mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-12-27 13:45:51 +01:00
86e5f3ed2e
Using https://github.com/asottile/pyupgrade 1. `__future__` imports and `coding: utf-8` were removed 2. Files were rewritten with `pyupgrade --py36-plus --keep-percent-format` 3. f-strings were cherry-picked from `pyupgrade --py36-plus` Extractors are left untouched (except removing header) to avoid unnecessary merge conflicts
95 lines
3.0 KiB
Python
95 lines
3.0 KiB
Python
import json
|
|
|
|
from .common import InfoExtractor
|
|
from ..utils import int_or_none
|
|
|
|
|
|
class DLiveVODIE(InfoExtractor):
|
|
IE_NAME = 'dlive:vod'
|
|
_VALID_URL = r'https?://(?:www\.)?dlive\.tv/p/(?P<uploader_id>.+?)\+(?P<id>[^/?#&]+)'
|
|
_TESTS = [{
|
|
'url': 'https://dlive.tv/p/pdp+3mTzOl4WR',
|
|
'info_dict': {
|
|
'id': '3mTzOl4WR',
|
|
'ext': 'mp4',
|
|
'title': 'Minecraft with james charles epic',
|
|
'upload_date': '20190701',
|
|
'timestamp': 1562011015,
|
|
'uploader_id': 'pdp',
|
|
}
|
|
}, {
|
|
'url': 'https://dlive.tv/p/pdpreplay+D-RD-xSZg',
|
|
'only_matching': True,
|
|
}]
|
|
|
|
def _real_extract(self, url):
|
|
uploader_id, vod_id = self._match_valid_url(url).groups()
|
|
broadcast = self._download_json(
|
|
'https://graphigo.prd.dlive.tv/', vod_id,
|
|
data=json.dumps({'query': '''query {
|
|
pastBroadcast(permlink:"%s+%s") {
|
|
content
|
|
createdAt
|
|
length
|
|
playbackUrl
|
|
title
|
|
thumbnailUrl
|
|
viewCount
|
|
}
|
|
}''' % (uploader_id, vod_id)}).encode())['data']['pastBroadcast']
|
|
title = broadcast['title']
|
|
formats = self._extract_m3u8_formats(
|
|
broadcast['playbackUrl'], vod_id, 'mp4', 'm3u8_native')
|
|
self._sort_formats(formats)
|
|
return {
|
|
'id': vod_id,
|
|
'title': title,
|
|
'uploader_id': uploader_id,
|
|
'formats': formats,
|
|
'description': broadcast.get('content'),
|
|
'thumbnail': broadcast.get('thumbnailUrl'),
|
|
'timestamp': int_or_none(broadcast.get('createdAt'), 1000),
|
|
'view_count': int_or_none(broadcast.get('viewCount')),
|
|
}
|
|
|
|
|
|
class DLiveStreamIE(InfoExtractor):
|
|
IE_NAME = 'dlive:stream'
|
|
_VALID_URL = r'https?://(?:www\.)?dlive\.tv/(?!p/)(?P<id>[\w.-]+)'
|
|
|
|
def _real_extract(self, url):
|
|
display_name = self._match_id(url)
|
|
user = self._download_json(
|
|
'https://graphigo.prd.dlive.tv/', display_name,
|
|
data=json.dumps({'query': '''query {
|
|
userByDisplayName(displayname:"%s") {
|
|
livestream {
|
|
content
|
|
createdAt
|
|
title
|
|
thumbnailUrl
|
|
watchingCount
|
|
}
|
|
username
|
|
}
|
|
}''' % display_name}).encode())['data']['userByDisplayName']
|
|
livestream = user['livestream']
|
|
title = livestream['title']
|
|
username = user['username']
|
|
formats = self._extract_m3u8_formats(
|
|
'https://live.prd.dlive.tv/hls/live/%s.m3u8' % username,
|
|
display_name, 'mp4')
|
|
self._sort_formats(formats)
|
|
return {
|
|
'id': display_name,
|
|
'title': title,
|
|
'uploader': display_name,
|
|
'uploader_id': username,
|
|
'formats': formats,
|
|
'description': livestream.get('content'),
|
|
'thumbnail': livestream.get('thumbnailUrl'),
|
|
'is_live': True,
|
|
'timestamp': int_or_none(livestream.get('createdAt'), 1000),
|
|
'view_count': int_or_none(livestream.get('watchingCount')),
|
|
}
|