#!/usr/bin/env python3
"""
直接測試人臉識別功能（不通過 HTTP API）
"""

import os
import sys
import numpy as np
import cv2

# 添加項目根目錄到 Python 路徑
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))


def test_direct_face_processing():
    """直接測試人臉處理功能"""
    print("=" * 60)
    print("直接人臉處理測試")
    print("=" * 60)

    try:
        # 1. 測試人臉註冊
        print("\n1. 測試人臉註冊...")
        from scripts.face_registration import FaceRegistration

        # 創建測試圖像
        img = np.zeros((480, 640, 3), dtype=np.uint8)
        img.fill(200)
        cv2.circle(img, (320, 240), 100, (255, 200, 150), -1)
        cv2.circle(img, (280, 200), 20, (0, 0, 0), -1)
        cv2.circle(img, (360, 200), 20, (0, 0, 0), -1)
        cv2.ellipse(img, (320, 280), (40, 20), 0, 0, 360, (0, 0, 0), -1)

        test_image_path = "/tmp/direct_test_face.jpg"
        cv2.imwrite(test_image_path, img)
        print(f"✅ 創建測試圖像: {test_image_path}")

        # 初始化註冊器
        registration = FaceRegistration()
        registration.load_models(use_mps=False)

        # 註冊人臉
        result = registration.register_face(
            image_path=test_image_path,
            name="Direct Test Person",
            metadata={"test": True, "method": "direct"},
        )

        if result["success"]:
            print("✅ 人臉註冊成功")
            print(f"  - Face ID: {result.get('face_id')}")
            print(f"  - 嵌入向量長度: {len(result.get('embedding', []))}")

            # 保存嵌入向量
            embedding = result.get("embedding", [])
            if embedding:
                np.save("/tmp/test_embedding.npy", embedding)
        else:
            print(f"❌ 人臉註冊失敗: {result.get('message')}")

        # 2. 測試人臉識別處理器
        print("\n2. 測試人臉識別處理器...")
        from scripts.face_recognition_processor import FaceRecognitionProcessor

        processor = FaceRecognitionProcessor(
            enable_recognition=True, enable_tracking=True, enable_clustering=True
        )
        processor.load_models(use_mps=False)

        # 讀取圖像
        image = cv2.imread(test_image_path)

        # 檢測人臉
        detections = processor.detect_faces(image)
        print(f"✅ 檢測到 {len(detections)} 個人臉")

        if len(detections) > 0:
            for i, detection in enumerate(detections):
                print(f"\n  人臉 {i + 1}:")
                print(
                    f"    - 位置: x={detection['x']}, y={detection['y']}, width={detection['width']}, height={detection['height']}"
                )
                print(f"    - 置信度: {detection['confidence']:.4f}")

                if "embedding" in detection and detection["embedding"] is not None:
                    embedding = detection["embedding"]
                    if hasattr(embedding, "shape"):
                        print(f"    - 嵌入向量形狀: {embedding.shape}")
                    else:
                        print(f"    - 嵌入向量長度: {len(embedding)}")

                if "attributes" in detection:
                    print(f"    - 屬性: {detection['attributes']}")

        # 3. 測試數據庫操作
        print("\n3. 測試數據庫操作...")
        import psycopg2
        from psycopg2.extras import Json

        conn = psycopg2.connect(
            host="localhost",
            port=5432,
            database="momentry",
            user="accusys",
            password="accusys",
        )

        cursor = conn.cursor()

        # 插入測試數據
        print("  插入測試檢測記錄...")
        cursor.execute(
            """
            INSERT INTO face_detections 
            (video_uuid, frame_number, timestamp_secs, face_id, x, y, width, height, confidence, attributes)
            VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
            RETURNING id;
        """,
            (
                "direct_test_video",
                1,
                0.0,
                "direct_test_face_001",
                100,
                100,
                200,
                200,
                0.95,
                Json({"test": True, "method": "direct", "source": "python_test"}),
            ),
        )

        detection_id = cursor.fetchone()[0]
        print(f"  ✅ 插入成功，ID: {detection_id}")

        # 查詢測試
        print("  查詢測試數據...")
        cursor.execute("SELECT COUNT(*) as total_faces FROM face_detections")
        total_faces = cursor.fetchone()[0]

        cursor.execute("SELECT COUNT(*) as total_identities FROM face_identities")
        total_identities = cursor.fetchone()[0]

        print("  ✅ 數據庫統計:")
        print(f"     - 總人臉檢測記錄: {total_faces}")
        print(f"     - 總人臉身份: {total_identities}")

        # 測試向量搜索
        print("  測試向量搜索...")
        if embedding:
            cursor.execute(
                """
                SELECT * FROM find_similar_faces(
                    %s::vector,
                    0.5,  -- similarity_threshold
                    5     -- limit_count
                );
            """,
                (embedding,),
            )

            similar_faces = cursor.fetchall()
            print(f"  ✅ 找到 {len(similar_faces)} 個相似人臉")

            for face in similar_faces:
                print(f"    - {face[0]}: {face[1]} (相似度: {face[2]:.4f})")

        # 清理測試數據
        print("\n4. 清理測試數據...")
        cursor.execute(
            "DELETE FROM face_detections WHERE video_uuid = 'direct_test_video';"
        )
        cursor.execute("DELETE FROM face_identities WHERE face_id LIKE 'test_%';")
        conn.commit()

        cursor.close()
        conn.close()

        print("✅ 測試數據清理完成")

        # 清理文件
        os.remove(test_image_path)
        if os.path.exists("/tmp/test_embedding.npy"):
            os.remove("/tmp/test_embedding.npy")

        print("\n" + "=" * 60)
        print("✅ 直接測試完成！所有功能正常工作。")
        print("=" * 60)

        return True

    except Exception as e:
        print(f"❌ 測試失敗: {e}")
        import traceback

        traceback.print_exc()
        return False


def test_mps_performance():
    """測試 MPS 性能"""
    print("\n" + "=" * 60)
    print("MPS 性能測試")
    print("=" * 60)

    try:
        import time
        from scripts.face_recognition_processor import FaceRecognitionProcessor

        # 創建測試圖像
        test_image = np.random.randint(0, 255, (640, 480, 3), dtype=np.uint8)

        # 測試 CPU
        print("測試 CPU 性能...")
        cpu_processor = FaceRecognitionProcessor()
        cpu_processor.load_models(use_mps=False)

        start_time = time.time()
        cpu_detections = cpu_processor.detect_faces(test_image)
        cpu_time = time.time() - start_time

        print(f"✅ CPU 檢測時間: {cpu_time:.3f} 秒")
        print(f"✅ CPU 檢測到 {len(cpu_detections)} 個人臉")

        # 測試 MPS（如果可用）
        print("\n測試 MPS 性能...")
        try:
            mps_processor = FaceRecognitionProcessor()
            mps_processor.load_models(use_mps=True)

            start_time = time.time()
            mps_detections = mps_processor.detect_faces(test_image)
            mps_time = time.time() - start_time

            print(f"✅ MPS 檢測時間: {mps_time:.3f} 秒")
            print(f"✅ MPS 檢測到 {len(mps_detections)} 個人臉")

            if mps_time > 0:
                speedup = cpu_time / mps_time
                print(f"✅ MPS 加速比: {speedup:.2f}x")
            else:
                print("⚠️  MPS 時間測量不準確")

        except Exception as e:
            print(f"⚠️  MPS 測試失敗: {e}")
            print("⚠️  回退到 CPU 模式")

        print("\n" + "=" * 60)
        print("✅ 性能測試完成")
        print("=" * 60)

        return True

    except Exception as e:
        print(f"❌ 性能測試失敗: {e}")
        return False


def main():
    """主測試函數"""
    print("人臉識別系統直接測試")
    print("=" * 60)

    # 測試直接處理
    if not test_direct_face_processing():
        print("❌ 直接處理測試失敗")
        return 1

    # 測試 MPS 性能
    if not test_mps_performance():
        print("⚠️  性能測試有問題，但系統仍可工作")

    print("\n" + "=" * 60)
    print("🎉 所有測試完成！系統功能正常。")
    print("=" * 60)
    print("\n系統狀態:")
    print("  ✅ 人臉註冊功能")
    print("  ✅ 人臉檢測功能")
    print("  ✅ 數據庫操作")
    print("  ✅ MPS 加速支援")
    print("\n下一步:")
    print("1. 配置 API 密鑰進行 HTTP 測試")
    print("2. 使用實際視頻進行測試")
    print("3. 部署到生產環境")

    return 0


if __name__ == "__main__":
    sys.exit(main())