Stdlib: html_entity_decode() numeric entities &#N;/&#xN; — not decoded (ext/standard/html.c)

[PurHur]

Summary

html_entity_decode() leaves numeric character references (A, A) unchanged. Zend decodes them to the corresponding UTF-8 character (A). Named-entity gaps are tracked separately (#10763, #10519); this issue is decimal/hex numeric refs only.

php-src reference: ext/standard/html.c (php_unescape_html_entities, numeric entity branch).

PHP implementation target: ext/standard/ HTML helpers (VmHtml / string builtins) — port decode table logic in PHP; no permanent C-only shortcut in runtime/.

Repro

./script/docker-exec.sh -- php test/repro/issue_html_entity_decode_numeric.php          # Zend: A / A
./script/docker-exec.sh -- php bin/vm.php test/repro/issue_html_entity_decode_numeric.php  # VM: A / A

Minimal:

html_entity_decode('A', ENT_QUOTES | ENT_HTML5);   // Zend: 'A'; VM: 'A'
html_entity_decode('A', ENT_QUOTES | ENT_HTML5);  // Zend: 'A'; VM: 'A'

Expected (php-src-strict)

• Decimal &#digits; and hex &#xhex; / &#Xhex; decode to UTF-8 bytes.
• Invalid/out-of-range numeric refs follow Zend (typically leave ref or emit warning per php-src).
• Works with combined ENT_* flag ints when those flags are accepted.

Done when

• test/repro/issue_html_entity_decode_numeric.php green on php bin/vm.php.
• Targeted compliance guard under test/compliance/ or test/repro/.

[PurHur]

claim: PHP Compiler STD Lib — batch (issues #11509, #11510, #11408)

[PurHur]

Merged in PR #11521. VM + AOT lint green: php bin/vm.php test/repro/issue_html_entity_decode_numeric.php → A/A.