Get ahead
VMware offers training and certification to turbo-charge your progress.
Learn moreIf you have an application that connects to an RSocket server at runtime, how do you test it? We need a way for a test to start a server and tell us where it is listening, and then we need to be able to register request and response examples (a.k.a. "contracts"). That’s what this project provides - it’s like Wiremock but for RSocket.
The easiest way to use the project is as a JUnit (Jupiter) extension, e.g:
@SpringBootTest
@ExtendWith(RSocketServerExtension.class)
class SocketsApplicationTests {
...
}
With this extension installed the Spring Boot tests will run with an RSocket server listening on a port given by test.rsocket.server.port
, so the test can connect directly to it, or (more likely) the code that it is testing will connect to it. You might need to tell it where to connect via the @SpringBootTest
annotation, e.g. if the application is looking for a property at runtime called rsocket.port
:
@SpringBootTest("rsocket.port=${test.rsocket.server.port}")
@ExtendWith(RSocketServerExtension.class)
class SocketsApplicationTests {
...
}
Test methods can inject an RSocketMessageCatalog
or RSocketMessageRegistry
and then use those to set up or inspect the state of the server. By default the server reads JSON contracts from the classpath at /catalog/*.json
, so you can set those up locally or in a test library that you share with the sarver. The structure of the JSON mirrors the MessageMapping
that is stored in the RSocketMessageCatalog
. Here’s an example (the request and response are just JSON objects):
{
"pattern": "events.response.*",
"frameType": "REQUEST_RESPONSE",
"request": {
"origin": "Client"
},
"response": {
"origin": "Server",
"interaction": "Response",
"index": 0
}
}
This mapping will match any REQUEST_RESPONSE
frame type on a route that matches the pattern, and which also has a request with a field "origin" equal to "Client". You can also match on a pattern in a field in the request by adding wildcards. Or you can leave the request out to match only on the route. If the frame type is REQUEST_RESPONSE
then the response is single valued. If the frametype is REQUEST_STREAM
you can provide a multi-valued "responses", for example:
{
"pattern": "my.stream.route",
"frameType": "REQUEST_STREAM",
"responses": [
{
"origin": "Server",
"interaction": "Stream",
"index": 0
},
{
"origin": "Server",
"interaction": "Stream",
"index": 1
},
{
"origin": "Server",
"interaction": "Stream",
"index": 2
}
]
}
In addition you can specify an integer field "repeat" to repeat the responses to form a longer stream. If the frame type is REQUEST_FNF
then there is no response, and the request will be ignored. And finally, if the frame type is REQUEST_CHANNEL
then the format of the mapping JSON is the same as the REQUEST_STREAM
, except that every time a message comes in from the input stream, the output stream is emitted again.
If you need more flexibility with the processing rules you can create your own MessageMapping
from the convenient static factory methods in the interface. You can supply a handler function (except for the fire and forget case), a pattern to match, and optionally a request to match. These methods can be used to dynamically register mappings to define the expected behaviour of the server at runtime.
You can inspect the state of the server by acquiring a MessageMapping
from the catalog, and then calling one of the drain()
methods to drain off the requests that have been received. For example:
@SpringBootTest
@ExtendWith(RSocketServerExtension.class)
class DynamicRouteTests {
private RSocketRequester rsocketRequester;
public DynamicRouteTests(@Autowired RSocketRequester.Builder rsocketRequesterBuilder,
@Value("${test.rsocket.server.port:7000}") int port) {
rsocketRequester = rsocketRequesterBuilder.tcp("localhost", port);
}
@Test
void response(RSocketMessageRegistry catalog) {
MessageMapping response = MessageMapping.response("response")
.response(new Foo("Server", "Response"));
catalog.register(response);
assertThat(rsocketRequester.route("response").data(new Foo("Client", "Request"))
.retrieveMono(Foo.class).doOnNext(foo -> {
System.err.println(foo);
assertThat(foo.getOrigin()).isEqualTo("Server");
}).block()).isNotNull();
assertThat(response.drain()).hasSize(1);
assertThat(response.drain()).hasSize(0);
}
}
The code here is still really a prototype, but it’s already potentially quite useful. So try it out and send feedback and maybe we can mature it to the point where we can release it.