yt_dlp/extractor/allocine.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..compat import compat_str
   6 from ..utils import (
   7     int_or_none,
   8     qualities,
   9     remove_end,
  10     strip_or_none,
  11     try_get,
  12     unified_timestamp,
  13     url_basename,
  14 )
  15
  16
  17 class AllocineIE(InfoExtractor):
  18     _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)?'
  19
  20     _TESTS = [{
  21         'url': 'http://www.allocine.fr/article/fichearticle_gen_carticle=18635087.html',
  22         'md5': '0c9fcf59a841f65635fa300ac43d8269',
  23         'info_dict': {
  24             'id': '19546517',
  25             'display_id': '18635087',
  26             'ext': 'mp4',
  27             'title': 'Astérix - Le Domaine des Dieux Teaser VF',
  28             'description': 'md5:4a754271d9c6f16c72629a8a993ee884',
  29             'thumbnail': r're:http://.*\.jpg',
  30             'duration': 39,
  31             'timestamp': 1404273600,
  32             'upload_date': '20140702',
  33             'view_count': int,
  34         },
  35     }, {
  36         'url': 'http://www.allocine.fr/video/player_gen_cmedia=19540403&cfilm=222257.html',
  37         'md5': 'd0cdce5d2b9522ce279fdfec07ff16e0',
  38         'info_dict': {
  39             'id': '19540403',
  40             'display_id': '19540403',
  41             'ext': 'mp4',
  42             'title': 'Planes 2 Bande-annonce VF',
  43             'description': 'Regardez la bande annonce du film Planes 2 (Planes 2 Bande-annonce VF). Planes 2, un film de Roberts Gannaway',
  44             'thumbnail': r're:http://.*\.jpg',
  45             'duration': 69,
  46             'timestamp': 1385659800,
  47             'upload_date': '20131128',
  48             'view_count': int,
  49         },
  50     }, {
  51         'url': 'http://www.allocine.fr/video/player_gen_cmedia=19544709&cfilm=181290.html',
  52         'md5': '101250fb127ef9ca3d73186ff22a47ce',
  53         'info_dict': {
  54             'id': '19544709',
  55             'display_id': '19544709',
  56             'ext': 'mp4',
  57             'title': 'Dragons 2 - Bande annonce finale VF',
  58             'description': 'md5:6cdd2d7c2687d4c6aafe80a35e17267a',
  59             'thumbnail': r're:http://.*\.jpg',
  60             'duration': 144,
  61             'timestamp': 1397589900,
  62             'upload_date': '20140415',
  63             'view_count': int,
  64         },
  65     }, {
  66         'url': 'http://www.allocine.fr/video/video-19550147/',
  67         'md5': '3566c0668c0235e2d224fd8edb389f67',
  68         'info_dict': {
  69             'id': '19550147',
  70             'ext': 'mp4',
  71             'title': 'Faux Raccord N°123 - Les gaffes de Cliffhanger',
  72             'description': 'md5:bc734b83ffa2d8a12188d9eb48bb6354',
  73             'thumbnail': r're:http://.*\.jpg',
  74         },
  75     }]
  76
  77     def _real_extract(self, url):
  78         display_id = self._match_id(url)
  79
  80         webpage = self._download_webpage(url, display_id)
  81
  82         formats = []
  83         quality = qualities(['ld', 'md', 'hd'])
  84
  85         model = self._html_search_regex(
  86             r'data-model="([^"]+)"', webpage, 'data model', default=None)
  87         if model:
  88             model_data = self._parse_json(model, display_id)
  89             video = model_data['videos'][0]
  90             title = video['title']
  91             for video_url in video['sources'].values():
  92                 video_id, format_id = url_basename(video_url).split('_')[:2]
  93                 formats.append({
  94                     'format_id': format_id,
  95                     'quality': quality(format_id),
  96                     'url': video_url,
  97                 })
  98             duration = int_or_none(video.get('duration'))
  99             view_count = int_or_none(video.get('view_count'))
 100             timestamp = unified_timestamp(try_get(
 101                 video, lambda x: x['added_at']['date'], compat_str))
 102         else:
 103             video_id = display_id
 104             media_data = self._download_json(
 105                 'http://www.allocine.fr/ws/AcVisiondataV5.ashx?media=%s' % video_id, display_id)
 106             title = remove_end(strip_or_none(self._html_extract_title(webpage), ' - AlloCiné'))
 107             for key, value in media_data['video'].items():
 108                 if not key.endswith('Path'):
 109                     continue
 110                 format_id = key[:-len('Path')]
 111                 formats.append({
 112                     'format_id': format_id,
 113                     'quality': quality(format_id),
 114                     'url': value,
 115                 })
 116             duration, view_count, timestamp = [None] * 3
 117
 118         self._sort_formats(formats)
 119
 120         return {
 121             'id': video_id,
 122             'display_id': display_id,
 123             'title': title,
 124             'description': self._og_search_description(webpage),
 125             'thumbnail': self._og_search_thumbnail(webpage),
 126             'duration': duration,
 127             'timestamp': timestamp,
 128             'view_count': view_count,
 129             'formats': formats,
 130         }