From 8dc799aac7b3c5f7be639c0f15d3f294a7b194a7 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Fri, 5 Jul 2024 20:11:34 +0200
Subject: [PATCH 1/5] Port XML_GetCurrentByteIndex to public APIs

This is necessary to avoid a deprecation break in libxml2 2.14.x.
---
 ext/xml/compat.c | 17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)

diff --git a/ext/xml/compat.c b/ext/xml/compat.c
index 7b463ebb511..bc46f001c05 100644
--- a/ext/xml/compat.c
+++ b/ext/xml/compat.c
@@ -705,8 +705,21 @@ XML_GetCurrentColumnNumber(XML_Parser parser)
 PHP_XML_API int
 XML_GetCurrentByteIndex(XML_Parser parser)
 {
-	return parser->parser->input->consumed +
-			(parser->parser->input->cur - parser->parser->input->base);
+	/* We have to temporarily disable the encoder to satisfy the note from the manual:
+	 * "This function returns byte index according to UTF-8 encoded text disregarding if input is in another encoding."
+	 * Although that should probably be corrected at one point? (TODO) */
+	xmlCharEncodingHandlerPtr encoder = NULL;
+	xmlParserInputPtr input = parser->parser->input;
+	if (input->buf) {
+		encoder = input->buf->encoder;
+		input->buf->encoder = NULL;
+	}
+	long result = xmlByteConsumed(parser->parser);
+	if (encoder) {
+		input->buf->encoder = encoder;
+	}
+	/* TODO: at one point this should return long probably to make sure that files greater than 2 GiB are handled correctly. */
+	return (int) result;
 }
 
 PHP_XML_API int

From 6490a1827100c0b5da01bf9de915921ea586a281 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Fri, 5 Jul 2024 20:32:09 +0200
Subject: [PATCH 2/5] Stop relying on the sax2 flag directly

Setting this directly will become deprecated in libxml2 2.14.
---
 ext/xml/compat.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/ext/xml/compat.c b/ext/xml/compat.c
index bc46f001c05..910063f6def 100644
--- a/ext/xml/compat.c
+++ b/ext/xml/compat.c
@@ -476,8 +476,9 @@ XML_ParserCreate_MM(const XML_Char *encoding, const XML_Memory_Handling_Suite *m
 	parser->parser->replaceEntities = 1;
 	parser->parser->wellFormed = 0;
 	if (sep != NULL) {
+		/* Note: sax2 flag will be set due to the magic number in `initialized` in php_xml_compat_handlers */
+		ZEND_ASSERT(parser->parser->sax->initialized == XML_SAX2_MAGIC);
 		parser->use_namespace = 1;
-		parser->parser->sax2 = 1;
 		parser->_ns_separator = xmlStrdup(sep);
 	} else {
 		/* Reset flag as XML_SAX2_MAGIC is needed for xmlCreatePushParserCtxt

From 823d0588f767b392ea74850c4586721b7079b588 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Fri, 5 Jul 2024 20:38:07 +0200
Subject: [PATCH 3/5] Stop relying on lastError directly

Reading this directly will become deprecated in libxml2 2.14.
---
 ext/xml/compat.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/ext/xml/compat.c b/ext/xml/compat.c
index 910063f6def..d8548bf4dc9 100644
--- a/ext/xml/compat.c
+++ b/ext/xml/compat.c
@@ -564,10 +564,14 @@ XML_SetEndNamespaceDeclHandler(XML_Parser parser, XML_EndNamespaceDeclHandler en
 PHP_XML_API int
 XML_Parse(XML_Parser parser, const XML_Char *data, int data_len, int is_final)
 {
-	int error;
+	int error = xmlParseChunk(parser->parser, (char *) data, data_len, is_final);
 
-	error = xmlParseChunk(parser->parser, (char *) data, data_len, is_final);
-	return !error && parser->parser->lastError.level <= XML_ERR_WARNING;
+	if (!error) {
+		const xmlError *error_data = xmlCtxtGetLastError(parser->parser);
+		return !error_data || error_data->level <= XML_ERR_WARNING;
+	}
+
+	return 0;
 }
 
 PHP_XML_API int

From e5e15fd2293e381212786d95790eecbd41630517 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Fri, 5 Jul 2024 20:52:56 +0200
Subject: [PATCH 4/5] Stop setting parse options directly

Setting this directly will be deprecated in libxml2 2.14.
---
 ext/xml/compat.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/ext/xml/compat.c b/ext/xml/compat.c
index d8548bf4dc9..242cc4ba7c4 100644
--- a/ext/xml/compat.c
+++ b/ext/xml/compat.c
@@ -471,9 +471,8 @@ XML_ParserCreate_MM(const XML_Char *encoding, const XML_Memory_Handling_Suite *m
 	}
 
 	php_libxml_sanitize_parse_ctxt_options(parser->parser);
-	xmlCtxtUseOptions(parser->parser, XML_PARSE_OLDSAX);
+	xmlCtxtUseOptions(parser->parser, XML_PARSE_OLDSAX | XML_PARSE_NOENT);
 
-	parser->parser->replaceEntities = 1;
 	parser->parser->wellFormed = 0;
 	if (sep != NULL) {
 		/* Note: sax2 flag will be set due to the magic number in `initialized` in php_xml_compat_handlers */

From a66afbbe09e91fcbd7d65a9b6d03a0af19aeb778 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Sat, 6 Jul 2024 13:46:42 +0200
Subject: [PATCH 5/5] NEWS for compatibility in XML

Closes GH-14836.
---
 NEWS | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/NEWS b/NEWS
index 74c017b3fbf..883b3c162b1 100644
--- a/NEWS
+++ b/NEWS
@@ -46,6 +46,9 @@ PHP                                                                        NEWS
 - Treewide:
   . Fix compatibility with libxml2 2.13.2. (nielsdos)
 
+- XML:
+  . Move away from to-be-deprecated libxml fields. (nielsdos)
+
 04 Jul 2024, PHP 8.2.21
 
 - Core: