-# encoding: utf-8
+# coding: utf-8
from __future__ import unicode_literals
+import re
+
from .common import InfoExtractor
from ..compat import compat_urlparse
from ..utils import (
webpage = self._download_webpage(url, video_id)
info = self._parse_json(self._search_regex(
- r'(?s)ntv.pageInfo.article =\s(\{.*?\});', webpage, 'info'),
+ r'(?s)ntv\.pageInfo\.article\s*=\s*(\{.*?\});', webpage, 'info'),
video_id, transform_source=js_to_json)
timestamp = int_or_none(info.get('publishedDateAsUnixTimeStamp'))
vdata = self._parse_json(self._search_regex(
r'(?s)\$\(\s*"\#player"\s*\)\s*\.data\(\s*"player",\s*(\{.*?\})\);',
- webpage, 'player data'),
- video_id, transform_source=js_to_json)
+ webpage, 'player data'), video_id,
+ transform_source=lambda s: js_to_json(re.sub(r'advertising:\s*{[^}]+},', '', s)))
duration = parse_duration(vdata.get('duration'))
formats = []