As with other string functions, there is a problem with Turkish "i" with this function. There is a bug report from 2015 about the issue but PHP team says "language-specific conditional special case mappings is not implemented", although actually it breaks the logic of the function and renders it non-usable for the purpose.
https://bugs.php.net/bug.php?id=70072
The problem arises from the letter "i" in Latin being a COMPLETELY different letter from "i" in Turkish. Turkish "ı" becomes "I" for capital; while Latin "I" capital is actually capital for "i" and not "ı".
PHP takes this into consideration in some cases and ignores it in other cases; which causes an unpredictable behavior. When the letters in question is in the middle or at the beginning of a word, when some of multibyte chars are next to standard Latin chars or another multibyte character etc. These all behave differently, which is simply wrong.
There are some user notes trying to cover this but not very efficiently. Because some of them doesn't cover word boundaries and some produce non-standard characters. Here is what I tested and have been using for quite a time:
<?php
function mb_convert_case_i(string $string, int $mode = MB_CASE_TITLE, string $encoding = 'UTF-8'): string {
// Turkish "i" is a special case
$string = match($mode) {
MB_CASE_UPPER, MB_CASE_UPPER_SIMPLE => str_replace(['i', 'ı'], ['İ', 'I'], $string),
MB_CASE_LOWER, MB_CASE_LOWER_SIMPLE => str_replace(['İ', 'I'], ['i', 'ı'], $string),
// PHP behaves differently when i and ı are at the beginning of the word
MB_CASE_TITLE, MB_CASE_TITLE_SIMPLE => preg_replace(['/İ/u', '/I/u', '/\b(i)/u'], ['i', 'ı', 'İ'], $string),
default => $string,
};
return mb_convert_case($string, $mode, $encoding);
}
?>
As you have noticed, it uses match syntax which requires PHP 8. For lower versions, you can replace it with switch properly. I haven't tested it for case folding. If you need it, just add another condition to the match.