Update app.py

app.py

@@ -1,15 +1,15 @@
 import streamlit as st
+import re
+import requests
+from youtube_transcript_api import YouTubeTranscriptApi
 from phi.agent import Agent
 from phi.model.google import Gemini
 from phi.tools.duckduckgo import DuckDuckGo
 from google.generativeai import upload_file, get_file
 import google.generativeai as genai
 import time
-from pathlib import Path
-import tempfile
 from dotenv import load_dotenv
 import os
-from pytube import YouTube  # Import pytube for downloading YouTube videos
 
 load_dotenv()
 
@@ -19,15 +19,15 @@ if API_KEY:
 
 # Page configuration
 st.set_page_config(
-    page_title="Multimodal AI Agent- Video Summarizer",
+    page_title="Video Summarizer with Transcript Extraction",
     page_icon="🎥",
     layout="wide"
 )
 
-st.title("Phidata Video AI Summarizer Agent 🎥🎤🖬")
+st.title("Video AI Summarizer with Transcript 🎥🎤🖬")
 st.header("Powered by Gemini 2.0 Flash Exp")
 
-
+# Initialize the agent
 @st.cache_resource
 def initialize_agent():
     return Agent(
@@ -37,69 +37,76 @@ def initialize_agent():
         markdown=True,
     )
 
-# Initialize the agent
 multimodal_Agent = initialize_agent()
 
-# YouTube video URL input
+# YouTube URL input
 video_url = st.text_input(
-    "Enter a YouTube video URL", 
-    help="Paste a YouTube video link for AI analysis"
+    "Enter a YouTube video URL",
+    help="Paste a YouTube video link to extract the transcript and generate a summary"
 )
 
+# Function to extract YouTube video ID from URL
+def extract_video_id(url):
+    match = re.search(r"v=([a-zA-Z0-9_-]+)", url)
+    if match:
+        return match.group(1)
+    else:
+        raise ValueError("Invalid YouTube URL")
+
+# Function to get the transcript using the YouTubeTranscriptApi
+def get_transcript(video_id):
+    transcript_raw = YouTubeTranscriptApi.get_transcript(video_id, languages=['en', 'es', 'ko'])
+    transcript_full = ' '.join([i['text'] for i in transcript_raw])
+    return transcript_full
+
+# Process the YouTube URL
 if video_url:
     try:
-        # Download the YouTube video
-        yt = YouTube(video_url)
-        video_stream = yt.streams.filter(progressive=True, file_extension='mp4').first()
-        
-        # Create a temporary file to save the video
-        with tempfile.NamedTemporaryFile(delete=False, suffix='.mp4') as temp_video:
-            video_stream.download(output_path=temp_video.name)
-            video_path = temp_video.name
-        
-        st.video(video_path, format="video/mp4", start_time=0)
+        # Extract video ID from the URL
+        video_id = extract_video_id(video_url)
+
+        # Get the transcript for the video
+        transcript = get_transcript(video_id)
+
+        # Display the transcript
+        st.subheader("Video Transcript")
+        st.text_area("Transcript", transcript, height=200)
 
         user_query = st.text_area(
-            "What insights are you seeking from the video?",
-            placeholder="Ask anything about the video content. The AI agent will analyze and gather additional context if needed.",
-            help="Provide specific questions or insights you want from the video."
+            "What insights or summary would you like from the transcript?",
+            placeholder="Ask anything about the transcript or request a summary.",
+            help="Provide specific questions or insights you want from the transcript."
         )
 
-        if st.button("🔍 Analyze Video", key="analyze_video_button"):
+        if st.button("🔍 Generate Summary", key="generate_summary_button"):
             if not user_query:
-                st.warning("Please enter a question or insight to analyze the video.")
+                st.warning("Please enter a query or request a summary of the transcript.")
             else:
                 try:
-                    with st.spinner("Processing video and gathering insights..."):
-                        # Upload and process the video file
-                        processed_video = upload_file(video_path)
-                        while processed_video.state.name == "PROCESSING":
-                            time.sleep(1)
-                            processed_video = get_file(processed_video.name)
-
+                    with st.spinner("Processing transcript and generating summary..."):
                         # Prompt generation for analysis
                         analysis_prompt = (
                             f"""
-                            Analyze the uploaded video for content and context.
-                            Respond to the following query using video insights and supplementary web research:
+                            Here is the transcript of the video:
+                            {transcript}
+
+                            The user has requested the following insight/summary:
                             {user_query}
-                            Provide a detailed, user-friendly, and actionable response.
+
+                            Please provide a detailed, user-friendly, and actionable summary based on the provided transcript.
                             """
                         )
 
                         # AI agent processing
-                        response = multimodal_Agent.run(analysis_prompt, videos=[processed_video])
+                        response = multimodal_Agent.run(analysis_prompt)
 
                     # Display the result
-                    st.subheader("Analysis Result")
+                    st.subheader("Summary Result")
                     st.markdown(response.content)
 
                 except Exception as error:
                     st.error(f"An error occurred during analysis: {error}")
-                finally:
-                    # Clean up temporary video file
-                    Path(video_path).unlink(missing_ok=True)
     except Exception as e:
-        st.error(f"Error downloading YouTube video: {e}")
+        st.error(f"Error processing the YouTube video: {e}")
 else:
-    st.info("Paste a YouTube video URL to begin analysis.")
+    st.info("Paste a YouTube video URL to begin transcript extraction and analysis.")