nokogiri 1.11.0.rc2 → 1.11.2

data/ext/nokogiri/test_global_handlers.c

@@ -0,0 +1,40 @@
+#include <nokogiri.h>
+
+static VALUE foreign_error_handler_block = Qnil;
+
+static void
+foreign_error_handler(void *user_data, xmlErrorPtr c_error)
+{
+  rb_funcall(foreign_error_handler_block, rb_intern("call"), 0);
+}
+
+/*
+ * call-seq:
+ *   __foreign_error_handler { ... } -> nil
+ *
+ * Override libxml2's global error handlers to call the block. This method thus has very little
+ * value except to test that Nokogiri is properly setting error handlers elsewhere in the code. See
+ * test/helper.rb for how this is being used.
+ */
+static VALUE
+rb_foreign_error_handler(VALUE klass)
+{
+  rb_need_block();
+  foreign_error_handler_block = rb_block_proc();
+  xmlSetStructuredErrorFunc(NULL, foreign_error_handler);
+  return Qnil;
+}
+
+/*
+ *  Document-module: Nokogiri::Test
+ *
+ *  The Nokogiri::Test module should only be used for testing Nokogiri.
+ *  Do NOT use this outside of the Nokogiri test suite.
+ */
+void
+noko_init_test_global_handlers()
+{
+  VALUE mNokogiriTest = rb_define_module_under(mNokogiri, "Test");
+
+  rb_define_singleton_method(mNokogiriTest, "__foreign_error_handler", rb_foreign_error_handler, 0);
+}

data/ext/nokogiri/xml_attr.c

@@ -1,4 +1,6 @@
-#include <xml_attr.h>
+#include <nokogiri.h>
+
+VALUE cNokogiriXmlAttr;
 
 /*
  * call-seq:
@@ -7,7 +9,8 @@
  * Set the value for this Attr to +content+. Use `nil` to remove the value
  * (e.g., a HTML boolean attribute).
  */
-static VALUE set_value(VALUE self, VALUE content)
+static VALUE
+set_value(VALUE self, VALUE content)
 {
   xmlAttrPtr attr;
   xmlChar *value;
@@ -49,7 +52,8 @@ static VALUE set_value(VALUE self, VALUE content)
  *
  * Create a new Attr element on the +document+ with +name+
  */
-static VALUE new(int argc, VALUE *argv, VALUE klass)
+static VALUE
+new (int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
   VALUE document;
@@ -72,9 +76,9 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
            NULL
          );
 
-  nokogiri_root_node((xmlNodePtr)node);
+  noko_xml_document_pin_node((xmlNodePtr)node);
 
-  rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
+  rb_node = noko_xml_node_wrap(klass, (xmlNodePtr)node);
   rb_obj_call_init(rb_node, argc, argv);
 
   if (rb_block_given_p()) {
@@ -84,20 +88,16 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   return rb_node;
 }
 
-VALUE cNokogiriXmlAttr;
-void init_xml_attr()
+void
+noko_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);
-
+  assert(cNokogiriXmlNode);
   /*
    * Attr represents a Attr node in an xml document.
    */
-  VALUE klass = rb_define_class_under(xml, "Attr", node);
+  cNokogiriXmlAttr = rb_define_class_under(mNokogiriXml, "Attr", cNokogiriXmlNode);
 
-  cNokogiriXmlAttr = klass;
+  rb_define_singleton_method(cNokogiriXmlAttr, "new", new, -1);
 
-  rb_define_singleton_method(klass, "new", new, -1);
-  rb_define_method(klass, "value=", set_value, 1);
+  rb_define_method(cNokogiriXmlAttr, "value=", set_value, 1);
 }

data/ext/nokogiri/xml_attribute_decl.c

@@ -1,4 +1,6 @@
-#include <xml_attribute_decl.h>
+#include <nokogiri.h>
+
+VALUE cNokogiriXmlAttributeDecl;
 
 /*
  * call-seq:
@@ -6,7 +8,8 @@
  *
  * The attribute_type for this AttributeDecl
  */
-static VALUE attribute_type(VALUE self)
+static VALUE
+attribute_type(VALUE self)
 {
   xmlAttributePtr node;
   Data_Get_Struct(self, xmlAttribute, node);
@@ -19,12 +22,13 @@ static VALUE attribute_type(VALUE self)
  *
  * The default value
  */
-static VALUE default_value(VALUE self)
+static VALUE
+default_value(VALUE self)
 {
   xmlAttributePtr node;
   Data_Get_Struct(self, xmlAttribute, node);
 
-  if(node->defaultValue) return NOKOGIRI_STR_NEW2(node->defaultValue);
+  if (node->defaultValue) { return NOKOGIRI_STR_NEW2(node->defaultValue); }
   return Qnil;
 }
 
@@ -34,7 +38,8 @@ static VALUE default_value(VALUE self)
  *
  * An enumeration of possible values
  */
-static VALUE enumeration(VALUE self)
+static VALUE
+enumeration(VALUE self)
 {
   xmlAttributePtr node;
   xmlEnumerationPtr enm;
@@ -45,7 +50,7 @@ static VALUE enumeration(VALUE self)
   list = rb_ary_new();
   enm = node->tree;
 
-  while(enm) {
+  while (enm) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(enm->name));
     enm = enm->next;
   }
@@ -53,18 +58,13 @@ static VALUE enumeration(VALUE self)
   return list;
 }
 
-VALUE cNokogiriXmlAttributeDecl;
-
-void init_xml_attribute_decl()
+void
+noko_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;
+  assert(cNokogiriXmlNode);
+  cNokogiriXmlAttributeDecl = rb_define_class_under(mNokogiriXml, "AttributeDecl", cNokogiriXmlNode);
 
-  rb_define_method(klass, "attribute_type", attribute_type, 0);
-  rb_define_method(klass, "default", default_value, 0);
-  rb_define_method(klass, "enumeration", enumeration, 0);
+  rb_define_method(cNokogiriXmlAttributeDecl, "attribute_type", attribute_type, 0);
+  rb_define_method(cNokogiriXmlAttributeDecl, "default", default_value, 0);
+  rb_define_method(cNokogiriXmlAttributeDecl, "enumeration", enumeration, 0);
 }

data/ext/nokogiri/xml_cdata.c

@@ -1,4 +1,6 @@
-#include <xml_cdata.h>
+#include <nokogiri.h>
+
+VALUE cNokogiriXmlCData;
 
 /*
  * call-seq:
@@ -9,7 +11,8 @@
  * If +content+ cannot be implicitly converted to a string, this method will
  * raise a TypeError exception.
  */
-static VALUE new(int argc, VALUE *argv, VALUE klass)
+static VALUE
+new (int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
   xmlNodePtr node;
@@ -31,32 +34,24 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   node = xmlNewCDataBlock(xml_doc->doc, content_str, content_str_len);
 
-  nokogiri_root_node(node);
+  noko_xml_document_pin_node(node);
 
-  rb_node = Nokogiri_wrap_xml_node(klass, node);
+  rb_node = noko_xml_node_wrap(klass, node);
   rb_obj_call_init(rb_node, argc, argv);
 
-  if(rb_block_given_p()) { rb_yield(rb_node); }
+  if (rb_block_given_p()) { rb_yield(rb_node); }
 
   return rb_node;
 }
 
-VALUE cNokogiriXmlCData;
-void init_xml_cdata()
+void
+noko_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);
-
+  assert(cNokogiriXmlText);
   /*
    * CData represents a CData node in an xml document.
    */
-  VALUE klass = rb_define_class_under(xml, "CDATA", text);
-
-
-  cNokogiriXmlCData = klass;
+  cNokogiriXmlCData = rb_define_class_under(mNokogiriXml, "CDATA", cNokogiriXmlText);
 
-  rb_define_singleton_method(klass, "new", new, -1);
+  rb_define_singleton_method(cNokogiriXmlCData, "new", new, -1);
 }

data/ext/nokogiri/xml_comment.c

@@ -1,4 +1,6 @@
-#include <xml_comment.h>
+#include <nokogiri.h>
+
+VALUE cNokogiriXmlComment;
 
 static ID document_id ;
 
@@ -9,7 +11,8 @@ static ID document_id ;
  * 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)
+static VALUE
+new (int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
   xmlNodePtr node;
@@ -20,50 +23,40 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   rb_scan_args(argc, argv, "2*", &document, &content, &rest);
 
-  if (rb_obj_is_kind_of(document, cNokogiriXmlNode))
-  {
+  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))
-  {
+  } 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)
-  );
+           xml_doc,
+           (const xmlChar *)StringValueCStr(content)
+         );
 
-  rb_node = Nokogiri_wrap_xml_node(klass, node);
+  rb_node = noko_xml_node_wrap(klass, node);
   rb_obj_call_init(rb_node, argc, argv);
 
-  nokogiri_root_node(node);
+  noko_xml_document_pin_node(node);
 
-  if(rb_block_given_p()) rb_yield(rb_node);
+  if (rb_block_given_p()) { rb_yield(rb_node); }
 
   return rb_node;
 }
 
-VALUE cNokogiriXmlComment;
-void init_xml_comment()
+void
+noko_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);
-
+  assert(cNokogiriXmlCharacterData);
   /*
    * Comment represents a comment node in an xml document.
    */
-  VALUE klass = rb_define_class_under(xml, "Comment", char_data);
-
-
-  cNokogiriXmlComment = klass;
+  cNokogiriXmlComment = rb_define_class_under(mNokogiriXml, "Comment", cNokogiriXmlCharacterData);
 
-  rb_define_singleton_method(klass, "new", new, -1);
+  rb_define_singleton_method(cNokogiriXmlComment, "new", new, -1);
 
   document_id = rb_intern("document");
 }

data/ext/nokogiri/xml_document.c

@@ -1,8 +1,11 @@
-#include <xml_document.h>
+#include <nokogiri.h>
 
-static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
+VALUE cNokogiriXmlDocument ;
+
+static int
+dealloc_node_i2(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
 {
-  switch(node->type) {
+  switch (node->type) {
   case XML_ATTRIBUTE_NODE:
     xmlFreePropList((xmlAttrPtr)node);
     break;
@@ -13,41 +16,52 @@ static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
     xmlFreeDtd((xmlDtdPtr)node);
     break;
   default:
-    if(node->parent == NULL) {
+    if (node->parent == NULL) {
       xmlAddChild((xmlNodePtr)doc, node);
     }
   }
   return ST_CONTINUE;
 }
 
-static void remove_private(xmlNodePtr node)
+static int
+dealloc_node_i(st_data_t key, st_data_t node, st_data_t doc)
+{
+  return dealloc_node_i2((xmlNodePtr)key, (xmlNodePtr)node, (xmlDocPtr)doc);
+}
+
+static void
+remove_private(xmlNodePtr node)
 {
   xmlNodePtr child;
 
-  for (child = node->children; child; child = child->next)
+  for (child = node->children; child; child = child->next) {
     remove_private(child);
+  }
 
   if ((node->type == XML_ELEMENT_NODE ||
        node->type == XML_XINCLUDE_START ||
        node->type == XML_XINCLUDE_END) &&
       node->properties) {
-    for (child = (xmlNodePtr)node->properties; child; child = child->next)
+    for (child = (xmlNodePtr)node->properties; child; child = child->next) {
       remove_private(child);
+    }
   }
 
   node->_private = NULL;
 }
 
-static void mark(xmlDocPtr doc)
+static void
+mark(xmlDocPtr doc)
 {
   nokogiriTuplePtr tuple = (nokogiriTuplePtr)doc->_private;
-  if(tuple) {
-      rb_gc_mark(tuple->doc);
-      rb_gc_mark(tuple->node_cache);
+  if (tuple) {
+    rb_gc_mark(tuple->doc);
+    rb_gc_mark(tuple->node_cache);
   }
 }
 
-static void dealloc(xmlDocPtr doc)
+static void
+dealloc(xmlDocPtr doc)
 {
   st_table *node_hash;
 
@@ -65,23 +79,26 @@ static void dealloc(xmlDocPtr doc)
    * xmlDeregisterNode callback from accessing VALUE pointers from ruby's GC
    * free context, which can result in segfaults.
    */
-  if (xmlDeregisterNodeDefaultValue)
+  if (xmlDeregisterNodeDefaultValue) {
     remove_private((xmlNodePtr)doc);
+  }
 
   xmlFreeDoc(doc);
 
   NOKOGIRI_DEBUG_END(doc);
 }
 
-static void recursively_remove_namespaces_from_node(xmlNodePtr node)
+static void
+recursively_remove_namespaces_from_node(xmlNodePtr node)
 {
   xmlNodePtr child ;
   xmlAttrPtr property ;
 
   xmlSetNs(node, NULL);
 
-  for (child = node->children ; child ; child = child->next)
+  for (child = node->children ; child ; child = child->next) {
     recursively_remove_namespaces_from_node(child);
+  }
 
   if (((node->type == XML_ELEMENT_NODE) ||
        (node->type == XML_XINCLUDE_START) ||
@@ -94,7 +111,7 @@ static void recursively_remove_namespaces_from_node(xmlNodePtr node)
   if (node->type == XML_ELEMENT_NODE && node->properties != NULL) {
     property = node->properties ;
     while (property != NULL) {
-      if (property->ns) property->ns = NULL ;
+      if (property->ns) { property->ns = NULL ; }
       property = property->next ;
     }
   }
@@ -106,12 +123,13 @@ static void recursively_remove_namespaces_from_node(xmlNodePtr node)
  *
  * Get the url name for this document.
  */
-static VALUE url(VALUE self)
+static VALUE
+url(VALUE self)
 {
   xmlDocPtr doc;
   Data_Get_Struct(self, xmlDoc, doc);
 
-  if(doc->URL) return NOKOGIRI_STR_NEW2(doc->URL);
+  if (doc->URL) { return NOKOGIRI_STR_NEW2(doc->URL); }
 
   return Qnil;
 }
@@ -122,7 +140,8 @@ static VALUE url(VALUE self)
  *
  * Set the root element on this document
  */
-static VALUE set_root(VALUE self, VALUE root)
+static VALUE
+set_root(VALUE self, VALUE root)
 {
   xmlDocPtr doc;
   xmlNodePtr new_root;
@@ -132,12 +151,12 @@ static VALUE set_root(VALUE self, VALUE root)
 
   old_root = NULL;
 
-  if(NIL_P(root)) {
+  if (NIL_P(root)) {
     old_root = xmlDocGetRootElement(doc);
 
-    if(old_root) {
+    if (old_root) {
       xmlUnlinkNode(old_root);
-      nokogiri_root_node(old_root);
+      noko_xml_document_pin_node(old_root);
     }
 
     return root;
@@ -148,7 +167,7 @@ static VALUE set_root(VALUE self, VALUE root)
 
   /* If the new root's document is not the same as the current document,
    * then we need to dup the node in to this document. */
-  if(new_root->doc != doc) {
+  if (new_root->doc != doc) {
     old_root = xmlDocGetRootElement(doc);
     if (!(new_root = xmlDocCopyNode(new_root, doc, 1))) {
       rb_raise(rb_eRuntimeError, "Could not reparent node (xmlDocCopyNode)");
@@ -156,7 +175,7 @@ static VALUE set_root(VALUE self, VALUE root)
   }
 
   xmlDocSetRootElement(doc, new_root);
-  if(old_root) nokogiri_root_node(old_root);
+  if (old_root) { noko_xml_document_pin_node(old_root); }
   return root;
 }
 
@@ -166,7 +185,8 @@ static VALUE set_root(VALUE self, VALUE root)
  *
  * Get the root node for this document.
  */
-static VALUE root(VALUE self)
+static VALUE
+root(VALUE self)
 {
   xmlDocPtr doc;
   xmlNodePtr root;
@@ -175,8 +195,8 @@ static VALUE root(VALUE self)
 
   root = xmlDocGetRootElement(doc);
 
-  if(!root) return Qnil;
-  return Nokogiri_wrap_xml_node(Qnil, root) ;
+  if (!root) { return Qnil; }
+  return noko_xml_node_wrap(Qnil, root) ;
 }
 
 /*
@@ -185,13 +205,15 @@ static VALUE root(VALUE self)
  *
  * Set the encoding string for this Document
  */
-static VALUE set_encoding(VALUE self, VALUE encoding)
+static VALUE
+set_encoding(VALUE self, VALUE encoding)
 {
   xmlDocPtr doc;
   Data_Get_Struct(self, xmlDoc, doc);
 
-  if (doc->encoding)
-      free((char *)(uintptr_t) doc->encoding); /* avoid gcc cast warning */
+  if (doc->encoding) {
+    free((char *)(uintptr_t) doc->encoding);  /* avoid gcc cast warning */
+  }
 
   doc->encoding = xmlStrdup((xmlChar *)StringValueCStr(encoding));
 
@@ -204,12 +226,13 @@ static VALUE set_encoding(VALUE self, VALUE encoding)
  *
  * Get the encoding for this Document
  */
-static VALUE encoding(VALUE self)
+static VALUE
+encoding(VALUE self)
 {
   xmlDocPtr doc;
   Data_Get_Struct(self, xmlDoc, doc);
 
-  if(!doc->encoding) return Qnil;
+  if (!doc->encoding) { return Qnil; }
   return NOKOGIRI_STR_NEW2(doc->encoding);
 }
 
@@ -219,12 +242,13 @@ static VALUE encoding(VALUE self)
  *
  * Get the XML version for this Document
  */
-static VALUE version(VALUE self)
+static VALUE
+version(VALUE self)
 {
   xmlDocPtr doc;
   Data_Get_Struct(self, xmlDoc, doc);
 
-  if(!doc->version) return Qnil;
+  if (!doc->version) { return Qnil; }
   return NOKOGIRI_STR_NEW2(doc->version);
 }
 
@@ -234,14 +258,15 @@ static VALUE version(VALUE self)
  *
  * Create a new document from an IO object
  */
-static VALUE read_io( VALUE klass,
-                      VALUE io,
-                      VALUE url,
-                      VALUE encoding,
-                      VALUE options )
+static VALUE
+read_io(VALUE klass,
+        VALUE io,
+        VALUE url,
+        VALUE encoding,
+        VALUE options)
 {
-  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
+  const char *c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char *c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
   VALUE error_list      = rb_ary_new();
   VALUE document;
   xmlDocPtr doc;
@@ -250,30 +275,31 @@ static VALUE read_io( VALUE klass,
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 
   doc = xmlReadIO(
-      (xmlInputReadCallback)io_read_callback,
-      (xmlInputCloseCallback)io_close_callback,
-      (void *)io,
-      c_url,
-      c_enc,
-      (int)NUM2INT(options)
-  );
+          (xmlInputReadCallback)noko_io_read,
+          (xmlInputCloseCallback)noko_io_close,
+          (void *)io,
+          c_url,
+          c_enc,
+          (int)NUM2INT(options)
+        );
   xmlSetStructuredErrorFunc(NULL, NULL);
 
-  if(doc == NULL) {
+  if (doc == NULL) {
     xmlErrorPtr error;
 
     xmlFreeDoc(doc);
 
     error = xmlGetLastError();
-    if(error)
+    if (error) {
       rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+    } else {
       rb_raise(rb_eRuntimeError, "Could not parse document");
+    }
 
     return Qnil;
   }
 
-  document = Nokogiri_wrap_xml_document(klass, doc);
+  document = noko_xml_document_wrap(klass, doc);
   rb_iv_set(document, "@errors", error_list);
   return document;
 }
@@ -284,15 +310,16 @@ static VALUE read_io( VALUE klass,
  *
  * Create a new document from a String
  */
-static VALUE read_memory( VALUE klass,
-                          VALUE string,
-                          VALUE url,
-                          VALUE encoding,
-                          VALUE options )
+static VALUE
+read_memory(VALUE klass,
+            VALUE string,
+            VALUE url,
+            VALUE encoding,
+            VALUE options)
 {
-  const char * c_buffer = StringValuePtr(string);
-  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
+  const char *c_buffer = StringValuePtr(string);
+  const char *c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char *c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
   int len               = (int)RSTRING_LEN(string);
   VALUE error_list      = rb_ary_new();
   VALUE document;
@@ -303,21 +330,22 @@ static VALUE read_memory( VALUE klass,
   doc = xmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
   xmlSetStructuredErrorFunc(NULL, NULL);
 
-  if(doc == NULL) {
+  if (doc == NULL) {
     xmlErrorPtr error;
 
     xmlFreeDoc(doc);
 
     error = xmlGetLastError();
-    if(error)
+    if (error) {
       rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+    } else {
       rb_raise(rb_eRuntimeError, "Could not parse document");
+    }
 
     return Qnil;
   }
 
-  document = Nokogiri_wrap_xml_document(klass, doc);
+  document = noko_xml_document_wrap(klass, doc);
   rb_iv_set(document, "@errors", error_list);
   return document;
 }
@@ -329,26 +357,26 @@ static VALUE read_memory( VALUE klass,
  * Copy this Document.  An optional depth may be passed in, but it defaults
  * to a deep copy.  0 is a shallow copy, 1 is a deep copy.
  */
-static VALUE duplicate_document(int argc, VALUE *argv, VALUE self)
+static VALUE
+duplicate_document(int argc, VALUE *argv, VALUE self)
 {
   xmlDocPtr doc, dup;
   VALUE copy;
   VALUE level;
-  VALUE error_list;
 
-  if(rb_scan_args(argc, argv, "01", &level) == 0)
+  if (rb_scan_args(argc, argv, "01", &level) == 0) {
     level = INT2NUM((long)1);
+  }
 
   Data_Get_Struct(self, xmlDoc, doc);
 
   dup = xmlCopyDoc(doc, (int)NUM2INT(level));
 
-  if(dup == NULL) return Qnil;
+  if (dup == NULL) { return Qnil; }
 
   dup->type = doc->type;
-  copy = Nokogiri_wrap_xml_document(rb_obj_class(self), dup);
-  error_list = rb_iv_get(self, "@errors");
-  rb_iv_set(copy, "@errors", error_list);
+  copy = noko_xml_document_wrap(rb_obj_class(self), dup);
+  rb_iv_set(copy, "@errors", rb_iv_get(self, "@errors"));
   return copy ;
 }
 
@@ -358,18 +386,18 @@ static VALUE duplicate_document(int argc, VALUE *argv, VALUE self)
  *
  * Create a new document with +version+ (defaults to "1.0")
  */
-static VALUE new(int argc, VALUE *argv, VALUE klass)
+static VALUE
+new (int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr doc;
   VALUE version, rest, rb_doc ;
 
   rb_scan_args(argc, argv, "0*", &rest);
   version = rb_ary_entry(rest, (long)0);
-  if (NIL_P(version)) version = rb_str_new2("1.0");
+  if (NIL_P(version)) { version = rb_str_new2("1.0"); }
 
   doc = xmlNewDoc((xmlChar *)StringValueCStr(version));
-  rb_doc = Nokogiri_wrap_xml_document(klass, doc);
-  rb_obj_call_init(rb_doc, argc, argv);
+  rb_doc = noko_xml_document_wrap_with_init_args(klass, doc, argc, argv);
   return rb_doc ;
 }
 
@@ -410,7 +438,8 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
  *  please direct your browser to
  *  http://tenderlovemaking.com/2009/04/23/namespaces-in-xml.html
  */
-VALUE remove_namespaces_bang(VALUE self)
+static VALUE
+remove_namespaces_bang(VALUE self)
 {
   xmlDocPtr doc ;
   Data_Get_Struct(self, xmlDoc, doc);
@@ -430,7 +459,8 @@ VALUE remove_namespaces_bang(VALUE self)
  * +external_id+, +system_id+, and +content+ set the External ID, System ID,
  * and content respectively.  All of these parameters are optional.
  */
-static VALUE create_entity(int argc, VALUE *argv, VALUE self)
+static VALUE
+create_entity(int argc, VALUE *argv, VALUE self)
 {
   VALUE name;
   VALUE type;
@@ -443,52 +473,50 @@ static VALUE create_entity(int argc, VALUE *argv, VALUE self)
   Data_Get_Struct(self, xmlDoc, doc);
 
   rb_scan_args(argc, argv, "14", &name, &type, &external_id, &system_id,
-      &content);
+               &content);
 
   xmlResetLastError();
   ptr = xmlAddDocEntity(
-      doc,
-      (xmlChar *)(NIL_P(name)        ? NULL                        : StringValueCStr(name)),
-      (int)      (NIL_P(type)        ? XML_INTERNAL_GENERAL_ENTITY : NUM2INT(type)),
-      (xmlChar *)(NIL_P(external_id) ? NULL                        : StringValueCStr(external_id)),
-      (xmlChar *)(NIL_P(system_id)   ? NULL                        : StringValueCStr(system_id)),
-      (xmlChar *)(NIL_P(content)     ? NULL                        : StringValueCStr(content))
-    );
-
-  if(NULL == ptr) {
+          doc,
+          (xmlChar *)(NIL_P(name)        ? NULL                        : StringValueCStr(name)),
+          (int)(NIL_P(type)        ? XML_INTERNAL_GENERAL_ENTITY : NUM2INT(type)),
+          (xmlChar *)(NIL_P(external_id) ? NULL                        : StringValueCStr(external_id)),
+          (xmlChar *)(NIL_P(system_id)   ? NULL                        : StringValueCStr(system_id)),
+          (xmlChar *)(NIL_P(content)     ? NULL                        : StringValueCStr(content))
+        );
+
+  if (NULL == ptr) {
     xmlErrorPtr error = xmlGetLastError();
-    if(error)
+    if (error) {
       rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+    } else {
       rb_raise(rb_eRuntimeError, "Could not create entity");
+    }
 
     return Qnil;
   }
 
-  return Nokogiri_wrap_xml_node(cNokogiriXmlEntityDecl, (xmlNodePtr)ptr);
+  return noko_xml_node_wrap(cNokogiriXmlEntityDecl, (xmlNodePtr)ptr);
 }
 
-static int block_caller(void * ctx, xmlNodePtr _node, xmlNodePtr _parent)
+static int
+block_caller(void *ctx, xmlNodePtr c_node, xmlNodePtr c_parent_node)
 {
-  VALUE block;
-  VALUE node;
-  VALUE parent;
+  VALUE block = (VALUE)ctx;
+  VALUE rb_node;
+  VALUE rb_parent_node;
   VALUE ret;
 
-  if(_node->type == XML_NAMESPACE_DECL){
-    node = Nokogiri_wrap_xml_namespace(_parent->doc, (xmlNsPtr) _node);
-  }
-  else{
-    node   = Nokogiri_wrap_xml_node(Qnil, _node);
+  if (c_node->type == XML_NAMESPACE_DECL) {
+    rb_node = noko_xml_namespace_wrap((xmlNsPtr)c_node, c_parent_node->doc);
+  } else {
+    rb_node = noko_xml_node_wrap(Qnil, c_node);
   }
-  parent = _parent ? Nokogiri_wrap_xml_node(Qnil, _parent) : Qnil;
-  block  = (VALUE)ctx;
+  rb_parent_node = c_parent_node ? noko_xml_node_wrap(Qnil, c_parent_node) : Qnil;
 
-  ret = rb_funcall(block, rb_intern("call"), 2, node, parent);
+  ret = rb_funcall(block, rb_intern("call"), 2, rb_node, rb_parent_node);
 
-  if(Qfalse == ret || Qnil == ret) return 0;
-
-  return 1;
+  return (Qfalse == ret || Qnil == ret) ? 0 : 1;
 }
 
 /* call-seq:
@@ -501,7 +529,8 @@ static int block_caller(void * ctx, xmlNodePtr _node, xmlNodePtr _parent)
  * The block must return a non-nil, non-false value if the +obj+ passed in
  * should be included in the canonicalized document.
  */
-static VALUE canonicalize(int argc, VALUE* argv, VALUE self)
+static VALUE
+rb_xml_document_canonicalize(int argc, VALUE *argv, VALUE self)
 {
   VALUE mode;
   VALUE incl_ns;
@@ -512,7 +541,7 @@ static VALUE canonicalize(int argc, VALUE* argv, VALUE self)
   xmlDocPtr doc;
   xmlOutputBufferPtr buf;
   xmlC14NIsVisibleCallback cb = NULL;
-  void * ctx = NULL;
+  void *ctx = NULL;
 
   VALUE rb_cStringIO;
   VALUE io;
@@ -525,93 +554,126 @@ static VALUE canonicalize(int argc, VALUE* argv, VALUE self)
   io           = rb_class_new_instance(0, 0, rb_cStringIO);
   buf          = xmlAllocOutputBuffer(NULL);
 
-  buf->writecallback = (xmlOutputWriteCallback)io_write_callback;
-  buf->closecallback = (xmlOutputCloseCallback)io_close_callback;
+  buf->writecallback = (xmlOutputWriteCallback)noko_io_write;
+  buf->closecallback = (xmlOutputCloseCallback)noko_io_close;
   buf->context       = (void *)io;
 
-  if(rb_block_given_p()) {
+  if (rb_block_given_p()) {
     cb = block_caller;
     ctx = (void *)rb_block_proc();
   }
 
-  if(NIL_P(incl_ns)){
+  if (NIL_P(incl_ns)) {
     ns = NULL;
-  }
-  else{
+  } else {
     Check_Type(incl_ns, T_ARRAY);
     ns_len = RARRAY_LEN(incl_ns);
-    ns = calloc((size_t)ns_len+1, sizeof(xmlChar *));
+    ns = calloc((size_t)ns_len + 1, sizeof(xmlChar *));
     for (i = 0 ; i < ns_len ; i++) {
       VALUE entry = rb_ary_entry(incl_ns, i);
-      ns[i] = (xmlChar*)StringValueCStr(entry);
+      ns[i] = (xmlChar *)StringValueCStr(entry);
     }
   }
 
 
   xmlC14NExecute(doc, cb, ctx,
-    (int)      (NIL_P(mode)        ? 0 : NUM2INT(mode)),
-    ns,
-    (int)      RTEST(with_comments),
-    buf);
+                 (int)(NIL_P(mode)        ? 0 : NUM2INT(mode)),
+                 ns,
+                 (int)      RTEST(with_comments),
+                 buf);
 
   xmlOutputBufferClose(buf);
 
   return rb_funcall(io, rb_intern("string"), 0);
 }
 
-VALUE cNokogiriXmlDocument ;
-void init_xml_document()
+VALUE
+noko_xml_document_wrap_with_init_args(VALUE klass, xmlDocPtr c_document, int argc, VALUE *argv)
 {
-  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 rb_document;
+  nokogiriTuplePtr tuple;
 
-  /*
-   * Nokogiri::XML::Document wraps an xml document.
-   */
-  VALUE klass = rb_define_class_under(xml, "Document", node);
-
-  cNokogiriXmlDocument = klass;
-
-  rb_define_singleton_method(klass, "read_memory", read_memory, 4);
-  rb_define_singleton_method(klass, "read_io", read_io, 4);
-  rb_define_singleton_method(klass, "new", new, -1);
-
-  rb_define_method(klass, "root", root, 0);
-  rb_define_method(klass, "root=", set_root, 1);
-  rb_define_method(klass, "encoding", encoding, 0);
-  rb_define_method(klass, "encoding=", set_encoding, 1);
-  rb_define_method(klass, "version", version, 0);
-  rb_define_method(klass, "canonicalize", canonicalize, -1);
-  rb_define_method(klass, "dup", duplicate_document, -1);
-  rb_define_method(klass, "url", url, 0);
-  rb_define_method(klass, "create_entity", create_entity, -1);
-  rb_define_method(klass, "remove_namespaces!", remove_namespaces_bang, 0);
+  if (!klass) {
+    klass = cNokogiriXmlDocument;
+  }
+
+  rb_document = Data_Wrap_Struct(klass, mark, dealloc, c_document);
+
+  tuple = (nokogiriTuplePtr)malloc(sizeof(nokogiriTuple));
+  tuple->doc = rb_document;
+  tuple->unlinkedNodes = st_init_numtable_with_size(128);
+  tuple->node_cache = rb_ary_new();
+
+  c_document->_private = tuple ;
+
+  rb_iv_set(rb_document, "@decorators", Qnil);
+  rb_iv_set(rb_document, "@errors", Qnil);
+  rb_iv_set(rb_document, "@node_cache", tuple->node_cache);
+
+  rb_obj_call_init(rb_document, argc, argv);
+
+  return rb_document ;
 }
 
 
-/* this takes klass as a param because it's used for HtmlDocument, too. */
-VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
+/* deprecated. use noko_xml_document_wrap() instead. */
+VALUE
+Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
 {
-  nokogiriTuplePtr tuple = (nokogiriTuplePtr)malloc(sizeof(nokogiriTuple));
+  /* TODO: deprecate this method in v2.0 */
+  return noko_xml_document_wrap_with_init_args(klass, doc, 0, NULL);
+}
 
-  VALUE rb_doc = Data_Wrap_Struct(
-      klass ? klass : cNokogiriXmlDocument,
-      mark,
-      dealloc,
-      doc
-  );
+VALUE
+noko_xml_document_wrap(VALUE klass, xmlDocPtr doc)
+{
+  return noko_xml_document_wrap_with_init_args(klass, doc, 0, NULL);
+}
 
-  VALUE cache = rb_ary_new();
-  rb_iv_set(rb_doc, "@decorators", Qnil);
-  rb_iv_set(rb_doc, "@node_cache", cache);
 
-  tuple->doc = rb_doc;
-  tuple->unlinkedNodes = st_init_numtable_with_size(128);
-  tuple->node_cache = cache;
-  doc->_private = tuple ;
+void
+noko_xml_document_pin_node(xmlNodePtr node)
+{
+  xmlDocPtr doc;
+  nokogiriTuplePtr tuple;
 
-  rb_obj_call_init(rb_doc, 0, NULL);
+  doc = node->doc;
+  tuple = (nokogiriTuplePtr)doc->_private;
+  st_insert(tuple->unlinkedNodes, (st_data_t)node, (st_data_t)node);
+}
 
-  return rb_doc ;
+
+void
+noko_xml_document_pin_namespace(xmlNsPtr ns, xmlDocPtr doc)
+{
+  nokogiriTuplePtr tuple;
+
+  tuple = (nokogiriTuplePtr)doc->_private;
+  st_insert(tuple->unlinkedNodes, (st_data_t)ns, (st_data_t)ns);
+}
+
+
+void
+noko_init_xml_document()
+{
+  assert(cNokogiriXmlNode);
+  /*
+   * Nokogiri::XML::Document wraps an xml document.
+   */
+  cNokogiriXmlDocument = rb_define_class_under(mNokogiriXml, "Document", cNokogiriXmlNode);
+
+  rb_define_singleton_method(cNokogiriXmlDocument, "read_memory", read_memory, 4);
+  rb_define_singleton_method(cNokogiriXmlDocument, "read_io", read_io, 4);
+  rb_define_singleton_method(cNokogiriXmlDocument, "new", new, -1);
+
+  rb_define_method(cNokogiriXmlDocument, "root", root, 0);
+  rb_define_method(cNokogiriXmlDocument, "root=", set_root, 1);
+  rb_define_method(cNokogiriXmlDocument, "encoding", encoding, 0);
+  rb_define_method(cNokogiriXmlDocument, "encoding=", set_encoding, 1);
+  rb_define_method(cNokogiriXmlDocument, "version", version, 0);
+  rb_define_method(cNokogiriXmlDocument, "canonicalize", rb_xml_document_canonicalize, -1);
+  rb_define_method(cNokogiriXmlDocument, "dup", duplicate_document, -1);
+  rb_define_method(cNokogiriXmlDocument, "url", url, 0);
+  rb_define_method(cNokogiriXmlDocument, "create_entity", create_entity, -1);
+  rb_define_method(cNokogiriXmlDocument, "remove_namespaces!", remove_namespaces_bang, 0);
 }