@digitak/esrun

esrun

esrun is a "work out of the box" library to execute Typescript (as well as modern Javascript with decorators and stuff) without having to use a bundler. This is useful for quick demonstrations or when launching your tests written in Typescript.

This library is a thin wrapper around esbuild which compiles Typescript almost instantly.

The harder work to run typescript is to deal with dependencies. For example, you may need to import other Typescript files, but also libraries written in Javascript and using either the CJS or the ESM format. All these use cases should be considered.

esrun is able to handle all the annoying stuff and make things work as you would expect.

Usage

Global installation

Install the library globally with your favorite package manager:

npm i -g @digitak/esrun

Then you can execute any Typescript file in the same way Node would execute a Javascript file:

esrun foo.ts

You can pass arguments like any process:

esrun foo.ts --option=bar --verbose -S

All file dependencies will be bundled and executed as well.

External module dependencies won't be bundled, it's up to the node engine to resolve dependencies.

Local installation

Install the library locally with your favorite package manager.

npm i -D @digitak/esrun

Then you can use it in your package.json scripts:

{
   "scripts": {
      "test": "esrun test"
   }
}

Running npm run test will run the first file that exists in the following list:

• /test.ts
• /test/index.ts
• /test/test.ts
• /test/main.ts
• /test.js
• /test/index.js
• /test/test.js
• /test/main.js

Custom tsconfig.json

You can pass a custom path to your tsconfig.json file from the CLI:

esrun --tsconfig:/custom/path/to/tsconfig.json foo.ts

Watch mode

You can also execute esrun in watch mode.

In watch mode, your file will automatically be re-executed every time itself or one of its dependencies is updated.

esrun --watch foo.ts

The --watch (or -w) option must be placed before the path of the file to execute. If you place it after the file path, it will be passed as an argument to foo.ts instead.

This feature is very useful when you are doing test-driven development. You can just run esrun --watch test.ts and enjoy a live output of your changes right into your console.

You may want to watch other files than your code files. For example, if you load data from a configuration file. In this case you can specify a glob (or a list of globs) that have to be watched:

esrun --watch:src/*.json foo.ts

Then any json file in the src/folder will re-trigger the run.

You can use several globs separated by a comma (but no space):

esrun --watch:src/*.json,test/*.json foo.ts

Preventing console clearing

In watch mode, every file change will trigger a console clearing. You can disable this behavior with the --preserveConsole option:

esrun --watch --preserveConsole foo.ts

Inspect mode

You can also execute esrun in inspect mode.

When run in inspect mode, your code will be connected to the Webkit DevTools to benefit the power of the browser console instead of the terminal console.

First, run your program in inspect mode:

esrun --inspect foo.ts

Then open about:inspect in a Chrome / Brave / Edge browser. You should see your program running in the Remote targets section.

Click on Open dedicated DevTools for Node and enjoy the browser console for your back-end program.

In case of troubleshooting, read the node documentation.

Inspect and watch mode are alas not compatible yet.

Importing a CJS module

If you import a CJS module (like the typescript library itself), it's likely that you will need to set the esModuleInterop flag in your tsconfig.json file:

{
	"compilerOptions": {
		"esModuleInterop": true
	}
}

This will suppress the import errors from the Typescript compiler and allow you to write import ts from "typescript" instead of import * as ts from "typescript" - the latest syntax being not standard ESM.

Using a directory as an entry point

If the given entry point is a directory, the following actions will be executed in order to find the right entry file:

• check if a package.json file exists with a main field. The entry file will be the value of the main field, relative to the package.json directory.
• check if an index.ts file exists in the given directory.
• check if an eponym file exists in the given directory.
• check if an eponym file with the .ts extension exists in the given directory.
• check if a main.ts file exists in the given directory.
• check if a index.js file exists in the given directory.
• check if an eponym file with the .js extension exists in the given directory.
• check if a main.js file exists in the given directory.

API

The library exports a single function that you can use to programmatically execute a Typescript file.

import esrun from '@digitak/esrun'

export async function esrun(filePath: string, options?: Options): Promise<void>

export type Options = {
   // arguments to pass to the script
	args?: string[] = []

   // if true, will reload the script on file changes
   // you can also pass an additional array of globs to watch
	watch?: boolean | string[] = false

   // if true, prevent console clearing on watch mode
	preserveConsole?: boolean

   // if true, turn on inspect mode to use browser's console
   inspect?: boolean = false

   // if false, do not transform __dirname and __filename
   // (the CLI option to disable file constants is --noFileConstants)
	fileConstants?: boolean = true

   // if false, external packages will be bundled
	makeAllPackagesExternal?: boolean = true

   // if false, process.exit() won't be called after execution
	exitAfterExecution?: boolean = true

   // enable use of process.send() from the children
	interProcessCommunication?: boolean = false

   // executed before the code is executed (after the build)
   beforeRun?: () => unknown

   // executed after the code is executed
   afterRun?: () => unknown
}

Create a new runner to get / transform generated code

To have full control, you can create your own script runner instance:

import { Runner } from '@digitak/esrun'

const runner = new Runner(inputFile: string, options?: Options)

// build the given file and all its dependencies
await runner.build(buildOptions?: BuildOptions)

// you can see what the generated code is
console.log("Generated javascript code:", runner.outputCode)

// you can apply transformations to the code
await runner.transform(code => `console.log('Hello world!');\n` + code)

// then execute the build and return the given status
const status = await runner.execute()

Receive data

You can receive data from a script you executed by turning on the option interProcessCommunication.

When the option is on, the script will be able to call process.send(message: string).

At the end of the execution, the sent messages will be disponible through runner.output.

Let's suppose you have the following file helloWorld.ts:

// helloWorld.ts
process.send('Hello world')

Then you can receive data fro this script this way:

const runner = new Runner('helloWorld.ts', { interProcessCommunication: true })

await runner.execute({ exitAfterExecution: false })

console.log("Received data:", runner.output)
// should log 'Received data: Hello world'

Changelog

3.2.8

• New strategy to detect external dependencies. Now check if paths are inside a parent node_modules directory instead of checking if the import start with ".", "/", "~"n "@/" or "$". The previous strategy used to fail for typescript aliases that didn't start with "@/", "~" or "$".