Skip to the content.

karate-grpc

Build Status Maven Central

simple gRPC testing with karate and a dynamic client using polyglot.

karate-grpc makes it easy to:

Hello World

karate-grpc-hello-world

Testing hello world

Requires maven to be installed

$ # compile and test the whole project
$ mvn clean install

$ # test demo
$ cd karate-grpc-demo
$ mvn test
$ # or run single test
$ mvn test -Dtest=HelloWorldNewRunner

When running tests, the hello world grpc server is started/stopped automatically in AbstractTestBase.java.

Karate also generates beautiful test reports:

karate-grpc-hello-world-report

Real World Examples

A set of real-life examples which includes single rpc, client stream rpc, server stream rpc and bidi stream rpc can be found here: karate-grpc-demo

Getting Started

karate-grpc requires Java 8 and then Maven to be installed, these also are required by karate and polyglot.

karate-grpc only support Maven currently.

Maven

You need to add the following <dependencies>:

<dependency>
    <groupId>com.github.thinkerou</groupId>
    <artifactId>karate-grpc-core</artifactId>
    <version>1.0.7</version>
</dependency>

Gradle

TODO: need to test!!!

Alternatively for Gradle you need to add the following entry:

testImplementation 'com.github.thinkerou:karate-grpc-core:1.0.7'

And simulates karate-grpc-helper and karate-grpc-demo build your redis helper project and test project.

What to need for testing grpc server

Testing one grpc server, we have the follow info:

So, we could test it based on the two point.

⚠️ Using karate-grpc we can perfect to solve it!

What to need in pom file

For testing your grpc server, as above, need protobuf jar dependency and protobuf build plugins - protobuf-maven-plugin.

MUST appoint descriptorSetFileName and protoSourceRoot params:

<descriptorSetFileName>karate-grpc.protobin</descriptorSetFileName>
<protoSourceRoot>${project.build.directory}/dependency/demo</protoSourceRoot>

Especially, descriptorSetFileName MUST equal karate-grpc.protobin, please see here about more details.

And other pom settings are the same as karate.

How to write karate feature

We need to use Java interop of Karate in order to call us define grpc client.

And use JSON.parse javascript function parse the response of grpc server return value.

So, use karate-grpc need the following steps:

  1. Calls into karate-grpc GrpcClient via Java Interop.
* def GrpcClient = Java.type('com.github.thinkerou.karate.GrpcClient')
  1. Builds one public Grpc client using your grpc ip and port.
* def client = new GrpcClient('localhost', 50051)

If you want to list protobuf by service name or/and message name, you should use:

* def client = new GrpcClient()

Because you don’t need grpc server ip/port when listing protobuf.

  1. Reads JSON data corresponding your protobuf definition.

  2. Calls your Grpc server using call of karate-grpc.

* def response = client.call('helloworld.Greeter/SayHello', payload, karate)

call has two required params, and one optional

  1. protobuf full name(format:<package-name>.<service-name>/<rpc-name>)
  2. JSON data.
  3. karate (optional, nullable) – if present, will add [request] and [response] to html report. karate variable is of type ScenarioBridge and is automatically created when running karate tests.

request-response-report

If you input protobuf full name error, call will fail and output protobuf message by list, like this:

When input helloworld.Greeter/SayHello1, it will fail and print log:

Oct 11, 2018 6:53:24 PM com.github.thinkerou.karate.service.GrpcCall invoke
警告: Call grpc failed, maybe you should see the follow grpc information.
Oct 11, 2018 6:53:24 PM com.github.thinkerou.karate.service.GrpcCall invoke
信息: [
  {
    "helloworld.Greeter/SayHelloBiStreaming":"",
    "helloworld.Greeter/RecordRoute":"",
    "helloworld.Greeter/RouteChat":"",
    "helloworld.Greeter/SayHelloServerStreaming":"",
    "helloworld.Greeter/ListFeatures":"",
    "helloworld.Greeter/SayHello":"",
    "helloworld.Greeter/AgainSayHello":"",
    "helloworld.Greeter/SayHelloClientStreaming":"",
    "helloworld.Greeter/GetFeature":""
  }
]

  1. Converts response string to JSON.
* def response = JSON.parse(response)

Because call of karate-grpc returns JSON string, we need to convert it and then can use match assertion.

  1. Asserts payload.

  2. (Optional) Saves response for second Grpc.

If have second Grpc use the response of first Grpc, we should save it, like:

* def message = response[0].message

And use it on JSON file:

[
  {
    "message": "#(message)",
    "address": "BeiJing"
  }
]
  1. (Optional) Second Grpc call using response data before.

One whole example likes this:

Feature: grpc helloworld example by grpc dynamic client

  Background:
    * def client = Java.type('demo.DemoGrpcClientSingleton').INSTANCE.getGrpcClient();

  Scenario: do it
    * string payload = read('helloworld.json')
    * def response = client.call('helloworld.Greeter/SayHello', payload)
    * def response = JSON.parse(response)
    * print response
    * match response[0].message == 'Hello thinkerou'
    * def message = response[0].message

    * string payload = read('again-helloworld.json')
    * def response = client.call('helloworld.Greeter/AgainSayHello', payload)
    * def response = JSON.parse(response)
    * match response[0].details == 'Details Hello thinkerou in BeiJing'

How to write JSON file

Because karate-grpc supports stream grpc, we use list JSON.

Input JSON file like:

[
  {
    "name": "thinkerou"
  },
  {
    "name": "thinkerou2"
  }
]

Output JSON string also like:

[
  {
    "message": "Hello thinkerou"
  },
  {
    "message": "Hello thinkerou2"
  }
]

That’s all!!!

Redis

Why use Redis?

Using redis is optional, but caching descriptor sets may save compile time, especially when your project has many protobuf jar package dependencies.

Mock Redis

You can even use jedis-mock so you don’t even need to install Redis. see MockRedisHelperSingleton.java:

Redis performance

Note: while the redis test implementation is thread-safe, Redis uses single-threaded execution so test performance may be degraded for high concurrency.

To use redis, use class com.github.thinkerou.karate.RedisGrpcClient instead of com.github.thinkerou.karate.GrpClient

example:

public enum DemoGrpcClientSingleton {
    INSTANCE;

    RedisGrpcClient redisGrpcClient;

    public GrpcClient getGrpcClient() {
        return redisGrpcClient;
    }

    DemoGrpcClientSingleton() {
        redisGrpcClient = new RedisGrpcClient("localhost", 50051, MockRedisHelperSingleton.INSTANCE.getRedisHelper());
    }
}

TODO:

How to write grpc client

Note:

You only need two steps:

Like this:

public class Client {

  // ...

  public static String greet(String input) {
    HelloRequest.Builder requestBuilder = HelloRequest.newBuilder();
    try {
      JsonFormat.parser().merge(input, requestBuilder);
    } catch (ProtocolBufferException e) {
      // ...
    }

    HelloReply response = null;
    try {
      response = blockingStub.sayHello(requestBuilder.build());
    } catch (StatusRuntimeException e) {
      // ...
    }

    String res = "";
    try {
      res = JsonFormat.printer().print(response);
    } catch (ProtocolBufferException e) {
      // ...
    }

    return res;
  }

  // ...
}

Thanks

Thanks Peter Thomas for his work for karate and his generous help, also thanks Dino Wernli for his contributions for polyglot. And the favicon of organization generate at favicon.io.

Reference

Maybe you want to know more information about Karate or other, please read the follow contents:

License

karate-grpc is licensed under MIT License.