yt_dlp/extractor/escapist.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     determine_ext,
   4     clean_html,
   5     int_or_none,
   6     float_or_none,
   7 )
   8
   9
  10 def _decrypt_config(key, string):
  11     a = ''
  12     i = ''
  13     r = ''
  14
  15     while len(a) < (len(string) / 2):
  16         a += key
  17
  18     a = a[0:int(len(string) / 2)]
  19
  20     t = 0
  21     while t < len(string):
  22         i += chr(int(string[t] + string[t + 1], 16))
  23         t += 2
  24
  25     icko = [s for s in i]
  26
  27     for t, c in enumerate(a):
  28         r += chr(ord(c) ^ ord(icko[t]))
  29
  30     return r
  31
  32
  33 class EscapistIE(InfoExtractor):
  34     _VALID_URL = r'https?://?(?:(?:www|v1)\.)?escapistmagazine\.com/videos/view/[^/]+/(?P<id>[0-9]+)'
  35     _TESTS = [{
  36         'url': 'http://www.escapistmagazine.com/videos/view/the-escapist-presents/6618-Breaking-Down-Baldurs-Gate',
  37         'md5': 'ab3a706c681efca53f0a35f1415cf0d1',
  38         'info_dict': {
  39             'id': '6618',
  40             'ext': 'mp4',
  41             'description': "Baldur's Gate: Original, Modded or Enhanced Edition? I'll break down what you can expect from the new Baldur's Gate: Enhanced Edition.",
  42             'title': "Breaking Down Baldur's Gate",
  43             'thumbnail': r're:^https?://.*\.jpg$',
  44             'duration': 264,
  45             'uploader': 'The Escapist',
  46         }
  47     }, {
  48         'url': 'http://www.escapistmagazine.com/videos/view/zero-punctuation/10044-Evolve-One-vs-Multiplayer',
  49         'md5': '9e8c437b0dbb0387d3bd3255ca77f6bf',
  50         'info_dict': {
  51             'id': '10044',
  52             'ext': 'mp4',
  53             'description': 'This week, Zero Punctuation reviews Evolve.',
  54             'title': 'Evolve - One vs Multiplayer',
  55             'thumbnail': r're:^https?://.*\.jpg$',
  56             'duration': 304,
  57             'uploader': 'The Escapist',
  58         }
  59     }, {
  60         'url': 'http://escapistmagazine.com/videos/view/the-escapist-presents/6618',
  61         'only_matching': True,
  62     }, {
  63         'url': 'https://v1.escapistmagazine.com/videos/view/the-escapist-presents/6618-Breaking-Down-Baldurs-Gate',
  64         'only_matching': True,
  65     }]
  66
  67     def _real_extract(self, url):
  68         video_id = self._match_id(url)
  69         webpage = self._download_webpage(url, video_id)
  70
  71         ims_video = self._parse_json(
  72             self._search_regex(
  73                 r'imsVideo\.play\(({.+?})\);', webpage, 'imsVideo'),
  74             video_id)
  75         video_id = ims_video['videoID']
  76         key = ims_video['hash']
  77
  78         config = self._download_webpage(
  79             'http://www.escapistmagazine.com/videos/vidconfig.php',
  80             video_id, 'Downloading video config', headers={
  81                 'Referer': url,
  82             }, query={
  83                 'videoID': video_id,
  84                 'hash': key,
  85             })
  86
  87         data = self._parse_json(_decrypt_config(key, config), video_id)
  88
  89         video_data = data['videoData']
  90
  91         title = clean_html(video_data['title'])
  92
  93         formats = [{
  94             'url': video['src'],
  95             'format_id': '%s-%sp' % (determine_ext(video['src']), video['res']),
  96             'height': int_or_none(video.get('res')),
  97         } for video in data['files']['videos']]
  98
  99         return {
 100             'id': video_id,
 101             'formats': formats,
 102             'title': title,
 103             'thumbnail': self._og_search_thumbnail(webpage) or data.get('poster'),
 104             'description': self._og_search_description(webpage),
 105             'duration': float_or_none(video_data.get('duration'), 1000),
 106             'uploader': video_data.get('publisher'),
 107             'series': video_data.get('show'),
 108         }