The purpose of this document is to cover the Relay APIs for testing Relay components.
The content is focused mostly on jest unit-tests (testing individual components) and integration tests (testing a combination of components). But these testing tools may be applied in different cases: screenshot-tests, production smoke-tests, "Redbox" tests, fuzz-tests, e2e test, etc.
What are the benefits of writing jest tests:
- It simplifies refactoring process: when properly written (testing public interface, not implementation) - tests really helps with changing the internal implementation of the components.
- It may speed up and improve the development workflow. Some people may call it Test Driven Development (TM). But essentially it's just writing tests for public interfaces of your components, and then writing the components that are implementing those interfaces. Jest —watch mode is really shining in this case.
- It will simplify the on-boarding process for new developers, having tests really help to ramp up on the new code base, fixing bugs, and delivering features.
One thing to notice - while jest unit- and integration tests will help improve the stability of the system, they should be considered as a part of a bigger stability infrastructure with multiple layers of automated testing: flow, e2e, screenshot, "Redbox", performance tests.
Testing applications that are using Relay may be challenging, because of the additional data fetching layer that is wrapping the actual product code.
And it's not always easy to understand the mechanics of all processes that are happening behind Relay, and how to properly handle interactions with the framework.
Fortunately, there are tools that aim to simplify the process of writing tests for Relay components, by providing imperative APIs for controlling the request/response flow and additional API for mock data generation.
There are two main modules that you may using in your tests:
createMockEnvironment, you will be able to create an instance of
RelayMockEnvironment, a Relay environment specifically for your tests. The instance created by
createMockEnvironment is implementing the Relay Environment Interface and it also has an additional Mock layer, with methods that allow to resolve/reject and control the flow of operations (queries/mutations/subscriptions).
The main purpose of
MockPayloadGenerator is to improve the process of creating and maintaining the mock data for tested components.
One of the patterns you may see in the tests for Relay components: 95% of the test code is the test preparation: the gigantic mock object with dummy data, manually created, or just a copy of a sample server response that needs to be passed as the network response. And rest 5% is actual test. As a result, people don't test much. It's hard to create and manage all these dummy payloads for different cases. Hence, writing tests are time-consuming and painful to maintain.
@relay_test_operation, we want to get rid of this pattern and switch the developer's focus from the preparation of the test to the actual testing.
RelayMockEnvironment is a special version of Relay Environment with additional API methods for controlling the operation flow: resolving and rejection operations, providing incremental payloads for subscriptions, working with the cache.
- Methods for finding operations executed on the environment
getAllOperations()- get all operation executed during the test by the current time
findOperation(findFn => boolean)- find particular operation in the list of all executed operations, this method will throw, if operation is not available. Maybe useful to find a particular operation when multiple operations executed at the same time
getMostRecentOperation() -return the most recent operation, this method will throw if no operations were executed prior this call.
- Methods for resolving or rejecting operations
nextValue(request | operation, data)- provide payload for operation(request), but not complete request. Practically useful when testing incremental updates and subscriptions
complete(request | operation)- complete the operation, no more payloads are expected for this operation, when it's completed.
resolve(request | operation, data)- resolve the request with provided GraphQL response. Essentially, it's nextValue(...) and complete(...)
reject(request | operation, error)- reject the request with particular error
resolveMostRecentOperation(operation => data)- resolve and getMostRecentOperation work together
rejectMostRecentOperation(operation => error)- reject and getMostRecentOperation work together
queueOperationResolver(operation => data | error)- adds an OperationResolver function to the queue. The passed resolver will be used to resolve/reject operations as they appear
queuePendingOperation(query, variables)- in order for the
usePreloadedQueryhook to not suspend, one must call these functions:
preloadQuery(mockEnvironment, query, variables)with the same
variablesthat were passed to
preloadQuerymust be called after
- Additional utility methods
isLoading(request | operation)- will return
trueif operations has not been completed, yet.
cachePayload(request | operation, variables, payload)- will add payload to QueryResponse cache
clearCache()- will clear QueryResponse cache
Mock Payload Generator and the
MockPayloadGenerator may drastically simplify the process of creating and maintaining mock data for your tests. MockPayloadGenerator is the module that can generate dummy data for the selection that you have in your operation. There is an API to modify the generated data - Mock Resolvers. With Mock Resolvers, you may adjust the data for your needs. Collection of Mock Resolvers it's an object where keys are names of GraphQL types (ID, String, User, Feedback, Comment, etc), and values are functions which will return the default data for the type.
Example of a simple Mock Resolver:
It is possible to define more resolvers for Object types
The first argument of the MockResolver is the object that contains Mock Resolver Context. It is possible to return dynamic values from mock resolvers based on the context - for instance, name or alias of the field, a path in the selection, arguments, or parent type.
The second argument of the Mock Resolver its a function that will generate a sequence of integers, useful to generate unique ids in the tests
Please note, that for production queries we don't have full type information for Scalar fields - like Boolean, Integer, Float. And in the MockResolvers, they map to String. You can use
context to adjust return values, based on the field name, alias, etc.
Most of GraphQL type information for a specific field in the selection is not available during Relay runtime. By default, Relay, cannot get type information for a scalar field in the selection, or an interface type of the object.
Operation with the @relay_test_operation directive will have additional metadata that will contain GraphQL type info for fields in the operation's selection. And it will improve the quality of the generated data. You also will be able to define Mock resolvers for Scalar (not only ID and String) and Abstract types:
MockPayloadGenerator allows writing concise tests for components that are using Relay hooks. Both those modules can be imported from
Essentially, in the example above will
resolveMostRecentOperation will generate data for all child fragment containers (pagination, refetch). But, usually the root component may have many child fragment components and you may want to exercise a specific component that uses
useFragment. The solution for that would be to wrap your fragment container with the
useLazyLoadQuery component that renders a Query that's spreads fragments from your fragment component:
Essentially, tests for pagination components (e.g. using
usePaginationFragment) are not different from fragment component tests. But we can do more here, we can actually see how the pagination works - we can assert the behavior of our components when performing pagination (load more, refetch).
We can use similar approach here with wrapping the component with a query. And for the sake of completeness, we will add example here:
Mutations itself are operations so we can test them independently (unit-test) for specific mutation, or in combination with the view from which this mutation is called.
useMutationAPI is an improvement over calling
useSubscriptionAPI is an improvement over calling
We can test subscriptions similarly to how we test mutations.
queueOperationResolver it possible to define responses for operations that will be executed on the environment
The examples in this guide should work for testing components both with Relay Hooks, Containers or Renderers. When writing tests that involve the
usePreloadedQuery hook, please also see the
queuePendingOperation note above.
Even though in all of the examples here you can see assertions with
toMatchSnapshot(), we keep it that way just to make examples concise. But it's not the recommended way to test your components.
React Testing Library is a set of helpers that let you test React components without relying on their implementation details. This approach makes refactoring a breeze and also nudges you towards best practices for accessibility. Although it doesn't provide a way to "shallowly" render a component without its children, a test runner like Jest lets you do this by mocking.
The best source of example tests is in the relay-experimental package.
Testing is good. You should definitely do it.
Is this page useful?