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):
49 _VALID_URL = r'http://www\.kuwo\.cn/yinyue/(?P<id>\d+?)/'
51 'url': 'http://www.kuwo.cn/yinyue/635632/',
57 'upload_date': '20080122',
58 'description': 'md5:ed13f58e3c3bf3f7fd9fbc4e5a7aa75c'
60 'skip': 'this song has been offline because of copyright issues',
62 'url': 'http://www.kuwo.cn/yinyue/6446136/',
68 'upload_date': '20150518',
75 def _real_extract(self, url):
76 song_id = self._match_id(url)
77 webpage = self._download_webpage(
78 url, song_id, note='Download song detail info',
79 errnote='Unable to get song detail info')
81 song_name = self._html_search_regex(
82 r'(?s)class="[^"]*title[^"]*".*?<h1[^>]+title="([^"]+)"', webpage, 'song name')
83 singer_name = self._html_search_regex(
84 r'<div[^>]+class="s_img">\s*<a[^>]+title="([^>]+)"',
85 webpage, 'singer name', fatal=False)
86 lrc_content = clean_html(get_element_by_id('lrcContent', webpage))
87 if lrc_content == '暂无': # indicates no lyrics
90 formats = self._get_formats(song_id)
92 album_id = self._html_search_regex(
93 r'<p[^>]+class="album"[^<]+<a[^>]+href="http://www\.kuwo\.cn/album/(\d+)/"',
94 webpage, 'album id', fatal=False)
97 if album_id is not None:
98 album_info_page = self._download_webpage(
99 'http://www.kuwo.cn/album/%s/' % album_id, song_id,
100 note='Download album detail info',
101 errnote='Unable to get album detail info')
103 publish_time = self._html_search_regex(
104 r'发行时间:(\d{4}-\d{2}-\d{2})', album_info_page,
105 'publish time', fatal=False)
107 publish_time = publish_time.replace('-', '')
112 'creator': singer_name,
113 'upload_date': publish_time,
114 'description': lrc_content,
119 class KuwoAlbumIE(InfoExtractor):
120 IE_NAME = 'kuwo:album'
121 IE_DESC = '酷我音乐 - 专辑'
122 _VALID_URL = r'http://www\.kuwo\.cn/album/(?P<id>\d+?)/'
124 'url': 'http://www.kuwo.cn/album/502294/',
128 'description': 'md5:6a7235a84cc6400ec3b38a7bdaf1d60c',
133 def _real_extract(self, url):
134 album_id = self._match_id(url)
136 webpage = self._download_webpage(
137 url, album_id, note='Download album info',
138 errnote='Unable to get album info')
140 album_name = self._html_search_regex(
141 r'<div[^>]+class="comm"[^<]+<h1[^>]+title="([^"]+)"', webpage,
143 album_intro = remove_start(
144 clean_html(get_element_by_id('intro', webpage)),
145 '%s简介:' % album_name)
148 self.url_result(song_url, 'Kuwo') for song_url in re.findall(
149 r'<p[^>]+class="listen"><a[^>]+href="(http://www\.kuwo\.cn/yinyue/\d+/)"',
152 return self.playlist_result(entries, album_id, album_name, album_intro)
155 class KuwoChartIE(InfoExtractor):
156 IE_NAME = 'kuwo:chart'
157 IE_DESC = '酷我音乐 - 排行榜'
158 _VALID_URL = r'http://yinyue\.kuwo\.cn/billboard_(?P<id>[^.]+).htm'
160 'url': 'http://yinyue.kuwo.cn/billboard_香港中文龙虎榜.htm',
164 'description': 're:\d{4}第\d{2}期',
166 'playlist_mincount': 10,
169 def _real_extract(self, url):
170 chart_id = self._match_id(url)
171 webpage = self._download_webpage(
172 url, chart_id, note='Download chart info',
173 errnote='Unable to get chart info')
175 chart_name = self._html_search_regex(
176 r'<h1[^>]+class="unDis">([^<]+)</h1>', webpage, 'chart name')
178 chart_desc = self._html_search_regex(
179 r'<p[^>]+class="tabDef">(\d{4}第\d{2}期)</p>', webpage, 'chart desc')
182 self.url_result(song_url, 'Kuwo') for song_url in re.findall(
183 r'<a[^>]+href="(http://www\.kuwo\.cn/yinyue/\d+)/"', webpage)
185 return self.playlist_result(entries, chart_id, chart_name, chart_desc)
188 class KuwoSingerIE(InfoExtractor):
189 IE_NAME = 'kuwo:singer'
190 IE_DESC = '酷我音乐 - 歌手'
191 _VALID_URL = r'http://www\.kuwo\.cn/mingxing/(?P<id>[^/]+)'
193 'url': 'http://www.kuwo.cn/mingxing/bruno+mars/',
196 'title': 'Bruno Mars',
198 'playlist_count': 10,
200 'url': 'http://www.kuwo.cn/mingxing/Ali/music.htm',
205 'playlist_mincount': 95,
206 'skip': 'Regularly stalls travis build', # See https://travis-ci.org/rg3/youtube-dl/jobs/78878540
209 def _real_extract(self, url):
210 singer_id = self._match_id(url)
211 webpage = self._download_webpage(
212 url, singer_id, note='Download singer info',
213 errnote='Unable to get singer info')
215 singer_name = self._html_search_regex(
216 r'<div class="title clearfix">\s*<h1>([^<]+)<span', webpage, 'singer name'
220 first_page_only = False if re.search(r'/music(?:_\d+)?\.htm', url) else True
221 for page_num in itertools.count(1):
222 webpage = self._download_webpage(
223 'http://www.kuwo.cn/mingxing/%s/music_%d.htm' % (singer_id, page_num),
224 singer_id, note='Download song list page #%d' % page_num,
225 errnote='Unable to get song list page #%d' % page_num)
228 self.url_result(song_url, 'Kuwo') for song_url in re.findall(
229 r'<p[^>]+class="m_name"><a[^>]+href="(http://www\.kuwo\.cn/yinyue/\d+)/',
231 ][:10 if first_page_only else None])
233 if first_page_only or not re.search(r'<a[^>]+href="[^"]+">下一页</a>', webpage):
236 return self.playlist_result(entries, singer_id, singer_name)
239 class KuwoCategoryIE(InfoExtractor):
240 IE_NAME = 'kuwo:category'
241 IE_DESC = '酷我音乐 - 分类'
242 _VALID_URL = r'http://yinyue\.kuwo\.cn/yy/cinfo_(?P<id>\d+?).htm'
244 'url': 'http://yinyue.kuwo.cn/yy/cinfo_86375.htm',
248 'description': '这些都是属于八十年代的回忆!',
250 'playlist_count': 30,
253 def _real_extract(self, url):
254 category_id = self._match_id(url)
255 webpage = self._download_webpage(
256 url, category_id, note='Download category info',
257 errnote='Unable to get category info')
259 category_name = self._html_search_regex(
260 r'<h1[^>]+title="([^<>]+?)">[^<>]+?</h1>', webpage, 'category name')
262 category_desc = remove_start(
263 get_element_by_id('intro', webpage).strip(),
264 '%s简介:' % category_name)
266 jsonm = self._parse_json(self._html_search_regex(
267 r'var\s+jsonm\s*=\s*([^;]+);', webpage, 'category songs'), category_id)
270 self.url_result('http://www.kuwo.cn/yinyue/%s/' % song['musicrid'], 'Kuwo')
271 for song in jsonm['musiclist']
273 return self.playlist_result(entries, category_id, category_name, category_desc)
276 class KuwoMvIE(KuwoBaseIE):
278 IE_DESC = '酷我音乐 - MV'
279 _VALID_URL = r'http://www\.kuwo\.cn/mv/(?P<id>\d+?)/'
281 'url': 'http://www.kuwo.cn/mv/6480076/',
289 _FORMATS = KuwoBaseIE._FORMATS + [
290 {'format': 'mkv', 'ext': 'mkv', 'preference': 250},
291 {'format': 'mp4', 'ext': 'mp4', 'preference': 200},
294 def _real_extract(self, url):
295 song_id = self._match_id(url)
296 webpage = self._download_webpage(
297 url, song_id, note='Download mv detail info: %s' % song_id,
298 errnote='Unable to get mv detail info: %s' % song_id)
301 r'<h1[^>]+title="(?P<song>[^"]+)">[^<]+<span[^>]+title="(?P<singer>[^"]+)"',
304 song_name = mobj.group('song')
305 singer_name = mobj.group('singer')
307 raise ExtractorError('Unable to find song or singer names')
309 formats = self._get_formats(song_id)
314 'creator': singer_name,