ocr_engine.py

import easyocr
import numpy as np
import cv2
import re

reader = easyocr.Reader(['en'], gpu=False)

def enhance_image(img):
    max_dim = 1000
    height, width = img.shape[:2]
    if max(height, width) > max_dim:
        scale = max_dim / max(height, width)
        img = cv2.resize(img, None, fx=scale, fy=scale, interpolation=cv2.INTER_AREA)

    gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
    gray = cv2.fastNlMeansDenoising(gray, h=15)

    kernel = np.array([[0, -1, 0], [-1, 5, -1], [0, -1, 0]])
    sharp = cv2.filter2D(gray, -1, kernel)

    clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
    enhanced = clahe.apply(sharp)

    return enhanced

def extract_weight_from_image(pil_img):
    try:
        img = np.array(pil_img)
        enhanced = enhance_image(img)

        results = reader.readtext(enhanced)
        print("DEBUG OCR RESULTS:", results)

        if not results:
            return "No text detected", 0.0, "OCR returned empty list"

        all_texts = []
        weight_candidates = []

        for _, text, conf in results:
            original = text
            cleaned = text.lower().strip()
            cleaned = cleaned.replace(",", ".")
            cleaned = cleaned.replace("o", "0").replace("O", "0")
            cleaned = cleaned.replace("s", "5").replace("S", "5")
            cleaned = cleaned.replace("g", "9").replace("G", "6")
            cleaned = cleaned.replace("kg", "").replace("kgs", "")
            cleaned = re.sub(r"[^\d\.]", "", cleaned)

            all_texts.append(f"{original} → {cleaned} (conf: {round(conf, 2)})")

            if re.fullmatch(r"\d{2,4}(\.\d{1,3})?", cleaned):
                weight_candidates.append((cleaned, conf))

        if not weight_candidates:
            return "Not detected", 0.0, "\n".join(all_texts)

        best_weight, best_conf = sorted(weight_candidates, key=lambda x: -x[1])[0]

        if "." in best_weight:
            parts = best_weight.split(".")
            parts[0] = parts[0].lstrip("0") or "0"
            best_weight = ".".join(parts)
        else:
            best_weight = best_weight.lstrip("0") or "0"

        return best_weight, round(best_conf * 100, 2), "\n".join(all_texts)

    except Exception as e:
        return f"Error: {str(e)}", 0.0, "OCR failed"