youtube_dlc/extractor/amcnetworks.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .theplatform import ThePlatformIE
   7 from ..utils import (
   8     int_or_none,
   9     parse_age_limit,
  10     try_get,
  11     update_url_query,
  12 )
  13
  14
  15 class AMCNetworksIE(ThePlatformIE):
  16     _VALID_URL = r'https?://(?:www\.)?(?P<site>amc|bbcamerica|ifc|(?:we|sundance)tv)\.com/(?P<id>(?:movies|shows(?:/[^/]+)+)/[^/?#&]+)'
  17     _TESTS = [{
  18         'url': 'https://www.bbcamerica.com/shows/the-graham-norton-show/videos/tina-feys-adorable-airline-themed-family-dinner--51631',
  19         'info_dict': {
  20             'id': '4Lq1dzOnZGt0',
  21             'ext': 'mp4',
  22             'title': "The Graham Norton Show - Season 28 - Tina Fey's Adorable Airline-Themed Family Dinner",
  23             'description': "It turns out child stewardesses are very generous with the wine! All-new episodes of 'The Graham Norton Show' premiere Fridays at 11/10c on BBC America.",
  24             'upload_date': '20201120',
  25             'timestamp': 1605904350,
  26             'uploader': 'AMCN',
  27         },
  28         'params': {
  29             # m3u8 download
  30             'skip_download': True,
  31         },
  32     }, {
  33         'url': 'http://www.bbcamerica.com/shows/the-hunt/full-episodes/season-1/episode-01-the-hardest-challenge',
  34         'only_matching': True,
  35     }, {
  36         'url': 'http://www.amc.com/shows/preacher/full-episodes/season-01/episode-00/pilot',
  37         'only_matching': True,
  38     }, {
  39         'url': 'http://www.wetv.com/shows/million-dollar-matchmaker/season-01/episode-06-the-dumped-dj-and-shallow-hal',
  40         'only_matching': True,
  41     }, {
  42         'url': 'http://www.ifc.com/movies/chaos',
  43         'only_matching': True,
  44     }, {
  45         'url': 'http://www.bbcamerica.com/shows/doctor-who/full-episodes/the-power-of-the-daleks/episode-01-episode-1-color-version',
  46         'only_matching': True,
  47     }, {
  48         'url': 'http://www.wetv.com/shows/mama-june-from-not-to-hot/full-episode/season-01/thin-tervention',
  49         'only_matching': True,
  50     }, {
  51         'url': 'http://www.wetv.com/shows/la-hair/videos/season-05/episode-09-episode-9-2/episode-9-sneak-peek-3',
  52         'only_matching': True,
  53     }, {
  54         'url': 'https://www.sundancetv.com/shows/riviera/full-episodes/season-1/episode-01-episode-1',
  55         'only_matching': True,
  56     }]
  57     _REQUESTOR_ID_MAP = {
  58         'amc': 'AMC',
  59         'bbcamerica': 'BBCA',
  60         'ifc': 'IFC',
  61         'sundancetv': 'SUNDANCE',
  62         'wetv': 'WETV',
  63     }
  64
  65     def _real_extract(self, url):
  66         site, display_id = re.match(self._VALID_URL, url).groups()
  67         requestor_id = self._REQUESTOR_ID_MAP[site]
  68         properties = self._download_json(
  69             'https://content-delivery-gw.svc.ds.amcn.com/api/v2/content/amcn/%s/url/%s' % (requestor_id.lower(), display_id),
  70             display_id)['data']['properties']
  71         query = {
  72             'mbr': 'true',
  73             'manifest': 'm3u',
  74         }
  75         tp_path = 'M_UwQC/media/' + properties['videoPid']
  76         media_url = 'https://link.theplatform.com/s/' + tp_path
  77         theplatform_metadata = self._download_theplatform_metadata(tp_path, display_id)
  78         info = self._parse_theplatform_metadata(theplatform_metadata)
  79         video_id = theplatform_metadata['pid']
  80         title = theplatform_metadata['title']
  81         rating = try_get(
  82             theplatform_metadata, lambda x: x['ratings'][0]['rating'])
  83         video_category = properties.get('videoCategory')
  84         if video_category and video_category.endswith('-Auth'):
  85             resource = self._get_mvpd_resource(
  86                 requestor_id, title, video_id, rating)
  87             query['auth'] = self._extract_mvpd_auth(
  88                 url, video_id, requestor_id, resource)
  89         media_url = update_url_query(media_url, query)
  90         formats, subtitles = self._extract_theplatform_smil(
  91             media_url, video_id)
  92         self._sort_formats(formats)
  93         info.update({
  94             'id': video_id,
  95             'subtitles': subtitles,
  96             'formats': formats,
  97             'age_limit': parse_age_limit(parse_age_limit(rating)),
  98         })
  99         ns_keys = theplatform_metadata.get('$xmlns', {}).keys()
 100         if ns_keys:
 101             ns = list(ns_keys)[0]
 102             series = theplatform_metadata.get(ns + '$show')
 103             season_number = int_or_none(
 104                 theplatform_metadata.get(ns + '$season'))
 105             episode = theplatform_metadata.get(ns + '$episodeTitle')
 106             episode_number = int_or_none(
 107                 theplatform_metadata.get(ns + '$episode'))
 108             if season_number:
 109                 title = 'Season %d - %s' % (season_number, title)
 110             if series:
 111                 title = '%s - %s' % (series, title)
 112             info.update({
 113                 'title': title,
 114                 'series': series,
 115                 'season_number': season_number,
 116                 'episode': episode,
 117                 'episode_number': episode_number,
 118             })
 119         return info