Fix issues related to libxml2 2.12.0 (#12802)

ndossche · web-flow · commit ae83d6ab077d · 2023-11-29T20:46:35.000+01:00
* Avoid passing NULL to xmlSwitchToEncoding This otherwise switches to UTF-8 on libxml2 2.12.0 * Split tests for different error reporting behaviour in libxml2 2.12.0 * Avoid deprecation warnings for libxml2 2.12.0 We can't fully get rid of the parser globals as there are still APIs that implicitly use them. * Temporarily disable part of test for libxml 2.12.0 regression See https://gitlab.gnome.org/GNOME/libxml2/-/issues/634 * Review fixes * [ci skip] Update test description
diff --git a/ext/dom/document.c b/ext/dom/document.c
@@ -1240,7 +1240,10 @@ xmlDocPtr dom_document_parser(zval *id, dom_load_mode mode, const char *source,
 		return(NULL);
 	}
 
-	(void) xmlSwitchToEncoding(ctxt, encoding);
+	if (encoding != NULL) {
+		/* Note: libxml 2.12+ doesn't handle NULL encoding well. */
+		(void) xmlSwitchToEncoding(ctxt, encoding);
+	}
 
 	/* If loading from memory, we need to set the base directory for the document */
 	if (mode != DOM_LOAD_FILE) {
diff --git a/ext/dom/tests/DOMDocument_loadXML_error1.phpt b/ext/dom/tests/DOMDocument_loadXML_error1.phpt
@@ -1,5 +1,9 @@
 --TEST--
 Test DOMDocument::loadXML() detects not-well formed XML
+--SKIPIF--
+<?php
+if (LIBXML_VERSION >= 21200) die('skip libxml2 test variant for version < 2.12');
+?>
 --DESCRIPTION--
 This test verifies the method detects an opening and ending tag mismatch
 Environment variables used in the test:
diff --git a/ext/dom/tests/DOMDocument_loadXML_error1_gte2_12.phpt b/ext/dom/tests/DOMDocument_loadXML_error1_gte2_12.phpt
@@ -0,0 +1,26 @@
+--TEST--
+Test DOMDocument::loadXML() detects not-well formed XML
+--SKIPIF--
+<?php
+if (LIBXML_VERSION < 21200) die('skip libxml2 test variant for version >= 2.12');
+?>
+--DESCRIPTION--
+This test verifies the method detects an opening and ending tag mismatch
+Environment variables used in the test:
+- XML_FILE: the xml file to load
+- LOAD_OPTIONS: the second parameter to pass to the method
+- EXPECTED_RESULT: the expected result
+--CREDITS--
+Antonio Diaz Ruiz <dejalatele@gmail.com>
+--EXTENSIONS--
+dom
+--ENV--
+XML_FILE=/not_well_formed.xml
+LOAD_OPTIONS=0
+EXPECTED_RESULT=0
+--FILE_EXTERNAL--
+domdocumentloadxml_test_method.inc
+--EXPECTF--
+Warning: DOMDocument::load%r(XML){0,1}%r(): Opening and ending tag mismatch: title line 5 and book %s
+
+Warning: DOMDocument::load%r(XML){0,1}%r(): %rexpected '>'|Opening and ending tag mismatch: book line (4|5) and books%r %s
diff --git a/ext/dom/tests/DOMDocument_loadXML_error2_gte2_11.phpt b/ext/dom/tests/DOMDocument_loadXML_error2_gte2_11.phpt
@@ -2,7 +2,7 @@
 Test DOMDocument::loadXML() detects not-well formed XML
 --SKIPIF--
 <?php
-if (LIBXML_VERSION < 21100) die('skip libxml2 test variant for version >= 2.11');
+if (LIBXML_VERSION < 21100 || LIBXML_VERSION >= 21200) die('skip libxml2 test variant for version >= 2.11 && <= 2.12');
 ?>
 --DESCRIPTION--
 This test verifies the method detects attributes values not closed between " or '
diff --git a/ext/dom/tests/DOMDocument_loadXML_error2_gte2_12.phpt b/ext/dom/tests/DOMDocument_loadXML_error2_gte2_12.phpt
@@ -0,0 +1,30 @@
+--TEST--
+Test DOMDocument::loadXML() detects not-well formed XML
+--SKIPIF--
+<?php
+if (LIBXML_VERSION < 21200) die('skip libxml2 test variant for version >= 2.12');
+?>
+--DESCRIPTION--
+This test verifies the method detects attributes values not closed between " or '
+Environment variables used in the test:
+- XML_FILE: the xml file to load
+- LOAD_OPTIONS: the second parameter to pass to the method
+- EXPECTED_RESULT: the expected result
+--CREDITS--
+Antonio Diaz Ruiz <dejalatele@gmail.com>
+--EXTENSIONS--
+dom
+--ENV--
+XML_FILE=/not_well_formed2.xml
+LOAD_OPTIONS=0
+EXPECTED_RESULT=0
+--FILE_EXTERNAL--
+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(): Couldn't find end of Start Tag book line 4 in Entity, line: 4 in %s on line %d
+
+Warning: DOMDocument::loadXML(): Opening and ending tag mismatch: books line 3 and book in Entity, line: 7 in %s on line %d
diff --git a/ext/dom/tests/DOMDocument_load_error1.phpt b/ext/dom/tests/DOMDocument_load_error1.phpt
@@ -1,5 +1,9 @@
 --TEST--
 Test DOMDocument::load() detects not-well formed XML
+--SKIPIF--
+<?php
+if (LIBXML_VERSION >= 21200) die('skip libxml2 test variant for version < 2.12');
+?>
 --DESCRIPTION--
 This test verifies the method detects an opening and ending tag mismatch
 Environment variables used in the test:
diff --git a/ext/dom/tests/DOMDocument_load_error1_gte2_12.phpt b/ext/dom/tests/DOMDocument_load_error1_gte2_12.phpt
@@ -0,0 +1,26 @@
+--TEST--
+Test DOMDocument::load() detects not-well formed XML
+--SKIPIF--
+<?php
+if (LIBXML_VERSION < 21200) die('skip libxml2 test variant for version >= 2.12');
+?>
+--DESCRIPTION--
+This test verifies the method detects an opening and ending tag mismatch
+Environment variables used in the test:
+- XML_FILE: the xml file to load
+- LOAD_OPTIONS: the second parameter to pass to the method
+- EXPECTED_RESULT: the expected result
+--CREDITS--
+Antonio Diaz Ruiz <dejalatele@gmail.com>
+--EXTENSIONS--
+dom
+--ENV--
+XML_FILE=/not_well_formed.xml
+LOAD_OPTIONS=0
+EXPECTED_RESULT=0
+--FILE_EXTERNAL--
+domdocumentload_test_method.inc
+--EXPECTF--
+Warning: DOMDocument::load%r(XML){0,1}%r(): Opening and ending tag mismatch: title line 5 and book %s
+
+Warning: DOMDocument::load%r(XML){0,1}%r(): %rexpected '>'|Opening and ending tag mismatch: book line (4|5) and books%r %s
diff --git a/ext/dom/tests/DOMDocument_load_error2_gte2_11.phpt b/ext/dom/tests/DOMDocument_load_error2_gte2_11.phpt
@@ -2,7 +2,7 @@
 Test DOMDocument::load() detects not-well formed
 --SKIPIF--
 <?php
-if (LIBXML_VERSION < 21100) die('skip libxml2 test variant for version >= 2.11');
+if (LIBXML_VERSION < 21100 || LIBXML_VERSION >= 21200) die('skip libxml2 test variant for version >= 2.11 && <= 2.12');
 ?>
 --DESCRIPTION--
 This test verifies the method detects attributes values not closed between " or '
diff --git a/ext/dom/tests/DOMDocument_load_error2_gte2_12.phpt b/ext/dom/tests/DOMDocument_load_error2_gte2_12.phpt
@@ -0,0 +1,30 @@
+--TEST--
+Test DOMDocument::load() detects not-well formed
+--SKIPIF--
+<?php
+if (LIBXML_VERSION < 21200) die('skip libxml2 test variant for version >= 2.12');
+?>
+--DESCRIPTION--
+This test verifies the method detects attributes values not closed between " or '
+Environment variables used in the test:
+- XML_FILE: the xml file to load
+- LOAD_OPTIONS: the second parameter to pass to the method
+- EXPECTED_RESULT: the expected result
+--CREDITS--
+Antonio Diaz Ruiz <dejalatele@gmail.com>
+--EXTENSIONS--
+dom
+--ENV--
+XML_FILE=/not_well_formed2.xml
+LOAD_OPTIONS=0
+EXPECTED_RESULT=0
+--FILE_EXTERNAL--
+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(): Couldn't find end of Start Tag book line 4 in %s on line %d
+
+Warning: DOMDocument::load(): Opening and ending tag mismatch: books line 3 and book in %s on line %d
diff --git a/ext/dom/tests/DOMNode_isEqualNode.phpt b/ext/dom/tests/DOMNode_isEqualNode.phpt
@@ -2,6 +2,12 @@
 DOMNode::isEqualNode()
 --EXTENSIONS--
 dom
+--SKIPIF--
+<?php
+if (LIBXML_VERSION >= 21200 && LIBXML_VERSION <= 21201) {
+    die("xfail Broken for libxml2 2.12.0 - 2.12.1 see https://gitlab.gnome.org/GNOME/libxml2/-/issues/634");
+}
+?>
 --FILE--
 <?php
 
diff --git a/ext/libxml/libxml.c b/ext/libxml/libxml.c
@@ -580,7 +580,11 @@ static void _php_libxml_free_error(void *ptr)
 	xmlResetError((xmlErrorPtr) ptr);
 }
 
-static void _php_list_set_error_structure(xmlErrorPtr error, const char *msg, int line, int column)
+#if LIBXML_VERSION >= 21200
+static void _php_list_set_error_structure(const xmlError *error, const char *msg, int line, int column)
+#else
+static void _php_list_set_error_structure(xmlError *error, const char *msg, int line, int column)
+#endif
 {
 	xmlError error_copy;
 	int ret;
@@ -824,7 +828,10 @@ PHP_LIBXML_API void php_libxml_pretend_ctx_error_ex(const char *file, int line,
 			if (!last->file) {
 				last->file = strdup(file);
 			}
+			/* Until there is a replacement */
+			ZEND_DIAGNOSTIC_IGNORED_START("-Wdeprecated-declarations")
 			xmlCopyError(last, &xmlLastError);
+			ZEND_DIAGNOSTIC_IGNORED_END
 		}
 	}
 }
@@ -845,11 +852,13 @@ PHP_LIBXML_API void php_libxml_ctx_warning(void *ctx, const char *msg, ...)
 	va_end(args);
 }
 
+#if LIBXML_VERSION >= 21200
+static void php_libxml_structured_error_handler(void *userData, const xmlError *error)
+#else
 static void php_libxml_structured_error_handler(void *userData, xmlErrorPtr error)
+#endif
 {
 	_php_list_set_error_structure(error, NULL, 0, 0);
-
-	return;
 }
 
 PHP_LIBXML_API void php_libxml_error_handler(void *ctx, const char *msg, ...)
@@ -1077,11 +1086,9 @@ PHP_FUNCTION(libxml_use_internal_errors)
 /* {{{ Retrieve last error from libxml */
 PHP_FUNCTION(libxml_get_last_error)
 {
-	xmlErrorPtr error;
-
 	ZEND_PARSE_PARAMETERS_NONE();
 
-	error = xmlGetLastError();
+	const xmlError* error = xmlGetLastError();
 
 	if (error) {
 		object_init_ex(return_value, libxmlerror_class_entry);
@@ -1108,7 +1115,6 @@ PHP_FUNCTION(libxml_get_last_error)
 /* {{{ Retrieve array of errors */
 PHP_FUNCTION(libxml_get_errors)
 {
-
 	xmlErrorPtr error;
 
 	ZEND_PARSE_PARAMETERS_NONE();
diff --git a/ext/libxml/php_libxml.h b/ext/libxml/php_libxml.h
@@ -160,23 +160,27 @@ ZEND_TSRMLS_CACHE_EXTERN()
  * See libxml2 globals.c and parserInternals.c.
  * The unique_name argument allows multiple sanitizes and restores within the
  * same function, even nested is necessary. */
-#define PHP_LIBXML_SANITIZE_GLOBALS(unique_name) \
+# define PHP_LIBXML_SANITIZE_GLOBALS(unique_name) \
+	ZEND_DIAGNOSTIC_IGNORED_START("-Wdeprecated-declarations") \
 	int xml_old_loadsubset_##unique_name = xmlLoadExtDtdDefaultValue; \
 	xmlLoadExtDtdDefaultValue = 0; \
 	int xml_old_validate_##unique_name = xmlDoValidityCheckingDefaultValue; \
 	xmlDoValidityCheckingDefaultValue = 0; \
 	int xml_old_pedantic_##unique_name = xmlPedanticParserDefault(0); \
 	int xml_old_substitute_##unique_name = xmlSubstituteEntitiesDefault(0); \
 	int xml_old_linenrs_##unique_name = xmlLineNumbersDefault(0); \
-	int xml_old_blanks_##unique_name = xmlKeepBlanksDefault(1);
+	int xml_old_blanks_##unique_name = xmlKeepBlanksDefault(1); \
+	ZEND_DIAGNOSTIC_IGNORED_END
 
-#define PHP_LIBXML_RESTORE_GLOBALS(unique_name) \
+# define PHP_LIBXML_RESTORE_GLOBALS(unique_name) \
+	ZEND_DIAGNOSTIC_IGNORED_START("-Wdeprecated-declarations") \
 	xmlLoadExtDtdDefaultValue = xml_old_loadsubset_##unique_name; \
 	xmlDoValidityCheckingDefaultValue = xml_old_validate_##unique_name; \
 	(void) xmlPedanticParserDefault(xml_old_pedantic_##unique_name); \
 	(void) xmlSubstituteEntitiesDefault(xml_old_substitute_##unique_name); \
 	(void) xmlLineNumbersDefault(xml_old_linenrs_##unique_name); \
-	(void) xmlKeepBlanksDefault(xml_old_blanks_##unique_name);
+	(void) xmlKeepBlanksDefault(xml_old_blanks_##unique_name); \
+	ZEND_DIAGNOSTIC_IGNORED_END
 
 /* Alternative for above, working directly on the context and not setting globals.
  * Generally faster because no locking is involved, and this has the advantage that it sets the options to a known good value. */
diff --git a/ext/xml/tests/bug81351.phpt b/ext/xml/tests/bug81351.phpt
@@ -21,6 +21,6 @@ $code = xml_get_error_code($parser);
 $error = xml_error_string($code);
 echo "xml_parse returned $success, xml_get_error_code = $code, xml_error_string = $error\r\n";
 ?>
---EXPECT--
+--EXPECTF--
 xml_parse returned 1, xml_get_error_code = 0, xml_error_string = No error
-xml_parse returned 0, xml_get_error_code = 5, xml_error_string = Invalid document end
+%rxml_parse returned 0, xml_get_error_code = 5, xml_error_string = Invalid document end|xml_parse returned 0, xml_get_error_code = 77, xml_error_string = Tag not finished%r
diff --git a/ext/xml/tests/xml_error_string_basic.phpt b/ext/xml/tests/xml_error_string_basic.phpt
@@ -21,9 +21,9 @@ foreach ($xmls as $xml) {
     xml_parser_free($xml_parser);
 }
 ?>
---EXPECT--
-int(5)
-string(20) "Invalid document end"
+--EXPECTF--
+int(%r5|77%r)
+string(%d) %r"Invalid document end"|"Tag not finished"%r
 int(47)
 string(35) "Processing Instruction not finished"
 int(57)
