[thisoldhouse] Improve video id extraction (closes #24549)
[youtube-dl] / youtube_dl / extractor / thisoldhouse.py
index 197258df141b4b6864afa0e4c1df7d0db431f64e..a3d9b4017b93b1d9381b896967fa9e68da59eaec 100644 (file)
@@ -5,31 +5,43 @@ from .common import InfoExtractor
 
 
 class ThisOldHouseIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?thisoldhouse\.com/(?:watch|how-to|tv-episode)/(?P<id>[^/?#]+)'
+    _VALID_URL = r'https?://(?:www\.)?thisoldhouse\.com/(?:watch|how-to|tv-episode|(?:[^/]+/)?\d+)/(?P<id>[^/?#]+)'
     _TESTS = [{
         'url': 'https://www.thisoldhouse.com/how-to/how-to-build-storage-bench',
-        'md5': '946f05bbaa12a33f9ae35580d2dfcfe3',
         'info_dict': {
-            'id': '2REGtUDQ',
+            'id': '5dcdddf673c3f956ef5db202',
             'ext': 'mp4',
             'title': 'How to Build a Storage Bench',
             'description': 'In the workshop, Tom Silva and Kevin O\'Connor build a storage bench for an entryway.',
             'timestamp': 1442548800,
             'upload_date': '20150918',
-        }
+        },
+        'params': {
+            'skip_download': True,
+        },
     }, {
         '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,
+    }, {
+        'url': 'https://www.thisoldhouse.com/furniture/21017078/how-to-build-a-storage-bench',
+        'only_matching': True,
+    }, {
+        'url': 'https://www.thisoldhouse.com/21113884/s41-e13-paradise-lost',
+        'only_matching': True,
+    }, {
+        # iframe www.thisoldhouse.com
+        'url': 'https://www.thisoldhouse.com/21083431/seaside-transformation-the-westerly-project',
+        'only_matching': True,
     }]
+    _ZYPE_TMPL = 'https://player.zype.com/embed/%s.html?api_key=hsOk_yMSPYNrT22e9pu8hihLXjaZf0JW5jsOWv4ZqyHJFvkJn6rtToHl09tbbsbe'
 
     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']
-        return self.url_result('jwplatform:' + video_id, 'JWPlatform', video_id)
+        video_id = self._search_regex(
+            r'<iframe[^>]+src=[\'"](?:https?:)?//(?:www\.)?thisoldhouse\.(?:chorus\.build|com)/videos/zype/([0-9a-f]{24})',
+            webpage, 'video id')
+        return self.url_result(self._ZYPE_TMPL % video_id, 'Zype', video_id)