Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/tvn24.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     int_or_none,
   7     unescapeHTML,
   8 )
   9
  10
  11 class TVN24IE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:(?:[^/]+)\.)?tvn24(?:bis)?\.pl/(?:[^/]+/)*(?P<id>[^/]+)'
  13     _TESTS = [{
  14         'url': 'http://www.tvn24.pl/wiadomosci-z-kraju,3/oredzie-artura-andrusa,702428.html',
  15         'md5': 'fbdec753d7bc29d96036808275f2130c',
  16         'info_dict': {
  17             'id': '1584444',
  18             'ext': 'mp4',
  19             'title': '"Święta mają być wesołe, dlatego, ludziska, wszyscy pod jemiołę"',
  20             'description': 'Wyjątkowe orędzie Artura Andrusa, jednego z gości "Szkła kontaktowego".',
  21             'thumbnail': 're:https?://.*[.]jpeg',
  22         }
  23     }, {
  24         'url': 'http://fakty.tvn24.pl/ogladaj-online,60/53-konferencja-bezpieczenstwa-w-monachium,716431.html',
  25         'only_matching': True,
  26     }, {
  27         'url': 'http://sport.tvn24.pl/pilka-nozna,105/ligue-1-kamil-glik-rozcial-glowe-monaco-tylko-remisuje-z-bastia,716522.html',
  28         'only_matching': True,
  29     }, {
  30         'url': 'http://tvn24bis.pl/poranek,146,m/gen-koziej-w-tvn24-bis-wracamy-do-czasow-zimnej-wojny,715660.html',
  31         'only_matching': True,
  32     }, {
  33         'url': 'https://www.tvn24.pl/magazyn-tvn24/angie-w-jednej-czwartej-polka-od-szarej-myszki-do-cesarzowej-europy,119,2158',
  34         'only_matching': True,
  35     }]
  36
  37     def _real_extract(self, url):
  38         video_id = self._match_id(url)
  39
  40         webpage = self._download_webpage(url, video_id)
  41
  42         title = self._og_search_title(webpage)
  43
  44         def extract_json(attr, name, fatal=True):
  45             return self._parse_json(
  46                 self._search_regex(
  47                     r'\b%s=(["\'])(?P<json>(?!\1).+?)\1' % attr, webpage,
  48                     name, group='json', fatal=fatal) or '{}',
  49                 video_id, transform_source=unescapeHTML, fatal=fatal)
  50
  51         quality_data = extract_json('data-quality', 'formats')
  52
  53         formats = []
  54         for format_id, url in quality_data.items():
  55             formats.append({
  56                 'url': url,
  57                 'format_id': format_id,
  58                 'height': int_or_none(format_id.rstrip('p')),
  59             })
  60         self._sort_formats(formats)
  61
  62         description = self._og_search_description(webpage)
  63         thumbnail = self._og_search_thumbnail(
  64             webpage, default=None) or self._html_search_regex(
  65             r'\bdata-poster=(["\'])(?P<url>(?!\1).+?)\1', webpage,
  66             'thumbnail', group='url')
  67
  68         share_params = extract_json(
  69             'data-share-params', 'share params', fatal=False)
  70         if isinstance(share_params, dict):
  71             video_id = share_params.get('id') or video_id
  72
  73         return {
  74             'id': video_id,
  75             'title': title,
  76             'description': description,
  77             'thumbnail': thumbnail,
  78             'formats': formats,
  79         }