report-detect/archive/temp_scripts/quick_validation_test.py

"""
Quick validation test for CMA template matching improvements.
Tests a subset of PDFs to verify the improvements.
"""
import sys
import os
import json
import logging
import fitz
import numpy as np
import cv2
from pathlib import Path

logging.basicConfig(level=logging.INFO, format='%(message)s')
logger = logging.getLogger(__name__)

# Add parent dir to path
sys.path.insert(0, os.path.dirname(__file__))

# Import from our module
from cma_extraction_template_primary import extract_cma_code_fullpage

# Disable model source check
os.environ["DISABLE_MODEL_SOURCE_CHECK"] = "True"
from paddleocr import PaddleOCR

PDF_DIR = Path("src/test/resources/data/pdfs")
RESULTS_FILE = Path("src/test/resources/data/results.json")

def main():
    # Load expected results
    with open(RESULTS_FILE, 'r', encoding='utf-8') as f:
        expected_results = json.load(f)

    # Test specific PDFs
    test_pdfs = [
        "WTS2025-21283.pdf",
        "YDQ23_001838.pdf",
        "YDQ23_001850.pdf",
        "YDQ25_001875.pdf",
        "YDQ25_002294.pdf",
        "1.pdf",
    ]

    # Initialize OCR
    logger.info("Initializing PaddleOCR...")
    ocr = PaddleOCR(lang='ch')

    results = []

    logger.info("=" * 80)
    logger.info("QUICK VALIDATION TEST FOR CMA TEMPLATE MATCHING")
    logger.info("=" * 80)

    for pdf_name in test_pdfs:
        pdf_path = PDF_DIR / pdf_name
        if not pdf_path.exists():
            logger.warning(f"PDF not found: {pdf_name}")
            continue

        logger.info(f"\nProcessing: {pdf_name}")
        logger.info("-" * 80)

        # Extract first page
        doc = fitz.open(str(pdf_path))
        page = doc[0]
        mat = fitz.Matrix(300 / 72, 300 / 72)
        pix = page.get_pixmap(matrix=mat)
        img_data = pix.tobytes("png")
        img_array = np.frombuffer(img_data, dtype=np.uint8)
        page_img = cv2.imdecode(img_array, cv2.IMREAD_COLOR)
        doc.close()

        # Get expected CMA
        expected_cma = expected_results.get(pdf_name, {}).get('cma')

        # Process with template matching
        result = extract_cma_code_fullpage(page_img, ocr, None)

        # Record result
        success = result.get('success', False)
        extracted_cma = result.get('code')

        logger.info(f"  Expected CMA: {expected_cma}")
        logger.info(f"  Extracted CMA: {extracted_cma}")
        logger.info(f"  Status: {'✓ PASS' if (success and extracted_cma == expected_cma) else '✗ FAIL'}")

        results.append({
            'pdf': pdf_name,
            'expected': expected_cma,
            'extracted': extracted_cma,
            'success': success and extracted_cma == expected_cma
        })

    # Summary
    logger.info("\n" + "=" * 80)
    logger.info("SUMMARY")
    logger.info("=" * 80)

    passed = sum(1 for r in results if r['success'])
    total = len(results)

    for r in results:
        status = "✓ PASS" if r['success'] else "✗ FAIL"
        logger.info(f"{status} | {r['pdf']:30s} | {r['extracted'] or 'None':15s} (expected: {r['expected']})")

    logger.info("-" * 80)
    logger.info(f"Accuracy: {passed}/{total} ({passed/total*100:.1f}%)")
    logger.info("=" * 80)

    return passed, total

if __name__ == "__main__":
    try:
        passed, total = main()
        sys.exit(0 if passed == total else 1)
    except Exception as e:
        logger.error(f"Test failed: {e}")
        import traceback
        traceback.print_exc()
        sys.exit(1)