yt_dlp/extractor/allocine.py

   1 from .common import InfoExtractor
   2 from ..compat import compat_str
   3 from ..utils import (
   4     int_or_none,
   5     qualities,
   6     remove_end,
   7     strip_or_none,
   8     try_get,
   9     unified_timestamp,
  10     url_basename,
  11 )
  12
  13
  14 class AllocineIE(InfoExtractor):
  15     _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)?'
  16
  17     _TESTS = [{
  18         'url': 'http://www.allocine.fr/article/fichearticle_gen_carticle=18635087.html',
  19         'md5': '0c9fcf59a841f65635fa300ac43d8269',
  20         'info_dict': {
  21             'id': '19546517',
  22             'display_id': '18635087',
  23             'ext': 'mp4',
  24             'title': 'Astérix - Le Domaine des Dieux Teaser VF',
  25             'description': 'md5:4a754271d9c6f16c72629a8a993ee884',
  26             'thumbnail': r're:http://.*\.jpg',
  27             'duration': 39,
  28             'timestamp': 1404273600,
  29             'upload_date': '20140702',
  30             'view_count': int,
  31         },
  32     }, {
  33         'url': 'http://www.allocine.fr/video/player_gen_cmedia=19540403&cfilm=222257.html',
  34         'md5': 'd0cdce5d2b9522ce279fdfec07ff16e0',
  35         'info_dict': {
  36             'id': '19540403',
  37             'display_id': '19540403',
  38             'ext': 'mp4',
  39             'title': 'Planes 2 Bande-annonce VF',
  40             'description': 'Regardez la bande annonce du film Planes 2 (Planes 2 Bande-annonce VF). Planes 2, un film de Roberts Gannaway',
  41             'thumbnail': r're:http://.*\.jpg',
  42             'duration': 69,
  43             'timestamp': 1385659800,
  44             'upload_date': '20131128',
  45             'view_count': int,
  46         },
  47     }, {
  48         'url': 'http://www.allocine.fr/video/player_gen_cmedia=19544709&cfilm=181290.html',
  49         'md5': '101250fb127ef9ca3d73186ff22a47ce',
  50         'info_dict': {
  51             'id': '19544709',
  52             'display_id': '19544709',
  53             'ext': 'mp4',
  54             'title': 'Dragons 2 - Bande annonce finale VF',
  55             'description': 'md5:6cdd2d7c2687d4c6aafe80a35e17267a',
  56             'thumbnail': r're:http://.*\.jpg',
  57             'duration': 144,
  58             'timestamp': 1397589900,
  59             'upload_date': '20140415',
  60             'view_count': int,
  61         },
  62     }, {
  63         'url': 'http://www.allocine.fr/video/video-19550147/',
  64         'md5': '3566c0668c0235e2d224fd8edb389f67',
  65         'info_dict': {
  66             'id': '19550147',
  67             'ext': 'mp4',
  68             'title': 'Faux Raccord N°123 - Les gaffes de Cliffhanger',
  69             'description': 'md5:bc734b83ffa2d8a12188d9eb48bb6354',
  70             'thumbnail': r're:http://.*\.jpg',
  71         },
  72     }]
  73
  74     def _real_extract(self, url):
  75         display_id = self._match_id(url)
  76
  77         webpage = self._download_webpage(url, display_id)
  78
  79         formats = []
  80         quality = qualities(['ld', 'md', 'hd'])
  81
  82         model = self._html_search_regex(
  83             r'data-model="([^"]+)"', webpage, 'data model', default=None)
  84         if model:
  85             model_data = self._parse_json(model, display_id)
  86             video = model_data['videos'][0]
  87             title = video['title']
  88             for video_url in video['sources'].values():
  89                 video_id, format_id = url_basename(video_url).split('_')[:2]
  90                 formats.append({
  91                     'format_id': format_id,
  92                     'quality': quality(format_id),
  93                     'url': video_url,
  94                 })
  95             duration = int_or_none(video.get('duration'))
  96             view_count = int_or_none(video.get('view_count'))
  97             timestamp = unified_timestamp(try_get(
  98                 video, lambda x: x['added_at']['date'], compat_str))
  99         else:
 100             video_id = display_id
 101             media_data = self._download_json(
 102                 'http://www.allocine.fr/ws/AcVisiondataV5.ashx?media=%s' % video_id, display_id)
 103             title = remove_end(strip_or_none(self._html_extract_title(webpage), ' - AlloCiné'))
 104             for key, value in media_data['video'].items():
 105                 if not key.endswith('Path'):
 106                     continue
 107                 format_id = key[:-len('Path')]
 108                 formats.append({
 109                     'format_id': format_id,
 110                     'quality': quality(format_id),
 111                     'url': value,
 112                 })
 113             duration, view_count, timestamp = [None] * 3
 114
 115         return {
 116             'id': video_id,
 117             'display_id': display_id,
 118             'title': title,
 119             'description': self._og_search_description(webpage),
 120             'thumbnail': self._og_search_thumbnail(webpage),
 121             'duration': duration,
 122             'timestamp': timestamp,
 123             'view_count': view_count,
 124             'formats': formats,
 125         }