Fix: edge-cases of semi-style

[mysticatea]

What is the purpose of this pull request? (put an "X" next to item)

[X] Bug fix

Tell us about your environment

• ESLint Version: 4.10.0
• Node Version: 8.9.0
• npm Version: 5.5.1

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

• default

Please show your full configuration:

• nothing

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

The online demo for "first" option

/*eslint semi-style: [error, first] */

{
	; // Should be ignored. It's reported by `no-extra-semi`.
}

while (a)
	; // Should be ignored to keep the indented loop body.
foo

do
	; // Should be ignored to keep the indented loop body.
while (a)

do
	foo; // Should be ignored instead of making `;while`.
while (a)

if (a)
	foo; // Should be ignored instead of making `;else`.
else
	bar

The online demo for "last" option

/*eslint semi-style: [error, last] */

{
	; // Should be ignored. It's reported by `no-extra-semi`.
}

switch (a) {
	case 1:
		;foo // Should be ignored. It's reported by `no-extra-semi`.
}

while (a)
	; // Should be ignored to keep the indented loop body.
foo

do
	; // Should be ignored to keep the indented loop body.
while (a)

What did you expect to happen?

No errors.

What actually happened? Please include the actual, raw output from ESLint.

Several errors are reported. See the demo pages.

What changes did you make? (Give an overview)

This PR fixes those false positives. Now the rule ignores semicolons at the first/last in a block/clause.

Is there anything you'd like reviewers to focus on?

Nothing in particular.

[not-an-aardvark]

Could this be !astUtils.STATEMENT_LIST_PARENTS.has(node.parent.type)?

[mysticatea]

Oh, that's simple. Yes, it works fine!

[mysticatea]

After I rethought, I'm wondering if I remove the check of EmptyStatement entirely. It makes only redundant consequent semicolons... E.g. foo;;\nbar or foo\n;;bar. No-extra-semi rule should remove that.

[mysticatea]

I updated this PR to simplify with dropping EmptyStatements. The behavior follows to Semi rule which ignores EmptyStatements.

[aladdin-add]

LGTM, just a question 😄

[aladdin-add]

could you please explain a little why this case is valid? IMHO, it's using options: ["last"], so it should be foo;

[mysticatea]

Sure. This rule checks the location of the semicolon between two statements. I.e.:

first();
second()
// OR
first()
;second()

The rule does not add/remove semicolons because it's the responsibility of semi and no-extra-semi rules. Instead, the rule handles whitespaces of around semicolons.

Now, the current behavior is that the case 1:\n;foo is warned and fixed to case 1:;\nfoo, it was wrong.