nokogiri 1.3.3-java → 1.4.0-java

data/ext/nokogiri/xml_node.h

@@ -7,7 +7,6 @@ void init_xml_node();
 
 extern VALUE cNokogiriXmlNode ;
 extern VALUE cNokogiriXmlElement ;
-extern VALUE cNokogiriXmlEntityDeclaration ;
 
 VALUE Nokogiri_wrap_xml_node(VALUE klass, xmlNodePtr node) ;
 void Nokogiri_xml_node_properties(xmlNodePtr node, VALUE attr_hash) ;

data/ext/nokogiri/xml_node_set.c

@@ -45,7 +45,7 @@ static VALUE push(VALUE self, VALUE rb_node)
   xmlNodeSetPtr node_set;
   xmlNodePtr node;
 
-  if(! rb_funcall(rb_node, rb_intern("is_a?"), 1, cNokogiriXmlNode))
+  if(!rb_obj_is_kind_of(rb_node, cNokogiriXmlNode))
     rb_raise(rb_eArgError, "node must be a Nokogiri::XML::Node");
 
   Data_Get_Struct(self, xmlNodeSet, node_set);
@@ -66,7 +66,7 @@ static VALUE delete(VALUE self, VALUE rb_node)
   xmlNodeSetPtr node_set ;
   xmlNodePtr node ;
 
-  if(! rb_funcall(rb_node, rb_intern("is_a?"), 1, cNokogiriXmlNode))
+  if(!rb_obj_is_kind_of(rb_node, cNokogiriXmlNode))
     rb_raise(rb_eArgError, "node must be a Nokogiri::XML::Node");
   
   Data_Get_Struct(self, xmlNodeSet, node_set);
@@ -92,7 +92,7 @@ static VALUE intersection(VALUE self, VALUE rb_other)
   xmlNodeSetPtr node_set;
   xmlNodeSetPtr other;
 
-  if(! rb_funcall(rb_other, rb_intern("is_a?"), 1, cNokogiriXmlNodeSet))
+  if(!rb_obj_is_kind_of(rb_other, cNokogiriXmlNodeSet))
     rb_raise(rb_eArgError, "node_set must be a Nokogiri::XML::NodeSet");
 
   Data_Get_Struct(self, xmlNodeSet, node_set);
@@ -113,7 +113,7 @@ static VALUE include_eh(VALUE self, VALUE rb_node)
   xmlNodeSetPtr node_set;
   xmlNodePtr node;
 
-  if(! rb_funcall(rb_node, rb_intern("is_a?"), 1, cNokogiriXmlNode))
+  if(!rb_obj_is_kind_of(rb_node, cNokogiriXmlNode))
     rb_raise(rb_eArgError, "node must be a Nokogiri::XML::Node");
 
   Data_Get_Struct(self, xmlNodeSet, node_set);
@@ -125,18 +125,18 @@ static VALUE include_eh(VALUE self, VALUE rb_node)
 
 /*
  * call-seq:
- *  +(node_set)
+ *  |(node_set)
  *
- *  Concatenation - returns a new NodeSet built by concatenating the node set
- *  with +node_set+ to produce a third NodeSet
+ * Returns a new set built by merging the set and the elements of the given
+ * set.
  */
-static VALUE plus(VALUE self, VALUE rb_other)
+static VALUE set_union(VALUE self, VALUE rb_other)
 {
   xmlNodeSetPtr node_set;
   xmlNodeSetPtr other;
   xmlNodeSetPtr new;
 
-  if(! rb_funcall(rb_other, rb_intern("is_a?"), 1, cNokogiriXmlNodeSet))
+  if(!rb_obj_is_kind_of(rb_other, cNokogiriXmlNodeSet))
     rb_raise(rb_eArgError, "node_set must be a Nokogiri::XML::NodeSet");
 
   Data_Get_Struct(self, xmlNodeSet, node_set);
@@ -166,7 +166,7 @@ static VALUE minus(VALUE self, VALUE rb_other)
   xmlNodeSetPtr new;
   int j ;
 
-  if(! rb_funcall(rb_other, rb_intern("is_a?"), 1, cNokogiriXmlNodeSet))
+  if(!rb_obj_is_kind_of(rb_other, cNokogiriXmlNodeSet))
     rb_raise(rb_eArgError, "node_set must be a Nokogiri::XML::NodeSet");
 
   Data_Get_Struct(self, xmlNodeSet, node_set);
@@ -238,6 +238,8 @@ static VALUE slice(int argc, VALUE *argv, VALUE self)
     if (beg < 0) {
       beg += node_set->nodeNr ;
     }
+    if (len > node_set->nodeNr) len = node_set->nodeNr;
+
     return subseq(self, beg, len);
   }
 
@@ -278,16 +280,21 @@ static VALUE to_array(VALUE self, VALUE rb_node)
   VALUE *elts = calloc((size_t)set->nodeNr, sizeof(VALUE *));
   int i;
   for(i = 0; i < set->nodeNr; i++) {
-    if(set->nodeTab[i]->_private) {
-      elts[i] = (VALUE)set->nodeTab[i]->_private;
+    xmlNodePtr node = set->nodeTab[i];
+
+    if(node->_private) {
+      if(node->type == XML_DOCUMENT_NODE || node->type == XML_HTML_DOCUMENT_NODE)
+        elts[i] = DOC_RUBY_OBJECT(node->doc);
+      else
+        elts[i] = (VALUE)node->_private;
     } else {
-      elts[i] = Nokogiri_wrap_xml_node(Qnil, set->nodeTab[i]);
+      elts[i] = Nokogiri_wrap_xml_node(Qnil, node);
     }
   }
 
-  VALUE list = rb_ary_new4(set->nodeNr, elts);
+  VALUE list = rb_ary_new4((long)set->nodeNr, elts);
 
-  free(elts);
+  //free(elts);
 
   return list;
 }
@@ -379,7 +386,7 @@ void init_xml_node_set(void)
   rb_define_method(klass, "[]", slice, -1);
   rb_define_method(klass, "slice", slice, -1);
   rb_define_method(klass, "push", push, 1);
-  rb_define_method(klass, "+", plus, 1);
+  rb_define_method(klass, "|", set_union, 1);
   rb_define_method(klass, "-", minus, 1);
   rb_define_method(klass, "unlink", unlink_nodeset, 0);
   rb_define_method(klass, "to_a", to_array, 0);

data/ext/nokogiri/xml_processing_instruction.c

@@ -28,7 +28,7 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   NOKOGIRI_ROOT_NODE(node);
 
   VALUE rb_node = Nokogiri_wrap_xml_node(klass, node);
-  rb_funcall2(rb_node, rb_intern("initialize"), argc, argv);
+  rb_obj_call_init(rb_node, argc, argv);
 
   if(rb_block_given_p()) rb_yield(rb_node);
 

data/ext/nokogiri/xml_reader.c

@@ -54,8 +54,8 @@ static void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash)
     }
 
     rb_hash_aset(attr_hash,
-        NOKOGIRI_STR_NEW2(key, node->doc->encoding),
-        (ns->href ? NOKOGIRI_STR_NEW2(ns->href, node->doc->encoding) : Qnil)
+        NOKOGIRI_STR_NEW2(key),
+        (ns->href ? NOKOGIRI_STR_NEW2(ns->href) : Qnil)
     );
     if (key != buffer) {
       free(key);
@@ -163,17 +163,6 @@ static VALUE attribute_nodes(VALUE self)
   xmlNodePtr ptr = xmlTextReaderExpand(reader);
   if(ptr == NULL) return Qnil;
 
-  VALUE enc = rb_iv_get(self, "@encoding");
-
-  if(enc != Qnil && NULL == ptr->doc->encoding) {
-    ptr->doc->encoding = calloc((size_t)RSTRING_LEN(enc), sizeof(char));
-    strncpy(
-      (char *)ptr->doc->encoding,
-      StringValuePtr(enc),
-      (size_t)RSTRING_LEN(enc)
-    );
-  }
-
   Nokogiri_xml_node_properties(ptr, attr);
 
   return attr ;
@@ -190,8 +179,8 @@ static VALUE attribute_at(VALUE self, VALUE index)
   xmlTextReaderPtr reader;
   Data_Get_Struct(self, xmlTextReader, reader);
 
-  if(index == Qnil) return Qnil;
-  index = rb_funcall(index, rb_intern("to_i"), 0);
+  if(NIL_P(index)) return Qnil;
+  index = rb_Integer(index);
 
   xmlChar * value = xmlTextReaderGetAttributeNo(
       reader,
@@ -199,9 +188,7 @@ static VALUE attribute_at(VALUE self, VALUE index)
   );
   if(value == NULL) return Qnil;
 
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  VALUE rb_value = NOKOGIRI_STR_NEW2(value,
-      RTEST(enc) ? StringValuePtr(enc) : NULL);
+  VALUE rb_value = NOKOGIRI_STR_NEW2(value);
   xmlFree(value);
   return rb_value;
 }
@@ -218,7 +205,7 @@ static VALUE reader_attribute(VALUE self, VALUE name)
   xmlChar *value ;
   Data_Get_Struct(self, xmlTextReader, reader);
 
-  if(name == Qnil) return Qnil;
+  if(NIL_P(name)) return Qnil;
   name = StringValue(name) ;
 
   value = xmlTextReaderGetAttribute(reader, (xmlChar*)StringValuePtr(name));
@@ -237,9 +224,7 @@ static VALUE reader_attribute(VALUE self, VALUE name)
   }
   if(value == NULL) return Qnil;
 
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  VALUE rb_value = NOKOGIRI_STR_NEW2(value,
-      RTEST(enc) ? StringValuePtr(enc) : NULL);
+  VALUE rb_value = NOKOGIRI_STR_NEW2(value);
   xmlFree(value);
   return rb_value;
 }
@@ -257,7 +242,7 @@ static VALUE attribute_count(VALUE self)
   int count = xmlTextReaderAttributeCount(reader);
   if(count == -1) return Qnil;
 
-  return INT2NUM(count);
+  return INT2NUM((long)count);
 }
 
 /*
@@ -273,7 +258,7 @@ static VALUE depth(VALUE self)
   int depth = xmlTextReaderDepth(reader);
   if(depth == -1) return Qnil;
 
-  return INT2NUM(depth);
+  return INT2NUM((long)depth);
 }
 
 /*
@@ -289,7 +274,7 @@ static VALUE xml_version(VALUE self)
   const char * version = (const char *)xmlTextReaderConstXmlVersion(reader);
   if(version == NULL) return Qnil;
 
-  return NOKOGIRI_STR_NEW2(version, "UTF-8");
+  return NOKOGIRI_STR_NEW2(version);
 }
 
 /*
@@ -305,9 +290,7 @@ static VALUE lang(VALUE self)
   const char * lang = (const char *)xmlTextReaderConstXmlLang(reader);
   if(lang == NULL) return Qnil;
 
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  return NOKOGIRI_STR_NEW2(lang,
-      RTEST(enc) ? StringValuePtr(enc) : NULL);
+  return NOKOGIRI_STR_NEW2(lang);
 }
 
 /*
@@ -323,9 +306,7 @@ static VALUE value(VALUE self)
   const char * value = (const char *)xmlTextReaderConstValue(reader);
   if(value == NULL) return Qnil;
 
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  return NOKOGIRI_STR_NEW2(value,
-      RTEST(enc) ? StringValuePtr(enc) : NULL);
+  return NOKOGIRI_STR_NEW2(value);
 }
 
 /*
@@ -341,9 +322,7 @@ static VALUE prefix(VALUE self)
   const char * prefix = (const char *)xmlTextReaderConstPrefix(reader);
   if(prefix == NULL) return Qnil;
 
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  return NOKOGIRI_STR_NEW2(prefix,
-      RTEST(enc) ? StringValuePtr(enc) : NULL);
+  return NOKOGIRI_STR_NEW2(prefix);
 }
 
 /*
@@ -359,9 +338,7 @@ static VALUE namespace_uri(VALUE self)
   const char * uri = (const char *)xmlTextReaderConstNamespaceUri(reader);
   if(uri == NULL) return Qnil;
 
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  return NOKOGIRI_STR_NEW2(uri,
-      RTEST(enc) ? StringValuePtr(enc) : NULL);
+  return NOKOGIRI_STR_NEW2(uri);
 }
 
 /*
@@ -377,9 +354,7 @@ static VALUE local_name(VALUE self)
   const char * name = (const char *)xmlTextReaderConstLocalName(reader);
   if(name == NULL) return Qnil;
 
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  return NOKOGIRI_STR_NEW2(name,
-      RTEST(enc) ? StringValuePtr(enc) : NULL);
+  return NOKOGIRI_STR_NEW2(name);
 }
 
 /*
@@ -395,9 +370,7 @@ static VALUE name(VALUE self)
   const char * name = (const char *)xmlTextReaderConstName(reader);
   if(name == NULL) return Qnil;
 
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  return NOKOGIRI_STR_NEW2(name,
-      RTEST(enc) ? StringValuePtr(enc) : NULL);
+  return NOKOGIRI_STR_NEW2(name);
 }
 
 /*
@@ -410,7 +383,20 @@ static VALUE state(VALUE self)
 {
   xmlTextReaderPtr reader;
   Data_Get_Struct(self, xmlTextReader, reader);
-  return INT2NUM(xmlTextReaderReadState(reader));
+  return INT2NUM((long)xmlTextReaderReadState(reader));
+}
+
+/*
+ * call-seq:
+ *   node_type
+ *
+ * Get the type of readers current node
+ */
+static VALUE node_type(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+  return INT2NUM((long)xmlTextReaderNodeType(reader));
 }
 
 /*
@@ -435,15 +421,51 @@ static VALUE read_more(VALUE self)
 
   xmlErrorPtr error = xmlGetLastError();
   if(error)
-    rb_funcall(rb_mKernel, rb_intern("raise"), 1,
-        Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error)
-    );
+    rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
   else
     rb_raise(rb_eRuntimeError, "Error pulling: %d", ret);
 
   return Qnil;
 }
 
+/*
+ * call-seq:
+ *   inner_xml
+ *
+ * Read the contents of the current node, including child nodes and markup.
+ */
+static VALUE inner_xml(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+
+  const char * value = (const char *)xmlTextReaderReadInnerXml(reader);
+
+  if(value == NULL)
+    return Qnil;
+  else
+    return NOKOGIRI_STR_NEW2(value);
+}
+
+/*
+ * call-seq:
+ *   outer_xml
+ *
+ * Read the current node and its contents, including child nodes and markup.
+ */
+static VALUE outer_xml(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  Data_Get_Struct(self, xmlTextReader, reader);
+
+  const char * value = (const char *)xmlTextReaderReadOuterXml(reader);
+
+  if(value == NULL)
+    return Qnil;
+  else
+    return NOKOGIRI_STR_NEW2(value);
+}
+
 /*
  * call-seq:
  *   from_memory(string, url = nil, encoding = nil, options = 0)
@@ -479,7 +501,8 @@ static VALUE from_memory(int argc, VALUE *argv, VALUE klass)
   }
 
   VALUE rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
-  rb_funcall(rb_reader, rb_intern("initialize"), 3, rb_buffer, rb_url, encoding);
+  VALUE args[3] = {rb_buffer, rb_url, encoding};
+  rb_obj_call_init(rb_reader, 3, args);
 
   return rb_reader;
 }
@@ -520,7 +543,8 @@ static VALUE from_io(int argc, VALUE *argv, VALUE klass)
   }
 
   VALUE rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
-  rb_funcall(rb_reader, rb_intern("initialize"), 3, rb_io, rb_url, encoding);
+  VALUE args[3] = {rb_io, rb_url, encoding};
+  rb_obj_call_init(rb_reader, 3, args);
 
   return rb_reader;
 }
@@ -545,7 +569,10 @@ void init_xml_reader()
   rb_define_singleton_method(klass, "from_io", from_io, -1);
 
   rb_define_method(klass, "read", read_more, 0);
+  rb_define_method(klass, "inner_xml", inner_xml, 0);
+  rb_define_method(klass, "outer_xml", outer_xml, 0);
   rb_define_method(klass, "state", state, 0);
+  rb_define_method(klass, "node_type", node_type, 0);
   rb_define_method(klass, "name", name, 0);
   rb_define_method(klass, "local_name", local_name, 0);
   rb_define_method(klass, "namespace_uri", namespace_uri, 0);
@@ -558,8 +585,9 @@ void init_xml_reader()
   rb_define_method(klass, "attribute", reader_attribute, 1);
   rb_define_method(klass, "namespaces", namespaces, 0);
   rb_define_method(klass, "attribute_at", attribute_at, 1);
-  rb_define_method(klass, "attribute_nodes", attribute_nodes, 0);
   rb_define_method(klass, "attributes?", attributes_eh, 0);
   rb_define_method(klass, "value?", value_eh, 0);
   rb_define_method(klass, "default?", default_eh, 0);
+
+  rb_define_private_method(klass, "attr_nodes", attribute_nodes, 0);
 }

data/ext/nokogiri/xml_sax_parser.c

@@ -1,122 +1,86 @@
-#include <nokogiri.h>
+#include <xml_sax_parser.h>
+
+static ID id_start_document, id_end_document, id_start_element, id_end_element;
+static ID id_start_element_namespace, id_end_element_namespace;
+static ID id_comment, id_characters, id_xmldecl, id_error, id_warning;
+static ID id_cdata_block, id_cAttribute;
 
 #define STRING_OR_NULL(str) \
    (RTEST(str) ? StringValuePtr(str) : NULL)
 
-#define RBSTR_OR_QNIL(_str, rb_enc) \
-  (_str ? NOKOGIRI_STR_NEW2(_str, STRING_OR_NULL(rb_enc)) : Qnil)
-
-/*
- * call-seq:
- *  parse_memory(data)
- *
- * Parse the document stored in +data+
- */
-static VALUE parse_memory(VALUE self, VALUE data)
-{
-  xmlSAXHandlerPtr handler;
-  Data_Get_Struct(self, xmlSAXHandler, handler);
-
-  if(Qnil == data) rb_raise(rb_eArgError, "data cannot be nil");
-
-  xmlSAXUserParseMemory(  handler,
-                          (void *)self,
-                          StringValuePtr(data),
-                          RSTRING_LEN(data)
-  );
-  return data;
-}
-
-/*
- * call-seq:
- *  native_parse_io(data, encoding)
- *
- * Parse the document accessable via +io+
- */
-static VALUE native_parse_io(VALUE self, VALUE io, VALUE encoding)
-{
-  xmlSAXHandlerPtr handler;
-  Data_Get_Struct(self, xmlSAXHandler, handler);
-
-  xmlCharEncoding enc = (xmlCharEncoding)NUM2INT(encoding); 
-
-  xmlParserCtxtPtr sax_ctx = xmlCreateIOParserCtxt(
-      handler,
-      (void *)self,
-      (xmlInputReadCallback)io_read_callback,
-      (xmlInputCloseCallback)io_close_callback,
-      (void *)io,
-      enc
-  );
-  xmlParseDocument(sax_ctx);
-  xmlFreeParserCtxt(sax_ctx);
-  return io;
-}
-
-/*
- * call-seq:
- *  native_parse_file(data)
- *
- * Parse the document stored in +data+
- */
-static VALUE native_parse_file(VALUE self, VALUE data)
-{
-  xmlSAXHandlerPtr handler;
-  Data_Get_Struct(self, xmlSAXHandler, handler);
-  xmlSAXUserParseFile(  handler,
-                        (void *)self,
-                        StringValuePtr(data)
-  );
-  return data;
-}
+#define RBSTR_OR_QNIL(_str) \
+  (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
 
 static void start_document(void * ctx)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  rb_funcall(doc, rb_intern("start_document"), 0);
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+
+  xmlParserCtxtPtr ctxt = NOKOGIRI_SAX_CTXT(ctx);
+
+  if(NULL != ctxt && ctxt->html != 1) {
+    if(ctxt->standalone != -1) {  // -1 means there was no declaration
+      VALUE encoding = ctxt->encoding ?
+        NOKOGIRI_STR_NEW2(ctxt->encoding) :
+        Qnil;
+
+      VALUE version = ctxt->version ?
+        NOKOGIRI_STR_NEW2(ctxt->version) :
+        Qnil;
+
+      VALUE standalone = Qnil;
+
+      switch(ctxt->standalone)
+      {
+        case 0:
+          standalone = NOKOGIRI_STR_NEW2("no");
+          break;
+        case 1:
+          standalone = NOKOGIRI_STR_NEW2("yes");
+          break;
+      }
+
+      rb_funcall(doc, id_xmldecl, 3, version, encoding, standalone);
+    }
+  }
+
+  rb_funcall(doc, id_start_document, 0);
 }
 
 static void end_document(void * ctx)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  rb_funcall(doc, rb_intern("end_document"), 0);
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  rb_funcall(doc, id_end_document, 0);
 }
 
 static void start_element(void * ctx, const xmlChar *name, const xmlChar **atts)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
   VALUE attributes = rb_ary_new();
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
   const xmlChar * attr;
   int i = 0;
   if(atts) {
     while((attr = atts[i]) != NULL) {
-      rb_funcall(attributes, rb_intern("<<"), 1,
-          NOKOGIRI_STR_NEW2(attr, STRING_OR_NULL(enc))
-      );
+      rb_ary_push(attributes, NOKOGIRI_STR_NEW2(attr));
       i++;
     }
   }
 
   rb_funcall( doc,
-              rb_intern("start_element"),
+              id_start_element,
               2,
-              NOKOGIRI_STR_NEW2(name, STRING_OR_NULL(enc)),
+              NOKOGIRI_STR_NEW2(name),
               attributes
   );
 }
 
 static void end_element(void * ctx, const xmlChar *name)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  rb_funcall(doc, rb_intern("end_element"), 1,
-      NOKOGIRI_STR_NEW2(name, STRING_OR_NULL(enc))
-  );
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  rb_funcall(doc, id_end_element, 1, NOKOGIRI_STR_NEW2(name));
 }
 
 static VALUE attributes_as_list(
@@ -124,29 +88,23 @@ static VALUE attributes_as_list(
   int nb_attributes,
   const xmlChar ** attributes)
 {
-  VALUE list = rb_ary_new2(nb_attributes);
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
+  VALUE list = rb_ary_new2((long)nb_attributes);
 
-  VALUE attr_klass = rb_const_get(cNokogiriXmlSaxParser, rb_intern("Attribute"));
+  VALUE attr_klass = rb_const_get(cNokogiriXmlSaxParser, id_cAttribute);
   if (attributes) {
     /* Each attribute is an array of [localname, prefix, URI, value, end] */
     int i;
     for (i = 0; i < nb_attributes * 5; i += 5) {
-      VALUE attribute = rb_funcall(attr_klass, rb_intern("new"), 4,
-        /* localname */
-        RBSTR_OR_QNIL(attributes[i + 0], enc),
-
-        /* prefix */
-        RBSTR_OR_QNIL(attributes[i + 1], enc),
+      VALUE argv[4];
+      argv[0] = RBSTR_OR_QNIL(attributes[i + 0]); /* localname */
+      argv[1] = RBSTR_OR_QNIL(attributes[i + 1]); /* prefix */
+      argv[2] = RBSTR_OR_QNIL(attributes[i + 2]); /* URI */
 
-        /* URI */
-        RBSTR_OR_QNIL(attributes[i + 2], enc),
+      /* value */
+      argv[3] = NOKOGIRI_STR_NEW((const char*)attributes[i+3],
+          (attributes[i+4] - attributes[i+3]));
 
-        /* value */
-        NOKOGIRI_STR_NEW((const char*)attributes[i+3],
-          (attributes[i+4] - attributes[i+3]),
-          STRING_OR_NULL(enc))
-      );
+      VALUE attribute = rb_class_new_instance(4, argv, attr_klass);
       rb_ary_push(list, attribute);
     }
   }
@@ -166,44 +124,33 @@ start_element_ns (
   int nb_defaulted,
   const xmlChar ** attributes)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
 
   VALUE attribute_list = attributes_as_list(self, nb_attributes, attributes);
 
-  VALUE ns_list = rb_ary_new2(nb_namespaces);
+  VALUE ns_list = rb_ary_new2((long)nb_namespaces);
 
   if (namespaces) {
     int i;
     for (i = 0; i < nb_namespaces * 2; i += 2)
     {
       rb_ary_push(ns_list,
-        rb_ary_new3(2,
-          RBSTR_OR_QNIL(namespaces[i + 0], enc),
-          RBSTR_OR_QNIL(namespaces[i + 1], enc)
+        rb_ary_new3((long)2,
+          RBSTR_OR_QNIL(namespaces[i + 0]),
+          RBSTR_OR_QNIL(namespaces[i + 1])
         )
       );
     }
   }
 
   rb_funcall( doc,
-              rb_intern("start_element_namespace"),
+              id_start_element_namespace,
               5,
-              NOKOGIRI_STR_NEW2(localname, STRING_OR_NULL(enc)),
+              NOKOGIRI_STR_NEW2(localname),
               attribute_list,
-              RBSTR_OR_QNIL(prefix, enc),
-              RBSTR_OR_QNIL(uri, enc),
-              ns_list
-  );
-
-  rb_funcall( self,
-              rb_intern("start_element_namespace"),
-              5,
-              NOKOGIRI_STR_NEW2(localname, STRING_OR_NULL(enc)),
-              attribute_list,
-              RBSTR_OR_QNIL(prefix, enc),
-              RBSTR_OR_QNIL(uri, enc),
+              RBSTR_OR_QNIL(prefix),
+              RBSTR_OR_QNIL(uri),
               ns_list
   );
 }
@@ -218,46 +165,36 @@ end_element_ns (
   const xmlChar * prefix,
   const xmlChar * uri)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-
-  rb_funcall(doc, rb_intern("end_element_namespace"), 3, 
-    NOKOGIRI_STR_NEW2(localname, STRING_OR_NULL(enc)),
-    RBSTR_OR_QNIL(prefix, enc),
-    RBSTR_OR_QNIL(uri, enc)
-  );
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
 
-  rb_funcall(self, rb_intern("end_element_namespace"), 3, 
-    NOKOGIRI_STR_NEW2(localname, STRING_OR_NULL(enc)),
-    RBSTR_OR_QNIL(prefix, enc),
-    RBSTR_OR_QNIL(uri, enc)
+  rb_funcall(doc, id_end_element_namespace, 3, 
+    NOKOGIRI_STR_NEW2(localname),
+    RBSTR_OR_QNIL(prefix),
+    RBSTR_OR_QNIL(uri)
   );
 }
 
 static void characters_func(void * ctx, const xmlChar * ch, int len)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  VALUE str = NOKOGIRI_STR_NEW(ch, len, STRING_OR_NULL(enc));
-  rb_funcall(doc, rb_intern("characters"), 1, str);
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  VALUE str = NOKOGIRI_STR_NEW(ch, len);
+  rb_funcall(doc, id_characters, 1, str);
 }
 
 static void comment_func(void * ctx, const xmlChar * value)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  VALUE str = NOKOGIRI_STR_NEW2(value, STRING_OR_NULL(enc));
-  rb_funcall(doc, rb_intern("comment"), 1, str);
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  VALUE str = NOKOGIRI_STR_NEW2(value);
+  rb_funcall(doc, id_comment, 1, str);
 }
 
 static void warning_func(void * ctx, const char *msg, ...)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
   char * message;
 
   va_list args;
@@ -265,17 +202,14 @@ static void warning_func(void * ctx, const char *msg, ...)
   vasprintf(&message, msg, args);
   va_end(args);
 
-  rb_funcall(doc, rb_intern("warning"), 1,
-      NOKOGIRI_STR_NEW2(message, STRING_OR_NULL(enc))
-  );
+  rb_funcall(doc, id_warning, 1, NOKOGIRI_STR_NEW2(message));
   free(message);
 }
 
 static void error_func(void * ctx, const char *msg, ...)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
   char * message;
 
   va_list args;
@@ -283,20 +217,16 @@ static void error_func(void * ctx, const char *msg, ...)
   vasprintf(&message, msg, args);
   va_end(args);
 
-  rb_funcall(doc, rb_intern("error"), 1,
-      NOKOGIRI_STR_NEW2(message, STRING_OR_NULL(enc))
-  );
+  rb_funcall(doc, id_error, 1, NOKOGIRI_STR_NEW2(message));
   free(message);
 }
 
 static void cdata_block(void * ctx, const xmlChar * value, int len)
 {
-  VALUE self = (VALUE)ctx;
-  VALUE MAYBE_UNUSED(enc) = rb_iv_get(self, "@encoding");
-  VALUE doc = rb_funcall(self, rb_intern("document"), 0);
-  VALUE string =
-    NOKOGIRI_STR_NEW(value, len, STRING_OR_NULL(enc));
-  rb_funcall(doc, rb_intern("cdata_block"), 1, string);
+  VALUE self = NOKOGIRI_SAX_SELF(ctx);
+  VALUE doc = rb_iv_get(self, "@document");
+  VALUE string = NOKOGIRI_STR_NEW(value, len);
+  rb_funcall(doc, id_cdata_block, 1, string);
 }
 
 static void deallocate(xmlSAXHandlerPtr handler)
@@ -339,7 +269,18 @@ void init_xml_sax_parser()
   cNokogiriXmlSaxParser = klass;
 
   rb_define_alloc_func(klass, allocate);
-  rb_define_method(klass, "parse_memory", parse_memory, 1);
-  rb_define_private_method(klass, "native_parse_file", native_parse_file, 1);
-  rb_define_private_method(klass, "native_parse_io", native_parse_io, 2);
+
+  id_start_document = rb_intern("start_document");
+  id_end_document   = rb_intern("end_document");
+  id_start_element  = rb_intern("start_element");
+  id_end_element    = rb_intern("end_element");
+  id_comment        = rb_intern("comment");
+  id_characters     = rb_intern("characters");
+  id_xmldecl        = rb_intern("xmldecl");
+  id_error          = rb_intern("error");
+  id_warning        = rb_intern("warning");
+  id_cdata_block    = rb_intern("cdata_block");
+  id_cAttribute     = rb_intern("Attribute");
+  id_start_element_namespace = rb_intern("start_element_namespace");
+  id_end_element_namespace = rb_intern("end_element_namespace");
 }