yt_dlp/extractor/axs.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     float_or_none,
   4     js_to_json,
   5     parse_iso8601,
   6     traverse_obj,
   7     url_or_none,
   8 )
   9
  10
  11 class AxsIE(InfoExtractor):
  12     IE_NAME = 'axs.tv'
  13     _VALID_URL = r'https?://(?:www\.)?axs\.tv/(?:channel/(?:[^/?#]+/)+)?video/(?P<id>[^/?#]+)'
  14
  15     _TESTS = [{
  16         'url': 'https://www.axs.tv/video/5f4dc776b70e4f1c194f22ef/',
  17         'md5': '8d97736ae8e50c64df528e5e676778cf',
  18         'info_dict': {
  19             'id': '5f4dc776b70e4f1c194f22ef',
  20             'title': 'Small Town',
  21             'ext': 'mp4',
  22             'description': 'md5:e314d28bfaa227a4d7ec965fae19997f',
  23             'upload_date': '20230602',
  24             'timestamp': 1685729564,
  25             'duration': 1284.216,
  26             'series': 'Rock & Roll Road Trip with Sammy Hagar',
  27             'season': 'Season 2',
  28             'season_number': 2,
  29             'episode': '3',
  30             'thumbnail': 'https://images.dotstudiopro.com/5f4e9d330a0c3b295a7e8394',
  31         },
  32     }, {
  33         'url': 'https://www.axs.tv/channel/rock-star-interview/video/daryl-hall',
  34         'md5': '300ae795cd8f9984652c0949734ffbdc',
  35         'info_dict': {
  36             'id': '5f488148b70e4f392572977c',
  37             'display_id': 'daryl-hall',
  38             'title': 'Daryl Hall',
  39             'ext': 'mp4',
  40             'description': 'md5:e54ecaa0f4b5683fc9259e9e4b196628',
  41             'upload_date': '20230214',
  42             'timestamp': 1676403615,
  43             'duration': 2570.668,
  44             'series': 'The Big Interview with Dan Rather',
  45             'season': 'Season 3',
  46             'season_number': 3,
  47             'episode': '5',
  48             'thumbnail': 'https://images.dotstudiopro.com/5f4d1901f340b50d937cec32',
  49         },
  50     }]
  51
  52     def _real_extract(self, url):
  53         display_id = self._match_id(url)
  54         webpage = self._download_webpage(url, display_id)
  55
  56         webpage_json_data = self._search_json(
  57             r'mountObj\s*=', webpage, 'video ID data', display_id,
  58             transform_source=js_to_json)
  59         video_id = webpage_json_data['video_id']
  60         company_id = webpage_json_data['company_id']
  61
  62         meta = self._download_json(
  63             f'https://api.myspotlight.tv/dotplayer/video/{company_id}/{video_id}',
  64             video_id, query={'device_type': 'desktop_web'})['video']
  65
  66         formats = self._extract_m3u8_formats(
  67             meta['video_m3u8'], video_id, 'mp4', m3u8_id='hls')
  68
  69         subtitles = {}
  70         for cc in traverse_obj(meta, ('closeCaption', lambda _, v: url_or_none(v['srtPath']))):
  71             subtitles.setdefault(cc.get('srtShortLang') or 'en', []).append(
  72                 {'ext': cc.get('srtExt'), 'url': cc['srtPath']})
  73
  74         return {
  75             'id': video_id,
  76             'display_id': display_id,
  77             'formats': formats,
  78             **traverse_obj(meta, {
  79                 'title': ('title', {str}),
  80                 'description': ('description', {str}),
  81                 'series': ('seriestitle', {str}),
  82                 'season_number': ('season', {int}),
  83                 'episode': ('episode', {str}),
  84                 'duration': ('duration', {float_or_none}),
  85                 'timestamp': ('updated_at', {parse_iso8601}),
  86                 'thumbnail': ('thumb', {url_or_none}),
  87             }),
  88             'subtitles': subtitles,
  89         }