Middleware does not accept a single function

[JamesMessinger]

According to the docs for the .middleware() method:

The callbacks parameter can be a function or a list of functions

However, it currently (as of v12.0.2) only works with an array of functions. Passing a single function does nothing.

yargs
  .command('myCommand', 'some command', {}, function(argv){
    console.log('Running myCommand!');
  })
  .middleware(() => console.log('Running middleware 1'))   // <--- this never runs
  .middleware(() => console.log('Running middleware 2'))   // <--- neither does this
  .parse(['myCommand']);

Here is a reproduction of the issue on RunKit.

The problem is caused by this block of code:

yargs/lib/middleware.js

Lines 3 to 7 in 6c6c27d

	if (Array.isArray(callback)) {
	Array.prototype.push.apply(globalMiddleware, callback)
	} else if (typeof callback === 'object') {
	globalMiddleware.push(callback)
	}

I believe that if (typeof callback === 'object') should be if (typeof callback === 'function').

I will submit a PR for this momentarily

[JimiC]

Confirmed. Currently, if you need to pass a single function, it's mandatory to add it to an Array and pass that.

What I don't get either, is why and when a callback is an object (See also https://github.com/yargs/yargs/blob/v12.0.2/test/middleware.js#L24).
From what I know a callback is always a function.

/cc @aorinevo

[JamesMessinger]

@JimiC - Agreed. I think the 'object' in the type check was just a typo or misunderstanding. It probably should have always been 'function' instead

[aorinevo]

Hi @BigstickCarpet thank you for raising this issue; looking forward to the PR.

@JimiC thank you for validating.

It was a misunderstanding. Although functions are objects, typeof function returns 'function'. For reference, I'm including a link to Todd Motto's post on use of typeof https://toddmotto.com/understanding-javascript-types-and-reliable-type-checking/#typeof-operator

@BigstickCarpet: As part of your PR, can you add a unit test for this to /test/middleware.js?

[JamesMessinger]

@aorinevo - Here's the PR: #1215

[JimiC]

And the related test is also adjusted.

[bcoe]

@JamesMessinger @JimiC please try npm i yargs@next, which should have a patch that fixes this; if the release is looking good, I'll promote it next week.

[JimiC]

LGTM 👍

[arkon]

Seems like this should be closed?

[bcoe]

@arkon thanks for helping us groom some of our old issues 👍