no-irregular-whitespace: Inside of regexp irregular whitespace

[terrierscript]

What version of ESLint are you using?
2.7.0

What parser (default, Babel-ESLint, etc.) are you using?
default

Please show your full configuration:

{
    "env": {
        "node": true,
    },
    "rules": {
        "no-irregular-whitespace": 2
    }
};

What did you do? Please include the actual source code causing the issue.

I wrote those code.

var reg = /^[\s\　]*$/
reg.test('foo')

What did you expect to happen?

  1:5  error  Parsing error: Unexpected token reg

I think this may bug because string literal is ignored on this rule.

If this is not bug, I want documentation or skipRegexp option.

Note: I can avoid this error with new RegExp('^[\s\　]*$')

[mysticatea]

Thank you for this issue.

Indeed, no-irregular-whitespace ignores spaces in string literals.
So I feel this is a bug.

[kaicataldo]

I actually don't think this is a bug - seems to be working as intended, as the rule states it causes warnings if irregular whitespace characters are found outside of strings or comments. From the docs:

With this rule enabled the following characters will cause warnings outside of strings and comments ...

I think that it's not working for your first example because it's a regex literal (not a string). The reason new RegExp('^[\s\　]*$') doesn't warn is because the irregular whitespace is in a string. Or that's my interpretation of the rules docs, anyway :)

[terrierscript]

Oh sorry, my explanation was lacking.

If this is regular behavior, I think this should documentation with incorrect example.
But I wish fix this behavior or append ignore option.

Because regexp with Ideographic Space' pattern is often used in multi byte language.

[kaicataldo]

Got it - I'm going to label this as an enhancement and let's see what @nzakas thinks. Thanks for the issue report!

[mysticatea]

Hmm, I cannot find reasons that irregular spaces in strings are good but irregular spaces in regular expressions are bad. And I got aware that template literals have the same issue:

/^[\s\　]*$/             /*error Irregular whitespace not allowed (no-irregular-whitespace)*/
new RegExp(`^[\s\　]*$`) /*error Irregular whitespace not allowed (no-irregular-whitespace)*/
new RegExp("^[\s\　]*$") // good.

I wish to fix this behavior since this does not have consistency, IMO.

What about adding options for each type?

{
    "no-irregular-whitespace": ["error", {
        "skipStrings": false,
        "skipTemplates": false,
        "skipRegExps": false,
        "skipComments": false // "skipComments" exists already.
    }]
}

In my ideal, all default are false, but it's a breaking change.

[nzakas]

Irregular whitespace in strings might be intentional for output. No objections to adding the options if @mysticatea wants to as long as the default behavior remains the same.

[rhysd]

I just encountered this bug today and sent a PR for this in #6018. Could you check it?

[mysticatea]

@rhysd Thank you for the PR 😄. But the consensus on this issue is to add options above. Could you update the PR with the options?

[rhysd]

@mysticatea I'll fix my PR today 👍

[rhysd]

        "skipStrings": false,
        "skipTemplates": false,

Should we separate these options? I think skipTemplates should be unified into skipStrings. What do you think?

[pedrottimark]

@rhysd If we were designing the rule today, unlikely to find a use case to separate options for string literals from template literals. The reason is to avoid breaking existing behavior, as requested in #5840 (comment).

Therefore, I think the default values of the options are as follows:

{
    "no-irregular-whitespace": ["error", {
        "skipStrings": true, // current default is to skip strings
        "skipTemplates": false,
        "skipRegExps": false,
        "skipComments": false // this option already exists
    }]
}

[rhysd]

@pedrottimark

OK, I'll maintain the previous behavior 👍