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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..compat import compat_urllib_parse_unquote_plus
   6 from ..utils import (
   7     js_to_json,
   8 )
   9
  10
  11 class KaraoketvIE(InfoExtractor):
  12     _VALID_URL = r'http://karaoketv\.co\.il/\?container=songs&id=(?P<id>[0-9]+)'
  13     _TEST = {
  14         'url': 'http://karaoketv.co.il/?container=songs&id=171568',
  15         'info_dict': {
  16             'id': '171568',
  17             'ext': 'mp4',
  18             'title': 'אל העולם שלך - רותם כהן - שרים קריוקי',
  19         }
  20     }
  21
  22     def _real_extract(self, url):
  23         video_id = self._match_id(url)
  24         webpage = self._download_webpage(url, video_id)
  25
  26         page_video_url = self._og_search_video_url(webpage, video_id)
  27         config_json = compat_urllib_parse_unquote_plus(self._search_regex(
  28             r'config=(.*)', page_video_url, 'configuration'))
  29
  30         urls_info_json = self._download_json(
  31             config_json, video_id, 'Downloading configuration',
  32             transform_source=js_to_json)
  33
  34         url = urls_info_json['playlist'][0]['url']
  35
  36         return {
  37             'id': video_id,
  38             'title': self._og_search_title(webpage),
  39             'url': url,
  40         }