nokogiri 1.10.9 → 1.18.3

data/ext/nokogiri/xml_reader.c

@@ -1,13 +1,30 @@
-#include <xml_reader.h>
+#include <nokogiri.h>
 
-static void dealloc(xmlTextReaderPtr reader)
+VALUE cNokogiriXmlReader;
+
+static void
+xml_reader_deallocate(void *data)
 {
-  NOKOGIRI_DEBUG_START(reader);
+  // free the document separately because we _may_ have triggered preservation by calling
+  // xmlTextReaderCurrentDoc during a read_more.
+  xmlTextReaderPtr reader = data;
+  xmlDocPtr doc = xmlTextReaderCurrentDoc(reader);
   xmlFreeTextReader(reader);
-  NOKOGIRI_DEBUG_END(reader);
+  if (doc) {
+    xmlFreeDoc(doc);
+  }
 }
 
-static int has_attributes(xmlTextReaderPtr reader)
+static const rb_data_type_t xml_text_reader_type = {
+  .wrap_struct_name = "xmlTextReader",
+  .function = {
+    .dfree = xml_reader_deallocate,
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
+static int
+has_attributes(xmlTextReaderPtr reader)
 {
   /*
    *  this implementation of xmlTextReaderHasAttributes explicitly includes
@@ -16,47 +33,40 @@ static int has_attributes(xmlTextReaderPtr reader)
    */
   xmlNodePtr node ;
   node = xmlTextReaderCurrentNode(reader);
-  if (node == NULL)
-    return(0);
+  if (node == NULL) {
+    return (0);
+  }
 
   if ((node->type == XML_ELEMENT_NODE) &&
-      ((node->properties != NULL) || (node->nsDef != NULL)))
-    return(1);
-  return(0);
+      ((node->properties != NULL) || (node->nsDef != NULL))) {
+    return (1);
+  }
+  return (0);
 }
 
-static void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash)
+// TODO: merge this function into the `namespaces` method implementation
+static void
+Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash)
 {
   xmlNsPtr ns;
-  static char buffer[XMLNS_BUFFER_LEN] ;
-  char *key ;
-  size_t keylen ;
+  VALUE key;
 
-  if (node->type != XML_ELEMENT_NODE) return ;
+  if (node->type != XML_ELEMENT_NODE) { return ; }
 
   ns = node->nsDef;
   while (ns != NULL) {
 
-    keylen = XMLNS_PREFIX_LEN + (ns->prefix ? (strlen((const char*)ns->prefix) + 1) : 0) ;
-    if (keylen > XMLNS_BUFFER_LEN) {
-      key = (char*)malloc(keylen) ;
-    } else {
-      key = buffer ;
-    }
-
+    key = rb_enc_str_new_cstr(XMLNS_PREFIX, rb_utf8_encoding());
     if (ns->prefix) {
-      sprintf(key, "%s:%s", XMLNS_PREFIX, ns->prefix);
-    } else {
-      sprintf(key, "%s", XMLNS_PREFIX);
+      rb_str_cat_cstr(key, ":");
+      rb_str_cat_cstr(key, (const char *)ns->prefix);
     }
 
+    key = rb_str_conv_enc(key, rb_utf8_encoding(), rb_default_internal_encoding());
     rb_hash_aset(attr_hash,
-        NOKOGIRI_STR_NEW2(key),
-        (ns->href ? NOKOGIRI_STR_NEW2(ns->href) : Qnil)
-    );
-    if (key != buffer) {
-      free(key);
-    }
+                 key,
+                 (ns->href ? NOKOGIRI_STR_NEW2(ns->href) : Qnil)
+                );
     ns = ns->next ;
   }
 }
@@ -68,15 +78,16 @@ static void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash)
  *
  * Was an attribute generated from the default value in the DTD or schema?
  */
-static VALUE default_eh(VALUE self)
+static VALUE
+default_eh(VALUE self)
 {
   xmlTextReaderPtr reader;
   int eh;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   eh = xmlTextReaderIsDefault(reader);
-  if(eh == 0) return Qfalse;
-  if(eh == 1) return Qtrue;
+  if (eh == 0) { return Qfalse; }
+  if (eh == 1) { return Qtrue; }
 
   return Qnil;
 }
@@ -87,15 +98,16 @@ static VALUE default_eh(VALUE self)
  *
  * Does this node have a text value?
  */
-static VALUE value_eh(VALUE self)
+static VALUE
+value_eh(VALUE self)
 {
   xmlTextReaderPtr reader;
   int eh;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   eh = xmlTextReaderHasValue(reader);
-  if(eh == 0) return Qfalse;
-  if(eh == 1) return Qtrue;
+  if (eh == 0) { return Qfalse; }
+  if (eh == 1) { return Qtrue; }
 
   return Qnil;
 }
@@ -106,15 +118,16 @@ static VALUE value_eh(VALUE self)
  *
  * Does this node have attributes?
  */
-static VALUE attributes_eh(VALUE self)
+static VALUE
+attributes_eh(VALUE self)
 {
   xmlTextReaderPtr reader;
   int eh;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   eh = has_attributes(reader);
-  if(eh == 0) return Qfalse;
-  if(eh == 1) return Qtrue;
+  if (eh == 0) { return Qfalse; }
+  if (eh == 1) { return Qtrue; }
 
   return Qnil;
 }
@@ -125,52 +138,94 @@ static VALUE attributes_eh(VALUE self)
  *
  * Get a hash of namespaces for this Node
  */
-static VALUE namespaces(VALUE self)
+static VALUE
+rb_xml_reader_namespaces(VALUE rb_reader)
 {
-  xmlTextReaderPtr reader;
-  xmlNodePtr ptr;
-  VALUE attr ;
+  VALUE rb_namespaces = rb_hash_new() ;
+  xmlTextReaderPtr c_reader;
+  xmlNodePtr c_node;
+  VALUE rb_errors;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
 
-  attr = rb_hash_new() ;
+  if (! has_attributes(c_reader)) {
+    return rb_namespaces ;
+  }
 
-  if (! has_attributes(reader))
-    return attr ;
+  rb_errors = rb_funcall(rb_reader, rb_intern("errors"), 0);
 
-  ptr = xmlTextReaderExpand(reader);
-  if(ptr == NULL) return Qnil;
+  xmlSetStructuredErrorFunc((void *)rb_errors, noko__error_array_pusher);
+  c_node = xmlTextReaderExpand(c_reader);
+  xmlSetStructuredErrorFunc(NULL, NULL);
 
-  Nokogiri_xml_node_namespaces(ptr, attr);
+  if (c_node == NULL) {
+    if (RARRAY_LEN(rb_errors) > 0) {
+      VALUE rb_error = rb_ary_entry(rb_errors, 0);
+      VALUE exception_message = rb_funcall(rb_error, rb_intern("to_s"), 0);
+      rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
+    }
+    return Qnil;
+  }
+
+  Nokogiri_xml_node_namespaces(c_node, rb_namespaces);
 
-  return attr ;
+  return rb_namespaces ;
 }
 
 /*
- * call-seq:
- *   attribute_nodes
- *
- * Get a list of attributes for this Node
+  :call-seq: attribute_hash() → Hash<String ⇒ String>
+
+  Get the attributes of the current node as a Hash of names and values.
+
+  See related: #attributes and #namespaces
  */
-static VALUE attribute_nodes(VALUE self)
+static VALUE
+rb_xml_reader_attribute_hash(VALUE rb_reader)
 {
-  xmlTextReaderPtr reader;
-  xmlNodePtr ptr;
-  VALUE attr ;
+  VALUE rb_attributes = rb_hash_new();
+  xmlTextReaderPtr c_reader;
+  xmlNodePtr c_node;
+  xmlAttrPtr c_property;
+  VALUE rb_errors;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
 
-  attr = rb_ary_new() ;
+  if (!has_attributes(c_reader)) {
+    return rb_attributes;
+  }
 
-  if (! has_attributes(reader))
-    return attr ;
+  rb_errors = rb_funcall(rb_reader, rb_intern("errors"), 0);
 
-  ptr = xmlTextReaderExpand(reader);
-  if(ptr == NULL) return Qnil;
+  xmlSetStructuredErrorFunc((void *)rb_errors, noko__error_array_pusher);
+  c_node = xmlTextReaderExpand(c_reader);
+  xmlSetStructuredErrorFunc(NULL, NULL);
 
-  Nokogiri_xml_node_properties(ptr, attr);
+  if (c_node == NULL) {
+    if (RARRAY_LEN(rb_errors) > 0) {
+      VALUE rb_error = rb_ary_entry(rb_errors, 0);
+      VALUE exception_message = rb_funcall(rb_error, rb_intern("to_s"), 0);
+      rb_exc_raise(rb_class_new_instance(1, &exception_message, cNokogiriXmlSyntaxError));
+    }
+    return Qnil;
+  }
+
+  c_property = c_node->properties;
+  while (c_property != NULL) {
+    VALUE rb_name = NOKOGIRI_STR_NEW2(c_property->name);
+    VALUE rb_value = Qnil;
+    xmlChar *c_value = xmlNodeGetContent((xmlNode *)c_property);
+
+    if (c_value) {
+      rb_value = NOKOGIRI_STR_NEW2(c_value);
+      xmlFree(c_value);
+    }
 
-  return attr ;
+    rb_hash_aset(rb_attributes, rb_name, rb_value);
+
+    c_property = c_property->next;
+  }
+
+  return rb_attributes;
 }
 
 /*
@@ -179,22 +234,23 @@ static VALUE attribute_nodes(VALUE self)
  *
  * Get the value of attribute at +index+
  */
-static VALUE attribute_at(VALUE self, VALUE index)
+static VALUE
+attribute_at(VALUE self, VALUE index)
 {
   xmlTextReaderPtr reader;
   xmlChar *value;
   VALUE rb_value;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
-  if(NIL_P(index)) return Qnil;
+  if (NIL_P(index)) { return Qnil; }
   index = rb_Integer(index);
 
   value = xmlTextReaderGetAttributeNo(
-      reader,
-      (int)NUM2INT(index)
-  );
-  if(value == NULL) return Qnil;
+            reader,
+            (int)NUM2INT(index)
+          );
+  if (value == NULL) { return Qnil; }
 
   rb_value = NOKOGIRI_STR_NEW2(value);
   xmlFree(value);
@@ -207,19 +263,20 @@ static VALUE attribute_at(VALUE self, VALUE index)
  *
  * Get the value of attribute named +name+
  */
-static VALUE reader_attribute(VALUE self, VALUE name)
+static VALUE
+reader_attribute(VALUE self, VALUE name)
 {
   xmlTextReaderPtr reader;
   xmlChar *value ;
   VALUE rb_value;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
-  if(NIL_P(name)) return Qnil;
+  if (NIL_P(name)) { return Qnil; }
   name = StringValue(name) ;
 
-  value = xmlTextReaderGetAttribute(reader, (xmlChar*)StringValueCStr(name));
-  if(value == NULL) return Qnil;
+  value = xmlTextReaderGetAttribute(reader, (xmlChar *)StringValueCStr(name));
+  if (value == NULL) { return Qnil; }
 
   rb_value = NOKOGIRI_STR_NEW2(value);
   xmlFree(value);
@@ -232,16 +289,17 @@ static VALUE reader_attribute(VALUE self, VALUE name)
  *
  * Get the number of attributes for the current node
  */
-static VALUE attribute_count(VALUE self)
+static VALUE
+attribute_count(VALUE self)
 {
   xmlTextReaderPtr reader;
   int count;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   count = xmlTextReaderAttributeCount(reader);
-  if(count == -1) return Qnil;
+  if (count == -1) { return Qnil; }
 
-  return INT2NUM((long)count);
+  return INT2NUM(count);
 }
 
 /*
@@ -250,16 +308,17 @@ static VALUE attribute_count(VALUE self)
  *
  * Get the depth of the node
  */
-static VALUE depth(VALUE self)
+static VALUE
+depth(VALUE self)
 {
   xmlTextReaderPtr reader;
   int depth;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   depth = xmlTextReaderDepth(reader);
-  if(depth == -1) return Qnil;
+  if (depth == -1) { return Qnil; }
 
-  return INT2NUM((long)depth);
+  return INT2NUM(depth);
 }
 
 /*
@@ -268,14 +327,15 @@ static VALUE depth(VALUE self)
  *
  * Get the XML version of the document being read
  */
-static VALUE xml_version(VALUE self)
+static VALUE
+xml_version(VALUE self)
 {
   xmlTextReaderPtr reader;
   const char *version;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   version = (const char *)xmlTextReaderConstXmlVersion(reader);
-  if(version == NULL) return Qnil;
+  if (version == NULL) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(version);
 }
@@ -286,14 +346,15 @@ static VALUE xml_version(VALUE self)
  *
  * Get the xml:lang scope within which the node resides.
  */
-static VALUE lang(VALUE self)
+static VALUE
+lang(VALUE self)
 {
   xmlTextReaderPtr reader;
   const char *lang;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   lang = (const char *)xmlTextReaderConstXmlLang(reader);
-  if(lang == NULL) return Qnil;
+  if (lang == NULL) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(lang);
 }
@@ -304,14 +365,15 @@ static VALUE lang(VALUE self)
  *
  * Get the text value of the node if present. Returns a utf-8 encoded string.
  */
-static VALUE value(VALUE self)
+static VALUE
+value(VALUE self)
 {
   xmlTextReaderPtr reader;
   const char *value;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   value = (const char *)xmlTextReaderConstValue(reader);
-  if(value == NULL) return Qnil;
+  if (value == NULL) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(value);
 }
@@ -322,14 +384,15 @@ static VALUE value(VALUE self)
  *
  * Get the shorthand reference to the namespace associated with the node.
  */
-static VALUE prefix(VALUE self)
+static VALUE
+prefix(VALUE self)
 {
   xmlTextReaderPtr reader;
   const char *prefix;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   prefix = (const char *)xmlTextReaderConstPrefix(reader);
-  if(prefix == NULL) return Qnil;
+  if (prefix == NULL) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(prefix);
 }
@@ -340,14 +403,15 @@ static VALUE prefix(VALUE self)
  *
  * Get the URI defining the namespace associated with the node
  */
-static VALUE namespace_uri(VALUE self)
+static VALUE
+namespace_uri(VALUE self)
 {
   xmlTextReaderPtr reader;
   const char *uri;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   uri = (const char *)xmlTextReaderConstNamespaceUri(reader);
-  if(uri == NULL) return Qnil;
+  if (uri == NULL) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(uri);
 }
@@ -358,14 +422,15 @@ static VALUE namespace_uri(VALUE self)
  *
  * Get the local name of the node
  */
-static VALUE local_name(VALUE self)
+static VALUE
+local_name(VALUE self)
 {
   xmlTextReaderPtr reader;
   const char *name;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   name = (const char *)xmlTextReaderConstLocalName(reader);
-  if(name == NULL) return Qnil;
+  if (name == NULL) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(name);
 }
@@ -376,14 +441,15 @@ static VALUE local_name(VALUE self)
  *
  * Get the name of the node. Returns a utf-8 encoded string.
  */
-static VALUE name(VALUE self)
+static VALUE
+name(VALUE self)
 {
   xmlTextReaderPtr reader;
   const char *name;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
   name = (const char *)xmlTextReaderConstName(reader);
-  if(name == NULL) return Qnil;
+  if (name == NULL) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(name);
 }
@@ -394,16 +460,24 @@ static VALUE name(VALUE self)
  *
  * Get the xml:base of the node
  */
-static VALUE base_uri(VALUE self)
+static VALUE
+rb_xml_reader_base_uri(VALUE rb_reader)
 {
-  xmlTextReaderPtr reader;
-  const char * base_uri;
+  VALUE rb_base_uri;
+  xmlTextReaderPtr c_reader;
+  xmlChar *c_base_uri;
+
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
+
+  c_base_uri = xmlTextReaderBaseUri(c_reader);
+  if (c_base_uri == NULL) {
+    return Qnil;
+  }
 
-  Data_Get_Struct(self, xmlTextReader, reader);
-  base_uri = (const char *)xmlTextReaderBaseUri(reader);
-  if (base_uri == NULL) return Qnil;
+  rb_base_uri = NOKOGIRI_STR_NEW2(c_base_uri);
+  xmlFree(c_base_uri);
 
-  return NOKOGIRI_STR_NEW2(base_uri);
+  return rb_base_uri;
 }
 
 /*
@@ -412,11 +486,12 @@ static VALUE base_uri(VALUE self)
  *
  * Get the state of the reader
  */
-static VALUE state(VALUE self)
+static VALUE
+state(VALUE self)
 {
   xmlTextReaderPtr reader;
-  Data_Get_Struct(self, xmlTextReader, reader);
-  return INT2NUM((long)xmlTextReaderReadState(reader));
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
+  return INT2NUM(xmlTextReaderReadState(reader));
 }
 
 /*
@@ -425,11 +500,12 @@ static VALUE state(VALUE self)
  *
  * Get the type of readers current node
  */
-static VALUE node_type(VALUE self)
+static VALUE
+node_type(VALUE self)
 {
   xmlTextReaderPtr reader;
-  Data_Get_Struct(self, xmlTextReader, reader);
-  return INT2NUM((long)xmlTextReaderNodeType(reader));
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
+  return INT2NUM(xmlTextReaderNodeType(reader));
 }
 
 /*
@@ -438,31 +514,42 @@ static VALUE node_type(VALUE self)
  *
  * Move the Reader forward through the XML document.
  */
-static VALUE read_more(VALUE self)
+static VALUE
+read_more(VALUE rb_reader)
 {
-  xmlTextReaderPtr reader;
-  xmlErrorPtr error;
-  VALUE error_list;
-  int ret;
+  xmlTextReaderPtr c_reader;
+  libxmlStructuredErrorHandlerState handler_state;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
 
-  error_list = rb_funcall(self, rb_intern("errors"), 0);
+  VALUE rb_errors = rb_funcall(rb_reader, rb_intern("errors"), 0);
+  noko__structured_error_func_save_and_set(&handler_state, (void *)rb_errors, noko__error_array_pusher);
 
-  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
-  ret = xmlTextReaderRead(reader);
-  xmlSetStructuredErrorFunc(NULL, NULL);
+  int status = xmlTextReaderRead(c_reader);
 
-  if(ret == 1) return self;
-  if(ret == 0) return Qnil;
+  noko__structured_error_func_restore(&handler_state);
 
-  error = xmlGetLastError();
-  if(error)
-    rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
-  else
-    rb_raise(rb_eRuntimeError, "Error pulling: %d", ret);
+  xmlDocPtr c_document = xmlTextReaderCurrentDoc(c_reader);
+  if (c_document && c_document->encoding == NULL) {
+    VALUE constructor_encoding = rb_iv_get(rb_reader, "@encoding");
+    if (RTEST(constructor_encoding)) {
+      c_document->encoding = xmlStrdup(BAD_CAST StringValueCStr(constructor_encoding));
+    } else {
+      rb_iv_set(rb_reader, "@encoding", NOKOGIRI_STR_NEW2("UTF-8"));
+      c_document->encoding = xmlStrdup(BAD_CAST "UTF-8");
+    }
+  }
 
-  return Qnil;
+  if (status == 1) { return rb_reader; }
+  if (status == 0) { return Qnil; }
+
+  /* if we're here, there was an error */
+  VALUE exception = rb_funcall(cNokogiriXmlSyntaxError, rb_intern("aggregate"), 1, rb_errors);
+  if (RB_TEST(exception)) {
+    rb_exc_raise(exception);
+  } else {
+    rb_raise(rb_eRuntimeError, "Error pulling: %d", status);
+  }
 }
 
 /*
@@ -472,19 +559,20 @@ static VALUE read_more(VALUE self)
  * Read the contents of the current node, including child nodes and markup.
  * Returns a utf-8 encoded string.
  */
-static VALUE inner_xml(VALUE self)
+static VALUE
+inner_xml(VALUE self)
 {
   xmlTextReaderPtr reader;
-  xmlChar* value;
+  xmlChar *value;
   VALUE str;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
   value = xmlTextReaderReadInnerXml(reader);
 
   str = Qnil;
-  if(value) {
-    str = NOKOGIRI_STR_NEW2((char*)value);
+  if (value) {
+    str = NOKOGIRI_STR_NEW2((char *)value);
     xmlFree(value);
   }
 
@@ -498,18 +586,19 @@ static VALUE inner_xml(VALUE self)
  * Read the current node and its contents, including child nodes and markup.
  * Returns a utf-8 encoded string.
  */
-static VALUE outer_xml(VALUE self)
+static VALUE
+outer_xml(VALUE self)
 {
   xmlTextReaderPtr reader;
   xmlChar *value;
   VALUE str = Qnil;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
   value = xmlTextReaderReadOuterXml(reader);
 
-  if(value) {
-    str = NOKOGIRI_STR_NEW2((char*)value);
+  if (value) {
+    str = NOKOGIRI_STR_NEW2((char *)value);
     xmlFree(value);
   }
   return str;
@@ -519,38 +608,41 @@ static VALUE outer_xml(VALUE self)
  * call-seq:
  *   from_memory(string, url = nil, encoding = nil, options = 0)
  *
- * Create a new reader that parses +string+
+ * Create a new Reader to parse a String.
  */
-static VALUE from_memory(int argc, VALUE *argv, VALUE klass)
+static VALUE
+from_memory(int argc, VALUE *argv, VALUE klass)
 {
+  /* TODO: deprecate this method, since Reader.new can handle both memory and IO. It can then
+   * become private. */
   VALUE rb_buffer, rb_url, encoding, rb_options;
   xmlTextReaderPtr reader;
-  const char * c_url      = NULL;
-  const char * c_encoding = NULL;
+  const char *c_url      = NULL;
+  const char *c_encoding = NULL;
   int c_options           = 0;
   VALUE rb_reader, args[3];
 
   rb_scan_args(argc, argv, "13", &rb_buffer, &rb_url, &encoding, &rb_options);
 
-  if (!RTEST(rb_buffer)) rb_raise(rb_eArgError, "string cannot be nil");
-  if (RTEST(rb_url)) c_url = StringValueCStr(rb_url);
-  if (RTEST(encoding)) c_encoding = StringValueCStr(encoding);
-  if (RTEST(rb_options)) c_options = (int)NUM2INT(rb_options);
+  if (!RTEST(rb_buffer)) { rb_raise(rb_eArgError, "string cannot be nil"); }
+  if (RTEST(rb_url)) { c_url = StringValueCStr(rb_url); }
+  if (RTEST(encoding)) { c_encoding = StringValueCStr(encoding); }
+  if (RTEST(rb_options)) { c_options = (int)NUM2INT(rb_options); }
 
   reader = xmlReaderForMemory(
-      StringValuePtr(rb_buffer),
-      (int)RSTRING_LEN(rb_buffer),
-      c_url,
-      c_encoding,
-      c_options
-  );
-
-  if(reader == NULL) {
+             StringValuePtr(rb_buffer),
+             (int)RSTRING_LEN(rb_buffer),
+             c_url,
+             c_encoding,
+             c_options
+           );
+
+  if (reader == NULL) {
     xmlFreeTextReader(reader);
     rb_raise(rb_eRuntimeError, "couldn't create a parser");
   }
 
-  rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
+  rb_reader = TypedData_Wrap_Struct(klass, &xml_text_reader_type, reader);
   args[0] = rb_buffer;
   args[1] = rb_url;
   args[2] = encoding;
@@ -563,39 +655,42 @@ static VALUE from_memory(int argc, VALUE *argv, VALUE klass)
  * call-seq:
  *   from_io(io, url = nil, encoding = nil, options = 0)
  *
- * Create a new reader that parses +io+
+ * Create a new Reader to parse an IO stream.
  */
-static VALUE from_io(int argc, VALUE *argv, VALUE klass)
+static VALUE
+from_io(int argc, VALUE *argv, VALUE klass)
 {
+  /* TODO: deprecate this method, since Reader.new can handle both memory and IO. It can then
+   * become private. */
   VALUE rb_io, rb_url, encoding, rb_options;
   xmlTextReaderPtr reader;
-  const char * c_url      = NULL;
-  const char * c_encoding = NULL;
+  const char *c_url      = NULL;
+  const char *c_encoding = NULL;
   int c_options           = 0;
   VALUE rb_reader, args[3];
 
   rb_scan_args(argc, argv, "13", &rb_io, &rb_url, &encoding, &rb_options);
 
-  if (!RTEST(rb_io)) rb_raise(rb_eArgError, "io cannot be nil");
-  if (RTEST(rb_url)) c_url = StringValueCStr(rb_url);
-  if (RTEST(encoding)) c_encoding = StringValueCStr(encoding);
-  if (RTEST(rb_options)) c_options = (int)NUM2INT(rb_options);
+  if (!RTEST(rb_io)) { rb_raise(rb_eArgError, "io cannot be nil"); }
+  if (RTEST(rb_url)) { c_url = StringValueCStr(rb_url); }
+  if (RTEST(encoding)) { c_encoding = StringValueCStr(encoding); }
+  if (RTEST(rb_options)) { c_options = (int)NUM2INT(rb_options); }
 
   reader = xmlReaderForIO(
-      (xmlInputReadCallback)io_read_callback,
-      (xmlInputCloseCallback)io_close_callback,
-      (void *)rb_io,
-      c_url,
-      c_encoding,
-      c_options
-  );
-
-  if(reader == NULL) {
+             (xmlInputReadCallback)noko_io_read,
+             (xmlInputCloseCallback)noko_io_close,
+             (void *)rb_io,
+             c_url,
+             c_encoding,
+             c_options
+           );
+
+  if (reader == NULL) {
     xmlFreeTextReader(reader);
     rb_raise(rb_eRuntimeError, "couldn't create a parser");
   }
 
-  rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
+  rb_reader = TypedData_Wrap_Struct(klass, &xml_text_reader_type, reader);
   args[0] = rb_io;
   args[1] = rb_url;
   args[2] = encoding;
@@ -610,59 +705,73 @@ static VALUE from_io(int argc, VALUE *argv, VALUE klass)
  *
  * Returns true if the current node is empty, otherwise false.
  */
-static VALUE empty_element_p(VALUE self)
+static VALUE
+empty_element_p(VALUE self)
 {
   xmlTextReaderPtr reader;
 
-  Data_Get_Struct(self, xmlTextReader, reader);
+  TypedData_Get_Struct(self, xmlTextReader, &xml_text_reader_type, reader);
 
-  if(xmlTextReaderIsEmptyElement(reader))
+  if (xmlTextReaderIsEmptyElement(reader)) {
     return Qtrue;
+  }
 
   return Qfalse;
 }
 
-VALUE cNokogiriXmlReader;
-
-void init_xml_reader()
+static VALUE
+rb_xml_reader_encoding(VALUE rb_reader)
 {
-  VALUE module = rb_define_module("Nokogiri");
-  VALUE xml = rb_define_module_under(module, "XML");
+  xmlTextReaderPtr c_reader;
+  const char *parser_encoding;
+  VALUE constructor_encoding;
+
+  TypedData_Get_Struct(rb_reader, xmlTextReader, &xml_text_reader_type, c_reader);
+  parser_encoding = (const char *)xmlTextReaderConstEncoding(c_reader);
+  if (parser_encoding) {
+    return NOKOGIRI_STR_NEW2(parser_encoding);
+  }
 
-  /*
-   * The Reader parser allows you to effectively pull parse an XML document.
-   * Once instantiated, call Nokogiri::XML::Reader#each to iterate over each
-   * node.  Note that you may only iterate over the document once!
-   */
-  VALUE klass = rb_define_class_under(xml, "Reader", rb_cObject);
-
-  cNokogiriXmlReader = klass;
-
-  rb_define_singleton_method(klass, "from_memory", from_memory, -1);
-  rb_define_singleton_method(klass, "from_io", from_io, -1);
-
-  rb_define_method(klass, "read", read_more, 0);
-  rb_define_method(klass, "inner_xml", inner_xml, 0);
-  rb_define_method(klass, "outer_xml", outer_xml, 0);
-  rb_define_method(klass, "state", state, 0);
-  rb_define_method(klass, "node_type", node_type, 0);
-  rb_define_method(klass, "name", name, 0);
-  rb_define_method(klass, "local_name", local_name, 0);
-  rb_define_method(klass, "namespace_uri", namespace_uri, 0);
-  rb_define_method(klass, "prefix", prefix, 0);
-  rb_define_method(klass, "value", value, 0);
-  rb_define_method(klass, "lang", lang, 0);
-  rb_define_method(klass, "xml_version", xml_version, 0);
-  rb_define_method(klass, "depth", depth, 0);
-  rb_define_method(klass, "attribute_count", attribute_count, 0);
-  rb_define_method(klass, "attribute", reader_attribute, 1);
-  rb_define_method(klass, "namespaces", namespaces, 0);
-  rb_define_method(klass, "attribute_at", attribute_at, 1);
-  rb_define_method(klass, "empty_element?", empty_element_p, 0);
-  rb_define_method(klass, "attributes?", attributes_eh, 0);
-  rb_define_method(klass, "value?", value_eh, 0);
-  rb_define_method(klass, "default?", default_eh, 0);
-  rb_define_method(klass, "base_uri", base_uri, 0);
-
-  rb_define_private_method(klass, "attr_nodes", attribute_nodes, 0);
+  constructor_encoding = rb_iv_get(rb_reader, "@encoding");
+  if (RTEST(constructor_encoding)) {
+    return constructor_encoding;
+  }
+
+  return Qnil;
+}
+
+void
+noko_init_xml_reader(void)
+{
+  cNokogiriXmlReader = rb_define_class_under(mNokogiriXml, "Reader", rb_cObject);
+
+  rb_undef_alloc_func(cNokogiriXmlReader);
+
+  rb_define_singleton_method(cNokogiriXmlReader, "from_memory", from_memory, -1);
+  rb_define_singleton_method(cNokogiriXmlReader, "from_io", from_io, -1);
+
+  rb_define_method(cNokogiriXmlReader, "attribute", reader_attribute, 1);
+  rb_define_method(cNokogiriXmlReader, "attribute_at", attribute_at, 1);
+  rb_define_method(cNokogiriXmlReader, "attribute_count", attribute_count, 0);
+  rb_define_method(cNokogiriXmlReader, "attribute_hash", rb_xml_reader_attribute_hash, 0);
+  rb_define_method(cNokogiriXmlReader, "attributes?", attributes_eh, 0);
+  rb_define_method(cNokogiriXmlReader, "base_uri", rb_xml_reader_base_uri, 0);
+  rb_define_method(cNokogiriXmlReader, "default?", default_eh, 0);
+  rb_define_method(cNokogiriXmlReader, "depth", depth, 0);
+  rb_define_method(cNokogiriXmlReader, "empty_element?", empty_element_p, 0);
+  rb_define_method(cNokogiriXmlReader, "encoding", rb_xml_reader_encoding, 0);
+  rb_define_method(cNokogiriXmlReader, "inner_xml", inner_xml, 0);
+  rb_define_method(cNokogiriXmlReader, "lang", lang, 0);
+  rb_define_method(cNokogiriXmlReader, "local_name", local_name, 0);
+  rb_define_method(cNokogiriXmlReader, "name", name, 0);
+  rb_define_method(cNokogiriXmlReader, "namespace_uri", namespace_uri, 0);
+  rb_define_method(cNokogiriXmlReader, "namespaces", rb_xml_reader_namespaces, 0);
+  rb_define_method(cNokogiriXmlReader, "node_type", node_type, 0);
+  rb_define_method(cNokogiriXmlReader, "outer_xml", outer_xml, 0);
+  rb_define_method(cNokogiriXmlReader, "prefix", prefix, 0);
+  rb_define_method(cNokogiriXmlReader, "read", read_more, 0);
+  rb_define_method(cNokogiriXmlReader, "state", state, 0);
+  rb_define_method(cNokogiriXmlReader, "value", value, 0);
+  rb_define_method(cNokogiriXmlReader, "value?", value_eh, 0);
+  rb_define_method(cNokogiriXmlReader, "xml_version", xml_version, 0);
 }