youtube_transcript_tool.py

from langchain_community.document_loaders import YoutubeLoader
from smolagents import Tool

class YoutubeTranscriptTool(Tool):
    name = "YouTubeTransciptTool"
    description = "This tool extracts transcripts from YouTube videos."

    inputs = {
        "url": {
            "type": "string",
            "description": "the URL for the YouTube video, like https://www.youtube.com/watch?v=1htKBjuUWec",
        }
    }

    output_type = "string"

    def forward(self, url: str) -> str:
        """
        Extracts the transcript from the YouTube video.

        https://python.langchain.com/docs/integrations/document_loaders/youtube_transcript/
        """
        loader = YoutubeLoader.from_youtube_url(url, add_video_info=False)
        document = loader.load()
        return document[0].page_content