MongoDB多重嵌套数组操作梳理

测试数据

导入

mongoimport -d grids -c schemas --file schemas.json

schemas.json:

{
  "_id" : ObjectId("5b9882a1629317fe30e4fd78"),
  "schema" : "attachment",
  "fields" : [{
      "interactive" : false,
      "name" : "name",
      "data" : ["attachment", "action"]
    }, {
      "interactive" : false,
      "name" : "1532921881691.jpg",
      "data" : ["", {
          "action" : true,
          "visible" : true,
          "type" : "button",
          "innerText" : "view",
          "value" : {
            "url" : "http://localhost/photos/1532921881691.jpg"
          }
        }]
    }]
}

{
  "_id" : ObjectId("5b9882a1629317fe30e4fd79"),
  "schema" : "report",
  "fields" : [{
      "interactive" : false,
      "name" : "name",
      "data" : ["statistic", "arts", "science"]
    }, {
      "interactive" : false,
      "name" : "subject",
      "data" : [{
          "innerText" : "total"
        }, ["history", "physical", "painting"], ["math", "chemistry", "geography"]]
    }, {
      "interactive" : true,
      "name" : "Wang",
      "data" : [545.0, [85.0, 100.0, 85.0], [92.0, 91.0, 92.0]]
    }, {
      "interactive" : true,
      "name" : "Liu",
      "data" : [527.0, [88.0, 99.0, 88.0], [66.0, 98.0, 88.0]]
    }]
}

{
  "_id" : ObjectId("5b9882a1629317fe30e4fd7a"),
  "schema" : "graph",
  "fields" : [{
      "interactive" : false,
      "name" : "shape",
      "data" : ["type"]
    }, {
      "interactive" : true,
      "name" : "tree",
      "data" : [{
          "innerText" : "draw",
          "value" : ["subjects", ["arts", ["history", "geography", "painting"]], ["science", ["math", "physical", "chemistry"]]]
        }]
    }, {
      "interactive" : true,
      "name" : "circle",
      "data" : [{
          "innerText" : "draw",
          "value" : {
            "point" : {
              "x" : 0.0,
              "y" : 0.0
            },
            "radius" : 666.0
          }
        }]
    }]
}

一、增

插入通用模版：

$push:{
    :{
        $position:, // 索引>=0
        [$slice]:, // 可选，>0 从头部开始截取（尾插） ; <0 从尾部向上截取（头插） ; 0 删除全部
        [$sort]:{:}, // 可选，-1 降序 ; 1 升序
        $each:[{: , ...},{: , ...} , ...]
    }
}

在$position位置拼接$each数组，并且将拼接得出的数组长度截断为$slice。
令数组拼接后长度为length，$slice>0截取[0,$slice-1]，$slice<0截取[length-|$slice|,length)，当length-|$slice|<=0时，数组清空。
$sort排序在截断$slice操作前执行，即先排序再截断。

为方便记叙，以下，直接在文档根部定义的数组约定为外层数组，不管是否带键值，外层数组内部的数组全部约定为内嵌数组，不管是否带键值。

1.1 外层数组插入

Visual Schema Attachment（可视化）

基本操作，往attachment模型插入一行数据，即往fields字段数组添加一个对象。

db.schemas.update(
{"schema" : "attachment"},
{
    $push:{
        fields:{
          "interactive" : false,
          "name" : "1565489.xlsx",
          "data" : ["-", {
              "action" : true,
              "visible" : true,
              "type" : "button",
              "innerText" : "view",
              "value" : {
                "url" : "http://localhost/files/1565489.xlsx"
              }
            }]
        }
    }
});

结果：

{
  "_id" : ObjectId("5b9882a1629317fe30e4fd78"),
  "schema" : "attachment",
  "fields" : [{
      "interactive" : false,
      "name" : "name",
      "data" : ["attachment", "action"]
    }, {
      "interactive" : false,
      "name" : "1532921881691.jpg",
      "data" : ["", {
          "action" : true,
          "visible" : true,
          "type" : "button",
          "innerText" : "view",
          "value" : {
            "url" : "http://localhost/photos/1532921881691.jpg"
          }
        }]
    }, {
      "interactive" : false,
      "name" : "1565489.xlsx",
      "data" : ["-", {
          "action" : true,
          "visible" : true,
          "type" : "button",
          "innerText" : "view",
          "value" : {
            "url" : "http://localhost/files/1565489.xlsx"
          }
        }]
    }]
}

1.2 带键值内嵌数组插入

这里要用到占位符，MongoDB 3.6+的特性，能解决旧版本无解的内嵌元素定位问题，在旧版只能修改外层数组，修改内嵌数组需要将外层数组元素整个替换。
已弃坑项目 Robotmongo / Robot3T 下不能运行以下查询，仅支持到MongoDB 3.4，应使用自带Shell环境或换用NoSQL Manager。

往attachment模型添加一列，即往所有的内嵌字段data添加一个对象，且仅当data字段存在且类型为数组时生效。

假如内嵌数组data的实际类型为String，那么查询就会报错，出于严谨性考虑，因此需要在arrayFilters中为data字段及其元素判断设置过滤条件，不符合条件的data字段不被执行插入操作。
注意，arrayFilters中的占位符i指代数组data的元素本体，并非数组元素索引，arrayFilters每个元素绑定一个占位符的子查询。
i的类型为Object时，可使用"Element Query Operators"进行过滤，即可以使用$exists和$type，当i的类型为数字或字符串时，可使用值域判断相关的操作符$in、$eq等，当i类型为数组时，可以使用"Array Query Operators"。

db.schemas.update(
{
    "schema" : "attachment"
},
{
    $push:{
        "fields.$[i].data":{
            innerText:"new column"
        }
    }
},{
    arrayFilters:[
        {
            i:{
                $type:"object"
            },
            "i.data":{
                $exists:true  // 若不设置该条件，当data字段不存在时，会自动创建data并插入元素
            },
            "i.data":{
                $type:"array" // 校验类型，实际上也起到了{$exists:true}的作用
            }
        }
    ]
});

结果：

{
  "_id" : ObjectId("5b9882a1629317fe30e4fd78"),
  "schema" : "attachment",
  "fields" : [{
      "interactive" : false,
      "name" : "name",
      "data" : ["attachment", "action", {
          "innerText" : "new column"
        }]
    }, {
      "interactive" : false,
      "name" : "1532921881691.jpg",
      "data" : ["", {
          "action" : true,
          "visible" : true,
          "type" : "button",
          "innerText" : "view",
          "value" : {
            "url" : "http://localhost/photos/1532921881691.jpg"
          }
        }, {
          "innerText" : "new column"
        }]
    }]
}

1.3 无键值内嵌元素插入

Visual Schema Graph

往graph模型的tree图表绑定数据的叶子元素添加一个节点，表现为["history","geography","painting"]里面多了个字符串。

db.schemas.update({schema:"graph"},{
    $push:{
        "fields.$[i].data.$[j].value.$[k].$[l]":"leaf"
    }
},{
    arrayFilters:[
        {
            i:{
                $type:"object"
            },
            "i.name":{
                $eq:"tree"
            },
            "i.data":{
                $type:"array"
            }
        },
        {
            j:{
                $type:"object"
            },
            "j.value":{
                $type:"array"
            }
        },
        {
            k:{
                $type:"array"
            }
        },
        {
            l:{
                $type:"array"
            }
        }
    ]
});

结果：

{
  "_id" : ObjectId("5b9882a1629317fe30e4fd7a"),
  "schema" : "graph",
  "fields" : [{
      "interactive" : false,
      "name" : "shape",
      "data" : ["type"]
    }, {
      "interactive" : true,
      "name" : "tree",
      "data" : [{
          "innerText" : "draw",
          "value" : ["subjects", ["arts", ["history", "geography", "painting", "leaf"]], ["science", ["math", "physical", "chemistry", "leaf"]]]
        }]
    }, {
      "interactive" : true,
      "name" : "circle",
      "data" : [{
          "innerText" : "draw",
          "value" : {
            "point" : {
              "x" : 0,
              "y" : 0
            },
            "radius" : 666
          }
        }]
    }]
}

由于使用k指定过滤为数组，可以使用$elemMatch数组操作符进行具体匹配，定向在"arts"类别下添加节点，也可以使用$all:["arts"]。
如何细化匹配视具体数据而言，如果所有元素数据都是相同的，那就无法做到单一定位修改。

db.schemas.update({schema:"graph"},{
    $push:{
        "fields.$[i].data.$[j].value.$[k].$[l]":"leaf"
    }
},{
    arrayFilters:[
        {
            i:{
                $type:"object"
            },
            "i.name":{
                $eq:"tree"
            },
            "i.data":{
                $type:"array"
            }
        },
        {
            j:{
                $type:"object"
            },
            "j.value":{
                $type:"array"
            }
        },
        {
            k:{
                $type:"array",
                $elemMatch:{
                    $eq:"arts"  // $all:["arts"]
                }
            }

        },
        {
            l:{
                $type:"array"
            }
        }
    ]
});

结果：

{
  "_id" : ObjectId("5b9882a1629317fe30e4fd7a"),
  "schema" : "graph",
  "fields" : [{
      "interactive" : false,
      "name" : "shape",
      "data" : ["type"]
    }, {
      "interactive" : true,
      "name" : "tree",
      "data" : [{
          "innerText" : "draw",
          "value" : ["subjects", ["arts", ["history", "geography", "painting", "leaf"]], ["science", ["math", "physical", "chemistry"]]]
        }]
    }, {
      "interactive" : true,
      "name" : "circle",
      "data" : [{
          "innerText" : "draw",
          "value" : {
            "point" : {
              "x" : 0,
              "y" : 0
            },
            "radius" : 666
          }
        }]
    }]
}

二、删

2.1 无键值内嵌数组的删除

Visual Schema Report

删除模型report中的physical和geography科目分数，即裁掉一列。
更新和删除是原子操作，暂没有类似removeAt(index)根据索引删除的操作符，需要先置空再删除。

db.schemas.update({"schema" : "report"},
{
    $pull:{
        "fields.$[i].data.$[j]":{
            $in:["physical","geography"]
        }
    },
    $unset:{
        "fields.$[i2].data.1.1":{}
    },
    $pop:{
        "fields.$[i2].data.2":1,
    }
},{
    arrayFilters:[
        {
            i:
            {
                $type:"object",
            },
            "i.name":{$eq:"subject"}
        },
        {
            j:
            {
                $type:"array"
            }
        },
            {
            "i2":
            {
                $type:"object"
            },
            "i2.name":{$not:{$in:["name","subject"]}}
        }
    ]
});

db.schemas.update({"schema" : "report"},
{
    $pull:{
        "fields.$[i].data.1":null
    }
},{
    arrayFilters:[
        {
            "i":
            {
                $type:"object"
            },
            "i.name":{$not:{$in:["name","subject"]}}
        }
    ]
});

结果：

{
  "_id" : ObjectId("5b9882a1629317fe30e4fd79"),
  "schema" : "report",
  "fields" : [{
      "interactive" : false,
      "name" : "name",
      "data" : ["statistic", "arts", " science"]
    }, {
      "interactive" : false,
      "name" : "subject",
      "data" : [{
          "innerText" : "total"
        }, ["history", "painting"], ["math", "chemistry"]]
    }, {
      "interactive" : true,
      "name" : "Wang",
      "data" : [545, [85, 85], [92, 91]]
    }, {
      "interactive" : true,
      "name" : "Liu",
      "data" : [527, [88, 88], [66, 98]]
    }]
}

三、查

3.1 常用聚集操作

重新统计2.1的的总分
聚集操作符$in、$not、$type单独一套，与同名的查询操作符用法不一样，尤其是$type，聚集操作符{$type:value}识别数据类型，返回字符串"string"、"object"等，与查询操作符{$type:type}刚好反过来。

db.schemas.aggregate([
    {
        $match:{"schema" : "report"}
    },
    {
        $project:{
            _id:false,
            scores:{
                $filter:{
                    input:"$fields",
                    as:"field",
                    cond:{
                        $not:{
                            $in:["$$field.name",["name","subject"]]
                        }
                    }
                }
            }
        }
    },
    {
        $unwind:{
            path:"$scores",
            preserveNullAndEmptyArrays:false
        }
    },
    {
        $project:
        {
            name:"$scores.name",
            scores:{
                $filter:{
                    input:"$scores.data",
                    as:"score",
                    cond:{
                        $eq:[{$type:"$$score"},"array"]
                    }
                }
            }
        }
    },
    {
        $project:{
            name:"$name",
            arts: { $arrayElemAt: [ "$scores", 0 ] },
            science: { $arrayElemAt: [ "$scores", 1 ] }
        }
    },
    {
        $project:{
            name:"$name",
            history:{ $arrayElemAt: [ "$arts", 0 ] },
            painting:{ $arrayElemAt: [ "$arts", 1 ] },
            math:{ $arrayElemAt: [ "$science", 0 ] },
            chemistry:{ $arrayElemAt: [ "$science", 1 ] },
            atrs: { $sum:"$arts" },
            science: { $sum:"$science" }
        }
    },
    {
        $addFields:{
            total: { $sum:["$atrs","$science"] }
        }
    },
    {
        $out:"scores"
    } 
]);

结果：

db.scores.find()

/* 1 */
{
  "_id" : ObjectId("5b9f97db927db7e834715b04"),
  "name" : "Wang",
  "history" : 85,
  "painting" : 85,
  "math" : 92,
  "chemistry" : 91,
  "atrs" : 170,
  "science" : 183,
  "total" : 353
}

/* 2 */
{
  "_id" : ObjectId("5b9f97db927db7e834715b05"),
  "name" : "Liu",
  "history" : 88,
  "painting" : 88,
  "math" : 66,
  "chemistry" : 98,
  "atrs" : 176,
  "science" : 164,
  "total" : 340
}

四、改

4.1 存储过程

更新 2.1总分

var atrs_Wang=db.scores.findOne({name:"Wang"}).atrs
var atrs_Liu=db.scores.findOne({name:"Liu"}).atrs

var science_Wang=db.scores.findOne({name:"Wang"}).science
var science_Liu=db.scores.findOne({name:"Liu"}).science


db.schemas.update(
{"schema" : "report"},
{
    $set:{
        "fields.$[i].data.0":(atrs_Wang+science_Wang),
        "fields.$[i2].data.0":(atrs_Liu+science_Liu)
    }
},{
    arrayFilters:[
        {
            i:
            {
                $type:"object"
            },
            "i.name":{$eq:"Wang"}
        },   
        {
            i2:
            {
                $type:"object"
            },
            "i2.name":{$eq:"Liu"}
        }
    ]
});

结果：

{
  "_id" : ObjectId("5b9882a1629317fe30e4fd79"),
  "schema" : "report",
  "fields" : [{
      "interactive" : false,
      "name" : "name",
      "data" : ["statistic", "arts", "science"]
    }, {
      "interactive" : false,
      "name" : "subject",
      "data" : [{
          "innerText" : "total"
        }, ["history", "painting"], ["math", "chemistry"]]
    }, {
      "interactive" : true,
      "name" : "Wang",
      "data" : [353, [85, 85], [92, 91]]
    }, {
      "interactive" : true,
      "name" : "Liu",
      "data" : [340, [88, 88], [66, 98]]
    }]
}