Stdlib: mb_scrub() — PHP 8.4 invalid-byte scrubbing missing (ext/mbstring/mbstring.c)

[PurHur]

Category

stdlib

Problem

PHP 8.4 adds mb_scrub(string $string, ?string $encoding = null): string — replace invalid byte sequences with ? (or U+FFFD for UTF-8) for the given encoding. Mail sanitizers, log pipelines, and HTTP body normalizers use it alongside mb_convert_encoding() (#3075).

Maintainer audit (Jun 2026): function_exists('mb_scrub') → false on bin/vm.php.

php-src reference

• php/php-src ext/mbstring/mbstring.c — PHP_FUNCTION(mb_scrub)
• php/php-src ext/mbstring/mbstring.stub.php — signature + $encoding default (null → internal encoding)
• php/php-src ext/mbstring/libmbfl — invalid-sequence detection / substitution

Repro (failure today)

Save as test/repro-maintainer/parity_mb_scrub.php:

<?php
var_export(function_exists('mb_scrub'));
// Invalid UTF-8 byte
echo mb_scrub("\xFF", 'UTF-8'), "\n";
// Valid UTF-8 unchanged
echo mb_scrub("café", 'UTF-8'), "\n";
// Default encoding (internal)
echo mb_scrub("\xFF"), "\n";

./script/docker-exec.sh -- bash -lc 'source script/php-env.sh
php test/repro-maintainer/parity_mb_scrub.php
php bin/vm.php test/repro-maintainer/parity_mb_scrub.php 2>&1 | head -5
'

Runtime	function_exists	scrub \xFF UTF-8	valid café
Zend PHP 8.4+	true	? or U+FFFD	unchanged
bin/vm.php	false	fatal	—

php-src-strict enum operand

<?php
enum E: string { case A = 'x'; }
try {
    mb_scrub(E::A, 'UTF-8');
} catch (Throwable $e) {
    echo get_class($e), ': ', $e->getMessage(), "\n";
}

Must throw catchable TypeError (mirror #5873 mb_strlen pattern), not silent backing-string coercion.

Scope (PHP-in-PHP first)

Path	Work
ext/mbstring/mb_scrub.php	Algorithm in PHP using existing mbstring module (#5695)
ext/mbstring/Module.php	Register function
ext/mbstring/VmMbstring.php	Shared invalid-byte scrub helper (reuse for #6038 MIME if applicable)
JIT/AOT	VM-only v1

Avoid: new logic in runtime/*.c.

Done when

• function_exists('mb_scrub') → true on VM
• Invalid-byte repro output byte-matches Zend on Docker harness
• Valid UTF-8 string unchanged
• Enum case $string → TypeError (php-src-strict)
• test/compliance/cases/stdlib/mb_scrub.phpt green
• ./script/ci-fast.sh --filter MbScrub green

Verification

./script/docker-exec.sh -- bash -lc 'source script/php-env.sh && vendor/bin/phpunit --filter MbScrub'

Related

• Stdlib: mb_convert_encoding / mb_detect_encoding (ext/mbstring parity) #3075 mb_convert_encoding · Stdlib: mb_encode_mimeheader()/mb_decode_mimeheader() — MIME header encoding missing (ext/mbstring/mbstring.c) #6038 MIME headers · Stdlib: mb_strlen() — enum case operand must TypeError (ext/mbstring/mbstring.c) #5873 mb_strlen enum TypeError · Bootstrap: ext/mbstring/ module skeleton — register mbstring builtins for self-host (M4) #5695 mbstring module · Epic: Self-host critical path — compiler compiles itself (M3→M5) #1492

[PurHur]

claim: PHP Compiler STD Lib — starting this run

[PurHur]

Merged via #8112

Verification (green on branch):

• VM: MbScrubVMTest — invalid UTF-8/ASCII scrubbing + enum TypeError
• AOT: AotTest --filter mb_scrub — compile-time literal folding
• Repro: php bin/vm.php test/repro-maintainer/parity_mb_scrub.php → true, ?, café, ?

Implementation: ext/mbstring/mb_scrub.php, VmMbstring::scrub(), JitMbScrub (PHP-in-PHP, no C runtime).
php-src ref: ext/mbstring/mbstring.c — PHP_FUNCTION(mb_scrub)