merge records in to one using mongodb - mongodb

Question

Here is my collection structure, tried $mergeObjects , but did not know how to use in right way, please help :
{
"_id" : ObjectId("5e39b407eb2b5e4c3c80c5b0"),
"groupId":"1",
"emp" : {
"roles" : [
{
"roleId" : "role1"
}
],
"designation" : [
"Manager"
],
"dept" : [
{
"deptId" : "dept1"
}
]
}
},
{
"_id" : ObjectId("5e39b435eb2b5e4c3c80c5b1"),
"groupId":"1",
"emp" : {
"roles" : [
{
"roleId" : "role2"
}
],
"designation" : [
"Developer"
],
"dept" : [
{
"deptId" : "dept2"
}
]
}
}
I want an aggregate query which merges the two documents w.r.t. "groupId" like this using mongodb
{
"_id" : <some id>,
"groupId":"1",
"emp" : {
"roles" : [
{
"roleId" : "role1"
},
{
"roleId" : "role2"
}
],
"designation" : [
"Manager","Developer"
],
"dept" : [
{
"deptId" : "dept1"
},
{
"deptId" : "dept2"
}
]
}
}
There are around 200 to 300 fields like arrays, array of an array , array of an array of an array and so on.

Answer 1

I don't think the $mergeObject operator would work in this given it overrides fields, for example the roleId in role would always be role2 in this case if $mergeObject is used. I are going to need a multistage solution combining $group to combine the record together and a $project to restructure the fields inside emp into what you want.
Try this:
db.collection.aggregate([
{ "$group": {
"_id": "$groupId",
"roles": { "$push": "$emp.roles" },
"designation": { "$push": "$emp.designation" },
"dept": { "$push": "$emp.dept" },
}},
{ "$project": {
"_id": 0,
"groupId": "$_id",
"roles": {
"$reduce": {
"input": "$roles",
"initialValue": [],
"in": { "$concatArrays": [ "$$value", "$$this" ] }
}
},
"designation": {
"$reduce": {
"input": "$designation",
"initialValue": [],
"in": { "$concatArrays": [ "$$value", "$$this" ] }
}
},
"dept": {
"$reduce": {
"input": "$dept",
"initialValue": [],
"in": { "$concatArrays": [ "$$value", "$$this" ] }
}
}
}}
]);

Answer 2

merging arrays together without repeating values is done using operator $concatArrays and this is done by pushing all arrays together in $group then using $reduce like in previous answer by Josh Balcitis
another operator that gives unique result isr $addToSet instead of $push but you have to $unwind all array records before $group to add unique single records.
ex:
db.collection.aggregate([{
$unwind: {
path: "$emp.roles",
preserveNullAndEmptyArrays: true
}
},
{
$unwind: {
path: "$emp.designation",
preserveNullAndEmptyArrays: true
}
},
{
$unwind: {
path: "$emp.dept",
preserveNullAndEmptyArrays: true
}
},
{
"$group": {
"_id": "$groupId",
"roles": {
$addToSet: "$emp.roles"
},
"designation": {
$addToSet: "$emp.designation"
},
"dept": {
$addToSet: "$emp.dept"
},
}
},
]);
which approach is better? that depends on your data and $group vs $reduce perfomance
note: preserveNullAndEmptyArrays option is used to $prevent empty arrays from affecting whole result, its optional