import mediapipe as mp
import cv2

class KeypointExtractor:
    def __init__(self):
        self.mp_drawing = mp.solutions.drawing_utils
        
        # hands extractor
        self.hands = mp.solutions.hands.Hands(
            min_detection_confidence=0.5,
            min_tracking_confidence=0.5,
            max_num_hands=2
        )

        # pose extractor
        self.pose = mp.solutions.pose.Pose(
            min_detection_confidence=0.5,
            min_tracking_confidence=0.5,
            model_complexity=2
        )

    def extract(self, image, video):
        # load video
        pass


    def extract_from_frame(self, image):
        # Convert the BGR image to RGB and process it with MediaPipe Pose.
        hand_results = self.hands.process(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))

        # Draw the hand annotations on the image.
        draw_image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
        
        draw_image.flags.writeable = False

        for hand_landmarks in hand_results.multi_hand_landmarks:
            self.mp_drawing.draw_landmarks(
                draw_image, hand_landmarks, mp.solutions.hands.HAND_CONNECTIONS)
    
        pose_results = self.pose.process(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))

        self.mp_drawing.draw_landmarks(
            draw_image, pose_results.pose_landmarks, mp.solutions.pose.POSE_CONNECTIONS)


        draw_image.flags.writeable = True
        draw_image = cv2.cvtColor(draw_image, cv2.COLOR_RGB2BGR)

        return draw_image


ke = KeypointExtractor()
image = cv2.imread('data/test_photo.jpg')

image = ke.extract_from_frame(image)

# save image
cv2.imwrite('test_output.jpg', image)