease:program'
316 _VALID_URL = r'https?://music\.163\.com/(#/?)program\?id=(?P<id>[0-9]+)'
318 'url': 'http://music.163.com/#/program?id=10109055',
322 'title': '不丹足球背后的故事',
323 'description': '喜马拉雅人的足球梦 ...',
325 'timestamp': 1434179342,
326 'upload_date': '20150613',
330 'note': 'This program has accompanying songs.',
331 'url': 'http://music.163.com/#/program?id=10141022',
334 'title': '25岁,你是自在如风的少年<27°C>',
335 'description': 'md5:8d594db46cc3e6509107ede70a4aaa3b',
339 'note': 'This program has accompanying songs.',
340 'url': 'http://music.163.com/#/program?id=10141022',
344 'title': '25岁,你是自在如风的少年<27°C>',
345 'description': 'md5:8d594db46cc3e6509107ede70a4aaa3b',
346 'timestamp': 1434450841,
347 'upload_date': '20150616',
354 def _real_extract(self, url):
355 program_id = self._match_id(url)
357 info = self.query_api(
358 'dj/program/detail?id=%s' % program_id,
359 program_id, 'Downloading program info')['program']
362 description = info['description']
364 if not info['songs'] or self._downloader.params.get('noplaylist'):
367 'Downloading just the main audio %s because of --no-playlist'
368 % info['mainSong']['id'])
370 formats = self.extract_formats(info['mainSong'])
371 self._sort_formats(formats)
376 'description': description,
377 'creator': info['dj']['brand'],
378 'timestamp': self.convert_milliseconds(info['createTime']),
379 'thumbnail': info['coverUrl'],
380 'duration': self.convert_milliseconds(info.get('duration', 0)),
385 'Downloading playlist %s - add --no-playlist to just download the main audio %s'
386 % (program_id, info['mainSong']['id']))
388 song_ids = [info['mainSong']['id']]
389 song_ids.extend([song['id'] for song in info['songs']])
391 self.url_result('http://music.163.com/#/song?id=%s' % song_id,
392 'NetEaseMusic', song_id)
393 for song_id in song_ids
395 return self.playlist_result(entries, program_id, name, description)
398 class NetEaseMusicDjRadioIE(NetEaseMusicBaseIE):
399 IE_NAME = 'netease:djradio'
400 _VALID_URL = r'https?://music\.163\.com/(#/)?djradio\?id=(?P<id>[0-9]+)'
402 'url': 'http://music.163.com/#/djradio?id=42',
406 'description': 'md5:766220985cbd16fdd552f64c578a6b15'
408 'playlist_mincount': 40,
412 def _real_extract(self, url):
413 dj_id = self._match_id(url)
418 for offset in itertools.count(start=0, step=self._PAGE_SIZE):
419 info = self.query_api(
420 'dj/program/byradio?asc=false&limit=%d&radioId=%s&offset=%d'
421 % (self._PAGE_SIZE, dj_id, offset),
422 dj_id, 'Downloading dj programs - %d' % offset)
426 'http://music.163.com/#/program?id=%s' % program['id'],
427 'NetEaseMusicProgram', program['id'])
428 for program in info['programs']
432 radio = info['programs'][0]['radio']
439 return self.playlist_result(entries, dj_id, name, desc)