libxml-ruby 1.1.4 → 2.0.0

data/ext/libxml/ruby_xml_encoding.h

@@ -9,4 +9,8 @@ extern VALUE mXMLEncoding;
 
 void rxml_init_encoding();
 
+#ifdef HAVE_RUBY_ENCODING_H
+VALUE rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding);
+#endif
+
 #endif

data/ext/libxml/ruby_xml_namespace.c

@@ -98,7 +98,7 @@ static VALUE rxml_namespace_href_get(VALUE self)
   if (xns->href == NULL)
     return Qnil;
   else
-    return rb_str_new2((const char*) xns->href);
+    return rxml_str_new2((const char*) xns->href, xns->context ? xns->context->encoding : NULL);
 }
 
 /*
@@ -133,7 +133,7 @@ static VALUE rxml_namespace_prefix_get(VALUE self)
   if (xns->prefix == NULL)
     return Qnil;
   else
-    return rb_str_new2((const char*) xns->prefix);
+    return rxml_str_new2((const char*) xns->prefix, xns->context ? xns->context->encoding : NULL);
 }
 
 /*

data/ext/libxml/ruby_xml_node.c

@@ -279,7 +279,7 @@ static VALUE rxml_node_base_uri_get(VALUE self)
   base_uri = xmlNodeGetBase(xnode->doc, xnode);
   if (base_uri)
   {
-    result = rb_str_new2((const char*) base_uri);
+    result = rxml_str_new2((const char*) base_uri, xnode->doc ? xnode->doc->encoding : NULL);
     xmlFree(base_uri);
   }
 
@@ -323,7 +323,7 @@ static VALUE rxml_node_content_get(VALUE self)
   content = xmlNodeGetContent(xnode);
   if (content)
   {
-    result = rb_str_new2((const char *) content);
+    result = rxml_str_new2((const char *) content, xnode->doc ? xnode->doc->encoding : NULL);
     xmlFree(content);
   }
 
@@ -370,7 +370,7 @@ static VALUE rxml_node_content_stripped_get(VALUE self)
   content = xmlNodeGetContent(xnode);
   if (content)
   {
-    result = rb_str_new2((const char*) content);
+    result = rxml_str_new2((const char*) content, xnode->doc ? xnode->doc->encoding : NULL);
     xmlFree(content);
   }
   return (result);
@@ -538,7 +538,7 @@ static VALUE rxml_node_to_s(int argc, VALUE *argv, VALUE self)
 
   int level = 0;
   int indent = 1;
-  const char *xencoding = NULL;
+  const char *xencoding = "UTF-8";
 
   rb_scan_args(argc, argv, "01", &options);
 
@@ -572,9 +572,9 @@ static VALUE rxml_node_to_s(int argc, VALUE *argv, VALUE self)
   xmlOutputBufferFlush(output);
 
   if (output->conv)
-    result = rb_str_new2((const char*) output->conv->content);
+    result = rxml_str_new2((const char*) output->conv->content, xencoding);
   else
-    result = rb_str_new2((const char*) output->buffer->content);
+    result = rxml_str_new2((const char*) output->buffer->content, xencoding);
 
   xmlOutputBufferClose(output);
   
@@ -678,7 +678,7 @@ static VALUE rxml_node_lang_get(VALUE self)
 
   if (lang)
   {
-    result = rb_str_new2((const char*) lang);
+    result = rxml_str_new2((const char*) lang, xnode->doc ? xnode->doc->encoding : NULL);
     xmlFree(lang);
   }
 
@@ -811,11 +811,11 @@ static VALUE rxml_node_xlink_type_name(VALUE self)
   case XLINK_TYPE_NONE:
     return (Qnil);
   case XLINK_TYPE_SIMPLE:
-    return (rb_str_new2("simple"));
+    return (rxml_str_new2("simple", xnode->doc ? xnode->doc->encoding : NULL));
   case XLINK_TYPE_EXTENDED:
-    return (rb_str_new2("extended"));
+    return (rxml_str_new2("extended", xnode->doc ? xnode->doc->encoding : NULL));
   case XLINK_TYPE_EXTENDED_SET:
-    return (rb_str_new2("extended_set"));
+    return (rxml_str_new2("extended_set", xnode->doc ? xnode->doc->encoding : NULL));
   default:
     rb_fatal("Unknowng xlink type, %d", xlt);
   }
@@ -866,7 +866,7 @@ static VALUE rxml_node_name_get(VALUE self)
   if (xnode->name == NULL)
     return (Qnil);
   else
-    return (rb_str_new2((const char*) name));
+    return (rxml_str_new2((const char*) name, xnode->doc ? xnode->doc->encoding : NULL));
 }
 
 /*
@@ -958,7 +958,7 @@ static VALUE rxml_node_path(VALUE self)
   if (path == NULL)
     return (Qnil);
   else
-    return (rb_str_new2((const char*) path));
+    return (rxml_str_new2((const char*) path, xnode->doc ? xnode->doc->encoding : NULL));
 }
 
 /*
@@ -1233,9 +1233,9 @@ static VALUE rxml_node_space_preserve_set(VALUE self, VALUE bool)
   Data_Get_Struct(self, xmlNode, xnode);
 
   if (TYPE(bool) == T_FALSE)
-    xmlNodeSetSpacePreserve(xnode, 1);
-  else
     xmlNodeSetSpacePreserve(xnode, 0);
+  else
+    xmlNodeSetSpacePreserve(xnode, 1);
 
   return (Qnil);
 }

data/ext/libxml/ruby_xml_parser_context.c

@@ -190,7 +190,7 @@ static VALUE rxml_parser_context_base_uri_get(VALUE self)
   Data_Get_Struct(self, xmlParserCtxt, ctxt);
 
   if (ctxt->input && ctxt->input->filename)
-    return rb_str_new2(ctxt->input->filename);
+    return rxml_str_new2(ctxt->input->filename, ctxt->encoding);
   else
     return Qnil;
 }
@@ -231,7 +231,7 @@ static VALUE rxml_parser_context_data_directory_get(VALUE self)
   if (ctxt->directory == NULL)
     return (Qnil);
   else
-    return (rb_str_new2(ctxt->directory));
+    return (rxml_str_new2(ctxt->directory, ctxt->encoding));
 }
 
 /*
@@ -489,7 +489,7 @@ static VALUE rxml_parser_context_name_node_get(VALUE self)
   if (ctxt->name == NULL)
     return (Qnil);
   else
-    return (rb_str_new2((const char*) ctxt->name));
+    return (rxml_str_new2((const char*) ctxt->name, ctxt->encoding));
 }
 
 /*
@@ -516,7 +516,7 @@ static VALUE rxml_parser_context_name_tab_get(VALUE self)
     if (ctxt->nameTab[i] == NULL)
       continue;
     else
-      rb_ary_push(tab_ary, rb_str_new2((const char*) ctxt->nameTab[i]));
+      rb_ary_push(tab_ary, rxml_str_new2((const char*) ctxt->nameTab[i], ctxt->encoding));
   }
 
   return (tab_ary);
@@ -782,7 +782,7 @@ static VALUE rxml_parser_context_subset_name_get(VALUE self)
   if (ctxt->intSubName == NULL)
     return (Qnil);
   else
-    return (rb_str_new2((const char*) ctxt->intSubName));
+    return (rxml_str_new2((const char*) ctxt->intSubName, ctxt->encoding));
 }
 
 /*
@@ -801,7 +801,7 @@ static VALUE rxml_parser_context_subset_external_uri_get(VALUE self)
   if (ctxt->extSubURI == NULL)
     return (Qnil);
   else
-    return (rb_str_new2((const char*) ctxt->extSubURI));
+    return (rxml_str_new2((const char*) ctxt->extSubURI, ctxt->encoding));
 }
 
 /*
@@ -820,7 +820,7 @@ static VALUE rxml_parser_context_subset_external_system_id_get(VALUE self)
   if (ctxt->extSubSystem == NULL)
     return (Qnil);
   else
-    return (rb_str_new2((const char*) ctxt->extSubSystem));
+    return (rxml_str_new2((const char*) ctxt->extSubSystem, ctxt->encoding));
 }
 
 /*
@@ -905,7 +905,7 @@ static VALUE rxml_parser_context_version_get(VALUE self)
   if (ctxt->version == NULL)
     return (Qnil);
   else
-    return (rb_str_new2((const char*) ctxt->version));
+    return (rxml_str_new2((const char*) ctxt->version, ctxt->encoding));
 }
 
 /*

data/ext/libxml/ruby_xml_reader.c

@@ -238,6 +238,7 @@ static VALUE rxml_reader_string(int argc, VALUE *argv, VALUE klass)
     VALUE encoding = Qnil;
     VALUE parserOptions = Qnil;
 
+    int foo;
     Check_Type(options, T_HASH);
 
     baseurl = rb_hash_aref(options, BASE_URI_SYMBOL);
@@ -458,8 +459,19 @@ static VALUE rxml_reader_read_attr_value(VALUE self)
  */
 static VALUE rxml_reader_read_inner_xml(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderReadInnerXml(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  VALUE result = Qnil;
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+
+  xmlChar *xml = xmlTextReaderReadInnerXml(xReader);
+
+  if (xml != NULL)
+  {
+    const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+    result = rxml_str_new2((const char*) xml, xencoding);
+    xmlFree(xml);
+  }
+
+  return result;
 }
 
 /*
@@ -473,8 +485,19 @@ static VALUE rxml_reader_read_inner_xml(VALUE self)
  */
 static VALUE rxml_reader_read_outer_xml(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderReadOuterXml(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  VALUE result = Qnil;
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+
+  xmlChar *xml = xmlTextReaderReadOuterXml(xReader);
+
+  if (xml != NULL)
+  {
+    const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+    result = rxml_str_new2((const char*) xml, xencoding);
+    xmlFree(xml);
+  }
+
+  return result;
 }
 
 /*
@@ -499,8 +522,19 @@ static VALUE rxml_reader_read_state(VALUE self)
  */
 static VALUE rxml_reader_read_string(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderReadString(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  VALUE result = Qnil;
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+
+  xmlChar *xml = xmlTextReaderReadString(xReader);
+
+  if (xml != NULL)
+  {
+    const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+    result = rxml_str_new2((const char*) xml, xencoding);
+    xmlFree(xml);
+  }
+
+  return result;
 }
 
 /*
@@ -549,8 +583,11 @@ rxml_reader_schema_validate(VALUE self, VALUE xsd)
  */
 static VALUE rxml_reader_name(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderConstName(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *result = xmlTextReaderConstName(xReader);
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+
+  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
 }
 
 /*
@@ -561,8 +598,11 @@ static VALUE rxml_reader_name(VALUE self)
  */
 static VALUE rxml_reader_local_name(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderConstLocalName(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *result = xmlTextReaderConstLocalName(xReader);
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+
+  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
 }
 
 /*
@@ -610,8 +650,11 @@ static VALUE rxml_reader_encoding(VALUE self)
  */
 static VALUE rxml_reader_base_uri(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderConstBaseUri(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *result = xmlTextReaderConstBaseUri(xReader);
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+
+  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
 }
 
 /*
@@ -622,8 +665,11 @@ static VALUE rxml_reader_base_uri(VALUE self)
  */
 static VALUE rxml_reader_namespace_uri(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderConstNamespaceUri(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *result = xmlTextReaderConstNamespaceUri(xReader);
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+
+  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
 }
 
 /*
@@ -634,8 +680,11 @@ static VALUE rxml_reader_namespace_uri(VALUE self)
  */
 static VALUE rxml_reader_value(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderConstValue(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *result = xmlTextReaderConstValue(xReader);
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+
+  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
 }
 
 /*
@@ -646,8 +695,11 @@ static VALUE rxml_reader_value(VALUE self)
  */
 static VALUE rxml_reader_prefix(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderConstPrefix(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *result = xmlTextReaderConstPrefix(xReader);
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+
+  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
 }
 
 /*
@@ -696,8 +748,11 @@ static VALUE rxml_reader_standalone(VALUE self)
  */
 static VALUE rxml_reader_xml_lang(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderConstXmlLang(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *result = xmlTextReaderConstXmlLang(xReader);
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+
+  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
 }
 
 /*
@@ -708,8 +763,11 @@ static VALUE rxml_reader_xml_lang(VALUE self)
  */
 static VALUE rxml_reader_xml_version(VALUE self)
 {
-  const xmlChar *result = xmlTextReaderConstXmlVersion(rxml_text_reader_get(self));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *result = xmlTextReaderConstXmlVersion(xReader);
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+
+  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
 }
 
 /*
@@ -747,18 +805,19 @@ static VALUE rxml_reader_attribute(VALUE self, VALUE key)
 {
   xmlTextReaderPtr reader;
   xmlChar *attr;
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
 
-  reader = rxml_text_reader_get(self);
 
   if (TYPE(key) == T_FIXNUM)
   {
-    attr = xmlTextReaderGetAttributeNo(reader, FIX2INT(key));
+    attr = xmlTextReaderGetAttributeNo(xReader, FIX2INT(key));
   }
   else
   {
-    attr = xmlTextReaderGetAttribute(reader, (const xmlChar *) StringValueCStr(key));
+    attr = xmlTextReaderGetAttribute(xReader, (const xmlChar *) StringValueCStr(key));
   }
-  return (attr == NULL ? Qnil : rb_str_new2((const char*)attr));
+  return (attr == NULL ? Qnil : rxml_str_new2(attr, xencoding));
 }
 
 /*
@@ -770,9 +829,11 @@ static VALUE rxml_reader_attribute(VALUE self, VALUE key)
  */
 static VALUE rxml_reader_lookup_namespace(VALUE self, VALUE prefix)
 {
-  const xmlChar *result = xmlTextReaderLookupNamespace(rxml_text_reader_get(
-      self), (const xmlChar *) StringValueCStr(prefix));
-  return (result == NULL ? Qnil : rb_str_new2((const char*)result));
+  xmlTextReaderPtr xReader = rxml_text_reader_get(self);
+  const xmlChar *result = xmlTextReaderLookupNamespace(xReader, (const xmlChar *) StringValueCStr(prefix));
+  const xmlChar *xencoding = xmlTextReaderConstEncoding(xReader);
+
+  return (result == NULL ? Qnil : rxml_str_new2(result, xencoding));
 }
 
 /*
@@ -786,25 +847,19 @@ static VALUE rxml_reader_lookup_namespace(VALUE self, VALUE prefix)
  */
 static VALUE rxml_reader_expand(VALUE self)
 {
-  xmlNodePtr node;
-  xmlDocPtr doc;
+  xmlNodePtr node, nodec;
   xmlTextReaderPtr reader = rxml_text_reader_get(self);
   node = xmlTextReaderExpand(reader);
 
   if (!node)
     return Qnil;
-
-  /* Okay this is tricky.  By accessing the returned node, we
-   take ownership of the reader's document.  Thus we need to
-   tell the reader to not free it.  Otherwise it will be
-   freed twice - once when the Ruby document wrapper goes
-   out of scope and once when the reader goes out of scope. */
-
-  xmlTextReaderPreserve(reader);
-  doc = xmlTextReaderCurrentDoc(reader);
-  rxml_document_wrap(doc);
-
-  return rxml_node_wrap(node);
+  else {
+    nodec = xmlCopyNode(node, 1);
+    if (!nodec)
+      return Qnil;
+    else
+      return rxml_node_wrap(nodec);
+  }
 }
 
 #if LIBXML_VERSION >= 20618

data/ext/libxml/ruby_xml_sax2_handler.c

@@ -253,7 +253,19 @@ static void start_element_ns_callback(void *ctx,
 
 static void structured_error_callback(void *ctx, xmlErrorPtr xerror)
 {
-  VALUE handler = (VALUE) ctx;
+  /* Older versions of Libxml will pass a NULL context from the sax parser.  Fixed on
+     Feb 23, 2011.  See:
+
+     http://git.gnome.org/browse/libxml2/commit/?id=241d4a1069e6bedd0ee2295d7b43858109c1c6d1 */
+
+  VALUE handler;
+
+  #if LIBXML_VERSION <= 20708
+    xmlParserCtxtPtr ctxt = (xmlParserCtxt*)(xerror->ctxt);
+    ctx = ctxt->userData;
+  #endif
+
+  handler = (VALUE) ctx;
 
   if (handler != Qnil)
   {