ZWJ emoji sequences (e.g. 💆‍♀️) are broken by clean() in v4.0.1

[seth-hayward]

Hello!

In v4.0.1, Blasp::check()->clean() appears to strip Unicode format characters (\p{Cf}), which breaks valid emoji ZWJ sequences.

For example, the emoji 💆‍♀️ (woman getting massage) is rendered as two separate emojis after calling clean(), instead of a single combined glyph.

This behavior did not occur in previous versions (4.0.0).

Reproduction:

$text = "Test 💆‍♀️ emoji";

$result = Blasp::check($text);

$output = $result->clean();

echo $output;

Expected Behavior

The emoji should remain intact:

Test 💆‍♀️ emoji

Actual Behavior

The ZWJ sequence is broken, resulting in something like:

Test 💆 ♀️ emoji

(or similar split rendering depending on platform)

Thank you for your package. This has been very helpful to me on my Laravel app.