How do I deal with an arbitrary hash returned from mongo in go (using mgo)?

Question

All of the references I can find construct a struct to hold the return values, assuming that each returned record has the same schema. If they're really documents and don't have a consistent schema other than maybe a few queryable consistent metadata attributes, how can I handle that return value?

e.g.: this https://groups.google.com/forum/#!msg/mgo-users/KirqfCSlKFc/t2l3l4yxFRwJ assumes that you have an array of timestamps. What if it's an array where some of the values are timestamps and some are strings?

In ruby, I'd just treat the whole returned record as an arbitrary hash and deal with it key by key. What can I do with it in go?

Answer 1

You can use a generic map, for example:

var result map[string]interface{}
err := collection.Find(query).One(&result)

You'll find people using bson.M, which is also a map with that same underlying type. There's nothing special about bson.M, though. It's just a short and convenient name for a map. You can create your own convenient name, or use a plain map as in the example above.

Another way to have access to documents without a defined schema is using bson.D in place of the map type above. bson.D is most useful when the order of elements in the document is relevant, or if you want to reduce the operation overhead slightly (maps are a bit more expensive to handle due to their nature). The bson.D type is a slice of struct values with Key/Value pairs. Unlike bson.M, bson.D is special and is handled internally by the mgo/bson package.