Stdlib: mb_ereg*() — multibyte POSIX regex API (ext/mbstring/mbregex.c parity)

[PurHur]

Category

stdlib

Problem

The mbregex subset of ext/mbstring is entirely absent. Only mb_strlen() (and a few other mb_* helpers) exist under ext/types/. Legacy apps and some frameworks still call mb_ereg(), mb_eregi(), mb_ereg_replace(), mb_split(), etc.

php-src reference

• php/php-src ext/mbstring/mbregex.c — PHP_FUNCTION(mb_ereg), mb_eregi, mb_ereg_replace, mb_ereg_replace_callback, mb_split
• php/php-src ext/mbstring/mbregex.h
• php/php-src ext/mbstring/mbstring.stub.php — signatures + mb_regex_encoding()

Repro (failure today)

<?php
var_dump(function_exists('mb_ereg'));
var_dump(function_exists('mb_regex_encoding'));
if (mb_ereg('^[a-z]+$', 'hello')) {
    echo "match\n";
}

./script/docker-exec.sh -- bash -lc 'source script/php-env.sh
php repro.php
php bin/vm.php repro.php'

Runtime	mb_ereg('^[a-z]+$', 'hello')
Zend (ext/mbstring)	1 + match
This compiler	undefined function

Scope (this repo)

Area	Path
v1 surface	mb_ereg, mb_eregi, mb_ereg_replace, mb_split, mb_regex_encoding, mb_regex_set_options
VM	Delegate to host \mb_* when ext/mbstring loaded; clear error when unavailable
Module	ext/mbstring/Module.php or extend ext/types/Module.php
JIT/AOT	VM-first; document matrix as VM-only v1
Tests	test/compliance/cases/stdlib/mb_ereg.phpt (skip when host lacks mbregex)

Done when

• Repro prints match on VM when host mbstring is available (Docker php-compiler:22.04-dev)
• mb_regex_encoding() get/set matches Zend for UTF-8 default
• No-match returns false (not exception) per Zend
• ./script/ci-fast.sh --filter mb_ereg green
• Capability matrix rows for each implemented function

Related

• Stdlib: mb_convert_encoding / mb_detect_encoding (ext/mbstring parity) #3075 mb_convert_encoding / mb_detect_encoding · Stdlib: mb_str_pad() — multibyte string padding (ext/mbstring/mbstring.c parity) #4006 mb_str_pad · Epic: Self-host critical path — compiler compiles itself (M3→M5) #1492

[PurHur]

claim: PHP Compiler Ext — I/O & Crypto — starting this run

[PurHur]

Merged via #18285

Implementation: ext/mbstring/mb_ereg.php, mb_eregi.php, mb_ereg_replace.php, mb_regex_encoding.php, mb_regex_set_options.php + shared helpers in VmMbstring (PCRE u-flag, same strategy as mb_split). By-ref $registers wired via BuiltinByRefParams.

Verification:

php bin/vm.php test/repro/maintainer_mb_ereg.php          # match
php bin/vm.php test/compliance/cases/stdlib/mb_ereg.phpt  # full Zend parity output
./script/phpunit.sh --filter 'MbEregTest|mb_split_basic'  # OK (3 tests)