[youtube] Fix extraction (closes #20758, closes #20759, closes #20761, closes #20762...

[youtube-dl] / youtube_dl / extractor / thisoldhouse.py

diff --git a/youtube_dl/extractor/thisoldhouse.py b/youtube_dl/extractor/thisoldhouse.py
index 7629f0d10e4ebc40bf25b0f02f52b2524ab9e303..6ab147ad726306ba9250599d34491a50e64e82d0 100644 (file)
--- a/youtube_dl/extractor/thisoldhouse.py
+++ b/youtube_dl/extractor/thisoldhouse.py
@@ -2,10 +2,12 @@
 from __future__ import unicode_literals
 
 from .common import InfoExtractor
+from ..compat import compat_str
+from ..utils import try_get
 
 
 class ThisOldHouseIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?thisoldhouse\.com/(?:watch|how-to)/(?P<id>[^/?#]+)'
+    _VALID_URL = r'https?://(?:www\.)?thisoldhouse\.com/(?:watch|how-to|tv-episode)/(?P<id>[^/?#]+)'
     _TESTS = [{
         'url': 'https://www.thisoldhouse.com/how-to/how-to-build-storage-bench',
         'md5': '568acf9ca25a639f0c4ff905826b662f',
@@ -20,13 +22,23 @@ class ThisOldHouseIE(InfoExtractor):
     }, {
         'url': 'https://www.thisoldhouse.com/watch/arlington-arts-crafts-arts-and-crafts-class-begins',
         'only_matching': True,
+    }, {
+        'url': 'https://www.thisoldhouse.com/tv-episode/ask-toh-shelf-rough-electric',
+        'only_matching': True,
     }]
 
     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 = drupal_settings['jwplatform']['video_id']
+        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)