PCRE2 10.45-10.47 contains a known regex character class compilation issue #20341
Description
Description
The current PHP 8.5 builds (RC3) bundle PCRE2 10.46, which contains a known issue in character class handling (PCRE2Project/pcre2#833),
This may lead to incorrect (non-)matches in certain patterns containing bracketed character class [...].
Affected PCRE2 versions are 10.45-10.47, dev-master.
The following code: https://3v4l.org/iZ91q
$patterns = [
'/[\x{ff}\x{100}\x{8000}\x{8002}\x{8004}\x{8006}]/u' => "\u{100}",
'/[\x{ff}\x{100}\x{8000}\x{8002}\x{8004}\x{8006}\x{8008}]/u' => "\u{100}",
'/[\x{ff}\x{101}\x{8000}\x{8002}\x{8004}\x{8006}\x{8008}]/u' => "\u{101}",
];
foreach ($patterns as $pattern => $str) {
if (preg_match($pattern, $str, $m)) {
echo "0: ", json_encode($m[0]), "\n";
} else {
echo "No match.\n";
}
}Resulted in this output:
0: "\u0100"
No match.
0: "\u0101"
But I expected this output instead:
0: "\u0100"
0: "\u0100"
0: "\u0101"
PHP Version
PHP 8.5.0RC3 (cli) (built: Oct 21 2025 21:38:34) (NTS Visual C++ 2022 x64)
Copyright (c) The PHP Group
Zend Engine v4.5.0RC3, Copyright (c) Zend Technologies
with Zend OPcache v8.5.0RC3, Copyright (c), by Zend Technologies
Operating System
Windows, but also confirmed on pcre2test -32 on Linux PCRE2 10.47