Georgia Tech MCP Server

#!/usr/bin/env python3 """Compare performance of old JSON cache vs new pickle_gz cache""" import json import time import sys from pathlib import Path sys.path.insert(0, '/home/wondermutt/gtmcp/src') from gtmcp.fast_cache import FastCache from gtmcp.models import CourseInfo from datetime import datetime print("Cache Performance Comparison: JSON vs Pickle+Gzip") print("=" * 60) # Create realistic test data (simulate full cache) def create_full_cache(): cache = {} timestamps = {} # Semesters cache['available_semesters'] = [{ 'code': f'20{25-i}{m:02d}', 'name': f'Term {i}', 'view_only': False, '__type__': 'Semester' } for i in range(3) for m in [2, 5, 8]] # Course lists (bulk of data) for term in ['202508', '202505', '202502']: for subject in ['CS', 'CSE', 'ECE', 'MATH', 'ISYE']: key = f'{term}_{subject}_all' cache[key] = [{ 'crn': str(20000 + i), 'title': f'{subject} {6000+i}: Long Course Title with Description of Advanced Topics in {subject}', 'subject': subject, 'course_number': str(6000 + i), 'section': chr(65 + (i % 26)), '__type__': 'CourseInfo' } for i in range(300)] # 300 courses per subject timestamps[key] = datetime.now().isoformat() # Individual course entries (1000+) for i in range(1500): key = f'course_202508_{10000+i}' cache[key] = { 'course': { 'crn': str(10000 + i), 'title': 'Very Long Course Title to Simulate Real Data', 'subject': 'CS', 'course_number': str(6000 + i), 'section': 'A', '__type__': 'CourseInfo' }, 'term_code': '202508', 'type': 'course' } timestamps[key] = datetime.now().isoformat() return {'cache': cache, 'timestamps': timestamps} print("Creating test data...") test_data = create_full_cache() total_entries = len(test_data['cache']) course_count = sum(len(v) if isinstance(v, list) else 0 for v in test_data['cache'].values()) print(f"Total cache entries: {total_entries}") print(f"Total courses: {course_count}") # Test old JSON approach print("\n1. OLD APPROACH (JSON with indent):") start_total = time.time() # Save start = time.time() json_str = json.dumps(test_data, indent=2) json_save_time = time.time() - start with open('/tmp/cache_json.json', 'w') as f: f.write(json_str) json_size = Path('/tmp/cache_json.json').stat().st_size # Load start = time.time() with open('/tmp/cache_json.json', 'r') as f: loaded = json.load(f) json_load_time = time.time() - start json_total_time = time.time() - start_total print(f" Save time: {json_save_time:.3f}s") print(f" Load time: {json_load_time:.3f}s") print(f" Total time: {json_total_time:.3f}s") print(f" File size: {json_size/1024/1024:.2f} MB") # Test new pickle_gz approach print("\n2. NEW APPROACH (Pickle + Gzip):") cache_dir = Path('/tmp/test_cache_perf') cache_dir.mkdir(exist_ok=True) fast_cache = FastCache(cache_dir, backend='pickle_gz') start_total = time.time() # Save start = time.time() fast_cache.save(test_data) pickle_save_time = time.time() - start # Load start = time.time() loaded = fast_cache.load() pickle_load_time = time.time() - start pickle_total_time = time.time() - start_total stats = fast_cache.get_stats() pickle_size = stats['size_mb'] * 1024 * 1024 print(f" Save time: {pickle_save_time:.3f}s") print(f" Load time: {pickle_load_time:.3f}s") print(f" Total time: {pickle_total_time:.3f}s") print(f" File size: {stats['size_mb']:.2f} MB") # Calculate improvements print("\n3. PERFORMANCE IMPROVEMENTS:") print("=" * 60) save_speedup = json_save_time / pickle_save_time load_speedup = json_load_time / pickle_load_time total_speedup = json_total_time / pickle_total_time size_reduction = (1 - pickle_size/json_size) * 100 print(f" Save: {save_speedup:.1f}x faster") print(f" Load: {load_speedup:.1f}x faster") print(f" Total: {total_speedup:.1f}x faster") print(f" Size: {size_reduction:.1f}% smaller") print(f"\n Time saved per operation: {json_total_time - pickle_total_time:.3f}s") print(f" Space saved: {(json_size - pickle_size)/1024/1024:.2f} MB") # Clean up import os os.remove('/tmp/cache_json.json') import shutil shutil.rmtree(cache_dir) print("\n4. CONCLUSION:") print("=" * 60) print(f"For a cache with {total_entries} entries and {course_count} courses:") print(f"- Old JSON approach: {json_total_time:.3f}s, {json_size/1024/1024:.2f} MB") print(f"- New Pickle+Gzip: {pickle_total_time:.3f}s, {stats['size_mb']:.2f} MB") print(f"\nThe new approach is {total_speedup:.1f}x faster and {size_reduction:.0f}% smaller!") print("\nThis means ChatGPT will experience much faster response times,") print("especially for repeated queries that benefit from caching.")