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 ..compat import compat_str
   9 from ..utils import (
  10     qualities,
  11     unescapeHTML,
  12     xpath_element,
  13 )
  14
  15
  16 class AllocineIE(InfoExtractor):
  17     _VALID_URL = r'https?://(?:www\.)?allocine\.fr/(?P<typ>article|video|film)/(fichearticle_gen_carticle=|player_gen_cmedia=|fichefilm_gen_cfilm=|video-)(?P<id>[0-9]+)(?:\.html)?'
  18
  19     _TESTS = [{
  20         'url': 'http://www.allocine.fr/article/fichearticle_gen_carticle=18635087.html',
  21         'md5': '0c9fcf59a841f65635fa300ac43d8269',
  22         'info_dict': {
  23             'id': '19546517',
  24             'ext': 'mp4',
  25             'title': 'Astérix - Le Domaine des Dieux Teaser VF',
  26             'description': 'md5:abcd09ce503c6560512c14ebfdb720d2',
  27             'thumbnail': 're:http://.*\.jpg',
  28         },
  29     }, {
  30         'url': 'http://www.allocine.fr/video/player_gen_cmedia=19540403&cfilm=222257.html',
  31         'md5': 'd0cdce5d2b9522ce279fdfec07ff16e0',
  32         'info_dict': {
  33             'id': '19540403',
  34             'ext': 'mp4',
  35             'title': 'Planes 2 Bande-annonce VF',
  36             'description': 'Regardez la bande annonce du film Planes 2 (Planes 2 Bande-annonce VF). Planes 2, un film de Roberts Gannaway',
  37             'thumbnail': 're:http://.*\.jpg',
  38         },
  39     }, {
  40         'url': 'http://www.allocine.fr/film/fichefilm_gen_cfilm=181290.html',
  41         'md5': '101250fb127ef9ca3d73186ff22a47ce',
  42         'info_dict': {
  43             'id': '19544709',
  44             'ext': 'mp4',
  45             'title': 'Dragons 2 - Bande annonce finale VF',
  46             'description': 'md5:601d15393ac40f249648ef000720e7e3',
  47             'thumbnail': 're:http://.*\.jpg',
  48         },
  49     }, {
  50         'url': 'http://www.allocine.fr/video/video-19550147/',
  51         'only_matching': True,
  52     }]
  53
  54     def _real_extract(self, url):
  55         mobj = re.match(self._VALID_URL, url)
  56         typ = mobj.group('typ')
  57         display_id = mobj.group('id')
  58
  59         webpage = self._download_webpage(url, display_id)
  60
  61         if typ == 'film':
  62             video_id = self._search_regex(r'href="/video/player_gen_cmedia=([0-9]+).+"', webpage, 'video id')
  63         else:
  64             player = self._search_regex(r'data-player=\'([^\']+)\'>', webpage, 'data player', default=None)
  65             if player:
  66                 player_data = json.loads(player)
  67                 video_id = compat_str(player_data['refMedia'])
  68             else:
  69                 model = self._search_regex(r'data-model="([^"]+)">', webpage, 'data model')
  70                 model_data = self._parse_json(unescapeHTML(model), display_id)
  71                 video_id = compat_str(model_data['id'])
  72
  73         xml = self._download_xml('http://www.allocine.fr/ws/AcVisiondataV4.ashx?media=%s' % video_id, display_id)
  74
  75         video = xpath_element(xml, './/AcVisionVideo').attrib
  76         quality = qualities(['ld', 'md', 'hd'])
  77
  78         formats = []
  79         for k, v in video.items():
  80             if re.match(r'.+_path', k):
  81                 format_id = k.split('_')[0]
  82                 formats.append({
  83                     'format_id': format_id,
  84                     'quality': quality(format_id),
  85                     'url': v,
  86                 })
  87         self._sort_formats(formats)
  88
  89         return {
  90             'id': video_id,
  91             'title': video['videoTitle'],
  92             'thumbnail': self._og_search_thumbnail(webpage),
  93             'formats': formats,
  94             'description': self._og_search_description(webpage),
  95         }