Question

I am developing a system where items can be shared with other users via an access key. I’m storing the access keys as fields within a shareinfo object (embedded within the item’s document), as shown below:

shareinfo:{
    ........
    <nth key>: <permissions object - may be complex and large>
    ........
}

When an item is accessed I check shareinfo.key and find if its valid or not.

Currently, to list the keys I am loading (in Java) the entire shareinfo object in memory and running keySet() on it to retrieve and return the keys while the rest of the data is wasted.

Here’s the problem: I want to get the list of keys (i.e. object field names) without the accompanying data (because in some cases the permissions object is noticeably large).

I could not find any query in the mongodb docs for such a query. I want to know whether it’s possible or not? Or is there an optimized way to load the list of field names into the application without the accompanying field values?

Answer 1

MongoDB has a schema-less design, which means that any document could have different fields contained within it from any other document. To get an exhaustive list of all the fields for all the documents, you would need to traverse all the documents in the collection and enumerate each field. For any reasonably sized collection this is going to be an extremely expensive operation.

There are a couple of helpers that make this more simple: Variety and schema.js . They both allow you to limit the documents inspected, and they report on coverage as well. Both amount to doing a map/reduce on the collections, but are at least simpler than having to write one yourself.