nokogiri 1.2.3-x86-mswin32-60 → 1.4.5-x86-mswin32-60

data/ext/nokogiri/html_document.c

@@ -8,15 +8,20 @@
  */
 static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
-  VALUE uri, external_id;
+  VALUE uri, external_id, rest, rb_doc;
+  htmlDocPtr doc;
 
-  rb_scan_args(argc, argv, "02", &uri, &external_id);
+  rb_scan_args(argc, argv, "0*", &rest);
+  uri         = rb_ary_entry(rest, (long)0);
+  external_id = rb_ary_entry(rest, (long)1);
 
-  htmlDocPtr doc = htmlNewDoc(
+  doc = htmlNewDoc(
       RTEST(uri) ? (const xmlChar *)StringValuePtr(uri) : NULL,
       RTEST(external_id) ? (const xmlChar *)StringValuePtr(external_id) : NULL
   );
-  return Nokogiri_wrap_xml_document(klass, doc);
+  rb_doc = Nokogiri_wrap_xml_document(klass, doc);
+  rb_obj_call_init(rb_doc, argc, argv);
+  return rb_doc ;
 }
 
 /*
@@ -32,40 +37,41 @@ static VALUE read_io( VALUE klass,
                       VALUE encoding,
                       VALUE options )
 {
-  const char * c_url    = (url == Qnil) ? NULL : StringValuePtr(url);
-  const char * c_enc    = (encoding == Qnil) ? NULL : StringValuePtr(encoding);
+  const char * c_url    = NIL_P(url)      ? NULL : StringValuePtr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
   VALUE error_list      = rb_ary_new();
+  VALUE document;
+  htmlDocPtr doc;
 
-  xmlInitParser();
   xmlResetLastError();
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 
-  htmlDocPtr doc = htmlReadIO(
+  doc = htmlReadIO(
       io_read_callback,
       io_close_callback,
       (void *)io,
       c_url,
       c_enc,
-      NUM2INT(options)
+      (int)NUM2INT(options)
   );
   xmlSetStructuredErrorFunc(NULL, NULL);
 
   if(doc == NULL) {
+    xmlErrorPtr error;
+
     xmlFreeDoc(doc);
 
-    xmlErrorPtr error = xmlGetLastError();
+    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, "Could not parse document");
 
     return Qnil;
   }
 
-  VALUE document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_funcall(document, rb_intern("errors="), 1, error_list);
+  document = Nokogiri_wrap_xml_document(klass, doc);
+  rb_iv_set(document, "@errors", error_list);
   return document;
 }
 
@@ -83,34 +89,35 @@ static VALUE read_memory( VALUE klass,
                           VALUE options )
 {
   const char * c_buffer = StringValuePtr(string);
-  const char * c_url    = (url == Qnil) ? NULL : StringValuePtr(url);
-  const char * c_enc    = (encoding == Qnil) ? NULL : StringValuePtr(encoding);
-  int len               = RSTRING_LEN(string);
+  const char * c_url    = NIL_P(url)      ? NULL : StringValuePtr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
+  int len               = (int)RSTRING_LEN(string);
   VALUE error_list      = rb_ary_new();
+  VALUE document;
+  htmlDocPtr doc;
 
-  xmlInitParser();
   xmlResetLastError();
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 
-  htmlDocPtr doc = htmlReadMemory(c_buffer, len, c_url, c_enc, NUM2INT(options));
+  doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
   xmlSetStructuredErrorFunc(NULL, NULL);
 
   if(doc == NULL) {
+    xmlErrorPtr error;
+
     xmlFreeDoc(doc);
 
-    xmlErrorPtr error = xmlGetLastError();
+    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, "Could not parse document");
 
     return Qnil;
   }
 
-  VALUE document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_funcall(document, rb_intern("errors="), 1, error_list);
+  document = Nokogiri_wrap_xml_document(klass, doc);
+  rb_iv_set(document, "@errors", error_list);
   return document;
 }
 
@@ -124,7 +131,7 @@ static VALUE type(VALUE self)
 {
   htmlDocPtr doc;
   Data_Get_Struct(self, xmlDoc, doc);
-  return INT2NUM((int)doc->type);
+  return INT2NUM((long)doc->type);
 }
 
 VALUE cNokogiriHtmlDocument ;

data/ext/nokogiri/html_document.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_HTML_DOCUMENT
 #define NOKOGIRI_HTML_DOCUMENT
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_html_document();
 

data/ext/nokogiri/html_element_description.c

@@ -0,0 +1,276 @@
+#include <html_element_description.h>
+
+/*
+ * call-seq:
+ *  required_attributes
+ *
+ * A list of required attributes for this element
+ */
+static VALUE required_attributes(VALUE self)
+{
+  htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->attrs_req) return list;
+
+  for(i = 0; description->attrs_depr[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  deprecated_attributes
+ *
+ * A list of deprecated attributes for this element
+ */
+static VALUE deprecated_attributes(VALUE self)
+{
+  htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->attrs_depr) return list;
+
+  for(i = 0; description->attrs_depr[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  optional_attributes
+ *
+ * A list of optional attributes for this element
+ */
+static VALUE optional_attributes(VALUE self)
+{
+  htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->attrs_opt) return list;
+
+  for(i = 0; description->attrs_opt[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  default_sub_element
+ *
+ * The default sub element for this element
+ */
+static VALUE default_sub_element(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  return NOKOGIRI_STR_NEW2(description->defaultsubelt);
+}
+
+/*
+ * call-seq:
+ *  sub_elements
+ *
+ * A list of allowed sub elements for this element.
+ */
+static VALUE sub_elements(VALUE self)
+{
+  htmlElemDesc * description;
+  VALUE list;
+  int i;
+
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  list = rb_ary_new();
+
+  if(NULL == description->subelts) return list;
+
+  for(i = 0; description->subelts[i]; i++) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
+  }
+
+  return list;
+}
+
+/*
+ * call-seq:
+ *  description
+ *
+ * The description for this element
+ */
+static VALUE description(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  return NOKOGIRI_STR_NEW2(description->desc);
+}
+
+/*
+ * call-seq:
+ *  inline?
+ *
+ * Is this element an inline element?
+ */
+static VALUE inline_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->isinline) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  deprecated?
+ *
+ * Is this element deprecated?
+ */
+static VALUE deprecated_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->depr) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  empty?
+ *
+ * Is this an empty element?
+ */
+static VALUE empty_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->empty) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  save_end_tag?
+ *
+ * Should the end tag be saved?
+ */
+static VALUE save_end_tag_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->saveEndTag) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  implied_end_tag?
+ *
+ * Can the end tag be implied for this tag?
+ */
+static VALUE implied_end_tag_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->endTag) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  implied_start_tag?
+ *
+ * Can the start tag be implied for this tag?
+ */
+static VALUE implied_start_tag_eh(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(description->startTag) return Qtrue;
+  return Qfalse;
+}
+
+/*
+ * call-seq:
+ *  name
+ *
+ * Get the tag name for this ElemementDescription
+ */
+static VALUE name(VALUE self)
+{
+  htmlElemDesc * description;
+  Data_Get_Struct(self, htmlElemDesc, description);
+
+  if(NULL == description->name) return Qnil;
+  return NOKOGIRI_STR_NEW2(description->name);
+}
+
+/*
+ * call-seq:
+ *  [](tag_name)
+ *
+ * Get ElemementDescription for +tag_name+
+ */
+static VALUE get_description(VALUE klass, VALUE tag_name)
+{
+  const htmlElemDesc * description = htmlTagLookup(
+      (const xmlChar *)StringValuePtr(tag_name)
+  );
+
+  if(NULL == description) return Qnil;
+  return Data_Wrap_Struct(klass, 0, 0, (void *)description);
+}
+
+VALUE cNokogiriHtmlElementDescription ;
+void init_html_element_description()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE html     = rb_define_module_under(nokogiri, "HTML");
+  VALUE klass    = rb_define_class_under(html, "ElementDescription",rb_cObject);
+
+  cNokogiriHtmlElementDescription = klass;
+
+  rb_define_singleton_method(klass, "[]", get_description, 1);
+
+  rb_define_method(klass, "name", name, 0);
+  rb_define_method(klass, "implied_start_tag?", implied_start_tag_eh, 0);
+  rb_define_method(klass, "implied_end_tag?", implied_end_tag_eh, 0);
+  rb_define_method(klass, "save_end_tag?", save_end_tag_eh, 0);
+  rb_define_method(klass, "empty?", empty_eh, 0);
+  rb_define_method(klass, "deprecated?", deprecated_eh, 0);
+  rb_define_method(klass, "inline?", inline_eh, 0);
+  rb_define_method(klass, "description", description, 0);
+  rb_define_method(klass, "sub_elements", sub_elements, 0);
+  rb_define_method(klass, "default_sub_element", default_sub_element, 0);
+  rb_define_method(klass, "optional_attributes", optional_attributes, 0);
+  rb_define_method(klass, "deprecated_attributes", deprecated_attributes, 0);
+  rb_define_method(klass, "required_attributes", required_attributes, 0);
+}

data/ext/nokogiri/html_element_description.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_HTML_ELEMENT_DESCRIPTION
+#define NOKOGIRI_HTML_ELEMENT_DESCRIPTION
+
+#include <nokogiri.h>
+
+void init_html_element_description();
+
+extern VALUE cNokogiriHtmlElementDescription ;
+
+#endif

data/ext/nokogiri/html_entity_lookup.c

@@ -10,14 +10,16 @@ static VALUE get(VALUE self, VALUE key)
 {
   const htmlEntityDesc * desc =
     htmlEntityLookup((const xmlChar *)StringValuePtr(key));
+  VALUE klass, args[3];
 
   if(NULL == desc) return Qnil;
-  VALUE klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
+  klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
 
-  return rb_funcall(klass, rb_intern("new"), 3,
-      INT2NUM((int)desc->value),
-      NOKOGIRI_STR_NEW2(desc->name, "UTF-8"),
-      NOKOGIRI_STR_NEW2(desc->desc, "UTF-8"));
+  args[0] = INT2NUM((long)desc->value);
+  args[1] = NOKOGIRI_STR_NEW2(desc->name);
+  args[2] = NOKOGIRI_STR_NEW2(desc->desc);
+
+  return rb_class_new_instance(3, args, klass);
 }
 
 void init_html_entity_lookup()