projects
/
youtube-dl
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
[generic] Extract Instagram embeds (#8817)
[youtube-dl]
/
youtube_dl
/
extractor
/
kuwo.py
diff --git
a/youtube_dl/extractor/kuwo.py
b/youtube_dl/extractor/kuwo.py
index f94804d06865c9e5b72c936818dbce5f4bf09229..a586308b2d31e8bbac83b8446c5e00dd9b2bdce9 100644
(file)
--- a/
youtube_dl/extractor/kuwo.py
+++ b/
youtube_dl/extractor/kuwo.py
@@
-2,13
+2,13
@@
from __future__ import unicode_literals
import re
from __future__ import unicode_literals
import re
-import itertools
from .common import InfoExtractor
from ..utils import (
get_element_by_id,
clean_html,
ExtractorError,
from .common import InfoExtractor
from ..utils import (
get_element_by_id,
clean_html,
ExtractorError,
+ InAdvancePagedList,
remove_start,
)
remove_start,
)
@@
-55,7
+55,7
@@
class KuwoBaseIE(InfoExtractor):
class KuwoIE(KuwoBaseIE):
IE_NAME = 'kuwo:song'
IE_DESC = '酷我音乐'
class KuwoIE(KuwoBaseIE):
IE_NAME = 'kuwo:song'
IE_DESC = '酷我音乐'
- _VALID_URL = r'http
://www\.kuwo\.cn/yinyue/(?P<id>\d+?)/
'
+ _VALID_URL = r'http
s?://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': {
@@
-80,6
+80,9
@@
class KuwoIE(KuwoBaseIE):
'params': {
'format': 'mp3-320'
},
'params': {
'format': 'mp3-320'
},
+ }, {
+ 'url': 'http://www.kuwo.cn/yinyue/3197154?catalog=yueku2016',
+ 'only_matching': True,
}]
def _real_extract(self, url):
}]
def _real_extract(self, url):
@@
-131,7
+134,7
@@
class KuwoIE(KuwoBaseIE):
class KuwoAlbumIE(InfoExtractor):
IE_NAME = 'kuwo:album'
IE_DESC = '酷我音乐 - 专辑'
class KuwoAlbumIE(InfoExtractor):
IE_NAME = 'kuwo:album'
IE_DESC = '酷我音乐 - 专辑'
- _VALID_URL = r'http://www\.kuwo\.cn/album/(?P<id>\d+?)/'
+ _VALID_URL = r'http
s?
://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': {
@@
-167,13
+170,11
@@
class KuwoAlbumIE(InfoExtractor):
class KuwoChartIE(InfoExtractor):
IE_NAME = 'kuwo:chart'
IE_DESC = '酷我音乐 - 排行榜'
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'
+ _VALID_URL = r'http
s?
://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',
_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:\d{4}第\d{2}期',
},
'playlist_mincount': 10,
}
},
'playlist_mincount': 10,
}
@@
-184,30
+185,24
@@
class KuwoChartIE(InfoExtractor):
url, chart_id, note='Download chart info',
errnote='Unable to get chart info')
url, chart_id, note='Download chart info',
errnote='Unable to get chart info')
- chart_name = self._html_search_regex(
- r'<h1[^>]+class="unDis">([^<]+)</h1>', webpage, 'chart name')
-
- chart_desc = self._html_search_regex(
- r'<p[^>]+class="tabDef">(\d{4}第\d{2}期)</p>', webpage, 'chart desc')
-
entries = [
self.url_result(song_url, 'Kuwo') for song_url in re.findall(
entries = [
self.url_result(song_url, 'Kuwo') for song_url in re.findall(
- r'<a[^>]+href="(http://www\.kuwo\.cn/yinyue/\d+)
/"
', webpage)
+ r'<a[^>]+href="(http://www\.kuwo\.cn/yinyue/\d+)', webpage)
]
]
- return self.playlist_result(entries, chart_id
, chart_name, chart_desc
)
+ return self.playlist_result(entries, chart_id)
class KuwoSingerIE(InfoExtractor):
IE_NAME = 'kuwo:singer'
IE_DESC = '酷我音乐 - 歌手'
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>[^/]+)'
+ _VALID_URL = r'http
s?
://www\.kuwo\.cn/mingxing/(?P<id>[^/]+)'
_TESTS = [{
'url': 'http://www.kuwo.cn/mingxing/bruno+mars/',
'info_dict': {
'id': 'bruno+mars',
'title': 'Bruno Mars',
},
_TESTS = [{
'url': 'http://www.kuwo.cn/mingxing/bruno+mars/',
'info_dict': {
'id': 'bruno+mars',
'title': 'Bruno Mars',
},
- 'playlist_
count': 10
,
+ 'playlist_
mincount': 329
,
}, {
'url': 'http://www.kuwo.cn/mingxing/Ali/music.htm',
'info_dict': {
}, {
'url': 'http://www.kuwo.cn/mingxing/Ali/music.htm',
'info_dict': {
@@
-218,6
+213,8
@@
class KuwoSingerIE(InfoExtractor):
'skip': 'Regularly stalls travis build', # See https://travis-ci.org/rg3/youtube-dl/jobs/78878540
}]
'skip': 'Regularly stalls travis build', # See https://travis-ci.org/rg3/youtube-dl/jobs/78878540
}]
+ PAGE_SIZE = 15
+
def _real_extract(self, url):
singer_id = self._match_id(url)
webpage = self._download_webpage(
def _real_extract(self, url):
singer_id = self._match_id(url)
webpage = self._download_webpage(
@@
-225,25
+222,28
@@
class KuwoSingerIE(InfoExtractor):
errnote='Unable to get singer info')
singer_name = self._html_search_regex(
errnote='Unable to get singer info')
singer_name = self._html_search_regex(
- r'<div class="title clearfix">\s*<h1>([^<]+)<span', webpage, 'singer name'
- )
+ r'<h1>([^<]+)</h1>', webpage, 'singer name')
+
+ artist_id = self._html_search_regex(
+ r'data-artistid="(\d+)"', webpage, 'artist id')
+
+ page_count = int(self._html_search_regex(
+ r'data-page="(\d+)"', webpage, 'page count'))
- entries = []
- first_page_only = False if re.search(r'/music(?:_\d+)?\.htm', url) else True
- for page_num in itertools.count(1):
+ def page_func(page_num):
webpage = self._download_webpage(
webpage = self._download_webpage(
- 'http://www.kuwo.cn/mingxing/%s/music_%d.htm' % (singer_id, page_num),
- singer_id, note='Download song list page #%d' % page_num,
- errnote='Unable to get song list page #%d' % page_num)
+ 'http://www.kuwo.cn/artist/contentMusicsAjax',
+ singer_id, note='Download song list page #%d' % (page_num + 1),
+ errnote='Unable to get song list page #%d' % (page_num + 1),
+ query={'artistId': artist_id, 'pn': page_num, 'rn': self.PAGE_SIZE})
-
entries.extend(
[
+
return
[
self.url_result(song_url, 'Kuwo') for song_url in re.findall(
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+)/
',
+ r'<
div[^>]+class="name"><a[^>]+href="(http://www\.kuwo\.cn/yinyue/\d+)
',
webpage)
webpage)
- ]
[:10 if first_page_only else None])
+ ]
- if first_page_only or not re.search(r'<a[^>]+href="[^"]+">下一页</a>', webpage):
- break
+ entries = InAdvancePagedList(page_func, page_count, self.PAGE_SIZE)
return self.playlist_result(entries, singer_id, singer_name)
return self.playlist_result(entries, singer_id, singer_name)
@@
-251,7
+251,7
@@
class KuwoSingerIE(InfoExtractor):
class KuwoCategoryIE(InfoExtractor):
IE_NAME = 'kuwo:category'
IE_DESC = '酷我音乐 - 分类'
class KuwoCategoryIE(InfoExtractor):
IE_NAME = 'kuwo:category'
IE_DESC = '酷我音乐 - 分类'
- _VALID_URL = r'http://yinyue\.kuwo\.cn/yy/cinfo_(?P<id>\d+?).htm'
+ _VALID_URL = r'http
s?
://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': {
@@
-288,7
+288,7
@@
class KuwoCategoryIE(InfoExtractor):
class KuwoMvIE(KuwoBaseIE):
IE_NAME = 'kuwo:mv'
IE_DESC = '酷我音乐 - MV'
class KuwoMvIE(KuwoBaseIE):
IE_NAME = 'kuwo:mv'
IE_DESC = '酷我音乐 - MV'
- _VALID_URL = r'http://www\.kuwo\.cn/mv/(?P<id>\d+?)/'
+ _VALID_URL = r'http
s?
://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': {