0
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-30 03:33:02 +00:00
yt-dlp/youtube_dl/extractor/azmedien.py
2018-11-03 01:33:42 +07:00

90 lines
3 KiB
Python

# coding: utf-8
from __future__ import unicode_literals
import json
import re
from .common import InfoExtractor
from .kaltura import KalturaIE
class AZMedienIE(InfoExtractor):
IE_DESC = 'AZ Medien videos'
_VALID_URL = r'''(?x)
https?://
(?:www\.)?
(?P<host>
telezueri\.ch|
telebaern\.tv|
telem1\.ch
)/
[^/]+/
(?P<id>
[^/]+-(?P<article_id>\d+)
)
(?:
\#video=
(?P<kaltura_id>
[_0-9a-z]+
)
)?
'''
_TESTS = [{
'url': 'https://www.telezueri.ch/sonntalk/bundesrats-vakanzen-eu-rahmenabkommen-133214569',
'info_dict': {
'id': '1_anruz3wy',
'ext': 'mp4',
'title': 'Bundesrats-Vakanzen / EU-Rahmenabkommen',
'description': 'md5:dd9f96751ec9c35e409a698a328402f3',
'uploader_id': 'TVOnline',
'upload_date': '20180930',
'timestamp': 1538328802,
},
'params': {
'skip_download': True,
},
}, {
'url': 'https://www.telebaern.tv/telebaern-news/montag-1-oktober-2018-ganze-sendung-133531189#video=0_7xjo9lf1',
'only_matching': True
}]
_PARTNER_ID = '1719221'
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
entry_id = mobj.group('kaltura_id')
if not entry_id:
webpage = self._download_webpage(url, video_id)
api_path = self._search_regex(
r'["\']apiPath["\']\s*:\s*["\']([^"^\']+)["\']',
webpage, 'api path')
api_url = 'https://www.%s%s' % (mobj.group('host'), api_path)
payload = {
'query': '''query VideoContext($articleId: ID!) {
article: node(id: $articleId) {
... on Article {
mainAssetRelation {
asset {
... on VideoAsset {
kalturaId
}
}
}
}
}
}''',
'variables': {'articleId': 'Article:%s' % mobj.group('article_id')},
}
json_data = self._download_json(
api_url, video_id, headers={
'Content-Type': 'application/json',
},
data=json.dumps(payload).encode())
entry_id = json_data['data']['article']['mainAssetRelation']['asset']['kalturaId']
return self.url_result(
'kaltura:%s:%s' % (self._PARTNER_ID, entry_id),
ie=KalturaIE.ie_key(), video_id=entry_id)