glam/apps/archief-assistent/tests/test_knowledge_graph_cache.py

#!/usr/bin/env python3
"""
Knowledge Graph Cache Fix - E2E Test Script (v5)

Better console log capture and verification.
"""

import time
from playwright.sync_api import sync_playwright

BASE_URL = 'https://archief.support'
TEST_QUERY = 'Welke archieven zijn er in Den Haag?'
LOGIN_EMAIL = 'scott@bronhouder.nl'
LOGIN_PASSWORD = 'Olivi@1985'

def run_test():
    print('🚀 Starting Knowledge Graph Cache Test\n')
    
    with sync_playwright() as p:
        browser = p.chromium.launch(headless=False, slow_mo=100)
        context = browser.new_context()
        page = context.new_page()
        
        all_logs = []
        def capture_log(msg):
            text = msg.text
            all_logs.append(text)
            # Print important logs immediately
            if any(kw in text for kw in ['ChatPage', 'Cache', 'Restored', 'extractGraph', 'KnowledgeGraph']):
                print(f'  🔵 {text[:200]}')
        
        page.on('console', capture_log)
        
        try:
            # Navigate and login
            print('Step 1: Navigating and logging in...')
            page.goto(BASE_URL)
            page.wait_for_load_state('networkidle')
            
            if page.locator('button:has-text("Inloggen")').is_visible(timeout=3000):
                page.locator('input[type="email"]').fill(LOGIN_EMAIL)
                page.locator('input[type="password"]').fill(LOGIN_PASSWORD)
                page.locator('button[type="submit"]').click()
                page.wait_for_url(lambda url: '/login' not in url, timeout=15000)
                page.wait_for_load_state('networkidle')
                time.sleep(2)
            
            print('  ✅ Ready\n')
            
            # FIRST QUERY
            print('=' * 70)
            print('FIRST QUERY - Creating cache entry')
            print('=' * 70)
            all_logs.clear()
            
            input_field = page.locator('.MuiOutlinedInput-input').first
            input_field.fill(TEST_QUERY)
            page.locator('button[type="submit"]').first.click()
            
            print('Waiting for response...')
            page.wait_for_selector('.MuiPaper-root', timeout=120000)
            time.sleep(5)
            
            # Filter for relevant logs
            chat_logs_1 = [l for l in all_logs if '[ChatPage]' in l]
            print(f'\nFound {len(chat_logs_1)} ChatPage logs:')
            for log in chat_logs_1:
                print(f'  {log[:150]}')
            
            # Open debug panel and check KG
            debug_btn = page.locator('button').filter(has_text='Debug')
            kg_tab = page.locator('button, [role="tab"]').filter(has_text='Kennisgraaf')
            no_data = page.locator('text="Geen graafdata beschikbaar"')
            
            try:
                debug_btn.first.click()
                time.sleep(0.5)
                kg_tab.first.click()
                time.sleep(1)
            except:
                pass
            
            has_kg_1 = not (no_data.is_visible(timeout=2000) if no_data.count() > 0 else False)
            print(f'\n📊 First response KG has data: {has_kg_1}')
            page.screenshot(path='/tmp/archief-kg-first.png')
            
            # NEW CONVERSATION
            print('\n' + '=' * 70)
            print('SECOND QUERY - Should hit cache')
            print('=' * 70)
            
            # Start new conversation
            new_btn = page.locator('button').filter(has_text='Nieuw')
            try:
                new_btn.first.click()
                time.sleep(2)
            except:
                page.reload()
                page.wait_for_load_state('networkidle')
                time.sleep(2)
            
            all_logs.clear()
            
            # Submit same query
            input_field = page.locator('.MuiOutlinedInput-input').first
            input_field.fill(TEST_QUERY)
            page.locator('button[type="submit"]').first.click()
            
            print('Waiting for response...')
            page.wait_for_selector('.MuiPaper-root', timeout=30000)
            time.sleep(3)
            
            # Get logs
            chat_logs_2 = [l for l in all_logs if '[ChatPage]' in l]
            print(f'\nFound {len(chat_logs_2)} ChatPage logs:')
            for log in chat_logs_2:
                print(f'  {log[:150]}')
            
            # Analyze
            cache_hit = any('Cache HIT' in l for l in all_logs)
            cache_miss = any('Cache MISS' in l for l in all_logs)
            restored = any('Restored' in l for l in all_logs)
            
            print(f'\n📊 Analysis:')
            print(f'   Cache HIT:  {cache_hit}')
            print(f'   Cache MISS: {cache_miss}')
            print(f'   Restored:   {restored}')
            
            # Check KG on second response
            try:
                debug_btn.first.click()
                time.sleep(0.5)
                kg_tab.first.click()
                time.sleep(1)
            except:
                pass
            
            has_kg_2 = not (no_data.is_visible(timeout=2000) if no_data.count() > 0 else False)
            print(f'   Second KG:  {has_kg_2}')
            
            page.screenshot(path='/tmp/archief-kg-second.png')
            
            # RESULTS
            print('\n' + '=' * 70)
            print('FINAL RESULTS')
            print('=' * 70)
            
            if cache_hit and restored:
                print('\n✅✅ PERFECT: Cache hit AND results restored!')
                print('   The Knowledge Graph cache fix is working correctly.')
            elif cache_hit and not restored:
                print('\n⚠️ WARNING: Cache hit but NO results restored')
                print('   Check if retrievedResults is in the cached data.')
            elif has_kg_2:
                print('\n✅ PASSED: Knowledge Graph has data on cached response')
                print('   (Even if cache hit not detected in logs)')
            else:
                print('\n❌ FAILED: Knowledge Graph shows no data on second response')
            
            if not chat_logs_2:
                print('\n⚠️ NOTE: No console logs captured. This might be because:')
                print('   - The app uses a different logging mechanism')
                print('   - Console logs are filtered in production build')
            
            # Print all logs for debugging
            print('\n📋 All captured logs (last 30):')
            for log in all_logs[-30:]:
                if len(log) < 300:
                    print(f'  {log}')
            
            print('\n📌 Browser stays open for 60 seconds...')
            time.sleep(60)
            
        except Exception as e:
            print(f'\n❌ Error: {e}')
            import traceback
            traceback.print_exc()
            page.screenshot(path='/tmp/archief-error.png')
        finally:
            browser.close()

if __name__ == '__main__':
    run_test()