fix: improve YouTube transcript handling for manual captions

backend/open_webui/retrieval/loaders/youtube.py

@@ -137,21 +137,42 @@ class YoutubeLoader:
             return []
 
         try:
-            # First try to get transcript in requested language
+            # First try to get manual transcript in requested language
-            transcript = transcript_list.find_transcript(self.language)
+            for lang in self.language:
-        except NoTranscriptFound:
+                try:
-            # Fallback: try to get any available transcript
+                    available_transcripts = (
-            available_transcripts = list(
+                        transcript_list._manually_created_transcripts
-                transcript_list._generated_transcripts.values()
+                    )
-            )
+                    if lang in available_transcripts:
-            if available_transcripts:
+                        transcript = available_transcripts[lang]
-                transcript = available_transcripts[0]
+                        log.info(f"Found manual transcript in language: {lang}")
-                log.info(
+                        break
-                    f"Using first available transcript in language: {transcript.language_code}"
+                except NoTranscriptFound:
-                )
+                    continue
             else:
-                log.error("No transcripts found for video")
+                # If no manual transcript found, try auto-generated ones
-                return []
+                try:
+                    transcript = transcript_list.find_transcript(self.language)
+                    log.info(
+                        f"Using auto-generated transcript in language: {transcript.language_code}"
+                    )
+                except NoTranscriptFound:
+                    # Final fallback: try to get any available transcript
+                    available_transcripts = list(
+                        transcript_list._manually_created_transcripts.values()
+                    ) + list(transcript_list._generated_transcripts.values())
+                    if available_transcripts:
+                        transcript = available_transcripts[0]
+                        log.info(
+                            f"Using first available transcript in language: {transcript.language_code}"
+                        )
+                    else:
+                        log.error("No transcripts found for video")
+                        return []
+
+        except Exception as e:
+            log.exception(f"Error fetching transcript: {str(e)}")
+            return []
 
         transcript_pieces: List[Dict[str, Any]] = transcript.fetch()
 

package-lock.json

@@ -12589,4 +12589,4 @@
 			}
 		}
 	}
-}
+}

package.json

@@ -110,4 +110,4 @@
 		"node": ">=18.13.0 <=22.x.x",
 		"npm": ">=6.0.0"
 	}
-}
+}

src/lib/i18n/locales/fr-FR/translation.json

@@ -1062,4 +1062,4 @@
 	"Your entire contribution will go directly to the plugin developer; Open WebUI does not take any percentage. However, the chosen funding platform might have its own fees.": "L'intégralité de votre contribution ira directement au développeur du plugin ; Open WebUI ne prend aucun pourcentage. Cependant, la plateforme de financement choisie peut avoir ses propres frais.",
 	"Youtube": "YouTube",
 	"Youtube Loader Settings": "Paramètres de l'outil de téléchargement YouTube"
-}
+}

src/lib/i18n/locales/ie-GA/translation.json

@@ -1062,4 +1062,4 @@
 	"Your entire contribution will go directly to the plugin developer; Open WebUI does not take any percentage. However, the chosen funding platform might have its own fees.": "Rachaidh do ranníocaíocht iomlán go díreach chuig an bhforbróir breiseán; Ní ghlacann Open WebUI aon chéatadán. Mar sin féin, d'fhéadfadh a tháillí féin a bheith ag an ardán maoinithe roghnaithe.",
 	"Youtube": "Youtube",
 	"Youtube Loader Settings": "Socruithe Luchtaire Youtube"
-}
+}

src/lib/i18n/locales/zh-CN/translation.json

@@ -1062,4 +1062,4 @@
 	"Your entire contribution will go directly to the plugin developer; Open WebUI does not take any percentage. However, the chosen funding platform might have its own fees.": "您的全部捐款将直接给到插件开发者，Open WebUI 不会收取任何比例。但众筹平台可能会有服务费、抽成。",
 	"Youtube": "YouTube",
 	"Youtube Loader Settings": "YouTube 爬取设置"
-}
+}