)
from .openload import PhantomJSwrapper
from ..utils import (
+ determine_ext,
ExtractorError,
int_or_none,
- js_to_json,
orderedSet,
remove_quotes,
str_to_int,
def dl_webpage(platform):
self._set_cookie(host, 'platform', platform)
return self._download_webpage(
- 'http://www.%s/view_video.php?viewkey=%s' % (host, video_id),
+ 'https://www.%s/view_video.php?viewkey=%s' % (host, video_id),
video_id, 'Downloading %s webpage' % platform)
webpage = dl_webpage('pc')
r'/(\d{6}/\d{2})/', video_url, 'upload data', default=None)
if upload_date:
upload_date = upload_date.replace('/', '')
+ if determine_ext(video_url) == 'mpd':
+ formats.extend(self._extract_mpd_formats(
+ video_url, video_id, mpd_id='dash', fatal=False))
+ continue
tbr = None
mobj = re.search(r'(?P<height>\d+)[pP]?_(?P<tbr>\d+)[kK]', video_url)
if mobj:
comment_count = self._extract_count(
r'All Comments\s*<span>\(([\d,.]+)\)', webpage, 'comment')
- page_params = self._parse_json(self._search_regex(
- r'page_params\.zoneDetails\[([\'"])[^\'"]+\1\]\s*=\s*(?P<data>{[^}]+})',
- webpage, 'page parameters', group='data', default='{}'),
- video_id, transform_source=js_to_json, fatal=False)
- tags = categories = None
- if page_params:
- tags = page_params.get('tags', '').split(',')
- categories = page_params.get('categories', '').split(',')
+ def extract_list(meta_key):
+ div = self._search_regex(
+ r'(?s)<div[^>]+\bclass=["\'].*?\b%sWrapper[^>]*>(.+?)</div>'
+ % meta_key, webpage, meta_key, default=None)
+ if div:
+ return re.findall(r'<a[^>]+\bhref=[^>]+>([^<]+)', div)
return {
'id': video_id,
'comment_count': comment_count,
'formats': formats,
'age_limit': 18,
- 'tags': tags,
- 'categories': categories,
+ 'tags': extract_list('tags'),
+ 'categories': extract_list('categories'),
'subtitles': subtitles,
}
def _extract_entries(self, webpage, host):
# Only process container div with main playlist content skipping
# drop-down menu that uses similar pattern for videos (see
- # https://github.com/rg3/youtube-dl/issues/11594).
+ # https://github.com/ytdl-org/youtube-dl/issues/11594).
container = self._search_regex(
r'(?s)(<div[^>]+class=["\']container.+)', webpage,
'container', default=webpage)