Stdlib: RecursiveRegexIterator — regex-filtered Iterator missing (ext/spl/spl_iterators.c)

[PurHur]

Category

stdlib

Problem

RecursiveRegexIterator wraps an Iterator and yields only elements matching a PCRE pattern (used with RecursiveIteratorIterator for directory trees, config scanners, etc.). Not registered in this compiler (class_exists false).

Distinct from #6639 (preg JIT lowering) and #6593 (base SPL iterators). Needs VM class + accept() regex match using existing VmPreg / host PCRE.

php-src reference

• php/php-src ext/spl/spl_iterators.c — spl_ce_RecursiveRegexIterator, SPL_REGEX_* modes
• php/php-src ext/pcre/php_pcre.c — pattern matching for accept()

Repro

./script/docker-exec.sh -- bash -lc 'source script/php-env.sh
php bin/vm.php -r "echo class_exists(\"RecursiveRegexIterator\")?\"yes\":\"no\", PHP_EOL;"
php -r "echo class_exists(\"RecursiveRegexIterator\")?\"yes\":\"no\", PHP_EOL;"
'

<?php
$it = new ArrayIterator(['a1', 'b2', 'a3']);
$rx = new RecursiveRegexIterator($it, '/^a/', RecursiveRegexIterator::MATCH);
foreach ($rx as $v) { echo $v, "\n"; } // Zend: a1, a3

Runtime	class exists	output
Zend	yes	a1/a3 lines
VM	no	—

Scope (this repo)

Area	Files
Class	ext/spl/RecursiveRegexIterator.php + constants (MATCH, GET_MATCH, …)
Register	ext/spl/Module.php, ext/spl/BuiltinClasses.php
VM	Extend filter-iterator base; delegate regex to preg_match parity layer
Tests	test/compliance/cases/stdlib/recursive_regex_iterator.phpt

Done when

• Class registered; repro foreach matches Zend.
• Invalid pattern → Error/ValueError per php-src mode flags.
• Enum case pattern operand → TypeError (php-src-strict), not silent backing coercion.

Related

• Stdlib: SPL EmptyIterator / RecursiveArrayIterator / OuterIterator — iterator graph missing (ext/spl/spl_iterators.c) #6593, Close #6676: bare continue outside loop compile fatal coverage #6684 (SPL / language backlog)
• php-in-php: preg_replace_callback()/preg_split() JIT — delete LLVM stubs, wire real lowering (ext/pcre/php_pcre.c) #6639 (preg JIT — VM may use PHP preg first)

[PurHur]

claim: cursor-agent-lane-a — starting this run

[PurHur]

Merged: #16662

Registered full RecursiveRegexIterator VM class (constructor + hasChildren/getChildren) in ext/spl/RegexIteratorBuiltin.php — stub-only registration left inner iterator uninitialized and fataled on rewind().

Verification:

php bin/vm.php test/repro/maintainer_gap_recursive_regex_iterator.php
# a1,a3

vendor/bin/phpunit --filter recursive_regexiterator_match
# OK (2 tests, 2 assertions)

[PurHur]

Merged inventory follow-up: #16668 — regenerated docs/bootstrap-inventory.md for RecursiveRegexIterator constructs in RegexIteratorBuiltin.php (11034→11042 warning count). php script/bootstrap-inventory.php --check green again.