Writing JSON REST Services
JSON is now the lingua franca between microservices.
In this guide, we see how you can get your REST services to consume and produce JSON payloads.
there is another guide if you need a REST client (including support for JSON). |
This is an introduction to writing JSON REST services with Quarkus. A more detailed guide about RESTEasy Reactive is available here. |
准备
要完成本指南,您需要:
-
Roughly 15 minutes
-
An IDE
-
JDK 11+ installed with
JAVA_HOME
configured appropriately -
Apache Maven 3.9.6
-
Optionally the Quarkus CLI if you want to use it
-
Optionally Mandrel or GraalVM installed and configured appropriately if you want to build a native executable (or Docker if you use a native container build)
架构
The application built in this guide is quite simple: the user can add elements in a list using a form and the list is updated.
All the information between the browser and the server are formatted as JSON.
完整源码
We recommend that you follow the instructions in the next sections and create the application step by step. However, you can go right to the completed example.
Clone the Git repository: git clone https://github.com/quarkusio/quarkus-quickstarts.git
, or download
an archive.
The solution is located in the rest-json-quickstart
directory.
Creating the Maven project
First, we need a new project. Create a new project with the following command:
For Windows users:
-
If using cmd, (don’t use backward slash
\
and put everything on the same line) -
If using Powershell, wrap
-D
parameters in double quotes e.g."-DprojectArtifactId=rest-json-quickstart"
This command generates a new project importing the RESTEasy Reactive/Jakarta REST and Jackson extensions, and in particular adds the following dependency:
<dependency>
<groupId>io.quarkus</groupId>
<artifactId>quarkus-resteasy-reactive-jackson</artifactId>
</dependency>
implementation("io.quarkus:quarkus-resteasy-reactive-jackson")
To improve user experience, Quarkus registers the three Jackson Java 8 modules so you don’t need to do it manually. |
Quarkus also supports JSON-B so, if you prefer JSON-B over Jackson, you can create a project relying on the RESTEasy Reactive JSON-B extension instead:
For Windows users:
-
If using cmd, (don’t use backward slash
\
and put everything on the same line) -
If using Powershell, wrap
-D
parameters in double quotes e.g."-DprojectArtifactId=rest-json-quickstart"
This command generates a new project importing the RESTEasy Reactive/Jakarta REST and JSON-B extensions, and in particular adds the following dependency:
<dependency>
<groupId>io.quarkus</groupId>
<artifactId>quarkus-resteasy-reactive-jsonb</artifactId>
</dependency>
implementation("io.quarkus:quarkus-resteasy-reactive-jsonb")
While named "reactive", RESTEasy Reactive supports equally well both traditional blocking patterns and reactive patterns. For more information about RESTEasy Reactive, please refer to the dedicated guide. |
Creating your first JSON REST service
In this example, we will create an application to manage a list of fruits.
First, let’s create the Fruit
bean as follows:
package org.acme.rest.json;
public class Fruit {
public String name;
public String description;
public Fruit() {
}
public Fruit(String name, String description) {
this.name = name;
this.description = description;
}
}
Nothing fancy. One important thing to note is that having a default constructor is required by the JSON serialization layer.
Now, create the org.acme.rest.json.FruitResource
class as follows:
package org.acme.rest.json;
import java.util.Collections;
import java.util.LinkedHashMap;
import java.util.Set;
import jakarta.ws.rs.DELETE;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.POST;
import jakarta.ws.rs.Path;
@Path("/fruits")
public class FruitResource {
private Set<Fruit> fruits = Collections.newSetFromMap(Collections.synchronizedMap(new LinkedHashMap<>()));
public FruitResource() {
fruits.add(new Fruit("Apple", "Winter fruit"));
fruits.add(new Fruit("Pineapple", "Tropical fruit"));
}
@GET
public Set<Fruit> list() {
return fruits;
}
@POST
public Set<Fruit> add(Fruit fruit) {
fruits.add(fruit);
return fruits;
}
@DELETE
public Set<Fruit> delete(Fruit fruit) {
fruits.removeIf(existingFruit -> existingFruit.name.contentEquals(fruit.name));
return fruits;
}
}
The implementation is pretty straightforward, and you just need to define your endpoints using the Jakarta REST annotations.
The Fruit
objects will be automatically serialized/deserialized by
JSON-B or
Jackson, depending on the extension
you chose when initializing the project.
When a JSON extension is installed such as
|
Configuring JSON support
Jackson
In Quarkus, the default Jackson ObjectMapper
obtained via CDI (and
consumed by the Quarkus extensions) is configured to ignore unknown
properties (by disabling the
DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES
feature).
You can restore the default behavior of Jackson by setting
quarkus.jackson.fail-on-unknown-properties=true
in your
application.properties
or on a per-class basis via
@JsonIgnoreProperties(ignoreUnknown = false)
.
Furthermore, the ObjectMapper
is configured to format dates and time in
ISO-8601 (by disabling the SerializationFeature.WRITE_DATES_AS_TIMESTAMPS
feature).
The default behaviour of Jackson can be restored by setting
quarkus.jackson.write-dates-as-timestamps=true
in your
application.properties
. If you want to change the format for a single
field, you can use the @JsonFormat
annotation.
Also, Quarkus makes it very easy to configure various Jackson settings via
CDI beans. The simplest (and suggested) approach is to define a CDI bean of
type io.quarkus.jackson.ObjectMapperCustomizer
inside of which any Jackson
configuration can be applied.
An example where a custom module needs to be registered would look like so:
import com.fasterxml.jackson.databind.ObjectMapper;
import io.quarkus.jackson.ObjectMapperCustomizer;
import jakarta.inject.Singleton;
@Singleton
public class RegisterCustomModuleCustomizer implements ObjectMapperCustomizer {
public void customize(ObjectMapper mapper) {
mapper.registerModule(new CustomModule());
}
}
Users can even provide their own ObjectMapper
bean if they so choose. If
this is done, it is very important to manually inject and apply all
io.quarkus.jackson.ObjectMapperCustomizer
beans in the CDI producer that
produces ObjectMapper
. Failure to do so will prevent Jackson specific
customizations provided by various extensions from being applied.
import com.fasterxml.jackson.databind.ObjectMapper;
import io.quarkus.arc.All;
import io.quarkus.jackson.ObjectMapperCustomizer;
import java.util.List;
import jakarta.inject.Singleton;
public class CustomObjectMapper {
// Replaces the CDI producer for ObjectMapper built into Quarkus
@Singleton
@Produces
ObjectMapper objectMapper(@All List<ObjectMapperCustomizer> customizers) {
ObjectMapper mapper = myObjectMapper(); // Custom `ObjectMapper`
// Apply all ObjectMapperCustomizer beans (incl. Quarkus)
for (ObjectMapperCustomizer customizer : customizers) {
customizer.customize(mapper);
}
return mapper;
}
}
Mixin support
Quarkus automates the registration of Jackson’s Mixin support, via the
io.quarkus.jackson.JacksonMixin
annotation. This annotation can be placed
on classes that are meant to be used as Jackson mixins while the classes
they are meant to customize are defined as the value of the annotation.
JSON-B
As stated above, Quarkus provides the option of using JSON-B instead of
Jackson via the use of the quarkus-resteasy-jsonb
extension.
Following the same approach as described in the previous section, JSON-B can
be configured using a io.quarkus.jsonb.JsonbConfigCustomizer
bean.
If for example a custom serializer named FooSerializer
for type
com.example.Foo
needs to be registered with JSON-B, the addition of a bean
like the following would suffice:
import io.quarkus.jsonb.JsonbConfigCustomizer;
import jakarta.inject.Singleton;
import jakarta.json.bind.JsonbConfig;
import jakarta.json.bind.serializer.JsonbSerializer;
@Singleton
public class FooSerializerRegistrationCustomizer implements JsonbConfigCustomizer {
public void customize(JsonbConfig config) {
config.withSerializers(new FooSerializer());
}
}
A more advanced option would be to directly provide a bean of
jakarta.json.bind.JsonbConfig
(with a Dependent
scope) or in the extreme
case to provide a bean of type jakarta.json.bind.Jsonb
(with a Singleton
scope). If the latter approach is leveraged it is very important to
manually inject and apply all io.quarkus.jsonb.JsonbConfigCustomizer
beans
in the CDI producer that produces jakarta.json.bind.Jsonb
. Failure to do
so will prevent JSON-B specific customizations provided by various
extensions from being applied.
import io.quarkus.jsonb.JsonbConfigCustomizer;
import jakarta.enterprise.context.Dependent;
import jakarta.enterprise.inject.Instance;
import jakarta.json.bind.JsonbConfig;
public class CustomJsonbConfig {
// Replaces the CDI producer for JsonbConfig built into Quarkus
@Dependent
JsonbConfig jsonConfig(Instance<JsonbConfigCustomizer> customizers) {
JsonbConfig config = myJsonbConfig(); // Custom `JsonbConfig`
// Apply all JsonbConfigCustomizer beans (incl. Quarkus)
for (JsonbConfigCustomizer customizer : customizers) {
customizer.customize(config);
}
return config;
}
}
Creating a frontend
Now let’s add a simple web page to interact with our FruitResource
.
Quarkus automatically serves static resources located under the
META-INF/resources
directory. In the
src/main/resources/META-INF/resources
directory, add a fruits.html
file
with the content from this
fruits.html
file in it.
You can now interact with your REST service:
-
start Quarkus with:
CLIquarkus dev
Maven./mvnw quarkus:dev
Gradle./gradlew --console=plain quarkusDev
-
open a browser to
http://localhost:8080/fruits.html
-
add new fruits to the list via the form
Building a native executable
You can build a native executable with the usual command:
quarkus build --native
./mvnw install -Dnative
./gradlew build -Dquarkus.package.type=native
Running it is as simple as executing
./target/rest-json-quickstart-1.0.0-SNAPSHOT-runner
.
You can then point your browser to http://localhost:8080/fruits.html
and
use your application.
About serialization
JSON serialization libraries use Java reflection to get the properties of an object and serialize them.
When using native executables with GraalVM, all classes that will be used
with reflection need to be registered. The good news is that Quarkus does
that work for you most of the time. So far, we haven’t registered any
class, not even Fruit
, for reflection usage and everything is working
fine.
Quarkus performs some magic when it is capable of inferring the serialized
types from the REST methods. When you have the following REST method,
Quarkus determines that Fruit
will be serialized:
@GET
public List<Fruit> list() {
// ...
}
Quarkus does that for you automatically by analyzing the REST methods at build time and that’s why we didn’t need any reflection registration in the first part of this guide.
Another common pattern in the Jakarta REST world is to use the Response
object. Response
comes with some nice perks:
-
you can return different entity types depending on what happens in your method (a
Legume
or anError
for instance); -
you can set the attributes of the
Response
(the status comes to mind in the case of an error).
Your REST method then looks like this:
@GET
public Response list() {
// ...
}
It is not possible for Quarkus to determine at build time the type included
in the Response
as the information is not available. In this case,
Quarkus won’t be able to automatically register for reflection the required
classes.
This leads us to our next section.
Using Response
Let’s create the Legume
class which will be serialized as JSON, following
the same model as for our Fruit
class:
package org.acme.rest.json;
public class Legume {
public String name;
public String description;
public Legume() {
}
public Legume(String name, String description) {
this.name = name;
this.description = description;
}
}
Now let’s create a LegumeResource
REST service with only one method which
returns the list of legumes.
This method returns a Response
and not a list of Legume
.
package org.acme.rest.json;
import java.util.Collections;
import java.util.LinkedHashSet;
import java.util.Set;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.core.Response;
@Path("/legumes")
public class LegumeResource {
private Set<Legume> legumes = Collections.synchronizedSet(new LinkedHashSet<>());
public LegumeResource() {
legumes.add(new Legume("Carrot", "Root vegetable, usually orange"));
legumes.add(new Legume("Zucchini", "Summer squash"));
}
@GET
public Response list() {
return Response.ok(legumes).build();
}
}
Now let’s add a simple web page to display our list of legumes. In the
src/main/resources/META-INF/resources
directory, add a legumes.html
file
with the content from this
legumes.html
file in it.
Open a browser to http://localhost:8080/legumes.html, and you will see our list of legumes.
The interesting part starts when running the application as a native executable:
-
create the native executable with:
CLIquarkus build --native
Maven./mvnw install -Dnative
Gradle./gradlew build -Dquarkus.package.type=native
-
execute it with
./target/rest-json-quickstart-1.0.0-SNAPSHOT-runner
-
open a browser and go to http://localhost:8080/legumes.html
No legumes there.
As mentioned above, the issue is that Quarkus was not able to determine the
Legume
class will require some reflection by analyzing the REST
endpoints. The JSON serialization library tries to get the list of fields
of Legume
and gets an empty list, so it does not serialize the fields'
data.
At the moment, when JSON-B or Jackson tries to get the list of fields of a class, if the class is not registered for reflection, no exception will be thrown. GraalVM will simply return an empty list of fields. Hopefully, this will change in the future and make the error more obvious. |
We can register Legume
for reflection manually by adding the
@RegisterForReflection
annotation on our Legume
class:
import io.quarkus.runtime.annotations.RegisterForReflection;
@RegisterForReflection
public class Legume {
// ...
}
The @RegisterForReflection annotation instructs Quarkus to keep the class
and its members during the native compilation. More details about the
@RegisterForReflection annotation can be found on the
native
application tips page.
|
Let’s do that and follow the same steps as before:
-
hit
Ctrl+C
to stop the application -
create the native executable with:
CLIquarkus build --native
Maven./mvnw install -Dnative
Gradle./gradlew build -Dquarkus.package.type=native
-
execute it with
./target/rest-json-quickstart-1.0.0-SNAPSHOT-runner
-
open a browser and go to http://localhost:8080/legumes.html
This time, you can see our list of legumes.
Being reactive
You can return reactive types to handle asynchronous processing. Quarkus recommends the usage of Mutiny to write reactive and asynchronous code.
RESTEasy Reactive is naturally integrated with Mutiny.
Your endpoints can return Uni
or Multi
instances:
@GET
@Path("/{name}")
public Uni<Fruit> getOne(String name) {
return findByName(name);
}
@GET
public Multi<Fruit> getAll() {
return findAll();
}
Use Uni
when you have a single result. Use Multi
when you have multiple
items that may be emitted asynchronously.
You can use Uni
and Response
to return asynchronous HTTP responses:
Uni<Response>
.
More details about Mutiny can be found in Mutiny - an intuitive reactive programming library.
Conclusion
Creating JSON REST services with Quarkus is easy as it relies on proven and well known technologies.
As usual, Quarkus further simplifies things under the hood when running your application as a native executable.
There is only one thing to remember: if you use Response
and Quarkus can’t
determine the beans that are serialized, you need to annotate them with
@RegisterForReflection
.