PSR12.Classes.ClassInstantiation incorrectly fix when using member vars and some variable formats

[jrfnl]

WPCS already contained a ClassInstantiation sniff, so after seeing #2047 I've run the PSR12 version over the WPCS unit tests for the sniff to see if we could switch over at some point.

The test run identified a number of bugs in the PSR12 sniffs:

Test cases:

$a = new self::$transport[$cap_string];
$renderer = new $this->inline_diff_renderer;
$a = new ${$varHoldingClassName};

Fixed as:

$a = new self()::$transport[$cap_string];
$renderer = new $this()->inline_diff_renderer;
$a = new $(){$varHoldingClassName};

Expected fixes:

$a = new self::$transport[$cap_string]();
$renderer = new $this->inline_diff_renderer();
$a = new ${$varHoldingClassName}();

👉 Would you like me to look into fixing these ?

It also raises the question Why does FIG ignore this rule for anonymous classes ?, though that's outside the scope of PHPCS.

Other than the above, the WPCS sniff has some additional features.

• Detect class instantiation without parentheses in JS.
• Throw error when "Assigning new by reference" is detected (PHP only).
• Throw fixable error for whitespace between class name and open parenthesis (PHP + JS).
• Metrics for all errors.

👉 Would you be interested in any of these ?

[gsherwood]

Would you like me to look into fixing these ?

I'm working on this sniff right now due to another bug report, so I can take these ones. Thanks for reporting them.

Would you be interested in any of these ?

They are good checks, but I don't think they belong in the PSR-12 sniff. I prefer these PSR ones to be as focused as possible due to the arguments that they tend to bring up.

[gsherwood]

I've fixed these cases. If you find others, please report new issues and I'll fix them up too.

[jrfnl]

@gsherwood Thanks for the quick turn-around. ❤️

They are good checks, but I don't think they belong in the PSR-12 sniff. I prefer these PSR ones to be as focused as possible due to the arguments that they tend to bring up.

So leave off ? or would you like me to pull these as separate sniffs in the Generic standard ?

[gsherwood]

or would you like me to pull these as separate sniffs in the Generic standard ?

If you want to contrib them, and these checks don't exist elsewhere, then absolutely.