Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/amcnetworks.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .theplatform import ThePlatformIE
   5 from ..utils import (
   6     int_or_none,
   7     parse_age_limit,
   8     try_get,
   9     update_url_query,
  10 )
  11
  12
  13 class AMCNetworksIE(ThePlatformIE):
  14     _VALID_URL = r'https?://(?:www\.)?(?:amc|bbcamerica|ifc|(?:we|sundance)tv)\.com/(?:movies|shows(?:/[^/]+)+)/(?P<id>[^/?#]+)'
  15     _TESTS = [{
  16         'url': 'http://www.ifc.com/shows/maron/season-04/episode-01/step-1',
  17         'md5': '',
  18         'info_dict': {
  19             'id': 's3MX01Nl4vPH',
  20             'ext': 'mp4',
  21             'title': 'Maron - Season 4 - Step 1',
  22             'description': 'In denial about his current situation, Marc is reluctantly convinced by his friends to enter rehab. Starring Marc Maron and Constance Zimmer.',
  23             'age_limit': 17,
  24             'upload_date': '20160505',
  25             'timestamp': 1462468831,
  26             'uploader': 'AMCN',
  27         },
  28         'params': {
  29             # m3u8 download
  30             'skip_download': True,
  31         },
  32         'skip': 'Requires TV provider accounts',
  33     }, {
  34         'url': 'http://www.bbcamerica.com/shows/the-hunt/full-episodes/season-1/episode-01-the-hardest-challenge',
  35         'only_matching': True,
  36     }, {
  37         'url': 'http://www.amc.com/shows/preacher/full-episodes/season-01/episode-00/pilot',
  38         'only_matching': True,
  39     }, {
  40         'url': 'http://www.wetv.com/shows/million-dollar-matchmaker/season-01/episode-06-the-dumped-dj-and-shallow-hal',
  41         'only_matching': True,
  42     }, {
  43         'url': 'http://www.ifc.com/movies/chaos',
  44         'only_matching': True,
  45     }, {
  46         'url': 'http://www.bbcamerica.com/shows/doctor-who/full-episodes/the-power-of-the-daleks/episode-01-episode-1-color-version',
  47         'only_matching': True,
  48     }, {
  49         'url': 'http://www.wetv.com/shows/mama-june-from-not-to-hot/full-episode/season-01/thin-tervention',
  50         'only_matching': True,
  51     }, {
  52         'url': 'http://www.wetv.com/shows/la-hair/videos/season-05/episode-09-episode-9-2/episode-9-sneak-peek-3',
  53         'only_matching': True,
  54     }, {
  55         'url': 'https://www.sundancetv.com/shows/riviera/full-episodes/season-1/episode-01-episode-1',
  56         'only_matching': True,
  57     }]
  58
  59     def _real_extract(self, url):
  60         display_id = self._match_id(url)
  61         webpage = self._download_webpage(url, display_id)
  62         query = {
  63             'mbr': 'true',
  64             'manifest': 'm3u',
  65         }
  66         media_url = self._search_regex(
  67             r'window\.platformLinkURL\s*=\s*[\'"]([^\'"]+)',
  68             webpage, 'media url')
  69         theplatform_metadata = self._download_theplatform_metadata(self._search_regex(
  70             r'link\.theplatform\.com/s/([^?]+)',
  71             media_url, 'theplatform_path'), display_id)
  72         info = self._parse_theplatform_metadata(theplatform_metadata)
  73         video_id = theplatform_metadata['pid']
  74         title = theplatform_metadata['title']
  75         rating = try_get(
  76             theplatform_metadata, lambda x: x['ratings'][0]['rating'])
  77         auth_required = self._search_regex(
  78             r'window\.authRequired\s*=\s*(true|false);',
  79             webpage, 'auth required')
  80         if auth_required == 'true':
  81             requestor_id = self._search_regex(
  82                 r'window\.requestor_id\s*=\s*[\'"]([^\'"]+)',
  83                 webpage, 'requestor id')
  84             resource = self._get_mvpd_resource(
  85                 requestor_id, title, video_id, rating)
  86             query['auth'] = self._extract_mvpd_auth(
  87                 url, video_id, requestor_id, resource)
  88         media_url = update_url_query(media_url, query)
  89         formats, subtitles = self._extract_theplatform_smil(
  90             media_url, video_id)
  91         self._sort_formats(formats)
  92         info.update({
  93             'id': video_id,
  94             'subtitles': subtitles,
  95             'formats': formats,
  96             'age_limit': parse_age_limit(parse_age_limit(rating)),
  97         })
  98         ns_keys = theplatform_metadata.get('$xmlns', {}).keys()
  99         if ns_keys:
 100             ns = list(ns_keys)[0]
 101             series = theplatform_metadata.get(ns + '$show')
 102             season_number = int_or_none(
 103                 theplatform_metadata.get(ns + '$season'))
 104             episode = theplatform_metadata.get(ns + '$episodeTitle')
 105             episode_number = int_or_none(
 106                 theplatform_metadata.get(ns + '$episode'))
 107             if season_number:
 108                 title = 'Season %d - %s' % (season_number, title)
 109             if series:
 110                 title = '%s - %s' % (series, title)
 111             info.update({
 112                 'title': title,
 113                 'series': series,
 114                 'season_number': season_number,
 115                 'episode': episode,
 116                 'episode_number': episode_number,
 117             })
 118         return info