test_performance.py

#!/usr/bin/env python3
"""Test script to measure performance improvements with SQLite database."""

import time
from pathlib import Path
from core.data_loader import TranscriptLoader
from core.database import TranscriptDatabase

def test_loading_performance():
    """Compare loading performance with and without database."""
    
    print("🔍 Testing Transcript Loading Performance\n")
    print("=" * 60)
    
    # Test 1: Initial load without database (baseline)
    print("\n📊 Test 1: Loading WITHOUT database (baseline)")
    loader_no_db = TranscriptLoader(use_database=False)
    loader_no_db.discover_transcript_files()
    print(f"Found {len(loader_no_db.file_paths)} transcript files")
    
    start = time.time()
    messages, conversations = loader_no_db.load_all_files(show_progress=True)
    no_db_time = time.time() - start
    
    print(f"✅ Loaded {len(messages)} messages in {no_db_time:.2f} seconds")
    print(f"   Speed: {len(messages) / no_db_time:.0f} messages/second")
    
    # Test 2: Initial load WITH database (first time - building cache)
    print("\n📊 Test 2: Initial load WITH database (building cache)")
    # Clear database first
    db = TranscriptDatabase()
    db.clear_database()
    
    loader_with_db = TranscriptLoader(use_database=True)
    loader_with_db.discover_transcript_files()
    
    start = time.time()
    messages, conversations = loader_with_db.load_all_files(show_progress=True)
    initial_db_time = time.time() - start
    
    print(f"✅ Loaded {len(messages)} messages in {initial_db_time:.2f} seconds")
    print(f"   Speed: {len(messages) / initial_db_time:.0f} messages/second")
    
    # Test 3: Subsequent load WITH database (using cache)
    print("\n📊 Test 3: Subsequent load WITH database (using cache)")
    loader_cached = TranscriptLoader(use_database=True)
    loader_cached.discover_transcript_files()
    
    start = time.time()
    messages, conversations = loader_cached.load_all_files(show_progress=True)
    cached_db_time = time.time() - start
    
    print(f"✅ Loaded {len(messages)} messages in {cached_db_time:.2f} seconds")
    print(f"   Speed: {len(messages) / cached_db_time:.0f} messages/second")
    
    # Get database statistics
    stats = loader_cached.get_statistics()
    
    # Print summary
    print("\n" + "=" * 60)
    print("📈 PERFORMANCE SUMMARY")
    print("=" * 60)
    print(f"\n1. Without Database:      {no_db_time:.2f} seconds")
    print(f"2. With Database (first): {initial_db_time:.2f} seconds")
    print(f"3. With Database (cached): {cached_db_time:.2f} seconds")
    
    if cached_db_time > 0:
        speedup = no_db_time / cached_db_time
        print(f"\n🚀 SPEEDUP: {speedup:.1f}x faster with cached database!")
    
    print(f"\n📊 Database Statistics:")
    print(f"   - Database size: {stats.get('database_size_mb', 0):.2f} MB")
    print(f"   - Total messages: {stats.get('total_messages', 0):,}")
    print(f"   - Total sessions: {stats.get('total_conversations', 0)}")
    print(f"   - Unique tools: {len(stats.get('unique_tools', []))}")
    
    # Test 4: Incremental update (modify one file)
    print("\n📊 Test 4: Incremental update (simulating file change)")
    # Touch a file to simulate modification
    if loader_cached.file_paths:
        test_file = loader_cached.file_paths[0]
        print(f"   Simulating change to: {test_file.name}")
        test_file.touch()
        
        start = time.time()
        messages, conversations = loader_cached.load_all_files(show_progress=True)
        incremental_time = time.time() - start
        
        print(f"✅ Incremental update took {incremental_time:.2f} seconds")
        print(f"   Only reloaded modified files!")

if __name__ == "__main__":
    test_loading_performance()