[pbs] Improve multipart video support (closes #12981)

This commit is contained in:
Sergey M․ 2017-05-04 22:42:49 +07:00
parent 7f09e523e8
commit 4947f13cd0
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D

View file

@ -10,6 +10,7 @@
int_or_none,
float_or_none,
js_to_json,
orderedSet,
strip_jsonp,
strip_or_none,
unified_strdate,
@ -264,6 +265,13 @@ class PBSIE(InfoExtractor):
},
'playlist_count': 2,
},
{
'url': 'http://www.pbs.org/wgbh/americanexperience/films/great-war/',
'info_dict': {
'id': 'great-war',
},
'playlist_count': 3,
},
{
'url': 'http://www.pbs.org/wgbh/americanexperience/films/death/player/',
'info_dict': {
@ -382,10 +390,10 @@ def _extract_webpage(self, url):
# tabbed frontline videos
MULTI_PART_REGEXES = (
r'<div[^>]+class="videotab[^"]*"[^>]+vid="(\d+)"',
r'<a[^>]+href=["\']#video-\d+["\'][^>]+data-coveid=["\'](\d+)',
r'<a[^>]+href=["\']#(?:video-|part)\d+["\'][^>]+data-cove[Ii]d=["\'](\d+)',
)
for p in MULTI_PART_REGEXES:
tabbed_videos = re.findall(p, webpage)
tabbed_videos = orderedSet(re.findall(p, webpage))
if tabbed_videos:
return tabbed_videos, presumptive_id, upload_date, description