diff options
| author | Sergey M․ <dstftw@gmail.com> | 2017-07-02 20:04:08 +0700 |
|---|---|---|
| committer | Sergey M․ <dstftw@gmail.com> | 2017-07-02 20:06:15 +0700 |
| commit | 50ae3f646e6c7eec187078dd43f1cb3777ccfb4c (patch) | |
| tree | 1f2b54928c44b2416a2e63ad1f315ab03847e647 | |
| parent | 99a7e76240f3e79d770bff0029dc3321e14b2a97 (diff) | |
[thisoldhouse] Add more fallbacks for video id (closes #13541)
| -rw-r--r-- | youtube_dl/extractor/thisoldhouse.py | 17 |
1 files 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<id>(?:(?!\1).)+)\1', + r'id=(["\'])inline-video-player-(?P<id>(?:(?!\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) |