yt_dlp/extractor/joj.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     format_field,
   4     int_or_none,
   5     js_to_json,
   6     try_get,
   7 )
   8
   9
  10 class JojIE(InfoExtractor):
  11     _VALID_URL = r'''(?x)
  12                     (?:
  13                         joj:|
  14                         https?://media\.joj\.sk/embed/
  15                     )
  16                     (?P<id>[^/?#^]+)
  17                 '''
  18     _EMBED_REGEX = [r'<iframe\b[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//media\.joj\.sk/embed/(?:(?!\1).)+)\1']
  19     _TESTS = [{
  20         'url': 'https://media.joj.sk/embed/a388ec4c-6019-4a4a-9312-b1bee194e932',
  21         'info_dict': {
  22             'id': 'a388ec4c-6019-4a4a-9312-b1bee194e932',
  23             'ext': 'mp4',
  24             'title': 'NOVÉ BÝVANIE',
  25             'thumbnail': r're:^https?://.*?$',
  26             'duration': 3118,
  27         },
  28     }, {
  29         'url': 'https://media.joj.sk/embed/CSM0Na0l0p1',
  30         'info_dict': {
  31             'id': 'CSM0Na0l0p1',
  32             'ext': 'mp4',
  33             'height': 576,
  34             'title': 'Extrémne rodiny 2 - POKRAČOVANIE (2012/04/09 21:30:00)',
  35             'duration': 3937,
  36             'thumbnail': r're:^https?://.*?$',
  37         },
  38     }, {
  39         'url': 'https://media.joj.sk/embed/9i1cxv',
  40         'only_matching': True,
  41     }, {
  42         'url': 'joj:a388ec4c-6019-4a4a-9312-b1bee194e932',
  43         'only_matching': True,
  44     }, {
  45         'url': 'joj:9i1cxv',
  46         'only_matching': True,
  47     }]
  48
  49     def _real_extract(self, url):
  50         video_id = self._match_id(url)
  51
  52         webpage = self._download_webpage(
  53             f'https://media.joj.sk/embed/{video_id}', video_id)
  54
  55         title = (self._search_json(r'videoTitle\s*:', webpage, 'title', video_id,
  56                                    contains_pattern=r'["\'].+["\']', default=None)
  57                  or self._html_extract_title(webpage, default=None)
  58                  or self._og_search_title(webpage))
  59
  60         bitrates = self._parse_json(
  61             self._search_regex(
  62                 r'(?s)(?:src|bitrates)\s*=\s*({.+?});', webpage, 'bitrates',
  63                 default='{}'),
  64             video_id, transform_source=js_to_json, fatal=False)
  65
  66         formats = []
  67         for format_url in try_get(bitrates, lambda x: x['mp4'], list) or []:
  68             if isinstance(format_url, str):
  69                 height = self._search_regex(
  70                     r'(\d+)[pP]|(pal)\.', format_url, 'height', default=None)
  71                 if height == 'pal':
  72                     height = 576
  73                 formats.append({
  74                     'url': format_url,
  75                     'format_id': format_field(height, None, '%sp'),
  76                     'height': int_or_none(height),
  77                 })
  78         if not formats:
  79             playlist = self._download_xml(
  80                 f'https://media.joj.sk/services/Video.php?clip={video_id}',
  81                 video_id)
  82             for file_el in playlist.findall('./files/file'):
  83                 path = file_el.get('path')
  84                 if not path:
  85                     continue
  86                 format_id = file_el.get('id') or file_el.get('label')
  87                 formats.append({
  88                     'url': 'http://n16.joj.sk/storage/{}'.format(path.replace(
  89                         'dat/', '', 1)),
  90                     'format_id': format_id,
  91                     'height': int_or_none(self._search_regex(
  92                         r'(\d+)[pP]', format_id or path, 'height',
  93                         default=None)),
  94                 })
  95
  96         thumbnail = self._og_search_thumbnail(webpage)
  97
  98         duration = int_or_none(self._search_regex(
  99             r'videoDuration\s*:\s*(\d+)', webpage, 'duration', fatal=False))
 100
 101         return {
 102             'id': video_id,
 103             'title': title,
 104             'thumbnail': thumbnail,
 105             'duration': duration,
 106             'formats': formats,
 107         }