【问题】

Hi,

i am trying to use mongodb aggregate query using $setUnion, $project and also $group to group the documents. The structure of document is

{
“_id” : ObjectId(“55014006e4b0333c9531043e”),
“acls” : {
“append” : {
“users” : [ObjectId(“54f5bfb0336a15084785c393”) ],
“groups” : [ ]
},
“edit” : {
“groups” : [ ],
“users” : [
ObjectId(“54f5bfb0336a15084785c392”)
]
},
“fullControl” : {
“users” : [ ],
“groups” : [ ]
},
“read” : {
“users” : [ObjectId(“54f5bfb0336a15084785c392”), ObjectId(“54f5bfb0336a15084785c398”)],
“groups” : [ ]
}
},
        name: “ABC”
}
{
“_id” : ObjectId(“55014006e4b0333c9531043f”),
“acls” : {
“append” : {
“users” : [ObjectId(“54f5bfb0336a15084785c365”) ],
“groups” : [ ]
},
“edit” : {
“groups” : [ ],
“users” : [
ObjectId(“54f5bfb0336a15084785c392”)
]
},
“fullControl” : {
“users” : [ ],
“groups” : [ ]
},
“read” : {
“users” : [ObjectId(“54f5bfb0336a15084785c392”), ObjectId(“54f5bfb0336a15084785c370”)],
“groups” : [ ]
}
},
        name: “ABC”
}

I need to query based on the name, then i need to use $setUnion for “acls.read.users, acls.edit.users, acls.append.users and acls.fullControl.users” and after that i need to group the documents by “name” and another field having the list of users in “user field” like below

{
  result : [
     {
          _id: “ABC”,
          readUsers : [
                  ObjectId(“54f5bfb0336a15084785c393”),
                  ObjectId(“54f5bfb0336a15084785c392”),
                  ObjectId(“54f5bfb0336a15084785c398”),
                  ObjectId(“54f5bfb0336a15084785c365”),
                  ObjectId(“54f5bfb0336a15084785c370”)
           ]
      }
  ]
}

The query i tried like below

db.abc.aggregate([
             {$match:{“name”:“ABC”}},
             {$project:{users : {$setUnion:[“$acls.read.users”,“$acls.edit.users”,“$acls.append.users”,“$acls.fullControl.users”]}}},
            {$group: {_id: “$owner”,“readuser”: “$user”}}
 ])
when i run the query i am getting error like 
aggregate failed: {
“errmsg” : “exception: invalid operator ‘$setUnion’”,
“code” : 15999,
“ok” : 0
}

Can anyone tell me is it possiible to get the result like above and how to achieve it?

【回答】

用 Mongodb 本身的 API 需要硬编码才能实现子文档的合并，不够直观且难度较大，这种情况下可以用集算器来实现，代码如下：

A2：

A3：数据根据 name 分组

A4：每组数据的 acls 列内数据合并成一个序列，作为 readUser 保存