ocr_engine.py

import pytesseract
import numpy as np
import cv2
import re
import logging
from PIL import Image

# Setup logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')


def preprocess_strong(img):
    """Sharpen and enhance contrast for blurry weight images."""
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

    # Resize up for OCR (scale x2)
    h, w = gray.shape
    gray = cv2.resize(gray, (w * 2, h * 2), interpolation=cv2.INTER_CUBIC)

    # CLAHE for contrast enhancement
    clahe = cv2.createCLAHE(clipLimit=3.0, tileGridSize=(8, 8))
    enhanced = clahe.apply(gray)

    # Strong sharpening
    kernel = np.array([[0, -1, 0],
                       [-1, 5, -1],
                       [0, -1, 0]])
    sharpened = cv2.filter2D(enhanced, -1, kernel)

    return sharpened


def extract_weight_from_image(pil_img):
    """Extract weight from an image using multiple Tesseract strategies."""
    try:
        img = np.array(pil_img)
        img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)

        processed = preprocess_strong(img)

        # OCR configs
        configs = [
            r'--oem 3 --psm 6 -c tessedit_char_whitelist=0123456789.',
            r'--oem 3 --psm 11 -c tessedit_char_whitelist=0123456789.'
        ]

        for config in configs:
            raw_text = pytesseract.image_to_string(processed, config=config)
            logging.info(f"[Tesseract Output {config}] Raw text: {raw_text}")
            cleaned = raw_text.strip().replace('\n', '').replace(' ', '')
            cleaned = re.sub(r"[^\d.]", "", cleaned)
            if cleaned.count('.') > 1:
                cleaned = cleaned.replace('.', '', cleaned.count('.') - 1)
            if cleaned.startswith('.'):
                cleaned = '0' + cleaned

            if cleaned and re.fullmatch(r"\d*\.?\d*", cleaned):
                value = float(cleaned)
                if 0.001 <= value <= 5000:
                    return str(round(value, 2)), 90.0

        return "Not detected", 0.0

    except Exception as e:
        logging.error(f"OCR failed: {e}")
        return "Not detected", 0.0