fix: improve YouTube transcript handling for manual captions

This commit is contained in:
Juanan Pereira 2025-01-09 09:36:24 +01:00
parent 49f462163a
commit f0f7a56c48
6 changed files with 40 additions and 19 deletions

View File

@ -137,21 +137,42 @@ class YoutubeLoader:
return []
try:
# First try to get transcript in requested language
transcript = transcript_list.find_transcript(self.language)
except NoTranscriptFound:
# Fallback: try to get any available transcript
available_transcripts = 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}"
)
# First try to get manual transcript in requested language
for lang in self.language:
try:
available_transcripts = (
transcript_list._manually_created_transcripts
)
if lang in available_transcripts:
transcript = available_transcripts[lang]
log.info(f"Found manual transcript in language: {lang}")
break
except NoTranscriptFound:
continue
else:
log.error("No transcripts found for video")
return []
# If no manual transcript found, try auto-generated ones
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()

2
package-lock.json generated
View File

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

View File

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

View File

@ -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"
}
}

View File

@ -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"
}
}

View File

@ -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"
}
}