yt_dlp/extractor/la7.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     smuggle_url,
   7 )
   8
   9
  10 class LA7IE(InfoExtractor):
  11     IE_NAME = 'la7.it'
  12     _VALID_URL = r'''(?x)(https?://)?(?:
  13         (?:www\.)?la7\.it/([^/]+)/(?:rivedila7|video)/|
  14         tg\.la7\.it/repliche-tgla7\?id=
  15     )(?P<id>.+)'''
  16
  17     _TESTS = [{
  18         # 'src' is a plain URL
  19         'url': 'http://www.la7.it/crozza/video/inccool8-02-10-2015-163722',
  20         'md5': '8b613ffc0c4bf9b9e377169fc19c214c',
  21         'info_dict': {
  22             'id': '0_42j6wd36',
  23             'ext': 'mp4',
  24             'title': 'Inc.Cool8',
  25             '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',
  26             'thumbnail': 're:^https?://.*',
  27             'uploader_id': 'kdla7pillole@iltrovatore.it',
  28             'timestamp': 1443814869,
  29             'upload_date': '20151002',
  30         },
  31     }, {
  32         'url': 'http://www.la7.it/omnibus/rivedila7/omnibus-news-02-07-2016-189077',
  33         'only_matching': True,
  34     }]
  35
  36     def _real_extract(self, url):
  37         video_id = self._match_id(url)
  38
  39         if not url.startswith('http'):
  40             url = '%s//%s' % (self.http_scheme(), url)
  41
  42         webpage = self._download_webpage(url, video_id)
  43
  44         player_data = self._search_regex(
  45             [r'(?s)videoParams\s*=\s*({.+?});', r'videoLa7\(({[^;]+})\);'],
  46             webpage, 'player data')
  47         vid = self._search_regex(r'vid\s*:\s*"(.+?)",', player_data, 'vid')
  48
  49         return {
  50             '_type': 'url_transparent',
  51             'url': smuggle_url('kaltura:103:%s' % vid, {
  52                 'service_url': 'http://nkdam.iltrovatore.it',
  53             }),
  54             'id': video_id,
  55             'title': self._og_search_title(webpage, default=None),
  56             'description': self._og_search_description(webpage, default=None),
  57             'thumbnail': self._og_search_thumbnail(webpage, default=None),
  58             'ie_key': 'Kaltura',
  59         }