@tapjs/create-plugin
Create a tap plugin with npm init @tapjs/plugin
(or yarn create @tapjs/plugin
).
This will create the basic scaffolding required to write a tap plugin.
Tap plugins are typically a function that takes a Test
object
and optionally an options
object, and returns an object that is
used as the extension.
For example, a very simple "hello, world" plugin:
import { TapPlugin, TestBase } from '@tapjs/core'
export interface HelloSayer {
hello: (who?: string) => string
}
export const plugin: TapPlugin<HelloSayer> = (t: TestBase) => {
return {
hello: (who: string = 'world') => {
console.error(`${t.name} says "Hello, ${who}!"`)
},
}
}
As a more realistic example, to add a isString
method to all
tests, you could define a plugin like this:
import {
TestBase,
TapPlugin,
Extra,
MessageExtra,
normalizeMessageExtra,
} from '@tapjs/core'
export class StringTester {
#t: TestBase
constructor (t: TestBase) {
this.#t = t
}
isString (
s: any,
...[msg, extra]: MessageExtra
) {
const args = [msg, extra] as MessageExtra
const me = normalizeMessageExtra('expect string', args)
// note: 'this' here is the StringTester plugin object
if (typeof s === 'string') {
return this.#t.pass(msg, extra)
} else {
return this.#t.fail(msg, {
...extra,
value: s,
wanted: 'string',
found: typeof s,
})
}
}
}
export const plugin: TapPlugin<StringTester> =
t => new StringTester(t)
The object returned by a plugin can be any sort of thing. If you
want to use a class with private properties, that's totally fine
as well. Whatever type is expected as the second argument will
be combined with the built-in TestBaseOpts
interface, and
required when tests are instantiated.
import { TestBase, TapPlugin, AssertionOpts } from '@tapjs/core'
import { cleanup, render } from '@testing-library/react'
import { ReactElement } from 'react'
import { RenderResult } from '@testing-library/react'
import userEvent from '@testing-library/user-event'
class ReactTest {
#result?: RenderResult
constructor(node: ReactElement) {
if (node) {
this.#result = render(node)
}
}
findByText(text: string) {
return this.#result?.findByText(text)
}
// add other helpful methods here...
}
export interface ReactTestOpts {
node?: ReactElement
}
export const plugin: TapPlugin<ReactTest, ReactTestOpts> = (t, opts) =>
new ReactTest(node)
When loaded, this plugin would make it so that every test may
provide a { node: ReactElement }
option, and would then have a
t.findByText()
method that returns the results.
If you need access to the constructed Test
object, you can get
that after the initial plugin load, via t.t
. However, it will
be undefined
until all plugins are done loading, so do not rely
on it being present in your plugin function itself or any
constructors it calls synchronously.
// my-plugin.ts
export const plugin = (t: TestBase) => {
// here, t.t === undefined
return {
someMethod() {
// here, t.t is the object with all the plugins applied
},
}
}
require()
and
import()
. This maintains tap's ability to run in all
JavaScript dialects.loader
, importLoader
,
plugin
, or config
.A common use-case for plugins is to manage some state that needs to be disposed at the end of the test.
In those cases, you can leverage the
@tapjs/after
plugin's functionality by testing to see whether it is loaded.
For example, if you wanted to ruin the process
object for some
reason:
import { TapPlugin, TestBase } from '@tapjs/core'
import { plugin as AfterPlugin } from '@tapjs/after'
export class BreakProcess {
// the method that restores the world to how it was before
#restore?: () => void
#t: TestBase
#didCleanup: boolean = false
constructor (t: TestBase) {
this.#t = t
}
restoreProcess() {
if (this.#restore) {
this.#restore()
}
}
breakProcess() {
// if we already broke it, nothing to do
if (this.#restore) return
const originalProcess = process
global.process = { not: 'the actual process object' }
this.#restore = () => {
global.process = originalProcess
this.#restore = undefined
}
if (this.#t.t.pluginLoaded(AfterPlugin) && !this.#didCleanup) {
this.#t.t.after(() => this.restoreProcess())
}
}
}
(Don't actually write this plugin, though. Just use
@tapjs/intercept
for this, it's included with tap already, and does the sort of
cleanup described here.)
The first plugin in a list that provides a given method or property will be the one that "wins", as far as the object presented in test code is concerned.
However, within a given plugin, it only sees itself and the
TestBase
object it's been given. For example, if returning an
object constructed from a class defined in the plugin, this
will refer to that object, always.
// first-plugin
export const plugin = (t: TestBase) => {
return {
// this is the first plugin to register this value
// so this is what shows up on the Test object
myVal: 4,
getFirstPluginVal() {
return this.myVal // always returns 4
},
// this is the first plugin to register this method
// so this is what shows up on the Test object
getFour() {
return 4
},
}
}
// second-plugin
export const plugin = (t: TestBase) => {
return {
// user will never see this, because first-plugin registered it
myVal: 5,
getSecondPluginValue() {
return this.myVal // always returns 5
},
// overridden, this isn't the 'getFour' that the user will see
getFour() {
return 'four'
},
}
}
Then in the test:
import t from 'tap'
console.log(t.myVal) // 4, not 5
console.log(t.getFour()) // 4, not 'four'
console.log(t.getFirstPluginVal()) // 4
console.log(t.getSecondPluginVal()) // 5
While collisions like this are usually not a big deal, this behavior can get confusing. So, it's best to name properties and methods somewhat uniquely, so as to make collisions less likely.