]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/douyutv.py
5 from .common
import InfoExtractor
14 class DouyuTVIE(InfoExtractor
):
16 _VALID_URL
= r
'https?://(?:www\.)?douyu(?:tv)?\.com/(?:[^/]+/)*(?P<id>[A-Za-z0-9]+)'
18 'url': 'http://www.douyutv.com/iseven',
21 'display_id': 'iseven',
23 'title': 're:^清晨醒脑!根本停不下来! [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
24 'description': r
're:.*m7show@163\.com.*',
25 'thumbnail': r
're:^https?://.*\.jpg$',
30 'skip_download': True,
33 'url': 'http://www.douyutv.com/85982',
36 'display_id': '85982',
38 'title': 're:^小漠从零单排记!——CSOL2躲猫猫 [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
39 'description': 'md5:746a2f7a253966a06755a912f0acc0d2',
40 'thumbnail': r
're:^https?://.*\.jpg$',
41 'uploader': 'douyu小漠',
45 'skip_download': True,
47 'skip': 'Room not found',
49 'url': 'http://www.douyutv.com/17732',
52 'display_id': '17732',
54 'title': 're:^清晨醒脑!根本停不下来! [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
55 'description': r
're:.*m7show@163\.com.*',
56 'thumbnail': r
're:^https?://.*\.jpg$',
61 'skip_download': True,
64 'url': 'http://www.douyu.com/xiaocang',
65 'only_matching': True,
68 'url': 'http://www.douyu.com/t/lpl',
69 'only_matching': True,
72 def _real_extract(self
, url
):
73 video_id
= self
._match
_id
(url
)
75 if video_id
.isdigit():
78 page
= self
._download
_webpage
(url
, video_id
)
79 room_id
= self
._html
_search
_regex
(
80 r
'"room_id\\?"\s*:\s*(\d+),', page
, 'room id')
82 # Grab metadata from mobile API
83 room
= self
._download
_json
(
84 'http://m.douyu.com/html5/live?roomId=%s' % room_id
, video_id
,
85 note
='Downloading room info')['data']
87 # 1 = live, 2 = offline
88 if room
.get('show_status') == '2':
89 raise ExtractorError('Live stream is offline', expected
=True)
91 # Grab the URL from PC client API
92 # The m3u8 url from mobile API requires re-authentication every 5 minutes
94 signContent
= 'lapi/live/thirdPart/getPlay/%s?aid=pcclient&rate=0&time=%d9TUk5fjjUjg9qIMH3sdnh' % (room_id
, tt
)
95 sign
= hashlib
.md5(signContent
.encode('ascii')).hexdigest()
96 video_url
= self
._download
_json
(
97 'http://coapi.douyucdn.cn/lapi/live/thirdPart/getPlay/' + room_id
,
98 video_id
, note
='Downloading video URL info',
99 query
={'rate': 0}
, headers
={
103 })['data']['live_url']
105 title
= unescapeHTML(room
['room_name'])
106 description
= room
.get('show_details')
107 thumbnail
= room
.get('room_src')
108 uploader
= room
.get('nickname')
112 'display_id': video_id
,
115 'description': description
,
116 'thumbnail': thumbnail
,
117 'uploader': uploader
,
122 class DouyuShowIE(InfoExtractor
):
123 _VALID_URL
= r
'https?://v(?:mobile)?\.douyu\.com/show/(?P<id>[0-9a-zA-Z]+)'
126 'url': 'https://v.douyu.com/show/rjNBdvnVXNzvE2yw',
127 'md5': '0c2cfd068ee2afe657801269b2d86214',
129 'id': 'rjNBdvnVXNzvE2yw',
131 'title': '陈一发儿:砒霜 我有个室友系列!04-01 22点场',
133 'thumbnail': r
're:^https?://.*\.jpg$',
135 'uploader_id': 'XrZwYelr5wbK',
136 'uploader_url': 'https://v.douyu.com/author/XrZwYelr5wbK',
137 'upload_date': '20170402',
140 'url': 'https://vmobile.douyu.com/show/rjNBdvnVXNzvE2yw',
141 'only_matching': True,
144 def _real_extract(self
, url
):
145 url
= url
.replace('vmobile.', 'v.')
146 video_id
= self
._match
_id
(url
)
148 webpage
= self
._download
_webpage
(url
, video_id
)
150 room_info
= self
._parse
_json
(self
._search
_regex
(
151 r
'var\s+\$ROOM\s*=\s*({.+});', webpage
, 'room info'), video_id
)
155 for trial
in range(5):
156 # Sometimes Douyu rejects our request. Let's try it more times
158 video_info
= self
._download
_json
(
159 'https://vmobile.douyu.com/video/getInfo', video_id
,
160 query
={'vid': video_id}
,
163 'x-requested-with': 'XMLHttpRequest',
166 except ExtractorError
:
167 self
._sleep
(1, video_id
)
170 raise ExtractorError('Can\'t fetch video info')
172 formats
= self
._extract
_m
3u8_formats
(
173 video_info
['data']['video_url'], video_id
,
174 entry_protocol
='m3u8_native', ext
='mp4')
176 upload_date
= unified_strdate(self
._html
_search
_regex
(
177 r
'<em>上传时间:</em><span>([^<]+)</span>', webpage
,
178 'upload date', fatal
=False))
180 uploader
= uploader_id
= uploader_url
= None
182 r
'(?m)<a[^>]+href="/author/([0-9a-zA-Z]+)".+?<strong[^>]+title="([^"]+)"',
185 uploader_id
, uploader
= mobj
.groups()
186 uploader_url
= urljoin(url
, '/author/' + uploader_id
)
190 'title': room_info
['name'],
192 'duration': room_info
.get('duration'),
193 'thumbnail': room_info
.get('pic'),
194 'upload_date': upload_date
,
195 'uploader': uploader
,
196 'uploader_id': uploader_id
,
197 'uploader_url': uploader_url
,