import re
from .common import InfoExtractor
-from ..compat import compat_urllib_request
from ..utils import (
int_or_none,
+ sanitized_Request,
str_to_int,
unescapeHTML,
unified_strdate,
'upload_date': '20101221',
'average_rating': int,
'view_count': int,
+ 'comment_count': int,
'categories': list,
'tags': list,
'age_limit': 18,
'upload_date': '20111125',
'average_rating': int,
'view_count': int,
+ 'comment_count': int,
'categories': list,
'tags': list,
'age_limit': 18,
video_id = mobj.group('id')
display_id = mobj.group('display_id')
- request = compat_urllib_request.Request(url)
+ request = sanitized_Request(url)
request.add_header('Cookie', 'age_verified=1')
webpage = self._download_webpage(request, display_id)
formats.append(f)
self._sort_formats(formats)
- description = self._html_search_regex(
- r'(?s)<div[^>]+class=["\']video-description["\'][^>]*>(.+?)</div>',
- webpage, 'description', default=None)
+ description = self._og_search_description(webpage, default=None)
thumbnail = self._search_regex(
r'(?:imageurl\s*=|poster\s*:)\s*(["\'])(?P<thumbnail>.+?)\1',
webpage, 'thumbnail', fatal=False, group='thumbnail')
uploader = self._html_search_regex(
- r'(?s)<div[^>]+class=["\']videoInfoBy["\'][^>]*>\s*By:\s*</div>(.+?)</(?:a|div)>',
+ r'(?s)<div[^>]+class=["\']videoInfoBy(?:\s+[^"\']+)?["\'][^>]*>\s*By:\s*</div>(.+?)</(?:a|div)>',
webpage, 'uploader', fatal=False)
upload_date = unified_strdate(self._html_search_regex(
r'(?s)<div[^>]+class=["\']videoInfoTime["\'][^>]*>(.+?)</div>',
view_count = str_to_int(self._search_regex(
r'(?s)<div[^>]+class=["\']videoInfoViews["\'][^>]*>.*?([\d,.]+)\s*</div>',
webpage, 'view count', fatal=False))
+ comment_count = str_to_int(self._search_regex(
+ r'>All [Cc]omments? \(([\d,.]+)\)',
+ webpage, 'comment count', fatal=False))
def extract_tag_box(title):
tag_box = self._search_regex(
'upload_date': upload_date,
'average_rating': average_rating,
'view_count': view_count,
+ 'comment_count': comment_count,
'categories': categories,
'tags': tags,
'age_limit': age_limit,