The English version of quarkus.io is the official project site. Translated sites are community supported on a best-effort basis.
Edit this Page

Writing REST Services with Quarkus REST (formerly RESTEasy Reactive)

This guide explains how to write REST Services with Quarkus REST in Quarkus.

This is the reference guide for Quarkus REST. For a more lightweight introduction, please refer to the Writing JSON REST services guides.

What is Quarkus REST?

Quarkus REST is a new Jakarta REST (formerly known as JAX-RS) implementation written from the ground up to work on our common Vert.x layer and is thus fully reactive, while also being very tightly integrated with Quarkus and consequently moving a lot of work to build time.

You should be able to use it in place of any Jakarta REST implementation, but on top of that it has great performance for both blocking and non-blocking endpoints, and a lot of new features on top of what Jakarta REST provides.

Writing endpoints

Getting started

Add the following import to your build file:

pom.xml
<dependency>
    <groupId>io.quarkus</groupId>
    <artifactId>quarkus-rest</artifactId>
</dependency>
build.gradle
implementation("io.quarkus:quarkus-rest")

You can now write your first endpoint in the org.acme.rest.Endpoint class:

package org.acme.rest;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

@Path("")
public class Endpoint {

    @GET
    public String hello() {
        return "Hello, World!";
    }
}

Terminology

REST

REpresentational State Transfer

Endpoint

Java method which is called to serve a REST call

URL / URI (Uniform Resource Locator / Identifier)

Used to identify the location of REST resources (specification)

Resource

Represents your domain object. This is what your API serves and modifies. Also called an entity in Jakarta REST.

Representation

How your resource is represented on the wire, can vary depending on content types

Content type

Designates a particular representation (also called a media type), for example text/plain or application/json

HTTP

Underlying wire protocol for routing REST calls (see HTTP specifications)

HTTP request

The request part of the HTTP call, consisting of an HTTP method, a target URI, headers and an optional message body

HTTP response

The response part of the HTTP call, consisting of an HTTP response status, headers and an optional message body

Declaring endpoints: URI mapping

Any class annotated with a @Path annotation can have its methods exposed as REST endpoints, provided they have an HTTP method annotation (see below).

That @Path annotation defines the URI prefix under which those methods will be exposed. It can be empty, or contain a prefix such as rest or rest/V1.

Each exposed endpoint method can in turn have another @Path annotation which adds to its containing class annotation. For example, this defines a rest/hello endpoint:

package org.acme.rest;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

@Path("rest")
public class Endpoint {

    @Path("hello")
    @GET
    public String hello() {
        return "Hello, World!";
    }
}

See URI parameters for more information about URI mapping.

You can set the root path for all rest endpoints using the @ApplicationPath annotation, as shown below.

package org.acme.rest;

import jakarta.ws.rs.ApplicationPath;
import jakarta.ws.rs.core.Application;

@ApplicationPath("/api")
public static class MyApplication extends Application {

}

This will cause all rest endpoints to be resolve relative to /api, so the endpoint above with @Path("rest") would be accessible at /api/rest/. You can also set the quarkus.rest.path build time property to set the root path if you don’t want to use an annotation.

Declaring endpoints: HTTP methods

Each endpoint method must be annotated with one of the following annotations, which defines which HTTP method will be mapped to the method:

Table 1. HTTP method annotations
Annotation Uso

@GET

Obtain a resource representation, should not modify state, idempotent (HTTP docs)

@HEAD

Obtain metadata about a resource, similar to GET with no body (HTTP docs)

@POST

Create a resource and obtain a link to it (HTTP docs)

@PUT

Replace a resource or create one, should be idempotent (HTTP docs)

@DELETE

Delete an existing resource, idempotent (HTTP docs)

@OPTIONS

Obtain information about a resource, idempotent (HTTP docs)

@PATCH

Update a resource, or create one, not idempotent (HTTP docs)

You can also declare other HTTP methods by declaring them as an annotation with the @HttpMethod annotation:

package org.acme.rest;

import java.lang.annotation.Retention;
import java.lang.annotation.RetentionPolicy;

import jakarta.ws.rs.HttpMethod;
import jakarta.ws.rs.Path;

@Retention(RetentionPolicy.RUNTIME)
@HttpMethod("CHEESE")
@interface CHEESE {
}

@Path("")
public class Endpoint {

    @CHEESE
    public String hello() {
        return "Hello, Cheese World!";
    }
}

Declaring endpoints: representation / content types

Each endpoint method may consume or produce specific resource representations, which are indicated by the HTTP Content-Type header, which in turn contains MIME (Media Type) values, such as the following:

  • text/plain which is the default for any endpoint returning a String.

  • text/html for HTML (such as with Qute templating)

  • application/json for a JSON REST endpoint

  • text/* which is a sub-type wildcard for any text media type

  • */* which is a wildcard for any media type

You may annotate your endpoint class with the @Produces or @Consumes annotations, which allow you to specify one or more media types that your endpoint may accept as HTTP request body or produce as HTTP response body. Those class annotations apply to each method.

Any method may also be annotated with the @Produces or @Consumes annotations, in which case they override any eventual class annotation.

The MediaType class has many constants you can use to point to specific pre-defined media types.

See the Negotiation section for more information.

Accessing request parameters

don’t forget to configure your compiler to generate parameter name information with -parameters (javac) or <parameters> or <maven.compiler.parameters> (Maven).

The following HTTP request elements may be obtained by your endpoint method:

Table 2. HTTP request parameter annotations
HTTP element Annotation Uso

Path parameter

@RestPath (or nothing)

URI template parameter (simplified version of the URI Template specification), see URI parameters for more information.

Query parameter

@RestQuery

The value of a URI query parameter

Header

@RestHeader

The value of an HTTP header

Cookie

@RestCookie

The value of an HTTP cookie

Form parameter

@RestForm

The value of an HTTP URL-encoded FORM

Matrix parameter

@RestMatrix

The value of an URI path segment parameter

For each of those annotations, you may specify the name of the element they refer to, otherwise, they will use the name of the annotated method parameter.

If a client made the following HTTP call:

POST /cheeses;variant=goat/tomme?age=matured HTTP/1.1
Content-Type: application/x-www-form-urlencoded
Cookie: level=hardcore
X-Cheese-Secret-Handshake: fist-bump

smell=strong

Then you could obtain all the various parameters with this endpoint method:

package org.acme.rest;

import jakarta.ws.rs.POST;
import jakarta.ws.rs.Path;

import org.jboss.resteasy.reactive.RestCookie;
import org.jboss.resteasy.reactive.RestForm;
import org.jboss.resteasy.reactive.RestHeader;
import org.jboss.resteasy.reactive.RestMatrix;
import org.jboss.resteasy.reactive.RestPath;
import org.jboss.resteasy.reactive.RestQuery;

@Path("/cheeses/{type}")
public class Endpoint {

    @POST
    public String allParams(@RestPath String type,
                            @RestMatrix String variant,
                            @RestQuery String age,
                            @RestCookie String level,
                            @RestHeader("X-Cheese-Secret-Handshake")
                            String secretHandshake,
                            @RestForm String smell) {
        return type + "/" + variant + "/" + age + "/" + level + "/"
            + secretHandshake + "/" + smell;
    }
}
The @RestPath annotation is optional: any parameter whose name matches an existing URI template variable will be automatically assumed to have @RestPath.

You can also use any of the Jakarta REST annotations @PathParam, @QueryParam, @HeaderParam, @CookieParam, @FormParam or @MatrixParam for this, but they require you to specify the parameter name.

See Parameter mapping for more advanced use-cases.

When an exception occurs in Quarkus REST request parameter handling code, the exception is not printed by default to the log (for security reasons). This can sometimes make it hard to understand why certain HTTP status codes are returned (as the Jakarta REST mandates the use of non-intuitive error codes in various cases). In such cases, users are encouraged to set the logging level for the org.jboss.resteasy.reactive.server.handlers.ParameterHandler category to DEBUG like so:

quarkus.log.category."org.jboss.resteasy.reactive.server.handlers.ParameterHandler".level=DEBUG

Grouping parameters in a custom class

You can group your request parameters in a container class instead of declaring them as method parameters to you endpoint, so we can rewrite the previous example like this:

package org.acme.rest;

import jakarta.ws.rs.POST;
import jakarta.ws.rs.Path;

import org.jboss.resteasy.reactive.RestCookie;
import org.jboss.resteasy.reactive.RestForm;
import org.jboss.resteasy.reactive.RestHeader;
import org.jboss.resteasy.reactive.RestMatrix;
import org.jboss.resteasy.reactive.RestPath;
import org.jboss.resteasy.reactive.RestQuery;

@Path("/cheeses/{type}")
public class Endpoint {

    public static class Parameters {
        @RestPath
        String type;

        @RestMatrix
        String variant;

        @RestQuery
        String age;

        @RestCookie
        String level;

        @RestHeader("X-Cheese-Secret-Handshake")
        String secretHandshake;

        @RestForm
        String smell;
    }

    @POST
    public String allParams(@BeanParam Parameters parameters) { (1)
        return parameters.type + "/" + parameters.variant + "/" + parameters.age
            + "/" + parameters.level + "/" + parameters.secretHandshake
            + "/" + parameters.smell;
    }
}
1 BeanParam is required to comply with the Jakarta REST specification so that libraries like OpenAPI can introspect the parameters.

Record classes are also supported, so you could rewrite the previous example as a record:

    public record Parameters(
        @RestPath
        String type,

        @RestMatrix
        String variant,

        @RestQuery
        String age,

        @RestCookie
        String level,

        @RestHeader("X-Cheese-Secret-Handshake")
        String secretHandshake,

        @RestForm
        String smell){}

Declaring URI parameters

You can declare URI parameters and use regular expressions in your path, so for instance the following endpoint will serve requests for /hello/stef/23 and /hello but not /hello/stef/0x23:

package org.acme.rest;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

@Path("hello")
public class Endpoint {

    @Path("{name}/{age:\\d+}")
    @GET
    public String personalisedHello(String name, int age) {
        return "Hello " + name + " is your age really " + age + "?";
    }

    @GET
    public String genericHello() {
        return "Hello stranger";
    }
}

Accessing the request body

Any method parameter with no annotation will receive the method body.[1], after it has been mapped from its HTTP representation to the Java type of the parameter.

The following parameter types will be supported out of the box:

Table 3. Request body parameter types
Tipo Uso

File

The entire request body in a temporary file

byte[]

The entire request body, not decoded

char[]

The entire request body, decoded

String

The entire request body, decoded

InputStream

The request body in a blocking stream

Reader

The request body in a blocking stream

All Java primitives and their wrapper classes

Java primitive types

BigDecimal, BigInteger

Large integers and decimals.

JsonArray, JsonObject, JsonStructure, JsonValue

JSON value types

Buffer

Vert.x Buffer

any other type

Will be mapped from JSON to that type

You can add support for more body parameter types.

Handling Multipart Form data

To handle HTTP requests that have multipart/form-data as their content type, you can use the regular @RestForm annotation, but we have special types that allow you to access the parts as files or as entities. Let us look at an example of its use.

Assuming an HTTP request containing a file upload, a JSON entity and a form value containing a string description, we could write the following endpoint:

import jakarta.ws.rs.POST;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.core.MediaType;

import org.jboss.resteasy.reactive.PartType;
import org.jboss.resteasy.reactive.RestForm;
import org.jboss.resteasy.reactive.multipart.FileUpload;

@Path("multipart")
public class MultipartResource {
    public static class Person {
        public String firstName;
        public String lastName;
    }

    @POST
    public void multipart(@RestForm String description,
            @RestForm("image") FileUpload file,
            @RestForm @PartType(MediaType.APPLICATION_JSON) Person person) {
        // do something
    }
}

The description parameter will contain the data contained in the part of HTTP request called description (because @RestForm does not define a value, the field name is used), while the file parameter will contain data about the uploaded file in the image part of HTTP request, and the person parameter will read the Person entity using the JSON body reader.

The size of every part in a multipart request must conform to the value of quarkus.http.limits.max-form-attribute-size, for which the default is 2048 bytes. Any request with a part of a size exceeding this configuration will result in HTTP status code 413.

FileUpload provides access to various metadata of the uploaded file. If however all you need is a handle to the uploaded file, java.nio.file.Path or java.io.File could be used.

If you need access to all uploaded files for all parts regardless of their names, you can do it with @RestForm(FileUpload.ALL) List<FileUpload>.

@PartType is used to aid in deserialization of the corresponding part of the request into the desired Java type. It is only required if you need to use a special body parameter type for that particular parameter.
Just like for any other request parameter type, you can also group them into a container class.
When handling file uploads, it is very important to move the file to permanent storage (like a database, a dedicated file system or a cloud storage) in your code that handles the POJO. Otherwise, the file will no longer be accessible when the request terminates. Moreover, if quarkus.http.body.delete-uploaded-files-on-end is set to true, Quarkus will delete the uploaded file when the HTTP response is sent. If the setting is disabled, the file will reside on the file system of the server (in the directory defined by the quarkus.http.body.uploads-directory configuration option), but as the uploaded files are saved with a UUID file name and no additional metadata is saved, these files are essentially a random dump of files.

When a Resource method needs to handle various types of multipart requests, then the org.jboss.resteasy.reactive.server.multipart.MultipartFormDataInput method type can be used as provides access to all the parts of the request.

The following code shows a simple example where we iterate over the parts and return a list of aggregated data:

@Path("/test")
public static class Resource {

    @POST
    @Consumes(MediaType.MULTIPART_FORM_DATA)
    @Produces(MediaType.APPLICATION_JSON)
    public List<Item> hello(MultipartFormDataInput input) throws IOException {
        Map<String, Collection<FormValue>> map = input.getValues();
        List<Item> items = new ArrayList<>();
        for (var entry : map.entrySet()) {
            for (FormValue value : entry.getValue()) {
                items.add(new Item(
                        entry.getKey(),
                        value.isFileItem() ? value.getFileItem().getFileSize() : value.getValue().length(),
                        value.getCharset(),
                        value.getFileName(),
                        value.isFileItem(),
                        value.getHeaders()));
            }

        }
        return items;
    }

    public static class Item {
        public final String name;
        public final long size;
        public final String charset;
        public final String fileName;
        public final boolean isFileItem;
        public final Map<String, List<String>> headers;

        public Item(String name, long size, String charset, String fileName, boolean isFileItem,
                Map<String, List<String>> headers) {
            this.name = name;
            this.size = size;
            this.charset = charset;
            this.fileName = fileName;
            this.isFileItem = isFileItem;
            this.headers = headers;
        }
    }
}

Handling malformed input

As part of reading the multipart body, Quarkus REST invokes the proper MessageBodyReaderMessageBodyReader for each part of the request. If an IOException occurs for one of these parts (for example if Jackson was unable to deserialize a JSON part), then a org.jboss.resteasy.reactive.server.multipart.MultipartPartReadingException is thrown. If this exception is not handled by the application as mentioned in Exception mapping, an HTTP 400 response is returned by default.

Multipart output

Similarly, Quarkus REST can produce Multipart Form data to allow users download files from the server. For example, we could write a POJO that will hold the information we want to expose as:

import jakarta.ws.rs.core.MediaType;

import org.jboss.resteasy.reactive.PartType;
import org.jboss.resteasy.reactive.RestForm;

public class DownloadFormData {

    @RestForm
    String name;

    @RestForm
    @PartType(MediaType.APPLICATION_OCTET_STREAM)
    File file;
}

And then expose this POJO via a Resource like so:

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.Produces;
import jakarta.ws.rs.core.MediaType;

@Path("multipart")
public class Endpoint {

    @GET
    @Produces(MediaType.MULTIPART_FORM_DATA)
    @Path("file")
    public DownloadFormData getFile() {
        // return something
    }
}

Additionally, you can also manually append the parts of the form using the class MultipartFormDataOutput as:

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.Produces;
import jakarta.ws.rs.core.MediaType;

import org.jboss.resteasy.reactive.server.multipart.MultipartFormDataOutput;

@Path("multipart")
public class Endpoint {

    @GET
    @Produces(MediaType.MULTIPART_FORM_DATA)
    @Path("file")
    public MultipartFormDataOutput getFile() {
        MultipartFormDataOutput form = new MultipartFormDataOutput();
        form.addFormData("person", new Person("John"), MediaType.APPLICATION_JSON_TYPE);
        form.addFormData("status", "a status", MediaType.TEXT_PLAIN_TYPE)
                .getHeaders().putSingle("extra-header", "extra-value");
        return form;
    }
}

This last approach allows you adding extra headers to the output part.

For the time being, returning Multipart data is limited to be blocking endpoints.

Returning a response body

In order to return an HTTP response, simply return the resource you want from your method. The method return type and its optional content type will be used to decide how to serialise it to the HTTP response (see the Negotiation section for more advanced information).

You can return any of the pre-defined types that you can read from the HTTP response, and any other type will be mapped from that type to JSON.

In addition, the following return types are also supported:

Table 4. Additional response body parameter types
Tipo Uso

Path

The contents of the file specified by the given path

PathPart

The partial contents of the file specified by the given path

FilePart

The partial contents of a file

AsyncFile

Vert.x AsyncFile, which can be in full, or partial

Alternately, you can also return a reactive type such as Uni, Multi or CompletionStage that resolve to one of the mentioned return types.

Setting other response properties

Manually setting the response

If you need to set more properties on the HTTP response than just the body, such as the status code or headers, you can make your method return org.jboss.resteasy.reactive.RestResponse from a resource method. An example of this could look like:

package org.acme.rest;

import java.time.Duration;
import java.time.Instant;
import java.util.Date;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.core.MediaType;
import jakarta.ws.rs.core.NewCookie;

import org.jboss.resteasy.reactive.RestResponse;
import org.jboss.resteasy.reactive.RestResponse.ResponseBuilder;

@Path("")
public class Endpoint {

    @GET
    public RestResponse<String> hello() {
        // HTTP OK status with text/plain content type
        return ResponseBuilder.ok("Hello, World!", MediaType.TEXT_PLAIN_TYPE)
         // set a response header
         .header("X-Cheese", "Camembert")
         // set the Expires response header to two days from now
         .expires(Date.from(Instant.now().plus(Duration.ofDays(2))))
         // send a new cookie
         .cookie(new NewCookie("Flavour", "chocolate"))
         // end of builder API
         .build();
    }
}
You can also use the Jakarta REST type Response but it is not strongly typed to your entity.

Using annotations

Alternatively, if you only need to set the status code and / or HTTP headers with static values, you can use @org.jboss.resteasy.reactive.ResponseStatus and /or ResponseHeader respectively. An example of this could look like:

package org.acme.rest;

import org.jboss.resteasy.reactive.Header;
import org.jboss.resteasy.reactive.ResponseHeader;
import org.jboss.resteasy.reactive.ResponseStatus;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

@Path("")
public class Endpoint {

    @ResponseStatus(201)
    @ResponseHeader(name = "X-Cheese", value = "Camembert")
    @GET
    public String hello() {
        return "Hello, World!";
    }
}

Redirect support

When handling a @POST, @PUT or @DELETE endpoint, it is common practice to redirect to a @GET endpoint after the action has been performed to allow the user to reload the page without triggering the action a second time. There are multiple ways to achieve this.

Using RestResponse

Using RestResponse as the return type while making sure the proper redirection URI is created can be done as in the following example:

package org.acme.rest;

import java.util.Map;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.atomic.AtomicLong;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.POST;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.core.Context;
import jakarta.ws.rs.core.UriInfo;
import org.jboss.resteasy.reactive.RestResponse;

@Path("/fruits")
public class FruitResource {

    public static class Fruit {
        public Long id;
        public String name;
        public String description;

        public Fruit() {
        }

        public Fruit(Long id, String name, String description) {
            this.id = id;
            this.name = name;
            this.description = description;
        }
    }

    private final Map<Long, Fruit> fruits = new ConcurrentHashMap<>();
    private final AtomicLong ids = new AtomicLong(0);


    public FruitResource() {
        Fruit apple = new Fruit(ids.incrementAndGet(), "Apple", "Winter fruit");
        fruits.put(apple.id, apple);
        Fruit pinneapple = new Fruit(ids.incrementAndGet(), "Pineapple", "Tropical fruit");
        fruits.put(pinneapple.id, pinneapple);
    }

    // when invoked, this method will result in an HTTP redirect to the GET method that obtains the fruit by id
    @POST
    public RestResponse<Fruit> add(Fruit fruit, @Context UriInfo uriInfo) {
        fruit.id = ids.incrementAndGet();
        fruits.put(fruit.id, fruit);
        // seeOther results in an HTTP 303 response with the Location header set to the value of the URI
        return RestResponse.seeOther(uriInfo.getAbsolutePathBuilder().path(Long.toString(fruit.id)).build());
    }

    @GET
    @Path("{id}")
    public Fruit byId(Long id) {
        return fruits.get(id);
    }
}

Using RedirectException

Users can also throw jakarta.ws.rs.RedirectionException from a method body to get Quarkus REST to perform the desired redirect.

Async/reactive support

If your endpoint method needs to accomplish an asynchronous or reactive task before being able to answer, you can declare your method to return the Uni type (from Mutiny), in which case the current HTTP request will be automatically suspended after your method, until the returned Uni instance resolves to a value, which will be mapped to a response exactly according to the previously described rules:

package org.acme.rest;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

import io.smallrye.mutiny.Uni;

@Path("escoffier")
public class Endpoint {

    @GET
    public Uni<Book> culinaryGuide() {
        return Book.findByIsbn("978-2081229297");
    }
}

This allows you to not block the event-loop thread while the book is being fetched from the database, and allows Quarkus to serve more requests until your book is ready to be sent to the client and terminate this request. See Execution Model documentation for more information.

The CompletionStage return type is also supported.

Streaming support

If you want to stream your response element by element, you can make your endpoint method return a Multi type (from Mutiny). This is especially useful for streaming text or binary data.

This example, using Reactive Messaging HTTP shows how to stream text data:

package org.acme.rest;

import jakarta.inject.Inject;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

import org.eclipse.microprofile.reactive.messaging.Channel;

import io.smallrye.mutiny.Multi;

@Path("logs")
public class Endpoint {

    @Inject
    @Channel("log-out")
    Multi<String> logs;

    @GET
    public Multi<String> streamLogs() {
        return logs;
    }
}
Response filters are not invoked on streamed responses, because they would give a false impression that you can set headers or HTTP status codes, which is not true after the initial response. Exception mappers are also not invoked because part of the response may already have been written.

Customizing headers and status

If you need to set custom HTTP headers and / or the HTTP response, then you can return org.jboss.resteasy.reactive.RestMulti instead, like this:

package org.acme.rest;

import jakarta.inject.Inject;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

import org.eclipse.microprofile.reactive.messaging.Channel;

import io.smallrye.mutiny.Multi;
import org.jboss.resteasy.reactive.RestMulti;

@Path("logs")
public class Endpoint {

    @Inject
    @Channel("log-out")
    Multi<String> logs;

    @GET
    public Multi<String> streamLogs() {
        return RestMulti.fromMultiData(logs).status(222).header("foo", "bar").build();
    }
}

In more advanced cases where the headers and / or status can only be obtained from the results of an async call, the RestMulti.fromUniResponse needs to be used. Here is an example of such a use case:

package org.acme.rest;

import jakarta.inject.Inject;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

import java.util.List;import java.util.Map;import org.eclipse.microprofile.reactive.messaging.Channel;

import io.smallrye.mutiny.Multi;
import org.jboss.resteasy.reactive.RestMulti;

@Path("logs")
public class Endpoint {

    interface SomeService {
        Uni<SomeResponse> get();
    }

    interface SomeResponse {
        Multi<byte[]> data;

        String myHeader();
    }

    private final SomeService someService;

    public Endpoint(SomeService someService) {
        this.someService = someService;
    }

    @GET
    public Multi<String> streamLogs() {
        return RestMulti.fromUniResponse(someService.get(), SomeResponse::data, (r -> Map.of("MyHeader", List.of(r.myHeader()))));
    }
}

Concurrent stream element processing

By default, RestMulti ensures serial/sequential order of the items/elements produced by the wrapped Multi by using a value of 1 for the demand signaled to the publishers. To enable concurrent processing/generation of multiple items, use withDemand(long demand).

Using a demand higher than 1 is useful when multiple items shall be returned and the production of each item takes some time, i.e. when parallel/concurrent production improves the service response time. Be aware the concurrent processing also requires more resources and puts a higher load on services or resources that are needed to produce the items. Also consider using Multi.capDemandsTo(long) and Multi.capDemandsUsing(LongFunction).

The example below produces 5 (JSON) strings, but the order of the strings in the returned JSON array is not guaranteed. The below example also works for JSON objects and not just simple types.

package org.acme.rest;

import jakarta.inject.Inject;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

import io.smallrye.mutiny.Multi;
import org.jboss.resteasy.reactive.RestMulti;

@Path("message-stream")
public class Endpoint {
    @GET
    public Multi<String> streamMessages() {
        Multi<String> sourceMulti = Multi
            .createBy()
            .merging()
            .streams(
                Multi.createFrom().items(
                    "message-1",
                    "message-2",
                    "message-3",
                    "message-4",
                    "message-5"
                )
            );

        return RestMulti
            .fromMultiData(sourceMulti)
            .withDemand(5)
            .build();
    }
}

Example response, the order is non-deterministic.

"message-3"
"message-5"
"message-4"
"message-1"
"message-2"

Returning multiple JSON objects

By default, RestMulti returns items/elements produced by the wrapped Multi as a JSON array, if the media-type is application/json. To return separate JSON objects that are not wrapped in a JSON array, use encodeAsArray(false) (encodeAsArray(true) is the default). Note that streaming multiple objects this way requires a slightly different parsing on the client side, but objects can be parsed and consumed as they appear without having to deserialize a possibly huge result at once.

The example below produces 5 (JSON) strings, that are not wrapped in an array, like this:

"message-1"
"message-2"
"message-3"
"message-4"
"message-5"
package org.acme.rest;

import jakarta.inject.Inject;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

import io.smallrye.mutiny.Multi;
import org.jboss.resteasy.reactive.RestMulti;

@Path("message-stream")
public class Endpoint {
    @GET
    public Multi<String> streamMessages() {
        Multi<String> sourceMulti = Multi
            .createBy()
            .merging()
            .streams(
                Multi.createFrom().items(
                    "message-1",
                    "message-2",
                    "message-3",
                    "message-4",
                    "message-5"
                )
            );

        return RestMulti
            .fromMultiData(sourceMulti)
            .encodeAsJsonArray(false)
            .build();
    }
}

Server-Sent Event (SSE) support

If you want to stream JSON objects in your response, you can use Server-Sent Events by just annotating your endpoint method with @Produces(MediaType.SERVER_SENT_EVENTS) and specifying that each element should be serialised to JSON with @RestStreamElementType(MediaType.APPLICATION_JSON).

package org.acme.rest;

import jakarta.inject.Inject;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.Produces;
import jakarta.ws.rs.core.MediaType;

import org.jboss.resteasy.reactive.RestStreamElementType;

import io.smallrye.mutiny.Multi;

import org.eclipse.microprofile.reactive.messaging.Channel;

@Path("escoffier")
public class Endpoint {

    // Inject our Book channel
    @Inject
    @Channel("book-out")
    Multi<Book> books;

    @GET
    // Each element will be sent as JSON
    @RestStreamElementType(MediaType.APPLICATION_JSON)
    // by using @RestStreamElementType, we don't need to add @Produces(MediaType.SERVER_SENT_EVENTS)
    public Multi<Book> stream() {
        return books;
    }
}

Sometimes it’s useful to create a customized SSE message, for example if you need to specify the event field of a SSE message to distinguish various event types. A resource method may return Multi<jakarta.ws.rs.sse.OutboundSseEvent> and an injected jakarta.ws.rs.sse.Sse can be used to create OutboundSseEvent instances.

package org.acme.rest;

import jakarta.inject.Inject;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.Produces;
import jakarta.ws.rs.core.MediaType;
import jakarta.ws.rs.sse.OutboundSseEvent;
import jakarta.ws.rs.sse.Sse;

import org.jboss.resteasy.reactive.RestStreamElementType;

import io.smallrye.mutiny.Multi;

import org.eclipse.microprofile.reactive.messaging.Channel;

@Path("escoffier")
public class Endpoint {

    @Inject
    @Channel("book-out")
    Multi<Book> books;

    @Inject
    Sse sse; (1)

    @GET
    @RestStreamElementType(MediaType.TEXT_PLAIN)
    public Multi<OutboundSseEvent> stream() {
        return books.map(book -> sse.newEventBuilder() (2)
            .name("book")  (3)
            .data(book.title) (4)
            .build());
    }
}
1 Inject the server-side entry point for creating OutboundSseEvents.
2 Create a new outbound event builder.
3 Set the event name, i.e. the value of the event field of a SSE message.
4 Set the data, i.e. the value of the data field of a SSE message.

Manipulation of the returned HTTP headers and status code is not possible via RestMulti.fromUniResponse because when returning SSE responses the headers and status code cannot be delayed until the response becomes available.

Controlling HTTP Caching features

Quarkus REST provides the @Cache and @NoCache annotations to facilitate handling HTTP caching semantics, i.e. setting the 'Cache-Control' HTTP header.

These annotations can be placed either on a Resource Method or a Resource Class (in which case it applies to all Resource Methods of the class that do not contain the same annotation) and allow users to return domain objects and not have to deal with building up the Cache-Control HTTP header explicitly.

While @Cache builds a complex Cache-Control header, @NoCache is a simplified notation to say that you don’t want anything cached; i.e. Cache-Control: nocache.

More information on the Cache-Control header and be found in RFC 7234

Accessing context objects

There are a number of contextual objects that the framework will give you, if your endpoint method takes parameters of the following type:

Table 5. Contextual objects
Tipo Uso

HttpHeaders

All the request headers

ResourceInfo

Information about the current endpoint method and class (requires reflection)

SecurityContext

Access to the current user and roles

SimpleResourceInfo

Information about the current endpoint method and class (no reflection required)

UriInfo

Provides information about the current endpoint and application URI

Application

Advanced: Current Jakarta REST application class

Configuration

Advanced: Configuration about the deployed Jakarta REST application

Providers

Advanced: Runtime access to Jakarta REST providers

Request

Advanced: Access to the current HTTP method and Preconditions

ResourceContext

Advanced: access to instances of endpoints

ServerRequestContext

Advanced: Quarkus REST access to the current request/response

Sse

Advanced: Complex SSE use-cases

HttpServerRequest

Advanced: Vert.x HTTP Request

HttpServerResponse

Advanced: Vert.x HTTP Response

For example, here is how you can return the name of the currently logged-in user:

package org.acme.rest;

import java.security.Principal;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.core.SecurityContext;

@Path("user")
public class Endpoint {

    @GET
    public String userName(SecurityContext security) {
        Principal user = security.getUserPrincipal();
        return user != null ? user.getName() : "<NOT LOGGED IN>";
    }
}

You can also inject those context objects using @Inject on fields of the same type:

package org.acme.rest;

import java.security.Principal;

import jakarta.inject.Inject;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.core.SecurityContext;

@Path("user")
public class Endpoint {

    @Inject
    SecurityContext security;

    @GET
    public String userName() {
        Principal user = security.getUserPrincipal();
        return user != null ? user.getName() : "<NOT LOGGED IN>";
    }
}

Or even on your endpoint constructor:

package org.acme.rest;

import java.security.Principal;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.core.SecurityContext;

@Path("user")
public class Endpoint {

    SecurityContext security;

    Endpoint(SecurityContext security) {
        this.security = security;
    }

    @GET
    public String userName() {
        Principal user = security.getUserPrincipal();
        return user != null ? user.getName() : "<NOT LOGGED IN>";
    }
}

JSON serialisation

Instead of importing io.quarkus:quarkus-rest, you can import either of the following modules to get support for JSON:

GAV Uso

io.quarkus:quarkus-rest-jackson

Jackson support

io.quarkus:quarkus-rest-jsonb

JSON-B support

In both cases, importing those modules will allow HTTP message bodies to be read from JSON and serialised to JSON, for all the types not already registered with a more specific serialisation.

Jackson-specific features

Exception handling

By default, Quarkus provides a built-in ExceptionMapper for MismatchedInputException which returns an HTTP 400 status code along with a good error message in Dev and Test modes, about what went wrong during serialization of an entity.

There are situations where various Jackson related exceptions need to handled in a uniform way.For example, the application may need to handle all JsonMappingException the same way. This becomes a problem when taking JAX-RS / Jakarta REST rules into account, because the exception mapper ExceptionMapper for MismatchedInputException would be used instead of the user provide ExceptionMapper for JsonMappingException (as MismatchedInputException is a subtype of JsonMappingException).

One solution for this case is to configure the following:

quarkus.class-loading.removed-resources."io.quarkus\:quarkus-rest-jackson"=io/quarkus/resteasy/reactive/jackson/runtime/mappers/BuiltinMismatchedInputExceptionMapper.class

which essentially makes Quarkus ignore the ExceptionMapper for MismatchedInputException completely.

Secure serialization

When used with Jackson to perform JSON serialization, Quarkus REST provides the ability to limit the set of fields that are serialized based on the roles of the current user. This is achieved by simply annotating the fields (or getters) of the POJO being returned with @io.quarkus.resteasy.reactive.jackson.SecureField.

A simple example could be the following:

Assume we have a POJO named Person which looks like so:

package org.acme.rest;

import io.quarkus.resteasy.reactive.jackson.SecureField;

public class Person {

    @SecureField(rolesAllowed = "admin")
    private final Long id;
    private final String first;
    private final String last;
    @SecureField(rolesAllowed = "${role:admin}")                 (1)
    private String address;

    public Person(Long id, String first, String last, String address) {
        this.id = id;
        this.first = first;
        this.last = last;
        this.address = address;
    }

    public Long getId() {
        return id;
    }

    public String getFirst() {
        return first;
    }

    public String getLast() {
        return last;
    }

    public String getAddress() {
        return address;
    }

    public void setAddress(String address) {
        this.address = address;
    }
}
1 The io.quarkus.resteasy.reactive.jackson.SecureField.rolesAllowed property supports property expressions exactly in the same fashion the jakarta.annotation.security.RolesAllowed annotation does. For more information, please refer to the Standard security annotations section of the Authorization of web endpoints guide.

A very simple Jakarta REST Resource that uses Person could be:

package org.acme.rest;

import static jakarta.ws.rs.core.MediaType.APPLICATION_JSON;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.Produces;
import jakarta.ws.rs.core.Response;

@Path("person")
public class PersonResource {

    @Path("{id}")
    @GET
    public Person getPerson(Long id) {
        return new Person(id, "foo", "bar", "Brick Lane");
    }

    @Produces(APPLICATION_JSON) (1)
    @Path("/friend/{id}")
    @GET
    public Response getPersonFriend(Long id) {
        var person = new Person(id, "foo", "bar", "Brick Lane");
        return Response.ok(person).build();
    }
}
1 The @SecureField annotation is only effective when Quarkus recognizes that produced content type is the 'application/json' type.
Currently you cannot use the @SecureField annotation to secure your data returned from resource methods returning the io.smallrye.mutiny.Multi reactive type.

All resource methods returning data secured with the @SecureField annotation should be tested. Please make sure data are secured as you intended. Quarkus always attempts to detect fields annotated with the @SecureField annotation, however it may fail to infer returned type and miss the @SecureField annotation instance. If that happens, please explicitly enable secure serialization on the resource endpoint with the @EnableSecureSerialization annotation.

Assuming security has been set up for the application (see our guide for more details), when a user with the admin role performs an HTTP GET on /person/1 they will receive:

{
  "id": 1,
  "first": "foo",
  "last": "bar",
  "address", "Brick Lane"
}

as the response.

Any user however that does not have the admin role will receive:

{
  "first": "foo",
  "last": "bar"
}
No additional configuration needs to be applied for this secure serialization to take place. However, users can use the @io.quarkus.resteasy.reactive.jackson.EnableSecureSerialization and @io.quarkus.resteasy.reactive.jackson.DisableSecureSerialization annotation to opt in or out for specific Jakarta REST Resource classes or methods.
Configuration expressions set with the SecureField.rolesAllowed property are validated during application startup even when the @io.quarkus.resteasy.reactive.jackson.DisableSecureSerialization annotation is used.
@JsonView support

Jakarta REST methods can be annotated with @JsonView in order to customize the serialization of the returned POJO, on a per method-basis. This is best explained with an example.

A typical use of @JsonView is to hide certain fields on certain methods. In that vein, let’s define two views:

public class Views {

    public static class Public {
    }

    public static class Private extends Public {
    }
}

Let’s assume we have the User POJO on which we want to hide some field during serialization. A simple example of this is:

public class User {

    @JsonView(Views.Private.class)
    public int id;

    @JsonView(Views.Public.class)
    public String name;
}

Depending on the Jakarta REST method that returns this user, we might want to exclude the id field from serialization. For example, you might want an insecure method to not expose this field. The way we can achieve that in Quarkus REST is shown in the following example:

@JsonView(Views.Public.class)
@GET
@Path("/public")
public User userPublic() {
    return testUser();
}

@JsonView(Views.Private.class)
@GET
@Path("/private")
public User userPrivate() {
    return testUser();
}

When the result the userPublic method is serialized, the id field will not be contained in the response as the Public view does not include it. The result of userPrivate however will include the id as expected when serialized.

Reflection-free Jackson serialization

Out-of-the-box Jackson serialization converts objects into their JSON representation by introspecting them through a heavy use of reflection. However, the general Quarkus philosophy is to avoid reflection as much as possible, often replacing it with build time code generation. For this reason it is possible to automatically generate at build time implementations of the Jackson StdSerializer, one for each class to be converted in JSON. These generated serializers can be subsequently used by Quarkus at runtime to perform the JSON serialization of the objects returned by a REST endpoint without any use of reflection.

This feature is turned off by default, but it can be enabled by setting the configuration property quarkus.rest.jackson.optimization.enable-reflection-free-serializers=true.

Completely customized per method serialization/deserialization

There are times when you need to completely customize the serialization/deserialization of a POJO on a per Jakarta REST method basis or on a per Jakarta REST resource basis. For such use cases, you can use the @io.quarkus.resteasy.reactive.jackson.CustomSerialization and @io.quarkus.resteasy.reactive.jackson.CustomDeserialization annotations in the REST method or in the REST resource at class level. These annotations allow you to fully configure the com.fasterxml.jackson.databind.ObjectWriter/com.fasterxml.jackson.databind.ObjectReader.

Here is an example use case to customize the com.fasterxml.jackson.databind.ObjectWriter:

@CustomSerialization(UnquotedFields.class)
@GET
@Path("/invalid-use-of-custom-serializer")
public User invalidUseOfCustomSerializer() {
    return testUser();
}

where UnquotedFields is a BiFunction defined as so:

public static class UnquotedFields implements BiFunction<ObjectMapper, Type, ObjectWriter> {

    @Override
    public ObjectWriter apply(ObjectMapper objectMapper, Type type) {
        return objectMapper.writer().without(JsonWriteFeature.QUOTE_FIELD_NAMES);
    }
}

Essentially what this class does is force Jackson to not include quotes in the field names.

It is important to note that this customization is only performed for the serialization of the Jakarta REST methods that use @CustomSerialization(UnquotedFields.class).

Following the previous example, let’s now customize the com.fasterxml.jackson.databind.ObjectReader to read JSON requests with unquoted field names:

@CustomDeserialization(SupportUnquotedFields.class)
@POST
@Path("/use-of-custom-deserializer")
public void useOfCustomSerializer(User request) {
    // ...
}

where SupportUnquotedFields is a BiFunction defined as so:

public static class SupportUnquotedFields implements BiFunction<ObjectMapper, Type, ObjectReader> {

    @Override
    public ObjectReader apply(ObjectMapper objectMapper, Type type) {
        return objectMapper.reader().with(JsonReadFeature.ALLOW_UNQUOTED_FIELD_NAMES);
    }
}

XML serialisation

To enable XML support, add the quarkus-rest-jaxb extension to your project.

GAV Uso

io.quarkus:quarkus-rest-jaxb

XML support

Importing this module will allow HTTP message bodies to be read from XML and serialised to XML, for all the types not already registered with a more specific serialisation.

The JAXB Quarkus REST extension will automatically detect the classes that are used in the resources and require JAXB serialization. Then, it will register these classes into the default JAXBContext which is internally used by the JAXB message reader and writer.

However, in some situations, these classes cause the JAXBContext to fail: for example, when you’re using the same class name in different java packages. In these cases, the application will fail at build time and print the JAXB exception that caused the issue, so you can properly fix it. Alternatively, you can also exclude the classes that cause the issue by using the property quarkus.jaxb.exclude-classes. When excluding classes that are required by any resource, the JAXB Quarkus REST extension will create and cache a custom JAXBContext that will include the excluded class, causing a minimal performance degradance.

The property quarkus.jaxb.exclude-classes accepts a comma separated list of either fully qualified class names or package names. Package names must be suffixed by .* and all classes in the specified package and its subpackages will be excluded.

For instance, when setting quarkus.jaxb.exclude-classes=org.acme.one.Model,org.acme.two.Model,org.acme.somemodel.*, the following elements are excluded:

  • The class org.acme.one.Model

  • The class org.acme.two.Model

  • All the classes in the org.acme.somemodel package and its subpackages

Advanced JAXB-specific features

When using the quarkus-resteasy-reactive-jaxb extension there are some advanced features that Quarkus REST supports.

Inject JAXB components

The JAXB Quarkus REST extension will serialize and unserialize requests and responses transparently for users. However, if you need finer grain control over JAXB components, you can inject either the JAXBContext, Marshaller, or Unmarshaller components into your beans:

@ApplicationScoped
public class MyService {

    @Inject
    JAXBContext jaxbContext;

    @Inject
    Marshaller marshaller;

    @Inject
    Unmarshaller unmarshaller;

    // ...
}

Quarkus will automatically find all the classes annotated with @XmlRootElement and then bound them to the JAXB context.

Customize the JAXB configuration

To customize the JAXB configuration for either the JAXB context, and/or the Marshaller/Unmarshaller components, the suggested approach is to define a CDI bean of type io.quarkus.jaxb.runtime.JaxbContextCustomizer.

An example where a custom module needs to be registered would look like so:

@Singleton
public class RegisterCustomModuleCustomizer implements JaxbContextCustomizer {

    // For JAXB context configuration
    @Override
    public void customizeContextProperties(Map<String, Object> properties) {

    }

    // For Marshaller configuration
    @Override
    public void customizeMarshaller(Marshaller marshaller) throws PropertyException {
        marshaller.setProperty("jaxb.formatted.output", Boolean.TRUE);
    }

    // For Unmarshaller configuration
    @Override
    public void customizeUnmarshaller(Unmarshaller unmarshaller) throws PropertyException {
        // ...
    }
}

It’s not necessary to implement all three methods, but only the want you need.

Alternatively, you can provide your own JAXBContext bean by doing:

public class CustomJaxbContext {

    // Replaces the CDI producer for JAXBContext built into Quarkus
    @Singleton
    @Produces
    JAXBContext jaxbContext() {
        // ...
    }
}

Note that if you provide your custom JAXB context instance, you will need to register the classes you want to use for the XML serialization. This means that Quarkus will not update your custom JAXB context instance with the auto-discovered classes.

GAV Uso

io.quarkus:quarkus-rest-links

Web Links support

Importing this module will allow injecting web links into the response HTTP headers by just annotating your endpoint resources with the @InjectRestLinks annotation. To declare the web links that will be returned, you must use the @RestLink annotation in the linked methods. Assuming a Record looks like:

public class Record {

    // The class must contain/inherit either and `id` field, an `@Id` or `@RestLinkId` annotated field.
    // When resolving the id the order of preference is: `@RestLinkId` > `@Id` > `id` field.
    private int id;

    public Record() {
    }

    protected Record(int id) {
        this.id = id;
    }

    public int getId() {
        return id;
    }

    public void setId(int id) {
        this.id = id;
    }
}

An example of enabling Web Links support would look like:

@Path("/records")
public class RecordsResource {

    @GET
    @RestLink(rel = "list")
    @InjectRestLinks
    public List<Record> getAll() {
        // ...
    }

    @GET
    @Path("/{id}")
    @RestLink(rel = "self")
    @InjectRestLinks(RestLinkType.INSTANCE)
    public Record get(@PathParam("id") int id) {
        // ...
    }

    @PUT
    @Path("/{id}")
    @RestLink
    @InjectRestLinks(RestLinkType.INSTANCE)
    public Record update(@PathParam("id") int id) {
        // ...
    }

    @DELETE
    @Path("/{id}")
    @RestLink
    public Record delete(@PathParam("id") int id) {
        // ...
    }
}

When calling the endpoint /records which is defined by the method getAll within the above resource using curl, you would get the web links header:

& curl -i localhost:8080/records
Link: <http://localhost:8080/records>; rel="list"

As this resource does not return a single instance of type Record, the links for the methods get, update, and delete are not injected. Now, when calling the endpoint /records/1, you would get the following web links:

& curl -i localhost:8080/records/1
Link: <http://localhost:8080/records>; rel="list"
Link: <http://localhost:8080/records/1>; rel="self"
Link: <http://localhost:8080/records/1>; rel="update"
Link: <http://localhost:8080/records/1>; rel="delete"

The get, update, and delete methods use the path param "id" and as the field "id" exists in the entity type "Record", the web link properly populates the value "1" in the returned links. In addition to this, we can also generate web links with path params that do not match with any field of the entity type. For example, the following method is using a path param "text" and the entity Record does not have any field named "text":

@Path("/records")
public class RecordsResource {

    // ...

    @GET
    @Path("/search/{text}")
    @RestLink(rel = "search records by free text")
    @InjectRestLinks
    public List<Record> search(@PathParam("text") String text) { (4)
        // ...
    }

    // ...
}

The generated web link for this resource is Link: <http://localhost:8080/search/{text}>; rel="search records by free text".

Finally, when calling the delete resource, you should not see any web links as the method delete is not annotated with the @InjectRestLinks annotation.

You can programmatically have access to the web links registry just by injecting the RestLinksProvider bean:

@Path("/records")
public class RecordsResource {

    @Inject
    RestLinksProvider linksProvider;

    // ...
}

Using this injected bean of type RestLinksProvider, you can get the links by type using the method RestLinksProvider.getTypeLinks or get the links by a concrete instance using the method RestLinksProvider.getInstanceLinks.

JSON Hypertext Application Language (HAL) support

The HAL standard is a simple format to represent web links.

To enable the HAL support, add the quarkus-hal extension to your project. Also, as HAL needs JSON support, you need to add either the quarkus-rest-jsonb or the quarkus-rest-jackson extension.

GAV Uso

io.quarkus:quarkus-hal

HAL

After adding the extensions, we can now annotate the REST resources to produce the media type application/hal+json (or use RestMediaType.APPLICATION_HAL_JSON). For example:

@Path("/records")
public class RecordsResource {

    @GET
    @Produces({ MediaType.APPLICATION_JSON, RestMediaType.APPLICATION_HAL_JSON })
    @RestLink(rel = "list")
    @InjectRestLinks
    public List<Record> getAll() {
        // ...
    }

    @GET
    @Produces({ MediaType.APPLICATION_JSON, RestMediaType.APPLICATION_HAL_JSON })
    @Path("/{id}")
    @RestLink(rel = "self")
    @InjectRestLinks(RestLinkType.INSTANCE)
    public Record get(@PathParam("id") int id) {
        // ...
    }
}

Now, the endpoints /records and /records/{id} will accept the media type both json and hal+json to print the records in Hal format.

For example, if we invoke the /records endpoint using curl to return a list of records, the HAL format will look like as follows:

& curl -H "Accept:application/hal+json" -i localhost:8080/records
{
    "_embedded": {
        "items": [
            {
                "id": 1,
                "slug": "first",
                "value": "First value",
                "_links": {
                    "self": {
                        "href": "http://localhost:8081/records/1"
                    },
                    "list": {
                        "href": "http://localhost:8081/records"
                    }
                }
            },
            {
                "id": 2,
                "slug": "second",
                "value": "Second value",
                "_links": {
                    "self": {
                        "href": "http://localhost:8081/records/2"
                    },
                    "list": {
                        "href": "http://localhost:8081/records"
                    }
                }
            }
        ]
    },
    "_links": {
        "list": {
            "href": "http://localhost:8081/records"
        }
    }
}

When we call a resource /records/1 that returns only one instance, then the output is:

& curl -H "Accept:application/hal+json" -i localhost:8080/records/1
{
    "id": 1,
    "slug": "first",
    "value": "First value",
    "_links": {
        "self": {
            "href": "http://localhost:8081/records/1"
        },
        "list": {
            "href": "http://localhost:8081/records"
        }
    }
}

Finally, you can also provide additional HAL links programmatically in your resource just by returning either HalCollectionWrapper<T> (to return a list of entities) or HalEntityWrapper<T> (to return a single object) as described in the following example:

@Path("/records")
public class RecordsResource {

    @Inject
    HalService halService;

    @GET
    @Produces({ MediaType.APPLICATION_JSON, RestMediaType.APPLICATION_HAL_JSON })
    @RestLink(rel = "list")
    public HalCollectionWrapper<Record> getAll() {
        List<Record> list = // ...
        HalCollectionWrapper<Record> halCollection = halService.toHalCollectionWrapper( list, "collectionName", Record.class);
        halCollection.addLinks(Link.fromPath("/records/1").rel("first-record").build());
        return halCollection;
    }

    @GET
    @Produces({ MediaType.APPLICATION_JSON, RestMediaType.APPLICATION_HAL_JSON })
    @Path("/{id}")
    @RestLink(rel = "self")
    @InjectRestLinks(RestLinkType.INSTANCE)
    public HalEntityWrapper<Record> get(@PathParam("id") int id) {
        Record entity = // ...
        HalEntityWrapper<Record> halEntity = halService.toHalWrapper(entity);
        halEntity.addLinks(Link.fromPath("/records/1/parent").rel("parent-record").build());
        return halEntity;
    }
}

CORS filter

Cross-origin resource sharing (CORS) is a mechanism that allows restricted resources on a web page to be requested from another domain outside the domain from which the first resource was served.

Quarkus includes a CORS filter at the HTTP layer level. For more information about the CORS filters and their usage, see the CORS filter section of the Quarkus "Cross-origin resource sharing" guide.

More advanced usage

Here are some more advanced topics that you may not need to know about initially, but could prove useful for more complex use cases.

Execution model, blocking, non-blocking

Quarkus REST is implemented using two main thread types:

  • Event-loop threads: which are responsible, among other things, for reading bytes from the HTTP request and writing bytes back to the HTTP response

  • Worker threads: they are pooled and can be used to offload long-running operations

The event-loop threads (also called IO threads) are responsible for actually performing all the IO operations in an asynchronous way, and to trigger any listener interested in the completion of those IO operations.

By default, the thread Quarkus REST will run endpoint methods on depends on the signature of the method. If a method returns one of the following types then it is considered non-blocking, and will be run on the IO thread by default:

  • io.smallrye.mutiny.Uni

  • io.smallrye.mutiny.Multi

  • java.util.concurrent.CompletionStage

  • org.reactivestreams.Publisher

  • Kotlin suspended methods

This 'best guess' approach means most operations will run on the correct thread by default. If you are writing reactive code, your method will generally return one of these types and will be executed on the IO thread. If you are writing blocking code, your methods will usually return the result directly, and these will be run on a worker thread.

You can override this behaviour using the @Blocking and @NonBlocking annotations. This can be applied at the method, class or jakarta.ws.rs.core.Application level.

The example below will override the default behaviour and always run on a worker thread, even though it returns a Uni.

package org.acme.rest;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

import io.smallrye.common.annotation.Blocking;

@Path("yawn")
public class Endpoint {

    @Blocking
    @GET
    public Uni<String> blockingHello() throws InterruptedException {
        // do a blocking operation
        Thread.sleep(1000);
        return Uni.createFrom().item("Yaaaawwwwnnnnnn…");
    }
}

Most of the time, there are ways to achieve the same blocking operations in an asynchronous/reactive way, using Mutiny, Hibernate Reactive or any of the Quarkus Reactive extensions for example:

package org.acme.rest;

import java.time.Duration;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;

import io.smallrye.mutiny.Uni;

@Path("yawn")
public class Endpoint {

    @GET
    public Uni<String> blockingHello() throws InterruptedException {
        return Uni.createFrom().item("Yaaaawwwwnnnnnn…")
                // do a non-blocking sleep
                .onItem().delayIt().by(Duration.ofSeconds(2));
    }
}

If a method or class is annotated with jakarta.transaction.Transactional then it will also be treated as a blocking method. This is because JTA is a blocking technology, and is generally used with other blocking technology such as Hibernate and JDBC. An explicit @Blocking or @NonBlocking on the class will override this behaviour.

Overriding the default behaviour

If you want to override the default behavior, you can annotate a jakarta.ws.rs.core.Application subclass in your application with @Blocking or @NonBlocking, and this will set the default for every method that does not have an explicit annotation.

Behavior can still be overridden on a class or method level by annotating them directly, however, all endpoints without an annotation will now follow the default, no matter their method signature.

Exception mapping

If your application needs to return non-nominal HTTP codes in error cases, the best is to throw exceptions that will result in the proper HTTP response being sent by the framework using WebApplicationException or any of its subtypes:

package org.acme.rest;

import jakarta.ws.rs.BadRequestException;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.NotFoundException;
import jakarta.ws.rs.Path;

@Path("cheeses/{cheese}")
public class Endpoint {

    @GET
    public String findCheese(String cheese) {
        if(cheese == null)
            // send a 400
            throw new BadRequestException();
        if(!cheese.equals("camembert"))
            // send a 404
            throw new NotFoundException("Unknown cheese: " + cheese);
        return "Camembert is a very nice cheese";
    }
}

You can change the log level of the thrown WebApplicationException exceptions by configuring the following property quarkus.log.category."WebApplicationException".level like so:

quarkus.log.category."WebApplicationException".level=DEBUG

If your endpoint method is delegating calls to another service layer which does not know of Jakarta REST, you need a way to turn service exceptions to an HTTP response, and you can do that using the @ServerExceptionMapper annotation on a method, with one parameter of the exception type you want to handle, and turning that exception into a RestResponse (or a Uni<RestResponse<?>>):

package org.acme.rest;

import java.util.Map;

import jakarta.enterprise.context.ApplicationScoped;
import jakarta.inject.Inject;
import jakarta.ws.rs.BadRequestException;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.core.Response;

import org.jboss.resteasy.reactive.server.ServerExceptionMapper;
import org.jboss.resteasy.reactive.RestResponse;

class UnknownCheeseException extends RuntimeException {
    public final String name;

    public UnknownCheeseException(String name) {
        this.name = name;
    }
}

@ApplicationScoped
class CheeseService {
    private static final Map<String, String> cheeses =
            Map.of("camembert", "Camembert is a very nice cheese",
                   "gouda", "Gouda is acceptable too, especially with cumin");

    public String findCheese(String name) {
        String ret = cheeses.get(name);
        if(ret != null)
            return ret;
        throw new UnknownCheeseException(name);
    }
}

@Path("cheeses/{cheese}")
public class Endpoint {

    @Inject
    CheeseService cheeses;

    @ServerExceptionMapper
    public RestResponse<String> mapException(UnknownCheeseException x) {
        return RestResponse.status(Response.Status.NOT_FOUND, "Unknown cheese: " + x.name);
    }

    @GET
    public String findCheese(String cheese) {
        if(cheese == null)
            // send a 400
            throw new BadRequestException();
        return cheeses.findCheese(cheese);
    }
}

By default, methods annotated with @ServerExceptionMapper do not run CDI interceptors that apply to the other methods of the class (like ones needed for implementing security method level security).

Users however can opt into interceptors by adding the corresponding annotations to the method.

When mapping an exception to a @ServerExceptionMapper method, the cause of the exception normally does not come into play.

However, some exception types in Java only serve as wrappers for other exceptions. Often, checked exceptions are wrapped into RuntimeException just to not have them declared in method throws parameters. Working with CompletionStage for example, will require CompletionException. There are many such exception types that are just wrappers around the real cause of the exception.

If you wish to make sure your exception mapper is called for your exception type even when it is wrapped by one of those wrapper exceptions, you can use @UnwrapException on the exception wrapper type:

public class MyExceptionWrapper extends RuntimeException {
    public MyExceptionWrapper(Exception cause) {
        super(cause);
    }
}

If you don’t control that exception wrapper type, you can place the annotation on any class and specify the exception wrapper types it applies to as annotation parameter:

@UnwrapException({CompletionException.class, RuntimeException.class})
public class Mapper {

    @ServerExceptionMapper
    public Response handleMyException(MyException x) {
        // ...
    }

}

Εxception mappers defined in REST endpoint classes will only be called if the exception is thrown in the same class. If you want to define global exception mappers, simply define them outside a REST endpoint class:

package org.acme.rest;

import org.jboss.resteasy.reactive.server.ServerExceptionMapper;
import org.jboss.resteasy.reactive.RestResponse;

class ExceptionMappers {
    @ServerExceptionMapper
    public RestResponse<String> mapException(UnknownCheeseException x) {
        return RestResponse.status(Response.Status.NOT_FOUND, "Unknown cheese: " + x.name);
    }
}

Your exception mapper may declare any of the following parameter types:

Table 6. Exception mapper parameters
Tipo Uso

An exception type

Defines the exception type you want to handle

Any of the Context objects

ContainerRequestContext

A context object to access the current request

It may declare any of the following return types:

Table 7. Exception mapper return types
Tipo Uso

RestResponse or Response

The response to send to the client when the exception occurs

Uni<RestResponse> or Uni<Response>

An asynchronous response to send to the client when the exception occurs

When an exception occurs, Quarkus REST does not log it by default (for security reasons). This can sometimes make it hard to understand why certain exception handling code was invoked (or not invoked). To make Quarkus REST log the actual exception before an exception mapping code is run the org.jboss.resteasy.reactive.common.core.AbstractResteasyReactiveContext log category can be set to DEBUG like so:

quarkus.log.category."org.jboss.resteasy.reactive.common.core.AbstractResteasyReactiveContext".level=DEBUG

Request or response filters

Via annotations

You can declare functions that are invoked in the following phases of the request processing:

  • Before the endpoint method is identified: pre-matching request filter

  • After routing, but before the endpoint method is called: normal request filter

  • After the endpoint method is called: response filter

These filters allow you to do various things such as examine the request URI, HTTP method, influence routing, look or change request headers, abort the request, or modify the response.

Request filters can be declared with the @ServerRequestFilter annotation:

import java.util.Optional;

class Filters {

    @ServerRequestFilter(preMatching = true)
    public void preMatchingFilter(ContainerRequestContext requestContext) {
        // make sure we don't lose cheese lovers
        if("yes".equals(requestContext.getHeaderString("Cheese"))) {
            requestContext.setRequestUri(URI.create("/cheese"));
        }
    }

    @ServerRequestFilter
    public Optional<RestResponse<Void>> getFilter(ContainerRequestContext ctx) {
        // only allow GET methods for now
        if(!ctx.getMethod().equals(HttpMethod.GET)) {
            return Optional.of(RestResponse.status(Response.Status.METHOD_NOT_ALLOWED));
        }
        return Optional.empty();
    }
}

Request filters are usually executed on the same thread that the method that handles the request will be executed. That means that if the method servicing the request is annotated with @Blocking, then the filters will also be run on the worker thread. If the method is annotated with @NonBlocking (or is not annotated at all), then the filters will also be run on the same event-loop thread.

If however a filter needs to be run on the event-loop despite the fact that the method servicing the request will be run on a worker thread, then @ServerRequestFilter(nonBlocking=true) can be used. Note however, that these filters need to be run before any filter that does not use that setting and would run on a worker thread.

Keep in mind however that the information above does not apply to pre-matching filters (@ServerRequestFilter(preMatching = true)). These filters are always run on an event-loop thread.

Similarly, response filters can be declared with the @ServerResponseFilter annotation:

class Filters {
    @ServerResponseFilter
    public void getFilter(ContainerResponseContext responseContext) {
        Object entity = responseContext.getEntity();
        if(entity instanceof String) {
            // make it shout
            responseContext.setEntity(((String)entity).toUpperCase());
        }
    }
}

Such a response filter will also be called for handled exceptions.

Your filters may declare any of the following parameter types:

Table 8. Filter parameters
Tipo Uso

Any of the Context objects

ContainerRequestContext

A context object to access the current request

ContainerResponseContext

A context object to access the current response

Throwable

Any thrown and handled exception, or null (only for response filters).

It may declare any of the following return types:

Table 9. Filter return types
Tipo Uso

RestResponse<?> or Response

The response to send to the client instead of continuing the filter chain, or null if the filter chain should proceed

Optional<RestResponse<?>> or Optional<Response>

An optional response to send to the client instead of continuing the filter chain, or an empty value if the filter chain should proceed

Uni<RestResponse<?>> or Uni<Response>

An asynchronous response to send to the client instead of continuing the filter chain, or null if the filter chain should proceed

You can restrict the Resource methods for which a filter runs, by using @NameBinding meta-annotations.

The Jakarta REST way

Both HTTP request and response can be intercepted by providing ContainerRequestFilter or ContainerResponseFilter implementations respectively. These filters are suitable for processing the metadata associated with a message: HTTP headers, query parameters, media type, and other metadata. They also have the capability to abort the request processing, for instance when the user does not have the permissions to access the endpoint.

Let’s use ContainerRequestFilter to add logging capability to our service. We can do that by implementing ContainerRequestFilter and annotating it with the @Provider annotation:

package org.acme.rest.json;

import io.vertx.core.http.HttpServerRequest;
import org.jboss.logging.Logger;

import jakarta.ws.rs.container.ContainerRequestContext;
import jakarta.ws.rs.container.ContainerRequestFilter;
import jakarta.ws.rs.core.Context;
import jakarta.ws.rs.core.UriInfo;
import jakarta.ws.rs.ext.Provider;

@Provider
public class LoggingFilter implements ContainerRequestFilter {

    private static final Logger LOG = Logger.getLogger(LoggingFilter.class);

    @Context
    UriInfo info;

    @Context
    HttpServerRequest request;

    @Override
    public void filter(ContainerRequestContext context) {

        final String method = context.getMethod();
        final String path = info.getPath();
        final String address = request.remoteAddress().toString();

        LOG.infof("Request %s %s from IP %s", method, path, address);
    }
}

Now, whenever a REST method is invoked, the request will be logged into the console:

2019-06-05 12:44:26,526 INFO  [org.acm.res.jso.LoggingFilter] (executor-thread-1) Request GET /legumes from IP 127.0.0.1
2019-06-05 12:49:19,623 INFO  [org.acm.res.jso.LoggingFilter] (executor-thread-1) Request GET /fruits from IP 0:0:0:0:0:0:0:1
2019-06-05 12:50:44,019 INFO  [org.acm.res.jso.LoggingFilter] (executor-thread-1) Request POST /fruits from IP 0:0:0:0:0:0:0:1
2019-06-05 12:51:04,485 INFO  [org.acm.res.jso.LoggingFilter] (executor-thread-1) Request GET /fruits from IP 127.0.0.1

A ContainerResponseFilter will also be called for handled exceptions.

Readers and Writers: mapping entities and HTTP bodies

Whenever your endpoint methods return an object (or when they return a RestResponse<?> or Response with an entity), Quarkus REST will look for a way to map that into an HTTP response body.

Similarly, whenever your endpoint method takes an object as parameter, we will look for a way to map the HTTP request body into that object.

This is done via a pluggable system of MessageBodyReader and MessageBodyWriter interfaces, which are responsible for defining which Java type they map from/to, for which media types, and how they turn HTTP bodies to/from Java instances of that type.

For example, if we have our own Cheese type on our endpoint:

package org.acme.rest;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.PUT;
import jakarta.ws.rs.Path;

class Cheese {
    public String name;

    public Cheese(String name) {
        this.name = name;
    }
}

@Path("cheese")
public class Endpoint {

    @GET
    public Cheese sayCheese() {
        return new Cheese("Cheeeeeese");
    }

    @PUT
    public void addCheese(Cheese cheese) {
        System.err.println("Received a new cheese: " + cheese.name);
    }
}

Then we can define how to read and write it with our body reader/writers, annotated with @Provider:

package org.acme.rest;

import java.io.IOException;
import java.io.InputStream;
import java.io.OutputStream;
import java.lang.annotation.Annotation;
import java.lang.reflect.Type;
import java.nio.charset.StandardCharsets;

import jakarta.ws.rs.WebApplicationException;
import jakarta.ws.rs.core.MediaType;
import jakarta.ws.rs.core.MultivaluedMap;
import jakarta.ws.rs.ext.MessageBodyReader;
import jakarta.ws.rs.ext.MessageBodyWriter;
import jakarta.ws.rs.ext.Provider;

@Provider
public class CheeseBodyHandler implements MessageBodyReader<Cheese>,
                                           MessageBodyWriter<Cheese> {

    @Override
    public boolean isWriteable(Class<?> type, Type genericType,
                               Annotation[] annotations, MediaType mediaType) {
        return type == Cheese.class;
    }

    @Override
    public void writeTo(Cheese t, Class<?> type, Type genericType,
                        Annotation[] annotations, MediaType mediaType,
                        MultivaluedMap<String, Object> httpHeaders,
                        OutputStream entityStream)
            throws IOException, WebApplicationException {
        entityStream.write(("[CheeseV1]" + t.name)
                           .getBytes(StandardCharsets.UTF_8));
    }

    @Override
    public boolean isReadable(Class<?> type, Type genericType,
                              Annotation[] annotations, MediaType mediaType) {
        return type == Cheese.class;
    }

    @Override
    public Cheese readFrom(Class<Cheese> type, Type genericType,
                            Annotation[] annotations, MediaType mediaType,
                            MultivaluedMap<String, String> httpHeaders,
                            InputStream entityStream)
            throws IOException, WebApplicationException {
        String body = new String(entityStream.readAllBytes(), StandardCharsets.UTF_8);
        if(body.startsWith("[CheeseV1]"))
            return new Cheese(body.substring(11));
        throw new IOException("Invalid cheese: " + body);
    }

}

If you want to get the most performance out of your writer, you can extend the ServerMessageBodyWriter instead of MessageBodyWriter where you will be able to use less reflection and bypass the blocking IO layer:

package org.acme.rest;

import java.io.IOException;
import java.io.InputStream;
import java.io.OutputStream;
import java.lang.annotation.Annotation;
import java.lang.reflect.Type;
import java.nio.charset.StandardCharsets;

import jakarta.ws.rs.WebApplicationException;
import jakarta.ws.rs.core.MediaType;
import jakarta.ws.rs.core.MultivaluedMap;
import jakarta.ws.rs.ext.MessageBodyReader;
import jakarta.ws.rs.ext.Provider;

import org.jboss.resteasy.reactive.server.spi.ResteasyReactiveResourceInfo;
import org.jboss.resteasy.reactive.server.spi.ServerMessageBodyWriter;
import org.jboss.resteasy.reactive.server.spi.ServerRequestContext;

@Provider
public class CheeseBodyHandler implements MessageBodyReader<Cheese>,
                                           ServerMessageBodyWriter<Cheese> {

    // …

    @Override
    public boolean isWriteable(Class<?> type, ResteasyReactiveResourceInfo target,
                               MediaType mediaType) {
        return type == Cheese.class;
    }

    @Override
    public void writeResponse(Cheese t, ServerRequestContext context)
      throws WebApplicationException, IOException {
        context.serverResponse().end("[CheeseV1]" + t.name);
    }
}
You can restrict which content-types your reader/writer apply to by adding Consumes/Produces annotations on your provider class.

Reader and Writer interceptors

Just as you can intercept requests and responses, you can also intercept readers and writers, by extending the ReaderInterceptor or WriterInterceptor on a class annotated with @Provider.

If we look at this endpoint:

package org.acme.rest;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.PUT;
import jakarta.ws.rs.Path;

@Path("cheese")
public class Endpoint {

    @GET
    public String sayCheese() {
        return "Cheeeeeese";
    }

    @PUT
    public void addCheese(String cheese) {
        System.err.println("Received a new cheese: " + cheese);
    }
}

We can add reader and writer interceptors like this:

package org.acme.rest;

import java.io.IOException;

import jakarta.ws.rs.WebApplicationException;
import jakarta.ws.rs.ext.Provider;
import jakarta.ws.rs.ext.ReaderInterceptor;
import jakarta.ws.rs.ext.ReaderInterceptorContext;
import jakarta.ws.rs.ext.WriterInterceptor;
import jakarta.ws.rs.ext.WriterInterceptorContext;

@Provider
public class CheeseIOInterceptor implements ReaderInterceptor, WriterInterceptor {

    @Override
    public void aroundWriteTo(WriterInterceptorContext context)
      throws IOException, WebApplicationException {
        System.err.println("Before writing " + context.getEntity());
        context.proceed();
        System.err.println("After writing " + context.getEntity());
    }

    @Override
    public Object aroundReadFrom(ReaderInterceptorContext context)
      throws IOException, WebApplicationException {
        System.err.println("Before reading " + context.getGenericType());
        Object entity = context.proceed();
        System.err.println("After reading " + entity);
        return entity;
    }
}

Quarkus REST and REST Client interactions

In Quarkus, the Quarkus REST extension and the REST Client extension share the same infrastructure. One important consequence of this consideration is that they share the same list of providers (in the Jakarta REST meaning of the word).

For instance, if you declare a WriterInterceptor, it will by default intercept both the servers calls and the client calls, which might not be the desired behavior.

However, you can change this default behavior and constrain a provider to:

  • only consider server calls by adding the @ConstrainedTo(RuntimeType.SERVER) annotation to your provider;

  • only consider client calls by adding the @ConstrainedTo(RuntimeType.CLIENT) annotation to your provider.

Parameter mapping

All Request Parameters can be declared as String, but also any of the following types:

  • Types for which a ParamConverter is available via a registered ParamConverterProvider.

  • Primitive types.

  • Types that have a constructor that accepts a single String argument.

  • Types that have a static method named valueOf or fromString with a single String argument that return an instance of the type. If both methods are present then valueOf will be used unless the type is an enum in which case fromString will be used.

  • List<T>, Set<T>, or SortedSet<T>, where T satisfies any above criterion.

The following example illustrates all those possibilities:

package org.acme.rest;

import java.lang.annotation.Annotation;
import java.lang.reflect.Type;
import java.util.List;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.ext.ParamConverter;
import jakarta.ws.rs.ext.ParamConverterProvider;
import jakarta.ws.rs.ext.Provider;

import org.jboss.resteasy.reactive.RestQuery;

@Provider
class MyConverterProvider implements ParamConverterProvider {

    @Override
    public <T> ParamConverter<T> getConverter(Class<T> rawType, Type genericType,
                                              Annotation[] annotations) {
        // declare a converter for this type
        if(rawType == Converter.class) {
            return (ParamConverter<T>) new MyConverter();
        }
        return null;
    }

}

// this is my custom converter
class MyConverter implements ParamConverter<Converter> {

    @Override
    public Converter fromString(String value) {
        return new Converter(value);
    }

    @Override
    public String toString(Converter value) {
        return value.value;
    }

}

// this uses a converter
class Converter {
    String value;
    Converter(String value) {
        this.value = value;
    }
}

class Constructor {
    String value;
    // this will use the constructor
    public Constructor(String value) {
        this.value = value;
    }
}

class ValueOf {
    String value;
    private ValueOf(String value) {
        this.value = value;
    }
    // this will use the valueOf method
    public static ValueOf valueOf(String value) {
        return new ValueOf(value);
    }
}

@Path("hello")
public class Endpoint {

    @Path("{converter}/{constructor}/{primitive}/{valueOf}")
    @GET
    public String conversions(Converter converter, Constructor constructor,
                              int primitive, ValueOf valueOf,
                              @RestQuery List<Constructor> list) {
        return converter + "/" + constructor + "/" + primitive
               + "/" + valueOf + "/" + list;
    }
}

Separating Query parameter values

Normally a collection of String values is used to capture the values used in multiple occurrences of the same query parameter. For example, for the following resource method:

@Path("hello")
public static class HelloResource {

    @GET
    public String hello(@RestQuery("name") List<String> names) {
        if (names.isEmpty()) {
            return "hello world";
        } else {
            return "hello " + String.join(" ", names);
        }
    }
}

and the following request:

GET /hello?name=foo&name=bar HTTP/1.1

the names variable will contain both foo and bar and the response will be hello foo bar.

It is not uncommon however to need to convert a single query parameter into a collection of values based on some delimiting character. That is where the @org.jboss.resteasy.reactive.Separator annotation comes into play.

If we update the resource method to:

@Path("hello")
public static class HelloResource {

    @GET
    public String hello(@RestQuery("name") @Separator(",") List<String> names) {
        if (names.isEmpty()) {
            return "hello world";
        } else {
            return "hello " + String.join(" ", names);
        }
    }
}

and use the following request:

GET /hello?name=foo,bar HTTP/1.1

then the response will be hello foo bar.

Handling dates

Quarkus REST supports the use of the implementations of java.time.Temporal (like java.time.LocalDateTime) as query, path, or form params. Furthermore, it provides the @org.jboss.resteasy.reactive.DateFormat annotation, which can be used to set a custom expected pattern. Otherwise, the JDK’s default format for each type is used implicitly.

Preconditions

HTTP allows requests to be conditional, based on a number of conditions, such as:

  • Date of last resource modification

  • A resource tag, similar to a hash code of the resource to designate its state or version

Let’s see how you can do conditional request validation using the Request context object:

package org.acme.rest;

import java.time.Instant;
import java.time.temporal.ChronoUnit;
import java.time.temporal.TemporalUnit;
import java.util.Date;

import jakarta.ws.rs.GET;
import jakarta.ws.rs.PUT;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.core.EntityTag;
import jakarta.ws.rs.core.Request;
import jakarta.ws.rs.core.Response;
import jakarta.ws.rs.core.Response.ResponseBuilder;

@Path("conditional")
public class Endpoint {

    // It's important to keep our date on seconds because that's how it's sent to the
    // user in the Last-Modified header
    private Date date = Date.from(Instant.now().truncatedTo(ChronoUnit.SECONDS));
    private int version = 1;
    private EntityTag tag = new EntityTag("v1");
    private String resource = "Some resource";

    @GET
    public Response get(Request request) {
        // first evaluate preconditions
        ResponseBuilder conditionalResponse = request.evaluatePreconditions(date, tag);
        if(conditionalResponse != null)
            return conditionalResponse.build();
        // preconditions are OK
        return Response.ok(resource)
                .lastModified(date)
                .tag(tag)
                .build();
    }

    @PUT
    public Response put(Request request, String body) {
        // first evaluate preconditions
        ResponseBuilder conditionalResponse = request.evaluatePreconditions(date, tag);
        if(conditionalResponse != null)
            return conditionalResponse.build();
        // preconditions are OK, we can update our resource
        resource = body;
        date = Date.from(Instant.now().truncatedTo(ChronoUnit.SECONDS));
        version++;
        tag = new EntityTag("v" + version);
        return Response.ok(resource)
                .lastModified(date)
                .tag(tag)
                .build();
    }
}

When we call GET /conditional the first time, we will get this response:

HTTP/1.1 200 OK
Content-Type: text/plain;charset=UTF-8
ETag: "v1"
Last-Modified: Wed, 09 Dec 2020 16:10:19 GMT
Content-Length: 13

Some resource

So now if we want to check if we need to fetch a new version, we can make the following request:

GET /conditional HTTP/1.1
Host: localhost:8080
If-Modified-Since: Wed, 09 Dec 2020 16:10:19 GMT

And we would get the following response:

HTTP/1.1 304 Not Modified

Because the resource has not been modified since that date, this saves on sending the resource but can also help your users detect the concurrent modification. For example, one client wants to update the resource, but another user has modified it since. You can follow the previous GET request with this update:

PUT /conditional HTTP/1.1
Host: localhost:8080
If-Unmodified-Since: Wed, 09 Dec 2020 16:25:43 GMT
If-Match: v1
Content-Length: 8
Content-Type: text/plain

newstuff

And if some other user has modified the resource between your GET and your PUT you would get this answer back:

HTTP/1.1 412 Precondition Failed
ETag: "v2"
Content-Length: 0

Negotiation

One of the main ideas of REST (and HTTP) is that your resource is independent of its representation, and that both the client and server are free to represent their resources in as many media types as they want. This allows the server to declare support for multiple representations and let the client declare which ones it supports and get served something appropriate.

The following endpoint supports serving cheese in plain text or JSON:

package org.acme.rest;

import jakarta.ws.rs.Consumes;
import jakarta.ws.rs.GET;
import jakarta.ws.rs.PUT;
import jakarta.ws.rs.Path;
import jakarta.ws.rs.Produces;
import jakarta.ws.rs.core.MediaType;

import com.fasterxml.jackson.annotation.JsonCreator;

class Cheese {
    public String name;
    @JsonCreator
    public Cheese(String name) {
        this.name = name;
    }
    @Override
    public String toString() {
        return "Cheese: " + name;
    }
}

@Path("negotiated")
public class Endpoint {

    @Produces({MediaType.APPLICATION_JSON, MediaType.TEXT_PLAIN})
    @GET
    public Cheese get() {
        return new Cheese("Morbier");
    }

    @Consumes(MediaType.TEXT_PLAIN)
    @PUT
    public Cheese putString(String cheese) {
        return new Cheese(cheese);
    }

    @Consumes(MediaType.APPLICATION_JSON)
    @PUT
    public Cheese putJson(Cheese cheese) {
        return cheese;
    }
}

The user will be able to select which representation it gets with the Accept header, in the case of JSON:

> GET /negotiated HTTP/1.1
> Host: localhost:8080
> Accept: application/json

< HTTP/1.1 200 OK
< Content-Type: application/json
< Content-Length: 18
<
< {"name":"Morbier"}

And for text:

> GET /negotiated HTTP/1.1
> Host: localhost:8080
> Accept: text/plain
>
< HTTP/1.1 200 OK
< Content-Type: text/plain
< Content-Length: 15
<
< Cheese: Morbier

Similarly, you can PUT two different representations. JSON:

> PUT /negotiated HTTP/1.1
> Host: localhost:8080
> Content-Type: application/json
> Content-Length: 16
>
> {"name": "brie"}

< HTTP/1.1 200 OK
< Content-Type: application/json;charset=UTF-8
< Content-Length: 15
<
< {"name":"brie"}

Or plain text:

> PUT /negotiated HTTP/1.1
> Host: localhost:8080
> Content-Type: text/plain
> Content-Length: 9
>
> roquefort

< HTTP/1.1 200 OK
< Content-Type: application/json;charset=UTF-8
< Content-Length: 20
<
< {"name":"roquefort"}

HTTP Compression

The body of an HTTP response is not compressed by default. You can enable the HTTP compression support by means of quarkus.http.enable-compression=true.

If compression support is enabled then the response body is compressed if:

  • the resource method is annotated with @io.quarkus.vertx.http.Compressed, or

  • the Content-Type header is set and the value is a compressed media type as configured via quarkus.http.compress-media-types.

The response body is never compressed if:

  • the resource method is annotated with @io.quarkus.vertx.http.Uncompressed, or

  • the Content-Type header is not set.

By default, the following list of media types is compressed: text/html, text/plain, text/xml, text/css, text/javascript, application/javascript, application/json, application/graphql+json and application/xhtml+xml.
If the client does not support HTTP compression then the response body is not compressed.

Include/Exclude Jakarta REST classes

Using Build time conditions

Quarkus enables the inclusion or exclusion of Jakarta REST Resources, Providers and Features directly thanks to build time conditions in the same that it does for CDI beans. Thus, the various Jakarta REST classes can be annotated with profile conditions (@io.quarkus.arc.profile.IfBuildProfile or @io.quarkus.arc.profile.UnlessBuildProfile) and/or with property conditions (io.quarkus.arc.properties.IfBuildProperty or io.quarkus.arc.properties.UnlessBuildProperty) to indicate to Quarkus at build time under which conditions these Jakarta REST classes should be included.

In the following example, Quarkus includes the ResourceForApp1Only Resource class if and only if the build profile app1 has been enabled.

@IfBuildProfile("app1")
public class ResourceForApp1Only {

    @GET
    @Path("sayHello")
    public String sayHello() {
        return "hello";
     }
}

Please note that if a Jakarta REST Application has been detected and the method getClasses() and/or getSingletons() has/have been overridden, Quarkus will ignore the build time conditions and consider only what has been defined in the Jakarta REST Application.

Using a runtime property

Quarkus can also conditionally disable Jakarta REST Resources based on the value of runtime properties using the @io.quarkus.resteasy.reactive.server.EndpointDisabled annotation.

In the following example, Quarkus will exclude RuntimeResource at runtime if the application has some.property configured to "disable".

@EndpointDisabled(name = "some.property", stringValue = "disable")
public class RuntimeResource {

    @GET
    @Path("sayHello")
    public String sayHello() {
        return "hello";
     }
}

This feature does not work when using native build.

REST Client

In addition to the Server side, Quarkus REST comes with a new MicroProfile REST Client implementation that is non-blocking at its core.

Please note that the quarkus-resteasy-client extension may not be used with Quarkus REST, use quarkus-rest-client instead.

See the REST Client Guide for more information about the REST client.


1. Unless it is a URI template parameter or a context object.

Related content