yt_dlp/extractor/sbs.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     smuggle_url,
   7     ExtractorError,
   8 )
   9
  10
  11 class SBSIE(InfoExtractor):
  12     IE_DESC = 'sbs.com.au'
  13     _VALID_URL = r'https?://(?:www\.)?sbs\.com\.au/(?:ondemand(?:/video/(?:single/)?|.*?\bplay=|/watch/)|news/(?:embeds/)?video/)(?P<id>[0-9]+)'
  14
  15     _TESTS = [{
  16         # Original URL is handled by the generic IE which finds the iframe:
  17         # http://www.sbs.com.au/thefeed/blog/2014/08/21/dingo-conservation
  18         'url': 'http://www.sbs.com.au/ondemand/video/single/320403011771/?source=drupal&vertical=thefeed',
  19         'md5': '3150cf278965eeabb5b4cea1c963fe0a',
  20         'info_dict': {
  21             'id': '_rFBPRPO4pMR',
  22             'ext': 'mp4',
  23             'title': 'Dingo Conservation (The Feed)',
  24             'description': 'md5:f250a9856fca50d22dec0b5b8015f8a5',
  25             'thumbnail': r're:http://.*\.jpg',
  26             'duration': 308,
  27             'timestamp': 1408613220,
  28             'upload_date': '20140821',
  29             'uploader': 'SBSC',
  30         },
  31     }, {
  32         'url': 'http://www.sbs.com.au/ondemand/video/320403011771/Dingo-Conservation-The-Feed',
  33         'only_matching': True,
  34     }, {
  35         'url': 'http://www.sbs.com.au/news/video/471395907773/The-Feed-July-9',
  36         'only_matching': True,
  37     }, {
  38         'url': 'https://www.sbs.com.au/ondemand/?play=1836638787723',
  39         'only_matching': True,
  40     }, {
  41         'url': 'https://www.sbs.com.au/ondemand/program/inside-windsor-castle?play=1283505731842',
  42         'only_matching': True,
  43     }, {
  44         'url': 'https://www.sbs.com.au/news/embeds/video/1840778819866',
  45         'only_matching': True,
  46     }, {
  47         'url': 'https://www.sbs.com.au/ondemand/watch/1698704451971',
  48         'only_matching': True,
  49     }]
  50
  51     def _real_extract(self, url):
  52         video_id = self._match_id(url)
  53         player_params = self._download_json(
  54             'http://www.sbs.com.au/api/video_pdkvars/id/%s?form=json' % video_id, video_id)
  55
  56         error = player_params.get('error')
  57         if error:
  58             error_message = 'Sorry, The video you are looking for does not exist.'
  59             video_data = error.get('results') or {}
  60             error_code = error.get('errorCode')
  61             if error_code == 'ComingSoon':
  62                 error_message = '%s is not yet available.' % video_data.get('title', '')
  63             elif error_code in ('Forbidden', 'intranetAccessOnly'):
  64                 error_message = 'Sorry, This video cannot be accessed via this website'
  65             elif error_code == 'Expired':
  66                 error_message = 'Sorry, %s is no longer available.' % video_data.get('title', '')
  67             raise ExtractorError('%s said: %s' % (self.IE_NAME, error_message), expected=True)
  68
  69         urls = player_params['releaseUrls']
  70         theplatform_url = (urls.get('progressive') or urls.get('html')
  71                            or urls.get('standard') or player_params['relatedItemsURL'])
  72
  73         return {
  74             '_type': 'url_transparent',
  75             'ie_key': 'ThePlatform',
  76             'id': video_id,
  77             'url': smuggle_url(self._proto_relative_url(theplatform_url), {'force_smil_url': True}),
  78         }