import cv2
import time
from PIL import Image
from inference import predict

CONF_THRESHOLD = 0.6
DECAY_SECONDS = 2.0

current_emotion = None
last_confident_ts = None

cap = cv2.VideoCapture(0, cv2.CAP_DSHOW)

cv2.namedWindow("Emotion Debug View", cv2.WINDOW_NORMAL)

print("Press 'q' to quit")

while True:
    ret, frame = cap.read()
    if not ret:
        print("Failed to grab frame")
        break

    rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
    pil_img = Image.fromarray(rgb)

    result = predict(pil_img)
    now = time.time()

    if result["confidence"] >= CONF_THRESHOLD:
        current_emotion = result["emotion"]
        last_confident_ts = now
    else:
        if last_confident_ts and (now - last_confident_ts > DECAY_SECONDS):
            current_emotion = None

    conf = round(result["confidence"], 2)
    label = f"{current_emotion if current_emotion else 'UNKNOWN'} | conf={conf}"

    state = "STABLE" if current_emotion else "DECAYED"
    cv2.putText(
        frame,
        f"State: {state} | {label}",
        (20, 40),
        cv2.FONT_HERSHEY_SIMPLEX,
        0.9,
        (0, 255, 0) if current_emotion else (0, 0, 255),
        2
    )

    cv2.imshow("Emotion Debug View", frame)

    if cv2.waitKey(1) & 0xFF == ord("q"):
        break

cap.release()
cv2.destroyAllWindows()