2 from __future__ import unicode_literals
4 from hashlib import md5
5 from base64 import b64encode
6 from datetime import datetime
9 from .common import InfoExtractor
10 from ..compat import (
13 compat_itertools_count,
15 from ..utils import sanitized_Request
18 class NetEaseMusicBaseIE(InfoExtractor):
19 _FORMATS = ['bMusic', 'mMusic', 'hMusic']
20 _NETEASE_SALT = '3go8&$8*3*3h0k(2)2'
21 _API_BASE = 'http://music.163.com/api/'
24 def _encrypt(cls, dfsid):
25 salt_bytes = bytearray(cls._NETEASE_SALT.encode('utf-8'))
26 string_bytes = bytearray(compat_str(dfsid).encode('ascii'))
27 salt_len = len(salt_bytes)
28 for i in range(len(string_bytes)):
29 string_bytes[i] = string_bytes[i] ^ salt_bytes[i % salt_len]
31 m.update(bytes(string_bytes))
32 result = b64encode(m.digest()).decode('ascii')
33 return result.replace('/', '_').replace('+', '-')
35 def extract_formats(self, info):
37 for song_format in self._FORMATS:
38 details = info.get(song_format)
41 song_file_path = '/%s/%s.%s' % (
42 self._encrypt(details['dfsId']), details['dfsId'], details['extension'])
44 # 203.130.59.9, 124.40.233.182, 115.231.74.139, etc is a reverse proxy-like feature
45 # from NetEase's CDN provider that can be used if m5.music.126.net does not
46 # work, especially for users outside of Mainland China
47 # via: https://github.com/JixunMoe/unblock-163/issues/3#issuecomment-163115880
48 for host in ('http://m5.music.126.net', 'http://115.231.74.139/m1.music.126.net',
49 'http://124.40.233.182/m1.music.126.net', 'http://203.130.59.9/m1.music.126.net'):
50 song_url = host + song_file_path
51 if self._is_valid_url(song_url, info['id'], 'song'):
54 'ext': details.get('extension'),
55 'abr': details.get('bitrate', 0) / 1000,
56 'format_id': song_format,
57 'filesize': details.get('size'),
58 'asr': details.get('sr')
64 def convert_milliseconds(cls, ms):
65 return int(round(ms / 1000.0))
67 def query_api(self, endpoint, video_id, note):
68 req = sanitized_Request('%s%s' % (self._API_BASE, endpoint))
69 req.add_header('Referer', self._API_BASE)
70 return self._download_json(req, video_id, note)
73 class NetEaseMusicIE(NetEaseMusicBaseIE):
74 IE_NAME = 'netease:song'
76 _VALID_URL = r'https?://music\.163\.com/(#/)?song\?id=(?P<id>[0-9]+)'
78 'url': 'http://music.163.com/#/song?id=32102397',
79 'md5': 'f2e97280e6345c74ba9d5677dd5dcb45',
83 'title': 'Bad Blood (feat. Kendrick Lamar)',
84 'creator': 'Taylor Swift / Kendrick Lamar',
85 'upload_date': '20150517',
86 'timestamp': 1431878400,
87 'description': 'md5:a10a54589c2860300d02e1de821eb2ef',
90 'note': 'No lyrics translation.',
91 'url': 'http://music.163.com/#/song?id=29822014',
97 'upload_date': '20141225',
98 'timestamp': 1419523200,
99 'description': 'md5:a4d8d89f44656af206b7b2555c0bce6c',
102 'note': 'No lyrics.',
103 'url': 'http://music.163.com/song?id=17241424',
108 'creator': 'Dustin O\'Halloran',
109 'upload_date': '20080211',
110 'timestamp': 1202745600,
113 'note': 'Has translated name.',
114 'url': 'http://music.163.com/#/song?id=22735043',
118 'title': '소원을 말해봐 (Genie)',
120 'description': 'md5:79d99cc560e4ca97e0c4d86800ee4184',
121 'upload_date': '20100127',
122 'timestamp': 1264608000,
123 'alt_title': '说出愿望吧(Genie)',
127 def _process_lyrics(self, lyrics_info):
128 original = lyrics_info.get('lrc', {}).get('lyric')
129 translated = lyrics_info.get('tlyric', {}).get('lyric')
134 lyrics_expr = r'(\[[0-9]{2}:[0-9]{2}\.[0-9]{2,}\])([^\n]+)'
135 original_ts_texts = re.findall(lyrics_expr, original)
136 translation_ts_dict = dict(
137 (time_stamp, text) for time_stamp, text in re.findall(lyrics_expr, translated)
140 '%s%s / %s' % (time_stamp, text, translation_ts_dict.get(time_stamp, ''))
141 for time_stamp, text in original_ts_texts
145 def _real_extract(self, url):
146 song_id = self._match_id(url)
150 'ids': '[%s]' % song_id
152 info = self.query_api(
153 'song/detail?' + compat_urllib_parse.urlencode(params),
154 song_id, 'Downloading song info')['songs'][0]
156 formats = self.extract_formats(info)
157 self._sort_formats(formats)
159 lyrics_info = self.query_api(
160 'song/lyric?id=%s&lv=-1&tv=-1' % song_id,
161 song_id, 'Downloading lyrics data')
162 lyrics = self._process_lyrics(lyrics_info)
165 if info.get('transNames'):
166 alt_title = '/'.join(info.get('transNames'))
170 'title': info['name'],
171 'alt_title': alt_title,
172 'creator': ' / '.join([artist['name'] for artist in info.get('artists', [])]),
173 'timestamp': self.convert_milliseconds(info.get('album', {}).get('publishTime')),
174 'thumbnail': info.get('album', {}).get('picUrl'),
175 'duration': self.convert_milliseconds(info.get('duration', 0)),
176 'description': lyrics,
181 class NetEaseMusicAlbumIE(NetEaseMusicBaseIE):
182 IE_NAME = 'netease:album'
183 IE_DESC = '网易云音乐 - 专辑'
184 _VALID_URL = r'https?://music\.163\.com/(#/)?album\?id=(?P<id>[0-9]+)'
186 'url': 'http://music.163.com/#/album?id=220780',
191 'playlist_count': 23,
194 def _real_extract(self, url):
195 album_id = self._match_id(url)
197 info = self.query_api(
198 'album/%s?id=%s' % (album_id, album_id),
199 album_id, 'Downloading album data')['album']
202 desc = info.get('description')
204 self.url_result('http://music.163.com/#/song?id=%s' % song['id'],
205 'NetEaseMusic', song['id'])
206 for song in info['songs']
208 return self.playlist_result(entries, album_id, name, desc)
211 class NetEaseMusicSingerIE(NetEaseMusicBaseIE):
212 IE_NAME = 'netease:singer'
213 IE_DESC = '网易云音乐 - 歌手'
214 _VALID_URL = r'https?://music\.163\.com/(#/)?artist\?id=(?P<id>[0-9]+)'
216 'note': 'Singer has aliases.',
217 'url': 'http://music.163.com/#/artist?id=10559',
220 'title': '张惠妹 - aMEI;阿密特',
222 'playlist_count': 50,
224 'note': 'Singer has translated name.',
225 'url': 'http://music.163.com/#/artist?id=124098',
228 'title': '李昇基 - 이승기',
230 'playlist_count': 50,
233 def _real_extract(self, url):
234 singer_id = self._match_id(url)
236 info = self.query_api(
237 'artist/%s?id=%s' % (singer_id, singer_id),
238 singer_id, 'Downloading singer data')
240 name = info['artist']['name']
241 if info['artist']['trans']:
242 name = '%s - %s' % (name, info['artist']['trans'])
243 if info['artist']['alias']:
244 name = '%s - %s' % (name, ';'.join(info['artist']['alias']))
247 self.url_result('http://music.163.com/#/song?id=%s' % song['id'],
248 'NetEaseMusic', song['id'])
249 for song in info['hotSongs']
251 return self.playlist_result(entries, singer_id, name)
254 class NetEaseMusicListIE(NetEaseMusicBaseIE):
255 IE_NAME = 'netease:playlist'
256 IE_DESC = '网易云音乐 - 歌单'
257 _VALID_URL = r'https?://music\.163\.com/(#/)?(playlist|discover/toplist)\?id=(?P<id>[0-9]+)'
259 'url': 'http://music.163.com/#/playlist?id=79177352',
262 'title': 'Billboard 2007 Top 100',
263 'description': 'md5:12fd0819cab2965b9583ace0f8b7b022'
265 'playlist_count': 99,
267 'note': 'Toplist/Charts sample',
268 'url': 'http://music.163.com/#/discover/toplist?id=3733003',
271 'title': 're:韩国Melon排行榜周榜 [0-9]{4}-[0-9]{2}-[0-9]{2}',
272 'description': 'md5:73ec782a612711cadc7872d9c1e134fc',
274 'playlist_count': 50,
277 def _real_extract(self, url):
278 list_id = self._match_id(url)
280 info = self.query_api(
281 'playlist/detail?id=%s&lv=-1&tv=-1' % list_id,
282 list_id, 'Downloading playlist data')['result']
285 desc = info.get('description')
287 if info.get('specialType') == 10: # is a chart/toplist
288 datestamp = datetime.fromtimestamp(
289 self.convert_milliseconds(info['updateTime'])).strftime('%Y-%m-%d')
290 name = '%s %s' % (name, datestamp)
293 self.url_result('http://music.163.com/#/song?id=%s' % song['id'],
294 'NetEaseMusic', song['id'])
295 for song in info['tracks']
297 return self.playlist_result(entries, list_id, name, desc)
300 class NetEaseMusicMvIE(NetEaseMusicBaseIE):
301 IE_NAME = 'netease:mv'
302 IE_DESC = '网易云音乐 - MV'
303 _VALID_URL = r'https?://music\.163\.com/(#/)?mv\?id=(?P<id>[0-9]+)'
305 'url': 'http://music.163.com/#/mv?id=415350',
309 'title': '이럴거면 그러지말지',
310 'description': '白雅言自作曲唱甜蜜爱情',
312 'upload_date': '20150520',
316 def _real_extract(self, url):
317 mv_id = self._match_id(url)
319 info = self.query_api(
320 'mv/detail?id=%s&type=mp4' % mv_id,
321 mv_id, 'Downloading mv info')['data']
324 {'url': mv_url, 'ext': 'mp4', 'format_id': '%sp' % brs, 'height': int(brs)}
325 for brs, mv_url in info['brs'].items()
327 self._sort_formats(formats)
331 'title': info['name'],
332 'description': info.get('desc') or info.get('briefDesc'),
333 'creator': info['artistName'],
334 'upload_date': info['publishTime'].replace('-', ''),
336 'thumbnail': info.get('cover'),
337 'duration': self.convert_milliseconds(info.get('duration', 0)),
341 class NetEaseMusicProgramIE(NetEaseMusicBaseIE):
342 IE_NAME = 'netease:program'
343 IE_DESC = '网易云音乐 - 电台节目'
344 _VALID_URL = r'https?://music\.163\.com/(#/?)program\?id=(?P<id>[0-9]+)'
346 'url': 'http://music.163.com/#/program?id=10109055',
350 'title': '不丹足球背后的故事',
351 'description': '喜马拉雅人的足球梦 ...',
353 'timestamp': 1434179342,
354 'upload_date': '20150613',
358 'note': 'This program has accompanying songs.',
359 'url': 'http://music.163.com/#/program?id=10141022',
362 'title': '25岁,你是自在如风的少年<27°C>',
363 'description': 'md5:8d594db46cc3e6509107ede70a4aaa3b',
367 'note': 'This program has accompanying songs.',
368 'url': 'http://music.163.com/#/program?id=10141022',
372 'title': '25岁,你是自在如风的少年<27°C>',
373 'description': 'md5:8d594db46cc3e6509107ede70a4aaa3b',
374 'timestamp': 1434450841,
375 'upload_date': '20150616',
382 def _real_extract(self, url):
383 program_id = self._match_id(url)
385 info = self.query_api(
386 'dj/program/detail?id=%s' % program_id,
387 program_id, 'Downloading program info')['program']
390 description = info['description']
392 if not info['songs'] or self._downloader.params.get('noplaylist'):
395 'Downloading just the main audio %s because of --no-playlist'
396 % info['mainSong']['id'])
398 formats = self.extract_formats(info['mainSong'])
399 self._sort_formats(formats)
404 'description': description,
405 'creator': info['dj']['brand'],
406 'timestamp': self.convert_milliseconds(info['createTime']),
407 'thumbnail': info['coverUrl'],
408 'duration': self.convert_milliseconds(info.get('duration', 0)),
413 'Downloading playlist %s - add --no-playlist to just download the main audio %s'
414 % (program_id, info['mainSong']['id']))
416 song_ids = [info['mainSong']['id']]
417 song_ids.extend([song['id'] for song in info['songs']])
419 self.url_result('http://music.163.com/#/song?id=%s' % song_id,
420 'NetEaseMusic', song_id)
421 for song_id in song_ids
423 return self.playlist_result(entries, program_id, name, description)
426 class NetEaseMusicDjRadioIE(NetEaseMusicBaseIE):
427 IE_NAME = 'netease:djradio'
428 IE_DESC = '网易云音乐 - 电台'
429 _VALID_URL = r'https?://music\.163\.com/(#/)?djradio\?id=(?P<id>[0-9]+)'
431 'url': 'http://music.163.com/#/djradio?id=42',
435 'description': 'md5:766220985cbd16fdd552f64c578a6b15'
437 'playlist_mincount': 40,
441 def _real_extract(self, url):
442 dj_id = self._match_id(url)
447 for offset in compat_itertools_count(start=0, step=self._PAGE_SIZE):
448 info = self.query_api(
449 'dj/program/byradio?asc=false&limit=%d&radioId=%s&offset=%d'
450 % (self._PAGE_SIZE, dj_id, offset),
451 dj_id, 'Downloading dj programs - %d' % offset)
455 'http://music.163.com/#/program?id=%s' % program['id'],
456 'NetEaseMusicProgram', program['id'])
457 for program in info['programs']
461 radio = info['programs'][0]['radio']
468 return self.playlist_result(entries, dj_id, name, desc)