[bbc] Add support for videos in news articles embedded with data-playable

This commit is contained in:
Sergey M․ 2015-10-10 20:34:06 +06:00
parent 6a959f2e52
commit dab062fb6e

View file

@ -11,6 +11,7 @@
int_or_none, int_or_none,
parse_duration, parse_duration,
parse_iso8601, parse_iso8601,
unescapeHTML,
) )
from ..compat import compat_HTTPError from ..compat import compat_HTTPError
@ -682,6 +683,21 @@ def _real_extract(self, url):
[r'data-video-player-vpid="([\da-z]{8})"', [r'data-video-player-vpid="([\da-z]{8})"',
r'<param[^>]+name="externalIdentifier"[^>]+value="([\da-z]{8})"'], r'<param[^>]+name="externalIdentifier"[^>]+value="([\da-z]{8})"'],
webpage, 'vpid', default=None) webpage, 'vpid', default=None)
duration = None
if not programme_id:
# single video in news article embedded with data-playable (e.g.
# http://www.bbc.com/news/world-us-canada-34473351)
data_playable = self._parse_json(
unescapeHTML(self._search_regex(
r'data-playable="({.+?})"', webpage, 'data playable', default='{}')),
programme_id, fatal=False)
if data_playable:
items = data_playable.get('settings', {}).get('playlistObject', {}).get('items')
if items and isinstance(items, list):
duration = int_or_none(items[0].get('duration'))
programme_id = items[0].get('vpid')
if programme_id: if programme_id:
formats, subtitles = self._download_media_selector(programme_id) formats, subtitles = self._download_media_selector(programme_id)
self._sort_formats(formats) self._sort_formats(formats)
@ -699,6 +715,7 @@ def _real_extract(self, url):
'title': title, 'title': title,
'description': description, 'description': description,
'timestamp': timestamp, 'timestamp': timestamp,
'duration': duration,
'formats': formats, 'formats': formats,
'subtitles': subtitles, 'subtitles': subtitles,
} }