MongoDB - How can I use MapReduce to merge a value from one collection into another collection on multiple keys of a second collection?

Question

I have two MongoDB collections: The first is a collection that includes frequency information for different IDs and is shown (truncated form) below:

[
    {
        "_id" : "A1",
        "value" : 19
    },
    {
        "_id" : "A2",
        "value" : 6
    },
    {
        "_id" : "A3",
        "value" : 12
    },
    {
        "_id" : "A4",
        "value" : 8
    },
    {
        "_id" : "A5",
        "value" : 4
    },
    ...
]

The second collection is more complex and contains information for each _id listed in the first collection (it's called frequency_collection_id in the second collection), but frequency_collection_id may be inside two lists (info.details_one, and info.details_two) for each record:

[
    {
        "_id" : ObjectId("53cfc1d086763c43723abb07"),
        "info" : {
            "status" : "pass",
            "details_one" : [ 
                {
                    "frequency_collection_id" : "A1",
                    "name" : "A1_object_name",
                    "class" : "known"
                }, 
                {
                    "frequency_collection_id" : "A2",
                    "name" : "A2_object_name",
                    "class" : "unknown"
                }
            ],
            "details_two" : [ 
                {
                    "frequency_collection_id" : "A1",
                    "name" : "A1_object_name",
                    "class" : "known"
                }, 
                {
                    "frequency_collection_id" : "A2",
                    "name" : "A2_object_name",
                    "class" : "unknown"
                }
            ],
        }
    }
    ...
]

What I'm looking to do, is merge the frequency information (from the first collection) into the second collection, in effect creating a collection that looks like:

[
    {
        "_id" : ObjectId("53cfc1d086763c43723abb07"),
        "info" : {
            "status" : "pass",
            "details_one" : [ 
                {
                    "frequency_collection_id" : "A1",
                    "name" : "A1_object_name",
                    "class" : "known",
                    **"value" : 19**
                }, 
                {
                    "frequency_collection_id" : "A2",
                    "name" : "A2_object_name",
                    "class" : "unknown",
                    **"value" : 6**
                }
            ],
            "details_two" : [ 
                {
                    "frequency_collection_id" : "A1",
                    "name" : "A1_object_name",
                    "class" : "known",
                    **"value" : 19**
                }, 
                {
                    "frequency_collection_id" : "A2",
                    "name" : "A2_object_name",
                    "class" : "unknown",
                    **"value" : 6**
                }
            ],
        }
    }
    ...
]

I know that this should be possible with MongoDB's MapReduce functions, but all the examples I've seen are either too minimal for my collection structure, or are answering different questions than I'm looking for.

Does anyone have any pointers? How can I merge my frequency information (from my first collection) into the records (inside my two lists in each record of the second collection)?

I know this is more or less a JOIN, which MongoDB does not support, but from my reading, it looks like this is a prime example of MapReduce.

I'm learning Mongo as best I can, so please forgive me if my question is too naive.

Answer 1

Just like all MongoDB operations, a MapReduce always operates only on a single collection and can not obtain info from another one. So you first step needs to be to dump both collections into one. Your documents have different _id's, so it should not be a problem for them to coexist in the same collection.

Then you do a MapReduce where the map function emits both kinds of documents for their common key, which is their frequency ID.

Your reduce function will then receive an array of two documents for each key: the two documents you have received. You then just have to merge these two documents into one. Keep in mind that the reduce-function can receive these two documents in any order. It can also happen that it gets called for a partial result (only one of the two documents) or for an already completed result. You need to handle these cases gracefully! A good implementation could be to create a new object and then iterate the input-documents copying all existing relevant fields with their values to the new object, so the resulting object is an amalgamation of the input documents.