report-detect/archive/temp_scripts/simple_test.py

66 lines
1.7 KiB
Python

"""
Simple test to see what CMA code is extracted
"""
import sys
import os
os.environ["DISABLE_MODEL_SOURCE_CHECK"] = "True"
# Clear cache
for module in list(sys.modules.keys()):
if 'cma_extraction' in module or 'test_accuracy' in module:
del sys.modules[module]
import fitz
import numpy as np
import cv2
from paddleocr import PaddleOCR
# Import CMA extraction
from cma_extraction_template_primary import extract_cma_code_fullpage, imread_unicode
pdf_path = "src/test/resources/data/pdfs/YDQ23_001838.pdf"
print(f"Processing: {pdf_path}")
print("=" * 80)
# Extract page
doc = fitz.open(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()
print(f"Page size: {page_img.shape}")
# Initialize OCR
print("\nInitializing OCR...")
ocr = PaddleOCR(lang='ch')
# Extract CMA
print("\nExtracting CMA code...")
output_dir = "test_debug"
os.makedirs(output_dir, exist_ok=True)
result = extract_cma_code_fullpage(page_img, ocr, output_dir=output_dir)
print("\n" + "=" * 80)
print("RESULT")
print("=" * 80)
print(f"Success: {result.get('success')}")
print(f"CMA Code: {result.get('code')}")
print(f"Confidence: {result.get('confidence')}")
print(f"Method: {result.get('method')}")
print(f"Position: {result.get('position')}")
print(f"Box: {result.get('box')}")
if result.get('code'):
if result['code'] == '210020349096':
print("\n✓ CORRECT CMA CODE EXTRACTED!")
elif result['code'] == '440023010130':
print("\n✗ WRONG CODE (440023010130) - This is the report number, not CMA!")
else:
print(f"\n? UNEXPECTED CODE: {result['code']}")