Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/la7.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     js_to_json,
   7     smuggle_url,
   8 )
   9
  10
  11 class LA7IE(InfoExtractor):
  12     IE_NAME = 'la7.it'
  13     _VALID_URL = r'''(?x)(https?://)?(?:
  14         (?:www\.)?la7\.it/([^/]+)/(?:rivedila7|video)/|
  15         tg\.la7\.it/repliche-tgla7\?id=
  16     )(?P<id>.+)'''
  17
  18     _TESTS = [{
  19         # 'src' is a plain URL
  20         'url': 'http://www.la7.it/crozza/video/inccool8-02-10-2015-163722',
  21         'md5': '8b613ffc0c4bf9b9e377169fc19c214c',
  22         'info_dict': {
  23             'id': 'inccool8-02-10-2015-163722',
  24             'ext': 'mp4',
  25             'title': 'Inc.Cool8',
  26             'description': 'Benvenuti nell\'incredibile mondo della INC. COOL. 8. dove “INC.” sta per “Incorporated” “COOL” sta per “fashion” ed Eight sta per il gesto  atletico',
  27             'thumbnail': 're:^https?://.*',
  28             'uploader_id': 'kdla7pillole@iltrovatore.it',
  29             'timestamp': 1443814869,
  30             'upload_date': '20151002',
  31         },
  32     }, {
  33         # 'src' is a dictionary
  34         'url': 'http://tg.la7.it/repliche-tgla7?id=189080',
  35         'md5': '6b0d8888d286e39870208dfeceaf456b',
  36         'info_dict': {
  37             'id': '189080',
  38             'ext': 'mp4',
  39             'title': 'TG LA7',
  40         },
  41     }, {
  42         'url': 'http://www.la7.it/omnibus/rivedila7/omnibus-news-02-07-2016-189077',
  43         'only_matching': True,
  44     }]
  45
  46     def _real_extract(self, url):
  47         video_id = self._match_id(url)
  48
  49         webpage = self._download_webpage(url, video_id)
  50
  51         player_data = self._parse_json(
  52             self._search_regex(r'videoLa7\(({[^;]+})\);', webpage, 'player data'),
  53             video_id, transform_source=js_to_json)
  54
  55         return {
  56             '_type': 'url_transparent',
  57             'url': smuggle_url('kaltura:103:%s' % player_data['vid'], {
  58                 'service_url': 'http://kdam.iltrovatore.it',
  59             }),
  60             'id': video_id,
  61             'title': player_data['title'],
  62             'description': self._og_search_description(webpage, default=None),
  63             'thumbnail': player_data.get('poster'),
  64             'ie_key': 'Kaltura',
  65         }