From 50ae3f646e6c7eec187078dd43f1cb3777ccfb4c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Sun, 2 Jul 2017 20:04:08 +0700 Subject: [PATCH] [thisoldhouse] Add more fallbacks for video id (closes #13541) --- youtube_dl/extractor/thisoldhouse.py | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/youtube_dl/extractor/thisoldhouse.py b/youtube_dl/extractor/thisoldhouse.py index 0c321354f..6ab147ad7 100644 --- a/youtube_dl/extractor/thisoldhouse.py +++ b/youtube_dl/extractor/thisoldhouse.py @@ -2,6 +2,8 @@ from __future__ import unicode_literals from .common import InfoExtractor +from ..compat import compat_str +from ..utils import try_get class ThisOldHouseIE(InfoExtractor): @@ -28,8 +30,15 @@ class ThisOldHouseIE(InfoExtractor): def _real_extract(self, url): display_id = self._match_id(url) webpage = self._download_webpage(url, display_id) - drupal_settings = self._parse_json(self._search_regex( - r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);', - webpage, 'drupal settings'), display_id) - video_id = list(drupal_settings['comScore'])[0] + video_id = self._search_regex( + (r'data-mid=(["\'])(?P(?:(?!\1).)+)\1', + r'id=(["\'])inline-video-player-(?P(?:(?!\1).)+)\1'), + webpage, 'video id', default=None, group='id') + if not video_id: + drupal_settings = self._parse_json(self._search_regex( + r'jQuery\.extend\(Drupal\.settings\s*,\s*({.+?})\);', + webpage, 'drupal settings'), display_id) + video_id = try_get( + drupal_settings, lambda x: x['jwplatform']['video_id'], + compat_str) or list(drupal_settings['comScore'])[0] return self.url_result('jwplatform:' + video_id, 'JWPlatform', video_id)