yt_dlp/extractor/tweakers.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     determine_ext,
   4     int_or_none,
   5     mimetype2ext,
   6 )
   7
   8
   9 class TweakersIE(InfoExtractor):
  10     _VALID_URL = r'https?://tweakers\.net/video/(?P<id>\d+)'
  11     _TEST = {
  12         'url': 'https://tweakers.net/video/9926/new-nintendo-3ds-xl-op-alle-fronten-beter.html',
  13         'md5': 'fe73e417c093a788e0160c4025f88b15',
  14         'info_dict': {
  15             'id': '9926',
  16             'ext': 'mp4',
  17             'title': 'New Nintendo 3DS XL - Op alle fronten beter',
  18             'description': 'md5:3789b21fed9c0219e9bcaacd43fab280',
  19             'thumbnail': r're:^https?://.*\.jpe?g$',
  20             'duration': 386,
  21             'uploader_id': 's7JeEm',
  22         },
  23     }
  24
  25     def _real_extract(self, url):
  26         video_id = self._match_id(url)
  27         video_data = self._download_json(
  28             f'https://tweakers.net/video/s1playlist/{video_id}/1920/1080/playlist.json',
  29             video_id)['items'][0]
  30
  31         title = video_data['title']
  32
  33         formats = []
  34         for location in video_data.get('locations', {}).get('progressive', []):
  35             format_id = location.get('label')
  36             width = int_or_none(location.get('width'))
  37             height = int_or_none(location.get('height'))
  38             for source in location.get('sources', []):
  39                 source_url = source.get('src')
  40                 if not source_url:
  41                     continue
  42                 ext = mimetype2ext(source.get('type')) or determine_ext(source_url)
  43                 formats.append({
  44                     'format_id': format_id,
  45                     'url': source_url,
  46                     'width': width,
  47                     'height': height,
  48                     'ext': ext,
  49                 })
  50
  51         return {
  52             'id': video_id,
  53             'title': title,
  54             'description': video_data.get('description'),
  55             'thumbnail': video_data.get('poster'),
  56             'duration': int_or_none(video_data.get('duration')),
  57             'uploader_id': video_data.get('account'),
  58             'formats': formats,
  59         }