MongoDB Aggregation: "Populate" (resolve) Ref in Nested Array - mongodb

Question

Background
I have a collection Items with documents such as
{
"_id": "5d9e3a5ced27230f68032e21",
... more fields
"foos": [
{
"_id": "5d9e3a5ced27230f68032e25",
... more fields
"bars": [
"5d9dab461bbb4db66db41f93"
],
},
{
"id": "5d9e3a5ced27230f68032e24",
... more fields
"bars": [
"5d9dab461bbb4db66db41f93",
"5d9e3a23ed27230f68032e1a"
]
}
]
}
with bars referring to another collection Bars.
Goal
I'd like to get a list of all documents (with all their fields) in Items, but with bars resolved to the document in Bars.
Small Catch
I want to be able to create a generic function to which I simply pass the path to resolve (e.g. foos.bars) and the collection from which to resolve (Bars) so that I can use it with different collections and arbitrary levels of nesting.
Initial Approach
I've found a rather complicated way to do this for my example, but before I generalize it, I'd like to know whether there really isn't a simpler way. Input is highly appreciated!
Here's what I've got:
[
{
"$unwind": {
"path": "$foos",
"includeArrayIndex": "foos_index"
}
},
{
"$unwind": {
"path": "$foos.bars"
}
},
{
"$lookup": {
"from": "Bars",
"localField": "foos.bars",
"foreignField": "_id",
"as": "foos.bars"
}
},
{
"$unwind": {
"path": "$foos.bars"
}
},
{
"$group": {
"_id": {
"id": "$_id",
"foo_index": "$foos_index"
},
"foos": {
"$first": "$foos"
},
"bars": {
"$push": "$foos.bars"
}
}
},
{
"$addFields": {
"foos": {
"$mergeObjects": [
"$foos",
{
"bars": "$bars"
}
]
}
}
},
{
"$group": {
"_id": "$_id.id",
"foos": {
"$push": "$foos"
}
}
},
{
"$lookup": {
"from": "Items",
"localField": "_id",
"foreignField": "_id",
"as": "original_doc"
}
},
{
"$unwind": {
"path": "$original_doc"
}
},
{
"$replaceRoot": {
"newRoot": {
"$mergeObjects": [
"$original_doc",
{
"foos": "$foos"
}
]
}
}
}
]
Update: First Iteration
I've realized I don't need the "leaf" level unwound, so I've now got a simplified version (but for deeper nesting, I'll still need what I had before, right?):
[
{
"$unwind": {
"path": "$foos",
"includeArrayIndex": "foos_index"
}
},
{
"$lookup": {
"from": "Bars",
"localField": "foos.bars",
"foreignField": "_id",
"as": "foos.bars"
}
},
{
"$group": {
"_id": "$_id",
"foos": {
"$push": "$foos"
}
}
},
{
"$lookup": {
"from": "Items",
"localField": "_id",
"foreignField": "_id",
"as": "original_doc"
}
},
{
"$unwind": {
"path": "$original_doc"
}
},
{
"$replaceRoot": {
"newRoot": {
"$mergeObjects": [
"$original_doc",
{
"foos": "$foos"
}
]
}
}
}
]

Answer 1

Maybe someone still finds something better, but until then, if someone comes across this, the following seems reasonable to me by now:
[
{
"$unwind": {
"path": "$foos",
"includeArrayIndex": "foos_index"
}
},
{
"$lookup": {
"from": "Bars",
"localField": "foos.bars",
"foreignField": "_id",
"as": "foos.bars"
}
},
{
"$group": {
"_id": "$_id",
"savepoint": {
"$first": "$$ROOT"
},
"foos": {
"$push": "$foos"
}
}
},
{
"$replaceRoot": {
"newRoot": {
"$mergeObjects": [
"$savepoint",
{
"foos": "$foos"
}
]
}
}
}
]