verify_models.py

"""
Final verification test - Models are downloaded, let's test them!
"""

from transformers import pipeline
from PIL import Image
import time

def test_downloaded_models():
    """Test the downloaded models quickly"""
    print("🚀 TESTING DOWNLOADED MODELS")
    print("=" * 40)
    
    try:
        # Test 1: Load TrOCR pipeline (should be fast now)
        print("📥 Loading TrOCR model...")
        start_time = time.time()
        
        pipe = pipeline("image-to-text", model="microsoft/trocr-base-handwritten")
        
        load_time = time.time() - start_time
        print(f"✅ Model loaded in {load_time:.1f} seconds")
        
        # Test 2: Process test image
        print("\n🔍 Processing test image...")
        
        if Image.open("test_singtel_bill.png"):
            img = Image.open("test_singtel_bill.png")
            print("✅ Test image loaded successfully")
            
            # Process with AI
            start_time = time.time()
            result = pipe(img)
            process_time = time.time() - start_time
            
            extracted_text = result[0]['generated_text']
            
            print(f"✅ Processing completed in {process_time:.1f} seconds")
            print(f"\n📝 EXTRACTED TEXT:")
            print(f"'{extracted_text}'")
            
            # Test 3: Simple parsing
            print(f"\n🔧 TESTING BILL PARSING:")
            
            # Look for key information
            import re
            
            # Check for Singtel
            if "singtel" in extracted_text.lower():
                print("✅ 'Singtel' detected in text")
            else:
                print("❌ 'Singtel' not clearly detected")
            
            # Check for numbers (amounts, account numbers, etc.)
            numbers = re.findall(r'\d+', extracted_text)
            if numbers:
                print(f"✅ Numbers detected: {numbers}")
            else:
                print("❌ No numbers detected")
            
            # Check for currency patterns
            currency = re.findall(r'\$[\d.]+', extracted_text)
            if currency:
                print(f"✅ Currency amounts: {currency}")
            else:
                print("⚠️  No currency patterns detected")
            
            print(f"\n🎉 SUCCESS! Your TrOCR model is working!")
            return True
            
        else:
            print("❌ Could not load test image")
            return False
            
    except Exception as e:
        print(f"❌ Error: {e}")
        return False

def test_real_bill_processing():
    """Show how to process a real bill"""
    print(f"\n🎯 READY FOR REAL BILLS!")
    print("=" * 30)
    print("To process your Singtel bill:")
    print()
    print("1. 📸 Take a clear photo of your bill")
    print("2. 💾 Save it as 'my_bill.jpg' in this folder")
    print("3. 🚀 Run this code:")
    print()
    print("```python")
    print("from transformers import pipeline")
    print("from PIL import Image")
    print()
    print("# Load the model (fast now - already downloaded!)")
    print("pipe = pipeline('image-to-text', model='microsoft/trocr-base-handwritten')")
    print()
    print("# Process your bill")
    print("image = Image.open('my_bill.jpg')")
    print("result = pipe(image)")
    print("text = result[0]['generated_text']")
    print()
    print("print(f'Extracted: {text}')")
    print("```")
    print()
    print("4. 🔧 Use singtel_scanner.py for advanced parsing!")

if __name__ == "__main__":
    print("FINAL MODEL VERIFICATION")
    print("Models are downloaded - testing now!")
    print("=" * 50)
    
    success = test_downloaded_models()
    
    if success:
        test_real_bill_processing()
        print(f"\n✅ COMPLETE! Your Singtel Bill Scanner is ready to use!")
        print(f"⚡ All future runs will be instant (models cached)")
    else:
        print(f"\n❌ Something went wrong. Check the error above.")
    
    print("\n" + "=" * 50)
    input("Press Enter to finish...")