nokogiri 1.0.0 → 1.6.8.1

data/ext/nokogiri/html_sax_push_parser.c

@@ -0,0 +1,87 @@
+#include <html_sax_push_parser.h>
+
+/*
+ * call-seq:
+ *  native_write(chunk, last_chunk)
+ *
+ * Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
+ */
+static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
+{
+  xmlParserCtxtPtr ctx;
+  const char * chunk  = NULL;
+  int size            = 0;
+
+
+  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  if(Qnil != _chunk) {
+    chunk = StringValuePtr(_chunk);
+    size = (int)RSTRING_LEN(_chunk);
+  }
+
+  if(htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
+    if (!(ctx->options & XML_PARSE_RECOVER)) {
+      xmlErrorPtr e = xmlCtxtGetLastError(ctx);
+      Nokogiri_error_raise(NULL, e);
+    }
+  }
+
+  return self;
+}
+
+/*
+ * call-seq:
+ *  initialize_native(xml_sax, filename)
+ *
+ * Initialize the push parser with +xml_sax+ using +filename+
+ */
+static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
+			       VALUE encoding)
+{
+  htmlSAXHandlerPtr sax;
+  const char * filename = NULL;
+  htmlParserCtxtPtr ctx;
+  xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
+
+  Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
+
+  if(_filename != Qnil) filename = StringValueCStr(_filename);
+
+  if (!NIL_P(encoding)) {
+    enc = xmlParseCharEncoding(StringValueCStr(encoding));
+    if (enc == XML_CHAR_ENCODING_ERROR)
+      rb_raise(rb_eArgError, "Unsupported Encoding");
+  }
+
+  ctx = htmlCreatePushParserCtxt(
+      sax,
+      NULL,
+      NULL,
+      0,
+      filename,
+      enc
+  );
+  if(ctx == NULL)
+    rb_raise(rb_eRuntimeError, "Could not create a parser context");
+
+  ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
+
+  ctx->sax2 = 1;
+  DATA_PTR(self) = ctx;
+  return self;
+}
+
+VALUE cNokogiriHtmlSaxPushParser;
+void init_html_sax_push_parser()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE html = rb_define_module_under(nokogiri, "HTML");
+  VALUE sax = rb_define_module_under(html, "SAX");
+  VALUE klass = rb_define_class_under(sax, "PushParser", cNokogiriXmlSaxPushParser);
+
+  cNokogiriHtmlSaxPushParser = klass;
+
+  rb_define_private_method(klass, "initialize_native", initialize_native, 3);
+  rb_define_private_method(klass, "native_write", native_write, 2);
+}

data/ext/nokogiri/html_sax_push_parser.h

@@ -0,0 +1,9 @@
+#ifndef NOKOGIRI_HTML_SAX_PUSH_PARSER
+#define NOKOGIRI_HTML_SAX_PUSH_PARSER
+
+#include <nokogiri.h>
+
+void init_html_sax_push_parser();
+
+extern VALUE cNokogiriHtmlSaxPushParser ;
+#endif

data/ext/nokogiri/nokogiri.c

@@ -0,0 +1,145 @@
+#include <nokogiri.h>
+
+VALUE mNokogiri ;
+VALUE mNokogiriXml ;
+VALUE mNokogiriHtml ;
+VALUE mNokogiriXslt ;
+VALUE mNokogiriXmlSax ;
+VALUE mNokogiriHtmlSax ;
+
+#ifdef USE_INCLUDED_VASPRINTF
+/*
+ * I srsly hate windows.  it doesn't have vasprintf.
+ * Thank you Geoffroy Couprie for this implementation of vasprintf!
+ */
+int vasprintf (char **strp, const char *fmt, va_list ap)
+{
+  /* Mingw32/64 have a broken vsnprintf implementation that fails when
+   * using a zero-byte limit in order to retrieve the required size for malloc.
+   * So we use a one byte buffer instead.
+   */
+  char tmp[1];
+  int len = vsnprintf (tmp, 1, fmt, ap) + 1;
+  char *res = (char *)malloc((unsigned int)len);
+  if (res == NULL)
+      return -1;
+  *strp = res;
+  return vsnprintf(res, (unsigned int)len, fmt, ap);
+}
+#endif
+
+void vasprintf_free (void *p)
+{
+  free(p);
+}
+
+#ifdef HAVE_RUBY_UTIL_H
+#include "ruby/util.h"
+#else
+#ifndef __MACRUBY__
+#include "util.h"
+#endif
+#endif
+
+void nokogiri_root_node(xmlNodePtr node)
+{
+  xmlDocPtr doc;
+  nokogiriTuplePtr tuple;
+
+  doc = node->doc;
+  if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
+  tuple = (nokogiriTuplePtr)doc->_private;
+  st_insert(tuple->unlinkedNodes, (st_data_t)node, (st_data_t)node);
+}
+
+void nokogiri_root_nsdef(xmlNsPtr ns, xmlDocPtr doc)
+{
+  nokogiriTuplePtr tuple;
+
+  if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
+  tuple = (nokogiriTuplePtr)doc->_private;
+  st_insert(tuple->unlinkedNodes, (st_data_t)ns, (st_data_t)ns);
+}
+
+void Init_nokogiri()
+{
+#ifndef __MACRUBY__
+  xmlMemSetup(
+      (xmlFreeFunc)ruby_xfree,
+      (xmlMallocFunc)ruby_xmalloc,
+      (xmlReallocFunc)ruby_xrealloc,
+      ruby_strdup
+  );
+#endif
+
+  mNokogiri         = rb_define_module("Nokogiri");
+  mNokogiriXml      = rb_define_module_under(mNokogiri, "XML");
+  mNokogiriHtml     = rb_define_module_under(mNokogiri, "HTML");
+  mNokogiriXslt     = rb_define_module_under(mNokogiri, "XSLT");
+  mNokogiriXmlSax   = rb_define_module_under(mNokogiriXml, "SAX");
+  mNokogiriHtmlSax  = rb_define_module_under(mNokogiriHtml, "SAX");
+
+  rb_const_set( mNokogiri,
+                rb_intern("LIBXML_VERSION"),
+                NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION)
+              );
+  rb_const_set( mNokogiri,
+                rb_intern("LIBXML_PARSER_VERSION"),
+                NOKOGIRI_STR_NEW2(xmlParserVersion)
+              );
+
+#ifdef NOKOGIRI_USE_PACKAGED_LIBRARIES
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_USE_PACKAGED_LIBRARIES"), Qtrue);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATH"), NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATH));
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATH"), NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATH));
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATCHES), " "));
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATCHES), " "));
+#else
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_USE_PACKAGED_LIBRARIES"), Qfalse);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATH"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATH"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATCHES"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATCHES"), Qnil);
+#endif
+
+#ifdef LIBXML_ICONV_ENABLED
+  rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qtrue);
+#else
+  rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qfalse);
+#endif
+
+  xmlInitParser();
+
+  init_xml_document();
+  init_html_document();
+  init_xml_node();
+  init_xml_document_fragment();
+  init_xml_text();
+  init_xml_cdata();
+  init_xml_processing_instruction();
+  init_xml_attr();
+  init_xml_entity_reference();
+  init_xml_comment();
+  init_xml_node_set();
+  init_xml_xpath_context();
+  init_xml_sax_parser_context();
+  init_xml_sax_parser();
+  init_xml_sax_push_parser();
+  init_xml_reader();
+  init_xml_dtd();
+  init_xml_element_content();
+  init_xml_attribute_decl();
+  init_xml_element_decl();
+  init_xml_entity_decl();
+  init_xml_namespace();
+  init_html_sax_parser_context();
+  init_html_sax_push_parser();
+  init_xslt_stylesheet();
+  init_xml_syntax_error();
+  init_html_entity_lookup();
+  init_html_element_description();
+  init_xml_schema();
+  init_xml_relax_ng();
+  init_nokogiri_io();
+  init_xml_encoding_handler();
+}

data/ext/nokogiri/nokogiri.h

@@ -0,0 +1,131 @@
+#ifndef NOKOGIRI_NATIVE
+#define NOKOGIRI_NATIVE
+
+#include <stdlib.h>
+#include <string.h>
+#include <assert.h>
+#include <stdarg.h>
+
+#ifdef USE_INCLUDED_VASPRINTF
+int vasprintf (char **strp, const char *fmt, va_list ap);
+#else
+
+#define _GNU_SOURCE
+#  include <stdio.h>
+#undef _GNU_SOURCE
+
+#endif
+
+#include <libxml/parser.h>
+#include <libxml/entities.h>
+#include <libxml/parserInternals.h>
+#include <libxml/xpath.h>
+#include <libxml/xpathInternals.h>
+#include <libxml/xmlreader.h>
+#include <libxml/xmlsave.h>
+#include <libxml/xmlschemas.h>
+#include <libxml/HTMLparser.h>
+#include <libxml/HTMLtree.h>
+#include <libxml/relaxng.h>
+#include <libxml/xinclude.h>
+#include <libxslt/extensions.h>
+#include <libxml/c14n.h>
+#include <ruby.h>
+#include <ruby/st.h>
+#include <ruby/encoding.h>
+
+#ifndef UNUSED
+# if defined(__GNUC__)
+#  define MAYBE_UNUSED(name) name __attribute__((unused))
+#  define UNUSED(name) MAYBE_UNUSED(UNUSED_ ## name)
+# else
+#  define MAYBE_UNUSED(name) name
+#  define UNUSED(name) name
+# endif
+#endif
+
+#ifndef NORETURN
+# if defined(__GNUC__)
+#  define NORETURN(name) __attribute__((noreturn)) name
+# else
+#  define NORETURN(name) name
+# endif
+#endif
+
+#define NOKOGIRI_STR_NEW2(str) \
+  NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
+
+#define NOKOGIRI_STR_NEW(str, len) \
+  rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
+
+#define RBSTR_OR_QNIL(_str) \
+  (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
+
+#include <xml_libxml2_hacks.h>
+
+#include <xml_io.h>
+#include <xml_document.h>
+#include <html_entity_lookup.h>
+#include <html_document.h>
+#include <xml_node.h>
+#include <xml_text.h>
+#include <xml_cdata.h>
+#include <xml_attr.h>
+#include <xml_processing_instruction.h>
+#include <xml_entity_reference.h>
+#include <xml_document_fragment.h>
+#include <xml_comment.h>
+#include <xml_node_set.h>
+#include <xml_dtd.h>
+#include <xml_attribute_decl.h>
+#include <xml_element_decl.h>
+#include <xml_entity_decl.h>
+#include <xml_xpath_context.h>
+#include <xml_element_content.h>
+#include <xml_sax_parser_context.h>
+#include <xml_sax_parser.h>
+#include <xml_sax_push_parser.h>
+#include <xml_reader.h>
+#include <html_sax_parser_context.h>
+#include <html_sax_push_parser.h>
+#include <xslt_stylesheet.h>
+#include <xml_syntax_error.h>
+#include <xml_schema.h>
+#include <xml_relax_ng.h>
+#include <html_element_description.h>
+#include <xml_namespace.h>
+#include <xml_encoding_handler.h>
+
+extern VALUE mNokogiri ;
+extern VALUE mNokogiriXml ;
+extern VALUE mNokogiriXmlSax ;
+extern VALUE mNokogiriHtml ;
+extern VALUE mNokogiriHtmlSax ;
+extern VALUE mNokogiriXslt ;
+
+void nokogiri_root_node(xmlNodePtr);
+void nokogiri_root_nsdef(xmlNsPtr, xmlDocPtr);
+
+#ifdef DEBUG
+
+#define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
+#define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
+
+#else
+
+#define NOKOGIRI_DEBUG_START(p)
+#define NOKOGIRI_DEBUG_END(p)
+
+#endif
+
+#ifndef __builtin_expect
+# if defined(__GNUC__)
+#  define __builtin_expect(expr, c) __builtin_expect((long)(expr), (long)(c))
+# endif
+#endif
+
+#define XMLNS_PREFIX "xmlns"
+#define XMLNS_PREFIX_LEN 6 /* including either colon or \0 */
+#define XMLNS_BUFFER_LEN 128
+
+#endif

data/ext/nokogiri/xml_attr.c

@@ -0,0 +1,94 @@
+#include <xml_attr.h>
+
+/*
+ * call-seq:
+ *  value=(content)
+ *
+ * Set the value for this Attr to +content+
+ */
+static VALUE set_value(VALUE self, VALUE content)
+{
+  xmlAttrPtr attr;
+  Data_Get_Struct(self, xmlAttr, attr);
+
+  if(attr->children) xmlFreeNodeList(attr->children);
+
+  attr->children = attr->last = NULL;
+
+  if(content) {
+    xmlChar *buffer;
+    xmlNode *tmp;
+
+    /* Encode our content */
+    buffer = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValueCStr(content));
+
+    attr->children = xmlStringGetNodeList(attr->doc, buffer);
+    attr->last = NULL;
+    tmp = attr->children;
+
+    /* Loop through the children */
+    for(tmp = attr->children; tmp; tmp = tmp->next) {
+      tmp->parent = (xmlNode *)attr;
+      tmp->doc = attr->doc;
+      if(tmp->next == NULL) attr->last = tmp;
+    }
+
+    /* Free up memory */
+    xmlFree(buffer);
+  }
+
+  return content;
+}
+
+/*
+ * call-seq:
+ *  new(document, name)
+ *
+ * Create a new Attr element on the +document+ with +name+
+ */
+static VALUE new(int argc, VALUE *argv, VALUE klass)
+{
+  xmlDocPtr xml_doc;
+  VALUE document;
+  VALUE name;
+  VALUE rest;
+  xmlAttrPtr node;
+  VALUE rb_node;
+
+  rb_scan_args(argc, argv, "2*", &document, &name, &rest);
+
+  Data_Get_Struct(document, xmlDoc, xml_doc);
+
+  node = xmlNewDocProp(
+      xml_doc,
+      (const xmlChar *)StringValueCStr(name),
+      NULL
+  );
+
+  nokogiri_root_node((xmlNodePtr)node);
+
+  rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
+  rb_obj_call_init(rb_node, argc, argv);
+
+  if(rb_block_given_p()) rb_yield(rb_node);
+
+  return rb_node;
+}
+
+VALUE cNokogiriXmlAttr;
+void init_xml_attr()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE xml = rb_define_module_under(nokogiri, "XML");
+  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
+
+  /*
+   * Attr represents a Attr node in an xml document.
+   */
+  VALUE klass = rb_define_class_under(xml, "Attr", node);
+
+  cNokogiriXmlAttr = klass;
+
+  rb_define_singleton_method(klass, "new", new, -1);
+  rb_define_method(klass, "value=", set_value, 1);
+}

data/ext/nokogiri/xml_attr.h

@@ -0,0 +1,9 @@
+#ifndef NOKOGIRI_XML_ATTR
+#define NOKOGIRI_XML_ATTR
+
+#include <nokogiri.h>
+
+void init_xml_attr();
+
+extern VALUE cNokogiriXmlAttr;
+#endif

data/ext/nokogiri/xml_attribute_decl.c

@@ -0,0 +1,70 @@
+#include <xml_attribute_decl.h>
+
+/*
+ * call-seq:
+ *  attribute_type
+ *
+ * The attribute_type for this AttributeDecl
+ */
+static VALUE attribute_type(VALUE self)
+{
+  xmlAttributePtr node;
+  Data_Get_Struct(self, xmlAttribute, node);
+  return INT2NUM((long)node->atype);
+}
+
+/*
+ * call-seq:
+ *  default
+ *
+ * The default value
+ */
+static VALUE default_value(VALUE self)
+{
+  xmlAttributePtr node;
+  Data_Get_Struct(self, xmlAttribute, node);
+
+  if(node->defaultValue) return NOKOGIRI_STR_NEW2(node->defaultValue);
+  return Qnil;
+}
+
+/*
+ * call-seq:
+ *  enumeration
+ *
+ * An enumeration of possible values
+ */
+static VALUE enumeration(VALUE self)
+{
+  xmlAttributePtr node;
+  xmlEnumerationPtr enm;
+  VALUE list;
+
+  Data_Get_Struct(self, xmlAttribute, node);
+
+  list = rb_ary_new();
+  enm = node->tree;
+
+  while(enm) {
+    rb_ary_push(list, NOKOGIRI_STR_NEW2(enm->name));
+    enm = enm->next;
+  }
+
+  return list;
+}
+
+VALUE cNokogiriXmlAttributeDecl;
+
+void init_xml_attribute_decl()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE xml = rb_define_module_under(nokogiri, "XML");
+  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
+  VALUE klass = rb_define_class_under(xml, "AttributeDecl", node);
+
+  cNokogiriXmlAttributeDecl = klass;
+
+  rb_define_method(klass, "attribute_type", attribute_type, 0);
+  rb_define_method(klass, "default", default_value, 0);
+  rb_define_method(klass, "enumeration", enumeration, 0);
+}

data/ext/nokogiri/xml_attribute_decl.h

@@ -0,0 +1,9 @@
+#ifndef NOKOGIRI_XML_ATTRIBUTE_DECL
+#define NOKOGIRI_XML_ATTRIBUTE_DECL
+
+#include <nokogiri.h>
+
+void init_xml_attribute_decl();
+
+extern VALUE cNokogiriXmlAttributeDecl;
+#endif

data/ext/nokogiri/xml_cdata.c

@@ -1,33 +1,34 @@
 #include <xml_cdata.h>
 
-static void dealloc(xmlNodePtr node)
-{
-  if (node->doc == NULL) {
-    NOKOGIRI_DEBUG_START_NODE(node);
-    xmlFreeNode(node);
-    NOKOGIRI_DEBUG_END(node);
-  }
-}
-
 /*
  * call-seq:
  *  new(document, content)
  *
- * Create a new CData element on the +document+ with +content+
+ * Create a new CDATA element on the +document+ with +content+
  */
-static VALUE new(VALUE klass, VALUE doc, VALUE content)
+static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
+  xmlNodePtr node;
+  VALUE doc;
+  VALUE content;
+  VALUE rest;
+  VALUE rb_node;
+
+  rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
+
   Data_Get_Struct(doc, xmlDoc, xml_doc);
 
-  xmlNodePtr node = xmlNewCDataBlock(
-      xml_doc,
-      (const xmlChar *)StringValuePtr(content),
-      NUM2INT(rb_funcall(content, rb_intern("length"), 0))
+  node = xmlNewCDataBlock(
+      xml_doc->doc,
+      NIL_P(content) ? NULL : (const xmlChar *)StringValuePtr(content),
+      NIL_P(content) ? 0 : (int)RSTRING_LEN(content)
   );
 
-  VALUE rb_node = Data_Wrap_Struct(klass, NULL, dealloc, node);
-  node->_private = (void *)rb_node;
+  nokogiri_root_node(node);
+
+  rb_node = Nokogiri_wrap_xml_node(klass, node);
+  rb_obj_call_init(rb_node, argc, argv);
 
   if(rb_block_given_p()) rb_yield(rb_node);
 
@@ -39,14 +40,17 @@ void init_xml_cdata()
 {
   VALUE nokogiri = rb_define_module("Nokogiri");
   VALUE xml = rb_define_module_under(nokogiri, "XML");
+  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
+  VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
+  VALUE text = rb_define_class_under(xml, "Text", char_data);
 
   /*
    * CData represents a CData node in an xml document.
    */
-  VALUE klass = rb_define_class_under(xml, "CData", cNokogiriXmlNode);
+  VALUE klass = rb_define_class_under(xml, "CDATA", text);
 
 
   cNokogiriXmlCData = klass;
 
-  rb_define_singleton_method(klass, "new", new, 2);
+  rb_define_singleton_method(klass, "new", new, -1);
 }

data/ext/nokogiri/xml_cdata.h

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

data/ext/nokogiri/xml_comment.c

@@ -0,0 +1,69 @@
+#include <xml_comment.h>
+
+static ID document_id ;
+
+/*
+ * call-seq:
+ *  new(document_or_node, content)
+ *
+ * Create a new Comment element on the +document+ with +content+.
+ * Alternatively, if a +node+ is passed, the +node+'s document is used.
+ */
+static VALUE new(int argc, VALUE *argv, VALUE klass)
+{
+  xmlDocPtr xml_doc;
+  xmlNodePtr node;
+  VALUE document;
+  VALUE content;
+  VALUE rest;
+  VALUE rb_node;
+
+  rb_scan_args(argc, argv, "2*", &document, &content, &rest);
+
+  if (rb_obj_is_kind_of(document, cNokogiriXmlNode))
+  {
+    document = rb_funcall(document, document_id, 0);
+  }
+  else if (   !rb_obj_is_kind_of(document, cNokogiriXmlDocument)
+           && !rb_obj_is_kind_of(document, cNokogiriXmlDocumentFragment))
+  {
+    rb_raise(rb_eArgError, "first argument must be a XML::Document or XML::Node");
+  }
+
+  Data_Get_Struct(document, xmlDoc, xml_doc);
+
+  node = xmlNewDocComment(
+      xml_doc,
+      (const xmlChar *)StringValueCStr(content)
+  );
+
+  rb_node = Nokogiri_wrap_xml_node(klass, node);
+  rb_obj_call_init(rb_node, argc, argv);
+
+  nokogiri_root_node(node);
+
+  if(rb_block_given_p()) rb_yield(rb_node);
+
+  return rb_node;
+}
+
+VALUE cNokogiriXmlComment;
+void init_xml_comment()
+{
+  VALUE nokogiri = rb_define_module("Nokogiri");
+  VALUE xml = rb_define_module_under(nokogiri, "XML");
+  VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
+  VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
+
+  /*
+   * Comment represents a comment node in an xml document.
+   */
+  VALUE klass = rb_define_class_under(xml, "Comment", char_data);
+
+
+  cNokogiriXmlComment = klass;
+
+  rb_define_singleton_method(klass, "new", new, -1);
+
+  document_id = rb_intern("document");
+}