Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/atvat.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     determine_ext,
   7     int_or_none,
   8     unescapeHTML,
   9 )
  10
  11
  12 class ATVAtIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?atv\.at/(?:[^/]+/){2}(?P<id>[dv]\d+)'
  14     _TESTS = [{
  15         'url': 'http://atv.at/aktuell/di-210317-2005-uhr/v1698449/',
  16         'md5': 'c3b6b975fb3150fc628572939df205f2',
  17         'info_dict': {
  18             'id': '1698447',
  19             'ext': 'mp4',
  20             'title': 'DI, 21.03.17 | 20:05 Uhr 1/1',
  21         }
  22     }, {
  23         'url': 'http://atv.at/aktuell/meinrad-knapp/d8416/',
  24         'only_matching': True,
  25     }]
  26
  27     def _real_extract(self, url):
  28         display_id = self._match_id(url)
  29         webpage = self._download_webpage(url, display_id)
  30         video_data = self._parse_json(unescapeHTML(self._search_regex(
  31             r'class="[^"]*jsb_video/FlashPlayer[^"]*"[^>]+data-jsb="([^"]+)"',
  32             webpage, 'player data')), display_id)['config']['initial_video']
  33
  34         video_id = video_data['id']
  35         video_title = video_data['title']
  36
  37         parts = []
  38         for part in video_data.get('parts', []):
  39             part_id = part['id']
  40             part_title = part['title']
  41
  42             formats = []
  43             for source in part.get('sources', []):
  44                 source_url = source.get('src')
  45                 if not source_url:
  46                     continue
  47                 ext = determine_ext(source_url)
  48                 if ext == 'm3u8':
  49                     formats.extend(self._extract_m3u8_formats(
  50                         source_url, part_id, 'mp4', 'm3u8_native',
  51                         m3u8_id='hls', fatal=False))
  52                 else:
  53                     formats.append({
  54                         'format_id': source.get('delivery'),
  55                         'url': source_url,
  56                     })
  57             self._sort_formats(formats)
  58
  59             parts.append({
  60                 'id': part_id,
  61                 'title': part_title,
  62                 'thumbnail': part.get('preview_image_url'),
  63                 'duration': int_or_none(part.get('duration')),
  64                 'is_live': part.get('is_livestream'),
  65                 'formats': formats,
  66             })
  67
  68         return {
  69             '_type': 'multi_video',
  70             'id': video_id,
  71             'title': video_title,
  72             'entries': parts,
  73         }