logo

youtube-dl

[mirror] Download/Watch videos from video hostersgit clone https://hacktivis.me/git/mirror/youtube-dl.git

s4c.py (4496B)

    

  1. # coding: utf-8
  3. from __future__ import unicode_literals
  5. from functools import partial as partial_f
  7. from .common import InfoExtractor
  8. from ..utils import (
  9.     float_or_none,
  10.     merge_dicts,
  11.     T,
  12.     traverse_obj,
  13.     txt_or_none,
  14.     url_or_none,
  15. )
  18. class S4CIE(InfoExtractor):
  19.     _VALID_URL = r'https?://(?:www\.)?s4c\.cymru/clic/programme/(?P<id>\d+)'
  20.     _TESTS = [{
  21.         'url': 'https://www.s4c.cymru/clic/programme/861362209',
  22.         'info_dict': {
  23.             'id': '861362209',
  24.             'ext': 'mp4',
  25.             'title': 'Y Swn',
  26.             'description': 'md5:f7681a30e4955b250b3224aa9fe70cf0',
  27.             'duration': 5340,
  28.             'thumbnail': 'https://www.s4c.cymru/amg/1920x1080/Y_Swn_2023S4C_099_ii.jpg',
  29.         },
  30.     }, {
  31.         'url': 'https://www.s4c.cymru/clic/programme/856636948',
  32.         'info_dict': {
  33.             'id': '856636948',
  34.             'ext': 'mp4',
  35.             'title': 'Am Dro',
  36.             'duration': 2880,
  37.             'description': 'md5:100d8686fc9a632a0cb2db52a3433ffe',
  38.             'thumbnail': 'https://www.s4c.cymru/amg/1920x1080/Am_Dro_2022-23S4C_P6_4005.jpg',
  39.         },
  40.     }]
  42.     def _real_extract(self, url):
  43.         video_id = self._match_id(url)
  44.         details = self._download_json(
  45.             'https://www.s4c.cymru/df/full_prog_details',
  46.             video_id, query={
  47.                 'lang': 'e',
  48.                 'programme_id': video_id,
  49.             }, fatal=False)
  51.         player_config = self._download_json(
  52.             'https://player-api.s4c-cdn.co.uk/player-configuration/prod', video_id, query={
  53.                 'programme_id': video_id,
  54.                 'signed': '0',
  55.                 'lang': 'en',
  56.                 'mode': 'od',
  57.                 'appId': 'clic',
  58.                 'streamName': '',
  59.             }, note='Downloading player config JSON')
  61.         m3u8_url = self._download_json(
  62.             'https://player-api.s4c-cdn.co.uk/streaming-urls/prod', video_id, query={
  63.                 'mode': 'od',
  64.                 'application': 'clic',
  65.                 'region': 'WW',
  66.                 'extra': 'false',
  67.                 'thirdParty': 'false',
  68.                 'filename': player_config['filename'],
  69.             }, note='Downloading streaming urls JSON')['hls']
  70.         formats = self._extract_m3u8_formats(m3u8_url, video_id, 'mp4', m3u8_id='hls', entry_protocol='m3u8_native')
  71.         self._sort_formats(formats)
  73.         subtitles = {}
  74.         for sub in traverse_obj(player_config, ('subtitles', lambda _, v: url_or_none(v['0']))):
  75.             subtitles.setdefault(sub.get('3', 'en'), []).append({
  76.                 'url': sub['0'],
  77.                 'name': sub.get('1'),
  78.             })
  80.         return merge_dicts({
  81.             'id': video_id,
  82.             'formats': formats,
  83.             'subtitles': subtitles,
  84.             'thumbnail': url_or_none(player_config.get('poster')),
  85.         }, traverse_obj(details, ('full_prog_details', 0, {
  86.             'title': (('programme_title', 'series_title'), T(txt_or_none)),
  87.             'description': ('full_billing', T(txt_or_none)),
  88.             'duration': ('duration', T(partial_f(float_or_none, invscale=60))),
  89.         }), get_all=False),
  90.             rev=True)
  93. class S4CSeriesIE(InfoExtractor):
  94.     _VALID_URL = r'https?://(?:www\.)?s4c\.cymru/clic/series/(?P<id>\d+)'
  95.     _TESTS = [{
  96.         'url': 'https://www.s4c.cymru/clic/series/864982911',
  97.         'playlist_mincount': 6,
  98.         'info_dict': {
  99.             'id': '864982911',
  100.             'title': 'Iaith ar Daith',
  101.         },
  102.     }, {
  103.         'url': 'https://www.s4c.cymru/clic/series/866852587',
  104.         'playlist_mincount': 8,
  105.         'info_dict': {
  106.             'id': '866852587',
  107.             'title': 'FFIT Cymru',
  108.         },
  109.     }]
  111.     def _real_extract(self, url):
  112.         series_id = self._match_id(url)
  113.         series_details = self._download_json(
  114.             'https://www.s4c.cymru/df/series_details', series_id, query={
  115.                 'lang': 'e',
  116.                 'series_id': series_id,
  117.                 'show_prog_in_series': 'Y'
  118.             }, note='Downloading series details JSON')
  120.         return self.playlist_result(
  121.             (self.url_result('https://www.s4c.cymru/clic/programme/' + episode_id, S4CIE, episode_id)
  122.              for episode_id in traverse_obj(series_details, ('other_progs_in_series', Ellipsis, 'id'))),
  123.             playlist_id=series_id, playlist_title=traverse_obj(
  124.                 series_details, ('full_prog_details', 0, 'series_title', T(txt_or_none))))