Add LIBXML_RECOVER (#13504)

Setting the recovery option by using a hardcoded value (1) worked
already for SimpleXML. For DOM, a small change is necessary because
otherwise the recover field overwrites the recovery option.

From a quick search on GitHub [1] it looks like this won't clash with
existing PHP code as no one seems to define (or use) a constant with
such a name.

[1] https://github.com/search?q=LIBXML_RECOVER+language%3APHP&type=code&l=PHP

Author: nielsdos

NEWS

@@ -70,6 +70,9 @@ PHP                                                                        NEWS
   . Added LDAP_OPT_X_TLS_PROTOCOL_MAX/LDAP_OPT_X_TLS_PROTOCOL_TLS1_3
     constants. (StephenWall)
 
+- LibXML:
+  . Added LIBXML_RECOVER constant. (nielsdos)
+
 - MBString:
   . Added mb_trim, mb_ltrim and mb_rtrim. (Yuya Hamada)
 

UPGRADING

@@ -484,6 +484,9 @@ PHP 8.4 UPGRADE NOTES
   . LDAP_OPT_X_TLS_PROTOCOL_MAX.
   . LDAP_OPT_X_TLS_PROTOCOL_TLS1_3.
 
+- LibXML:
+  . LIBXML_RECOVER.
+
 - OpenSSL:
   . X509_PURPOSE_OCSP_HELPER.
   . X509_PURPOSE_TIMESTAMP_SIGN.

ext/dom/document.c

@@ -1261,7 +1261,7 @@ xmlDocPtr dom_document_parser(zval *id, dom_load_mode mode, const char *source,
 	resolve_externals = doc_props->resolveexternals;
 	keep_blanks = doc_props->preservewhitespace;
 	substitute_ent = doc_props->substituteentities;
-	recover = doc_props->recover;
+	recover = doc_props->recover || (options & XML_PARSE_RECOVER) == XML_PARSE_RECOVER;
 
 	xmlInitParser();
 

ext/dom/tests/modern/xml/XMLDocument_fromString_03.phpt

@@ -6,7 +6,7 @@ dom
 <?php
 
 $flags = [
-    LIBXML_NOENT, LIBXML_DTDLOAD, LIBXML_DTDATTR, LIBXML_DTDVALID, LIBXML_NOERROR, LIBXML_NOWARNING, LIBXML_NOBLANKS, LIBXML_XINCLUDE, LIBXML_NSCLEAN, LIBXML_NOCDATA, LIBXML_NONET, LIBXML_PEDANTIC, LIBXML_COMPACT, LIBXML_PARSEHUGE, LIBXML_BIGLINES
+    LIBXML_RECOVER, LIBXML_NOENT, LIBXML_DTDLOAD, LIBXML_DTDATTR, LIBXML_DTDVALID, LIBXML_NOERROR, LIBXML_NOWARNING, LIBXML_NOBLANKS, LIBXML_XINCLUDE, LIBXML_NSCLEAN, LIBXML_NOCDATA, LIBXML_NONET, LIBXML_PEDANTIC, LIBXML_COMPACT, LIBXML_PARSEHUGE, LIBXML_BIGLINES
 ];
 
 try {
@@ -21,7 +21,8 @@ foreach ($flags as $flag) {
 
 ?>
 --EXPECT--
-DOM\XMLDocument::createFromString(): Argument #2 ($options) contains invalid flags (allowed flags: LIBXML_NOENT, LIBXML_DTDLOAD, LIBXML_DTDATTR, LIBXML_DTDVALID, LIBXML_NOERROR, LIBXML_NOWARNING, LIBXML_NOBLANKS, LIBXML_XINCLUDE, LIBXML_NSCLEAN, LIBXML_NOCDATA, LIBXML_NONET, LIBXML_PEDANTIC, LIBXML_COMPACT, LIBXML_PARSEHUGE, LIBXML_BIGLINES)bool(true)
+DOM\XMLDocument::createFromString(): Argument #2 ($options) contains invalid flags (allowed flags: LIBXML_RECOVER, LIBXML_NOENT, LIBXML_DTDLOAD, LIBXML_DTDATTR, LIBXML_DTDVALID, LIBXML_NOERROR, LIBXML_NOWARNING, LIBXML_NOBLANKS, LIBXML_XINCLUDE, LIBXML_NSCLEAN, LIBXML_NOCDATA, LIBXML_NONET, LIBXML_PEDANTIC, LIBXML_COMPACT, LIBXML_PARSEHUGE, LIBXML_BIGLINES)bool(true)
+bool(true)
 bool(true)
 bool(true)
 bool(true)

ext/dom/tests/xml_parsing_LIBXML_RECOVER.phpt

@@ -0,0 +1,23 @@
+--TEST--
+XML parsing with LIBXML_RECOVER
+--EXTENSIONS--
+dom
+--FILE--
+<?php
+
+$dom = new DOMDocument;
+$dom->loadXML('<root><child/>', options: LIBXML_RECOVER);
+echo $dom->saveXML();
+
+$dom = DOM\XMLDocument::createFromString('<root><child/>', options: LIBXML_RECOVER);
+echo $dom->saveXML(), "\n";
+
+?>
+--EXPECTF--
+Warning: DOMDocument::loadXML(): %s
+<?xml version="1.0"?>
+<root><child/></root>
+
+Warning: DOM\XMLDocument::createFromString(): %s
+<?xml version="1.0" encoding="UTF-8"?>
+<root><child/></root>

ext/dom/xml_document.c

@@ -25,7 +25,8 @@
 
 static bool check_options_validity(uint32_t arg_num, zend_long options)
 {
-	const zend_long VALID_OPTIONS = XML_PARSE_NOENT
+	const zend_long VALID_OPTIONS = XML_PARSE_RECOVER
+								  | XML_PARSE_NOENT
 								  | XML_PARSE_DTDLOAD
 								  | XML_PARSE_DTDATTR
 								  | XML_PARSE_DTDVALID
@@ -42,6 +43,7 @@ static bool check_options_validity(uint32_t arg_num, zend_long options)
 								  | XML_PARSE_BIG_LINES;
 	if ((options & ~VALID_OPTIONS) != 0) {
 		zend_argument_value_error(2, "contains invalid flags (allowed flags: "
+									 "LIBXML_RECOVER, "
 									 "LIBXML_NOENT, "
 									 "LIBXML_DTDLOAD, "
 									 "LIBXML_DTDATTR, "

ext/libxml/libxml.stub.php

@@ -18,6 +18,11 @@
  */
 const LIBXML_LOADED_VERSION = UNKNOWN;
 
+/**
+ * @var int
+ * @cvalue XML_PARSE_RECOVER
+ */
+const LIBXML_RECOVER = UNKNOWN;
 /**
  * @var int
  * @cvalue XML_PARSE_NOENT

ext/libxml/libxml_arginfo.h

@@ -0,0 +1,21 @@
+--TEST--
+XML parsing with LIBXML_RECOVER
+--EXTENSIONS--
+simplexml
+--FILE--
+<?php
+
+var_dump(simplexml_load_string('<root><child/>', options: LIBXML_RECOVER));
+
+?>
+--EXPECTF--
+Warning: simplexml_load_string(): %s
+
+Warning: simplexml_load_string(): <root><child/> in %s on line %d
+
+Warning: simplexml_load_string():               ^ in %s on line %d
+object(SimpleXMLElement)#1 (1) {
+  ["child"]=>
+  object(SimpleXMLElement)#2 (0) {
+  }
+}