Since MongoDB uses BSON, a binary form of JSON, to store its documents, a JSON mapper is a perfect mechanism for mapping Java objects to MongoDB documents. And the best Java JSON mapper is Jackson. Jackson's parsing/generating interface fits serialising to MongoDBs documents like a glove. Its plugins, custom creators, serialisers, views, pluggable annotators and so on give this mapping library a massive head start, making it powerful, performant, and robust.
The official documentation for this project lives here.
Looking specifically for the Maven dependency snippet? It's here.
The MongoDB Jackson Mapper users mailing list is hosted here.
Inserting objects is done like this:
JacksonDBCollection<MyObject, String> coll = JacksonDBCollection.wrap(dbCollection, MyObject.class,
String.class);
MyObject myObject = ...
WriteResult<MyObject, String> result = coll.insert(myObject);
String id = result.getSavedId();
MyObject savedObject = result.getSavedObject();
Both the object itself and the id of the object are strongly typed. If the id is generated, you can easily obtain it from the write result. Finding an object by ID is also simple:
MyObject foundObject = coll.findOneById(id);
Querying can be done using chained query builder methods on the DBCursor:
DBCursor<MyObject> cursor = coll.find().is("prop", "value");
if (cursor.hasNext()) {
MyObject firstObject = cursor.next();
}
The collection, cursor and write result interfaces are very similar to the standard Java MongoDB driver. Most methods have been copied across, with generic typing added where appropriate, and overloading to use the generic type where sometimes the generic type is not powerful enough, such as for queries and specifying fields for partial objects.
When it comes to mapping your objects, generally all you need to use is the Jackson annotations, such as @JsonProperty
and @JsonCreator
. If you want a type of ObjectId
, you have two options, either make your field be of type ObjectId
, or you can also use String
, as long as you annotate both the serialising and deserialising properties with @net.vz.mongodb.jackson.ObjectId
. For example:
public class MyObject {
private String id;
@ObjectId
@JsonProperty("_id")
public String getId() {
return id;
}
@ObjectId
@JsonProperty("_id")
public void setId(String id) {
this.id = id;
}
}
Now your id property will be stored in the database as an object ID, and you can let MongoDB generate it for you. You might not like annotating your ids with @JsonProperty("_id")
, the mapper supports @javax.persistence.Id
as a short hand for this:
public class MyObject {
@Id
public Long id;
}
Another useful implication of this is if you want to use the same object for database objects and objects to return on the web, you can name the id whatever you want for the web, and you don't need to use Jackson views to to specify which property gets mapped to what name for the database and for the web.
The only limitation to using the id annotation is if you are using @Creator
annotated constructors or factory methods, because @javax.persistence.Id
is not supported on method parameters. For this reason, the mapper provides the annotation @net.vz.mongodb.jackson.Id
, and it can be used like so:
public class MyObject {
private final String id;
@JsonCreator
public MyObject(@Id @ObjectId id) {
this.id = id;
}
@Id
@ObjectId
public String getId() {
return id;
}
}
As you can see, immutable objects are also supported because Jackson supports them, something that most other frameworks don't support.
If you're using your data objects for both storage and web views, you might want to take advantage of Jacksons views feature, so that generated/transient properties aren't persisted, and properties that you don't want leaked and serialised to the web. The mapper supports this easily, by letting you pass in a view to the wrap method:
JacksonDBCollection<MyObject, String> coll = JacksonDBCollection.wrap(DBCollection dbCollection, MyObject.class,
String.class, DatabaseView.class);
Of course, if you really want to control things and Jackson's annotations aren't enough, the wrap method is also overloaded to accept an ObjectMapper
. For convenience, you should add the object ID module in order to get the object id and id annotation mapping features:
ObjectMapper myObjectMapper = ...
myObjectMapper.withModule(net.vz.mongodb.jackson.internal.MongoJacksonMapperModule.INSTANCE);
JacksonDBCollection<MyObject, String> coll = JacksonDBCollection.wrap(DBCollection dbCollection, MyObject.class,
String.class, myObjectMapper);