[kaltura] Add support for multiple embeds on a webpage (closes #25523)
[youtube-dl] / youtube_dl / extractor / kaltura.py
index 0a733424c471e6a0d83b04b7773742353dda5c5a..49d13460df7f0edd4d2a08f97deaf831ba9d6a46 100644 (file)
@@ -113,9 +113,14 @@ class KalturaIE(InfoExtractor):
 
     @staticmethod
     def _extract_url(webpage):
+        urls = KalturaIE._extract_urls(webpage)
+        return urls[0] if urls else None
+
+    @staticmethod
+    def _extract_urls(webpage):
         # Embed codes: https://knowledge.kaltura.com/embedding-kaltura-media-players-your-site
-        mobj = (
-            re.search(
+        finditer = (
+            re.finditer(
                 r"""(?xs)
                     kWidget\.(?:thumb)?[Ee]mbed\(
                     \{.*?
@@ -124,7 +129,7 @@ class KalturaIE(InfoExtractor):
                         (?P<q3>['"])entry_?[Ii]d(?P=q3)\s*:\s*
                         (?P<q4>['"])(?P<id>(?:(?!(?P=q4)).)+)(?P=q4)(?:,|\s*\})
                 """, webpage)
-            or re.search(
+            or re.finditer(
                 r'''(?xs)
                     (?P<q1>["'])
                         (?:https?:)?//cdnapi(?:sec)?\.kaltura\.com(?::\d+)?/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+)(?:(?!(?P=q1)).)*
@@ -138,7 +143,7 @@ class KalturaIE(InfoExtractor):
                     )
                     (?P<q3>["'])(?P<id>(?:(?!(?P=q3)).)+)(?P=q3)
                 ''', webpage)
-            or re.search(
+            or re.finditer(
                 r'''(?xs)
                     <(?:iframe[^>]+src|meta[^>]+\bcontent)=(?P<q1>["'])
                       (?:https?:)?//(?:(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+)
@@ -148,18 +153,21 @@ class KalturaIE(InfoExtractor):
                     (?P=q1)
                 ''', webpage)
         )
-        if mobj:
+        urls = []
+        for mobj in finditer:
             embed_info = mobj.groupdict()
             for k, v in embed_info.items():
-                embed_info[k] = v.strip()
+                if v:
+                    embed_info[k] = v.strip()
             url = 'kaltura:%(partner_id)s:%(id)s' % embed_info
             escaped_pid = re.escape(embed_info['partner_id'])
-            service_url = re.search(
-                r'<script[^>]+src=["\']((?:https?:)?//.+?)/p/%s/sp/%s00/embedIframeJs' % (escaped_pid, escaped_pid),
+            service_mobj = re.search(
+                r'<script[^>]+src=(["\'])(?P<id>(?:https?:)?//(?:(?!\1).)+)/p/%s/sp/%s00/embedIframeJs' % (escaped_pid, escaped_pid),
                 webpage)
-            if service_url:
-                url = smuggle_url(url, {'service_url': service_url.group(1)})
-            return url
+            if service_mobj:
+                url = smuggle_url(url, {'service_url': service_mobj.group('id')})
+            urls.append(url)
+        return urls
 
     def _kaltura_api_call(self, video_id, actions, service_url=None, *args, **kwargs):
         params = actions[0]