[wdr] Relax media link regex (closes #14447)

This commit is contained in:
Sergey M․ 2017-10-08 21:36:50 +07:00
parent b0dde6686c
commit 8992331621
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D

View file

@ -22,8 +22,13 @@ def _extract_wdr_video(self, webpage, display_id):
# for wdrmaus, in a tag with the class "videoButton" (previously a link # for wdrmaus, in a tag with the class "videoButton" (previously a link
# to the page in a multiline "videoLink"-tag) # to the page in a multiline "videoLink"-tag)
json_metadata = self._html_search_regex( json_metadata = self._html_search_regex(
r'class=(?:"(?:mediaLink|wdrrPlayerPlayBtn|videoButton)\b[^"]*"[^>]+|"videoLink\b[^"]*"[\s]*>\n[^\n]*)data-extension="([^"]+)"', r'''(?sx)class=
webpage, 'media link', default=None, flags=re.MULTILINE) (?:
(["\'])(?:mediaLink|wdrrPlayerPlayBtn|videoButton)\b.*?\1[^>]+|
(["\'])videoLink\b.*?\2[\s]*>\n[^\n]*
)data-extension=(["\'])(?P<data>(?:(?!\3).)+)\3
''',
webpage, 'media link', default=None, group='data')
if not json_metadata: if not json_metadata:
return return