]>
Commit | Line | Data |
---|---|---|
a2043572 | 1 | import random |
8afff9f8 | 2 | import re |
19ada898 | 3 | import time |
a2043572 | 4 | |
5d98908b | 5 | from .common import InfoExtractor |
5edea45f | 6 | from ..utils import ( |
2fffb1dc | 7 | ExtractorError, |
e897bd82 | 8 | clean_html, |
25a6e769 YCH |
9 | strip_jsonp, |
10 | unescapeHTML, | |
5edea45f | 11 | ) |
5d98908b | 12 | |
5d98908b YCH |
13 | |
14 | class QQMusicIE(InfoExtractor): | |
7ec676bb | 15 | IE_NAME = 'qqmusic' |
a7ada46b | 16 | IE_DESC = 'QQ音乐' |
19ada898 | 17 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/song/(?P<id>[0-9A-Za-z]+)\.html' |
5d98908b | 18 | _TESTS = [{ |
19ada898 | 19 | 'url': 'https://y.qq.com/n/yqq/song/004295Et37taLD.html', |
25a6e769 | 20 | 'md5': '5f1e6cea39e182857da7ffc5ef5e6bb8', |
5d98908b YCH |
21 | 'info_dict': { |
22 | 'id': '004295Et37taLD', | |
55e5841f | 23 | 'ext': 'mp3', |
5d98908b | 24 | 'title': '可惜没如果', |
a5d09d68 | 25 | 'release_date': '20141227', |
5d98908b | 26 | 'creator': '林俊杰', |
25a6e769 | 27 | 'description': 'md5:d85afb3051952ecc50a1ee8a286d1eac', |
ec85ded8 | 28 | 'thumbnail': r're:^https?://.*\.jpg$', |
add96eb9 | 29 | }, |
5e3915cb | 30 | }, { |
31 | 'note': 'There is no mp3-320 version of this song.', | |
19ada898 | 32 | 'url': 'https://y.qq.com/n/yqq/song/004MsGEo3DdNxV.html', |
5e3915cb | 33 | 'md5': 'fa3926f0c585cda0af8fa4f796482e3e', |
34 | 'info_dict': { | |
35 | 'id': '004MsGEo3DdNxV', | |
36 | 'ext': 'mp3', | |
37 | 'title': '如果', | |
a5d09d68 | 38 | 'release_date': '20050626', |
5e3915cb | 39 | 'creator': '李季美', |
40 | 'description': 'md5:46857d5ed62bc4ba84607a805dccf437', | |
ec85ded8 | 41 | 'thumbnail': r're:^https?://.*\.jpg$', |
add96eb9 | 42 | }, |
cd1bb549 S |
43 | }, { |
44 | 'note': 'lyrics not in .lrc format', | |
19ada898 | 45 | 'url': 'https://y.qq.com/n/yqq/song/001JyApY11tIp6.html', |
cd1bb549 S |
46 | 'info_dict': { |
47 | 'id': '001JyApY11tIp6', | |
48 | 'ext': 'mp3', | |
49 | 'title': 'Shadows Over Transylvania', | |
a5d09d68 | 50 | 'release_date': '19970225', |
cd1bb549 | 51 | 'creator': 'Dark Funeral', |
25a6e769 | 52 | 'description': 'md5:c9b20210587cbcd6836a1c597bab4525', |
ec85ded8 | 53 | 'thumbnail': r're:^https?://.*\.jpg$', |
cd1bb549 S |
54 | }, |
55 | 'params': { | |
56 | 'skip_download': True, | |
57 | }, | |
5d98908b YCH |
58 | }] |
59 | ||
55e5841f | 60 | _FORMATS = { |
8b8cde21 | 61 | 'mp3-320': {'prefix': 'M800', 'ext': 'mp3', 'preference': 40, 'abr': 320}, |
62 | 'mp3-128': {'prefix': 'M500', 'ext': 'mp3', 'preference': 30, 'abr': 128}, | |
add96eb9 | 63 | 'm4a': {'prefix': 'C200', 'ext': 'm4a', 'preference': 10}, |
55e5841f | 64 | } |
65 | ||
a2043572 YCH |
66 | # Reference: m_r_GetRUin() in top_player.js |
67 | # http://imgcache.gtimg.cn/music/portal_v3/y/top_player.js | |
68 | @staticmethod | |
69 | def m_r_get_ruin(): | |
add96eb9 | 70 | cur_ms = int(time.time() * 1000) % 1000 |
71 | return int(round(random.random() * 2147483647) * cur_ms % 1E10) | |
a2043572 | 72 | |
5d98908b YCH |
73 | def _real_extract(self, url): |
74 | mid = self._match_id(url) | |
75 | ||
76 | detail_info_page = self._download_webpage( | |
add96eb9 | 77 | f'http://s.plcloud.music.qq.com/fcgi-bin/fcg_yqq_song_detail_info.fcg?songmid={mid}&play=0', |
8afff9f8 | 78 | mid, note='Download song detail info', |
c9a77969 | 79 | errnote='Unable to get song detail info', encoding='gbk') |
5d98908b YCH |
80 | |
81 | song_name = self._html_search_regex( | |
82 | r"songname:\s*'([^']+)'", detail_info_page, 'song name') | |
83 | ||
84 | publish_time = self._html_search_regex( | |
85 | r'发行时间:(\d{4}-\d{2}-\d{2})', detail_info_page, | |
a685ae51 YCH |
86 | 'publish time', default=None) |
87 | if publish_time: | |
88 | publish_time = publish_time.replace('-', '') | |
5d98908b YCH |
89 | |
90 | singer = self._html_search_regex( | |
a685ae51 YCH |
91 | r"singer:\s*'([^']+)", detail_info_page, 'singer', default=None) |
92 | ||
93 | lrc_content = self._html_search_regex( | |
94 | r'<div class="content" id="lrc_content"[^<>]*>([^<>]+)</div>', | |
95 | detail_info_page, 'LRC lyrics', default=None) | |
b813d8ca YCH |
96 | if lrc_content: |
97 | lrc_content = lrc_content.replace('\\n', '\n') | |
5d98908b | 98 | |
5e3915cb | 99 | thumbnail_url = None |
100 | albummid = self._search_regex( | |
0392ac98 | 101 | [r'albummid:\'([0-9a-zA-Z]+)\'', r'"albummid":"([0-9a-zA-Z]+)"'], |
102 | detail_info_page, 'album mid', default=None) | |
5e3915cb | 103 | if albummid: |
add96eb9 | 104 | thumbnail_url = f'http://i.gtimg.cn/music/photo/mid_album_500/{albummid[-2:-1]}/{albummid[-1]}/{albummid}.jpg' |
5e3915cb | 105 | |
a2043572 YCH |
106 | guid = self.m_r_get_ruin() |
107 | ||
5d98908b | 108 | vkey = self._download_json( |
add96eb9 | 109 | f'http://base.music.qq.com/fcgi-bin/fcg_musicexpress.fcg?json=3&guid={guid}', |
5d98908b YCH |
110 | mid, note='Retrieve vkey', errnote='Unable to get vkey', |
111 | transform_source=strip_jsonp)['key'] | |
55e5841f | 112 | |
113 | formats = [] | |
e8ac61e8 | 114 | for format_id, details in self._FORMATS.items(): |
55e5841f | 115 | formats.append({ |
add96eb9 | 116 | 'url': 'http://cc.stream.qqmusic.qq.com/{}{}.{}?vkey={}&guid={}&fromtag=0'.format( |
117 | details['prefix'], mid, details['ext'], vkey, guid), | |
e8ac61e8 YCH |
118 | 'format': format_id, |
119 | 'format_id': format_id, | |
f983b875 | 120 | 'quality': details['preference'], |
e8ac61e8 | 121 | 'abr': details.get('abr'), |
55e5841f | 122 | }) |
4d58b24c | 123 | self._check_formats(formats, mid) |
5d98908b | 124 | |
961c5cbf S |
125 | actual_lrc_lyrics = ''.join( |
126 | line + '\n' for line in re.findall( | |
95c5e101 | 127 | r'(?m)^(\[[0-9]{2}:[0-9]{2}(?:\.[0-9]{2,})?\][^\n]*|\[[^\]]*\])', lrc_content)) |
b65e5bb7 QF |
128 | |
129 | info_dict = { | |
5d98908b | 130 | 'id': mid, |
55e5841f | 131 | 'formats': formats, |
5d98908b | 132 | 'title': song_name, |
a5d09d68 | 133 | 'release_date': publish_time, |
5d98908b | 134 | 'creator': singer, |
a685ae51 | 135 | 'description': lrc_content, |
add96eb9 | 136 | 'thumbnail': thumbnail_url, |
5d98908b | 137 | } |
b65e5bb7 QF |
138 | if actual_lrc_lyrics: |
139 | info_dict['subtitles'] = { | |
140 | 'origin': [{ | |
141 | 'ext': 'lrc', | |
142 | 'data': actual_lrc_lyrics, | |
add96eb9 | 143 | }], |
b65e5bb7 QF |
144 | } |
145 | return info_dict | |
8afff9f8 YCH |
146 | |
147 | ||
5edea45f YCH |
148 | class QQPlaylistBaseIE(InfoExtractor): |
149 | @staticmethod | |
150 | def qq_static_url(category, mid): | |
add96eb9 | 151 | return f'http://y.qq.com/y/static/{category}/{mid[-2]}/{mid[-1]}/{mid}.html' |
5edea45f | 152 | |
5c037c0d | 153 | def get_singer_all_songs(self, singmid, num): |
154 | return self._download_webpage( | |
25a6e769 YCH |
155 | r'https://c.y.qq.com/v8/fcg-bin/fcg_v8_singer_track_cp.fcg', singmid, |
156 | query={ | |
157 | 'format': 'json', | |
158 | 'inCharset': 'utf8', | |
159 | 'outCharset': 'utf-8', | |
160 | 'platform': 'yqq', | |
161 | 'needNewCode': 0, | |
162 | 'singermid': singmid, | |
163 | 'order': 'listen', | |
164 | 'begin': 0, | |
165 | 'num': num, | |
166 | 'songstatus': 1, | |
167 | }) | |
5c037c0d | 168 | |
169 | def get_entries_from_page(self, singmid): | |
5edea45f YCH |
170 | entries = [] |
171 | ||
5c037c0d | 172 | default_num = 1 |
173 | json_text = self.get_singer_all_songs(singmid, default_num) | |
134d85a7 | 174 | json_obj_all_songs = self._parse_json(json_text, singmid) |
5c037c0d | 175 | |
134d85a7 | 176 | if json_obj_all_songs['code'] == 0: |
177 | total = json_obj_all_songs['data']['total'] | |
5c037c0d | 178 | json_text = self.get_singer_all_songs(singmid, total) |
134d85a7 | 179 | json_obj_all_songs = self._parse_json(json_text, singmid) |
5c037c0d | 180 | |
134d85a7 | 181 | for item in json_obj_all_songs['data']['list']: |
182 | if item['musicData'].get('songmid') is not None: | |
5c037c0d | 183 | songmid = item['musicData']['songmid'] |
25a6e769 | 184 | entries.append(self.url_result( |
add96eb9 | 185 | rf'https://y.qq.com/n/yqq/song/{songmid}.html', 'QQMusic', songmid)) |
5edea45f YCH |
186 | |
187 | return entries | |
188 | ||
189 | ||
190 | class QQMusicSingerIE(QQPlaylistBaseIE): | |
7ec676bb | 191 | IE_NAME = 'qqmusic:singer' |
181c4cca | 192 | IE_DESC = 'QQ音乐 - 歌手' |
19ada898 | 193 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/singer/(?P<id>[0-9A-Za-z]+)\.html' |
8afff9f8 | 194 | _TEST = { |
19ada898 | 195 | 'url': 'https://y.qq.com/n/yqq/singer/001BLpXF2DyJe2.html', |
8afff9f8 YCH |
196 | 'info_dict': { |
197 | 'id': '001BLpXF2DyJe2', | |
198 | 'title': '林俊杰', | |
141a273a | 199 | 'description': 'md5:870ec08f7d8547c29c93010899103751', |
8afff9f8 | 200 | }, |
25a6e769 | 201 | 'playlist_mincount': 12, |
8afff9f8 YCH |
202 | } |
203 | ||
204 | def _real_extract(self, url): | |
205 | mid = self._match_id(url) | |
206 | ||
5c037c0d | 207 | entries = self.get_entries_from_page(mid) |
208 | singer_page = self._download_webpage(url, mid, 'Download singer page') | |
25a6e769 YCH |
209 | singer_name = self._html_search_regex( |
210 | r"singername\s*:\s*'(.*?)'", singer_page, 'singer name', default=None) | |
8afff9f8 YCH |
211 | singer_desc = None |
212 | ||
5c037c0d | 213 | if mid: |
8afff9f8 | 214 | singer_desc_page = self._download_xml( |
25a6e769 YCH |
215 | 'http://s.plcloud.music.qq.com/fcgi-bin/fcg_get_singer_desc.fcg', mid, |
216 | 'Donwload singer description XML', | |
217 | query={'utf8': 1, 'outCharset': 'utf-8', 'format': 'xml', 'singermid': mid}, | |
218 | headers={'Referer': 'https://y.qq.com/n/yqq/singer/'}) | |
8afff9f8 YCH |
219 | |
220 | singer_desc = singer_desc_page.find('./data/info/desc').text | |
221 | ||
222 | return self.playlist_result(entries, mid, singer_name, singer_desc) | |
5edea45f YCH |
223 | |
224 | ||
225 | class QQMusicAlbumIE(QQPlaylistBaseIE): | |
7ec676bb | 226 | IE_NAME = 'qqmusic:album' |
181c4cca | 227 | IE_DESC = 'QQ音乐 - 专辑' |
19ada898 | 228 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/album/(?P<id>[0-9A-Za-z]+)\.html' |
5edea45f | 229 | |
29b809de | 230 | _TESTS = [{ |
19ada898 | 231 | 'url': 'https://y.qq.com/n/yqq/album/000gXCTb2AhRR1.html', |
5edea45f YCH |
232 | 'info_dict': { |
233 | 'id': '000gXCTb2AhRR1', | |
234 | 'title': '我们都是这样长大的', | |
fc7ae675 | 235 | 'description': 'md5:179c5dce203a5931970d306aa9607ea6', |
5edea45f YCH |
236 | }, |
237 | 'playlist_count': 4, | |
29b809de | 238 | }, { |
19ada898 | 239 | 'url': 'https://y.qq.com/n/yqq/album/002Y5a3b3AlCu3.html', |
29b809de | 240 | 'info_dict': { |
241 | 'id': '002Y5a3b3AlCu3', | |
242 | 'title': '그리고...', | |
fc7ae675 | 243 | 'description': 'md5:a48823755615508a95080e81b51ba729', |
29b809de | 244 | }, |
245 | 'playlist_count': 8, | |
246 | }] | |
5edea45f YCH |
247 | |
248 | def _real_extract(self, url): | |
249 | mid = self._match_id(url) | |
250 | ||
29b809de | 251 | album = self._download_json( |
add96eb9 | 252 | f'http://i.y.qq.com/v8/fcg-bin/fcg_v8_album_info_cp.fcg?albummid={mid}&format=json', |
29b809de | 253 | mid, 'Download album page')['data'] |
254 | ||
255 | entries = [ | |
256 | self.url_result( | |
add96eb9 | 257 | 'https://y.qq.com/n/yqq/song/' + song['songmid'] + '.html', 'QQMusic', song['songmid'], |
29b809de | 258 | ) for song in album['list'] |
259 | ] | |
dfc4eca2 | 260 | album_name = album.get('name') |
29b809de | 261 | album_detail = album.get('desc') |
fc7ae675 YCH |
262 | if album_detail is not None: |
263 | album_detail = album_detail.strip() | |
5edea45f YCH |
264 | |
265 | return self.playlist_result(entries, mid, album_name, album_detail) | |
41333b97 | 266 | |
267 | ||
268 | class QQMusicToplistIE(QQPlaylistBaseIE): | |
7ec676bb | 269 | IE_NAME = 'qqmusic:toplist' |
181c4cca | 270 | IE_DESC = 'QQ音乐 - 排行榜' |
5d1bd3b9 | 271 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/toplist/(?P<id>[0-9]+)\.html' |
54889739 | 272 | |
41333b97 | 273 | _TESTS = [{ |
19ada898 | 274 | 'url': 'https://y.qq.com/n/yqq/toplist/123.html', |
41333b97 | 275 | 'info_dict': { |
25a6e769 | 276 | 'id': '123', |
eedda32e | 277 | 'title': '美国iTunes榜', |
25a6e769 | 278 | 'description': 'md5:89db2335fdbb10678dee2d43fe9aba08', |
41333b97 | 279 | }, |
25a6e769 | 280 | 'playlist_count': 100, |
41333b97 | 281 | }, { |
19ada898 | 282 | 'url': 'https://y.qq.com/n/yqq/toplist/3.html', |
41333b97 | 283 | 'info_dict': { |
25a6e769 | 284 | 'id': '3', |
141a273a | 285 | 'title': '巅峰榜·欧美', |
25a6e769 | 286 | 'description': 'md5:5a600d42c01696b26b71f8c4d43407da', |
41333b97 | 287 | }, |
288 | 'playlist_count': 100, | |
fd4eefed | 289 | }, { |
19ada898 | 290 | 'url': 'https://y.qq.com/n/yqq/toplist/106.html', |
fd4eefed | 291 | 'info_dict': { |
25a6e769 | 292 | 'id': '106', |
eedda32e | 293 | 'title': '韩国Mnet榜', |
25a6e769 | 294 | 'description': 'md5:cb84b325215e1d21708c615cac82a6e7', |
fd4eefed | 295 | }, |
296 | 'playlist_count': 50, | |
41333b97 | 297 | }] |
298 | ||
41333b97 | 299 | def _real_extract(self, url): |
300 | list_id = self._match_id(url) | |
301 | ||
29ea5728 | 302 | toplist_json = self._download_json( |
25a6e769 YCH |
303 | 'http://i.y.qq.com/v8/fcg-bin/fcg_v8_toplist_cp.fcg', list_id, |
304 | note='Download toplist page', | |
305 | query={'type': 'toplist', 'topid': list_id, 'format': 'json'}) | |
41333b97 | 306 | |
25a6e769 YCH |
307 | entries = [self.url_result( |
308 | 'https://y.qq.com/n/yqq/song/' + song['data']['songmid'] + '.html', 'QQMusic', | |
309 | song['data']['songmid']) | |
310 | for song in toplist_json['songlist']] | |
41333b97 | 311 | |
9d4f213f YCH |
312 | topinfo = toplist_json.get('topinfo', {}) |
313 | list_name = topinfo.get('ListName') | |
314 | list_description = topinfo.get('info') | |
eedda32e | 315 | return self.playlist_result(entries, list_id, list_name, list_description) |
0d0d5d37 | 316 | |
317 | ||
318 | class QQMusicPlaylistIE(QQPlaylistBaseIE): | |
319 | IE_NAME = 'qqmusic:playlist' | |
181c4cca | 320 | IE_DESC = 'QQ音乐 - 歌单' |
19ada898 | 321 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/playlist/(?P<id>[0-9]+)\.html' |
0d0d5d37 | 322 | |
2fffb1dc | 323 | _TESTS = [{ |
19ada898 | 324 | 'url': 'http://y.qq.com/n/yqq/playlist/3462654915.html', |
0d0d5d37 | 325 | 'info_dict': { |
326 | 'id': '3462654915', | |
327 | 'title': '韩国5月新歌精选下旬', | |
328 | 'description': 'md5:d2c9d758a96b9888cf4fe82f603121d4', | |
329 | }, | |
330 | 'playlist_count': 40, | |
2fffb1dc YCH |
331 | 'skip': 'playlist gone', |
332 | }, { | |
19ada898 | 333 | 'url': 'https://y.qq.com/n/yqq/playlist/1374105607.html', |
2fffb1dc YCH |
334 | 'info_dict': { |
335 | 'id': '1374105607', | |
336 | 'title': '易入人心的华语民谣', | |
337 | 'description': '民谣的歌曲易于传唱、、歌词朗朗伤口、旋律简单温馨。属于那种才入耳孔。却上心头的感觉。没有太多的复杂情绪。简单而直接地表达乐者的情绪,就是这样的简单才易入人心。', | |
338 | }, | |
339 | 'playlist_count': 20, | |
340 | }] | |
0d0d5d37 | 341 | |
342 | def _real_extract(self, url): | |
343 | list_id = self._match_id(url) | |
344 | ||
345 | list_json = self._download_json( | |
25a6e769 YCH |
346 | 'http://i.y.qq.com/qzone-music/fcg-bin/fcg_ucc_getcdinfo_byids_cp.fcg', |
347 | list_id, 'Download list page', | |
348 | query={'type': 1, 'json': 1, 'utf8': 1, 'onlysong': 0, 'disstid': list_id}, | |
2fffb1dc YCH |
349 | transform_source=strip_jsonp) |
350 | if not len(list_json.get('cdlist', [])): | |
351 | if list_json.get('code'): | |
352 | raise ExtractorError( | |
353 | 'QQ Music said: error %d in fetching playlist info' % list_json['code'], | |
354 | expected=True) | |
355 | raise ExtractorError('Unable to get playlist info') | |
356 | ||
357 | cdlist = list_json['cdlist'][0] | |
25a6e769 YCH |
358 | entries = [self.url_result( |
359 | 'https://y.qq.com/n/yqq/song/' + song['songmid'] + '.html', 'QQMusic', song['songmid']) | |
360 | for song in cdlist['songlist']] | |
0d0d5d37 | 361 | |
2fffb1dc YCH |
362 | list_name = cdlist.get('dissname') |
363 | list_description = clean_html(unescapeHTML(cdlist.get('desc'))) | |
0d0d5d37 | 364 | return self.playlist_result(entries, list_id, list_name, list_description) |