fix bug #38454 (warning upon disabling handler via xml_set_element_handler)

fix bug #38427 (unicode causes xml_parser to misbehave)
remove need for TSRMLS_FETCH() call
add test

diff --git a/ext/xml/tests/xml011.phpt b/ext/xml/tests/xml011.phpt
new file mode 100644 (file)
index 0000000..9c4cfca
--- /dev/null
+++ b/ext/xml/tests/xml011.phpt
@@ -0,0 +1,71 @@
+--TEST--
+XML Parser test: concat character data and set empty handlers
+--SKIPIF--
+<?php
+require_once("skipif.inc");
+?>
+--FILE--
+<?php
+function start_elem($parser,$name,$attribs) {
+   echo "<$name>";
+}
+function end_elem()
+{
+   echo "</$name>";
+}
+
+$xml = '<text>start<b /> This &amp; that</text>';
+
+$parser = xml_parser_create();
+xml_parse_into_struct($parser, $xml, $vals, $index);
+print_r($vals);
+xml_parser_free($parser);
+
+echo "\nChange to empty end handler\n";
+$parser = xml_parser_create();
+xml_parser_set_option($parser,XML_OPTION_CASE_FOLDING,0);
+xml_set_element_handler($parser,'start_elem','end_elem');
+xml_set_element_handler($parser,'start_elem',NULL);
+xml_parse($parser, $xml, TRUE);
+
+xml_parser_free($parser);
+echo "\nDone\n";
+?>
+--EXPECT--
+Array
+(
+    [0] => Array
+        (
+            [tag] => TEXT
+            [type] => open
+            [level] => 1
+            [value] => start
+        )
+
+    [1] => Array
+        (
+            [tag] => B
+            [type] => complete
+            [level] => 2
+        )
+
+    [2] => Array
+        (
+            [tag] => TEXT
+            [value] =>  This & that
+            [type] => cdata
+            [level] => 1
+        )
+
+    [3] => Array
+        (
+            [tag] => TEXT
+            [type] => close
+            [level] => 1
+        )
+
+)
+
+Change to empty end handler
+<text><b>
+Done

diff --git a/ext/xml/xml.c b/ext/xml/xml.c
index 5a257c60a476d502ac1da39199651a33694443aa..2b6c45a6880cbdcf7a9638488e4aba282a7bbe39 100644 (file)
--- a/ext/xml/xml.c
+++ b/ext/xml/xml.c
@@ -74,7 +74,7 @@ PHP_MINFO_FUNCTION(xml);
 static PHP_GINIT_FUNCTION(xml);
 
 static void xml_parser_dtor(zend_rsrc_list_entry *rsrc TSRMLS_DC);
-static void xml_set_handler(zval **, zval **);
+static void xml_set_handler(zval **, zval **  TSRMLS_DC);
 inline static unsigned short xml_encode_iso_8859_1(unsigned char);
 inline static char xml_decode_iso_8859_1(unsigned short);
 inline static unsigned short xml_encode_us_ascii(unsigned char);
@@ -371,7 +371,7 @@ static void xml_parser_dtor(zend_rsrc_list_entry *rsrc TSRMLS_DC)
 /* }}} */
 
 /* {{{ xml_set_handler() */
-static void xml_set_handler(zval **handler, zval **data)
+static void xml_set_handler(zval **handler, zval **data TSRMLS_DC)
 {
        /* If we have already a handler, release it */
        if (*handler) {
@@ -380,9 +380,13 @@ static void xml_set_handler(zval **handler, zval **data)
 
        /* IS_ARRAY might indicate that we're using array($obj, 'method') syntax */
        if (Z_TYPE_PP(data) != IS_ARRAY) {
-               TSRMLS_FETCH();
-
                convert_to_text_ex(data);
+               if (((Z_TYPE_PP(data)==IS_UNICODE) && (Z_USTRLEN_PP(data) == 0)) ||
+                       ((Z_TYPE_PP(data)==IS_STRING) && (Z_STRLEN_PP(data) == 0))) {
+
+                       *handler = NULL;
+                       return;
+               }
        }
 
        zval_add_ref(data);
@@ -859,6 +863,25 @@ void _xml_characterDataHandler(void *userData, const XML_Char *s, int len)
                                        
                                } else {
                                        zval *tag;
+                                       zval **curtag, **mytype, **myval;
+                                       HashPosition hpos=NULL;
+
+                                       zend_hash_internal_pointer_end_ex(Z_ARRVAL_P(parser->data), &hpos);
+
+                                       if (hpos && (zend_hash_get_current_data_ex(Z_ARRVAL_P(parser->data), (void **) &curtag, &hpos) == SUCCESS)) {
+                                               if (zend_hash_find(Z_ARRVAL_PP(curtag),"type",sizeof("type"),(void **) &mytype) == SUCCESS) {
+                                                       if (!strcmp(Z_STRVAL_PP(mytype), "cdata")) {
+                                                               if (zend_hash_find(Z_ARRVAL_PP(curtag),"value",sizeof("value"),(void **) &myval) == SUCCESS) {
+                                                                       int newlen = Z_STRLEN_PP(myval) + decoded_len;
+                                                                       Z_STRVAL_PP(myval) = erealloc(Z_STRVAL_PP(myval),newlen+1);
+                                                                       strcpy(Z_STRVAL_PP(myval) + Z_STRLEN_PP(myval),decoded_value);
+                                                                       Z_STRLEN_PP(myval) += decoded_len;
+                                                                       efree(decoded_value);
+                                                                       return;
+                                                               }
+                                                       }
+                                               }
+                                       }
 
                                        MAKE_STD_ZVAL(tag);
                                        
@@ -1158,8 +1181,8 @@ PHP_FUNCTION(xml_set_element_handler)
 
        ZEND_FETCH_RESOURCE(parser,xml_parser *,pind, -1, "XML Parser", le_xml_parser);
 
-       xml_set_handler(&parser->startElementHandler, shdl);
-       xml_set_handler(&parser->endElementHandler, ehdl);
+       xml_set_handler(&parser->startElementHandler, shdl TSRMLS_CC);
+       xml_set_handler(&parser->endElementHandler, ehdl TSRMLS_CC);
        XML_SetElementHandler(parser->parser, _xml_startElementHandler, _xml_endElementHandler);
        RETVAL_TRUE;
 }
@@ -1178,7 +1201,7 @@ PHP_FUNCTION(xml_set_character_data_handler)
 
        ZEND_FETCH_RESOURCE(parser,xml_parser *, pind, -1, "XML Parser", le_xml_parser);
 
-       xml_set_handler(&parser->characterDataHandler, hdl);
+       xml_set_handler(&parser->characterDataHandler, hdl TSRMLS_CC);
        XML_SetCharacterDataHandler(parser->parser, _xml_characterDataHandler);
        RETVAL_TRUE;
 }
@@ -1197,7 +1220,7 @@ PHP_FUNCTION(xml_set_processing_instruction_handler)
 
        ZEND_FETCH_RESOURCE(parser,xml_parser *, pind, -1, "XML Parser", le_xml_parser);
 
-       xml_set_handler(&parser->processingInstructionHandler, hdl);
+       xml_set_handler(&parser->processingInstructionHandler, hdl TSRMLS_CC);
        XML_SetProcessingInstructionHandler(parser->parser, _xml_processingInstructionHandler);
        RETVAL_TRUE;
 }
@@ -1215,7 +1238,7 @@ PHP_FUNCTION(xml_set_default_handler)
        }
        ZEND_FETCH_RESOURCE(parser,xml_parser *, pind, -1, "XML Parser", le_xml_parser);
 
-       xml_set_handler(&parser->defaultHandler, hdl);
+       xml_set_handler(&parser->defaultHandler, hdl TSRMLS_CC);
        XML_SetDefaultHandler(parser->parser, _xml_defaultHandler);
        RETVAL_TRUE;
 }
@@ -1234,7 +1257,7 @@ PHP_FUNCTION(xml_set_unparsed_entity_decl_handler)
 
        ZEND_FETCH_RESOURCE(parser,xml_parser *, pind, -1, "XML Parser", le_xml_parser);
 
-       xml_set_handler(&parser->unparsedEntityDeclHandler, hdl);
+       xml_set_handler(&parser->unparsedEntityDeclHandler, hdl TSRMLS_CC);
        XML_SetUnparsedEntityDeclHandler(parser->parser, _xml_unparsedEntityDeclHandler);
        RETVAL_TRUE;
 }
@@ -1252,7 +1275,7 @@ PHP_FUNCTION(xml_set_notation_decl_handler)
        }
        ZEND_FETCH_RESOURCE(parser,xml_parser *, pind, -1, "XML Parser", le_xml_parser);
 
-       xml_set_handler(&parser->notationDeclHandler, hdl);
+       xml_set_handler(&parser->notationDeclHandler, hdl TSRMLS_CC);
        XML_SetNotationDeclHandler(parser->parser, _xml_notationDeclHandler);
        RETVAL_TRUE;
 }
@@ -1270,7 +1293,7 @@ PHP_FUNCTION(xml_set_external_entity_ref_handler)
        }
        ZEND_FETCH_RESOURCE(parser,xml_parser *, pind, -1, "XML Parser", le_xml_parser);
 
-       xml_set_handler(&parser->externalEntityRefHandler, hdl);
+       xml_set_handler(&parser->externalEntityRefHandler, hdl TSRMLS_CC);
        XML_SetExternalEntityRefHandler(parser->parser, (void *) _xml_externalEntityRefHandler);
        RETVAL_TRUE;
 }
@@ -1289,7 +1312,7 @@ PHP_FUNCTION(xml_set_start_namespace_decl_handler)
 
        ZEND_FETCH_RESOURCE(parser,xml_parser *, pind, -1, "XML Parser", le_xml_parser);
 
-       xml_set_handler(&parser->startNamespaceDeclHandler, hdl);
+       xml_set_handler(&parser->startNamespaceDeclHandler, hdl TSRMLS_CC);
        XML_SetStartNamespaceDeclHandler(parser->parser, _xml_startNamespaceDeclHandler);
        RETVAL_TRUE;
 }
@@ -1308,7 +1331,7 @@ PHP_FUNCTION(xml_set_end_namespace_decl_handler)
 
        ZEND_FETCH_RESOURCE(parser,xml_parser *, pind, -1, "XML Parser", le_xml_parser);
 
-       xml_set_handler(&parser->endNamespaceDeclHandler, hdl);
+       xml_set_handler(&parser->endNamespaceDeclHandler, hdl TSRMLS_CC);
        XML_SetEndNamespaceDeclHandler(parser->parser, _xml_endNamespaceDeclHandler);
        RETVAL_TRUE;
 }