Documents

The scala driver includes two scala specific representations for BSON documents. Following convention from the scala collections library, there are immutable and mutable implementations of documents. The underlying implementations of the scala Document use the type safe BsonDocument class. The scala bson classes are available from the org.mongodb.scala.bson namespace, which includes type aliases and companion objects. In general this should suffice but for advanced use cases you may need to use org.bson directly.

Note

The scala Document classes implement TraversableLike[(String, BsonValue)] and the general API mirrors that of a Map[String, BsonValue]. However, unlike Map implementations of TraversableLike enables strict type safety as there is no variance in the value type.

BsonValue is the type safe representation of a Bson type from the org.bson library, it represents specific value types. The most commonly used value types are:

BSON type Scala type
Document org.mongodb.scala.bson.Document
Array List
Date Date or int (ms since epoch)
Boolean Boolean
Double Double
Int32 Integer
Int64 Long
String String
Binary Array[Byte]
ObjectId ObjectId
Null None

It is actually possible to change or extend these mappings, this will be covered in detail below.

There are two main Document classes:

Immutable Documents

Like the Scala collections library the immutable class is the favoured class. For convenience it is aliased to org.mongodb.scala.Document and org.mongodb.scala.bson.Document as well as being available from org.mongodb.scala.bson.collection.immutable.Document. Instances of this type are guaranteed to be immutable for everyone. Such a collection will never change after it is created. Therefore, you can rely on the fact that accessing the same collection value repeatedly at different points in time will always yield a collection with the same elements.

import org.mongodb.scala.bson._

val doc1 = Document("AL" -> BsonString("Alabama"))
val doc2 = doc1 + ("AK" -> BsonString("Alaska"))
val doc3 = doc2 ++ Document("AR" -> BsonString("Arkansas"), "AZ" -> BsonString("Arizona"))

Mutable Documents

To get the mutable Document version, you need to import it explicitly from org.mongodb.scala.collections.mutable.Document. The mutable Document can be updated or extended in place. This means you can change, add, or remove elements of the Document as a side effect. Like scala collections, when dealing with mutable types you need to understand which code changes which collection and when.

import org.mongodb.scala.bson._
import org.mongodb.scala.bson.collection.mutable.Document

val doc = Document("AL" -> BsonString("Alabama"))
val doc1 = doc + ("AK" -> BsonString("Alaska"))   // doc not mutated but new doc created
doc1 ++= Document("AR" -> BsonString("Arkansas"), 
                  "AZ" -> BsonString("Arizona"))  // doc1 mutated as ++= changes in place. 

Implicit conversions

For many of the BsonValue types there are obvious direct mappings from a Scala type. For example, a String maps to BsonString, an Int maps to BsonInt32 and a Long maps to a BsonInt64. For convenience these types can be used directly with Documents and they are converted by the contract traits in the BsonMagnets object. As long as there is an implicit BsonTransformer in scope for any given type, then that type can be converted into a BsonValue.

The following BsonTransformers are in scope by default:

Scala type BsonValue
Boolean => BsonBoolean
String => BsonString
Array[Byte] => BsonBinary
Regex => BsonRegex
Date => BsonDateTime
ObjectId => BsonObjectId
Int => BsonInt32
Long => BsonInt64
Double => BsonDouble
None => BsonNull
immutable.Document => BsonDocument
mutable.Document => BsonDocument
Option[T] => BsonValue where T has a BsonTransformer
Seq[(String, T)] => BsonDocument where T has a BsonTransformer
Seq[T] => BsonArray where T has a BsonTransformer
BsonValue => BsonValue
import org.mongodb.scala.Document

val doc1 = Document("AL" -> "Alabama")
val doc2 = doc1 + ("AK" -> "Alaska")
val doc3 = doc2 ++ Document("AR" -> "Arkansas", "population" -> 2.966)

This is achieved by making use of the “Magnet Pattern”:

The magnet pattern is an alternative approach to method overloading. Rather than defining several identically named methods with different parameter lists you define only one method with only one parameter.

This parameter is called the magnet. Its type is the magnet type, a dedicated type constructed purely as the target of a number of implicit conversions defined in the magnets companion object, which are called the magnet branches and which model the various “overloads”.

Source: The Magnet Pattern

In the API where we would normally expect a single value or a key value pair or many key value pairs eg: (BsonValue, (String, BsonValue) or Iterable[(String, BsonValue)]) we require anything that can become those types via CanBeX traits that handle the implicit conversions necessary to conform to the correct types. These traits are CanBeBsonValue, [CanBeBsonElement](