@app.route('/video_feed')
def video_feed():
    return Response(generate_frames(), 
                    mimetype='multipart/x-mixed-replace; boundary=frame')

def generate_frames():
    camera = cv2.VideoCapture("http://192.168.144.241:81/stream")
    
    while True:
        success, frame = camera.read()
        if not success:
            break
        
        ret, buffer = cv2.imencode('.jpg', frame)
        frame = buffer.tobytes()
        
        yield (b'--frame\r\n'
               b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n')
        
        # 5초마다 프레임 저장 → AI 분석 트리거
        if int(time.time()) % 5 == 0:
            cv2.imwrite('static/assets/img/pre_picture.jpg', img)
            client_socket.send("recog".encode())

# Roboflow API 호출
rf = Roboflow(api_key="YOUR_API_KEY")
project = rf.workspace().project("lg-cns")
model = project.version(2).model

# 객체 감지 (confidence 40%, overlap 30%)
predict_json = model.predict("pre_picture.jpg", 
                              confidence=40, 
                              overlap=30).json()

# 결과 저장
with open("prediction.json", "w") as json_file:
    json.dump(predict_json, json_file)

import seaborn as sns
import matplotlib.pyplot as plt

# 승객 위치를 그리드로 변환
heatmap_data = np.zeros((desired_height // 10, desired_width // 10))

for result in results:
    x, y = int(result[1]), int(result[2])
    grid_x, grid_y = x // 10, y // 10
    heatmap_data[grid_y, grid_x] += 1

# 히트맵 생성
plt.figure(figsize=(12, 8))
sns.heatmap(heatmap_data, cmap="YlOrRd", cbar=True)
plt.title("Congestion Heatmap")
plt.savefig('static/assets/img/heatmap.jpg')

client_sockets = []

def threaded(client_socket, addr):
    while True:
        data = client_socket.recv(1024)
        
        # 다른 클라이언트에게 브로드캐스트
        for client in client_sockets: 
            if client != client_socket:
                client.send(data)
        
        if data. decode() == "recog":
            # AI 분석 실행
            analyze_frame()

# python/ai_server.py
rf = Roboflow(api_key="YOUR_API_KEY_HERE")
project = rf.workspace().project("lg-cns")
model = project.version(2).model

# 메인 페이지
@app.route('/')
def index():
    # 초기화:  빈 이미지로 설정
    shutil.copy('static/assets/img/white. jpg', 
                'static/assets/img/saved_picture.jpg')
    return render_template('index.html')

# 비디오 스트리밍
@app.route('/video_feed')
def video_feed():
    return Response(generate_frames(), 
                    mimetype='multipart/x-mixed-replace; boundary=frame')

# 결과 이미지 제공
@app.route('/get_image/<string:no>')
def get_image(no):
    if no == "1":
        image_path = 'static/assets/img/saved_picture.jpg'  # Bbox 이미지
    elif no == "2":
        image_path = 'static/assets/img/heatmap.jpg'  # 히트맵
    return send_file(image_path, mimetype='image/jpg')

def threaded(client_socket, addr):
    while True:
        data = client_socket.recv(1024)
        
        if data.decode() == "recog":
            # 1.  Roboflow API 호출
            predict_json = model.predict("pre_picture. jpg", 
                                          confidence=40, 
                                          overlap=30).json()
            
            # 2. JSON 저장
            with open("prediction.json", "w") as f:
                json.dump(predict_json, f)
            
            # 3. 이미지 로드
            image = cv2.imread('static/assets/img/pre_picture.jpg')
            
            # 4. Bounding Box 그리기
            for prediction in predict_json["predictions"]:
                x, y = prediction["x"], prediction["y"]
                w, h = prediction["width"], prediction["height"]
                
                x1, y1 = int(x - w/2), int(y - h/2)
                x2, y2 = int(x + w/2), int(y + h/2)
                
                cv2.rectangle(image, (x1, y1), (x2, y2), (0, 255, 0), 2)
                cv2.putText(image, "Person", (x1, y1-10), 
                            cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0,255,0), 2)
            
            # 5. 저장
            cv2.imwrite('static/assets/img/saved_picture.jpg', image)
            
            # 6. 히트맵 생성
            generate_heatmap(predict_json["predictions"])
            
            # 7. 응답 전송
            client_socket. send(f"analysis:{person_num}: level". encode())

# 1. API 키 확인
rf = Roboflow(api_key="YOUR_VALID_API_KEY")

# 2. 프로젝트명 확인
project = rf.workspace().project("YOUR_PROJECT_NAME")

# 3. 모델 버전 확인
model = project.version(YOUR_VERSION).model

# 4. 네트워크 확인
# 인터넷 연결 필요 (Roboflow 클라우드 API)

# 예외 처리 추가
try:
    client_socket.send(message.encode())
except BrokenPipeError:
    print("Client disconnected")
    client_sockets.remove(client_socket)
    client_socket.close()