fix for bug #29657 xml_* functions throw non descriptive error, compared to php4

fix for bug #29711 libxml and non iso-8859-1.
If no encoding specified, it defaults to UTF-8 now

2 files changed, 66 insertions, 66 deletions

diff --git a/ext/xml/compat.c b/ext/xml/compat.c
index 396393ee51..a5797ad37d 100644
--- a/ext/xml/compat.c
+++ b/ext/xml/compat.c
@@ -463,9 +463,9 @@ const XML_Char *error_mapping[] = {
 	"No error",
 	"Internal error",
 	"No memory",
-	"XML_ERR_DOCUMENT_START",
+	"Invalid document start",
 	"Empty document",
-	"XML_ERR_DOCUMENT_END",
+	"Invalid document end",
 	"Invalid hexadecimal character reference",
 	"Invalid decimal character reference",
 	"Invalid character reference",
@@ -478,14 +478,14 @@ const XML_Char *error_mapping[] = {
 	"XML_ERR_ENTITYREF_IN_PROLOG",
 	"XML_ERR_ENTITYREF_IN_EPILOG",
 	"XML_ERR_ENTITYREF_IN_DTD",
-	"XML_ERR_PEREF_AT_EOF",
-	"XML_ERR_PEREF_IN_PROLOG",
-	"XML_ERR_PEREF_IN_EPILOG",
-	"XML_ERR_PEREF_IN_INT_SUBSET",
+	"PEReference at end of document",
+	"PEReference in prolog",
+	"PEReference in epilog",
+	"PEReference: forbidden within markup decl in internal subset",
 	"XML_ERR_ENTITYREF_NO_NAME",
-	"XML_ERR_ENTITYREF_SEMICOL_MISSING",
-	"XML_ERR_PEREF_NO_NAME",
-	"XML_ERR_PEREF_SEMICOL_MISSING",
+	"EntityRef: expecting ';'",
+	"PEReference: no name",
+	"PEReference: expecting ';'",
 	"Undeclared entity error",
 	"Undeclared entity warning",
 	"Unparsed Entity",
@@ -493,82 +493,82 @@ const XML_Char *error_mapping[] = {
 	"XML_ERR_ENTITY_IS_PARAMETER",
 	"Unknown encoding",
 	"Unsupported encoding",
-	"XML_ERR_STRING_NOT_STARTED",
-	"XML_ERR_STRING_NOT_CLOSED",
+	"String not started expecting ' or \"",
+	"String not closed expecting \" or '",
 	"Namespace declaration error",
-	"XML_ERR_ENTITY_NOT_STARTED",
-	"XML_ERR_ENTITY_NOT_FINISHED",
-	"XML_ERR_LT_IN_ATTRIBUTE",
-	"XML_ERR_ATTRIBUTE_NOT_STARTED",
-	"XML_ERR_ATTRIBUTE_NOT_FINISHED",
-	"XML_ERR_ATTRIBUTE_WITHOUT_VALUE",
-	"XML_ERR_ATTRIBUTE_REDEFINED",
-	"XML_ERR_LITERAL_NOT_STARTED",
-	"XML_ERR_LITERAL_NOT_FINISHED",
+	"EntityValue: \" or ' expected",
+	"EntityValue: \" or ' expected",
+	"< in attribute",
+	"Attribute not started",
+	"Attribute not finished",
+	"Attribute without value",
+	"Attribute redefined",
+	"SystemLiteral \" or ' expected",
+	"SystemLiteral \" or ' expected",
 	/* "XML_ERR_COMMENT_NOT_STARTED", <= eliminated on purpose */
-	"XML_ERR_COMMENT_NOT_FINISHED",
-	"XML_ERR_PI_NOT_STARTED",
-	"XML_ERR_PI_NOT_FINISHED",
-	"XML_ERR_NOTATION_NOT_STARTED",
-	"XML_ERR_NOTATION_NOT_FINISHED",
-	"XML_ERR_ATTLIST_NOT_STARTED",
-	"XML_ERR_ATTLIST_NOT_FINISHED",
-	"XML_ERR_MIXED_NOT_STARTED",
+	"Comment not finished",
+	"Processing Instruction not started",
+	"Processing Instruction not finished",
+	"NOTATION: Name expected here",
+	"'>' required to close NOTATION declaration",
+	"'(' required to start ATTLIST enumeration",
+	"'(' required to start ATTLIST enumeration",
+	"MixedContentDecl : '|' or ')*' expected",
 	"XML_ERR_MIXED_NOT_FINISHED",
-	"XML_ERR_ELEMCONTENT_NOT_STARTED",
-	"XML_ERR_ELEMCONTENT_NOT_FINISHED",
-	"XML_ERR_XMLDECL_NOT_STARTED",
-	"XML_ERR_XMLDECL_NOT_FINISHED",
+	"ELEMENT in DTD not started",
+	"ELEMENT in DTD not finished",
+	"XML declaration not started",
+	"XML declaration not finished",
 	"XML_ERR_CONDSEC_NOT_STARTED",
-	"XML_ERR_CONDSEC_NOT_FINISHED",
-	"XML_ERR_EXT_SUBSET_NOT_FINISHED",
-	"XML_ERR_DOCTYPE_NOT_FINISHED",
-	"XML_ERR_MISPLACED_CDATA_END",
-	"XML_ERR_CDATA_NOT_FINISHED",
-	"XML_ERR_RESERVED_XML_NAME",
-	"XML_ERR_SPACE_REQUIRED",
+	"XML conditional section not closed",
+	"Content error in the external subset",
+	"DOCTYPE not finished",
+	"Sequence ']]>' not allowed in content",
+	"CDATA not finished",
+	"Reserved XML Name",
+	"Space required",
 	"XML_ERR_SEPARATOR_REQUIRED",
-	"XML_ERR_NMTOKEN_REQUIRED",
+	"NmToken expected in ATTLIST enumeration",
 	"XML_ERR_NAME_REQUIRED",
-	"XML_ERR_PCDATA_REQUIRED",
-	"XML_ERR_URI_REQUIRED",
-	"XML_ERR_PUBID_REQUIRED",
-	"XML_ERR_LT_REQUIRED",
-	"XML_ERR_GT_REQUIRED",
-	"XML_ERR_LTSLASH_REQUIRED",
-	"XML_ERR_EQUAL_REQUIRED",
-	"XML_ERR_TAG_NAME_MISMATCH",
-	"XML_ERR_TAG_NOT_FINISHED",
-	"XML_ERR_STANDALONE_VALUE",
-	"XML_ERR_ENCODING_NAME",
-	"XML_ERR_HYPHEN_IN_COMMENT",
+	"MixedContentDecl : '#PCDATA' expected",
+	"SYSTEM or PUBLIC, the URI is missing",
+	"PUBLIC, the Public Identifier is missing",
+	"< required",
+	"> required",
+	"</ required",
+	"= required",
+	"Mismatched tag",
+	"Tag not finished",
+	"standalone accepts only 'yes' or 'no'",
+	"Invalid XML encoding name",
+	"Comment must not contain '--' (double-hyphen)",
 	"Invalid encoding",
-	"XML_ERR_EXT_ENTITY_STANDALONE",
-	"XML_ERR_CONDSEC_INVALID",
-	"XML_ERR_VALUE_REQUIRED",
-	"XML_ERR_NOT_WELL_BALANCED",
-	"XML_ERR_EXTRA_CONTENT",
+	"external parsed entities cannot be standalone",
+	"XML conditional section '[' expected",
+	"Entity value required",
+	"chunk is not well balanced",
+	"extra content at the end of well balanced chunk",
     "XML_ERR_ENTITY_CHAR_ERROR",
-    "XML_ERR_ENTITY_PE_INTERNAL",
-    "XML_ERR_ENTITY_LOOP",
+    "PEReferences forbidden in internal subset",
+    "Detected an entity reference loop",
     "XML_ERR_ENTITY_BOUNDARY",
     "Invalid URI",
-    "XML_ERR_URI_FRAGMENT",
+    "Fragment not allowed",
     "XML_WAR_CATALOG_PI",
     "XML_ERR_NO_DTD",
-    "XML_ERR_CONDSEC_INVALID_KEYWORD", /* 95 */
-    "XML_ERR_VERSION_MISSING", /* 96 */
+    "conditional section INCLUDE or IGNORE keyword expected", /* 95 */
+    "Version in XML Declaration missing", /* 96 */
     "XML_WAR_UNKNOWN_VERSION", /* 97 */
     "XML_WAR_LANG_VALUE", /* 98 */
     "XML_WAR_NS_URI", /* 99 */
     "XML_WAR_NS_URI_RELATIVE", /* 100 */
-    "XML_ERR_MISSING_ENCODING" /* 101 */
+    "Missing encoding in text declaration" /* 101 */
 };
 
 PHPAPI const XML_Char *
 XML_ErrorString(int code)
 {
-	if (code < 0 || code >= (int)sizeof(error_mapping)) {
+	if (code < 0 || code >= (int)(sizeof(error_mapping) / sizeof(error_mapping[0]))) {
 		return "Unknown";
 	}
 	return error_mapping[code];
diff --git a/ext/xml/xml.c b/ext/xml/xml.c
index f17601f8eb..fba42e2e71 100644
--- a/ext/xml/xml.c
+++ b/ext/xml/xml.c
@@ -176,7 +176,7 @@ static int le_xml_parser;
 #ifdef ZTS
 static void php_xml_init_globals(php_xml_globals *xml_globals_p TSRMLS_DC)
 {
-	XML(default_encoding) = "ISO-8859-1";
+	XML(default_encoding) = "UTF-8";
 }
 #endif
 
@@ -204,7 +204,7 @@ PHP_MINIT_FUNCTION(xml)
 #ifdef ZTS
 	ts_allocate_id(&xml_globals_id, sizeof(php_xml_globals), (ts_allocate_ctor) php_xml_init_globals, NULL);
 #else
-	XML(default_encoding) = "ISO-8859-1";
+	XML(default_encoding) = "UTF-8";
 #endif
 
 	REGISTER_LONG_CONSTANT("XML_ERROR_NONE", XML_ERROR_NONE, CONST_CS|CONST_PERSISTENT);