Selecting all objects from complex model - mongodb

Question

I have aggregation pipeline stage:
$project: {
'school': {
'id': '$_id',
'name': '$name',
'manager': '$manager'
},
'students': '$groups.students',
'teachers': '$groups.teachers'
}
Need something like this:
{
'users': // manager + students + teachers
}
Tried:
{
'users': {
$push: {
$each: ['$school.manager', '$students', '$teachers']
}
}
}

Answer 1

I'm presuming that "students" and "teachers" are both arrays here and located under a common sub-document heading like so:
{
"_id": 123,
"name": "This school",
"manager": "Bill"
"groups": {
"teachers": ["Ted"],
"students": ["Missy"]
}
}
So in order to get all of those in a singular array such as "users" then it depends on your MongoDB version and the "uniqueness" of your data. For true "sets" and where you have MongoDB 2.6 or greater available, there is the $setUnion operator, albeit with an additional level of $group to make "manager" and array:
db.collection.aggregate([
{ "$group": {
"_id": { "_id": "$_id", "name": "$name" },
"manager": { "$push": "$manager" },
"groups": { "$first": "$groups" }
}},
{ "$project": {
"users": {
"$setUnion": [ "$manager", "$groups.teachers", "$groups.students" ]
}
}}
])
Or otherwise where that operator is not available or there is a "unique" problem then there is this way to handle "combining":
db.collection.aggregate([
{ "$group": {
"_id": { "_id": "_id", "name": "$name" },
"manager": { "$push": "$manager" },
"teachers": { "$first": "$groups.teachers" },
"students": { "$first": "$groups.students" },
"type": { "$first": { "$const": ["M","T","S"] } }
}},
{ "$unwind": "$type" },
{ "$project": {
"users": {
"$cond": [
{ "$eq": [ "$type", "M" ] },
"$manager",
{ "$cond": [
{ "$eq": [ "$type", "T" ] },
"$teachers",
"$students"
]}
]
}
}},
{ "$unwind": "$users" },
{ "$group": {
"_id": "$_id",
"users": { "$push": "$users" }
}}
])
This essentially "tags" each field by a "type" for which the document is copied in the pipeline. Then placed into a single "users" field depending on which "type" matched. The single array then from the resulting three documents from each original can then be safely "unwound" and combined in a final $group operation.
So "sets" are your fastest option where available or where not available or not unique you can use the later technique in order to combine these to a single list.