[svtplay] Improve extraction (closes #28507, closes #28876)

This commit is contained in:
Sergey M․ 2021-05-01 18:09:32 +07:00
parent e33dfb445c
commit d2f72c40db
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -146,7 +146,7 @@ class SVTPlayIE(SVTPlayBaseIE):
) )
(?P<svt_id>[^/?#&]+)| (?P<svt_id>[^/?#&]+)|
https?://(?:www\.)?(?:svtplay|oppetarkiv)\.se/(?:video|klipp|kanaler)/(?P<id>[^/?#&]+) https?://(?:www\.)?(?:svtplay|oppetarkiv)\.se/(?:video|klipp|kanaler)/(?P<id>[^/?#&]+)
(?:.*?modalId=(?P<modal_id>[\da-zA-Z-]+))? (?:.*?(?:modalId|id)=(?P<modal_id>[\da-zA-Z-]+))?
) )
''' '''
_TESTS = [{ _TESTS = [{
@ -177,6 +177,9 @@ class SVTPlayIE(SVTPlayBaseIE):
}, { }, {
'url': 'https://www.svtplay.se/video/30479064/husdrommar/husdrommar-sasong-8-designdrommar-i-stenungsund?modalId=8zVbDPA', 'url': 'https://www.svtplay.se/video/30479064/husdrommar/husdrommar-sasong-8-designdrommar-i-stenungsund?modalId=8zVbDPA',
'only_matching': True, 'only_matching': True,
}, {
'url': 'https://www.svtplay.se/video/30684086/rapport/rapport-24-apr-18-00-7?id=e72gVpa',
'only_matching': True,
}, { }, {
# geo restricted to Sweden # geo restricted to Sweden
'url': 'http://www.oppetarkiv.se/video/5219710/trollflojten', 'url': 'http://www.oppetarkiv.se/video/5219710/trollflojten',
@ -259,7 +262,7 @@ class SVTPlayIE(SVTPlayBaseIE):
if not svt_id: if not svt_id:
svt_id = self._search_regex( svt_id = self._search_regex(
(r'<video[^>]+data-video-id=["\']([\da-zA-Z-]+)', (r'<video[^>]+data-video-id=["\']([\da-zA-Z-]+)',
r'<[^>]+\bdata-rt=["\']top-area-play-button["\'][^>]+\bhref=["\'][^"\']*video/%s/[^"\']*\bmodalId=([\da-zA-Z-]+)' % re.escape(video_id), r'<[^>]+\bdata-rt=["\']top-area-play-button["\'][^>]+\bhref=["\'][^"\']*video/%s/[^"\']*\b(?:modalId|id)=([\da-zA-Z-]+)' % re.escape(video_id),
r'["\']videoSvtId["\']\s*:\s*["\']([\da-zA-Z-]+)', r'["\']videoSvtId["\']\s*:\s*["\']([\da-zA-Z-]+)',
r'["\']videoSvtId\\?["\']\s*:\s*\\?["\']([\da-zA-Z-]+)', r'["\']videoSvtId\\?["\']\s*:\s*\\?["\']([\da-zA-Z-]+)',
r'"content"\s*:\s*{.*?"id"\s*:\s*"([\da-zA-Z-]+)"', r'"content"\s*:\s*{.*?"id"\s*:\s*"([\da-zA-Z-]+)"',