Unicode support

12 files changed, 63 insertions, 64 deletions

diff --git a/ext/simplexml/simplexml.c b/ext/simplexml/simplexml.c
index 6d2b2b2efe..f4e97e6016 100644
--- a/ext/simplexml/simplexml.c
+++ b/ext/simplexml/simplexml.c
@@ -746,16 +746,6 @@ static void sxe_dimension_delete(zval *object, zval *offset TSRMLS_DC)
 }
 /* }}} */
 
-static inline char * sxe_xmlNodeListGetString(xmlDocPtr doc, xmlNodePtr list, int inLine)
-{
-	xmlChar *tmp = xmlNodeListGetString(doc, list, inLine);
-	char    *res = estrdup((char*)tmp);
-
-	xmlFree(tmp);
-	
-	return res;
-}
-
 /* {{{ _get_base_node_value()
  */
 static void _get_base_node_value(php_sxe_object *sxe_ref, xmlNodePtr node, zval **value, char *prefix TSRMLS_DC)
@@ -768,7 +758,7 @@ static void _get_base_node_value(php_sxe_object *sxe_ref, xmlNodePtr node, zval
 	if (node->children && node->children->type == XML_TEXT_NODE && !xmlIsBlankNode(node->children)) {
 		contents = xmlNodeListGetString(node->doc, node->children, 1);
 		if (contents) {
-			ZVAL_STRING(*value, contents, 1);
+			ZVAL_U_STRING(ZEND_U_CONVERTER(UG(runtime_encoding_conv)), *value, contents, 1);
 			xmlFree(contents);
 		}
 	} else {
@@ -846,8 +836,11 @@ static HashTable * sxe_properties_get(zval *object TSRMLS_DC)
 		test = sxe->iter.name && sxe->iter.type == SXE_ITER_ATTRLIST;
 		while (attr) {
 			if ((!test || !xmlStrcmp(attr->name, sxe->iter.name)) && match_ns(sxe, (xmlNodePtr)attr, sxe->iter.nsprefix)) {
+				xmlChar *tmp;
+
 				MAKE_STD_ZVAL(value);
-				ZVAL_STRING(value, sxe_xmlNodeListGetString((xmlDocPtr) sxe->document->ptr, attr->children, 1), 0);
+				ZVAL_U_STRING(ZEND_U_CONVERTER(UG(runtime_encoding_conv)), value, tmp = xmlNodeListGetString((xmlDocPtr) sxe->document->ptr, attr->children, 1), 1);
+				xmlFree(tmp);
 				namelen = xmlStrlen(attr->name) + 1;
 				if (!zattr) {
 					MAKE_STD_ZVAL(zattr);
@@ -864,8 +857,11 @@ static HashTable * sxe_properties_get(zval *object TSRMLS_DC)
 	node = php_sxe_get_first_node(sxe, node TSRMLS_CC);
 	if (node && sxe->iter.type != SXE_ITER_ATTRLIST) {
 		if (node->type == XML_ATTRIBUTE_NODE) {
+			xmlChar *tmp;
+
 			MAKE_STD_ZVAL(value);
-			ZVAL_U_STRING(ZEND_U_CONVERTER(UG(runtime_encoding_conv)), value, sxe_xmlNodeListGetString(node->doc, node->children, 1), 0);
+			ZVAL_U_STRING(ZEND_U_CONVERTER(UG(runtime_encoding_conv)), value, tmp = xmlNodeListGetString(node->doc, node->children, 1), 1);
+			xmlFree(tmp);
 			zend_hash_next_index_insert(rv, &value, sizeof(zval *), NULL);
 			node = NULL;
 		} else {
@@ -877,8 +873,11 @@ static HashTable * sxe_properties_get(zval *object TSRMLS_DC)
 				SKIP_TEXT(node);
 			} else {
 				if (node->type == XML_TEXT_NODE) {
+					xmlChar *tmp;
+
 					MAKE_STD_ZVAL(value);
-					ZVAL_U_STRING(ZEND_U_CONVERTER(UG(runtime_encoding_conv)), value, sxe_xmlNodeListGetString(node->doc, node, 1), 0);
+					ZVAL_U_STRING(ZEND_U_CONVERTER(UG(runtime_encoding_conv)), value, tmp = xmlNodeListGetString(node->doc, node, 1), 1);
+					xmlFree(tmp);
 					zend_hash_next_index_insert(rv, &value, sizeof(zval *), NULL);
 					goto next_iter;
 				}
diff --git a/ext/simplexml/tests/000.phpt b/ext/simplexml/tests/000.phpt
index 4af67010b3..d11c271dac 100755
--- a/ext/simplexml/tests/000.phpt
+++ b/ext/simplexml/tests/000.phpt
@@ -261,20 +261,20 @@ object(SimpleXMLElement)#%d (3) {
   [u"@attributes"]=>
   array(1) {
     [u"id"]=>
-    string(3) "123"
+    unicode(3) "123"
   }
   [u"elem1"]=>
   array(2) {
     [0]=>
-    string(36) "There is some text.Here is some more"
+    unicode(36) "There is some text.Here is some more"
     [1]=>
     object(SimpleXMLElement)#%d (1) {
       [u"@attributes"]=>
       array(2) {
         [u"attr1"]=>
-        string(2) "11"
+        unicode(2) "11"
         [u"attr2"]=>
-        string(2) "12"
+        unicode(2) "12"
       }
     }
   }
@@ -296,9 +296,9 @@ object(SimpleXMLElement)#%d (3) {
   [u"@attributes"]=>
   array(2) {
     [u"attr1"]=>
-    string(5) "first"
+    unicode(5) "first"
     [u"attr2"]=>
-    string(6) "second"
+    unicode(6) "second"
   }
   [u"comment"]=>
   object(SimpleXMLElement)#%d (0) {
@@ -308,9 +308,9 @@ object(SimpleXMLElement)#%d (3) {
     [u"@attributes"]=>
     array(2) {
       [u"att25"]=>
-      string(2) "25"
+      unicode(2) "25"
       [u"att42"]=>
-      string(2) "42"
+      unicode(2) "42"
     }
     [u"elem3"]=>
     object(SimpleXMLElement)#%d (1) {
@@ -331,9 +331,9 @@ object(SimpleXMLElement)#%d (3) {
   [u"@attributes"]=>
   array(2) {
     [u"attr1"]=>
-    string(5) "first"
+    unicode(5) "first"
     [u"attr2"]=>
-    string(6) "second"
+    unicode(6) "second"
   }
   [u"comment"]=>
   object(SimpleXMLElement)#%d (0) {
@@ -343,9 +343,9 @@ object(SimpleXMLElement)#%d (3) {
     [u"@attributes"]=>
     array(2) {
       [u"att25"]=>
-      string(2) "25"
+      unicode(2) "25"
       [u"att42"]=>
-      string(2) "42"
+      unicode(2) "42"
     }
     [u"elem3"]=>
     object(SimpleXMLElement)#%d (1) {
@@ -366,9 +366,9 @@ object(SimpleXMLElement)#%d (2) {
   [u"@attributes"]=>
   array(2) {
     [u"att25"]=>
-    string(2) "25"
+    unicode(2) "25"
     [u"att42"]=>
-    string(2) "42"
+    unicode(2) "42"
   }
   [u"elem3"]=>
   object(SimpleXMLElement)#%d (1) {
@@ -400,9 +400,9 @@ object(SimpleXMLElement)#%d (1) {
   [u"@attributes"]=>
   array(2) {
     [u"attr1"]=>
-    string(2) "11"
+    unicode(2) "11"
     [u"attr2"]=>
-    string(2) "12"
+    unicode(2) "12"
   }
 }
 ===sxe->elem1[2]
diff --git a/ext/simplexml/tests/001.phpt b/ext/simplexml/tests/001.phpt
index 5dc9abfd6f..b5c51ff960 100644
--- a/ext/simplexml/tests/001.phpt
+++ b/ext/simplexml/tests/001.phpt
@@ -46,14 +46,14 @@ object(SimpleXMLElement)#%d (2) {
   [u"@attributes"]=>
   array(1) {
     [u"id"]=>
-    string(5) "elem1"
+    unicode(5) "elem1"
   }
   [u"elem1"]=>
   object(SimpleXMLElement)#%d (3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr1"]=>
-      string(5) "first"
+      unicode(5) "first"
     }
     [u"comment"]=>
     object(SimpleXMLElement)#%d (0) {
diff --git a/ext/simplexml/tests/002.phpt b/ext/simplexml/tests/002.phpt
index 8632207d24..6d1679faa4 100644
--- a/ext/simplexml/tests/002.phpt
+++ b/ext/simplexml/tests/002.phpt
@@ -67,14 +67,14 @@ object(SimpleXMLElement)#%d (2) {
   [u"@attributes"]=>
   array(1) {
     [u"id"]=>
-    string(5) "elem1"
+    unicode(5) "elem1"
   }
   [u"elem1"]=>
   object(SimpleXMLElement)#%d (3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr1"]=>
-      string(5) "first"
+      unicode(5) "first"
     }
     [u"comment"]=>
     object(SimpleXMLElement)#%d (0) {
diff --git a/ext/simplexml/tests/003.phpt b/ext/simplexml/tests/003.phpt
index 978768a366..9cb5c0e5dc 100755
--- a/ext/simplexml/tests/003.phpt
+++ b/ext/simplexml/tests/003.phpt
@@ -72,14 +72,14 @@ object(SimpleXMLElement)#%d (2) {
   [u"@attributes"]=>
   array(1) {
     [u"id"]=>
-    string(5) "elem1"
+    unicode(5) "elem1"
   }
   [u"elem1"]=>
   object(SimpleXMLElement)#%d (3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr1"]=>
-      string(5) "first"
+      unicode(5) "first"
     }
     [u"comment"]=>
     object(SimpleXMLElement)#%d (0) {
@@ -91,7 +91,7 @@ object(SimpleXMLElement)#%d (2) {
         [u"included-entity"]=>
         object(SimpleXMLElement)#%d (1) {
           [u"included-entity"]=>
-          string(36) "This is text included from an entity"
+          unicode(36) "This is text included from an entity"
         }
         [u"elem4"]=>
         object(SimpleXMLElement)#%d (1) {
diff --git a/ext/simplexml/tests/004.phpt b/ext/simplexml/tests/004.phpt
index c2f462be6b..bb5916440b 100755
--- a/ext/simplexml/tests/004.phpt
+++ b/ext/simplexml/tests/004.phpt
@@ -71,14 +71,14 @@ object(SimpleXMLElement)#%d (2) {
   [u"@attributes"]=>
   array(1) {
     [u"id"]=>
-    string(5) "elem1"
+    unicode(5) "elem1"
   }
   [u"elem1"]=>
   object(SimpleXMLElement)#%d (3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr1"]=>
-      string(5) "first"
+      unicode(5) "first"
     }
     [u"comment"]=>
     object(SimpleXMLElement)#%d (0) {
diff --git a/ext/simplexml/tests/007.phpt b/ext/simplexml/tests/007.phpt
index 5a7ba29deb..5a2cf6f9c0 100755
--- a/ext/simplexml/tests/007.phpt
+++ b/ext/simplexml/tests/007.phpt
@@ -101,7 +101,7 @@ object(SimpleXMLElement)#%d (3) {
   [u"@attributes"]=>
   array(1) {
     [u"attr1"]=>
-    string(5) "first"
+    unicode(5) "first"
   }
   [u"comment"]=>
   object(SimpleXMLElement)#%d (0) {
diff --git a/ext/simplexml/tests/010.phpt b/ext/simplexml/tests/010.phpt
index a71618ba01..5318a525a3 100644
--- a/ext/simplexml/tests/010.phpt
+++ b/ext/simplexml/tests/010.phpt
@@ -67,14 +67,14 @@ object(simplexml_inherited)#%d (2) {
   [u"@attributes"]=>
   array(1) {
     [u"id"]=>
-    string(5) "elem1"
+    unicode(5) "elem1"
   }
   [u"elem1"]=>
   object(simplexml_inherited)#%d (3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr1"]=>
-      string(5) "first"
+      unicode(5) "first"
     }
     [u"comment"]=>
     object(simplexml_inherited)#%d (0) {
diff --git a/ext/simplexml/tests/022.phpt b/ext/simplexml/tests/022.phpt
index 36cc0145b3..5374bbaa2d 100755
--- a/ext/simplexml/tests/022.phpt
+++ b/ext/simplexml/tests/022.phpt
@@ -68,7 +68,7 @@ object(SimpleXMLElement)#%d (1) {
     [u"@attributes"]=>
     array(1) {
       [u"glob"]=>
-      string(11) "slide_*.xml"
+      unicode(11) "slide_*.xml"
     }
   }
 }
@@ -77,7 +77,7 @@ object(SimpleXMLElement)#%d (1) {
   [u"@attributes"]=>
   array(1) {
     [u"glob"]=>
-    string(11) "slide_*.xml"
+    unicode(11) "slide_*.xml"
   }
 }
 ===FOREACH===
@@ -85,7 +85,7 @@ object(SimpleXMLElement)#%d (1) {
   [u"@attributes"]=>
   array(1) {
     [u"glob"]=>
-    string(11) "slide_*.xml"
+    unicode(11) "slide_*.xml"
   }
 }
 object(SimpleXMLElement)#%d (1) {
diff --git a/ext/simplexml/tests/023.phpt b/ext/simplexml/tests/023.phpt
index 449daf47b7..be82b78f38 100755
--- a/ext/simplexml/tests/023.phpt
+++ b/ext/simplexml/tests/023.phpt
@@ -39,7 +39,7 @@ object(SimpleXMLElement)#%d (1) {
   [u"@attributes"]=>
   array(1) {
     [u"attr"]=>
-    string(%d) "foo%sbar%sbaz"
+    unicode(%d) "foo%sbar%sbaz"
   }
 }
 object(SimpleXMLElement)#%d (1) {
diff --git a/ext/simplexml/tests/024.phpt b/ext/simplexml/tests/024.phpt
index a136cfb9cc..018d7e50b6 100755
--- a/ext/simplexml/tests/024.phpt
+++ b/ext/simplexml/tests/024.phpt
@@ -181,7 +181,7 @@ array(3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "12"
+      unicode(2) "12"
     }
   }
   [1]=>
@@ -189,7 +189,7 @@ array(3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "22"
+      unicode(2) "22"
     }
   }
   [2]=>
@@ -197,7 +197,7 @@ array(3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "32"
+      unicode(2) "32"
     }
   }
 }
@@ -208,7 +208,7 @@ array(3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "12"
+      unicode(2) "12"
     }
   }
   [1]=>
@@ -216,7 +216,7 @@ array(3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "22"
+      unicode(2) "22"
     }
   }
   [2]=>
@@ -224,7 +224,7 @@ array(3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "32"
+      unicode(2) "32"
     }
   }
 }
@@ -235,7 +235,7 @@ array(9) {
     [u"@attributes"]=>
     array(1) {
       [u"attr1"]=>
-      string(2) "11"
+      unicode(2) "11"
     }
   }
   [1]=>
@@ -243,7 +243,7 @@ array(9) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "12"
+      unicode(2) "12"
     }
   }
   [2]=>
@@ -251,7 +251,7 @@ array(9) {
     [u"@attributes"]=>
     array(1) {
       [u"attr3"]=>
-      string(2) "13"
+      unicode(2) "13"
     }
   }
   [3]=>
@@ -259,7 +259,7 @@ array(9) {
     [u"@attributes"]=>
     array(1) {
       [u"attr1"]=>
-      string(2) "21"
+      unicode(2) "21"
     }
   }
   [4]=>
@@ -267,7 +267,7 @@ array(9) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "22"
+      unicode(2) "22"
     }
   }
   [5]=>
@@ -275,7 +275,7 @@ array(9) {
     [u"@attributes"]=>
     array(1) {
       [u"attr3"]=>
-      string(2) "23"
+      unicode(2) "23"
     }
   }
   [6]=>
@@ -283,7 +283,7 @@ array(9) {
     [u"@attributes"]=>
     array(1) {
       [u"attr1"]=>
-      string(2) "31"
+      unicode(2) "31"
     }
   }
   [7]=>
@@ -291,7 +291,7 @@ array(9) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "32"
+      unicode(2) "32"
     }
   }
   [8]=>
@@ -299,7 +299,7 @@ array(9) {
     [u"@attributes"]=>
     array(1) {
       [u"attr3"]=>
-      string(2) "33"
+      unicode(2) "33"
     }
   }
 }
@@ -310,7 +310,7 @@ array(1) {
     [u"@attributes"]=>
     array(1) {
       [u"attr2"]=>
-      string(2) "22"
+      unicode(2) "22"
     }
   }
 }
diff --git a/ext/spl/tests/sxe_001.phpt b/ext/spl/tests/sxe_001.phpt
index 11d8eeee23..0f5530249e 100755
--- a/ext/spl/tests/sxe_001.phpt
+++ b/ext/spl/tests/sxe_001.phpt
@@ -67,14 +67,14 @@ object(SimpleXMLIterator)#%d (2) {
   [u"@attributes"]=>
   array(1) {
     [u"id"]=>
-    string(5) "elem1"
+    unicode(5) "elem1"
   }
   [u"elem1"]=>
   object(SimpleXMLIterator)#%d (3) {
     [u"@attributes"]=>
     array(1) {
       [u"attr1"]=>
-      string(5) "first"
+      unicode(5) "first"
     }
     [u"comment"]=>
     object(SimpleXMLIterator)#%d (0) {