[francetvinfo] Improve video id extraction (closes #28792)
authorSergey M․ <dstftw@gmail.com>
Sun, 25 Apr 2021 15:52:28 +0000 (22:52 +0700)
committerSergey M․ <dstftw@gmail.com>
Sun, 25 Apr 2021 15:52:28 +0000 (22:52 +0700)
youtube_dl/extractor/francetv.py

index 7cc88bf1811f2b526ed0301e17b638d73ee22750..e4ec2e2004ceaf6cef12986838376b4417b46270 100644 (file)
@@ -383,6 +383,10 @@ class FranceTVInfoIE(FranceTVBaseInfoExtractor):
     }, {
         'url': 'http://france3-regions.francetvinfo.fr/limousin/emissions/jt-1213-limousin',
         'only_matching': True,
+    }, {
+        # "<figure id=" pattern (#28792)
+        'url': 'https://www.francetvinfo.fr/culture/patrimoine/incendie-de-notre-dame-de-paris/notre-dame-de-paris-de-l-incendie-de-la-cathedrale-a-sa-reconstruction_4372291.html',
+        'only_matching': True,
     }]
 
     def _real_extract(self, url):
@@ -400,7 +404,7 @@ class FranceTVInfoIE(FranceTVBaseInfoExtractor):
             (r'player\.load[^;]+src:\s*["\']([^"\']+)',
              r'id-video=([^@]+@[^"]+)',
              r'<a[^>]+href="(?:https?:)?//videos\.francetv\.fr/video/([^@]+@[^"]+)"',
-             r'data-id=["\']([\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})'),
+             r'(?:data-id|<figure[^<]+\bid)=["\']([\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})'),
             webpage, 'video id')
 
         return self._make_url_result(video_id)