Create youtube_transcription_tool.py
Browse files
tools/youtube_transcription_tool.py
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from smolagents import Tool
|
| 2 |
+
from youtube_transcript_api import YouTubeTranscriptApi
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
class YouTubeTranscriptionTool(Tool):
|
| 6 |
+
"""
|
| 7 |
+
Tool to fetch the transcript of a YouTube video given its URL.
|
| 8 |
+
Args:
|
| 9 |
+
video_url (str): YouTube video URL.
|
| 10 |
+
Returns:
|
| 11 |
+
str: Transcript of the video as a single string.
|
| 12 |
+
"""
|
| 13 |
+
|
| 14 |
+
name = "youtube_transcription"
|
| 15 |
+
description = "Fetches the transcript of a YouTube video given its URL"
|
| 16 |
+
inputs = {
|
| 17 |
+
"video_url": {"type": "string", "description": "YouTube video URL"},
|
| 18 |
+
}
|
| 19 |
+
output_type = "string"
|
| 20 |
+
|
| 21 |
+
def forward(self, video_url: str) -> str:
|
| 22 |
+
video_id = video_url.strip().split("v=")[-1]
|
| 23 |
+
transcript = YouTubeTranscriptApi.get_transcript(video_id)
|
| 24 |
+
return " ".join([entry["text"] for entry in transcript])
|