yt_dlp/extractor/ondemandkorea.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     ExtractorError,
   9     js_to_json,
  10 )
  11
  12
  13 class OnDemandKoreaIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:www\.)?ondemandkorea\.com/(?P<id>[^/]+)\.html'
  15     _GEO_COUNTRIES = ['US', 'CA']
  16     _TESTS = [{
  17         'url': 'https://www.ondemandkorea.com/ask-us-anything-e43.html',
  18         'info_dict': {
  19             'id': 'ask-us-anything-e43',
  20             'ext': 'mp4',
  21             'title': 'Ask Us Anything : Gain, Ji Soo - 09/24/2016',
  22             'description': 'A talk show/game show with a school theme where celebrity guests appear as “transfer students.”',
  23             'thumbnail': r're:^https?://.*\.jpg$',
  24         },
  25         'params': {
  26             'skip_download': 'm3u8 download'
  27         }
  28     }, {
  29         'url': 'https://www.ondemandkorea.com/confession-e01-1.html',
  30         'info_dict': {
  31             'id': 'confession-e01-1',
  32             'ext': 'mp4',
  33             'title': 'Confession : E01',
  34             'description': 'Choi Do-hyun, a criminal attorney, is the son of a death row convict. Ever since Choi Pil-su got arrested for murder, Do-hyun has wanted to solve his ',
  35             'thumbnail': r're:^https?://.*\.jpg$',
  36             'subtitles': {
  37                 'English': 'mincount:1',
  38             },
  39         },
  40         'params': {
  41             'skip_download': 'm3u8 download'
  42         }
  43     }]
  44
  45     def _real_extract(self, url):
  46         video_id = self._match_id(url)
  47         webpage = self._download_webpage(url, video_id, fatal=False)
  48
  49         if not webpage:
  50             # Page sometimes returns captcha page with HTTP 403
  51             raise ExtractorError(
  52                 'Unable to access page. You may have been blocked.',
  53                 expected=True)
  54
  55         if 'msg_block_01.png' in webpage:
  56             self.raise_geo_restricted(
  57                 msg='This content is not available in your region',
  58                 countries=self._GEO_COUNTRIES)
  59
  60         if 'This video is only available to ODK PLUS members.' in webpage:
  61             raise ExtractorError(
  62                 'This video is only available to ODK PLUS members.',
  63                 expected=True)
  64
  65         if 'ODK PREMIUM Members Only' in webpage:
  66             raise ExtractorError(
  67                 'This video is only available to ODK PREMIUM members.',
  68                 expected=True)
  69
  70         title = self._search_regex(
  71             r'class=["\']episode_title["\'][^>]*>([^<]+)',
  72             webpage, 'episode_title', fatal=False) or self._og_search_title(webpage)
  73
  74         jw_config = self._parse_json(
  75             self._search_regex(
  76                 r'playlist\s*=\s*\[(?P<options>.+)];?$',
  77                 webpage, 'jw config', flags=re.MULTILINE, group='options'),
  78             video_id, transform_source=js_to_json)
  79         info = self._parse_jwplayer_data(
  80             jw_config, video_id, require_title=False, m3u8_id='hls',
  81             base_url=url)
  82
  83         info.update({
  84             'title': title,
  85             'description': self._og_search_description(webpage),
  86             'thumbnail': self._og_search_thumbnail(webpage)
  87         })
  88         return info