ESM | Modern javascript tooling |
TypeScript | Strongly-typed test definitions |
Puppeteer | Cross-runtime testing (Node & Browser) |
esbuild | Fast test bundling |
Aider.ai | AI-powered test fixing |
ESLint | Static analysis of test files |
tsc | Type checking of test files |
npm install testeranto
rectangle.test.ts
):import { Given, When, Then } from "testeranto";
type Rectangle = { width: number; height: number };
const RectangleSpec = (Suite, Given, When, Then) => [
Suite.Default("Rectangle tests", {
test1: Given.Default(
["Basic rectangle operations"],
[When.setWidth(5), When.setHeight(10)],
[Then.getWidth(5), Then.getHeight(10)]
),
}),
];
To start testeranto in dev mode, build your tests in one terminal and execute them in another
# Terminal 1 - Build in watch mode yarn t-build rectangle.test.ts yourProject dev # Terminal 2 - Run in watch mode yarn t-run rectangle.test.ts yourProject dev
or build and run your tests only once
yarn t-build rectangle.test.ts yourProject once && yarn t-run rectangle.test.ts yourProject once
You should use this runtime for... | Important differences | |
---|---|---|
Node : node V8 with fork | testing backend code, Node APIs (like fs and crypto ), or anything needing filesystem access |
has access to the filesystem and io |
Web : chrome browser | testing frontend code, anything that uses document or window , UI interactions, or visual regression |
can take screenshots/recordings |
Pure : node v8, dynamically imported | testing code that can run on both node-v8 and the browser | Very similar to "Node" but has no IO access and thus no console.log. This runtime is theoretically faster. |
yarn t-init |
Create a new testeranto project |
yarn t-build <YOUR_TESTS> dev|once |
Build test bundles (watch or single-run mode) |
yarn t-run <YOUR_TESTS> dev|once |
Run tests (watch or single-run mode) |
yarn t-report |
Launch test report server |
yarn t-aider |
Fix failing tests with AI |
# Initialize project yarn t-init # Write tests in test/*.test.ts # In terminal 1 - Build tests (watch mode) yarn t-build test/rectangle.test.ts dev # In terminal 2 - Run tests (watch mode) yarn t-run test/rectangle.test.ts dev # Or for single-run mode: yarn t-build test/rectangle.test.ts once yarn t-run test/rectangle.test.ts once # Get AI help with failures yarn t-aider
Testeranto generates a "prompt" alongside test results. This prompt is passed to aider as input.
// input src files that can be edited by Aider /add test/node.ts // test report files that inform aider but should not be edited /read testeranto/reports/allTests/node/test/node/tests.json /read testeranto/reports/allTests/test/node/node/lint_errors.json /read testeranto/reports/allTests/test/node/node/type_errors.txt // A list of features which can inform aider. /load testeranto/reports/allTests/node/test/node/featurePrompt.txt // tell the AI what to do Fix the failing tests described in testeranto/reports/allTests/node/test/node/tests.json. Correct any type signature errors described in the files testeranto/reports/allTests/test/node/node/type_errors.txt. Implement any method which throws "Function not implemented. Resolve the lint errors described in testeranto/reports/allTests/test/node/node/lint_errors.json"
Testeranto connects "features" to tests. The features may be simple strings, but they can also take the form of local markdown files, or remote URLs to external feature tracking systems. For instance, this could be a jira ticket or a github issue. These features are used to inform the AI context.
import someMarkdownFile from "someMarkdownFile.md";
...
test0: Given.Default(
[
"https://api.github.com/repos/adamwong246/testeranto/issues/8",
"you can set the width and height of a Rectangle",
someMarkdownFile
],
[When.setWidth(4), When.setHeight(19)],
[Then.getWidth(4), Then.getHeight(19)]
),
...
Alongside the BDD tests, Testeranto runs ESLint and tsc on the input files to generate a list of static analysis errors and type errors, respectively. Aider will use this to resolve both.
Testeranto has a core repo, but there are also sub-projects which implement tests by type and by technology
Test a solidity contract. Also included is an example of deploying a contract to a ganache server.
Tests a Redux store.
Tests a Node HTTP server.
Test a react component. You can choose from a variety of types (jsx functions, class style, etc) and you can test with react
, react-dom
, or react-test-renderer
Alongside your test, you can include a number of "sidecars" - other bundled JavaScript assets upon which your test depends. For example, suppose you have an app with frontend and backend components. You could run a React test in the web and include the Node HTTP server as a sidecar.