Stdlib: htmlspecialchars()/htmlentities() — scalar-to-string coercion for $string (ext/standard/html.c parity)

[PurHur]

Category

stdlib

Problem

htmlspecialchars() and htmlentities() require TYPE_STRING for argument #1 ($string). Zend coerces int/float/bool (and objects with __toString) via zend_parse_parameter / convert_to_string before escaping.

This compiler throws LogicException: ... only supports strings in this compiler build (ext/standard/htmlspecialchars.php, ext/standard/htmlentities.php).

Related: #4296 (closed) fixed null $encoding; this issue is argument #1 coercion only.

php-src reference

• php/php-src ext/standard/html.c — PHP_FUNCTION(htmlspecialchars), PHP_FUNCTION(htmlentities)
• php/php-src Zend/zend_operators.c — scalar convert_to_string

Repro (failure today)

<?php
echo htmlspecialchars(42), "\n";
echo htmlentities(true), "\n";

./script/docker-exec.sh -- bash -lc 'source script/php-env.sh
php -r "echo htmlspecialchars(42), \"\\n\";"
php bin/vm.php -r "echo htmlspecialchars(42), \"\\n\";"
'

Runtime	Output
Zend PHP	42 then 1
bin/vm.php	LogicException: htmlspecialchars() only supports strings...

double_encode bool coercion is separate (existing PHPT covers bool); align flags/encoding with #4130 when touching VmString::htmlspecialchars.

Scope

Module	Path
VM	ext/standard/htmlspecialchars.php, htmlentities.php — use shared scalar→string helper (match nl2br() / preg_quote() parity issues)
Core	ext/standard/VmString.php — htmlspecialchars() / htmlentities() entry
JIT	JitStringArg / JitHtmlspecialchars — coerce non-string args like other string builtins
Tests	test/compliance/cases/stdlib/htmlspecialchars_scalar_coerce.phpt

Done when

• htmlspecialchars(42) and htmlentities(true) match Zend on VM
• Invalid types (array/resource) raise TypeError like Zend, not LogicException
• JIT path coerces scalars when not compile-time literal
• ./script/ci-fast.sh --filter htmlspecialchars_scalar green

Links

• Stdlib: nl2br() — scalar-to-string and bool coercion for $use_xhtml (ext/standard/string.c parity) #4293 nl2br coercion pattern · Stdlib: html_entity_decode() / htmlspecialchars_decode() — ENT_HTML5 named entities #4130 ENT_HTML5 decode · Language: __toString() exceptions during string coercion must reach try/catch (zend_object_handlers.c parity) #4284 __toString during coercion · Epic: Self-host critical path — compiler compiles itself (M3→M5) #1492

[PurHur]

claim: PHP Compiler STD Lib — starting this run

[PurHur]

Unclaiming: htmlspecialchars/htmlentities scalar coercion already green on master (VmString::coerceStringBuiltinArg + JitStringBuiltinArg). VM/JIT repro matches Zend. Recommend closing as duplicate unless repro differs.

[PurHur]

Already fixed — closing (lane C audit 2026-07-18)

Verified on current master — Zend and php bin/vm.php both coerce int/float/bool for htmlspecialchars/htmlentities (e.g. htmlspecialchars(42) → '42'). Matches prior unclaim (VmString::coerceStringBuiltinArg).

No PR needed.