Backport 4fe82131: Backport libxml2 2.13.2 fixes (#14816)

Backproted from php/php-src#14789

(cherry picked from commit bb46b4b799b583528025a775af45308133bfd4c1)

Author: nielsdos

ext/dom/document.c

@@ -1253,11 +1253,13 @@ 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;
@@ -1267,7 +1269,7 @@ 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_pre2_11.phpt

@@ -2,6 +2,7 @@
 Test DOMDocument::loadXML() detects not-well formed XML
 --SKIPIF--
 <?php
+include('skipif.inc');
 if (LIBXML_VERSION >= 21100) die('skip libxml2 test variant for version < 2.11');
 ?>
 --DESCRIPTION--
@@ -14,8 +15,6 @@ Environment variables used in the test:
 Antonio Diaz Ruiz <[email protected]>
 --INI--
 assert.bail=true
---SKIPIF--
-<?php include('skipif.inc'); ?>
 --ENV--
 XML_FILE=/not_well_formed2.xml
 LOAD_OPTIONS=0

ext/dom/tests/DOMDocument_load_error2_pre2_11.phpt

@@ -2,6 +2,7 @@
 Test DOMDocument::load() detects not-well formed XML
 --SKIPIF--
 <?php
+include('skipif.inc');
 if (LIBXML_VERSION >= 21100) die('skip libxml2 test variant for version < 2.11');
 ?>
 --DESCRIPTION--
@@ -14,8 +15,6 @@ Environment variables used in the test:
 Antonio Diaz Ruiz <[email protected]>
 --INI--
 assert.bail=true
---SKIPIF--
-<?php include('skipif.inc'); ?>
 --ENV--
 XML_FILE=/not_well_formed2.xml
 LOAD_OPTIONS=0

ext/dom/tests/DOMDocument_relaxNGValidate_error2.phpt

@@ -22,7 +22,7 @@ $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

@@ -6,6 +6,7 @@ Knut Urdalen <[email protected]>
 --SKIPIF--
 <?php
 require_once __DIR__ .'/skipif.inc';
+if (LIBXML_VERSION >= 21300) die("skip see https://gitlab.gnome.org/GNOME/libxml2/-/issues/756");
 ?>
 --FILE--
 <?php

ext/dom/tests/DOMDocument_saveHTMLFile_formatOutput.phpt

@@ -6,6 +6,7 @@ Knut Urdalen <[email protected]>
 --SKIPIF--
 <?php
 require_once __DIR__ .'/skipif.inc';
+if (LIBXML_VERSION >= 21300) die("skip see https://gitlab.gnome.org/GNOME/libxml2/-/issues/756");
 ?>
 --FILE--
 <?php

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 <[email protected]>
+#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 <[email protected]>
+#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 */
-print "29 DOMElement::__construct('xml:valid', '', 'http://www.w3.org/XML/1998/namespace')\n";
-try {
-    $element = new DomElement('xml:valid', '', 'http://www.w3.org/XML/1998/namespace');
-    print "valid\n";
-} catch (Exception $e) {
-    print $e->getMessage() . "\n";
-}
+/* There used to be a 29 here that tested DOMElement::__construct('xml:valid', '', 'http://www.w3.org/XML/1998/namespace').
+ * 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
+ * a document. Starting from libxml2 version 2.13 it does actually work because the XML namespace is statically defined.
+ * The behaviour from version 2.13 is actually the desired behaviour anyway. */
 
 
 /* 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

@@ -428,8 +428,10 @@ 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/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%00foo" 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

@@ -3379,7 +3379,6 @@ 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);
@@ -3393,9 +3392,15 @@ 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 +92,14 @@ xmlDocPtr soap_xmlParseFile(const char *filename)
 		zend_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. */
 		ctxt->keepBlanks = 0;
+		ctxt->options |= XML_PARSE_HUGE;
 		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);
@@ -146,6 +147,7 @@ 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. */
 		ctxt->options |= XML_PARSE_HUGE;
 		old = php_libxml_disable_entity_loader(1);
 		xmlParseDocument(ctxt);

ext/soap/tests/bugs/bug42151.phpt

@@ -25,8 +25,8 @@ try {
 }
 echo "ok\n";
 ?>
---EXPECT--
-SOAP-ERROR: Parsing WSDL: Couldn't load from 'httpx://' : failed to load external entity "httpx://"
+--EXPECTF--
+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 +
-			(parser->parser->input->cur - parser->parser->input->base);
+	return XML_GetCurrentByteIndex(parser);
 }
 
 PHP_XML_API const XML_Char *XML_ExpatVersion(void)

ext/xmlwriter/php_xmlwriter.c

@@ -1001,7 +1001,8 @@ 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);
 			}