formats = []
for file_format in self._FORMATS:
song_url = self._download_webpage(
- "http://antiserver.kuwo.cn/anti.s?format=%s&br=%s&rid=MUSIC_%s&type=convert_url&response=url" %
+ 'http://antiserver.kuwo.cn/anti.s?format=%s&br=%s&rid=MUSIC_%s&type=convert_url&response=url' %
(file_format['ext'], file_format.get('br', ''), song_id),
- song_id, note="Download %s url info" % file_format["format"],
+ song_id, note='Download %s url info' % file_format['format'],
)
if song_url.startswith('http://') or song_url.startswith('https://'):
formats.append({
class KuwoIE(KuwoBaseIE):
IE_NAME = 'kuwo:song'
- _VALID_URL = r'http://www\.kuwo\.cn/yinyue/(?P<id>[0-9]+?)/'
+ IE_DESC = '酷我音乐'
+ _VALID_URL = r'http://www\.kuwo\.cn/yinyue/(?P<id>\d+?)/'
_TESTS = [{
'url': 'http://www.kuwo.cn/yinyue/635632/',
'info_dict': {
'upload_date': '20080122',
'description': 'md5:ed13f58e3c3bf3f7fd9fbc4e5a7aa75c'
},
+ 'skip': 'this song has been offline because of copyright issues',
}, {
'url': 'http://www.kuwo.cn/yinyue/6446136/',
'info_dict': {
webpage = self._download_webpage(
url, song_id, note='Download song detail info',
errnote='Unable to get song detail info')
+ if '对不起,该歌曲由于版权问题已被下线,将返回网站首页' in webpage:
+ raise ExtractorError('this song has been offline because of copyright issues', expected=True)
song_name = self._html_search_regex(
- r'<h1[^>]+title="([^"]+)">', webpage, 'song name')
+ r'(?s)class="(?:[^"\s]+\s+)*title(?:\s+[^"\s]+)*".*?<h1[^>]+title="([^"]+)"', webpage, 'song name')
singer_name = self._html_search_regex(
r'<div[^>]+class="s_img">\s*<a[^>]+title="([^>]+)"',
- webpage, 'singer name', default=None)
- lrc_content = clean_html(get_element_by_id("lrcContent", webpage))
+ webpage, 'singer name', fatal=False)
+ lrc_content = clean_html(get_element_by_id('lrcContent', webpage))
if lrc_content == '暂无': # indicates no lyrics
lrc_content = None
album_id = self._html_search_regex(
r'<p[^>]+class="album"[^<]+<a[^>]+href="http://www\.kuwo\.cn/album/(\d+)/"',
- webpage, 'album id', default=None, fatal=False)
+ webpage, 'album id', fatal=False)
publish_time = None
if album_id is not None:
album_info_page = self._download_webpage(
- "http://www.kuwo.cn/album/%s/" % album_id, song_id,
+ 'http://www.kuwo.cn/album/%s/' % album_id, song_id,
note='Download album detail info',
errnote='Unable to get album detail info')
publish_time = self._html_search_regex(
r'发行时间:(\d{4}-\d{2}-\d{2})', album_info_page,
- 'publish time', default=None)
+ 'publish time', fatal=False)
if publish_time:
publish_time = publish_time.replace('-', '')
class KuwoAlbumIE(InfoExtractor):
IE_NAME = 'kuwo:album'
- _VALID_URL = r'http://www\.kuwo\.cn/album/(?P<id>[0-9]+?)/'
+ IE_DESC = '酷我音乐 - 专辑'
+ _VALID_URL = r'http://www\.kuwo\.cn/album/(?P<id>\d+?)/'
_TEST = {
'url': 'http://www.kuwo.cn/album/502294/',
'info_dict': {
r'<div[^>]+class="comm"[^<]+<h1[^>]+title="([^"]+)"', webpage,
'album name')
album_intro = remove_start(
- clean_html(get_element_by_id("intro", webpage)),
+ clean_html(get_element_by_id('intro', webpage)),
'%s简介:' % album_name)
entries = [
- self.url_result("http://www.kuwo.cn/yinyue/%s/" % song_id, 'Kuwo', song_id)
- for song_id in re.findall(
- r'<p[^>]+class="listen"><a[^>]+href="http://www\.kuwo\.cn/yinyue/(\d+)/"',
+ self.url_result(song_url, 'Kuwo') for song_url in re.findall(
+ r'<p[^>]+class="listen"><a[^>]+href="(http://www\.kuwo\.cn/yinyue/\d+/)"',
webpage)
]
return self.playlist_result(entries, album_id, album_name, album_intro)
class KuwoChartIE(InfoExtractor):
IE_NAME = 'kuwo:chart'
+ IE_DESC = '酷我音乐 - 排行榜'
_VALID_URL = r'http://yinyue\.kuwo\.cn/billboard_(?P<id>[^.]+).htm'
_TEST = {
'url': 'http://yinyue.kuwo.cn/billboard_香港中文龙虎榜.htm',
'info_dict': {
'id': '香港中文龙虎榜',
'title': '香港中文龙虎榜',
- 'description': 're:[0-9]{4}第[0-9]{2}期',
+ 'description': 're:\d{4}第\d{2}期',
},
'playlist_mincount': 10,
}
r'<p[^>]+class="tabDef">(\d{4}第\d{2}期)</p>', webpage, 'chart desc')
entries = [
- self.url_result("http://www.kuwo.cn/yinyue/%s/" % song_id, 'Kuwo', song_id)
- for song_id in re.findall(
- r'<a[^>]+href="http://www\.kuwo\.cn/yinyue/(\d+)/"', webpage)
+ self.url_result(song_url, 'Kuwo') for song_url in re.findall(
+ r'<a[^>]+href="(http://www\.kuwo\.cn/yinyue/\d+)/"', webpage)
]
return self.playlist_result(entries, chart_id, chart_name, chart_desc)
class KuwoSingerIE(InfoExtractor):
IE_NAME = 'kuwo:singer'
+ IE_DESC = '酷我音乐 - 歌手'
_VALID_URL = r'http://www\.kuwo\.cn/mingxing/(?P<id>[^/]+)'
_TESTS = [{
'url': 'http://www.kuwo.cn/mingxing/bruno+mars/',
'title': 'Ali',
},
'playlist_mincount': 95,
+ 'skip': 'Regularly stalls travis build', # See https://travis-ci.org/rg3/youtube-dl/jobs/78878540
}]
def _real_extract(self, url):
)
entries = []
- first_page_only = False if re.search(r'/music(?:_[0-9]+)?\.htm', url) else True
+ first_page_only = False if re.search(r'/music(?:_\d+)?\.htm', url) else True
for page_num in itertools.count(1):
webpage = self._download_webpage(
'http://www.kuwo.cn/mingxing/%s/music_%d.htm' % (singer_id, page_num),
errnote='Unable to get song list page #%d' % page_num)
entries.extend([
- self.url_result("http://www.kuwo.cn/yinyue/%s/" % song_id, 'Kuwo', song_id)
- for song_id in re.findall(
- r'<p[^>]+class="m_name"><a[^>]+href="http://www\.kuwo\.cn/yinyue/([0-9]+)/',
+ self.url_result(song_url, 'Kuwo') for song_url in re.findall(
+ r'<p[^>]+class="m_name"><a[^>]+href="(http://www\.kuwo\.cn/yinyue/\d+)/',
webpage)
][:10 if first_page_only else None])
class KuwoCategoryIE(InfoExtractor):
IE_NAME = 'kuwo:category'
- _VALID_URL = r'http://yinyue\.kuwo\.cn/yy/cinfo_(?P<id>[0-9]+?).htm'
+ IE_DESC = '酷我音乐 - 分类'
+ _VALID_URL = r'http://yinyue\.kuwo\.cn/yy/cinfo_(?P<id>\d+?).htm'
_TEST = {
'url': 'http://yinyue.kuwo.cn/yy/cinfo_86375.htm',
'info_dict': {
r'<h1[^>]+title="([^<>]+?)">[^<>]+?</h1>', webpage, 'category name')
category_desc = remove_start(
- get_element_by_id("intro", webpage).strip(),
+ get_element_by_id('intro', webpage).strip(),
'%s简介:' % category_name)
jsonm = self._parse_json(self._html_search_regex(
r'var\s+jsonm\s*=\s*([^;]+);', webpage, 'category songs'), category_id)
entries = [
- self.url_result(
- "http://www.kuwo.cn/yinyue/%s/" % song['musicrid'],
- 'Kuwo', song['musicrid'])
+ self.url_result('http://www.kuwo.cn/yinyue/%s/' % song['musicrid'], 'Kuwo')
for song in jsonm['musiclist']
]
return self.playlist_result(entries, category_id, category_name, category_desc)
class KuwoMvIE(KuwoBaseIE):
IE_NAME = 'kuwo:mv'
- _VALID_URL = r'http://www\.kuwo\.cn/mv/(?P<id>[0-9]+?)/'
+ IE_DESC = '酷我音乐 - MV'
+ _VALID_URL = r'http://www\.kuwo\.cn/mv/(?P<id>\d+?)/'
_TEST = {
'url': 'http://www.kuwo.cn/mv/6480076/',
'info_dict': {
song_name = mobj.group('song')
singer_name = mobj.group('singer')
else:
- raise ExtractorError("Unable to find song or singer names")
+ raise ExtractorError('Unable to find song or singer names')
formats = self._get_formats(song_id)