yt_dlp/extractor/allocine.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     int_or_none,
   4     qualities,
   5     remove_end,
   6     strip_or_none,
   7     try_get,
   8     unified_timestamp,
   9     url_basename,
  10 )
  11
  12
  13 class AllocineIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:www\.)?allocine\.fr/(?:article|video|film)/(?:fichearticle_gen_carticle=|player_gen_cmedia=|fichefilm_gen_cfilm=|video-)(?P<id>[0-9]+)(?:\.html)?'
  15
  16     _TESTS = [{
  17         'url': 'http://www.allocine.fr/article/fichearticle_gen_carticle=18635087.html',
  18         'md5': '0c9fcf59a841f65635fa300ac43d8269',
  19         'info_dict': {
  20             'id': '19546517',
  21             'display_id': '18635087',
  22             'ext': 'mp4',
  23             'title': 'Astérix - Le Domaine des Dieux Teaser VF',
  24             'description': 'md5:4a754271d9c6f16c72629a8a993ee884',
  25             'thumbnail': r're:http://.*\.jpg',
  26             'duration': 39,
  27             'timestamp': 1404273600,
  28             'upload_date': '20140702',
  29             'view_count': int,
  30         },
  31     }, {
  32         'url': 'http://www.allocine.fr/video/player_gen_cmedia=19540403&cfilm=222257.html',
  33         'md5': 'd0cdce5d2b9522ce279fdfec07ff16e0',
  34         'info_dict': {
  35             'id': '19540403',
  36             'display_id': '19540403',
  37             'ext': 'mp4',
  38             'title': 'Planes 2 Bande-annonce VF',
  39             'description': 'Regardez la bande annonce du film Planes 2 (Planes 2 Bande-annonce VF). Planes 2, un film de Roberts Gannaway',
  40             'thumbnail': r're:http://.*\.jpg',
  41             'duration': 69,
  42             'timestamp': 1385659800,
  43             'upload_date': '20131128',
  44             'view_count': int,
  45         },
  46     }, {
  47         'url': 'http://www.allocine.fr/video/player_gen_cmedia=19544709&cfilm=181290.html',
  48         'md5': '101250fb127ef9ca3d73186ff22a47ce',
  49         'info_dict': {
  50             'id': '19544709',
  51             'display_id': '19544709',
  52             'ext': 'mp4',
  53             'title': 'Dragons 2 - Bande annonce finale VF',
  54             'description': 'md5:6cdd2d7c2687d4c6aafe80a35e17267a',
  55             'thumbnail': r're:http://.*\.jpg',
  56             'duration': 144,
  57             'timestamp': 1397589900,
  58             'upload_date': '20140415',
  59             'view_count': int,
  60         },
  61     }, {
  62         'url': 'http://www.allocine.fr/video/video-19550147/',
  63         'md5': '3566c0668c0235e2d224fd8edb389f67',
  64         'info_dict': {
  65             'id': '19550147',
  66             'ext': 'mp4',
  67             'title': 'Faux Raccord N°123 - Les gaffes de Cliffhanger',
  68             'description': 'md5:bc734b83ffa2d8a12188d9eb48bb6354',
  69             'thumbnail': r're:http://.*\.jpg',
  70         },
  71     }]
  72
  73     def _real_extract(self, url):
  74         display_id = self._match_id(url)
  75
  76         webpage = self._download_webpage(url, display_id)
  77
  78         formats = []
  79         quality = qualities(['ld', 'md', 'hd'])
  80
  81         model = self._html_search_regex(
  82             r'data-model="([^"]+)"', webpage, 'data model', default=None)
  83         if model:
  84             model_data = self._parse_json(model, display_id)
  85             video = model_data['videos'][0]
  86             title = video['title']
  87             for video_url in video['sources'].values():
  88                 video_id, format_id = url_basename(video_url).split('_')[:2]
  89                 formats.append({
  90                     'format_id': format_id,
  91                     'quality': quality(format_id),
  92                     'url': video_url,
  93                 })
  94             duration = int_or_none(video.get('duration'))
  95             view_count = int_or_none(video.get('view_count'))
  96             timestamp = unified_timestamp(try_get(
  97                 video, lambda x: x['added_at']['date'], str))
  98         else:
  99             video_id = display_id
 100             media_data = self._download_json(
 101                 f'http://www.allocine.fr/ws/AcVisiondataV5.ashx?media={video_id}', display_id)
 102             title = remove_end(strip_or_none(self._html_extract_title(webpage), ' - AlloCiné'))
 103             for key, value in media_data['video'].items():
 104                 if not key.endswith('Path'):
 105                     continue
 106                 format_id = key[:-len('Path')]
 107                 formats.append({
 108                     'format_id': format_id,
 109                     'quality': quality(format_id),
 110                     'url': value,
 111                 })
 112             duration, view_count, timestamp = [None] * 3
 113
 114         return {
 115             'id': video_id,
 116             'display_id': display_id,
 117             'title': title,
 118             'description': self._og_search_description(webpage),
 119             'thumbnail': self._og_search_thumbnail(webpage),
 120             'duration': duration,
 121             'timestamp': timestamp,
 122             'view_count': view_count,
 123             'formats': formats,
 124         }