]>
Commit | Line | Data |
---|---|---|
5d98908b YCH |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
a2043572 | 4 | import random |
8afff9f8 | 5 | import re |
19ada898 | 6 | import time |
a2043572 | 7 | |
5d98908b | 8 | from .common import InfoExtractor |
5edea45f | 9 | from ..utils import ( |
5c2266df | 10 | sanitized_Request, |
5edea45f YCH |
11 | strip_jsonp, |
12 | unescapeHTML, | |
0d0d5d37 | 13 | clean_html, |
2fffb1dc | 14 | ExtractorError, |
5edea45f | 15 | ) |
5d98908b | 16 | |
5d98908b YCH |
17 | |
18 | class QQMusicIE(InfoExtractor): | |
7ec676bb | 19 | IE_NAME = 'qqmusic' |
a7ada46b | 20 | IE_DESC = 'QQ音乐' |
19ada898 | 21 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/song/(?P<id>[0-9A-Za-z]+)\.html' |
5d98908b | 22 | _TESTS = [{ |
19ada898 | 23 | 'url': 'https://y.qq.com/n/yqq/song/004295Et37taLD.html', |
55e5841f | 24 | 'md5': '9ce1c1c8445f561506d2e3cfb0255705', |
5d98908b YCH |
25 | 'info_dict': { |
26 | 'id': '004295Et37taLD', | |
55e5841f | 27 | 'ext': 'mp3', |
5d98908b | 28 | 'title': '可惜没如果', |
a5d09d68 | 29 | 'release_date': '20141227', |
5d98908b | 30 | 'creator': '林俊杰', |
b813d8ca | 31 | 'description': 'md5:d327722d0361576fde558f1ac68a7065', |
ec85ded8 | 32 | 'thumbnail': r're:^https?://.*\.jpg$', |
5e3915cb | 33 | } |
34 | }, { | |
35 | 'note': 'There is no mp3-320 version of this song.', | |
19ada898 | 36 | 'url': 'https://y.qq.com/n/yqq/song/004MsGEo3DdNxV.html', |
5e3915cb | 37 | 'md5': 'fa3926f0c585cda0af8fa4f796482e3e', |
38 | 'info_dict': { | |
39 | 'id': '004MsGEo3DdNxV', | |
40 | 'ext': 'mp3', | |
41 | 'title': '如果', | |
a5d09d68 | 42 | 'release_date': '20050626', |
5e3915cb | 43 | 'creator': '李季美', |
44 | 'description': 'md5:46857d5ed62bc4ba84607a805dccf437', | |
ec85ded8 | 45 | 'thumbnail': r're:^https?://.*\.jpg$', |
5d98908b | 46 | } |
cd1bb549 S |
47 | }, { |
48 | 'note': 'lyrics not in .lrc format', | |
19ada898 | 49 | 'url': 'https://y.qq.com/n/yqq/song/001JyApY11tIp6.html', |
cd1bb549 S |
50 | 'info_dict': { |
51 | 'id': '001JyApY11tIp6', | |
52 | 'ext': 'mp3', | |
53 | 'title': 'Shadows Over Transylvania', | |
a5d09d68 | 54 | 'release_date': '19970225', |
cd1bb549 S |
55 | 'creator': 'Dark Funeral', |
56 | 'description': 'md5:ed14d5bd7ecec19609108052c25b2c11', | |
ec85ded8 | 57 | 'thumbnail': r're:^https?://.*\.jpg$', |
cd1bb549 S |
58 | }, |
59 | 'params': { | |
60 | 'skip_download': True, | |
61 | }, | |
5d98908b YCH |
62 | }] |
63 | ||
55e5841f | 64 | _FORMATS = { |
8b8cde21 | 65 | 'mp3-320': {'prefix': 'M800', 'ext': 'mp3', 'preference': 40, 'abr': 320}, |
66 | 'mp3-128': {'prefix': 'M500', 'ext': 'mp3', 'preference': 30, 'abr': 128}, | |
55e5841f | 67 | 'm4a': {'prefix': 'C200', 'ext': 'm4a', 'preference': 10} |
68 | } | |
69 | ||
a2043572 YCH |
70 | # Reference: m_r_GetRUin() in top_player.js |
71 | # http://imgcache.gtimg.cn/music/portal_v3/y/top_player.js | |
72 | @staticmethod | |
73 | def m_r_get_ruin(): | |
74 | curMs = int(time.time() * 1000) % 1000 | |
75 | return int(round(random.random() * 2147483647) * curMs % 1E10) | |
76 | ||
5d98908b YCH |
77 | def _real_extract(self, url): |
78 | mid = self._match_id(url) | |
79 | ||
80 | detail_info_page = self._download_webpage( | |
81 | 'http://s.plcloud.music.qq.com/fcgi-bin/fcg_yqq_song_detail_info.fcg?songmid=%s&play=0' % mid, | |
8afff9f8 | 82 | mid, note='Download song detail info', |
c9a77969 | 83 | errnote='Unable to get song detail info', encoding='gbk') |
5d98908b YCH |
84 | |
85 | song_name = self._html_search_regex( | |
86 | r"songname:\s*'([^']+)'", detail_info_page, 'song name') | |
87 | ||
88 | publish_time = self._html_search_regex( | |
89 | r'发行时间:(\d{4}-\d{2}-\d{2})', detail_info_page, | |
a685ae51 YCH |
90 | 'publish time', default=None) |
91 | if publish_time: | |
92 | publish_time = publish_time.replace('-', '') | |
5d98908b YCH |
93 | |
94 | singer = self._html_search_regex( | |
a685ae51 YCH |
95 | r"singer:\s*'([^']+)", detail_info_page, 'singer', default=None) |
96 | ||
97 | lrc_content = self._html_search_regex( | |
98 | r'<div class="content" id="lrc_content"[^<>]*>([^<>]+)</div>', | |
99 | detail_info_page, 'LRC lyrics', default=None) | |
b813d8ca YCH |
100 | if lrc_content: |
101 | lrc_content = lrc_content.replace('\\n', '\n') | |
5d98908b | 102 | |
5e3915cb | 103 | thumbnail_url = None |
104 | albummid = self._search_regex( | |
0392ac98 | 105 | [r'albummid:\'([0-9a-zA-Z]+)\'', r'"albummid":"([0-9a-zA-Z]+)"'], |
106 | detail_info_page, 'album mid', default=None) | |
5e3915cb | 107 | if albummid: |
108 | thumbnail_url = "http://i.gtimg.cn/music/photo/mid_album_500/%s/%s/%s.jpg" \ | |
109 | % (albummid[-2:-1], albummid[-1], albummid) | |
110 | ||
a2043572 YCH |
111 | guid = self.m_r_get_ruin() |
112 | ||
5d98908b YCH |
113 | vkey = self._download_json( |
114 | 'http://base.music.qq.com/fcgi-bin/fcg_musicexpress.fcg?json=3&guid=%s' % guid, | |
115 | mid, note='Retrieve vkey', errnote='Unable to get vkey', | |
116 | transform_source=strip_jsonp)['key'] | |
55e5841f | 117 | |
118 | formats = [] | |
e8ac61e8 | 119 | for format_id, details in self._FORMATS.items(): |
55e5841f | 120 | formats.append({ |
121 | 'url': 'http://cc.stream.qqmusic.qq.com/%s%s.%s?vkey=%s&guid=%s&fromtag=0' | |
e8ac61e8 YCH |
122 | % (details['prefix'], mid, details['ext'], vkey, guid), |
123 | 'format': format_id, | |
124 | 'format_id': format_id, | |
125 | 'preference': details['preference'], | |
126 | 'abr': details.get('abr'), | |
55e5841f | 127 | }) |
4d58b24c | 128 | self._check_formats(formats, mid) |
55e5841f | 129 | self._sort_formats(formats) |
5d98908b | 130 | |
961c5cbf S |
131 | actual_lrc_lyrics = ''.join( |
132 | line + '\n' for line in re.findall( | |
95c5e101 | 133 | r'(?m)^(\[[0-9]{2}:[0-9]{2}(?:\.[0-9]{2,})?\][^\n]*|\[[^\]]*\])', lrc_content)) |
b65e5bb7 QF |
134 | |
135 | info_dict = { | |
5d98908b | 136 | 'id': mid, |
55e5841f | 137 | 'formats': formats, |
5d98908b | 138 | 'title': song_name, |
a5d09d68 | 139 | 'release_date': publish_time, |
5d98908b | 140 | 'creator': singer, |
a685ae51 | 141 | 'description': lrc_content, |
b65e5bb7 | 142 | 'thumbnail': thumbnail_url |
5d98908b | 143 | } |
b65e5bb7 QF |
144 | if actual_lrc_lyrics: |
145 | info_dict['subtitles'] = { | |
146 | 'origin': [{ | |
147 | 'ext': 'lrc', | |
148 | 'data': actual_lrc_lyrics, | |
149 | }] | |
150 | } | |
151 | return info_dict | |
8afff9f8 YCH |
152 | |
153 | ||
5edea45f YCH |
154 | class QQPlaylistBaseIE(InfoExtractor): |
155 | @staticmethod | |
156 | def qq_static_url(category, mid): | |
157 | return 'http://y.qq.com/y/static/%s/%s/%s/%s.html' % (category, mid[-2], mid[-1], mid) | |
158 | ||
5edea45f YCH |
159 | @classmethod |
160 | def get_entries_from_page(cls, page): | |
161 | entries = [] | |
162 | ||
163 | for item in re.findall(r'class="data"[^<>]*>([^<>]+)</', page): | |
164 | song_mid = unescapeHTML(item).split('|')[-5] | |
165 | entries.append(cls.url_result( | |
19ada898 | 166 | # https://y.qq.com/n/yqq/song/004Dbsoo1yCbNZ.html |
167 | 'https://y.qq.com/n/yqq/song/' + song_mid + ".html", 'QQMusic', | |
a685ae51 | 168 | song_mid)) |
5edea45f YCH |
169 | |
170 | return entries | |
171 | ||
172 | ||
173 | class QQMusicSingerIE(QQPlaylistBaseIE): | |
7ec676bb | 174 | IE_NAME = 'qqmusic:singer' |
181c4cca | 175 | IE_DESC = 'QQ音乐 - 歌手' |
19ada898 | 176 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/singer/(?P<id>[0-9A-Za-z]+)\.html' |
8afff9f8 | 177 | _TEST = { |
19ada898 | 178 | 'url': 'https://y.qq.com/n/yqq/singer/001BLpXF2DyJe2.html', |
8afff9f8 YCH |
179 | 'info_dict': { |
180 | 'id': '001BLpXF2DyJe2', | |
181 | 'title': '林俊杰', | |
141a273a | 182 | 'description': 'md5:870ec08f7d8547c29c93010899103751', |
8afff9f8 YCH |
183 | }, |
184 | 'playlist_count': 12, | |
185 | } | |
186 | ||
187 | def _real_extract(self, url): | |
188 | mid = self._match_id(url) | |
189 | ||
190 | singer_page = self._download_webpage( | |
5edea45f | 191 | self.qq_static_url('singer', mid), mid, 'Download singer page') |
8afff9f8 | 192 | |
5edea45f | 193 | entries = self.get_entries_from_page(singer_page) |
8afff9f8 YCH |
194 | |
195 | singer_name = self._html_search_regex( | |
196 | r"singername\s*:\s*'([^']+)'", singer_page, 'singer name', | |
197 | default=None) | |
198 | ||
199 | singer_id = self._html_search_regex( | |
200 | r"singerid\s*:\s*'([0-9]+)'", singer_page, 'singer id', | |
201 | default=None) | |
202 | ||
203 | singer_desc = None | |
204 | ||
205 | if singer_id: | |
5c2266df | 206 | req = sanitized_Request( |
8afff9f8 YCH |
207 | 'http://s.plcloud.music.qq.com/fcgi-bin/fcg_get_singer_desc.fcg?utf8=1&outCharset=utf-8&format=xml&singerid=%s' % singer_id) |
208 | req.add_header( | |
209 | 'Referer', 'http://s.plcloud.music.qq.com/xhr_proxy_utf8.html') | |
210 | singer_desc_page = self._download_xml( | |
5edea45f | 211 | req, mid, 'Donwload singer description XML') |
8afff9f8 YCH |
212 | |
213 | singer_desc = singer_desc_page.find('./data/info/desc').text | |
214 | ||
215 | return self.playlist_result(entries, mid, singer_name, singer_desc) | |
5edea45f YCH |
216 | |
217 | ||
218 | class QQMusicAlbumIE(QQPlaylistBaseIE): | |
7ec676bb | 219 | IE_NAME = 'qqmusic:album' |
181c4cca | 220 | IE_DESC = 'QQ音乐 - 专辑' |
19ada898 | 221 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/album/(?P<id>[0-9A-Za-z]+)\.html' |
5edea45f | 222 | |
29b809de | 223 | _TESTS = [{ |
19ada898 | 224 | 'url': 'https://y.qq.com/n/yqq/album/000gXCTb2AhRR1.html', |
5edea45f YCH |
225 | 'info_dict': { |
226 | 'id': '000gXCTb2AhRR1', | |
227 | 'title': '我们都是这样长大的', | |
fc7ae675 | 228 | 'description': 'md5:179c5dce203a5931970d306aa9607ea6', |
5edea45f YCH |
229 | }, |
230 | 'playlist_count': 4, | |
29b809de | 231 | }, { |
19ada898 | 232 | 'url': 'https://y.qq.com/n/yqq/album/002Y5a3b3AlCu3.html', |
29b809de | 233 | 'info_dict': { |
234 | 'id': '002Y5a3b3AlCu3', | |
235 | 'title': '그리고...', | |
fc7ae675 | 236 | 'description': 'md5:a48823755615508a95080e81b51ba729', |
29b809de | 237 | }, |
238 | 'playlist_count': 8, | |
239 | }] | |
5edea45f YCH |
240 | |
241 | def _real_extract(self, url): | |
242 | mid = self._match_id(url) | |
243 | ||
29b809de | 244 | album = self._download_json( |
245 | 'http://i.y.qq.com/v8/fcg-bin/fcg_v8_album_info_cp.fcg?albummid=%s&format=json' % mid, | |
246 | mid, 'Download album page')['data'] | |
247 | ||
248 | entries = [ | |
249 | self.url_result( | |
19ada898 | 250 | 'https://y.qq.com/n/yqq/song/' + song['songmid'] + ".html", 'QQMusic', song['songmid'] |
29b809de | 251 | ) for song in album['list'] |
252 | ] | |
dfc4eca2 | 253 | album_name = album.get('name') |
29b809de | 254 | album_detail = album.get('desc') |
fc7ae675 YCH |
255 | if album_detail is not None: |
256 | album_detail = album_detail.strip() | |
5edea45f YCH |
257 | |
258 | return self.playlist_result(entries, mid, album_name, album_detail) | |
41333b97 | 259 | |
260 | ||
261 | class QQMusicToplistIE(QQPlaylistBaseIE): | |
7ec676bb | 262 | IE_NAME = 'qqmusic:toplist' |
181c4cca | 263 | IE_DESC = 'QQ音乐 - 排行榜' |
19ada898 | 264 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/toplist/(?P<id>(top|global)_[0-9]+)\.html' |
54889739 | 265 | |
41333b97 | 266 | _TESTS = [{ |
19ada898 | 267 | 'url': 'https://y.qq.com/n/yqq/toplist/123.html', |
41333b97 | 268 | 'info_dict': { |
eedda32e | 269 | 'id': 'global_123', |
270 | 'title': '美国iTunes榜', | |
41333b97 | 271 | }, |
272 | 'playlist_count': 10, | |
273 | }, { | |
19ada898 | 274 | 'url': 'https://y.qq.com/n/yqq/toplist/3.html', |
41333b97 | 275 | 'info_dict': { |
eedda32e | 276 | 'id': 'top_3', |
141a273a | 277 | 'title': '巅峰榜·欧美', |
eedda32e | 278 | 'description': 'QQ音乐巅峰榜·欧美根据用户收听行为自动生成,集结当下最流行的欧美新歌!:更新时间:每周四22点|统' |
279 | '计周期:一周(上周四至本周三)|统计对象:三个月内发行的欧美歌曲|统计数量:100首|统计算法:根据' | |
280 | '歌曲在一周内的有效播放次数,由高到低取前100名(同一歌手最多允许5首歌曲同时上榜)|有效播放次数:' | |
281 | '登录用户完整播放一首歌曲,记为一次有效播放;同一用户收听同一首歌曲,每天记录为1次有效播放' | |
41333b97 | 282 | }, |
283 | 'playlist_count': 100, | |
fd4eefed | 284 | }, { |
19ada898 | 285 | 'url': 'https://y.qq.com/n/yqq/toplist/106.html', |
fd4eefed | 286 | 'info_dict': { |
eedda32e | 287 | 'id': 'global_106', |
288 | 'title': '韩国Mnet榜', | |
fd4eefed | 289 | }, |
290 | 'playlist_count': 50, | |
41333b97 | 291 | }] |
292 | ||
41333b97 | 293 | def _real_extract(self, url): |
294 | list_id = self._match_id(url) | |
295 | ||
29ea5728 | 296 | list_type, num_id = list_id.split("_") |
41333b97 | 297 | |
29ea5728 | 298 | toplist_json = self._download_json( |
eedda32e | 299 | 'http://i.y.qq.com/v8/fcg-bin/fcg_v8_toplist_cp.fcg?type=%s&topid=%s&format=json' |
300 | % (list_type, num_id), | |
301 | list_id, 'Download toplist page') | |
41333b97 | 302 | |
eedda32e | 303 | entries = [ |
304 | self.url_result( | |
19ada898 | 305 | 'https://y.qq.com/n/yqq/song/' + song['data']['songmid'] + ".html", 'QQMusic', song['data']['songmid'] |
eedda32e | 306 | ) for song in toplist_json['songlist'] |
307 | ] | |
41333b97 | 308 | |
9d4f213f YCH |
309 | topinfo = toplist_json.get('topinfo', {}) |
310 | list_name = topinfo.get('ListName') | |
311 | list_description = topinfo.get('info') | |
eedda32e | 312 | return self.playlist_result(entries, list_id, list_name, list_description) |
0d0d5d37 | 313 | |
314 | ||
315 | class QQMusicPlaylistIE(QQPlaylistBaseIE): | |
316 | IE_NAME = 'qqmusic:playlist' | |
181c4cca | 317 | IE_DESC = 'QQ音乐 - 歌单' |
19ada898 | 318 | _VALID_URL = r'https?://y\.qq\.com/n/yqq/playlist/(?P<id>[0-9]+)\.html' |
0d0d5d37 | 319 | |
2fffb1dc | 320 | _TESTS = [{ |
19ada898 | 321 | 'url': 'http://y.qq.com/n/yqq/playlist/3462654915.html', |
0d0d5d37 | 322 | 'info_dict': { |
323 | 'id': '3462654915', | |
324 | 'title': '韩国5月新歌精选下旬', | |
325 | 'description': 'md5:d2c9d758a96b9888cf4fe82f603121d4', | |
326 | }, | |
327 | 'playlist_count': 40, | |
2fffb1dc YCH |
328 | 'skip': 'playlist gone', |
329 | }, { | |
19ada898 | 330 | 'url': 'https://y.qq.com/n/yqq/playlist/1374105607.html', |
2fffb1dc YCH |
331 | 'info_dict': { |
332 | 'id': '1374105607', | |
333 | 'title': '易入人心的华语民谣', | |
334 | '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。', | |
335 | }, | |
336 | 'playlist_count': 20, | |
337 | }] | |
0d0d5d37 | 338 | |
339 | def _real_extract(self, url): | |
340 | list_id = self._match_id(url) | |
341 | ||
342 | list_json = self._download_json( | |
343 | 'http://i.y.qq.com/qzone-music/fcg-bin/fcg_ucc_getcdinfo_byids_cp.fcg?type=1&json=1&utf8=1&onlysong=0&disstid=%s' | |
344 | % list_id, list_id, 'Download list page', | |
2fffb1dc YCH |
345 | transform_source=strip_jsonp) |
346 | if not len(list_json.get('cdlist', [])): | |
347 | if list_json.get('code'): | |
348 | raise ExtractorError( | |
349 | 'QQ Music said: error %d in fetching playlist info' % list_json['code'], | |
350 | expected=True) | |
351 | raise ExtractorError('Unable to get playlist info') | |
352 | ||
353 | cdlist = list_json['cdlist'][0] | |
0d0d5d37 | 354 | entries = [ |
355 | self.url_result( | |
19ada898 | 356 | 'https://y.qq.com/n/yqq/song/' + song['songmid'] + ".html", 'QQMusic', song['songmid'] |
2fffb1dc | 357 | ) for song in cdlist['songlist'] |
0d0d5d37 | 358 | ] |
359 | ||
2fffb1dc YCH |
360 | list_name = cdlist.get('dissname') |
361 | list_description = clean_html(unescapeHTML(cdlist.get('desc'))) | |
0d0d5d37 | 362 | return self.playlist_result(entries, list_id, list_name, list_description) |