_ Git - youtube-dl/blob - youtube_dl/extractor/ynet.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 compat_urllib_parse
   9
  10
  11 class YnetIE(InfoExtractor):
  12     _VALID_URL = r'http://.*ynet\.co\.il/.*/0,7340,(?P<id>L(?:-[0-9]+)+),00\.html'
  13     _TESTS = [
  14         {
  15             'url': 'http://hot.ynet.co.il/home/0,7340,L-11659-99244,00.html',
  16             'md5': '002b44ee2f33d50363a1c153bed524cf',
  17             'info_dict': {
  18                 'id': 'L-11659-99244',
  19                 'ext': 'flv',
  20                 'title': 'איש לא יודע מאיפה באנו',
  21                 'thumbnail': 're:^https?://.*\.jpg',
  22             }
  23         }, {
  24             'url': 'http://hot.ynet.co.il/home/0,7340,L-8859-84418,00.html',
  25             'md5': '6455046ae1b48cf7e2b7cae285e53a16',
  26             'info_dict': {
  27                 'id': 'L-8859-84418',
  28                 'ext': 'flv',
  29                 'title': "צפו: הנשיקה הלוהטת של תורגי' ויוליה פלוטקין",
  30                 'thumbnail': 're:^https?://.*\.jpg',
  31             }
  32         }
  33     ]
  34
  35     def _real_extract(self, url):
  36         mobj = re.match(self._VALID_URL, url)
  37         video_id = mobj.group('id')
  38
  39         webpage = self._download_webpage(url, video_id)
  40
  41         content = compat_urllib_parse.unquote_plus(self._og_search_video_url(webpage))
  42         config = json.loads(self._search_regex(r'config=({.+?})$', content, 'video config'))
  43         f4m_url = config['clip']['url']
  44         title = self._og_search_title(webpage)
  45         m = re.search(r'ynet - HOT -- (["\']+)(?P<title>.+?)\1', title)
  46         if m:
  47             title = m.group('title')
  48
  49         return {
  50             'id': video_id,
  51             'title': title,
  52             'formats': self._extract_f4m_formats(f4m_url, video_id),
  53             'thumbnail': self._og_search_thumbnail(webpage),
  54         }