nokogiri 1.13.8 → 1.15.7

data/ext/nokogiri/nokogiri.h

@@ -1,6 +1,8 @@
 #ifndef NOKOGIRI_NATIVE
 #define NOKOGIRI_NATIVE
 
+#include <ruby/defines.h> // https://github.com/sparklemotion/nokogiri/issues/2696
+
 #ifdef _MSC_VER
 #  ifndef WIN32_LEAN_AND_MEAN
 #    define WIN32_LEAN_AND_MEAN
@@ -23,7 +25,6 @@
 #  define NOKOPUBVAR extern
 #endif
 
-
 #include <stdlib.h>
 #include <string.h>
 #include <assert.h>
@@ -50,6 +51,7 @@
 #include <libxslt/xsltconfig.h>
 #include <libxslt/xsltutils.h>
 #include <libxslt/transform.h>
+#include <libxslt/imports.h>
 #include <libxslt/xsltInternals.h>
 
 #include <libexslt/exslt.h>
@@ -75,22 +77,25 @@ xmlNodePtr xmlLastElementChild(xmlNodePtr parent);
 #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)
 
-#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)
+#ifndef NORETURN_DECL
+#  if defined(__GNUC__)
+#    define NORETURN_DECL __attribute__ ((noreturn))
+#  else
+#    define NORETURN_DECL
+#  endif
 #endif
 
-#ifndef NORETURN
+#ifndef PRINTFLIKE_DECL
 #  if defined(__GNUC__)
-#    define NORETURN(name) __attribute__((noreturn)) name
+#    define PRINTFLIKE_DECL(stringidx, argidx) __attribute__ ((format(printf,stringidx,argidx)))
 #  else
-#    define NORETURN(name) name
+#    define PRINTFLIKE_DECL(stringidx, argidx)
 #  endif
 #endif
 
+#if defined(TRUFFLERUBY) && !defined(NOKOGIRI_PACKAGED_LIBRARIES)
+#  define TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
+#endif
 
 NOKOPUBVAR VALUE mNokogiri ;
 NOKOPUBVAR VALUE mNokogiriGumbo ;
@@ -162,15 +167,16 @@ typedef struct _nokogiriXsltStylesheetTuple {
   VALUE func_instances;
 } nokogiriXsltStylesheetTuple;
 
-int vasprintf(char **strp, const char *fmt, va_list ap);
 void noko_xml_document_pin_node(xmlNodePtr);
 void noko_xml_document_pin_namespace(xmlNsPtr, xmlDocPtr);
+int noko_xml_document_has_wrapped_blank_nodes_p(xmlDocPtr c_document);
 
 int noko_io_read(void *ctx, char *buffer, int len);
 int noko_io_write(void *ctx, char *buffer, int len);
 int noko_io_close(void *ctx);
 
 #define Noko_Node_Get_Struct(obj,type,sval) ((sval) = (type*)DATA_PTR(obj))
+#define Noko_Namespace_Get_Struct(obj,type,sval) ((sval) = (type*)DATA_PTR(obj))
 
 VALUE noko_xml_node_wrap(VALUE klass, xmlNodePtr node) ;
 VALUE noko_xml_node_wrap_node_set_result(xmlNodePtr node, VALUE node_set) ;
@@ -182,12 +188,21 @@ VALUE noko_xml_namespace_wrap_xpath_copy(xmlNsPtr node);
 VALUE noko_xml_element_content_wrap(VALUE doc, xmlElementContentPtr element);
 
 VALUE noko_xml_node_set_wrap(xmlNodeSetPtr node_set, VALUE document) ;
+xmlNodeSetPtr noko_xml_node_set_unwrap(VALUE rb_node_set) ;
 
 VALUE noko_xml_document_wrap_with_init_args(VALUE klass, xmlDocPtr doc, int argc, VALUE *argv);
 VALUE noko_xml_document_wrap(VALUE klass, xmlDocPtr doc);
+xmlDocPtr noko_xml_document_unwrap(VALUE rb_document);
 NOKOPUBFUN VALUE Nokogiri_wrap_xml_document(VALUE klass,
     xmlDocPtr doc); /* deprecated. use noko_xml_document_wrap() instead. */
 
+xmlSAXHandlerPtr noko_sax_handler_unwrap(VALUE rb_sax_handler);
+
+xmlParserCtxtPtr noko_xml_sax_push_parser_unwrap(VALUE rb_parser);
+
+VALUE noko_xml_sax_parser_context_wrap(VALUE klass, xmlParserCtxtPtr c_context);
+xmlParserCtxtPtr noko_xml_sax_parser_context_unwrap(VALUE rb_context);
+
 #define DOC_RUBY_OBJECT_TEST(x) ((nokogiriTuplePtr)(x->_private))
 #define DOC_RUBY_OBJECT(x) (((nokogiriTuplePtr)(x->_private))->doc)
 #define DOC_UNLINKED_NODE_HASH(x) (((nokogiriTuplePtr)(x->_private))->unlinkedNodes)
@@ -197,15 +212,15 @@ NOKOPUBFUN VALUE Nokogiri_wrap_xml_document(VALUE klass,
 #define NOKOGIRI_SAX_SELF(_ctxt) ((nokogiriSAXTuplePtr)(_ctxt))->self
 #define NOKOGIRI_SAX_CTXT(_ctxt) ((nokogiriSAXTuplePtr)(_ctxt))->ctxt
 #define NOKOGIRI_SAX_TUPLE_NEW(_ctxt, _self) nokogiri_sax_tuple_new(_ctxt, _self)
-#define NOKOGIRI_SAX_TUPLE_DESTROY(_tuple) free(_tuple)
+#define NOKOGIRI_SAX_TUPLE_DESTROY(_tuple) ruby_xfree(_tuple)
 
 #define DISCARD_CONST_QUAL(t, v) ((t)(uintptr_t)(v))
 #define DISCARD_CONST_QUAL_XMLCHAR(v) DISCARD_CONST_QUAL(xmlChar *, v)
 
 #if HAVE_RB_CATEGORY_WARNING
-#  define NOKO_WARN_DEPRECATION(message) rb_category_warning(RB_WARN_CATEGORY_DEPRECATED, message)
+#  define NOKO_WARN_DEPRECATION(message...) rb_category_warning(RB_WARN_CATEGORY_DEPRECATED, message)
 #else
-#  define NOKO_WARN_DEPRECATION(message) rb_warning(message)
+#  define NOKO_WARN_DEPRECATION(message...) rb_warning(message)
 #endif
 
 void Nokogiri_structured_error_func_save(libxmlStructuredErrorHandlerState *handler_state);
@@ -214,7 +229,7 @@ void Nokogiri_structured_error_func_save_and_set(libxmlStructuredErrorHandlerSta
 void Nokogiri_structured_error_func_restore(libxmlStructuredErrorHandlerState *handler_state);
 VALUE Nokogiri_wrap_xml_syntax_error(xmlErrorPtr error);
 void Nokogiri_error_array_pusher(void *ctx, xmlErrorPtr error);
-NORETURN(void Nokogiri_error_raise(void *ctx, xmlErrorPtr error));
+NORETURN_DECL void Nokogiri_error_raise(void *ctx, xmlErrorPtr error);
 void Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr ctx, int nargs, VALUE handler,
     const char *function_name) ;
 
@@ -222,7 +237,7 @@ static inline
 nokogiriSAXTuplePtr
 nokogiri_sax_tuple_new(xmlParserCtxtPtr ctxt, VALUE self)
 {
-  nokogiriSAXTuplePtr tuple = malloc(sizeof(nokogiriSAXTuple));
+  nokogiriSAXTuplePtr tuple = ruby_xmalloc(sizeof(nokogiriSAXTuple));
   tuple->self = self;
   tuple->ctxt = ctxt;
   return tuple;

data/ext/nokogiri/test_global_handlers.c

@@ -32,7 +32,7 @@ rb_foreign_error_handler(VALUE klass)
  *  Do NOT use this outside of the Nokogiri test suite.
  */
 void
-noko_init_test_global_handlers()
+noko_init_test_global_handlers(void)
 {
   VALUE mNokogiriTest = rb_define_module_under(mNokogiri, "Test");
 

data/ext/nokogiri/xml_attr.c

@@ -68,7 +68,7 @@ new (int argc, VALUE *argv, VALUE klass)
     rb_raise(rb_eArgError, "parameter must be a Nokogiri::XML::Document");
   }
 
-  Noko_Node_Get_Struct(document, xmlDoc, xml_doc);
+  xml_doc = noko_xml_document_unwrap(document);
 
   node = xmlNewDocProp(
            xml_doc,
@@ -89,7 +89,7 @@ new (int argc, VALUE *argv, VALUE klass)
 }
 
 void
-noko_init_xml_attr()
+noko_init_xml_attr(void)
 {
   assert(cNokogiriXmlNode);
   /*

data/ext/nokogiri/xml_attribute_decl.c

@@ -13,7 +13,7 @@ attribute_type(VALUE self)
 {
   xmlAttributePtr node;
   Noko_Node_Get_Struct(self, xmlAttribute, node);
-  return INT2NUM((long)node->atype);
+  return INT2NUM(node->atype);
 }
 
 /*
@@ -59,7 +59,7 @@ enumeration(VALUE self)
 }
 
 void
-noko_init_xml_attribute_decl()
+noko_init_xml_attribute_decl(void)
 {
   assert(cNokogiriXmlNode);
   cNokogiriXmlAttributeDecl = rb_define_class_under(mNokogiriXml, "AttributeDecl", cNokogiriXmlNode);

data/ext/nokogiri/xml_cdata.c

@@ -12,31 +12,45 @@ VALUE cNokogiriXmlCData;
  * raise a TypeError exception.
  */
 static VALUE
-new (int argc, VALUE *argv, VALUE klass)
+rb_xml_cdata_s_new(int argc, VALUE *argv, VALUE klass)
 {
-  xmlDocPtr xml_doc;
-  xmlNodePtr node;
-  VALUE doc;
-  VALUE content;
-  VALUE rest;
+  xmlDocPtr c_document;
+  xmlNodePtr c_node;
+  VALUE rb_document;
+  VALUE rb_content;
+  VALUE rb_rest;
   VALUE rb_node;
-  xmlChar *content_str = NULL;
-  int content_str_len = 0;
+  xmlChar *c_content = NULL;
+  int c_content_len = 0;
 
-  rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
+  rb_scan_args(argc, argv, "2*", &rb_document, &rb_content, &rb_rest);
 
-  Noko_Node_Get_Struct(doc, xmlDoc, xml_doc);
+  if (!rb_obj_is_kind_of(rb_document, cNokogiriXmlNode)) {
+    rb_raise(rb_eTypeError,
+             "expected first parameter to be a Nokogiri::XML::Document, received %"PRIsVALUE,
+             rb_obj_class(rb_document));
+  }
+
+  if (!rb_obj_is_kind_of(rb_document, cNokogiriXmlDocument)) {
+    xmlNodePtr deprecated_node_type_arg;
+    // TODO: deprecate allowing Node
+    NOKO_WARN_DEPRECATION("Passing a Node as the first parameter to CDATA.new is deprecated. Please pass a Document instead. This will become an error in a future release of Nokogiri.");
+    Noko_Node_Get_Struct(rb_document, xmlNode, deprecated_node_type_arg);
+    c_document = deprecated_node_type_arg->doc;
+  } else {
+    c_document = noko_xml_document_unwrap(rb_document);
+  }
 
-  if (!NIL_P(content)) {
-    content_str = (xmlChar *)StringValuePtr(content);
-    content_str_len = RSTRING_LEN(content);
+  if (!NIL_P(rb_content)) {
+    c_content = (xmlChar *)StringValuePtr(rb_content);
+    c_content_len = RSTRING_LENINT(rb_content);
   }
 
-  node = xmlNewCDataBlock(xml_doc->doc, content_str, content_str_len);
+  c_node = xmlNewCDataBlock(c_document, c_content, c_content_len);
 
-  noko_xml_document_pin_node(node);
+  noko_xml_document_pin_node(c_node);
 
-  rb_node = noko_xml_node_wrap(klass, node);
+  rb_node = noko_xml_node_wrap(klass, c_node);
   rb_obj_call_init(rb_node, argc, argv);
 
   if (rb_block_given_p()) { rb_yield(rb_node); }
@@ -45,7 +59,7 @@ new (int argc, VALUE *argv, VALUE klass)
 }
 
 void
-noko_init_xml_cdata()
+noko_init_xml_cdata(void)
 {
   assert(cNokogiriXmlText);
   /*
@@ -53,5 +67,5 @@ noko_init_xml_cdata()
    */
   cNokogiriXmlCData = rb_define_class_under(mNokogiriXml, "CDATA", cNokogiriXmlText);
 
-  rb_define_singleton_method(cNokogiriXmlCData, "new", new, -1);
+  rb_define_singleton_method(cNokogiriXmlCData, "new", rb_xml_cdata_s_new, -1);
 }

data/ext/nokogiri/xml_comment.c

@@ -30,7 +30,7 @@ new (int argc, VALUE *argv, VALUE klass)
     rb_raise(rb_eArgError, "first argument must be a XML::Document or XML::Node");
   }
 
-  Data_Get_Struct(document, xmlDoc, xml_doc);
+  xml_doc = noko_xml_document_unwrap(document);
 
   node = xmlNewDocComment(
            xml_doc,
@@ -48,7 +48,7 @@ new (int argc, VALUE *argv, VALUE klass)
 }
 
 void
-noko_init_xml_comment()
+noko_init_xml_comment(void)
 {
   assert(cNokogiriXmlCharacterData);
   /*

data/ext/nokogiri/xml_document.c

@@ -51,8 +51,9 @@ remove_private(xmlNodePtr node)
 }
 
 static void
-mark(xmlDocPtr doc)
+mark(void *data)
 {
+  xmlDocPtr doc = (xmlDocPtr)data;
   nokogiriTuplePtr tuple = (nokogiriTuplePtr)doc->_private;
   if (tuple) {
     rb_gc_mark(tuple->doc);
@@ -61,33 +62,83 @@ mark(xmlDocPtr doc)
 }
 
 static void
-dealloc(xmlDocPtr doc)
+dealloc(void *data)
 {
+  xmlDocPtr doc = (xmlDocPtr)data;
   st_table *node_hash;
 
-  NOKOGIRI_DEBUG_START(doc);
-
   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.
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wdeprecated-declarations" // xmlDeregisterNodeDefault is deprecated as of libxml2 2.11.0
+  /*
+   * 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) {
     remove_private((xmlNodePtr)doc);
   }
+#pragma GCC diagnostic pop
 
   xmlFreeDoc(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 += 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 += xmlStrlen(node->content);
+  }
+  for (child = node->children; child; child = child->next) {
+    memsize += sizeof(xmlNode) + memsize_node(child);
+  }
+  return memsize;
+}
 
-  NOKOGIRI_DEBUG_END(doc);
+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 const rb_data_type_t noko_xml_document_data_type = {
+  .wrap_struct_name = "Nokogiri::XML::Document",
+  .function = {
+    .dmark = mark,
+    .dfree = dealloc,
+    .dsize = memsize,
+  },
+  // .flags = RUBY_TYPED_FREE_IMMEDIATELY, // TODO see https://github.com/sparklemotion/nokogiri/issues/2822
+};
+
 static void
 recursively_remove_namespaces_from_node(xmlNodePtr node)
 {
@@ -104,7 +155,11 @@ recursively_remove_namespaces_from_node(xmlNodePtr 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;
   }
 
@@ -126,8 +181,7 @@ recursively_remove_namespaces_from_node(xmlNodePtr node)
 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); }
 
@@ -146,7 +200,7 @@ rb_xml_document_root_set(VALUE self, VALUE rb_new_root)
   xmlDocPtr c_document;
   xmlNodePtr c_new_root = NULL, c_current_root;
 
-  Data_Get_Struct(self, xmlDoc, c_document);
+  c_document = noko_xml_document_unwrap(self);
 
   c_current_root = xmlDocGetRootElement(c_document);
   if (c_current_root) {
@@ -190,7 +244,7 @@ rb_xml_document_root(VALUE self)
   xmlDocPtr c_document;
   xmlNodePtr c_root;
 
-  Data_Get_Struct(self, xmlDoc, c_document);
+  c_document = noko_xml_document_unwrap(self);
 
   c_root = xmlDocGetRootElement(c_document);
   if (!c_root) {
@@ -209,8 +263,7 @@ rb_xml_document_root(VALUE self)
 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) {
     xmlFree(DISCARD_CONST_QUAL_XMLCHAR(doc->encoding));
@@ -230,8 +283,7 @@ set_encoding(VALUE self, VALUE encoding)
 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; }
   return NOKOGIRI_STR_NEW2(doc->encoding);
@@ -246,8 +298,7 @@ encoding(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; }
   return NOKOGIRI_STR_NEW2(doc->version);
@@ -369,7 +420,7 @@ duplicate_document(int argc, VALUE *argv, VALUE self)
     level = INT2NUM((long)1);
   }
 
-  Data_Get_Struct(self, xmlDoc, doc);
+  doc = noko_xml_document_unwrap(self);
 
   dup = xmlCopyDoc(doc, (int)NUM2INT(level));
 
@@ -442,8 +493,7 @@ new (int argc, VALUE *argv, VALUE klass)
 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;
@@ -471,7 +521,7 @@ create_entity(int argc, VALUE *argv, VALUE self)
   xmlEntityPtr ptr;
   xmlDocPtr doc ;
 
-  Data_Get_Struct(self, xmlDoc, doc);
+  doc = noko_xml_document_unwrap(self);
 
   rb_scan_args(argc, argv, "14", &name, &type, &external_id, &system_id,
                &content);
@@ -536,6 +586,7 @@ rb_xml_document_canonicalize(int argc, VALUE *argv, VALUE self)
   VALUE rb_mode;
   VALUE rb_namespaces;
   VALUE rb_comments_p;
+  int c_mode = 0;
   xmlChar **c_namespaces;
 
   xmlDocPtr c_doc;
@@ -547,10 +598,18 @@ rb_xml_document_canonicalize(int argc, VALUE *argv, VALUE self)
   VALUE rb_io;
 
   rb_scan_args(argc, argv, "03", &rb_mode, &rb_namespaces, &rb_comments_p);
-  if (!NIL_P(rb_mode)) { Check_Type(rb_mode, T_FIXNUM); }
-  if (!NIL_P(rb_namespaces)) { Check_Type(rb_namespaces, T_ARRAY); }
+  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, c_doc);
+  c_doc = noko_xml_document_unwrap(self);
 
   rb_cStringIO = rb_const_get_at(rb_cObject, rb_intern("StringIO"));
   rb_io = rb_class_new_instance(0, 0, rb_cStringIO);
@@ -569,7 +628,7 @@ rb_xml_document_canonicalize(int argc, VALUE *argv, VALUE self)
     c_namespaces = NULL;
   } else {
     long ns_len = RARRAY_LEN(rb_namespaces);
-    c_namespaces = calloc((size_t)ns_len + 1, sizeof(xmlChar *));
+    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);
@@ -577,12 +636,12 @@ rb_xml_document_canonicalize(int argc, VALUE *argv, VALUE self)
   }
 
   xmlC14NExecute(c_doc, c_callback_wrapper, rb_callback,
-                 (int)(NIL_P(rb_mode) ? 0 : NUM2INT(rb_mode)),
+                 c_mode,
                  c_namespaces,
                  (int)RTEST(rb_comments_p),
                  c_obuf);
 
-  free(c_namespaces);
+  ruby_xfree(c_namespaces);
   xmlOutputBufferClose(c_obuf);
 
   return rb_funcall(rb_io, rb_intern("string"), 0);
@@ -598,9 +657,9 @@ noko_xml_document_wrap_with_init_args(VALUE klass, xmlDocPtr c_document, int arg
     klass = cNokogiriXmlDocument;
   }
 
-  rb_document = Data_Wrap_Struct(klass, mark, dealloc, c_document);
+  rb_document = TypedData_Wrap_Struct(klass, &noko_xml_document_data_type, c_document);
 
-  tuple = (nokogiriTuplePtr)malloc(sizeof(nokogiriTuple));
+  tuple = (nokogiriTuplePtr)ruby_xmalloc(sizeof(nokogiriTuple));
   tuple->doc = rb_document;
   tuple->unlinkedNodes = st_init_numtable_with_size(128);
   tuple->node_cache = rb_ary_new();
@@ -631,6 +690,40 @@ 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, &noko_xml_document_data_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;
+}
 
 void
 noko_xml_document_pin_node(xmlNodePtr node)
@@ -655,7 +748,7 @@ noko_xml_document_pin_namespace(xmlNsPtr ns, xmlDocPtr doc)
 
 
 void
-noko_init_xml_document()
+noko_init_xml_document(void)
 {
   assert(cNokogiriXmlNode);
   /*

data/ext/nokogiri/xml_document_fragment.c

@@ -19,7 +19,7 @@ new (int argc, VALUE *argv, VALUE klass)
 
   rb_scan_args(argc, argv, "1*", &document, &rest);
 
-  Data_Get_Struct(document, xmlDoc, xml_doc);
+  xml_doc = noko_xml_document_unwrap(document);
 
   node = xmlNewDocFragment(xml_doc->doc);
 
@@ -32,7 +32,7 @@ new (int argc, VALUE *argv, VALUE klass)
 }
 
 void
-noko_init_xml_document_fragment()
+noko_init_xml_document_fragment(void)
 {
   assert(cNokogiriXmlNode);
   /*

data/ext/nokogiri/xml_dtd.c

@@ -139,7 +139,7 @@ validate(VALUE self, VALUE document)
   VALUE error_list;
 
   Noko_Node_Get_Struct(self, xmlDtd, dtd);
-  Noko_Node_Get_Struct(document, xmlDoc, doc);
+  doc = noko_xml_document_unwrap(document);
   error_list = rb_ary_new();
 
   ctxt = xmlNewValidCtxt();
@@ -190,7 +190,7 @@ external_id(VALUE self)
 }
 
 void
-noko_init_xml_dtd()
+noko_init_xml_dtd(void)
 {
   assert(cNokogiriXmlNode);
   /*