[kaltura] Improve widget ID extraction (closes #11480)

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

diff --git a/youtube_dl/extractor/kaltura.py b/youtube_dl/extractor/kaltura.py
index 5a8403777ad45c547632ff3ad239876d7b3cc68c..c0ddad6f9629b054f3e0064a9b3d60dc0f41dcf0 100644 (file)
--- a/youtube_dl/extractor/kaltura.py
+++ b/youtube_dl/extractor/kaltura.py
@@ -105,20 +105,20 @@ class KalturaIE(InfoExtractor):
                     kWidget\.(?:thumb)?[Ee]mbed\(
                     \{.*?
                         (?P<q1>['\"])wid(?P=q1)\s*:\s*
-                        (?P<q2>['\"])_?(?P<partner_id>[^'\"]+)(?P=q2),.*?
+                        (?P<q2>['\"])_?(?P<partner_id>(?:(?!(?P=q2)).)+)(?P=q2),.*?
                         (?P<q3>['\"])entry_?[Ii]d(?P=q3)\s*:\s*
-                        (?P<q4>['\"])(?P<id>[^'\"]+)(?P=q4),
+                        (?P<q4>['\"])(?P<id>(?:(?!(?P=q4)).)+)(?P=q4)(?:,|\s*\})
                 """, webpage) or
             re.search(
                 r'''(?xs)
                     (?P<q1>["\'])
-                        (?:https?:)?//cdnapi(?:sec)?\.kaltura\.com/.*?(?:p|partner_id)/(?P<partner_id>\d+).*?
+                        (?:https?:)?//cdnapi(?:sec)?\.kaltura\.com/(?:(?!(?P=q1)).)*(?:p|partner_id)/(?P<partner_id>\d+)(?:(?!(?P=q1)).)*
                     (?P=q1).*?
                     (?:
                         entry_?[Ii]d|
                         (?P<q2>["\'])entry_?[Ii]d(?P=q2)
                     )\s*:\s*
-                    (?P<q3>["\'])(?P<id>.+?)(?P=q3)
+                    (?P<q3>["\'])(?P<id>(?:(?!(?P=q3)).)+)(?P=q3)
                 ''', webpage))
         if mobj:
             embed_info = mobj.groupdict()
@@ -266,6 +266,9 @@ class KalturaIE(InfoExtractor):
             # skip for now.
             if f.get('fileExt') == 'chun':
                 continue
+            if not f.get('fileExt') and f.get('containerFormat') == 'qt':
+                # QT indicates QuickTime; some videos have broken fileExt
+                f['fileExt'] = 'mov'
             video_url = sign_url(
                 '%s/flavorId/%s' % (data_url, f['id']))
             # audio-only has no videoCodecId (e.g. kaltura:1926081:0_c03e1b5g