Update app.py

app.py

@@ -1,10 +1,8 @@
 import streamlit as st
-from pytube import YouTube
 from crewai import Agent, Task, Crew
 from transformers import pipeline
 import torch
 import whisper
-import yt_dlp
 import os
 
 st.title("AI Blog Writer from YouTube")
@@ -12,20 +10,20 @@ st.title("AI Blog Writer from YouTube")
 # Whisper model setup (Base Model for better speed)
 whisper_model = whisper.load_model("base")
 
-def download_audio(video_url):
-    """Downloads audio from the YouTube video using yt-dlp with cookies."""
-    # Get the path to the cookies file from the current directory
-    cookies_file = 'cookies.json'
-    
-    ydl_opts = {
-        'format': 'bestaudio/best',
-        'outtmpl': 'audio.mp4',
-        'cookies': cookies_file,  # Path to the cookies file
-    }
+def get_video_transcript(video_url):
+    """Fetch video transcript using CrewAI's YouTubeChannelSearch tool."""
+    from crewai.tools.youtubechannelsearch import YouTubeChannelSearch
+
+    # Initialize the YouTube Channel Search tool
+    search_tool = YouTubeChannelSearch()
+
+    # Search for video and fetch transcript
+    video_data = search_tool.search(video_url)
     
-    with yt_dlp.YoutubeDL(ydl_opts) as ydl:
-        info_dict = ydl.extract_info(video_url, download=True)
-    return 'audio.mp4'
+    if video_data and 'transcript' in video_data:
+        return video_data['transcript']
+    else:
+        raise Exception("Transcript not found for this video.")
 
 def transcribe_audio(audio_path):
     """Transcribes the audio using Whisper AI."""
@@ -51,12 +49,8 @@ video_url = st.text_input("Enter YouTube Video URL:")
 if st.button("Generate Blog"):
     if video_url:
         try:
-            st.info("Downloading audio...")
-            audio_path = download_audio(video_url)  # No need to ask for cookies file path anymore
-            st.success("Audio downloaded successfully!")
-            
-            st.info("Transcribing audio using Whisper...")
-            transcript_text = transcribe_audio(audio_path)
+            st.info("Fetching transcript...")
+            transcript_text = get_video_transcript(video_url)  # Fetch transcript using CrewAI
             st.text_area("Transcript Extracted", transcript_text, height=200)
 
             # Generate Blog with CrewAI