projects
/
youtube-dl
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
[extractor/common] Make ie_key and IE_NAME return unicode string
[youtube-dl]
/
youtube_dl
/
extractor
/
kuwo.py
diff --git
a/youtube_dl/extractor/kuwo.py
b/youtube_dl/extractor/kuwo.py
index 69afacac909b1bd2903681faf8a3ecca6e88cfdb..0c8ed5d07258d463375c2848d2f93c79885bdaae 100644
(file)
--- a/
youtube_dl/extractor/kuwo.py
+++ b/
youtube_dl/extractor/kuwo.py
@@
-45,7
+45,8
@@
class KuwoBaseIE(InfoExtractor):
class KuwoIE(KuwoBaseIE):
IE_NAME = 'kuwo:song'
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': {
_TESTS = [{
'url': 'http://www.kuwo.cn/yinyue/635632/',
'info_dict': {
@@
-56,6
+57,7
@@
class KuwoIE(KuwoBaseIE):
'upload_date': '20080122',
'description': 'md5:ed13f58e3c3bf3f7fd9fbc4e5a7aa75c'
},
'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': {
}, {
'url': 'http://www.kuwo.cn/yinyue/6446136/',
'info_dict': {
@@
-75,12
+77,14
@@
class KuwoIE(KuwoBaseIE):
webpage = self._download_webpage(
url, song_id, note='Download song detail info',
errnote='Unable to get song detail info')
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(
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="([^>]+)"',
singer_name = self._html_search_regex(
r'<div[^>]+class="s_img">\s*<a[^>]+title="([^>]+)"',
- webpage, 'singer name',
default=Non
e)
+ webpage, 'singer name',
fatal=Fals
e)
lrc_content = clean_html(get_element_by_id('lrcContent', webpage))
if lrc_content == '暂无': # indicates no lyrics
lrc_content = None
lrc_content = clean_html(get_element_by_id('lrcContent', webpage))
if lrc_content == '暂无': # indicates no lyrics
lrc_content = None
@@
-89,7
+93,7
@@
class KuwoIE(KuwoBaseIE):
album_id = self._html_search_regex(
r'<p[^>]+class="album"[^<]+<a[^>]+href="http://www\.kuwo\.cn/album/(\d+)/"',
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:
publish_time = None
if album_id is not None:
@@
-100,7
+104,7
@@
class KuwoIE(KuwoBaseIE):
publish_time = self._html_search_regex(
r'发行时间:(\d{4}-\d{2}-\d{2})', album_info_page,
publish_time = self._html_search_regex(
r'发行时间:(\d{4}-\d{2}-\d{2})', album_info_page,
- 'publish time',
default=Non
e)
+ 'publish time',
fatal=Fals
e)
if publish_time:
publish_time = publish_time.replace('-', '')
if publish_time:
publish_time = publish_time.replace('-', '')
@@
-116,7
+120,8
@@
class KuwoIE(KuwoBaseIE):
class KuwoAlbumIE(InfoExtractor):
IE_NAME = 'kuwo:album'
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': {
_TEST = {
'url': 'http://www.kuwo.cn/album/502294/',
'info_dict': {
@@
-151,13
+156,14
@@
class KuwoAlbumIE(InfoExtractor):
class KuwoChartIE(InfoExtractor):
IE_NAME = 'kuwo:chart'
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',
_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,
}
},
'playlist_mincount': 10,
}
@@
-183,6
+189,7
@@
class KuwoChartIE(InfoExtractor):
class KuwoSingerIE(InfoExtractor):
IE_NAME = 'kuwo:singer'
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/',
_VALID_URL = r'http://www\.kuwo\.cn/mingxing/(?P<id>[^/]+)'
_TESTS = [{
'url': 'http://www.kuwo.cn/mingxing/bruno+mars/',
@@
-198,6
+205,7
@@
class KuwoSingerIE(InfoExtractor):
'title': 'Ali',
},
'playlist_mincount': 95,
'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):
}]
def _real_extract(self, url):
@@
-211,7
+219,7
@@
class KuwoSingerIE(InfoExtractor):
)
entries = []
)
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),
for page_num in itertools.count(1):
webpage = self._download_webpage(
'http://www.kuwo.cn/mingxing/%s/music_%d.htm' % (singer_id, page_num),
@@
-232,7
+240,8
@@
class KuwoSingerIE(InfoExtractor):
class KuwoCategoryIE(InfoExtractor):
IE_NAME = 'kuwo:category'
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': {
_TEST = {
'url': 'http://yinyue.kuwo.cn/yy/cinfo_86375.htm',
'info_dict': {
@@
-268,7
+277,8
@@
class KuwoCategoryIE(InfoExtractor):
class KuwoMvIE(KuwoBaseIE):
IE_NAME = 'kuwo:mv'
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': {
_TEST = {
'url': 'http://www.kuwo.cn/mv/6480076/',
'info_dict': {