libxml-ruby 2.1.2-x86-mingw32 → 2.2.0-x86-mingw32

data/HISTORY

@@ -1,5 +1,12 @@
 = Release History
 
+== 2.2.0 / 2011-08-09 Charlie Savage
+
+* Update encoding support for Ruby 1.9 so that libxml-ruby returns
+  strings encoded in UTF-8.  This change was required since libxml internally
+  stores strings in UTF-8.  The exceptions to this rule are the #to_s methods
+  which return UTF-8 by default but can return other encodings if requested.
+
 == 2.1.2 / 2011-08-03 Charlie Savage
 
 * Fix segmentation fault that could occur when an XPathContext was marked

data/ext/libxml/libxml.c

@@ -21,21 +21,6 @@ static void rxml_init_memory(void)
   );*/
 }
 
-VALUE rxml_str_new2(const char* xstr, const char* xencoding)
-{
-#ifdef HAVE_RUBY_ENCODING_H
-  if (xencoding)
-  {
-    xmlCharEncoding xmlEncoding = xmlParseCharEncoding(xencoding);
-    VALUE encoding = rxml_xml_encoding_to_rb_encoding(mXMLEncoding, xmlEncoding);
-    rb_encoding* xencodingPtr = (rb_encoding*) RDATA(encoding)->data;
-    return rb_external_str_new_with_enc(xstr, strlen(xstr), xencodingPtr);
-  }
-#endif
-  return rb_str_new2(xstr);
-}
-
-
 void Init_libxml_ruby(void)
 {
 /* The libxml gem provides Ruby language bindings for GNOME's Libxml2

data/ext/libxml/ruby_libxml.h

@@ -26,9 +26,6 @@
 #include <ruby/encoding.h>
 #endif
 
-// Ruby 1.8/1.9 encoding compatibility
-VALUE rxml_str_new2(const char* xstr, const char* xencoding);
-
 #ifdef LIBXML_DEBUG_ENABLED
 #include <libxml/xpathInternals.h>
 #endif

data/ext/libxml/ruby_xml_attr.c

@@ -166,7 +166,7 @@ static VALUE rxml_attr_name_get(VALUE self)
   if (xattr->name == NULL)
     return Qnil;
   else
-    return rxml_str_new2((const char*) xattr->name, (xattr->doc ? xattr->doc->encoding : NULL));
+    return rxml_new_cstr((const char*) xattr->name, NULL);
 }
 
 /*
@@ -285,7 +285,7 @@ VALUE rxml_attr_value_get(VALUE self)
 
   if (value != NULL)
   {
-    result = rxml_str_new2((const char*) value, (xattr->doc ? xattr->doc->encoding : NULL));
+    result = rxml_new_cstr((const char*) value, NULL);
     xmlFree(value);
   }
   return result;

data/ext/libxml/ruby_xml_attr_decl.c

@@ -54,7 +54,7 @@ static VALUE rxml_attr_decl_name_get(VALUE self)
   if (xattr->name == NULL)
     return Qnil;
   else
-    return rxml_str_new2((const char*) xattr->name, xattr->doc->encoding);
+    return rxml_new_cstr((const char*) xattr->name, xattr->doc->encoding);
 }
 
 /*
@@ -135,7 +135,7 @@ VALUE rxml_attr_decl_value_get(VALUE self)
   Data_Get_Struct(self, xmlAttribute, xattr);
 
   if (xattr->defaultValue)
-    return rxml_str_new2((const char *)xattr->defaultValue, xattr->doc->encoding);
+    return rxml_new_cstr((const char *)xattr->defaultValue, NULL);
   else
     return Qnil;
 }

data/ext/libxml/ruby_xml_document.c

@@ -174,7 +174,7 @@ static VALUE rxml_document_canonicalize(int argc, VALUE *argv, VALUE self)
 
   if (buffer)
   {
-    result = rxml_str_new2((const char*) buffer, (const char*)xdoc->encoding);
+    result = rxml_new_cstr((const char*) buffer, NULL);
     xmlFree(buffer);
   }
 
@@ -349,10 +349,12 @@ static VALUE rxml_document_rb_encoding_get(VALUE self)
 {
   xmlDocPtr xdoc;
   const char *xencoding;
+  rb_encoding* rbencoding;
   Data_Get_Struct(self, xmlDoc, xdoc);
 
   xencoding = (const char*)xdoc->encoding;
-  return rxml_xml_encoding_to_rb_encoding(mXMLEncoding, xmlParseCharEncoding(xencoding));
+  rbencoding = rxml_xml_encoding_to_rb_encoding(mXMLEncoding, xmlParseCharEncoding(xencoding));
+  return rb_enc_from_encoding(rbencoding);
 }
 #endif 
 
@@ -734,7 +736,7 @@ static VALUE rxml_document_to_s(int argc, VALUE *argv, VALUE self)
   Data_Get_Struct(self, xmlDoc, xdoc);
   xmlDocDumpFormatMemoryEnc(xdoc, &buffer, &length, xencoding, indent);
 
-  result = rxml_str_new2((const char*) buffer, xencoding);
+  result = rxml_new_cstr((const char*) buffer, xencoding);
   xmlFree(buffer);
   return result;
 }
@@ -753,7 +755,7 @@ static VALUE rxml_document_url_get(VALUE self)
   if (xdoc->URL == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) xdoc->URL, xdoc->encoding));
+    return (rxml_new_cstr((const char*) xdoc->URL, NULL));
 }
 
 /*
@@ -770,7 +772,7 @@ static VALUE rxml_document_version_get(VALUE self)
   if (xdoc->version == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) xdoc->version, xdoc->encoding));
+    return (rxml_new_cstr((const char*) xdoc->version, NULL));
 }
 
 /*

data/ext/libxml/ruby_xml_dtd.c

@@ -90,7 +90,7 @@ static VALUE rxml_dtd_external_id_get(VALUE self)
   if (xdtd->ExternalID == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) xdtd->ExternalID, xdtd->doc ? xdtd->doc->encoding : NULL));
+    return (rxml_new_cstr((const char*) xdtd->ExternalID, NULL));
 }
 
 /*
@@ -108,7 +108,7 @@ static VALUE rxml_dtd_name_get(VALUE self)
   if (xdtd->name == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) xdtd->name, xdtd->doc ? xdtd->doc->encoding : NULL));
+    return (rxml_new_cstr((const char*) xdtd->name, NULL));
 }
 
 
@@ -127,7 +127,7 @@ static VALUE rxml_dtd_uri_get(VALUE self)
   if (xdtd->SystemID == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) xdtd->SystemID, xdtd->doc ? xdtd->doc->encoding : NULL));
+    return (rxml_new_cstr((const char*) xdtd->SystemID, NULL));
 }
 
 /*

data/ext/libxml/ruby_xml_encoding.c

@@ -71,7 +71,7 @@ static VALUE rxml_encoding_to_s(VALUE klass, VALUE encoding)
   if (!xencoding)
     return Qnil;
   else
-    return rxml_str_new2(xencoding, xencoding);
+    return rxml_new_cstr(xencoding, xencoding);
 }
 
 #ifdef HAVE_RUBY_ENCODING_H
@@ -79,7 +79,7 @@ static VALUE rxml_encoding_to_s(VALUE klass, VALUE encoding)
  * Converts an xmlCharEncoding enum value into a Ruby Encoding object (available
  * on Ruby 1.9.* and higher).
  */
-VALUE rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding)
+rb_encoding* rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding)
 {
   const char* encodingName;
 
@@ -148,7 +148,7 @@ VALUE rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding)
       break;
   }
 
-  return rb_enc_from_encoding(rb_enc_find(encodingName));
+  return rb_enc_find(encodingName);
 }
 
 /*
@@ -160,10 +160,31 @@ VALUE rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding)
  */
 VALUE rxml_encoding_to_rb_encoding(VALUE klass, VALUE encoding)
 {
-  return rxml_xml_encoding_to_rb_encoding(klass, NUM2INT(encoding));
+  xmlCharEncoding xmlEncoding = (xmlCharEncoding)NUM2INT(encoding);
+  rb_encoding* rbencoding = rxml_xml_encoding_to_rb_encoding(klass, xmlEncoding);
+  return rb_enc_from_encoding(rbencoding);
 }
 #endif
 
+
+VALUE rxml_new_cstr(const char* xstr, const char* xencoding)
+{
+#ifdef HAVE_RUBY_ENCODING_H
+  rb_encoding* rbencoding;
+  if (xencoding)
+  {
+    xmlCharEncoding xmlEncoding = xmlParseCharEncoding(xencoding);
+    rbencoding = rxml_xml_encoding_to_rb_encoding(mXMLEncoding, xmlEncoding);
+  }
+  else
+  {
+    rbencoding = rb_utf8_encoding();
+  }
+  return rb_external_str_new_with_enc(xstr, strlen(xstr), rbencoding);
+#endif
+  return rb_str_new2(xstr);
+}
+
 void rxml_init_encoding(void)
 {
   mXMLEncoding = rb_define_module_under(mXML, "Encoding");

data/ext/libxml/ruby_xml_encoding.h

@@ -7,8 +7,11 @@ extern VALUE mXMLEncoding;
 
 void rxml_init_encoding();
 
+// Ruby 1.8/1.9 encoding compatibility
+VALUE rxml_new_cstr(const char* xstr, const char* xencoding);
+
 #ifdef HAVE_RUBY_ENCODING_H
-VALUE rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding);
+rb_encoding* rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding);
 #endif
 
 #endif

data/ext/libxml/ruby_xml_namespace.c

@@ -34,15 +34,6 @@ VALUE rxml_namespace_wrap(xmlNsPtr xns)
   return Data_Wrap_Struct(cXMLNamespace, NULL, NULL, xns);
 }
 
-static VALUE rxml_namespace_string(xmlNsPtr xns, const char* buffer)
-{
-   const char* xencoding = NULL;
-#if LIBXML_VERSION >= 20628
-   xencoding = xns->context ? xns->context->encoding : NULL;
-#endif
-  return rxml_str_new2(buffer, xencoding);
-}
-
 
 /*
  * call-seq:
@@ -90,7 +81,7 @@ static VALUE rxml_namespace_href_get(VALUE self)
   if (xns->href == NULL)
     return Qnil;
   else
-    return rxml_namespace_string(xns, (const char*) xns->href);
+    return rxml_new_cstr((const char*) xns->href, NULL);
 }
 
 /*
@@ -125,7 +116,7 @@ static VALUE rxml_namespace_prefix_get(VALUE self)
   if (xns->prefix == NULL)
     return Qnil;
   else
-    return rxml_namespace_string(xns, (const char*) xns->prefix);
+    return rxml_new_cstr((const char*) xns->prefix, NULL);
 }
 
 /*

data/ext/libxml/ruby_xml_node.c

@@ -300,7 +300,7 @@ static VALUE rxml_node_base_uri_get(VALUE self)
   base_uri = xmlNodeGetBase(xnode->doc, xnode);
   if (base_uri)
   {
-    result = rxml_str_new2((const char*) base_uri, xnode->doc ? xnode->doc->encoding : NULL);
+    result = rxml_new_cstr((const char*) base_uri, NULL);
     xmlFree(base_uri);
   }
 
@@ -344,7 +344,7 @@ static VALUE rxml_node_content_get(VALUE self)
   content = xmlNodeGetContent(xnode);
   if (content)
   {
-    result = rxml_str_new2((const char *) content, xnode->doc ? xnode->doc->encoding : NULL);
+    result = rxml_new_cstr((const char *) content, NULL);
     xmlFree(content);
   }
 
@@ -391,7 +391,7 @@ static VALUE rxml_node_content_stripped_get(VALUE self)
   content = xmlNodeGetContent(xnode);
   if (content)
   {
-    result = rxml_str_new2((const char*) content, xnode->doc ? xnode->doc->encoding : NULL);
+    result = rxml_new_cstr((const char*) content, NULL);
     xmlFree(content);
   }
   return (result);
@@ -582,9 +582,9 @@ static VALUE rxml_node_to_s(int argc, VALUE *argv, VALUE self)
   xmlOutputBufferFlush(output);
 
   if (output->conv)
-    result = rxml_str_new2((const char*) output->conv->content, xencoding);
+    result = rxml_new_cstr((const char*) output->conv->content, xencoding);
   else
-    result = rxml_str_new2((const char*) output->buffer->content, xencoding);
+    result = rxml_new_cstr((const char*) output->buffer->content, xencoding);
 
   xmlOutputBufferClose(output);
   
@@ -688,7 +688,7 @@ static VALUE rxml_node_lang_get(VALUE self)
 
   if (lang)
   {
-    result = rxml_str_new2((const char*) lang, xnode->doc ? xnode->doc->encoding : NULL);
+    result = rxml_new_cstr((const char*) lang, NULL);
     xmlFree(lang);
   }
 
@@ -821,11 +821,11 @@ static VALUE rxml_node_xlink_type_name(VALUE self)
   case XLINK_TYPE_NONE:
     return (Qnil);
   case XLINK_TYPE_SIMPLE:
-    return (rxml_str_new2("simple", xnode->doc ? xnode->doc->encoding : NULL));
+    return (rxml_new_cstr("simple", NULL));
   case XLINK_TYPE_EXTENDED:
-    return (rxml_str_new2("extended", xnode->doc ? xnode->doc->encoding : NULL));
+    return (rxml_new_cstr("extended", NULL));
   case XLINK_TYPE_EXTENDED_SET:
-    return (rxml_str_new2("extended_set", xnode->doc ? xnode->doc->encoding : NULL));
+    return (rxml_new_cstr("extended_set", NULL));
   default:
     rb_fatal("Unknowng xlink type, %d", xlt);
   }
@@ -876,7 +876,7 @@ static VALUE rxml_node_name_get(VALUE self)
   if (xnode->name == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) name, xnode->doc ? xnode->doc->encoding : NULL));
+    return (rxml_new_cstr((const char*) name, NULL));
 }
 
 /*
@@ -968,7 +968,7 @@ static VALUE rxml_node_path(VALUE self)
   if (path == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) path, xnode->doc ? xnode->doc->encoding : NULL));
+    return (rxml_new_cstr((const char*) path, NULL));
 }
 
 /*

data/ext/libxml/ruby_xml_parser_context.c

@@ -188,7 +188,7 @@ static VALUE rxml_parser_context_base_uri_get(VALUE self)
   Data_Get_Struct(self, xmlParserCtxt, ctxt);
 
   if (ctxt->input && ctxt->input->filename)
-    return rxml_str_new2(ctxt->input->filename, ctxt->encoding);
+    return rxml_new_cstr(ctxt->input->filename, ctxt->encoding);
   else
     return Qnil;
 }
@@ -229,7 +229,7 @@ static VALUE rxml_parser_context_data_directory_get(VALUE self)
   if (ctxt->directory == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2(ctxt->directory, ctxt->encoding));
+    return (rxml_new_cstr(ctxt->directory, ctxt->encoding));
 }
 
 /*
@@ -487,7 +487,7 @@ static VALUE rxml_parser_context_name_node_get(VALUE self)
   if (ctxt->name == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) ctxt->name, ctxt->encoding));
+    return (rxml_new_cstr((const char*) ctxt->name, ctxt->encoding));
 }
 
 /*
@@ -514,7 +514,7 @@ static VALUE rxml_parser_context_name_tab_get(VALUE self)
     if (ctxt->nameTab[i] == NULL)
       continue;
     else
-      rb_ary_push(tab_ary, rxml_str_new2((const char*) ctxt->nameTab[i], ctxt->encoding));
+      rb_ary_push(tab_ary, rxml_new_cstr((const char*) ctxt->nameTab[i], ctxt->encoding));
   }
 
   return (tab_ary);
@@ -780,7 +780,7 @@ static VALUE rxml_parser_context_subset_name_get(VALUE self)
   if (ctxt->intSubName == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) ctxt->intSubName, ctxt->encoding));
+    return (rxml_new_cstr((const char*) ctxt->intSubName, ctxt->encoding));
 }
 
 /*
@@ -799,7 +799,7 @@ static VALUE rxml_parser_context_subset_external_uri_get(VALUE self)
   if (ctxt->extSubURI == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) ctxt->extSubURI, ctxt->encoding));
+    return (rxml_new_cstr((const char*) ctxt->extSubURI, ctxt->encoding));
 }
 
 /*
@@ -818,7 +818,7 @@ static VALUE rxml_parser_context_subset_external_system_id_get(VALUE self)
   if (ctxt->extSubSystem == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) ctxt->extSubSystem, ctxt->encoding));
+    return (rxml_new_cstr((const char*) ctxt->extSubSystem, ctxt->encoding));
 }
 
 /*
@@ -903,7 +903,7 @@ static VALUE rxml_parser_context_version_get(VALUE self)
   if (ctxt->version == NULL)
     return (Qnil);
   else
-    return (rxml_str_new2((const char*) ctxt->version, ctxt->encoding));
+    return (rxml_new_cstr((const char*) ctxt->version, ctxt->encoding));
 }
 
 /*

data/ext/libxml/ruby_xml_reader.c

@@ -489,7 +489,7 @@ static VALUE rxml_reader_read_inner_xml(VALUE self)
   if (xml)
   {
     const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
-    result = rxml_str_new2((const char*) xml, xencoding);
+    result = rxml_new_cstr((const char*) xml, xencoding);
     xmlFree(xml);
   }
 
@@ -515,7 +515,7 @@ static VALUE rxml_reader_read_outer_xml(VALUE self)
   if (xml)
   {
     const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
-    result = rxml_str_new2((const char*) xml, xencoding);
+    result = rxml_new_cstr((const char*) xml, xencoding);
     xmlFree(xml);
   }
 
@@ -553,7 +553,7 @@ static VALUE rxml_reader_read_string(VALUE self)
   if (xml)
   {
     const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
-    result = rxml_str_new2((const char*) xml, xencoding);
+    result = rxml_new_cstr((const char*) xml, xencoding);
     xmlFree(xml);
   }
 
@@ -612,7 +612,7 @@ static VALUE rxml_reader_name(VALUE self)
   const xmlChar *result = xmlTextReaderConstName(xReader);
   const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
 
-  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
+  return (result == NULL ? Qnil : rxml_new_cstr(result, xencoding));
 }
 
 /*
@@ -627,7 +627,7 @@ static VALUE rxml_reader_local_name(VALUE self)
   const xmlChar *result = xmlTextReaderConstLocalName(xReader);
   const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
 
-  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
+  return (result == NULL ? Qnil : rxml_new_cstr(result, xencoding));
 }
 
 /*
@@ -680,7 +680,7 @@ static VALUE rxml_reader_base_uri(VALUE self)
   const xmlChar *result = xmlTextReaderConstBaseUri(xReader);
   const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
 
-  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
+  return (result == NULL ? Qnil : rxml_new_cstr(result, xencoding));
 }
 
 /*
@@ -695,7 +695,7 @@ static VALUE rxml_reader_namespace_uri(VALUE self)
   const xmlChar *result = xmlTextReaderConstNamespaceUri(xReader);
   const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
 
-  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
+  return (result == NULL ? Qnil : rxml_new_cstr(result, xencoding));
 }
 
 /*
@@ -710,7 +710,7 @@ static VALUE rxml_reader_value(VALUE self)
   const xmlChar *result = xmlTextReaderConstValue(xReader);
   const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
 
-  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
+  return (result == NULL ? Qnil : rxml_new_cstr(result, xencoding));
 }
 
 /*
@@ -725,7 +725,7 @@ static VALUE rxml_reader_prefix(VALUE self)
   const xmlChar *result = xmlTextReaderConstPrefix(xReader);
   const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
 
-  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
+  return (result == NULL ? Qnil : rxml_new_cstr(result, xencoding));
 }
 
 /*
@@ -781,7 +781,7 @@ static VALUE rxml_reader_xml_lang(VALUE self)
   const xmlChar *result = xmlTextReaderConstXmlLang(xReader);
   const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
 
-  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
+  return (result == NULL ? Qnil : rxml_new_cstr(result, xencoding));
 }
 
 /*
@@ -796,7 +796,7 @@ static VALUE rxml_reader_xml_version(VALUE self)
   const xmlChar *result = xmlTextReaderConstXmlVersion(xReader);
   const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
 
-  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
+  return (result == NULL ? Qnil : rxml_new_cstr(result, xencoding));
 }
 
 /*
@@ -849,7 +849,7 @@ static VALUE rxml_reader_attribute(VALUE self, VALUE key)
 
   if (xattr)
   {
-    result = rxml_str_new2(xattr, xencoding);
+    result = rxml_new_cstr(xattr, xencoding);
     xmlFree(xattr);
   }
   return result;
@@ -871,7 +871,7 @@ static VALUE rxml_reader_lookup_namespace(VALUE self, VALUE prefix)
 
   if (xnamespace)
   {
-    result = rxml_str_new2((const char*)xnamespace, (const char*)xencoding);
+    result = rxml_new_cstr((const char*)xnamespace, (const char*)xencoding);
     xmlFree((void *)xnamespace);
   }
   return result;

data/ext/libxml/ruby_xml_version.h

@@ -1,9 +1,9 @@
 /* Don't nuke this block!  It is used for automatically updating the
  * versions below. VERSION = string formatting, VERNUM = numbered
  * version for inline testing: increment both or none at all.*/
-#define RUBY_LIBXML_VERSION  "2.1.2"
-#define RUBY_LIBXML_VERNUM   212
+#define RUBY_LIBXML_VERSION  "2.2.0"
+#define RUBY_LIBXML_VERNUM   220
 #define RUBY_LIBXML_VER_MAJ   2
-#define RUBY_LIBXML_VER_MIN   1
-#define RUBY_LIBXML_VER_MIC   2
+#define RUBY_LIBXML_VER_MIN   2
+#define RUBY_LIBXML_VER_MIC   0
 #define RUBY_LIBXML_VER_PATCH 0

data/ext/libxml/ruby_xml_xpath.c

@@ -104,7 +104,7 @@ rxml_xpath_to_value(xmlXPathContextPtr xctxt, xmlXPathObjectPtr xobject) {
       xmlXPathFreeObject(xobject);
       break;
     case XPATH_STRING:
-      result = rxml_str_new2((const char*)xobject->stringval, xctxt->doc->encoding);
+      result = rxml_new_cstr((const char*)xobject->stringval, xctxt->doc->encoding);
       xmlXPathFreeObject(xobject);
       break;
     default:

data/ext/libxml/ruby_xml_xpath_context.c

@@ -159,8 +159,8 @@ static VALUE rxml_xpath_context_register_namespaces_from_node(VALUE self,
        Skip it for now. */
       if (xns->prefix)
       {
-        VALUE prefix = rxml_str_new2((const char*)xns->prefix, xctxt->doc->encoding);
-        VALUE uri = rxml_str_new2((const char*)xns->href, xctxt->doc->encoding);
+        VALUE prefix = rxml_new_cstr((const char*)xns->prefix, xctxt->doc->encoding);
+        VALUE uri = rxml_new_cstr((const char*)xns->href, xctxt->doc->encoding);
         rxml_xpath_context_register_namespace(self, prefix, uri);
       }
       xns = xns->next;
@@ -215,7 +215,7 @@ static VALUE rxml_xpath_context_register_namespaces(VALUE self, VALUE nslist)
     {
       rprefix = rb_str_new(StringValuePtr(nslist), (int) ((long) cp
           - (long) StringValuePtr(nslist)));
-      ruri = rxml_str_new2(&cp[1], xctxt->doc->encoding);
+      ruri = rxml_new_cstr(&cp[1], xctxt->doc->encoding);
     }
     /* Should test the results of this */
     rxml_xpath_context_register_namespace(self, rprefix, ruri);

data/ext/libxml/ruby_xml_xpath_object.c

@@ -292,7 +292,7 @@ static VALUE rxml_xpath_object_string(VALUE self)
   if (rxpop->xpop->stringval == NULL)
     return Qnil;
 
-  return rxml_str_new2((const char*) rxpop->xpop->stringval, rxpop->xdoc->encoding);
+  return rxml_new_cstr((const char*) rxpop->xpop->stringval, rxpop->xdoc->encoding);
 }
 
 /*

data/test/new_main.rb

@@ -0,0 +1,29 @@
+# encoding: utf-8
+
+string1 = "(ISC)²"
+puts string1.encoding
+puts string1.bytes.to_a.join(' ')
+puts string1
+
+puts ""
+string2 = string1.encode('iso-8859-1')
+puts string2.encoding
+puts string2.bytes.to_a.join(' ')
+puts string2
+
+puts ""
+string3 = string1.force_encoding('iso-8859-1').encode('utf-8')
+puts string3.encoding
+puts string3.bytes.to_a.join(' ')
+puts string3
+
+#UTF-8
+#40 73 83 67 41 194 178
+#(ISC)²
+#
+#ISO-8859-1
+#40 73 83 67 41 178
+#(ISC)²
+
+
+#40 73 83 67 41 195 130 194 178

data/test/tc_attr.rb

@@ -48,7 +48,7 @@ class AttrNodeTest < Test::Unit::TestCase
   def test_name
     attribute = city_member.attributes.get_attribute('name')
     assert_equal('name', attribute.name)
-    assert_equal(Encoding::ASCII_8BIT, attribute.name.encoding) if defined?(Encoding)
+    assert_equal(Encoding::UTF_8, attribute.name.encoding) if defined?(Encoding)
 
     attribute = city_member.attributes.get_attribute('href')
     assert_equal('href', attribute.name)
@@ -64,7 +64,7 @@ class AttrNodeTest < Test::Unit::TestCase
   def test_value
     attribute = city_member.attributes.get_attribute('name')
     assert_equal('Cambridge', attribute.value)
-    assert_equal(Encoding::ASCII_8BIT, attribute.value.encoding) if defined?(Encoding)
+    assert_equal(Encoding::UTF_8, attribute.value.encoding) if defined?(Encoding)
 
     attribute = city_member.attributes.get_attribute('href')
     assert_equal('http://www.foo.net/cgi-bin/wfs?FeatureID=C10239', attribute.value)
@@ -74,12 +74,12 @@ class AttrNodeTest < Test::Unit::TestCase
     attribute = city_member.attributes.get_attribute('name')
     attribute.value = 'London'
     assert_equal('London', attribute.value)
-    assert_equal(Encoding::ASCII_8BIT, attribute.value.encoding) if defined?(Encoding)
+    assert_equal(Encoding::UTF_8, attribute.value.encoding) if defined?(Encoding)
 
     attribute = city_member.attributes.get_attribute('href')
     attribute.value = 'http://i.have.changed'
     assert_equal('http://i.have.changed', attribute.value)
-    assert_equal(Encoding::ASCII_8BIT, attribute.value.encoding) if defined?(Encoding)
+    assert_equal(Encoding::UTF_8, attribute.value.encoding) if defined?(Encoding)
   end
 
   def test_set_nil

data/test/tc_document.rb

@@ -88,20 +88,6 @@ class TestDocument < Test::Unit::TestCase
                  @doc.to_s(:indent => false))
   end
 
-  def test_encoding
-    doc = XML::Document.new
-    assert_equal(XML::Encoding::NONE, doc.encoding)
-    assert_equal(Encoding::ASCII_8BIT, doc.rb_encoding) if defined?(Encoding)
-
-    file = File.expand_path(File.join(File.dirname(__FILE__), 'model/bands.xml'))
-    doc = XML::Document.file(file)
-    assert_equal(XML::Encoding::UTF_8, doc.encoding)
-    assert_equal(Encoding::UTF_8, doc.rb_encoding) if defined?(Encoding)
-
-    doc.encoding = XML::Encoding::ISO_8859_1
-    assert_equal(XML::Encoding::ISO_8859_1, doc.encoding)
-    assert_equal(Encoding::ISO8859_1, doc.rb_encoding) if defined?(Encoding)
-  end
 
   def test_doc_node_type
     assert_equal(XML::Node::DOCUMENT_NODE, XML::Document.new.node_type)

data/test/tc_document_write.rb

@@ -6,17 +6,7 @@ require 'test/unit'
 
 class TestDocumentWrite < Test::Unit::TestCase
   def setup
-    load_encoding("utf-8")
-  end
-
-  def teardown
-    XML.default_keep_blanks = true
-    @doc = nil
-  end
-
-  def load_encoding(name)
-    @encoding = Encoding.find(name) if defined?(Encoding)
-    @file_name = "model/bands.#{name.downcase}.xml"
+    @file_name = "model/bands.utf-8.xml"
 
     # Strip spaces to make testing easier
     XML.default_keep_blanks = false
@@ -24,6 +14,11 @@ class TestDocumentWrite < Test::Unit::TestCase
     @doc = XML::Document.file(file)
   end
 
+  def teardown
+    XML.default_keep_blanks = true
+    @doc = nil
+  end
+
   # ---  to_s tests  ---
   def test_to_s_default
     # Default to_s has indentation

data/test/tc_encoding.rb

@@ -0,0 +1,119 @@
+# encoding: UTF-8
+
+if defined?(Encoding)
+require './test_helper'
+require 'test/unit'
+
+# Code  UTF8        Latin1      Hex
+# m      109          109        6D
+# ö      195 182      246        C3 B6 / F6
+# t      116          116        74
+# l      108          108        6C
+# e      101          101        65
+# y      121          121        79
+# _       95           95        5F
+# c       99           99        63
+# r      114          114        72
+# ü      195 188      252        C3 BC / FC
+# e      101          101        65
+
+# See:
+#  http://en.wikipedia.org/wiki/ISO/IEC_8859-1
+#  http://en.wikipedia.org/wiki/List_of_Unicode_characters
+
+class TestEncoding < Test::Unit::TestCase
+  def setup
+    Encoding.default_internal = nil
+  end
+
+  def load_encoding(encoding)
+    @encoding = encoding
+    file_name = "model/bands.#{@encoding.name.downcase}.xml"
+
+    # Strip spaces to make testing easier
+    XML.default_keep_blanks = false
+    file = File.join(File.dirname(__FILE__), file_name)
+    @doc = XML::Document.file(file)
+  end
+  
+  def test_encoding
+    doc = XML::Document.new
+    assert_equal(XML::Encoding::NONE, doc.encoding)
+    assert_equal(Encoding::ASCII_8BIT, doc.rb_encoding) if defined?(Encoding)
+
+    file = File.expand_path(File.join(File.dirname(__FILE__), 'model/bands.xml'))
+    doc = XML::Document.file(file)
+    assert_equal(XML::Encoding::UTF_8, doc.encoding)
+    assert_equal(Encoding::UTF_8, doc.rb_encoding) if defined?(Encoding)
+
+    doc.encoding = XML::Encoding::ISO_8859_1
+    assert_equal(XML::Encoding::ISO_8859_1, doc.encoding)
+    assert_equal(Encoding::ISO8859_1, doc.rb_encoding) if defined?(Encoding)
+  end
+
+  def test_no_internal_encoding_iso_8859_1
+    Encoding.default_internal = nil
+    load_encoding(Encoding::ISO_8859_1)
+    node = @doc.root.children.first
+
+    name = node.name
+    assert_equal(Encoding::UTF_8, name.encoding)
+    assert_equal("m\u00F6tley_cr\u00FCe", name)
+    assert_equal("109 195 182 116 108 101 121 95 99 114 195 188 101",
+                 name.bytes.to_a.join(" "))
+    assert_equal("M\u00F6tley Cr\u00FCe is an American heavy metal band formed in Los Angeles, California in 1981.",
+                 node.content)
+
+    name = name.encode(Encoding::ISO_8859_1)
+    assert_equal(Encoding::ISO_8859_1, name.encoding)
+    assert_equal("m\xF6tley_cr\xFCe".force_encoding(Encoding::ISO_8859_1), name)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+    assert_equal("M\xF6tley Cr\xFCe is an American heavy metal band formed in Los Angeles, California in 1981.".force_encoding(Encoding::ISO_8859_1),
+                node.content.encode(Encoding::ISO_8859_1))
+  end
+
+  def test_internal_encoding_iso_8859_1
+    Encoding.default_internal = Encoding::ISO_8859_1
+    load_encoding(Encoding::ISO_8859_1)
+    node = @doc.root.children.first
+
+    name = node.name
+    assert_equal(Encoding::ISO_8859_1, name.encoding)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+    assert_equal("m\xF6tley_cr\xFCe".force_encoding(Encoding::ISO_8859_1), name)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+    assert_equal("M\xF6tley Cr\xFCe is an American heavy metal band formed in Los Angeles, California in 1981.".force_encoding(Encoding::ISO_8859_1),
+                node.content.encode(Encoding::ISO_8859_1))
+  end
+
+  def test_no_internal_encoding_utf_8
+    Encoding.default_internal = nil
+    load_encoding(Encoding::UTF_8)
+    node = @doc.root.children.first
+
+    name = node.name
+    assert_equal(@encoding, name.encoding)
+    assert_equal("109 195 182 116 108 101 121 95 99 114 195 188 101",
+                 name.bytes.to_a.join(" "))
+
+    name = name.encode(Encoding::ISO_8859_1)
+    assert_equal(Encoding::ISO_8859_1, name.encoding)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+  end
+
+  def test_internal_encoding_utf_8
+    Encoding.default_internal = Encoding::ISO_8859_1
+    load_encoding(Encoding::UTF_8)
+    node = @doc.root.children.first
+
+    name = node.name
+    assert_equal(Encoding::ISO_8859_1, name.encoding)
+    assert_equal("109 246 116 108 101 121 95 99 114 252 101",
+                 name.bytes.to_a.join(" "))
+  end
+end
+end

data/test/tc_node.rb

@@ -5,23 +5,18 @@ require 'test/unit'
 
 class TestNode < Test::Unit::TestCase
   def setup
-    load_encoding("utf-8")
-  end
-  
-  def teardown
-    XML.default_keep_blanks = true
-    @doc = nil
-  end
-
-  def load_encoding(name)
-    @encoding = Encoding.find(name) if defined?(Encoding)
-    @file_name = "model/bands.#{name.downcase}.xml"
+    @file_name = "model/bands.utf-8.xml"
 
     # Strip spaces to make testing easier
     XML.default_keep_blanks = false
     file = File.join(File.dirname(__FILE__), @file_name)
     @doc = XML::Document.file(file)
   end
+  
+  def teardown
+    XML.default_keep_blanks = true
+    @doc = nil
+  end
 
   def nodes
     # Find all nodes with a country attributes
@@ -65,7 +60,7 @@ class TestNode < Test::Unit::TestCase
     assert_instance_of(XML::Node, @doc.root.child)
 
     if defined?(Encoding)
-      assert_equal(@encoding, @doc.root.child.name.encoding)
+      assert_equal(Encoding::UTF_8, @doc.root.child.name.encoding)
       assert_equal("m\u00F6tley_cr\u00FCe", @doc.root.child.name)
     else
       assert_equal("m\303\266tley_cr\303\274e", @doc.root.child.name)
@@ -79,25 +74,8 @@ class TestNode < Test::Unit::TestCase
   end
   
   def test_name
-    if defined?(Encoding)
-      assert_equal(@encoding, nodes[0].name.encoding)
-      assert_equal("m\u00F6tley_cr\u00FCe", nodes[0].name)
-    else
-      assert_equal("m\303\266tley_cr\303\274e", nodes[0].name)
-    end
-    assert_equal("iron_maiden", nodes[1].name)
-  end
-
-  def test_name_iso_8859_1
-    load_encoding("iso-8859-1")
-    
-    if defined?(Encoding)
-      assert_equal(@encoding, nodes[0].name.encoding)
-      assert_equal("m\303\266tley_cr\303\274e".force_encoding(@encoding), nodes[0].name)
-    else
-      assert_equal("m\303\266tley_cr\303\274e", nodes[0].name)
-    end
-    assert_equal("iron_maiden", nodes[1].name)
+    node = @doc.root.children.last
+    assert_equal("iron_maiden", node.name)
   end
 
   def test_node_find
@@ -140,47 +118,9 @@ class TestNode < Test::Unit::TestCase
   end
 
   def test_content
-    if defined?(Encoding)
-      assert_equal(@encoding, @doc.root.content.encoding)
-      assert_equal("M\u00F6tley Cr\u00FCe is an American heavy metal band formed in Los Angeles, California in 1981.Iron Maiden is a British heavy metal band formed in 1975.",
-                   @doc.root.content)
-
-      first = @doc.root.child
-      assert_equal("M\u00F6tley Cr\u00FCe is an American heavy metal band formed in Los Angeles, California in 1981.",
-                   first.content)
-      assert_equal("Iron Maiden is a British heavy metal band formed in 1975.", first.next.content)
-    else
-      assert_equal("M\303\266tley Cr\303\274e is an American heavy metal band formed in Los Angeles, California in 1981.Iron Maiden is a British heavy metal band formed in 1975.",
-                   @doc.root.content)
-
-      first = @doc.root.child
-      assert_equal("M\303\266tley Cr\303\274e is an American heavy metal band formed in Los Angeles, California in 1981.",
-                   first.content)
-      assert_equal("Iron Maiden is a British heavy metal band formed in 1975.",
-                   first.next.content)
-    end
-  end
-
-  def test_content_iso_8859_1
-    load_encoding("iso-8859-1")
-    if defined?(Encoding)
-      assert_equal(@encoding, @doc.root.content.encoding)
-      assert_equal("M\xC3\xB6tley Cr\xC3\xBCe is an American heavy metal band formed in Los Angeles, California in 1981.Iron Maiden is a British heavy metal band formed in 1975.".force_encoding(@encoding),
-                   @doc.root.content)
-
-      first = @doc.root.child
-      assert_equal("M\xC3\xB6tley Cr\xC3\xBCe is an American heavy metal band formed in Los Angeles, California in 1981.".force_encoding(@encoding),
-                   first.content)
-      assert_equal("Iron Maiden is a British heavy metal band formed in 1975.", first.next.content)
-    else
-      assert_equal("M\303\266tley Cr\303\274e is an American heavy metal band formed in Los Angeles, California in 1981.Iron Maiden is a British heavy metal band formed in 1975.",
-                   @doc.root.content)
-
-      first = @doc.root.child
-      assert_equal("M\303\266tley Cr\303\274e is an American heavy metal band formed in Los Angeles, California in 1981.",
-                   first.content)
-      assert_equal("Iron Maiden is a British heavy metal band formed in 1975.", first.next.content)
-    end
+    node = @doc.root.last
+    assert_equal("Iron Maiden is a British heavy metal band formed in 1975.",
+                 node.content)
   end
 
   def test_base

data/test/tc_parser.rb

@@ -227,8 +227,8 @@ class TestParser < Test::Unit::TestCase
     doc = parser.parse
     node = doc.find_first('//metal')
     if defined?(Encoding)
-      assert_equal(Encoding::ISO8859_1, node.content.encoding)
-      assert_equal("m\303\266tley_cr\303\274e".force_encoding(Encoding::ISO8859_1), node.content)
+      assert_equal(Encoding::UTF_8, node.content.encoding)
+      assert_equal("m\303\266tley_cr\303\274e", node.content)
     else
       assert_equal("m\303\266tley_cr\303\274e", node.content)
     end

data/test/test_suite.rb

@@ -11,6 +11,7 @@ require './tc_attributes'
 require './tc_document'
 require './tc_document_write'
 require './tc_dtd'
+require './tc_encoding'
 require './tc_error'
 require './tc_html_parser'
 require './tc_html_parser_context'

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: libxml-ruby
 version: !ruby/object:Gem::Version 
-  hash: 15
+  hash: 7
   prerelease: 
   segments: 
   - 2
-  - 1
   - 2
-  version: 2.1.2
+  - 0
+  version: 2.2.0
 platform: x86-mingw32
 authors: 
 - Ross Bamform
@@ -20,7 +20,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-08-03 00:00:00 Z
+date: 2011-08-09 00:00:00 Z
 dependencies: []
 
 description: "    The Libxml-Ruby project provides Ruby language bindings for the GNOME\n    Libxml2 XML toolkit. It is free software, released under the MIT License.\n    Libxml-ruby's primary advantage over REXML is performance - if speed\n    is your need, these are good libraries to consider, as demonstrated\n    by the informal benchmark below.\n"
@@ -156,6 +156,7 @@ files:
 - test/model/shiporder.xsd
 - test/model/soap.xml
 - test/model/xinclude.xml
+- test/new_main.rb
 - test/tc_attr.rb
 - test/tc_attributes.rb
 - test/tc_attr_decl.rb
@@ -163,6 +164,7 @@ files:
 - test/tc_document.rb
 - test/tc_document_write.rb
 - test/tc_dtd.rb
+- test/tc_encoding.rb
 - test/tc_error.rb
 - test/tc_gc.rb
 - test/tc_html_parser.rb
@@ -241,6 +243,7 @@ test_files:
 - test/tc_document.rb
 - test/tc_document_write.rb
 - test/tc_dtd.rb
+- test/tc_encoding.rb
 - test/tc_error.rb
 - test/tc_gc.rb
 - test/tc_html_parser.rb