nyr (npm-yarn-run)

A tiny helper that knows if you are running a script via npm or Yarn, to let you chain your commands for either.

Use as a replacement for x run in your package scripts.

$ yarn add nyr
$ npm install --save nyr

Example

The intention is for this to be used as a CLI in your package.json:

{
  "name": "example-project",
  "main": "index.js",
  "scripts": {
    "lint:eslint": "eslint",
    "lint:stylelint": "stylelint",
    "lint": "nyr lint:eslint && nyr lint:stylelint"
  }
}

Now if you run npm run lint or yarn lint, then lint:eslint and lint:stylelint will be executed with npm run x or yarn run x depending on whichever you called the script with.

Note: run is implied before all arguments. Yarn is clever and will check for user-defined script if you run an unrecognized command, but npm only works like that with a handful of pre-defined script names (for good reasons, as this avoids conflicts). For compatibility reasons, all commands will be executed with run (prepended if not manually provided) for both npm and Yarn.

Node API

You can use this package as a node module as well:

const {
  bin,    // (String)  path to executed binary (yarn or npm) 
  args,   // (Array)   arguments passed to the script (with "run" first)
  cwd,    // (String)  where the script was executed from
  isYarn, // (Boolean) if it was executed with Yarn
  isNPM,  // (Boolean) if it was executed with npm
} = require('nyr')

This would probably only be useful if it's a script you intend to call from your scripts in package.json, as calling it directly with node wouldn't provide much useful information (and args would likely be off).

Why?

I like Yarn, and I wish everyone would use it. But I also don't want to make any assumptions about what tool people are running yet, so I'd like to keep my scripts in package.json compatible with both, while still having them neat and tidy.

There's actually already a similar tool (which I've been using) available called yarn-or-npm. However, it relies on checking if Yarn is available on your system, and not which tool you're actually using.

I also needed a tool which correctly pipes all i/o and returns the correct exit codes, so tests and CI pass or fail as expected.

It does however cover more use-cases, so I'd recommend checking it out if you need something more extensive!

License

MIT © 2017 Simon Kjellberg