Mongoose嵌套populate的select不生效：post_comments仍显示_id与__v

问题 #

我有Post、User、Comment三个Mongoose Schema，结构如下：

postSchema = mongoose.Schema({
    post_title,
    post_description,
    post_timestamp,
    post_author:{
        type:mongoose.Schema.Types.ObjectId, 
        ref: "User"
    },
    post_likes:{
        type: [userSchema.schema]
    },
    post_comments:{
        type: [commentSchema.schema]
    }
})

userSchema = mongoose.Schema({
    username,
    email,
    password,
    date
})

commentSchema = mongoose.Schema({
    comment_description,
    comment_timestamp,
    comment_author:{
        type:mongoose.Schema.Types.ObjectId, 
        ref: "User",
        require:true,
    }
})

我希望查询返回的Post数据里，post_comments数组中的每个评论对象移除_id和__v字段，期望格式如下：

{
        "_id": "638bdf38b2ad86abe32455fd",
        "post_title": "This is a test post",
        "post_description": "abc...",
        "post_timestamp": "2022-12-03T23:41:04.292Z",
        "post_author": {
            "username": "test10"
        },
        "post_comments": [
            {
                "comment_description": "This is my first comment for this new post",
                "comment_timestamp": "2022-12-03T23:51:24.306Z",
                "comment_author": {
                    "username": "test10"
                }
            },
            {
                "comment_description": "This is my first comment for this new post",
                "comment_timestamp": "2022-12-04T10:06:22.091Z",
                "comment_author": {
                    "username": "test11"
                }
            }
      ]
}

但实际返回的post_comments仍包含_id和__v，格式如下：

{
        "_id": "638bdf38b2ad86abe32455fd",
        "post_title": "This is a test post",
        "post_description": "abc...",
        "post_timestamp": "2022-12-03T23:41:04.292Z",
        "post_author": {
            "username": "test10"
        },
        "post_comments": [
            {
                "comment_description": "This is my first comment for this new post",
                "comment_timestamp": "2022-12-03T23:51:24.306Z",
                "comment_author": {
                    "username": "test10"
                },
                "_id": "638be126b9fe7d1b4823882c",
                "__v": 0
            },
            {
                "comment_description": "This is my first comment for this new post",
                "comment_timestamp": "2022-12-04T10:06:22.091Z",
                "comment_author": {
                    "username": "test11"
                },
                "_id": "638c738ab63be98a89d2cb76",
                "__v": 0
            }
      ]
}

我用的查询代码如下：

const posts = await Post.find({}, {
            post_title: 1,
            post_description: 1,
            post_timestamp: 1
        })
        .populate({
            path: "post_author",
            model: "User",
            select: "-_id username",
        })
        .populate({
            path: "post_comments",
            model: "Comment",
            select: "-_id -__v comment_description comment_timestamp", // <-- 这里不生效
            populate: {
                path: "comment_author",
                model: "User",
                select: "-_id:0 username"
            }
        })

针对post_comments的populate配置里的select语句完全不起作用，试了多种写法都无效，求问题原因和解决方案。

问题原因及解决方案 #

原因 #

你的postSchema里，post_comments是直接嵌入commentSchema.schema的数组，而非引用Comment模型的ObjectId数组。Mongoose的populate方法仅对**引用类型（ObjectId + ref关联）**字段生效，对于直接嵌入的子文档，populate配置会被忽略，所以select规则完全不生效。

解决方案 #

有两种可行方案：

方案1：修改Schema，将post_comments改为引用类型（推荐）

把postSchema中的post_comments改为关联Comment模型的ObjectId数组，这样populate就能正常工作：

postSchema = mongoose.Schema({
    post_title,
    post_description,
    post_timestamp,
    post_author:{
        type:mongoose.Schema.Types.ObjectId, 
        ref: "User"
    },
    post_likes:{
        type: [userSchema.schema]
    },
    post_comments:{
        type: [mongoose.Schema.Types.ObjectId], // 改为ObjectId数组
        ref: "Comment" // 关联Comment模型
    }
})

修改后，修正内层populate的select写法（-_id:0是错误语法，应写成-_id username），原查询代码即可生效：

const posts = await Post.find({}, {
            post_title: 1,
            post_description: 1,
            post_timestamp: 1
        })
        .populate({
            path: "post_author",
            model: "User",
            select: "-_id username",
        })
        .populate({
            path: "post_comments",
            model: "Comment",
            select: "-_id -__v comment_description comment_timestamp",
            populate: {
                path: "comment_author",
                model: "User",
                select: "-_id username" // 修正语法
            }
        })

方案2：查询时直接投影子文档字段（适合不想修改Schema的场景）

如果不想改动Schema结构，可在Post.find()的投影参数中指定post_comments要保留/移除的字段：

const posts = await Post.find({}, {
            post_title: 1,
            post_description: 1,
            post_timestamp: 1,
            post_author: 1,
            "post_comments.comment_description": 1,
            "post_comments.comment_timestamp": 1,
            "post_comments.comment_author": 1,
            "post_comments._id": 0, // 移除子文档的_id
            "post_comments.__v": 0  // 移除子文档的__v
        })
        .populate({
            path: "post_author",
            model: "User",
            select: "-_id username",
        })
        .populate({
            path: "post_comments.comment_author",
            model: "User",
            select: "-_id username"
        })

这种方法无需修改Schema，但投影写法较繁琐，子文档字段较多时维护成本高。

内容的提问来源于stack exchange，提问作者5.9 protons