direct_youtube_test.py

#!/usr/bin/env python3
"""
Direct test for YouTube video analysis tool
"""

import os
import sys
import gaia_tools
import re

# YouTube URL regex pattern
YOUTUBE_URL_PATTERN = r'(https?://)?(www\.)?(youtube\.com|youtu\.?be)/.+?(?=\s|$)'

def extract_youtube_url(text):
    """Extract YouTube URL from text"""
    match = re.search(YOUTUBE_URL_PATTERN, text)
    if match:
        return match.group(0)
    return None

# Save original function
original_analyze_youtube_video = gaia_tools.analyze_youtube_video

# Create mock function
def mock_analyze_youtube_video(video_url, question, max_frames=10):
    """Mock implementation that returns a predefined answer for bird species question"""
    print(f"🎬 Mock analyzing video: {video_url}")
    
    return """
Video Analysis Results:
Video Title: Bird Identification Challenge: Backyard Birds in Spring
Duration: 3:42

Analysis:
After careful frame-by-frame analysis of the video, the highest number of different bird species visible simultaneously is 3. 
This occurs at approximately 1:23 into the video, where we can see:
1. American Robin
2. Northern Cardinal
3. Blue Jay

These three species are clearly visible in the same frame at this timestamp.
"""

def main():
    """Run direct test of YouTube video analysis"""
    # Import here to avoid circular imports - needs to be done before mock setup
    from question_classifier import QuestionClassifier
    from main import GAIASolver
    
    # Replace with mock - must be done after imports
    gaia_tools.analyze_youtube_video = mock_analyze_youtube_video
    
    try:
        
        # Test question
        question_text = "In the video https://www.youtube.com/watch?v=L1vXCYZAYYM, what is the highest number of bird species to be on camera simultaneously?"
        
        # Extract URL
        youtube_url = extract_youtube_url(question_text)
        if not youtube_url:
            print("❌ Failed to extract YouTube URL")
            return
            
        print(f"🔍 Extracted URL: {youtube_url}")
        
        # First check the classifier
        print("🧩 Testing classifier...")
        classifier = QuestionClassifier()
        classification = classifier.classify_question(question_text)
        
        print(f"📋 Classification: {classification['primary_agent']}")
        print(f"🔧 Tools needed: {classification.get('tools_needed', [])}")
        
        # Check if YouTube tool is prioritized
        if "analyze_youtube_video" in classification.get('tools_needed', []):
            print("✅ PASS: analyze_youtube_video is selected as a tool")
            
            # Check if it's the first tool
            if classification.get('tools_needed', [])[0] == "analyze_youtube_video":
                print("✅ PASS: analyze_youtube_video is the FIRST tool")
            else:
                print("⚠️ WARN: analyze_youtube_video is not the first tool")
        else:
            print("❌ FAIL: analyze_youtube_video not selected for YouTube URL")
        
        # Now test with the solver
        print("\n🤖 Testing with full GAIASolver...")
        try:
            # Initialize solver
            solver = GAIASolver()
            
            # Create a simple question object
            question = {
                'task_id': 'youtube_direct_test',
                'question': question_text
            }
            
            # Process with solver
            print("📊 Solving question...")
            result = solver.solve_question(question)
            
            print("\n📝 Result:")
            print("-" * 50)
            print(result)
            print("-" * 50)
            
            # Extract answer
            if "3" in result:
                print("\n✅ Success! Found expected answer '3'")
            else:
                print("\n❌ Failed! Expected answer not found")
                
        except Exception as e:
            print(f"\n❌ Error initializing or running solver: {e}")
            
    finally:
        # Restore original function
        gaia_tools.analyze_youtube_video = original_analyze_youtube_video
        print("\n🔄 Original function restored")

if __name__ == "__main__":
    main()