nokogiri 1.5.10 → 1.12.5

data/ext/nokogiri/xml_document.c

@@ -1,29 +1,71 @@
-#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) {
-  case XML_ATTRIBUTE_NODE:
-    xmlFreePropList((xmlAttrPtr)node);
-    break;
-  case XML_NAMESPACE_DECL:
-    xmlFree(node);
-    break;
-  default:
-    if(node->parent == NULL) {
-      xmlAddChild((xmlNodePtr)doc, node);
-    }
+  switch (node->type) {
+    case XML_ATTRIBUTE_NODE:
+      xmlFreePropList((xmlAttrPtr)node);
+      break;
+    case XML_NAMESPACE_DECL:
+      xmlFreeNs((xmlNsPtr)node);
+      break;
+    case XML_DTD_NODE:
+      xmlFreeDtd((xmlDtdPtr)node);
+      break;
+    default:
+      if (node->parent == NULL) {
+        xmlAddChild((xmlNodePtr)doc, node);
+      }
   }
   return ST_CONTINUE;
 }
 
-static void dealloc(xmlDocPtr doc)
+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) {
+    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) {
+      remove_private(child);
+    }
+  }
+
+  node->_private = NULL;
+}
+
+static void
+mark(xmlDocPtr doc)
+{
+  nokogiriTuplePtr tuple = (nokogiriTuplePtr)doc->_private;
+  if (tuple) {
+    rb_gc_mark(tuple->doc);
+    rb_gc_mark(tuple->node_cache);
+  }
+}
+
+static void
+dealloc(xmlDocPtr doc)
 {
-  xmlDeregisterNodeFunc func;
   st_table *node_hash;
 
   NOKOGIRI_DEBUG_START(doc);
-  func = xmlDeregisterNodeDefault(NULL);
 
   node_hash  = DOC_UNLINKED_NODE_HASH(doc);
 
@@ -31,22 +73,32 @@ static void dealloc(xmlDocPtr doc)
   st_free_table(node_hash);
 
   free(doc->_private);
-  doc->_private = NULL;
+
+  /* When both Nokogiri and libxml-ruby are loaded, make sure that all nodes
+   * have their _private pointers cleared. This is to avoid libxml-ruby's
+   * xmlDeregisterNode callback from accessing VALUE pointers from ruby's GC
+   * free context, which can result in segfaults.
+   */
+  if (xmlDeregisterNodeDefaultValue) {
+    remove_private((xmlNodePtr)doc);
+  }
+
   xmlFreeDoc(doc);
 
-  xmlDeregisterNodeDefault(func);
   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) ||
@@ -59,7 +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 ;
     }
   }
@@ -71,12 +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;
 }
@@ -87,42 +140,42 @@ static VALUE url(VALUE self)
  *
  * Set the root element on this document
  */
-static VALUE set_root(VALUE self, VALUE root)
+static VALUE
+rb_xml_document_root_set(VALUE self, VALUE rb_new_root)
 {
-  xmlDocPtr doc;
-  xmlNodePtr new_root;
-  xmlNodePtr old_root;
+  xmlDocPtr c_document;
+  xmlNodePtr c_new_root = NULL, c_current_root;
 
-  Data_Get_Struct(self, xmlDoc, doc);
-
-  old_root = NULL;
-
-  if(NIL_P(root)) {
-    old_root = xmlDocGetRootElement(doc);
-
-    if(old_root) {
-      xmlUnlinkNode(old_root);
-      nokogiri_root_node(old_root);
-    }
+  Data_Get_Struct(self, xmlDoc, c_document);
 
-    return root;
+  c_current_root = xmlDocGetRootElement(c_document);
+  if (c_current_root) {
+    xmlUnlinkNode(c_current_root);
+    noko_xml_document_pin_node(c_current_root);
   }
 
-  Data_Get_Struct(root, xmlNode, new_root);
+  if (!NIL_P(rb_new_root)) {
+    if (!rb_obj_is_kind_of(rb_new_root, cNokogiriXmlNode)) {
+      rb_raise(rb_eArgError,
+               "expected Nokogiri::XML::Node but received %"PRIsVALUE,
+               rb_obj_class(rb_new_root));
+    }
 
+    Data_Get_Struct(rb_new_root, xmlNode, c_new_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) {
-    old_root = xmlDocGetRootElement(doc);
-    if (!(new_root = xmlDocCopyNode(new_root, doc, 1))) {
-      rb_raise(rb_eRuntimeError, "Could not reparent node (xmlDocCopyNode)");
+    /* 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 (c_new_root->doc != c_document) {
+      c_new_root = xmlDocCopyNode(c_new_root, c_document, 1);
+      if (!c_new_root) {
+        rb_raise(rb_eRuntimeError, "Could not reparent node (xmlDocCopyNode)");
+      }
     }
   }
 
-  xmlDocSetRootElement(doc, new_root);
-  if(old_root) nokogiri_root_node(old_root);
-  return root;
+  xmlDocSetRootElement(c_document, c_new_root);
+
+  return rb_new_root;
 }
 
 /*
@@ -131,17 +184,20 @@ static VALUE set_root(VALUE self, VALUE root)
  *
  * Get the root node for this document.
  */
-static VALUE root(VALUE self)
+static VALUE
+rb_xml_document_root(VALUE self)
 {
-  xmlDocPtr doc;
-  xmlNodePtr root;
+  xmlDocPtr c_document;
+  xmlNodePtr c_root;
 
-  Data_Get_Struct(self, xmlDoc, doc);
+  Data_Get_Struct(self, xmlDoc, c_document);
 
-  root = xmlDocGetRootElement(doc);
+  c_root = xmlDocGetRootElement(c_document);
+  if (!c_root) {
+    return Qnil;
+  }
 
-  if(!root) return Qnil;
-  return Nokogiri_wrap_xml_node(Qnil, root) ;
+  return noko_xml_node_wrap(Qnil, c_root) ;
 }
 
 /*
@@ -150,15 +206,17 @@ 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 *) doc->encoding); /* this may produce a gcc cast warning */
+  if (doc->encoding) {
+    xmlFree(DISCARD_CONST_QUAL_XMLCHAR(doc->encoding));
+  }
 
-  doc->encoding = xmlStrdup((xmlChar *)StringValuePtr(encoding));
+  doc->encoding = xmlStrdup((xmlChar *)StringValueCStr(encoding));
 
   return encoding;
 }
@@ -169,12 +227,13 @@ 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);
 }
 
@@ -184,12 +243,13 @@ 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);
 }
 
@@ -199,14 +259,15 @@ 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 : StringValuePtr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(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;
@@ -215,30 +276,31 @@ 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)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
-    else
+    if (error) {
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
+    } 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;
 }
@@ -249,15 +311,16 @@ 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 : StringValuePtr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(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;
@@ -268,21 +331,22 @@ 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)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
-    else
+    if (error) {
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
+    } 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;
 }
@@ -294,21 +358,27 @@ 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_node(int argc, VALUE *argv, VALUE self)
+static VALUE
+duplicate_document(int argc, VALUE *argv, VALUE self)
 {
   xmlDocPtr doc, dup;
+  VALUE copy;
   VALUE level;
 
-  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;
-  return Nokogiri_wrap_xml_document(rb_obj_class(self), dup);
+  copy = noko_xml_document_wrap(rb_obj_class(self), dup);
+  rb_iv_set(copy, "@errors", rb_iv_get(self, "@errors"));
+  return copy ;
 }
 
 /*
@@ -317,18 +387,18 @@ static VALUE duplicate_node(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 *)StringValuePtr(version));
-  rb_doc = Nokogiri_wrap_xml_document(klass, doc);
-  rb_obj_call_init(rb_doc, argc, argv);
+  doc = xmlNewDoc((xmlChar *)StringValueCStr(version));
+  rb_doc = noko_xml_document_wrap_with_init_args(klass, doc, argc, argv);
   return rb_doc ;
 }
 
@@ -354,13 +424,13 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
  *         </bicycle>
  *       </root>
  *       EOXML
- *    
+ *
  *    doc.xpath("//tire").to_s # => ""
  *    doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => "<part:tire>Michelin Model XGV</part:tire>"
  *    doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => "<part:tire>I'm a bicycle tire!</part:tire>"
- *    
+ *
  *    doc.remove_namespaces!
- *    
+ *
  *    doc.xpath("//tire").to_s # => "<tire>Michelin Model XGV</tire><tire>I'm a bicycle tire!</tire>"
  *    doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => ""
  *    doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => ""
@@ -369,7 +439,8 @@ 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);
@@ -389,7 +460,8 @@ 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;
@@ -402,52 +474,50 @@ 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                        : StringValuePtr(name)),
-      (int)      (NIL_P(type)        ? XML_INTERNAL_GENERAL_ENTITY : NUM2INT(type)),
-      (xmlChar *)(NIL_P(external_id) ? NULL                        : StringValuePtr(external_id)),
-      (xmlChar *)(NIL_P(system_id)   ? NULL                        : StringValuePtr(system_id)),
-      (xmlChar *)(NIL_P(content)     ? NULL                        : StringValuePtr(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)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
-    else
+    if (error) {
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
+    } 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;
-
-  ret = rb_funcall(block, rb_intern("call"), 2, node, parent);
+  rb_parent_node = c_parent_node ? noko_xml_node_wrap(Qnil, c_parent_node) : Qnil;
 
-  if(Qfalse == ret || Qnil == ret) return 0;
+  ret = rb_funcall(block, rb_intern("call"), 2, rb_node, rb_parent_node);
 
-  return 1;
+  return (Qfalse == ret || Qnil == ret) ? 0 : 1;
 }
 
 /* call-seq:
@@ -455,12 +525,13 @@ static int block_caller(void * ctx, xmlNodePtr _node, xmlNodePtr _parent)
  *  doc.canonicalize { |obj, parent| ... }
  *
  * Canonicalize a document and return the results.  Takes an optional block
- * that takes two parameters: the +obj+ and that node's +parent+.  
+ * that takes two parameters: the +obj+ and that node's +parent+.
  * The  +obj+ will be either a Nokogiri::XML::Node, or a Nokogiri::XML::Namespace
- * The block must return a non-nil, non-false value if the +obj+ passed in 
+ * 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;
@@ -471,7 +542,7 @@ 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;
@@ -484,93 +555,126 @@ 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);
-      const char * ptr = StringValuePtr(entry);
-      ns[i] = (xmlChar*) ptr;
+      ns[i] = (xmlChar *)StringValueCStr(entry);
     }
   }
 
 
-  xmlC14NExecute(doc, cb, ctx, 
-    (int)      (NIL_P(mode)        ? 0 : NUM2INT(mode)), 
-    ns,
-    (int)      (NIL_P(with_comments)        ? 0 : 1),
-    buf);
+  xmlC14NExecute(doc, cb, ctx,
+                 (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_node, -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,
-      0,
-      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", rb_xml_document_root, 0);
+  rb_define_method(cNokogiriXmlDocument, "root=", rb_xml_document_root_set, 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);
 }