#!/usr/bin/env python3
"""
Test script to verify the Firebase loading and content processing improvements.
"""

import os
import sys

# Add the parent directory to Python path
current_dir = os.path.dirname(os.path.abspath(__file__))
sys.path.insert(0, current_dir)

def test_firebase_loading():
    """Test Firebase loading with language filtering."""
    print("🧪 Testing Firebase Loading and Content Processing")
    print("=" * 60)
    
    try:
        from cve_factchecker.firebase_loader import FirebaseNewsLoader
        
        # Test Firebase loader
        loader = FirebaseNewsLoader()
        print(f"✅ Firebase loader initialized - Project: {loader.project_id}")
        
        # Test with very small limit first
        print("\n🔍 Testing with small sample (5 articles)...")
        articles = loader.fetch_articles(limit=5)
        
        if articles:
            print(f"✅ Fetched {len(articles)} sample articles")
            
            # Analyze the first article
            first_article = articles[0]
            print(f"\n📋 Sample Article Analysis:")
            print(f"   Title: {first_article.title[:100]}...")
            print(f"   Content length: {len(first_article.content)} characters")
            print(f"   URL: {first_article.url}")
            print(f"   Source: {first_article.source}")
            print(f"   Article ID: {first_article.article_id}")
            
            # Show content preview
            if first_article.content:
                content_preview = first_article.content[:300].replace('\n', ' ')
                print(f"   Content preview: {content_preview}...")
            else:
                print("   ⚠️ No content found!")
        
        # Test language-specific fetching
        print("\n🌐 Testing English language filtering...")
        english_articles = loader.fetch_articles_by_language("English", limit=10)
        
        if english_articles:
            print(f"✅ Fetched {len(english_articles)} English articles")
            
            # Check content quality
            valid_content = 0
            for article in english_articles[:3]:  # Check first 3
                if article.content and len(article.content) > 100:
                    valid_content += 1
                    print(f"   📄 '{article.title[:50]}...' - {len(article.content)} chars")
                else:
                    print(f"   ⚠️ '{article.title[:50]}...' - insufficient content")
            
            print(f"   Content quality: {valid_content}/{min(3, len(english_articles))} articles have substantial content")
        else:
            print("❌ No English articles found")
        
        return english_articles
        
    except Exception as e:
        print(f"❌ Firebase test failed: {e}")
        import traceback
        traceback.print_exc()
        return []

def test_vector_processing(articles):
    """Test vector store processing."""
    print("\n🔍 Testing Vector Store Processing")
    print("=" * 60)
    
    try:
        from cve_factchecker.retriever import VectorRetriever
        
        # Test vector retriever
        retriever = VectorRetriever(persist_directory="/tmp/test_vector_db")
        print("✅ Vector retriever initialized")
        
        # Test article storage
        print(f"\n📦 Testing storage of {len(articles)} articles...")
        retriever.store_articles_in_vector_db(articles, clear_first=True)
        
        # Test retrieval
        print("\n🔍 Testing document retrieval...")
        test_query = "security vulnerability"
        results = retriever.search(test_query, k=3)
        
        if results:
            print(f"✅ Found {len(results)} relevant documents for '{test_query}'")
            for i, doc in enumerate(results):
                content_preview = doc.page_content[:100].replace('\n', ' ')
                print(f"   {i+1}. {content_preview}...")
        else:
            print("⚠️ No documents found for test query")
        
        return True
        
    except Exception as e:
        print(f"❌ Vector processing test failed: {e}")
        import traceback
        traceback.print_exc()
        return False

def test_full_system():
    """Test the complete system integration."""
    print("\n🚀 Testing Full System Integration")
    print("=" * 60)
    
    try:
        from cve_factchecker.orchestrator import FactCheckSystem
        
        # Test system initialization
        system = FactCheckSystem(vector_dir="/tmp/test_system_vector")
        print("✅ Fact check system initialized")
        
        # Test Firebase ingestion
        print("\n🔄 Testing Firebase ingestion...")
        result = system.ingest_firebase(limit=10)
        
        if result.get("success"):
            print(f"✅ Ingestion successful: {result.get('synced')} articles")
        else:
            print(f"⚠️ Ingestion issues: {result.get('error', 'Unknown error')}")
        
        # Test fact checking
        print("\n🧠 Testing fact checking...")
        test_claim = "A new security vulnerability was discovered in popular software"
        fact_result = system.fact_check(test_claim)
        
        print(f"   Claim: {test_claim}")
        print(f"   Verdict: {fact_result.get('verdict', 'Unknown')}")
        print(f"   Confidence: {fact_result.get('confidence', 0)}")
        reasoning = fact_result.get('reasoning', 'No reasoning provided')
        print(f"   Reasoning: {reasoning[:200]}...")
        
        return True
        
    except Exception as e:
        print(f"❌ Full system test failed: {e}")
        import traceback
        traceback.print_exc()
        return False

def main():
    """Run all tests."""
    print("🧪 CVE Fact Checker - Comprehensive Test Suite")
    print("=" * 80)
    
    # Test 1: Firebase loading
    articles = test_firebase_loading()
    
    # Test 2: Vector processing (if we have articles)
    if articles:
        vector_success = test_vector_processing(articles)
    else:
        print("\n⚠️ Skipping vector processing test - no articles loaded")
        vector_success = False
    
    # Test 3: Full system integration
    system_success = test_full_system()
    
    # Summary
    print("\n📊 Test Results Summary")
    print("=" * 80)
    print(f"Firebase Loading: {'✅ PASS' if articles else '❌ FAIL'}")
    print(f"Vector Processing: {'✅ PASS' if vector_success else '❌ FAIL'}")
    print(f"System Integration: {'✅ PASS' if system_success else '❌ FAIL'}")
    
    overall_success = bool(articles) and vector_success and system_success
    print(f"\nOverall Result: {'✅ ALL TESTS PASSED' if overall_success else '❌ SOME TESTS FAILED'}")
    
    return overall_success

if __name__ == "__main__":
    success = main()
    sys.exit(0 if success else 1)