How to get documents with non unique array elements?

Question

I have the following MongoDB documents:

{
   _id: ObjectId('09de14821345dda65c471c99'),
   items: [
        _id: ObjectId('34de64871345dfa655471c99'),
        _id: ObjectId('34de64871345dfa655471c91'),
        _id: ObjectId('34de64871345dfa655471c99'),       
   ]
},
{
   _id: ObjectId('09de14821345dda65c471c98'),
   items: [
        _id: ObjectId('24de64871345dfa61271c10'),
        _id: ObjectId('24de64871345dfa61271c11'),
        _id: ObjectId('24de64871345dfa61271c11'),       
   ]
},
{
   _id: ObjectId('09de14821345dda65c471c07'),
   items: [
        _id: ObjectId('24de64871345dfa61271c05'),
        _id: ObjectId('24de64871345dfa61271c06'),
        _id: ObjectId('24de64871345dfa61271c07'),       
   ]
}

I need to find all documents with repeated items array elements. So from the documents above I want to get the following result:

db.collection.documents.find({/** need query*/}).toArray(function (err, documents) {
    console.dir(documents); // documents with id's 09de14821345dda65c471c99 and 09de14821345dda65c471c98
});

How could I do that?

Answer 1

In order to group and match results you will need to use the Aggregation Framework or Map/Reduce rather than a simple find() query.

Example data

Your example document include some errors: a few of the ObjectIDs are too short and the array elements should either be embedded documents ({_id: ObjectId(...)}) or simple values.

For test data I've used:

db.mydocs.insert([
{
   _id: ObjectId('09de14821345dda65c471c99'),
   items: [
        ObjectId('34de64871345dfa655471c99'),
        ObjectId('34de64871345dfa655471c91'),
        ObjectId('34de64871345dfa655471c99')      
   ]
},
{
   _id: ObjectId('09de14821345dda65c471c98'),
   items: [
        ObjectId('24de64871345ddfa61271c10'),
        ObjectId('24de64871345ddfa61271c11'),
        ObjectId('24de64871345ddfa61271c11')       
   ]
},
{
   _id: ObjectId('09de14821345dda65c471c07'),
   items: [
        ObjectId('24de64871345ddfa61271c05'),
        ObjectId('24de64871345ddfa61271c06'),
        ObjectId('24de64871345ddfa61271c07')       
   ]
}])

Aggregation query

Here is an aggregation query using the mongo shell:

db.mydocs.aggregate(

    // Unpack items array into stream of documents
    { $unwind: "$items" },

    // Group by original document _id and item
    { $group: {
        _id: { _id: "$_id", item: "$items" },
        count: { $sum: 1 }
    }},

    // Limit to duplicated array items (1 or more count per document _id)
    { $match: {
        count: { $gt: 1 }
    }},

    // (Optional) clean up the result formatting
    { $project: {
        _id: "$_id._id",
        item: "$_id.item",
        count: "$count"
    }}
)

Sample results

{
    "_id" : ObjectId("09de14821345dda65c471c98"),
    "count" : 2,
    "item" : ObjectId("24de64871345ddfa61271c11")
}
{
    "_id" : ObjectId("09de14821345dda65c471c99"),
    "count" : 2,
    "item" : ObjectId("34de64871345dfa655471c99")
}