Backport libxml2 2.13.2 fixes (#14816)

Backproted from https://github.com/php/php-src/pull/14789

ext/dom/document.c

@@ -1292,11 +1292,13 @@ static xmlDocPtr dom_document_parser(zval *id, int mode, char *source, size_t so
 	if (keep_blanks == 0 && ! (options & XML_PARSE_NOBLANKS)) {
 		options |= XML_PARSE_NOBLANKS;
 	}
+	if (recover) {
+		options |= XML_PARSE_RECOVER;
+	}
 
 	php_libxml_sanitize_parse_ctxt_options(ctxt);
 	xmlCtxtUseOptions(ctxt, options);
 
-	ctxt->recovery = recover;
 	if (recover) {
 		old_error_reporting = EG(error_reporting);
 		EG(error_reporting) = old_error_reporting | E_WARNING;
@@ -1306,7 +1308,7 @@ static xmlDocPtr dom_document_parser(zval *id, int mode, char *source, size_t so
 
 	if (ctxt->wellFormed || recover) {
 		ret = ctxt->myDoc;
-		if (ctxt->recovery) {
+		if (recover) {
 			EG(error_reporting) = old_error_reporting;
 		}
 		/* If loading from memory, set the base reference uri for the document */

ext/dom/tests/DOMDocument_loadHTMLfile_error1.phpt

@@ -15,4 +15,4 @@ $result = $doc->loadHTMLFile(__DIR__ . "/ffff/test.html");
 assert($result === false);
 ?>
 --EXPECTF--
-%r(PHP ){0,1}%rWarning: DOMDocument::loadHTMLFile(): I/O warning : failed to load external entity %s
+%r(PHP ){0,1}%rWarning: DOMDocument::loadHTMLFile(): I/O %s

ext/dom/tests/DOMDocument_loadXML_error2_gte2_12.phpt

@@ -23,7 +23,7 @@ domdocumentloadxml_test_method.inc
 --EXPECTF--
 Warning: DOMDocument::loadXML(): AttValue: " or ' expected in Entity, line: 4 in %s on line %d
 
-Warning: DOMDocument::loadXML(): internal error: xmlParseStartTag: problem parsing attributes in Entity, line: 4 in %s on line %d
+Warning: DOMDocument::loadXML():%sattributes%s
 
 Warning: DOMDocument::loadXML(): Couldn't find end of Start Tag book line 4 in Entity, line: 4 in %s on line %d
 

ext/dom/tests/DOMDocument_load_error2_gte2_12.phpt

@@ -23,7 +23,7 @@ domdocumentload_test_method.inc
 --EXPECTF--
 Warning: DOMDocument::load(): AttValue: " or ' expected in %s on line %d
 
-Warning: DOMDocument::load(): internal error: xmlParseStartTag: problem parsing attributes in %s on line %d
+Warning: DOMDocument::load():%sattributes%s
 
 Warning: DOMDocument::load(): Couldn't find end of Start Tag book line 4 in %s on line %d
 

ext/dom/tests/DOMDocument_relaxNGValidate_error2.phpt

@@ -20,7 +20,7 @@ $result = $doc->relaxNGValidate($rng);
 var_dump($result);
 ?>
 --EXPECTF--
-Warning: DOMDocument::relaxNGValidate(): I/O warning : failed to load external entity "%s/foo.rng" in %s on line %d
+Warning: DOMDocument::relaxNGValidate(): I/O %s : failed to load %s
 
 Warning: DOMDocument::relaxNGValidate(): xmlRelaxNGParse: could not load %s/foo.rng in %s on line %d
 

ext/dom/tests/DOMDocument_saveHTMLFile_basic.phpt

@@ -5,6 +5,10 @@ Knut Urdalen <knut@php.net>
 #PHPTestFest2009 Norway 2009-06-09 \o/
 --EXTENSIONS--
 dom
+--SKIPIF--
+<?php
+if (LIBXML_VERSION >= 21300) die("skip see https://gitlab.gnome.org/GNOME/libxml2/-/issues/756");
+?>
 --FILE--
 <?php
 $filename = __DIR__."/DOMDocument_saveHTMLFile_basic.html";

ext/dom/tests/DOMDocument_saveHTMLFile_formatOutput.phpt

@@ -5,6 +5,10 @@ Knut Urdalen <knut@php.net>
 #PHPTestFest2009 Norway 2009-06-09 \o/
 --EXTENSIONS--
 dom
+--SKIPIF--
+<?php
+if (LIBXML_VERSION >= 21300) die("skip see https://gitlab.gnome.org/GNOME/libxml2/-/issues/756");
+?>
 --FILE--
 <?php
 $filename = __DIR__."/DOMDocument_saveHTMLFile_formatOutput.html";

ext/dom/tests/DOMDocument_saveHTMLFile_formatOutput_gte_2_13.phpt

@@ -0,0 +1,32 @@
+--TEST--
+DOMDocument::saveHTMLFile() should format output on demand
+--CREDITS--
+Knut Urdalen <knut@php.net>
+#PHPTestFest2009 Norway 2009-06-09 \o/
+--EXTENSIONS--
+dom
+--SKIPIF--
+<?php
+if (LIBXML_VERSION < 21300) die("skip see https://gitlab.gnome.org/GNOME/libxml2/-/issues/756");
+?>
+--FILE--
+<?php
+$filename = __DIR__."/DOMDocument_saveHTMLFile_formatOutput_gte_2_13.html";
+$doc = new DOMDocument('1.0');
+$doc->formatOutput = true;
+$root = $doc->createElement('html');
+$root = $doc->appendChild($root);
+$head = $doc->createElement('head');
+$head = $root->appendChild($head);
+$title = $doc->createElement('title');
+$title = $head->appendChild($title);
+$text = $doc->createTextNode('This is the title');
+$text = $title->appendChild($text);
+$bytes = $doc->saveHTMLFile($filename);
+var_dump($bytes);
+echo file_get_contents($filename);
+unlink($filename);
+?>
+--EXPECT--
+int(59)
+<html><head><title>This is the title</title></head></html>

ext/dom/tests/DOMDocument_saveHTML_basic_gte_2_13.phpt

@@ -0,0 +1,31 @@
+--TEST--
+DOMDocument::saveHTMLFile() should dump the internal document into a file using HTML formatting
+--CREDITS--
+Knut Urdalen <knut@php.net>
+#PHPTestFest2009 Norway 2009-06-09 \o/
+--EXTENSIONS--
+dom
+--SKIPIF--
+<?php
+if (LIBXML_VERSION < 21300) die("skip see https://gitlab.gnome.org/GNOME/libxml2/-/issues/756");
+?>
+--FILE--
+<?php
+$filename = __DIR__."/DOMDocument_saveHTMLFile_basic_gte_2_13.html";
+$doc = new DOMDocument('1.0');
+$root = $doc->createElement('html');
+$root = $doc->appendChild($root);
+$head = $doc->createElement('head');
+$head = $root->appendChild($head);
+$title = $doc->createElement('title');
+$title = $head->appendChild($title);
+$text = $doc->createTextNode('This is the title');
+$text = $title->appendChild($text);
+$bytes = $doc->saveHTMLFile($filename);
+var_dump($bytes);
+echo file_get_contents($filename);
+unlink($filename);
+?>
+--EXPECT--
+int(59)
+<html><head><title>This is the title</title></head></html>

ext/dom/tests/DOMDocument_schemaValidate_error5.phpt

@@ -17,7 +17,7 @@ var_dump($result);
 
 ?>
 --EXPECTF--
-Warning: DOMDocument::schemaValidate(): I/O warning : failed to load external entity "%snon-existent-file" in %s.php on line %d
+Warning: DOMDocument::schemaValidate(): I/O %s : failed to load %s
 
 Warning: DOMDocument::schemaValidate(): Failed to locate the main schema resource at '%s/non-existent-file'. in %s.php on line %d
 

ext/dom/tests/dom_create_element.phpt

@@ -251,14 +251,10 @@ try {
     print $e->getMessage() . "\n";
 }
 
-/* This isn't because the xml namespace isn't there and we can't create it */
+/* There used to be a 29 here that tested DOMElement::__construct('xml:valid', '', 'http://www.w3.org/XML/1998/namespace').
-print "29 DOMElement::__construct('xml:valid', '', 'http://www.w3.org/XML/1998/namespace')\n";
+ * In libxml2 version 2.12 or prior this didn't work because the xml namespace isn't there and you can't create it without
-try {
+ * a document. Starting from libxml2 version 2.13 it does actually work because the XML namespace is statically defined.
-    $element = new DomElement('xml:valid', '', 'http://www.w3.org/XML/1998/namespace');
+ * The behaviour from version 2.13 is actually the desired behaviour anyway. */
-    print "valid\n";
-} catch (Exception $e) {
-    print $e->getMessage() . "\n";
-}
 
 
 /* the qualifiedName or its prefix is "xmlns" and the  namespaceURI is
@@ -378,8 +374,6 @@ Namespace Error
 Namespace Error
 28 DOMDocument::createElementNS('http://www.w3.org/XML/1998/namespace', 'xml:valid')
 valid
-29 DOMElement::__construct('xml:valid', '', 'http://www.w3.org/XML/1998/namespace')
-Namespace Error
 30 DOMDocument::createElementNS('http://wrong.namespaceURI.com', 'xmlns:valid')
 Namespace Error
 31 DOMElement::__construct('xmlns:valid', '', 'http://wrong.namespaceURI.com')

ext/libxml/libxml.c

@@ -419,8 +419,10 @@ php_libxml_input_buffer_create_filename(const char *URI, xmlCharEncoding enc)
 static xmlOutputBufferPtr
 php_libxml_output_buffer_create_filename(const char *URI,
                               xmlCharEncodingHandlerPtr encoder,
-                              int compression ATTRIBUTE_UNUSED)
+                              int compression)
 {
+	ZEND_IGNORE_VALUE(compression);
+
 	xmlOutputBufferPtr ret;
 	xmlURIPtr puri;
 	void *context = NULL;

ext/libxml/php_libxml.h

@@ -167,6 +167,7 @@ ZEND_TSRMLS_CACHE_EXTERN()
  * Generally faster because no locking is involved, and this has the advantage that it sets the options to a known good value. */
 static zend_always_inline void php_libxml_sanitize_parse_ctxt_options(xmlParserCtxtPtr ctxt)
 {
+	PHP_LIBXML_IGNORE_DEPRECATIONS_START
 	ctxt->loadsubset = 0;
 	ctxt->validate = 0;
 	ctxt->pedantic = 0;
@@ -174,6 +175,7 @@ static zend_always_inline void php_libxml_sanitize_parse_ctxt_options(xmlParserC
 	ctxt->linenumbers = 0;
 	ctxt->keepBlanks = 1;
 	ctxt->options = 0;
+	PHP_LIBXML_IGNORE_DEPRECATIONS_END
 }
 
 #else /* HAVE_LIBXML */

ext/simplexml/tests/bug79971_1.phpt

@@ -20,7 +20,7 @@ var_dump($sxe->asXML("$uri.out%00foo"));
 --EXPECTF--
 Warning: simplexml_load_file(): URI must not contain percent-encoded NUL bytes in %s on line %d
 
-Warning: simplexml_load_file(): I/O warning : failed to load external entity "%s/bug79971_1.xml%%r00%rfoo" in %s on line %d
+Warning: simplexml_load_file(): I/O warning : failed to load %s
 bool(false)
 
 Warning: SimpleXMLElement::asXML(): URI must not contain percent-encoded NUL bytes in %s on line %d

ext/soap/php_encoding.c

@@ -3374,7 +3374,6 @@ xmlNsPtr encode_add_ns(xmlNodePtr node, const char* ns)
 		} else {
 			smart_str prefix = {0};
 			int num = ++SOAP_GLOBAL(cur_uniq_ns);
-			xmlChar *enc_ns;
 
 			while (1) {
 				smart_str_appendl(&prefix, "ns", 2);
@@ -3388,9 +3387,15 @@ xmlNsPtr encode_add_ns(xmlNodePtr node, const char* ns)
 				num = ++SOAP_GLOBAL(cur_uniq_ns);
 			}
 
-			enc_ns = xmlEncodeSpecialChars(node->doc, BAD_CAST(ns));
+			/* Starting with libxml 2.13, we don't have to do this workaround anymore, otherwise we get double-encoded
+			 * entities. See libxml2 commit f506ec66547ef9bac97a2bf306d368ecea8c0c9e. */
+#if LIBXML_VERSION < 21300
+			xmlChar *enc_ns = xmlEncodeSpecialChars(node->doc, BAD_CAST(ns));
 			xmlns = xmlNewNs(node->doc->children, enc_ns, BAD_CAST(prefix.s ? ZSTR_VAL(prefix.s) : ""));
 			xmlFree(enc_ns);
+#else
+			xmlns = xmlNewNs(node->doc->children, BAD_CAST(ns), BAD_CAST(prefix.s ? ZSTR_VAL(prefix.s) : ""));
+#endif
 			smart_str_free(&prefix);
 		}
 	}

ext/soap/php_xml.c

@@ -92,13 +92,16 @@ xmlDocPtr soap_xmlParseFile(const char *filename)
 		bool old;
 
 		php_libxml_sanitize_parse_ctxt_options(ctxt);
+		/* TODO: In libxml2 2.14.0 change this to the new options API so we don't rely on deprecated APIs. */
+		PHP_LIBXML_IGNORE_DEPRECATIONS_START
 		ctxt->keepBlanks = 0;
+		ctxt->options |= XML_PARSE_HUGE;
+		PHP_LIBXML_IGNORE_DEPRECATIONS_END
 		ctxt->sax->ignorableWhitespace = soap_ignorableWhitespace;
 		ctxt->sax->comment = soap_Comment;
 		ctxt->sax->warning = NULL;
 		ctxt->sax->error = NULL;
 		/*ctxt->sax->fatalError = NULL;*/
-		ctxt->options |= XML_PARSE_HUGE;
 		old = php_libxml_disable_entity_loader(1);
 		xmlParseDocument(ctxt);
 		php_libxml_disable_entity_loader(old);
@@ -146,7 +149,10 @@ xmlDocPtr soap_xmlParseMemory(const void *buf, size_t buf_size)
 		ctxt->sax->warning = NULL;
 		ctxt->sax->error = NULL;
 		/*ctxt->sax->fatalError = NULL;*/
+		/* TODO: In libxml2 2.14.0 change this to the new options API so we don't rely on deprecated APIs. */
+		PHP_LIBXML_IGNORE_DEPRECATIONS_START
 		ctxt->options |= XML_PARSE_HUGE;
+		PHP_LIBXML_IGNORE_DEPRECATIONS_END
 		old = php_libxml_disable_entity_loader(1);
 		xmlParseDocument(ctxt);
 		php_libxml_disable_entity_loader(old);

ext/soap/tests/bugs/bug42151.phpt

@@ -25,8 +25,8 @@ try {
 }
 echo "ok\n";
 ?>
---EXPECT--
+--EXPECTF--
-SOAP-ERROR: Parsing WSDL: Couldn't load from 'httpx://' : failed to load external entity "httpx://"
+SOAP-ERROR: Parsing WSDL: Couldn't load from 'httpx://' : failed to load %s
 
 ok
 I don't get executed either.

ext/xml/compat.c

@@ -714,8 +714,7 @@ XML_GetCurrentByteCount(XML_Parser parser)
 {
 	/* WARNING: this is identical to ByteIndex; it should probably
 	 * be different */
-	return parser->parser->input->consumed +
+	return XML_GetCurrentByteIndex(parser);
-			(parser->parser->input->cur - parser->parser->input->base);
 }
 
 PHP_XML_API const XML_Char *XML_ExpatVersion(void)

ext/xmlwriter/php_xmlwriter.c

@@ -1004,7 +1004,8 @@ static void php_xmlwriter_flush(INTERNAL_FUNCTION_PARAMETERS, int force_string)
 		}
 		output_bytes = xmlTextWriterFlush(ptr);
 		if (buffer) {
-			RETVAL_STRING((char *) buffer->content);
+			const xmlChar *content = xmlBufferContent(buffer);
+			RETVAL_STRING((const char *) content);
 			if (empty) {
 				xmlBufferEmpty(buffer);
 			}