1
1
mirror of https://github.com/ytdl-org/youtube-dl synced 2024-12-15 03:47:48 +01:00
youtube-dl/youtube_dl/extractor/sbs.py

52 lines
1.8 KiB
Python
Raw Normal View History

2014-08-23 15:20:49 +02:00
# -*- coding: utf-8 -*-
from __future__ import unicode_literals
from .common import InfoExtractor
class SBSIE(InfoExtractor):
IE_DESC = 'sbs.com.au'
2015-07-17 22:43:18 +02:00
_VALID_URL = r'https?://(?:www\.)?sbs\.com\.au/(?:ondemand|news)/video/(?:single/)?(?P<id>[0-9]+)'
2014-08-23 15:20:49 +02:00
_TESTS = [{
# Original URL is handled by the generic IE which finds the iframe:
# http://www.sbs.com.au/thefeed/blog/2014/08/21/dingo-conservation
'url': 'http://www.sbs.com.au/ondemand/video/single/320403011771/?source=drupal&vertical=thefeed',
'md5': '3150cf278965eeabb5b4cea1c963fe0a',
'info_dict': {
'id': '320403011771',
'ext': 'mp4',
2015-07-17 22:43:18 +02:00
'title': 'Dingo Conservation (The Feed)',
'description': 'md5:f250a9856fca50d22dec0b5b8015f8a5',
2014-08-23 15:20:49 +02:00
'thumbnail': 're:http://.*\.jpg',
2015-07-17 22:43:18 +02:00
'duration': 308,
2014-08-23 15:20:49 +02:00
},
2014-11-23 21:39:15 +01:00
}, {
'url': 'http://www.sbs.com.au/ondemand/video/320403011771/Dingo-Conservation-The-Feed',
'only_matching': True,
2015-07-17 22:43:18 +02:00
}, {
'url': 'http://www.sbs.com.au/news/video/471395907773/The-Feed-July-9',
'only_matching': True,
2014-08-23 15:20:49 +02:00
}]
def _real_extract(self, url):
2015-05-16 17:07:29 +02:00
video_id = self._match_id(url)
2015-07-17 22:43:18 +02:00
webpage = self._download_webpage(
'http://www.sbs.com.au/ondemand/video/single/%s?context=web' % video_id, video_id)
2015-05-16 17:07:29 +02:00
2015-07-17 22:43:18 +02:00
player_params = self._parse_json(
self._search_regex(
r'(?s)var\s+playerParams\s*=\s*({.+?});', webpage, 'playerParams'),
video_id)
2014-08-23 15:20:49 +02:00
2015-07-17 22:43:18 +02:00
urls = player_params['releaseUrls']
theplatform_url = (urls.get('progressive') or urls.get('standard') or
urls.get('html') or player_params['relatedItemsURL'])
2014-08-23 15:20:49 +02:00
return {
'_type': 'url_transparent',
'id': video_id,
'url': theplatform_url,
}