Mongodb Aggregated documents into a nested hierarchy - mongodb

Question

I am trying to aggregate documents in a collection to a nested graph. Below is the sample data from comments store application I am working with -
Posts Collection:
{
"_id" : "1",
"text" : "hey",
}
{
"_id" : "2",
"text" : "hello",
"replyTo" : "1"
}
{
"_id" : "3",
"text" : "What's up?",
"replyTo" : "2"
}
{
"_id" : "4",
"text" : "How are you",
"replyTo" : "1"
}
{
"_id" : "5",
"text" : "Knock, knock!",
}
The result I am expecting should look like :
{
"_id": "1",
"text": "hey",
"replies": [
{
"_id": "2",
"text": "hello",
"replyTo": "1",
"replies": [
{
"_id": "3",
"text": "What's up?",
"replyTo": "2"
}
]
},
{
"_id": "4",
"text": "How are you",
"replyTo": "1"
}
]
}
{
"_id" : "5",
"text" : "Knock, knock!",
}
I tried $graphLookup which is processing documents recursively but the results is a flat array which is not what I expect -
db.posts.aggregate(
[
{
"$graphLookup" : {
"from" : "posts",
"startWith" : "$_id",
"connectFromField" : "_id",
"connectToField" : "replyTo",
"as" : "replies"
}
}
],
{
"allowDiskUse" : false
}
);
Result:
{
"_id" : "1",
"text" : "hey",
"slug_path" : "1",
"replies" : [
{
"_id" : "2",
"text" : "hello",
"slug_path" : "1/2",
"replyTo" : "1"
},
{
"_id" : "4",
"text" : "How are you",
"slug_path" : "1/4",
"replyTo" : "1"
},
{
"_id" : "3",
"text" : "what's up?",
"slug_path" : "1/2/3",
"replyTo" : "2"
}
]
}
{
"_id" : "2",
"text" : "hello",
"slug_path" : "1/2",
"replyTo" : "1",
"replies" : [
{
"_id" : "3",
"text" : "what's up?",
"slug_path" : "1/2/3",
"replyTo" : "2"
}
]
}
{
"_id" : "4",
"text" : "How are you",
"slug_path" : "1/4",
"replyTo" : "1",
"replies" : [
]
}
{
"_id" : "3",
"text" : "what's up?",
"slug_path" : "1/2/3",
"replyTo" : "2",
"replies" : [
]
}
{
"_id" : "5",
"text" : "Knock, knock!",
"replies" : [
]`
}
Is there a way I can achieve the type of aggregation I am expecting?
Thanks
Davinder

Answer 1

I ended up ditching $graphLookup as it can only give 1-level parent-child relationships and can't recursively create a tree like structure as results. I was able to create tree like structure completely in Java and program is running fast from performance perspective.