tonic is a great crate to build GRPC applications. However, testing RPC built with tonic is not straightforward, especially for the streaming interface. If you have an RPC like this:
rpc Push(stream RequestPush) returns (stream ResponsePush);
Testing it usually involves lots of effort on properly mocking the data. This little crate helps to make it easier to mock the incoming data and to manipulate the response so that you could focus on testing the logic itself. For example:
#[tokio::test]
async fn service_push_works() -> anyhow::Result<()> {
let mut events: Vec<RequestPush> = Vec::with_capacity(3);
for i in 0..3 {
events.push(RequestPush::new(id: Bytes::from(i.to_string), data: Bytes::from("a".repeat(10))));
}
// preparing the streaming request
let req = tonic_mock::streaming_request(events);
let server = start_server();
// call the service
let res = server.push(req).await?;
// iterate the response and assert the result
tonic_mock::process_streaming_response(result, |msg, i| {
assert!(msg.is_ok());
assert_eq!(msg.as_ref().unwrap().code, i as i32);
})
.await;
Ok(())
}
Seven main functions provided:
streaming_request
: build streaming requests based on a vector of messages.streaming_request_with_interceptor
: build streaming requests with an interceptor function.request_with_interceptor
: create a standard (non-streaming) request with an interceptor.process_streaming_response
: iterate the streaming response and call the closure user provided.process_streaming_response_with_timeout
: iterate the streaming response with a timeout for each message.stream_to_vec
: iterate the streaming response and generate a vector for further processing.stream_to_vec_with_timeout
: iterate the streaming response with a timeout and generate a vector.
The crate provides a BidirectionalStreamingTest
utility for fine-grained testing of bidirectional streaming services:
use std::time::Duration;
use tokio::runtime::Runtime;
use tonic_mock::{BidirectionalStreamingTest, test_utils::TestRequest, test_utils::TestResponse};
// Create a runtime
let rt = Runtime::new().unwrap();
rt.block_on(async {
// Create a test context with your service function
let mut test = BidirectionalStreamingTest::new(my_bidirectional_service);
// Send messages one by one and check responses
test.send_client_message(TestRequest::new("msg1", "First message")).await;
// Get response
if let Some(response) = test.get_server_response().await {
// Verify response
assert_eq!(response.code, 200);
assert!(response.message.contains("First message"));
}
// You can also use timeouts
match test.get_server_response_with_timeout(Duration::from_millis(100)).await {
Ok(Some(resp)) => {
// Handle response
},
Ok(None) => {
// No more responses
},
Err(status) => {
// Timeout or other error
},
}
// Complete the test when done
test.complete().await;
});
Key features of BidirectionalStreamingTest
:
- Send messages one by one to the service
- Receive and test responses individually
- Support for timeouts when receiving responses
- More control over message flow compared to using a pre-collected vector
For a complete example, see examples/bidirectional_streaming_test_api.rs
.
The crate provides support for request interceptors, which allow you to modify requests before they are sent. This is useful for adding metadata, headers, or performing other customizations:
use tonic::metadata::MetadataValue;
use tonic_mock::streaming_request_with_interceptor;
// Create a streaming request with an interceptor that adds headers
let request = streaming_request_with_interceptor(messages, |req| {
// Add authentication header
req.metadata_mut().insert(
"authorization",
MetadataValue::from_static("Bearer token123")
);
// Add tracing header
req.metadata_mut().insert(
"x-request-id",
MetadataValue::from_static("trace-456")
);
});
// For non-streaming requests
let request = request_with_interceptor(message, |req| {
req.metadata_mut().insert(
"authorization",
MetadataValue::from_static("Bearer token123")
);
});
The crate provides functions for handling timeouts in streaming responses:
// Process streaming response with a 1-second timeout for each message
process_streaming_response_with_timeout(
response,
Duration::from_secs(1),
|msg, idx| {
if msg.is_ok() {
// Handle successful message
let response = msg.as_ref().unwrap();
// ...
} else {
// Handle error (could be timeout or other error)
let error = msg.as_ref().err().unwrap();
if error.code() == tonic::Code::DeadlineExceeded {
println!("Timeout occurred: {}", error.message());
}
}
}
).await;
// Convert stream to vector with timeout
let results = stream_to_vec_with_timeout(response, Duration::from_secs(1)).await;
for result in results {
// Check for timeout or other errors
if let Err(status) = &result {
if status.code() == tonic::Code::DeadlineExceeded {
println!("Timeout occurred: {}", status.message());
break;
}
}
}
The crate also provides optional test utilities to help with testing gRPC services. Enable these with the test-utils
feature (enabled by default):
[dependencies]
tonic-mock = "0.4" # Test utilities are included by default
# If you want to disable test utilities
tonic-mock = { version = "0.4", default-features = false }
Test utilities include:
TestRequest
andTestResponse
types: Simple message types for testingcreate_test_messages
: Generate test messages with sequential IDscreate_stream_response
: Create a streaming response from a vector of messagescreate_stream_response_with_errors
: Create a streaming response with errors at specified indicesassert_message_eq
: Assert that a message matches expected valuesassert_response_eq
: Assert that a response matches expected values
Example using test utilities:
use tonic_mock::test_utils::{TestRequest, TestResponse, create_test_messages, create_stream_response};
use tonic_mock::{streaming_request, process_streaming_response};
#[tokio::test]
async fn test_my_service() {
// Create test messages
let messages = create_test_messages(5);
// Create a streaming request
let request = streaming_request(messages);
// Call your service
let response = my_service.call(request).await.unwrap();
// Process the response
process_streaming_response(response, |msg: Result<TestResponse, Status>, idx| {
assert!(msg.is_ok());
// Assertions on the response
}).await;
}
For a more complete example, check the examples/grpc_test_demo.rs
file which demonstrates:
- Testing client streaming RPC (client sends multiple messages, server sends one response)
- Testing server streaming RPC (client sends one message, server sends multiple responses)
- Testing bidirectional streaming RPC (client and server both send multiple messages)
- Testing error handling in streaming RPCs
- Testing timeouts in streaming responses
- Using request interceptors to modify requests
Note these functions are for testing purpose only. DO NOT use them in other cases.
The crate provides utilities for low-level mocking of gRPC messages:
use tonic_mock::grpc_mock::{encode_grpc_request, decode_grpc_message, mock_grpc_call};
use tonic_mock::test_utils::{TestRequest, TestResponse};
use tonic::Status;
// Encode a gRPC request
let request = TestRequest::new("request-id", "test-data");
let encoded = encode_grpc_request(request);
// Decode a gRPC message
let decoded: TestRequest = decode_grpc_message(&encoded).unwrap();
assert_eq!(decoded.id, "request-id".as_bytes());
// Mock a gRPC call with a handler function
let response = mock_grpc_call(
"example.TestService",
"TestMethod",
request,
|req: TestRequest| {
// Process the request
let id_str = String::from_utf8_lossy(&req.id).to_string();
Ok(TestResponse::new(200, format!("Response for: {}", id_str)))
}
).unwrap();
These utilities give you fine-grained control over gRPC message encoding/decoding, which is useful for:
- Testing gRPC handlers directly without setting up a full service
- Implementing custom gRPC client/server logic
- Debugging gRPC message format issues
- Testing custom error handling in gRPC services
For a complete example, see examples/grpc_mock_example.rs
.
The crate provides a MockableGrpcClient
utility for mocking gRPC clients:
use tonic_mock::client_mock::{MockableGrpcClient, MockResponseDefinition, GrpcClientExt};
use tonic::{Request, Status, Code};
// Define a client type that will use the mock
#[derive(Debug, Clone)]
struct MyServiceClient<T> {
inner: T,
}
// Implement the GrpcClientExt trait for your client
impl GrpcClientExt<MyServiceClient<MockableGrpcClient>> for MyServiceClient<MockableGrpcClient> {
fn with_mock(mock: MockableGrpcClient) -> Self {
Self { inner: mock }
}
}
// Create a mock client and configure mock responses
#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
let mock = MockableGrpcClient::new();
// Configure mock responses
mock.mock::<MyRequestType, MyResponseType>("my.package.MyService", "MyMethod")
.respond_with(MockResponseDefinition::ok(MyResponseType { /* fields */ }))
.await
.respond_when(
|req| req.some_field == "special-value",
MockResponseDefinition::err(Status::new(Code::InvalidArgument, "Invalid value"))
)
.await;
// Create a client using the mock
let mut client = MyServiceClient::with_mock(mock.clone());
// Now your client will return the configured responses when called
let response = client.my_method(Request::new(my_request)).await?;
// Reset the mock to clear all handlers
mock.reset().await;
Ok(())
}
Here's a more complete example with proper error handling:
use tonic_mock::client_mock::{MockableGrpcClient, MockResponseDefinition, GrpcClientExt};
use tonic::{Request, Response, Status, Code};
use prost::Message;
use std::error::Error;
// Define message types
#[derive(Clone, PartialEq, Message)]
pub struct UserRequest {
#[prost(string, tag = "1")]
pub user_id: String,
}
#[derive(Clone, PartialEq, Message)]
pub struct UserResponse {
#[prost(string, tag = "1")]
pub name: String,
#[prost(int32, tag = "2")]
pub age: i32,
}
// Define a client that will use the mock
#[derive(Debug, Clone)]
struct UserServiceClient<T> {
inner: T,
}
// Implement the extension trait
impl GrpcClientExt<UserServiceClient<MockableGrpcClient>>
for UserServiceClient<MockableGrpcClient>
{
fn with_mock(mock: MockableGrpcClient) -> Self {
Self { inner: mock }
}
}
// Implement client methods
impl UserServiceClient<MockableGrpcClient> {
pub async fn get_user(
&mut self,
request: Request<UserRequest>
) -> Result<Response<UserResponse>, Status> {
// Extract request data
let request_data = request.into_inner();
// Encode the request
let encoded = tonic_mock::grpc_mock::encode_grpc_request(request_data);
// Call the mock service
let (response_bytes, headers) = self.inner
.handle_request("user.UserService", "GetUser", &encoded)
.await?;
// Decode the response
let response: UserResponse =
tonic_mock::grpc_mock::decode_grpc_message(&response_bytes)?;
// Return the response
Ok(Response::new(response))
}
}
#[tokio::main]
async fn main() -> Result<(), Box<dyn Error>> {
// Create the mock
let mock = MockableGrpcClient::new();
// Configure responses
mock.mock::<UserRequest, UserResponse>("user.UserService", "GetUser")
// For user_id = "user1"
.respond_when(
|req| req.user_id == "user1",
MockResponseDefinition::ok(UserResponse {
name: "Alice".to_string(),
age: 30,
})
)
.await
// For user_id = "user2"
.respond_when(
|req| req.user_id == "user2",
MockResponseDefinition::ok(UserResponse {
name: "Bob".to_string(),
age: 25,
})
)
.await
// Default case - not found
.respond_with(
MockResponseDefinition::err(Status::new(
Code::NotFound,
"User not found"
))
)
.await;
// Create client with the mock
let mut client = UserServiceClient::with_mock(mock);
// Test with existing users
let alice_request = Request::new(UserRequest {
user_id: "user1".to_string()
});
match client.get_user(alice_request).await {
Ok(response) => {
println!("Found user: {} (age {})",
response.get_ref().name,
response.get_ref().age);
},
Err(status) => {
println!("Error: {}", status.message());
}
}
// Test with unknown user
let unknown_request = Request::new(UserRequest {
user_id: "unknown".to_string()
});
match client.get_user(unknown_request).await {
Ok(response) => {
println!("Found user: {} (age {})",
response.get_ref().name,
response.get_ref().age);
},
Err(status) => {
println!("Error: {} (code: {:?})",
status.message(),
status.code());
}
}
Ok(())
}
Key features of the mockable client:
- Configure responses for specific service methods
- Return different responses based on request content with predicates
- Simulate error responses
- Include custom metadata in responses
- Add delays to simulate network latency
- Reset mock configurations between tests
For a complete example, see examples/mockable_client_example.rs
.
tonic-mock
is distributed under the terms of MIT.
See LICENSE for details.
Copyright 2021 Tyr Chen