[nrktv:episode] Improve video id extraction (closes #25594, closes #26369, closes #26409)

This commit is contained in:
Sergey M․ 2020-09-06 12:42:57 +07:00
parent 16ee69c1b7
commit 50e9fcc1fd
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D
1 changed files with 13 additions and 7 deletions

View File

@ -11,7 +11,6 @@ from ..compat import (
from ..utils import ( from ..utils import (
ExtractorError, ExtractorError,
int_or_none, int_or_none,
JSON_LD_RE,
js_to_json, js_to_json,
NO_DEFAULT, NO_DEFAULT,
parse_age_limit, parse_age_limit,
@ -425,13 +424,20 @@ class NRKTVEpisodeIE(InfoExtractor):
webpage = self._download_webpage(url, display_id) webpage = self._download_webpage(url, display_id)
nrk_id = self._parse_json( info = self._search_json_ld(webpage, display_id, default={})
self._search_regex(JSON_LD_RE, webpage, 'JSON-LD', group='json_ld'), nrk_id = info.get('@id') or self._html_search_meta(
display_id)['@id'] 'nrk:program-id', webpage, default=None) or self._search_regex(
r'data-program-id=["\'](%s)' % NRKTVIE._EPISODE_RE, webpage,
'nrk id')
assert re.match(NRKTVIE._EPISODE_RE, nrk_id) assert re.match(NRKTVIE._EPISODE_RE, nrk_id)
return self.url_result(
'nrk:%s' % nrk_id, ie=NRKIE.ie_key(), video_id=nrk_id) info.update({
'_type': 'url_transparent',
'id': nrk_id,
'url': 'nrk:%s' % nrk_id,
'ie_key': NRKIE.ie_key(),
})
return info
class NRKTVSerieBaseIE(InfoExtractor): class NRKTVSerieBaseIE(InfoExtractor):