import easyocr
import numpy as np
import cv2
import re

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

def enhance_image(img):
    # Resize large image down to avoid OCR failure
    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)

    # Convert to gray
    gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)

    # Denoise
    gray = cv2.fastNlMeansDenoising(gray, h=15)

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

    # Contrast
    clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8,8))
    contrast = clahe.apply(sharp)

    return contrast

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)

        ocr_texts = [text for _, text, _ in results]
        weight_candidates = []

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

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

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

        best_weight, best_conf = sorted(weight_candidates, key=lambda x: -x[1])[0]
        return best_weight, round(best_conf * 100, 2), "\n".join(ocr_texts)

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