Welcome to Keploy 👋

Keploy

Keploy is a no-code testing platform that generates tests from API calls.

It converts API calls into testcases. Mocks are automatically generated with the actual request/responses.

Features

Generates test cases from API calls. Say B-Bye! to writing unit and API test cases.

Native interoperability with popular testing libraries like go-test. Code coverage will be reported with existing and Keploy recorded test cases and can also be integrated in CI pipelines/infrastructure.

Other Features

Automatically mock network/external dependencies with correct responses. No more manually writing mocks for dependencies like DBs, internal services, or third party services like twilio, shopify or stripe.
Safely replay writes or mutations by capturing from local or other environments. Idempotency guarantees are also not required in the application. Multiple Read after write operations can be replicated automatically too.
Accurate noise detection in responses like (timestamps, random values) to ensure high quality tests.
Statistical deduplication ensures that redundant testcases are not generated. We're planning to make this more robust (ref #27).
Web Console to visually understand the results, update behaviour and share findings across your team.
Automatic instrumentation for popular libraries/drivers like sql, http, grpc, etc.
Instrumentation/Integration framework to easily add the new libraries/drivers with ~100 lines of code.

How it works?

Note: You can generate test cases from any environment which has all the infrastructure dependencies setup. Please consider using this to generate tests from low-traffic environments first. The deduplication feature necessary for high-traffic environments is currently experimental.

Installation

Start keploy server

git clone https://github.com/keploy/keploy.git && cd keploy
docker-compose up

The UI can be accessed at http://localhost:8081

Keeping keploy up-to-date

docker-compose pull

Helm chart

Keploy can also be installed to your your Kubernetes cluster using the Helm chart available here

Sample application

Demos using Echo/PostgreSQL and Gin/MongoDB are available here. For this example, we will use the Echo/PostgreSQL sample.

git clone https://github.com/keploy/samples-go && cd samples-go/echo-sql
go mod download

Start PostgreSQL instance

docker-compose up -d

Run the application

go run handler.go main.go

Generate testcases

To genereate testcases we just need to make some API calls. You can use Postman, Hoppscotch, or simply curl

1. Generate shortned url

curl --request POST \
  --url http://localhost:8080/url \
  --header 'content-type: application/json' \
  --data '{
  "url": "https://github.com"
}'

this will return the shortened url. The ts would automatically be ignored during testing because it'll always be different.

{
	"ts": 1647802058801841100,
	"url": "http://localhost:8080/GuwHCgoQ"
}

2. Redirect to original url from shortened url

curl --request GET \
  --url http://localhost:8080/GuwHCgoQ

Run the testcases

Note: Before running tests stop the sample application

go test -coverpkg=./... -covermode=atomic  ./...

this should show you have 74.4% coverage without writing any code!

ok      echo-psql-url-shortener 5.820s  coverage: 74.4% of statements in ./...

All of these can be visualised here - http://localhost:8081/testlist

Integration with native Go test framework

You just need 3 lines of code in your unit test file and that's it!!🔥🔥🔥

import (
	"github.com/keploy/go-sdk/keploy"
	"testing"
)

func TestKeploy(t *testing.T) {
	keploy.SetTestMode()
	go main()
	keploy.AssertTests(t)
}

Language Support

Go SDK
Java SDK - WIP #51
Typescript/Javascript SDK - WIP #61
Python SDK - WIP #58
Need another language support? Please raise an issue or discuss on our slack channel

FAQs

Is Keploy a unit testing framework?

No, keploy is designed to reduce time writing tests manually. It integrates with exising unit testing frameworks like (eg: go test, Junit, pytest, etc.) to ensure compatibility with existing tooling like code coverage, IDE support and CI pipeline/infrastructure support.

Does Keploy replace unit tests entirely?

If all your code paths can be invoked from API calls then yes, else you can still write testcases for some methods, but the idea is to save at least 80% of the effort.

What code changes do I need to do?

Web Framework/Router middleware needs to be added to ensure keploy can intercept incoming request and inject instrumentation data in the request context.
Wrapping External calls like database queries, http/gRPC calls needs to be done to ensure they are captured and correct mocks are generated for testing those requests.

How do I run keploy in my CI pipeline?

No changes necessary. You can reuse the pipeline which runs unit tests.

Does Keploy support read after write to DB scenarios?

Yes. Keploy records the write requests and read requests in the correct order. It then expects the application to perform the writes and reads in the same order. It would return the same database responses as captured earlier.

How does keploy handle fields like timestamps, random numbers (eg: uuids)?

A request only becomes a testcase if it passes our deduplication algorithm. If its becoming a testcase, a second request is sent to the same application instance (with the same request params) to check for difference in responses. Fields such as timestamps, uuids would be automatically flagged by comparing the second response with the first response. These fields are then ignored during testing going forward.

Can I use keploy to generate tests from production environments automatically?

Not yet. We are working on making our deduplication algorithm scalable enough to be used safely in production. If you are interested in this use-case, please connect with us on slack. We'd love to work with you to build the deduplication system and load test it with your systems.

What if my application behaviour changes?

If your application behaviour changes, the respective testcases would fail. You can then mark the new behaviour as normal by clicking on the normalise button.

Would keploy know if an external service changes?

Not yet. Unless that application is also using keploy, keploy would only test the functionality of the current application. We are working to detect scanning for API contract violations and adding multiple application to perform comprehensive integration tests. All contributions are welcome.

Contributing

There's a separate docker-compose file which helps with exposing the mongo server and also builds the dockerfile from local code. The build flag ensures that the binary is built again to reflect the latest code changes. There's also docker-compose-debug.yaml which can help remote debugging the go server on port 40000.

docker-compose -f docker-compose-dev.yaml up --build

If you are not using docker, you can build and run the keploy server directly. Ensure to provide the Mongo connection string via the KEPLOY_MONGO_URI env variable.

export KEPLOY_MONGO_URI="mongodb://mongo:27017"
go run cmd/server/main.go

Keploy exposes GraphQL API for the frontend based on gqlgen. After changing the schema you can autogenerate graphQL handlers using

go generate ./...

Community support

We'd love to collaborate with you to make Keploy great. To get started:

Slack - Discussions with the community and the team.
GitHub - For bug reports and feature requests.

poovaraj/keploy