nokogiri 1.10.9 → 1.18.3

data/ext/nokogiri/xml_document.c

@@ -1,91 +1,225 @@
-#include <xml_document.h>
-
-static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
-{
-  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);
-    }
+#include <nokogiri.h>
+
+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:
+      xmlFreeNs((xmlNsPtr)node);
+      break;
+    case XML_DTD_NODE:
+      xmlFreeDtd((xmlDtdPtr)node);
+      break;
+    default:
+      if (node->parent == NULL) {
+        node->next = NULL;
+        node->prev = 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 dealloc(xmlDocPtr doc)
+static void
+mark(void *data)
 {
-  st_table *node_hash;
+  xmlDocPtr doc = (xmlDocPtr)data;
+  nokogiriTuplePtr tuple = (nokogiriTuplePtr)doc->_private;
+  if (tuple) {
+    rb_gc_mark(tuple->doc);
+    rb_gc_mark(tuple->node_cache);
+  }
+}
 
-  NOKOGIRI_DEBUG_START(doc);
+static void
+dealloc(void *data)
+{
+  xmlDocPtr doc = (xmlDocPtr)data;
+  st_table *node_hash;
 
   node_hash  = DOC_UNLINKED_NODE_HASH(doc);
 
   st_foreach(node_hash, dealloc_node_i, (st_data_t)doc);
   st_free_table(node_hash);
 
-  free(doc->_private);
+  ruby_xfree(doc->_private);
 
-  /* 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 defined(__GNUC__) && __GNUC__ >= 5
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wdeprecated-declarations" // xmlDeregisterNodeDefault is deprecated as of libxml2 2.11.0
+#endif
+  /*
+   * libxml-ruby < 3.0.0 uses xmlDeregisterNodeDefault. If the user is using one of those older
+   * versions, the registered callback from libxml-ruby will access the _private pointers set by
+   * nokogiri, which will result in segfaults.
+   *
+   * To avoid this, we need to clear the _private pointers from all nodes in this document tree
+   * before that callback gets invoked.
+   *
+   * libxml-ruby 3.0.0 was released in 2017-02, so at some point we can probably safely remove this
+   * safeguard (though probably pairing with a runtime check on the libxml-ruby version).
    */
-  if (xmlDeregisterNodeDefaultValue)
+  if (xmlDeregisterNodeDefaultValue) {
     remove_private((xmlNodePtr)doc);
+  }
+#if defined(__GNUC__) && __GNUC__ >= 5
+#pragma GCC diagnostic pop
+#endif
 
   xmlFreeDoc(doc);
+}
 
-  NOKOGIRI_DEBUG_END(doc);
+static size_t
+memsize_node(const xmlNodePtr node)
+{
+  /* note we don't count namespace definitions, just going for a good-enough number here */
+  xmlNodePtr child;
+  xmlAttrPtr property;
+  size_t memsize = 0;
+
+  memsize += (size_t)xmlStrlen(node->name);
+
+  if (node->type == XML_ELEMENT_NODE) {
+    for (property = node->properties; property; property = property->next) {
+      memsize += sizeof(xmlAttr) + memsize_node((xmlNodePtr)property);
+    }
+  }
+  if (node->type == XML_TEXT_NODE) {
+    memsize += (size_t)xmlStrlen(node->content);
+  }
+  for (child = node->children; child; child = child->next) {
+    memsize += sizeof(xmlNode) + memsize_node(child);
+  }
+  return memsize;
+}
+
+static size_t
+memsize(const void *data)
+{
+  xmlDocPtr doc = (const xmlDocPtr)data;
+  size_t memsize = sizeof(xmlDoc);
+  /* This may not account for all memory use */
+  memsize += memsize_node((xmlNodePtr)doc);
+  return memsize;
 }
 
-static void recursively_remove_namespaces_from_node(xmlNodePtr node)
+static const rb_data_type_t xml_doc_type = {
+  .wrap_struct_name = "xmlDoc",
+  .function = {
+    .dmark = mark,
+    .dfree = dealloc,
+    .dsize = memsize,
+  },
+  // .flags = RUBY_TYPED_FREE_IMMEDIATELY, // TODO see https://github.com/sparklemotion/nokogiri/issues/2822
+};
+
+static VALUE
+_xml_document_alloc(VALUE klass)
+{
+  return TypedData_Wrap_Struct(klass, &xml_doc_type, NULL);
+}
+
+static void
+_xml_document_data_ptr_set(VALUE rb_document, xmlDocPtr c_document)
+{
+  nokogiriTuplePtr tuple;
+
+  assert(DATA_PTR(rb_document) == NULL);
+  assert(c_document->_private == NULL);
+
+  DATA_PTR(rb_document) = c_document;
+
+  tuple = (nokogiriTuplePtr)ruby_xmalloc(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, "@node_cache", tuple->node_cache);
+
+  return;
+}
+
+/* :nodoc: */
+static VALUE
+rb_xml_document_initialize_copy_with_args(VALUE rb_self, VALUE rb_other, VALUE rb_level)
+{
+  xmlDocPtr c_other, c_self;
+  int c_level;
+
+  c_other = noko_xml_document_unwrap(rb_other);
+  c_level = (int)NUM2INT(rb_level);
+
+  c_self = xmlCopyDoc(c_other, c_level);
+  if (c_self == NULL) { return Qnil; }
+
+  c_self->type = c_other->type;
+  _xml_document_data_ptr_set(rb_self, c_self);
+
+  return rb_self ;
+}
+
+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) ||
        (node->type == XML_XINCLUDE_END)) &&
       node->nsDef) {
-    xmlFreeNsList(node->nsDef);
+    xmlNsPtr curr = node->nsDef;
+    while (curr) {
+      noko_xml_document_pin_namespace(curr, node->doc);
+      curr = curr->next;
+    }
     node->nsDef = NULL;
   }
 
   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 ;
     }
   }
@@ -97,12 +231,12 @@ 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);
+  xmlDocPtr doc = noko_xml_document_unwrap(self);
 
-  if(doc->URL) return NOKOGIRI_STR_NEW2(doc->URL);
+  if (doc->URL) { return NOKOGIRI_STR_NEW2(doc->URL); }
 
   return Qnil;
 }
@@ -113,42 +247,42 @@ 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;
-
-  Data_Get_Struct(self, xmlDoc, doc);
-
-  old_root = NULL;
+  xmlDocPtr c_document;
+  xmlNodePtr c_new_root = NULL, c_current_root;
 
-  if(NIL_P(root)) {
-    old_root = xmlDocGetRootElement(doc);
+  c_document = noko_xml_document_unwrap(self);
 
-    if(old_root) {
-      xmlUnlinkNode(old_root);
-      nokogiri_root_node(old_root);
-    }
-
-    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));
+    }
 
+    Noko_Node_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;
 }
 
 /*
@@ -157,17 +291,20 @@ 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);
+  c_document = noko_xml_document_unwrap(self);
 
-  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) ;
 }
 
 /*
@@ -176,13 +313,14 @@ 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);
+  xmlDocPtr doc = noko_xml_document_unwrap(self);
 
-  if (doc->encoding)
-      free((char *)(uintptr_t) doc->encoding); /* avoid gcc cast warning */
+  if (doc->encoding) {
+    xmlFree(DISCARD_CONST_QUAL_XMLCHAR(doc->encoding));
+  }
 
   doc->encoding = xmlStrdup((xmlChar *)StringValueCStr(encoding));
 
@@ -195,12 +333,12 @@ 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);
+  xmlDocPtr doc = noko_xml_document_unwrap(self);
 
-  if(!doc->encoding) return Qnil;
+  if (!doc->encoding) { return Qnil; }
   return NOKOGIRI_STR_NEW2(doc->encoding);
 }
 
@@ -210,12 +348,12 @@ 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);
+  xmlDocPtr doc = noko_xml_document_unwrap(self);
 
-  if(!doc->version) return Qnil;
+  if (!doc->version) { return Qnil; }
   return NOKOGIRI_STR_NEW2(doc->version);
 }
 
@@ -225,48 +363,47 @@ 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 )
-{
-  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;
-
-  xmlResetLastError();
-  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)
-  );
-  xmlSetStructuredErrorFunc(NULL, NULL);
-
-  if(doc == NULL) {
-    xmlErrorPtr error;
-
-    xmlFreeDoc(doc);
-
-    error = xmlGetLastError();
-    if(error)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+static VALUE
+noko_xml_document_s_read_io(VALUE rb_class,
+                            VALUE rb_io,
+                            VALUE rb_url,
+                            VALUE rb_encoding,
+                            VALUE rb_options)
+{
+  /* TODO: deprecate this method, parse should be the preferred entry point. then we can make this
+     private. */
+  libxmlStructuredErrorHandlerState handler_state;
+  VALUE rb_errors = rb_ary_new();
+
+  noko__structured_error_func_save_and_set(&handler_state, (void *)rb_errors, noko__error_array_pusher);
+
+  const char *c_url    = NIL_P(rb_url)      ? NULL : StringValueCStr(rb_url);
+  const char *c_enc    = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
+  xmlDocPtr c_document = xmlReadIO(
+                           (xmlInputReadCallback)noko_io_read,
+                           (xmlInputCloseCallback)noko_io_close,
+                           (void *)rb_io,
+                           c_url,
+                           c_enc,
+                           (int)NUM2INT(rb_options)
+                         );
+
+  noko__structured_error_func_restore(&handler_state);
+
+  if (c_document == NULL) {
+    xmlFreeDoc(c_document);
+
+    VALUE exception = rb_funcall(cNokogiriXmlSyntaxError, rb_intern("aggregate"), 1, rb_errors);
+    if (RB_TEST(exception)) {
+      rb_exc_raise(exception);
+    } else {
       rb_raise(rb_eRuntimeError, "Could not parse document");
-
-    return Qnil;
+    }
   }
 
-  document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_iv_set(document, "@errors", error_list);
-  return document;
+  VALUE rb_document = noko_xml_document_wrap(rb_class, c_document);
+  rb_iv_set(rb_document, "@errors", rb_errors);
+  return rb_document;
 }
 
 /*
@@ -275,92 +412,58 @@ 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 )
-{
-  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;
-  xmlDocPtr doc;
-
-  xmlResetLastError();
-  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
-  doc = xmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
-  xmlSetStructuredErrorFunc(NULL, NULL);
+static VALUE
+noko_xml_document_s_read_memory(VALUE rb_class,
+                                VALUE rb_input,
+                                VALUE rb_url,
+                                VALUE rb_encoding,
+                                VALUE rb_options)
+{
+  /* TODO: deprecate this method, parse should be the preferred entry point. then we can make this
+     private. */
+  VALUE rb_errors = rb_ary_new();
+  xmlSetStructuredErrorFunc((void *)rb_errors, noko__error_array_pusher);
 
-  if(doc == NULL) {
-    xmlErrorPtr error;
+  const char *c_buffer = StringValuePtr(rb_input);
+  const char *c_url    = NIL_P(rb_url)      ? NULL : StringValueCStr(rb_url);
+  const char *c_enc    = NIL_P(rb_encoding) ? NULL : StringValueCStr(rb_encoding);
+  int c_buffer_len     = (int)RSTRING_LEN(rb_input);
+  xmlDocPtr c_document = xmlReadMemory(c_buffer, c_buffer_len, c_url, c_enc, (int)NUM2INT(rb_options));
 
-    xmlFreeDoc(doc);
+  xmlSetStructuredErrorFunc(NULL, NULL);
 
-    error = xmlGetLastError();
-    if(error)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+  if (c_document == NULL) {
+    VALUE exception = rb_funcall(cNokogiriXmlSyntaxError, rb_intern("aggregate"), 1, rb_errors);
+    if (RB_TEST(exception)) {
+      rb_exc_raise(exception);
+    } else {
       rb_raise(rb_eRuntimeError, "Could not parse document");
-
-    return Qnil;
+    }
   }
 
-  document = Nokogiri_wrap_xml_document(klass, doc);
-  rb_iv_set(document, "@errors", error_list);
+  VALUE document = noko_xml_document_wrap(rb_class, c_document);
+  rb_iv_set(document, "@errors", rb_errors);
   return document;
 }
 
 /*
  * call-seq:
- *  dup
+ *  new(version = "1.0")
  *
- * 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.
+ * Create a new empty document declaring XML version +version+.
  */
-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)
-    level = INT2NUM((long)1);
-
-  Data_Get_Struct(self, xmlDoc, doc);
-
-  dup = xmlCopyDoc(doc, (int)NUM2INT(level));
-
-  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);
-  return copy ;
-}
-
-/*
- * call-seq:
- *  new(version = default)
- *
- * 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 ;
 }
 
@@ -401,85 +504,87 @@ 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);
+  xmlDocPtr doc = noko_xml_document_unwrap(self);
 
   recursively_remove_namespaces_from_node((xmlNodePtr)doc);
   return self;
 }
 
-/* call-seq: doc.create_entity(name, type, external_id, system_id, content)
+/* call-seq:
+ *   doc.create_entity(name, type, external_id, system_id, content)
  *
  * Create a new entity named +name+.
  *
- * +type+ is an integer representing the type of entity to be created, and it
- * defaults to Nokogiri::XML::EntityDecl::INTERNAL_GENERAL.  See
- * the constants on Nokogiri::XML::EntityDecl for more information.
+ * +type+ is an integer representing the type of entity to be created, and it defaults to
+ * +Nokogiri::XML::EntityDecl::INTERNAL_GENERAL+. See the constants on Nokogiri::XML::EntityDecl for
+ * more information.
  *
  * +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)
-{
-  VALUE name;
-  VALUE type;
-  VALUE external_id;
-  VALUE system_id;
-  VALUE content;
-  xmlEntityPtr ptr;
-  xmlDocPtr doc ;
-
-  Data_Get_Struct(self, xmlDoc, doc);
-
-  rb_scan_args(argc, argv, "14", &name, &type, &external_id, &system_id,
-      &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) {
-    xmlErrorPtr error = xmlGetLastError();
-    if(error)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-    else
+static VALUE
+noko_xml_document__create_entity(int argc, VALUE *argv, VALUE rb_document)
+{
+  VALUE rb_name;
+  VALUE rb_type;
+  VALUE rb_ext_id;
+  VALUE rb_sys_id;
+  VALUE rb_content;
+
+  rb_scan_args(argc, argv, "14",
+               &rb_name,
+               &rb_type, &rb_ext_id, &rb_sys_id, &rb_content);
+
+  xmlDocPtr c_document = noko_xml_document_unwrap(rb_document);
+
+  libxmlStructuredErrorHandlerState handler_state;
+  VALUE rb_errors = rb_ary_new();
+  noko__structured_error_func_save_and_set(&handler_state, (void *)rb_errors, noko__error_array_pusher);
+
+  xmlEntityPtr c_entity = xmlAddDocEntity(
+                            c_document,
+                            (xmlChar *)(NIL_P(rb_name) ? NULL : StringValueCStr(rb_name)),
+                            (int)(NIL_P(rb_type) ? XML_INTERNAL_GENERAL_ENTITY : NUM2INT(rb_type)),
+                            (xmlChar *)(NIL_P(rb_ext_id) ? NULL : StringValueCStr(rb_ext_id)),
+                            (xmlChar *)(NIL_P(rb_sys_id) ? NULL : StringValueCStr(rb_sys_id)),
+                            (xmlChar *)(NIL_P(rb_content) ? NULL : StringValueCStr(rb_content))
+                          );
+
+  noko__structured_error_func_restore(&handler_state);
+
+  if (NULL == c_entity) {
+    VALUE exception = rb_funcall(cNokogiriXmlSyntaxError, rb_intern("aggregate"), 1, rb_errors);
+    if (RB_TEST(exception)) {
+      rb_exc_raise(exception);
+    } 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)c_entity);
 }
 
-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:
@@ -492,117 +597,188 @@ 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;
-  VALUE with_comments;
-  xmlChar **ns;
-  long ns_len, i;
+  VALUE rb_mode;
+  VALUE rb_namespaces;
+  VALUE rb_comments_p;
+  int c_mode = 0;
+  xmlChar **c_namespaces;
 
-  xmlDocPtr doc;
-  xmlOutputBufferPtr buf;
-  xmlC14NIsVisibleCallback cb = NULL;
-  void * ctx = NULL;
+  xmlDocPtr c_doc;
+  xmlOutputBufferPtr c_obuf;
+  xmlC14NIsVisibleCallback c_callback_wrapper = NULL;
+  void *rb_callback = NULL;
 
   VALUE rb_cStringIO;
-  VALUE io;
+  VALUE rb_io;
 
-  rb_scan_args(argc, argv, "03", &mode, &incl_ns, &with_comments);
+  rb_scan_args(argc, argv, "03", &rb_mode, &rb_namespaces, &rb_comments_p);
+  if (!NIL_P(rb_mode)) {
+    Check_Type(rb_mode, T_FIXNUM);
+    c_mode = NUM2INT(rb_mode);
+  }
+  if (!NIL_P(rb_namespaces)) {
+    Check_Type(rb_namespaces, T_ARRAY);
+    if (c_mode == XML_C14N_1_0 || c_mode == XML_C14N_1_1) {
+      rb_raise(rb_eRuntimeError, "This canonicalizer does not support this operation");
+    }
+  }
 
-  Data_Get_Struct(self, xmlDoc, doc);
+  c_doc = noko_xml_document_unwrap(self);
 
   rb_cStringIO = rb_const_get_at(rb_cObject, rb_intern("StringIO"));
-  io           = rb_class_new_instance(0, 0, rb_cStringIO);
-  buf          = xmlAllocOutputBuffer(NULL);
+  rb_io = rb_class_new_instance(0, 0, rb_cStringIO);
+  c_obuf = xmlAllocOutputBuffer(NULL);
 
-  buf->writecallback = (xmlOutputWriteCallback)io_write_callback;
-  buf->closecallback = (xmlOutputCloseCallback)io_close_callback;
-  buf->context       = (void *)io;
+  c_obuf->writecallback = (xmlOutputWriteCallback)noko_io_write;
+  c_obuf->closecallback = (xmlOutputCloseCallback)noko_io_close;
+  c_obuf->context = (void *)rb_io;
 
-  if(rb_block_given_p()) {
-    cb = block_caller;
-    ctx = (void *)rb_block_proc();
+  if (rb_block_given_p()) {
+    c_callback_wrapper = block_caller;
+    rb_callback = (void *)rb_block_proc();
   }
 
-  if(NIL_P(incl_ns)){
-    ns = NULL;
-  }
-  else{
-    Check_Type(incl_ns, T_ARRAY);
-    ns_len = RARRAY_LEN(incl_ns);
-    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);
+  if (NIL_P(rb_namespaces)) {
+    c_namespaces = NULL;
+  } else {
+    long ns_len = RARRAY_LEN(rb_namespaces);
+    c_namespaces = ruby_xcalloc((size_t)ns_len + 1, sizeof(xmlChar *));
+    for (int j = 0 ; j < ns_len ; j++) {
+      VALUE entry = rb_ary_entry(rb_namespaces, j);
+      c_namespaces[j] = (xmlChar *)StringValueCStr(entry);
     }
   }
 
+  xmlC14NExecute(c_doc, c_callback_wrapper, rb_callback,
+                 c_mode,
+                 c_namespaces,
+                 (int)RTEST(rb_comments_p),
+                 c_obuf);
 
-  xmlC14NExecute(doc, cb, ctx,
-    (int)      (NIL_P(mode)        ? 0 : NUM2INT(mode)),
-    ns,
-    (int)      RTEST(with_comments),
-    buf);
+  ruby_xfree(c_namespaces);
+  xmlOutputBufferClose(c_obuf);
 
-  xmlOutputBufferClose(buf);
+  return rb_funcall(rb_io, rb_intern("string"), 0);
+}
 
-  return rb_funcall(io, rb_intern("string"), 0);
+VALUE
+noko_xml_document_wrap_with_init_args(VALUE klass, xmlDocPtr c_document, int argc, VALUE *argv)
+{
+  VALUE rb_document;
+
+  if (!klass) {
+    klass = cNokogiriXmlDocument;
+  }
+
+  rb_document = _xml_document_alloc(klass);
+  _xml_document_data_ptr_set(rb_document, c_document);
+
+  rb_iv_set(rb_document, "@decorators", Qnil);
+  rb_iv_set(rb_document, "@errors", Qnil);
+
+  rb_obj_call_init(rb_document, argc, argv);
+
+  return rb_document ;
 }
 
-VALUE cNokogiriXmlDocument ;
-void init_xml_document()
+
+/* deprecated. use noko_xml_document_wrap() instead. */
+VALUE
+Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
 {
-  VALUE nokogiri  = rb_define_module("Nokogiri");
-  VALUE xml       = rb_define_module_under(nokogiri, "XML");
-  VALUE node      = rb_define_class_under(xml, "Node", rb_cObject);
+  /* TODO: deprecate this method in v2.0 */
+  return noko_xml_document_wrap_with_init_args(klass, doc, 0, NULL);
+}
 
-  /*
-   * Nokogiri::XML::Document wraps an xml document.
-   */
-  VALUE klass = rb_define_class_under(xml, "Document", node);
+VALUE
+noko_xml_document_wrap(VALUE klass, xmlDocPtr doc)
+{
+  return noko_xml_document_wrap_with_init_args(klass, doc, 0, NULL);
+}
+
+xmlDocPtr
+noko_xml_document_unwrap(VALUE rb_document)
+{
+  xmlDocPtr c_document;
+  TypedData_Get_Struct(rb_document, xmlDoc, &xml_doc_type, c_document);
+  return c_document;
+}
+
+/* Schema creation will remove and deallocate "blank" nodes.
+ * If those blank nodes have been exposed to Ruby, they could get freed
+ * out from under the VALUE pointer.  This function checks to see if any of
+ * those nodes have been exposed to Ruby, and if so we should raise an exception.
+ */
+int
+noko_xml_document_has_wrapped_blank_nodes_p(xmlDocPtr c_document)
+{
+  VALUE cache = DOC_NODE_CACHE(c_document);
+
+  if (NIL_P(cache)) {
+    return 0;
+  }
+
+  for (long jnode = 0; jnode < RARRAY_LEN(cache); jnode++) {
+    xmlNodePtr node;
+    VALUE element = rb_ary_entry(cache, jnode);
+
+    Noko_Node_Get_Struct(element, xmlNode, node);
+    if (xmlIsBlankNode(node)) {
+      return 1;
+    }
+  }
+
+  return 0;
+}
 
-  cNokogiriXmlDocument = klass;
+void
+noko_xml_document_pin_node(xmlNodePtr node)
+{
+  xmlDocPtr doc;
+  nokogiriTuplePtr tuple;
+
+  doc = node->doc;
+  tuple = (nokogiriTuplePtr)doc->_private;
+  st_insert(tuple->unlinkedNodes, (st_data_t)node, (st_data_t)node);
+}
 
-  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);
+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);
 }
 
 
-/* this takes klass as a param because it's used for HtmlDocument, too. */
-VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
+void
+noko_init_xml_document(void)
 {
-  nokogiriTuplePtr tuple = (nokogiriTuplePtr)malloc(sizeof(nokogiriTuple));
+  assert(cNokogiriXmlNode);
 
-  VALUE rb_doc = Data_Wrap_Struct(
-      klass ? klass : cNokogiriXmlDocument,
-      0,
-      dealloc,
-      doc
-  );
+  cNokogiriXmlDocument = rb_define_class_under(mNokogiriXml, "Document", cNokogiriXmlNode);
 
-  VALUE cache = rb_ary_new();
-  rb_iv_set(rb_doc, "@decorators", Qnil);
-  rb_iv_set(rb_doc, "@node_cache", cache);
+  rb_define_alloc_func(cNokogiriXmlDocument, _xml_document_alloc);
 
-  tuple->doc = rb_doc;
-  tuple->unlinkedNodes = st_init_numtable_with_size(128);
-  tuple->node_cache = cache;
-  doc->_private = tuple ;
+  rb_define_singleton_method(cNokogiriXmlDocument, "read_memory", noko_xml_document_s_read_memory, 4);
+  rb_define_singleton_method(cNokogiriXmlDocument, "read_io", noko_xml_document_s_read_io, 4);
+  rb_define_singleton_method(cNokogiriXmlDocument, "new", new, -1);
 
-  rb_obj_call_init(rb_doc, 0, NULL);
+  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, "url", url, 0);
+  rb_define_method(cNokogiriXmlDocument, "create_entity", noko_xml_document__create_entity, -1);
+  rb_define_method(cNokogiriXmlDocument, "remove_namespaces!", remove_namespaces_bang, 0);
 
-  return rb_doc ;
+  rb_define_protected_method(cNokogiriXmlDocument, "initialize_copy_with_args", rb_xml_document_initialize_copy_with_args,
+                             2);
 }