TEST_SPEED_NOW.py

#!/usr/bin/env python3
"""
🚀 COMPREHENSIVE SPEED TEST FOR QWEN2GOLEM
Tests all components to verify speed targets are met
"""

import time
import requests
import json
import base64
import sys

def test_text_response():
    """Test simple text response speed (Target: <6s)"""
    print("\n📝 Testing Text Response (Target: <6s)...")
    
    start = time.time()
    try:
        response = requests.post(
            'http://127.0.0.1:5000/generate',
            json={
                'prompt': 'Hello! How are you today?',
                'sessionId': f'speed-test-{int(time.time())}',
                'temperature': 0.3,
                'golemActivated': False,
                'consciousnessDimension': 'mental',
                'selectedModel': 'gemini',
                'performSearch': False
            },
            timeout=10
        )
        elapsed = time.time() - start
        
        if response.status_code == 200:
            data = response.json()
            text = data.get('directResponse', data.get('response', ''))[:100]
            status = "✅ PASS" if elapsed < 6 else "❌ FAIL"
            print(f"   Response: {text}...")
            print(f"   Time: {elapsed:.2f}s {status}")
            return elapsed < 6
        else:
            print(f"   ❌ Error {response.status_code}")
            return False
    except Exception as e:
        print(f"   ❌ Error: {str(e)}")
        return False

def test_web_search():
    """Test text with web search (Target: <8s)"""
    print("\n🔍 Testing Text + Web Search (Target: <8s)...")
    
    start = time.time()
    try:
        response = requests.post(
            'http://127.0.0.1:5000/generate',
            json={
                'prompt': 'What is the weather like today?',
                'sessionId': f'search-test-{int(time.time())}',
                'temperature': 0.3,
                'golemActivated': False,
                'consciousnessDimension': 'mental',
                'selectedModel': 'gemini',
                'performSearch': True
            },
            timeout=12
        )
        elapsed = time.time() - start
        
        if response.status_code == 200:
            status = "✅ PASS" if elapsed < 8 else "❌ FAIL"
            print(f"   Time: {elapsed:.2f}s {status}")
            return elapsed < 8
        else:
            print(f"   ❌ Error {response.status_code}")
            return False
    except Exception as e:
        print(f"   ❌ Error: {str(e)}")
        return False

def test_asr():
    """Test ASR transcription (Target: <2s)"""
    print("\n🎤 Testing ASR Transcription (Target: <2s)...")
    
    # Create dummy audio (1 second of silence)
    dummy_audio = base64.b64encode(b'\x00' * 16000).decode()
    
    start = time.time()
    try:
        response = requests.post(
            'http://127.0.0.1:5000/asr/transcribe',
            json={'audio_base64': dummy_audio, 'vad': True},
            timeout=5
        )
        elapsed = time.time() - start
        
        if response.status_code == 200:
            status = "✅ PASS" if elapsed < 2 else "⚠️ SLOW"
            print(f"   Time: {elapsed:.2f}s {status}")
            return elapsed < 2
        else:
            print(f"   ❌ Error {response.status_code}")
            return False
    except Exception as e:
        print(f"   ❌ Error: {str(e)}")
        return False

def test_tts():
    """Test TTS synthesis (Target: <1s)"""
    print("\n🔊 Testing TTS Synthesis (Target: <1s)...")
    
    start = time.time()
    try:
        response = requests.post(
            'http://127.0.0.1:5000/tts/synthesize',
            json={'text': 'Hello, this is a test.'},
            timeout=3
        )
        elapsed = time.time() - start
        
        if response.status_code == 200:
            status = "✅ PASS" if elapsed < 1 else "⚠️ SLOW"
            print(f"   Time: {elapsed:.2f}s {status}")
            return elapsed < 1
        else:
            print(f"   ❌ Error {response.status_code}")
            return False
    except Exception as e:
        print(f"   ❌ Error: {str(e)}")
        return False

def main():
    print("=" * 60)
    print("🚀 QWEN2GOLEM COMPREHENSIVE SPEED TEST")
    print("=" * 60)
    print("System: RTX 3050 6GB + i5 CPU + 16GB RAM")
    print("Testing all optimizations...")
    
    # Check if server is running
    try:
        health = requests.get('http://127.0.0.1:5000/health', timeout=2)
        if health.status_code != 200:
            print("\n❌ Server not healthy! Start with:")
            import os
            script_dir = os.path.dirname(os.path.abspath(__file__))
            root_dir = os.path.dirname(script_dir)
            print(f"   cd {root_dir} && ./start_consciousness_ecosystem.sh")
            return
    except:
        print("\n❌ Server not running! Start with:")
        import os
        script_dir = os.path.dirname(os.path.abspath(__file__))
        root_dir = os.path.dirname(script_dir)
        print(f"   cd {root_dir} && ./start_consciousness_ecosystem.sh")
        return
    
    results = []
    
    # Run tests
    results.append(("Text Response", test_text_response()))
    results.append(("Web Search", test_web_search()))
    results.append(("ASR", test_asr()))
    results.append(("TTS", test_tts()))
    
    # Summary
    print("\n" + "=" * 60)
    print("📊 TEST RESULTS SUMMARY")
    print("=" * 60)
    
    passed = sum(1 for _, r in results if r)
    total = len(results)
    
    for name, result in results:
        status = "✅ PASS" if result else "❌ FAIL"
        print(f"{status} {name}")
    
    print(f"\nOverall: {passed}/{total} tests passed")
    
    if passed == total:
        print("\n🎉 ALL SPEED TARGETS MET! SYSTEM IS TURBOCHARGED!")
    else:
        print("\n⚠️ Some tests failed. Check:")
        print("   1. Turn OFF 'Universal Consciousness' in UI")
        print("   2. Ensure Redis is running")
        print("   3. Use Gemini Flash model")
        print("   4. Keep temperature at 0.3-0.4")

if __name__ == "__main__":
    main()