How to group nested posts with MongoDB aggregation framework?

The following pipeline should work for you:

var pipeline = [
    {
        "$project": {
            "title": 1, "body": 1, 
            "post_id": { "$ifNull": [ "$_post", "$_id" ] }
        }
    },  
    {
        "$group": {
            "_id": "$post_id",
            "title": { "$first": "$title" },
            "body": { "$first": "$body" },
            "comments": {
                "$push": {
                    "_id": "$_id",
                    "_post": "$post_id",
                    "body": "$body"
                }
            }
        }
    },
    {
        "$project": {
            "title": 1, "body": 1,
            "comments": {
                "$setDifference": [
                    {
                        "$map": {
                            "input": "$comments",
                            "as": "el",
                            "in": {
                                "$cond": [
                                    { "$ne": [ "$$el._id", "$$el._post" ] },
                                    "$$el",
                                    false
                                ]
                            }
                        }
                    },
                    [false]
                ]
            }
        }
    }
];

Post.aggregate(pipeline, function (err, result) {
    if (err) { /* handle error */ };
    console.log(result);
});

The pipeline is structured in such a way that your first step, the $project operator stage, is to project the field post_id to be used as the group by key in the next pipeline stage. Since your schema is hierarchical, you'd need this field for parent/root documents. The $ifNull operator will act as the coalesce operator and return the replacement value if the field does not exist in the documents.

The next pipeline step, the $group pipeline stage tries to group the data to process them. The $group pipeline operator is similar to the SQL's GROUP BY clause. In SQL, we can't use GROUP BY unless we use any of the aggregation functions. The same way, we have to use an aggregation function in MongoDB as well. In this case you need the $push operator to create the comments array. The other fields are then accumulated using the $first operator.

The final step involves fitering the comments array so that you remove the document with the post details, which is definitely not of a comment type. This is made possible through the $setDifference and $map operators. The $map operator in essence creates a new array field that holds values as a result of the evaluated logic in a subexpression to each element of an array. The $setDifference operator then returns a set with elements that appear in the first set but not in the second set; i.e. performs a relative complement of the second set relative to the first. In this case it will return the final comments array that has elements not related to the parent documents via the _id property.