2 from __future__ import unicode_literals
7 from .common import InfoExtractor
16 class KuwoBaseIE(InfoExtractor):
18 {'format': 'ape', 'ext': 'ape', 'preference': 100},
19 {'format': 'mp3-320', 'ext': 'mp3', 'br': '320kmp3', 'abr': 320, 'preference': 80},
20 {'format': 'mp3-192', 'ext': 'mp3', 'br': '192kmp3', 'abr': 192, 'preference': 70},
21 {'format': 'mp3-128', 'ext': 'mp3', 'br': '128kmp3', 'abr': 128, 'preference': 60},
22 {'format': 'wma', 'ext': 'wma', 'preference': 20},
23 {'format': 'aac', 'ext': 'aac', 'abr': 48, 'preference': 10}
26 def _get_formats(self, song_id):
28 for file_format in self._FORMATS:
29 song_url = self._download_webpage(
30 'http://antiserver.kuwo.cn/anti.s?format=%s&br=%s&rid=MUSIC_%s&type=convert_url&response=url' %
31 (file_format['ext'], file_format.get('br', ''), song_id),
32 song_id, note='Download %s url info' % file_format['format'],
34 if song_url.startswith('http://') or song_url.startswith('https://'):
37 'format_id': file_format['format'],
38 'format': file_format['format'],
39 'preference': file_format['preference'],
40 'abr': file_format.get('abr'),
42 self._sort_formats(formats)
46 class KuwoIE(KuwoBaseIE):
48 _VALID_URL = r'http://www\.kuwo\.cn/yinyue/(?P<id>[0-9]+?)/'
50 'url': 'http://www.kuwo.cn/yinyue/635632/',
56 'upload_date': '20080122',
57 'description': 'md5:ed13f58e3c3bf3f7fd9fbc4e5a7aa75c'
60 'url': 'http://www.kuwo.cn/yinyue/6446136/',
66 'upload_date': '20150518',
73 def _real_extract(self, url):
74 song_id = self._match_id(url)
75 webpage = self._download_webpage(
76 url, song_id, note='Download song detail info',
77 errnote='Unable to get song detail info')
79 song_name = self._html_search_regex(
80 r'<h1[^>]+title="([^"]+)">', webpage, 'song name')
81 singer_name = self._html_search_regex(
82 r'<div[^>]+class="s_img">\s*<a[^>]+title="([^>]+)"',
83 webpage, 'singer name', default=None)
84 lrc_content = clean_html(get_element_by_id('lrcContent', webpage))
85 if lrc_content == '暂无': # indicates no lyrics
88 formats = self._get_formats(song_id)
90 album_id = self._html_search_regex(
91 r'<p[^>]+class="album"[^<]+<a[^>]+href="http://www\.kuwo\.cn/album/(\d+)/"',
92 webpage, 'album id', default=None, fatal=False)
95 if album_id is not None:
96 album_info_page = self._download_webpage(
97 'http://www.kuwo.cn/album/%s/' % album_id, song_id,
98 note='Download album detail info',
99 errnote='Unable to get album detail info')
101 publish_time = self._html_search_regex(
102 r'发行时间:(\d{4}-\d{2}-\d{2})', album_info_page,
103 'publish time', default=None)
105 publish_time = publish_time.replace('-', '')
110 'creator': singer_name,
111 'upload_date': publish_time,
112 'description': lrc_content,
117 class KuwoAlbumIE(InfoExtractor):
118 IE_NAME = 'kuwo:album'
119 _VALID_URL = r'http://www\.kuwo\.cn/album/(?P<id>[0-9]+?)/'
121 'url': 'http://www.kuwo.cn/album/502294/',
125 'description': 'md5:6a7235a84cc6400ec3b38a7bdaf1d60c',
130 def _real_extract(self, url):
131 album_id = self._match_id(url)
133 webpage = self._download_webpage(
134 url, album_id, note='Download album info',
135 errnote='Unable to get album info')
137 album_name = self._html_search_regex(
138 r'<div[^>]+class="comm"[^<]+<h1[^>]+title="([^"]+)"', webpage,
140 album_intro = remove_start(
141 clean_html(get_element_by_id('intro', webpage)),
142 '%s简介:' % album_name)
145 self.url_result('http://www.kuwo.cn/yinyue/%s/' % song_id, 'Kuwo', song_id)
146 for song_id in re.findall(
147 r'<p[^>]+class="listen"><a[^>]+href="http://www\.kuwo\.cn/yinyue/(\d+)/"',
150 return self.playlist_result(entries, album_id, album_name, album_intro)
153 class KuwoChartIE(InfoExtractor):
154 IE_NAME = 'kuwo:chart'
155 _VALID_URL = r'http://yinyue\.kuwo\.cn/billboard_(?P<id>[^.]+).htm'
157 'url': 'http://yinyue.kuwo.cn/billboard_香港中文龙虎榜.htm',
161 'description': 're:[0-9]{4}第[0-9]{2}期',
163 'playlist_mincount': 10,
166 def _real_extract(self, url):
167 chart_id = self._match_id(url)
168 webpage = self._download_webpage(
169 url, chart_id, note='Download chart info',
170 errnote='Unable to get chart info')
172 chart_name = self._html_search_regex(
173 r'<h1[^>]+class="unDis">([^<]+)</h1>', webpage, 'chart name')
175 chart_desc = self._html_search_regex(
176 r'<p[^>]+class="tabDef">(\d{4}第\d{2}期)</p>', webpage, 'chart desc')
179 self.url_result('http://www.kuwo.cn/yinyue/%s/' % song_id, 'Kuwo', song_id)
180 for song_id in re.findall(
181 r'<a[^>]+href="http://www\.kuwo\.cn/yinyue/(\d+)/"', webpage)
183 return self.playlist_result(entries, chart_id, chart_name, chart_desc)
186 class KuwoSingerIE(InfoExtractor):
187 IE_NAME = 'kuwo:singer'
188 _VALID_URL = r'http://www\.kuwo\.cn/mingxing/(?P<id>[^/]+)'
190 'url': 'http://www.kuwo.cn/mingxing/bruno+mars/',
193 'title': 'Bruno Mars',
195 'playlist_count': 10,
197 'url': 'http://www.kuwo.cn/mingxing/Ali/music.htm',
202 'playlist_mincount': 95,
205 def _real_extract(self, url):
206 singer_id = self._match_id(url)
207 webpage = self._download_webpage(
208 url, singer_id, note='Download singer info',
209 errnote='Unable to get singer info')
211 singer_name = self._html_search_regex(
212 r'<div class="title clearfix">\s*<h1>([^<]+)<span', webpage, 'singer name'
216 first_page_only = False if re.search(r'/music(?:_[0-9]+)?\.htm', url) else True
217 for page_num in itertools.count(1):
218 webpage = self._download_webpage(
219 'http://www.kuwo.cn/mingxing/%s/music_%d.htm' % (singer_id, page_num),
220 singer_id, note='Download song list page #%d' % page_num,
221 errnote='Unable to get song list page #%d' % page_num)
224 self.url_result('http://www.kuwo.cn/yinyue/%s/' % song_id, 'Kuwo', song_id)
225 for song_id in re.findall(
226 r'<p[^>]+class="m_name"><a[^>]+href="http://www\.kuwo\.cn/yinyue/([0-9]+)/',
228 ][:10 if first_page_only else None])
230 if first_page_only or not re.search(r'<a[^>]+href="[^"]+">下一页</a>', webpage):
233 return self.playlist_result(entries, singer_id, singer_name)
236 class KuwoCategoryIE(InfoExtractor):
237 IE_NAME = 'kuwo:category'
238 _VALID_URL = r'http://yinyue\.kuwo\.cn/yy/cinfo_(?P<id>[0-9]+?).htm'
240 'url': 'http://yinyue.kuwo.cn/yy/cinfo_86375.htm',
244 'description': '这些都是属于八十年代的回忆!',
246 'playlist_count': 30,
249 def _real_extract(self, url):
250 category_id = self._match_id(url)
251 webpage = self._download_webpage(
252 url, category_id, note='Download category info',
253 errnote='Unable to get category info')
255 category_name = self._html_search_regex(
256 r'<h1[^>]+title="([^<>]+?)">[^<>]+?</h1>', webpage, 'category name')
258 category_desc = remove_start(
259 get_element_by_id('intro', webpage).strip(),
260 '%s简介:' % category_name)
262 jsonm = self._parse_json(self._html_search_regex(
263 r'var\s+jsonm\s*=\s*([^;]+);', webpage, 'category songs'), category_id)
267 'http://www.kuwo.cn/yinyue/%s/' % song['musicrid'],
268 'Kuwo', song['musicrid'])
269 for song in jsonm['musiclist']
271 return self.playlist_result(entries, category_id, category_name, category_desc)
274 class KuwoMvIE(KuwoBaseIE):
276 _VALID_URL = r'http://www\.kuwo\.cn/mv/(?P<id>[0-9]+?)/'
278 'url': 'http://www.kuwo.cn/mv/6480076/',
286 _FORMATS = KuwoBaseIE._FORMATS + [
287 {'format': 'mkv', 'ext': 'mkv', 'preference': 250},
288 {'format': 'mp4', 'ext': 'mp4', 'preference': 200},
291 def _real_extract(self, url):
292 song_id = self._match_id(url)
293 webpage = self._download_webpage(
294 url, song_id, note='Download mv detail info: %s' % song_id,
295 errnote='Unable to get mv detail info: %s' % song_id)
298 r'<h1[^>]+title="(?P<song>[^"]+)">[^<]+<span[^>]+title="(?P<singer>[^"]+)"',
301 song_name = mobj.group('song')
302 singer_name = mobj.group('singer')
304 raise ExtractorError('Unable to find song or singer names')
306 formats = self._get_formats(song_id)
311 'creator': singer_name,