Mongoose pre-validate中间件中categories数组为空问题求助

我是JS、Mongoose和Express的新手，想在pre-validate中间件里处理categories数组——Schema要求数组是ObjectID类型，但为了用户方便（分类名称唯一），允许传入分类名称而非ObjectID。但发现调用Project.create前的路由里categories数组有预期的字符串值，到pre-validate中间件里就变成空数组了，其他字段都正常传递。

相关代码与日志 #

路由接收multipart form-data的代码 #

router.post('/', uploadFiles.fields([{ name: 'gszfile', maxCount: 1 }, { name: 'imagefile', maxCount: 1 }]), async (req, res) => {
  try {
    const { projectdata, versiondata } = req.body;
    const projectData = JSON.parse(projectdata); 
    console.log(`projectdata: ${JSON.stringify(projectData)}`)
    console.log(`projectData.categories: ${projectData.categories}`) //此处categories不为空
    const newProject = await Project.create(projectData);

Schema中的pre-validate中间件代码 #

project_schema.pre('validate', async function(next) {
    const categories = this.categories;
    console.log(`this: ${JSON.stringify(this)}`) //此处categories为空
    console.log(`categories: ${categories}`)
});

路由中的控制台日志 #

projectdata: {"title":"activator","shortdescription":"A stochastic model of the activator-substrate process","gszfile":"activator.gsz","author":"","description":"","howitworks":"","thingstonotice":"","thingstotry":"","extendingthemodel":"","relatedmodels":"","creditsandreferences":"","createdAt":"2009-04-20T00:00:00.000+01:00","updatedAt":"2009-04-20T00:00:00.000+01:00","categories":["Intern","FormalSystems"],"access":{"public":true,"readAccess":[],"writeAccess":[]},"imagefile":"activator.png"}
projectData.categories: Intern,FormalSystems

中间件中的控制台日志 #

this: {"title":"activator","shortdescription":"A stochastic model of the activator-substrate process","description":"","howitworks":"","thingstonotice":"","thingstotry":"","extendingthemodel":"","relatedmodels":"","creditsandreferences":"","categories":[],"tags":[],"imagefile":"activator.png","versions":[],"access":{"public":true,"readAccess":[],"writeAccess":[]},"_id":"65e98a15a58ee683b2bfa0e8","createdAt":"2009-04-19T23:00:00.000Z","updatedAt":"2009-04-19T23:00:00.000Z"}
categories:

原因分析 #

这种情况核心原因是Mongoose的Schema类型校验逻辑：你的categories字段被定义为[mongoose.Types.ObjectId]类型，但传入的是字符串数组。Mongoose在实例化文档时，会先执行类型转换与基础校验，无法转换为ObjectID的字符串会被直接过滤，而pre-validate钩子是在这一步之后才触发的，所以此时数组已经被清空。

解决办法 #

方法1：路由层提前转换（推荐） #

在调用Project.create之前，手动将分类名称转换为对应的ObjectID，确保传入的数组符合Schema类型要求：

// 假设存在Category模型，用于根据名称查询分类ID
const categoryIds = await Promise.all(
  projectData.categories.map(async name => {
    const category = await Category.findOne({ name });
    if (!category) {
      throw new Error(`分类 ${name} 不存在`); // 处理分类不存在的异常情况
    }
    return category._id;
  })
);
projectData.categories = categoryIds;

// 再执行创建操作
const newProject = await Project.create(projectData);

方法2：改用pre-save钩子处理 #

如果希望在模型层完成转换，可以改用pre('save')钩子（它在类型转换后、文档保存前触发），但需要确保转换后的元素都是有效ObjectID：

project_schema.pre('save', async function(next) {
  // 仅在新增文档且categories有值时处理
  if (this.isNew && this.categories.length > 0) {
    const categoryIds = await Promise.all(
      this.categories.map(async name => {
        const category = await Category.findOne({ name });
        return category?._id;
      })
    ).then(ids => ids.filter(id => id)); // 过滤掉不存在的分类ID

    this.categories = categoryIds;
  }
  next();
});

方法3：临时放宽Schema类型（不推荐） #

暂时将categories字段类型改为[mongoose.Schema.Types.Mixed]，让Mongoose不自动过滤字符串，待中间件转换完成后再改回ObjectID数组。这种方式会丢失Schema的类型约束，仅适合临时调试。

内容的提问来源于stack exchange，提问作者Dodo