#!/usr/bin/env python
"""
从已有的pkl结果文件进行评估，跳过推理步骤
"""
import os
import pickle
import sys
import torch

# 设置环境
os.environ['PATH'] = '/opt/conda/bin:' + os.environ.get('PATH', '')
os.environ['LD_LIBRARY_PATH'] = '/opt/conda/lib/python3.8/site-packages/torch/lib:/opt/conda/lib:/usr/local/cuda/lib64:' + os.environ.get('LD_LIBRARY_PATH', '')
os.environ['PYTHONPATH'] = '/workspace/bevfusion:' + os.environ.get('PYTHONPATH', '')

def main():
    import mmcv
    from mmcv import Config
    from mmdet3d.datasets import NuScenesDataset

    # 加载配置
    config_file = 'configs/nuscenes/det/transfusion/secfpn/camera+lidar/swint_v0p075/multitask_BEV2X_phase4b_rmtppad_segmentation.yaml'
    cfg = Config.fromfile(config_file)

    # 设置为测试模式
    cfg.data.val.test_mode = True

    # 创建数据集
    print("正在创建数据集...")
    from mmdet3d.datasets import build_dataset

    # 构建完整的val配置 (参考nuscenes/default.yaml)
    val_config = {
        'type': 'NuScenesDataset',
        'dataset_root': cfg.dataset_root,
        'ann_file': cfg.dataset_root + 'nuscenes_infos_val.pkl',
        'pipeline': cfg.evaluation.pipeline,
        'object_classes': cfg.object_classes,
        'map_classes': cfg.map_classes,
        'modality': {
            'use_lidar': True,
            'use_camera': True,
            'use_radar': False,
            'use_map': False,
            'use_external': False
        },
        'test_mode': True,
        'use_valid_flag': False,
        'box_type_3d': 'LiDAR',
        'load_interval': cfg.data.val.load_interval if hasattr(cfg.data.val, 'load_interval') else 1
    }

    dataset = build_dataset(val_config)
    print(f"数据集大小: {len(dataset)}")

    # 加载已有的结果
    results_file = '/data/eval_fast/epoch1_fast_20251119_133104/fast_results.pkl'
    print(f"正在加载结果文件: {results_file}")

    with open(results_file, 'rb') as f:
        results = pickle.load(f)

    print(f"结果文件包含 {len(results)} 个样本")

    # 检查结果格式
    if results:
        sample = results[0]
        print(f"样本键: {list(sample.keys())}")
        if 'boxes_3d' in sample:
            print(f"检测框数量: {len(sample['boxes_3d'])}")
        if 'masks_bev' in sample:
            print(f"BEV分割mask形状: {sample['masks_bev'].shape}")

    # 进行评估
    print("开始评估...")
    eval_kwargs = dict(
        metric=['bbox', 'map'],
        save_best=None,
        rule=None,
        logger=None
    )

    try:
        eval_results = dataset.evaluate(results, **eval_kwargs)
        print("\n" + "="*50)
        print("评估结果:")
        print("="*50)
        for key, value in eval_results.items():
            if isinstance(value, float):
                print("30")
            else:
                print(f"{key}: {value}")
    except Exception as e:
        print(f"评估失败: {e}")
        import traceback
        traceback.print_exc()

if __name__ == '__main__':
    main()