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'''(?x)
  14         https?://(?:www\.)?sbs\.com\.au/(?:
  15             ondemand(?:
  16                 /video/(?:single/)?|
  17                 /movie/[^/]+/|
  18                 .*?\bplay=|/watch/
  19             )|news/(?:embeds/)?video/
  20         )(?P<id>[0-9]+)'''
  21
  22     _TESTS = [{
  23         # Original URL is handled by the generic IE which finds the iframe:
  24         # http://www.sbs.com.au/thefeed/blog/2014/08/21/dingo-conservation
  25         'url': 'http://www.sbs.com.au/ondemand/video/single/320403011771/?source=drupal&vertical=thefeed',
  26         'md5': '3150cf278965eeabb5b4cea1c963fe0a',
  27         'info_dict': {
  28             'id': '_rFBPRPO4pMR',
  29             'ext': 'mp4',
  30             'title': 'Dingo Conservation (The Feed)',
  31             'description': 'md5:f250a9856fca50d22dec0b5b8015f8a5',
  32             'thumbnail': r're:http://.*\.jpg',
  33             'duration': 308,
  34             'timestamp': 1408613220,
  35             'upload_date': '20140821',
  36             'uploader': 'SBSC',
  37         },
  38     }, {
  39         'url': 'http://www.sbs.com.au/ondemand/video/320403011771/Dingo-Conservation-The-Feed',
  40         'only_matching': True,
  41     }, {
  42         'url': 'http://www.sbs.com.au/news/video/471395907773/The-Feed-July-9',
  43         'only_matching': True,
  44     }, {
  45         'url': 'https://www.sbs.com.au/ondemand/?play=1836638787723',
  46         'only_matching': True,
  47     }, {
  48         'url': 'https://www.sbs.com.au/ondemand/program/inside-windsor-castle?play=1283505731842',
  49         'only_matching': True,
  50     }, {
  51         'url': 'https://www.sbs.com.au/news/embeds/video/1840778819866',
  52         'only_matching': True,
  53     }, {
  54         'url': 'https://www.sbs.com.au/ondemand/watch/1698704451971',
  55         'only_matching': True,
  56     }, {
  57         'url': 'https://www.sbs.com.au/ondemand/movie/coherence/1469404227931',
  58         'only_matching': True,
  59     }, {
  60         'note': 'Live stream',
  61         'url': 'https://www.sbs.com.au/ondemand/video/1726824003663/sbs-24x7-live-stream-nsw',
  62         'only_matching': True,
  63     }]
  64
  65     def _real_extract(self, url):
  66         video_id = self._match_id(url)
  67         player_params = self._download_json(
  68             'http://www.sbs.com.au/api/video_pdkvars/id/%s?form=json' % video_id, video_id)
  69
  70         error = player_params.get('error')
  71         if error:
  72             error_message = 'Sorry, The video you are looking for does not exist.'
  73             video_data = error.get('results') or {}
  74             error_code = error.get('errorCode')
  75             if error_code == 'ComingSoon':
  76                 error_message = '%s is not yet available.' % video_data.get('title', '')
  77             elif error_code in ('Forbidden', 'intranetAccessOnly'):
  78                 error_message = 'Sorry, This video cannot be accessed via this website'
  79             elif error_code == 'Expired':
  80                 error_message = 'Sorry, %s is no longer available.' % video_data.get('title', '')
  81             raise ExtractorError('%s said: %s' % (self.IE_NAME, error_message), expected=True)
  82
  83         urls = player_params['releaseUrls']
  84         theplatform_url = (urls.get('progressive') or urls.get('html')
  85                            or urls.get('standard') or player_params['relatedItemsURL'])
  86
  87         return {
  88             '_type': 'url_transparent',
  89             'ie_key': 'ThePlatform',
  90             'id': video_id,
  91             'url': smuggle_url(self._proto_relative_url(theplatform_url), {'force_smil_url': True}),
  92             'is_live': player_params.get('streamType') == 'live',
  93         }