mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-11-18 06:09:17 +00:00
51 lines
1.6 KiB
Python
51 lines
1.6 KiB
Python
# coding: utf-8
|
||
from __future__ import unicode_literals
|
||
|
||
from .common import InfoExtractor
|
||
|
||
|
||
class FootyRoomIE(InfoExtractor):
|
||
_VALID_URL = r'https?://footyroom\.com/(?P<id>[^/]+)'
|
||
_TESTS = [{
|
||
'url': 'http://footyroom.com/schalke-04-0-2-real-madrid-2015-02/',
|
||
'info_dict': {
|
||
'id': 'schalke-04-0-2-real-madrid-2015-02',
|
||
'title': 'Schalke 04 0 – 2 Real Madrid',
|
||
},
|
||
'playlist_count': 3,
|
||
'skip': 'Video for this match is not available',
|
||
}, {
|
||
'url': 'http://footyroom.com/georgia-0-2-germany-2015-03/',
|
||
'info_dict': {
|
||
'id': 'georgia-0-2-germany-2015-03',
|
||
'title': 'Georgia 0 – 2 Germany',
|
||
},
|
||
'playlist_count': 1,
|
||
}]
|
||
|
||
def _real_extract(self, url):
|
||
playlist_id = self._match_id(url)
|
||
|
||
webpage = self._download_webpage(url, playlist_id)
|
||
|
||
playlist = self._parse_json(
|
||
self._search_regex(
|
||
r'VideoSelector\.load\((\[.+?\])\);', webpage, 'video selector'),
|
||
playlist_id)
|
||
|
||
playlist_title = self._og_search_title(webpage)
|
||
|
||
entries = []
|
||
for video in playlist:
|
||
payload = video.get('payload')
|
||
if not payload:
|
||
continue
|
||
playwire_url = self._search_regex(
|
||
r'data-config="([^"]+)"', payload,
|
||
'playwire url', default=None)
|
||
if playwire_url:
|
||
entries.append(self.url_result(self._proto_relative_url(
|
||
playwire_url, 'http:'), 'Playwire'))
|
||
|
||
return self.playlist_result(entries, playlist_id, playlist_title)
|