critter

Critter will look at your source code and generate type safe criteria builders for each model object. To use it, you simply need to add a plugin to your maven pom:

<plugin>
    <groupId>dev.morphia.critter</groupId>
    <artifactId>critter-maven</artifactId>
    <version>${critter.version}</version>
    <executions>
        <execution>
            <id>critter</id>
            <goals>
                <goal>generate</goal>
            </goals>
            <configuration>
                <force>false</force>
                <criteriaPackage>com.bob.bar</criteriaPackage>
            </configuration>
        </execution>
    </executions>
</plugin>

This will generate your criteria classes in target/generated-sources/critter and add the directory as a source directory of your maven project. If the criteriaPackage option is left out, the code will be generated using the package of your entities with .criteria appended. (The options are shown here for documentation purposes using the default values. They can be left out altogether if the defaults are acceptable.)

Include the dependency in your pom.xml like this:

<dependency>
    <groupId>dev.morphia.critter</groupId>
    <artifactId>critter-core</artifactId>
    <version>${critter.version}</version>
</dependency>

Critter also requires Morphia 2.0.0 and is built for Java 11.

What difference does it make?

Before critter, your criteria might look something like this:

Query<Book> query = ds.find(Book.class)
    .filter(and(
        eq("bookmark",bookmark),
        eq("database",database)));

But using critter, it would look like this:

Query<Book> query = ds.find(Book.class)
    .filter(and(
        Person.bookmark().eq(bookmark),
        Person.database().eq(database)));

Notice how bookmark() and database() methods were created based on the model object Book's fields. The query filters you're familiar with from Morphia's filters API are all there but now only take the type of the field itself. With this code in place if the model object changes, the code above runs the risk of failing to compile allowing you to catch model/query conflicts at compile time rather than waiting for things to fail at runtime (or in your tests if you're lucky enough to have those).

You can see a working example in the tests.

IDEA Users

IDEA users will need to enable the plugin registry in the maven configuration options for IDEA to pick up the plugin.