数据准备

[
    {
        "name": {
            "first_name": "qingquan",
            "last_name": "zeng"
        },
        "balance": 100
    },
    {
        "name": {
            "first_name": "fengxia",
            "last_name": "yu"
        },
        "balance": 200
    }
]

插入数据

db.accounts.insert([{"name": {"first_name": "qingquan","last_name": "zeng"},"balance": 100},{"name": {"first_name": "fengxia","last_name": "yu"},"balance": 200}])

数据查询

$project

# aggregate 中的 $project 除了可以实现投影效果，还直接使用了一个不存在的字段 client_name ，相当于 mysql 中的 as 语法
> db.accounts.aggregate([{
... $project:{
... _id:0,
... balance:1,
... client_name:"$name.first_name"
... }
... }]);
{ "balance" : 100, "client_name" : "qingquan" }
{ "balance" : 200, "client_name" : "fengxia" }

# 由于 middle_name 不存在，产生的结果就为 null 了
> db.accounts.aggregate([{
... $project:{
... _id:0,
... balance:1,
... name_arr:["$name.first_name","$name.middle_name","$name.first_name"]
... }
... }]);
{ "balance" : 100, "name_arr" : [ "qingquan", null, "qingquan" ] }
{ "balance" : 200, "name_arr" : [ "fengxia", null, "fengxia" ] }

$match 中使用的文档筛选语法，和读取文档时的筛选语法相同

db.accounts.aggregate([
    {
        $match: {
            "name.first_name": ‘fengxia‘
        }
    }
])
{ "_id" : ObjectId("5d80fd4471c6b2236fb80de9"), "name" : { "first_name" : "fengxia", "last_name" : "yu" }, "balance" : 200 }

将 $project 、 $match 、$skip、$limit 相结合

db.accounts.aggregate([
    {
        $match: {
            $or: [
                {
                    "name.first_name": ‘fengxia‘
                },
                {
                    "name.first_name": ‘qingquan‘
                },          
            ]
        }
    },
    {
        $project: {
            _id: 0
        }
    },
    {
        $skip: 1
    },
    {
        $limit: 1
    }
])
{ "name" : { "first_name" : "fengxia", "last_name" : "yu" }, "balance" : 200 }

$unwind
对本节的数据进行修改

db.accounts.update({‘name.first_name‘:‘qingquan‘},{
 $set:{
  "currency":["CNY","USD"]
 }
})

db.accounts.update({‘name.first_name‘:‘fengxia‘},{
 $set:{
  "currency":"GBP"
 }
})

修改后的数据如下，一个用户的currency是数组，另一个用户的currency是字符串

> db.accounts.find()
{ "_id" : ObjectId("5d80fd4471c6b2236fb80de8"), "name" : { "first_name" : "qingquan", "last_name" : "zeng" }, "balance" : 100, "currency" : [ "CNY", "USD" ] }
{ "_id" : ObjectId("5d80fd4471c6b2236fb80de9"), "name" : { "first_name" : "fengxia", "last_name" : "yu" }, "balance" : 200, "currency" : "GBP" }

使用unwind对数组元素进行平铺，可以将currency为数组的记录，从一条记录拆分为多条记录

db.accounts.aggregate([
    {
        $unwind: {
            path: "$currency"
        }
    }
])
{ "_id" : ObjectId("5d80fd4471c6b2236fb80de8"), "name" : { "first_name" : "qingquan", "last_name" : "zeng" }, "balance" : 100, "currency" : "CNY" }
{ "_id" : ObjectId("5d80fd4471c6b2236fb80de8"), "name" : { "first_name" : "qingquan", "last_name" : "zeng" }, "balance" : 100, "currency" : "USD" }
{ "_id" : ObjectId("5d80fd4471c6b2236fb80de9"), "name" : { "first_name" : "fengxia", "last_name" : "yu" }, "balance" : 200, "currency" : "GBP" }

为了方便排查，还可以在设定一个字段，用于数组展开后标记每个元素在原数组的位置

db.accounts.aggregate([
    {
        $unwind: {
            path: "$currency",
            includeArrayIndex:"origin_index"
        }
    }
])
{ "_id" : ObjectId("5d80c37349f3060f1212a055"), "name" : { "first_name" : "qingquan", "last_name" : "zeng" }, "balance" : 100, "currency" : "CNY", "origin_index" : NumberLong(0) }
{ "_id" : ObjectId("5d80c37349f3060f1212a055"), "name" : { "first_name" : "qingquan", "last_name" : "zeng" }, "balance" : 100, "currency" : "USD", "origin_index" : NumberLong(1) }
{ "_id" : ObjectId("5d80c37349f3060f1212a056"), "name" : { "first_name" : "fengxia", "last_name" : "yu" }, "balance" : 200, "currency" : "GBP", "origin_index" : null }

还有一点需要注意的是，$unwind 在产生结果前，默认会直接过滤掉如下记录：

• currency字段为空数组
• currency字段不存在
• currency字段为null

如果不想过滤的话，可以设定 preserveNullAndEmptyArrays 为 true

db.accounts.aggregate([
    {
        $unwind: {
            path: "$currency",
            includeArrayIndex: "origin_index",
            preserveNullAndEmptyArrays: true
        }
    }
])

$sort

• 1 从小到大
• -1 从大到小

db.accounts.aggregate([
    {
        $sort: {
            balance: -1
        }
    }
])
{ "_id" : ObjectId("5d80c37349f3060f1212a056"), "name" : { "first_name" : "fengxia", "last_name" : "yu" }, "balance" : 200, "currency" : "GBP" }
{ "_id" : ObjectId("5d80c37349f3060f1212a055"), "name" : { "first_name" : "qingquan", "last_name" : "zeng" }, "balance" : 100, "currency" : [ "CNY", "USD" ] }

MongoDB 聚合操作重复问题https://jacoobwang.github.io/2018/01/08/MongoDb%E8%81%9A%E5%90%88%E6%93%8D%E4%BD%9C%E9%87%8D%E5%A4%8D%E9%97%AE%E9%A2%98/

MongoDB 聚合管道使用