Question

I was wonderning if there was any way to make complex selections.
Considering the following document collection:

{
  image = "cab"
  tags = [
            [ "NNP", 0 ],
            [ "NN", 1 ]
         ]
},
{
  image = "castle"
  tags = [
            [ "NNP", 2 ],
            [ "NN", 1 ],
            [ "VB", 0 ]
         ]
}, 
(etc)

How would I express a query that checks if tags contains any element like, for instance, ["NN", ?] where I simply do not care about the integer-part of the inner array? Note that the inner array is unordered and integer values are arbitrarily assigned.

Note that I’m, absolutely new to mongodb, I might have overlooked a way to do that in the documentation. However, I was unable to use $elemMatch successfully 🙁

Answer 1

I suggest a little another schema:

{
  image = "cab"
  tags = [
            { name: "NNP", order: 0 },
            { name: "NN", order: 1 }
         ]
}

There are three reasons why i am suggest another schema:

1. First it’s because two level nesting usual harder to query/update
2. Array within tags array containts items of different types(int, string).
3. Array within tags array looks like for me as ‘tag object’ with two fields

With above schema you can easy query for tags with some name and order:

db.images.find({ "tags.name" : "NN", "tags.order" : 1 })

Update:

Dot notation documentation

Hope this help!