ZeroMock

Http Mock Server with (mostly) zero dependencies.

Right now the unique dependencies are minimal-json parser and jaxb, used for object serialization. But this project doesn't depends on any java servlet container or something like that. It uses the simple http server implementation that comes on every jdk.

Why?

Usually, the sorter answer should be 'because I can', but I want to explain a bit why I developed this library, and the principal reason is frustration.

• Frustration 1: Existing alternatives (like Wiremock, and others) are, in my opinion, too complex and dificult to read. In my daily work if I have to read an existing test that needs many stubs to work, is too difficult to understand the purpose of the test.

• Frustration 2: Why I need different libraries to implement a REST API and stubs for this API, in the end, you are doing the same, but instead to compute a result, you are returning a mock result. But finally, both are the same thing.

• Frustration 3: I have another project, called resp-server, is an implementation of REdis Serialization Protocol, and I thought that it would be nice to use this protocol to develop REST API applications. I tried to call spring-mvc controller using RESP protocol but the result was an ugly hack. I tried the same using spring-webflux, but the result was the same. So, why I cannot decouple HTTP protocol from the controller implementation?

So, these are the objetives of this library:

• provide an independent/decoupled/reusable implementation of REST API controllers.
• provide a fluent API to make easy to read and maintain applications/tests.
• keep the library small with the minimum of dependencies.

How?

First of all, I implemented the model, HttpRequest and HttpResponse, immutable objects, and then I implemented the functions to control the model transformations to obtain a response from a request. In the end the controller is a simple Function<HttpRequest, HttpResponse>.

Then, I created the server based in the HttpServer included inside the JRE. This server is under a com.sun package, so, it's not recomended to use, but for my test purposes it's more than enough.

Also I implemented a simple HttpClient using HttpURLConnection, initially only for test purposes.

And finally, I implemented a Junit5 extension for Junit integration.

Examples

This is the simplest example, a ping application. It uses the Junit5 extension. The test receives an instance of the server. Then you can create the stubs you need.

  @Test
  public void ping(MockHttpServer server) {
    server.when(Matchers.get("/ping")).then(Handlers.ok("pong"));
    
    HttpResponse response = HttpClient.connectTo(BASE_URL).request(Requests.get("/ping"));
    
    assertEquals("pong", Bytes.asString(response.body()));
  }

MockHttpServer is the principal class, it listen for requests, finds for a handler, and then executes it. If no handler is found, a NOT_FOUND(404) is returned.

The class Matchers contains predicates of type Predicate<HttpRequest>, with these predicates you can create the matcher function. These functions can be composes using the existing Predicate combinators like and, or and negate. In this case, it defines a predicate that matches a GET command with /ping path.

The class Handlers contains functions of type Function<HttpRequest, HttpResponse> to represent the response. Also you can combine different functions using Function combinators like andThen or compose. In this case, it defines a OK(200) response with pong content.

The class Bytes is the wrapper class I use for request and response bodies. It contains a ByteBuffer with the raw data. Also this class defines utillity methods to create and convert Bytes. In this case, it converts the response body to a String.

Another example of echo:

  @Test
  public void echoQueryParam(MockHttpServer server) {
    server.when(Matchers.get("/echo").and(Matchers.param("say")))
          .then(Handlers.ok(Extractors.queryParam("say").andThen(Serializers.plain())));
    
    HttpResponse response = HttpClient.connectTo(BASE_URL)
        .request(Requests.get("/echo").withParam("say", "Hello World!"));
    
    assertEquals("Hello World!", Bytes.asString(response.body()));
  }

The class Extractors contains functions of type Function<HttpRequest, T> to extract data from an HttpRequest. In this case extracts the content of the query param say.

The class Serializers contains functions of type Function<T, Bytes> to convert results to Bytes. In this case, it uses a plain text serializer.

The same example using path params:

  @Test
  public void echoPathParam(MockHttpServer server) {
    server.when(Matchers.get("/echo/:message")) 
          .then(Handlers.ok(Extractors.pathParam(1).andThen(Serializers.plain())));
    
    HttpResponse response = HttpClient.connectTo(BASE_URL).request(Requests.get("/echo/saysomething"));
    
    assertEquals("saysomething", Bytes.asString(response.body()));
  }

To declare a path param, you have to use the colon prefix : before the name of the param. To access the value you have to use the position of the parameter in the path.

And the final example, by now, is an implementation of the echo server using json serialization

  @Test
  public void pojoSerialization(MockHttpServer server) {
    server.when(Matchers.get("/echo").and(Matchers.param("say"))) 
          .then(Handlers.ok(Extractors.queryParam("say").andThen(Say::new).andThen(Serializers.json())));
    
    HttpResponse response = HttpClient.connectTo(BASE_URL)
        .request(Requests.get("/echo").withParam("say", "Hello World!"));
    
    assertEquals(new Say("Hello World!"), asObject(response.body()));
  }
  
  private Say asObject(Bytes body) {
    return Deserializers.json(Say.class).apply(body);
  }

In this example, a class Say is created with the say param content, and finally it converts the pojo using json.

In the other side, the client can use Deserializers class in order to create a Say class again. This class contains functions of type Function<Bytes, T> that converts Bytes to arbitrary objects.

Of course, you can use static imports for a clearer code if you want, I have added the full names only for explanatory purposes. All examples taken from this test class

One Line Server

It's pretty simple, using jbang

//usr/bin/env jbang "$0" "$@" ; exit $?

//DEPS com.github.tonivade:zeromock-server:0.18.0

import static com.github.tonivade.zeromock.server.MockHttpServer.listenAt;

public class OneLineApplication {

  public static void main(String[] args) {
    listenAt(8080)
        .get("/ping").ok("pong")
        .start();
  }
}

and with kscript:

#!/usr/bin/env kscript

//DEPS com.github.tonivade:zeromock-server:0.18.0

import com.github.tonivade.zeromock.server.MockHttpServer.listenAt

listenAt(8101)
  .get("/ping").ok("pong")
  .start()

and with scala-cli:

#!/usr/bin/env -S scala-cli shebang

//> using lib "com.github.tonivade:zeromock-server:0.18.0"

import com.github.tonivade.zeromock.server.MockHttpServer.listenAt

@main def server() =
  listenAt(8101)
    .get("/ping").ok("pong")
    .start()

License

This project is released under MIT License