> ## Documentation Index
> Fetch the complete documentation index at: https://all.mimedal.cn/llms.txt
> Use this file to discover all available pages before exploring further.

# 编译/解释器

> 说明如何把实验目标和设备接口整理成可验证的执行步骤。

编译/解释器读取科学目标、对象初始状态和 FSP 设备定义，将设备操作转换为类型化函数，并生成候选中间程序。它不直接调用物理设备。

## 输入

```json theme={null}
{
  "goal": {
    "name": "向目标容器加入液体",
    "success_criteria": [
      "实际加液量为 0.5 mL",
      "容器状态可确认"
    ]
  },
  "initial_objects": [
    {
      "container_id": "container-01",
      "object_type_id": "container.centrifuge_tube",
      "container_status": "无盖",
      "sample_status": "空",
      "revision": "object-21"
    },
    {
      "container_id": "source-01",
      "object_type_id": "container.reservoir",
      "sample_status": "液体",
      "sample_volume_mL": 10,
      "revision": "source-8"
    }
  ],
  "fsp_definitions": [
    {
      "device_id": "device-01",
      "revision": "12",
      "operations": [
        "add_liquid"
      ],
      "objects": [
        "container.centrifuge_tube"
      ],
      "system_states": [
        "idle",
        "busy",
        "fault"
      ]
    }
  ]
}
```

## 处理过程

1. 校验操作是否完整定义对象要求、参数 Schema、执行前条件、系统安全限制和响应结构；
2. 校验实验目标所需容器与操作兼容；
3. 将参数类型、范围、枚举和状态变化编译为类型化函数契约；
4. 组织带依赖关系的候选中间程序；
5. 固定设备定义、对象状态和系统状态版本号。

## 输出

```json theme={null}
{
  "program_id": "program-001",
  "status": "compiled",
  "definition_revisions": {
    "device-01": "12"
  },
  "steps": [
    {
      "step_id": "step-1",
      "device_id": "device-01",
      "object": {
        "containers": [
          {
            "container_id": "source-01",
            "role": "source",
            "expected_revision": "source-8"
          },
          {
            "container_id": "container-01",
            "role": "target",
            "expected_revision": "object-21",
            "before_state": {
              "container_status": "无盖"
            }
          }
        ]
      },
      "operation": {
        "operation_id": "add_liquid",
        "contract_version": "1",
        "parameters": {
          "source_container_id": "source-01",
          "volume_mL": 0.5
        }
      },
      "system": {
        "expected_definition_revision": "12",
        "expected_system_revision": "system-21",
        "expected_device_state": "idle",
        "execution_mode": "async"
      }
    }
  ],
  "diagnostics": []
}
```

编译/解释器必须拒绝未在 FSP 定义中声明的操作或参数。`compiled` 只表示结构通过，不表示已仿真或已执行。

编译结果必须写明使用哪台设备、哪一版操作和设备定义，并记录步骤顺序、所需资源和完成条件。`implemented` 或 `ready` 的缓存值不能当成执行许可。

中间步骤使用与请求相同的三部分格式。执行器创建执行任务时，再生成并保存幂等键；编译阶段不受理设备动作。