commit: 15ed5a27840e748d9f786c50b78a4c6326e9f186
parent 2e1280ed432257244ea52a47efe6a7f0e226b897
Author: Alexander Seiler <seileralex@gmail.com>
Date: Tue, 20 Nov 2018 20:50:40 +0100
[nzz] Relax kaltura regex
Diffstat:
1 file changed, 10 insertions(+), 3 deletions(-)
diff --git a/youtube_dl/extractor/nzz.py b/youtube_dl/extractor/nzz.py
@@ -11,20 +11,27 @@ from ..utils import (
class NZZIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?nzz\.ch/(?:[^/]+/)*[^/?#]+-ld\.(?P<id>\d+)'
- _TEST = {
+ _TESTS = [{
'url': 'http://www.nzz.ch/zuerich/gymizyte/gymizyte-schreiben-schueler-heute-noch-diktate-ld.9153',
'info_dict': {
'id': '9153',
},
'playlist_mincount': 6,
- }
+ }, {
+ 'url': 'https://www.nzz.ch/video/nzz-standpunkte/cvp-auf-der-suche-nach-dem-mass-der-mitte-ld.1368112',
+ 'info_dict': {
+ 'id': '1368112',
+ },
+ 'playlist_count': 1,
+ }]
def _real_extract(self, url):
page_id = self._match_id(url)
webpage = self._download_webpage(url, page_id)
entries = []
- for player_element in re.findall(r'(<[^>]+class="kalturaPlayer"[^>]*>)', webpage):
+ for player_element in re.findall(
+ r'(<[^>]+class="kalturaPlayer[^"]*"[^>]*>)', webpage):
player_params = extract_attributes(player_element)
if player_params.get('data-type') not in ('kaltura_singleArticle',):
self.report_warning('Unsupported player type')