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

   1 # -*- coding: utf-8 -*-
   2 from __future__ import unicode_literals
   3
   4 import re
   5 import json
   6
   7 from .common import InfoExtractor
   8 from ..utils import (
   9     compat_str,
  10     qualities,
  11     determine_ext,
  12 )
  13
  14
  15 class AllocineIE(InfoExtractor):
  16     _VALID_URL = r'https?://(?:www\.)?allocine\.fr/(?P<typ>article|video|film)/(fichearticle_gen_carticle=|player_gen_cmedia=|fichefilm_gen_cfilm=)(?P<id>[0-9]+)(?:\.html)?'
  17
  18     _TESTS = [{
  19         'url': 'http://www.allocine.fr/article/fichearticle_gen_carticle=18635087.html',
  20         'md5': '0c9fcf59a841f65635fa300ac43d8269',
  21         'info_dict': {
  22             'id': '19546517',
  23             'ext': 'mp4',
  24             'title': 'Astérix - Le Domaine des Dieux Teaser VF',
  25             'description': 'md5:4a754271d9c6f16c72629a8a993ee884',
  26             'thumbnail': 're:http://.*\.jpg',
  27         },
  28     }, {
  29         'url': 'http://www.allocine.fr/video/player_gen_cmedia=19540403&cfilm=222257.html',
  30         'md5': 'd0cdce5d2b9522ce279fdfec07ff16e0',
  31         'info_dict': {
  32             'id': '19540403',
  33             'ext': 'mp4',
  34             'title': 'Planes 2 Bande-annonce VF',
  35             'description': 'md5:c4b1f7bd682a91de6491ada267ec0f4d',
  36             'thumbnail': 're:http://.*\.jpg',
  37         },
  38     }, {
  39         'url': 'http://www.allocine.fr/film/fichefilm_gen_cfilm=181290.html',
  40         'md5': '101250fb127ef9ca3d73186ff22a47ce',
  41         'info_dict': {
  42             'id': '19544709',
  43             'ext': 'mp4',
  44             'title': 'Dragons 2 - Bande annonce finale VF',
  45             'description': 'md5:e74a4dc750894bac300ece46c7036490',
  46             'thumbnail': 're:http://.*\.jpg',
  47         },
  48     }]
  49
  50     def _real_extract(self, url):
  51         mobj = re.match(self._VALID_URL, url)
  52         typ = mobj.group('typ')
  53         display_id = mobj.group('id')
  54
  55         webpage = self._download_webpage(url, display_id)
  56
  57         if typ == 'film':
  58             video_id = self._search_regex(r'href="/video/player_gen_cmedia=([0-9]+).+"', webpage, 'video id')
  59         else:
  60             player = self._search_regex(r'data-player=\'([^\']+)\'>', webpage, 'data player')
  61
  62             player_data = json.loads(player)
  63             video_id = compat_str(player_data['refMedia'])
  64
  65         xml = self._download_xml('http://www.allocine.fr/ws/AcVisiondataV4.ashx?media=%s' % video_id, display_id)
  66
  67         video = xml.find('.//AcVisionVideo').attrib
  68         quality = qualities(['ld', 'md', 'hd'])
  69
  70         formats = []
  71         for k, v in video.items():
  72             if re.match(r'.+_path', k):
  73                 format_id = k.split('_')[0]
  74                 formats.append({
  75                     'format_id': format_id,
  76                     'quality': quality(format_id),
  77                     'url': v,
  78                     'ext': determine_ext(v),
  79                 })
  80
  81         self._sort_formats(formats)
  82
  83         return {
  84             'id': video_id,
  85             'title': video['videoTitle'],
  86             'thumbnail': self._og_search_thumbnail(webpage),
  87             'formats': formats,
  88             'description': self._og_search_description(webpage),
  89         }