# signature_verification/signature_detector.py
import cv2
import numpy as np

class SignatureDetector:
    """
    Detects and crops a signature from a prescription image using refined OpenCV contour analysis.
    This method is fast, offline, and does not require a pre-trained model.
    """
    def crop(self, img: np.ndarray) -> np.ndarray | None:
        if img is None: return None

        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
        blurred = cv2.GaussianBlur(gray, (5, 5), 0)
        _, thresh = cv2.threshold(blurred, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)

        contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
        if not contours: return None

        height, width = img.shape[:2]
        significant_contours = []
        for cnt in contours:
            x, y, w, h = cv2.boundingRect(cnt)
            # Filter for contours that are reasonably large and in the bottom 70% of the page
            if w > 50 and h > 15 and y > height * 0.3:
                if w < width * 0.8 and h < height * 0.4:
                    significant_contours.append(cnt)

        if not significant_contours: return None

        largest_contour = max(significant_contours, key=cv2.contourArea)
        x, y, w, h = cv2.boundingRect(largest_contour)
        
        padding = 15
        x1 = max(0, x - padding)
        y1 = max(0, y - padding)
        x2 = min(width, x + w + padding)
        y2 = min(height, y + h + padding)
        
        return img[y1:y2, x1:x2]