libxml-ruby 0.9.4 → 0.9.5

data/ext/libxml/ruby_xml_html_parser.c

@@ -1,4 +1,4 @@
-/* $Id: ruby_xml_html_parser.c 630 2008-11-24 06:53:01Z cfis $ */
+/* $Id: ruby_xml_html_parser.c 650 2008-11-30 03:40:22Z cfis $ */
 
 /* Please see the LICENSE file for copyright and distribution information */
 
@@ -8,77 +8,78 @@ VALUE cXMLHTMLParser;
 static ID INPUT_ATTR;
 
 /*
-* Document-class: LibXML::XML::HTMLParser
-*
-* The HTML parser implements an HTML 4.0 non-verifying parser with an API 
-* compatible with the XML::Parser.  In contrast with the XML::Parser, 
-* it can parse "real world" HTML, even if it severely broken from a
-* specification point of view. */
-
+ * Document-class: LibXML::XML::HTMLParser
+ *
+ * The HTML parser implements an HTML 4.0 non-verifying parser with an API
+ * compatible with the XML::Parser.  In contrast with the XML::Parser,
+ * it can parse "real world" HTML, even if it severely broken from a
+ * specification point of view. */
 
 /*
  * call-seq:
  *    XML::HTMLParser.initialize -> parser
- * 
+ *
  * Initializes a new parser instance with no pre-determined source.
  */
-static VALUE
-rxml_html_parser_initialize(VALUE self) {
+static VALUE rxml_html_parser_initialize(VALUE self)
+{
   VALUE input = rb_class_new_instance(0, NULL, cXMLInput);
   rb_iv_set(self, "@input", input);
   return self;
 }
 
-static htmlDocPtr
-rxml_html_parser_read_file(VALUE input) {
+static htmlDocPtr rxml_html_parser_read_file(VALUE input)
+{
   VALUE file = rb_ivar_get(input, FILE_ATTR);
   VALUE encoding = rb_ivar_get(input, ENCODING_ATTR);
   VALUE encoding_str = rxml_input_encoding_to_s(Qnil, encoding);
-  char *xencoding_str = (encoding_str == Qnil ? NULL : StringValuePtr(encoding_str));
+  char *xencoding_str = (encoding_str == Qnil ? NULL : StringValuePtr(
+      encoding_str));
   int options = 0;
 
   return htmlReadFile(StringValuePtr(file), xencoding_str, options);
 }
 
-static htmlDocPtr
-rxml_html_parser_read_string(VALUE input) {
+static htmlDocPtr rxml_html_parser_read_string(VALUE input)
+{
   VALUE string = rb_ivar_get(input, STRING_ATTR);
   VALUE base_url = rb_ivar_get(input, BASE_URL_ATTR);
   char *xbase_url = (base_url == Qnil ? NULL : StringValuePtr(base_url));
   VALUE encoding = rb_ivar_get(input, ENCODING_ATTR);
   VALUE encoding_str = rxml_input_encoding_to_s(Qnil, encoding);
-  char *xencoding_str = (encoding_str == Qnil ? NULL : StringValuePtr(encoding_str));
+  char *xencoding_str = (encoding_str == Qnil ? NULL : StringValuePtr(
+      encoding_str));
   int options = 0;
 
   return htmlReadMemory(StringValuePtr(string), RSTRING_LEN(string),
-                        xbase_url, xencoding_str, options);
+  xbase_url, xencoding_str, options);
 }
 
-static htmlDocPtr
-rxml_html_parser_read_io(VALUE input) {
+static htmlDocPtr rxml_html_parser_read_io(VALUE input)
+{
   VALUE io = rb_ivar_get(input, IO_ATTR);
   VALUE base_url = rb_ivar_get(input, BASE_URL_ATTR);
   char *xbase_url = (base_url == Qnil ? NULL : StringValuePtr(base_url));
   VALUE encoding = rb_ivar_get(input, ENCODING_ATTR);
   VALUE encoding_str = rxml_input_encoding_to_s(Qnil, encoding);
-  char *xencoding_str = (encoding_str == Qnil ? NULL : StringValuePtr(encoding_str));
+  char *xencoding_str = (encoding_str == Qnil ? NULL : StringValuePtr(
+      encoding_str));
   int options = 0;
-  
+
   return htmlReadIO((xmlInputReadCallback) rxml_read_callback, NULL,
-        				     (void *)io, 
-                     xbase_url, xencoding_str, options);
+      (void *) io, xbase_url, xencoding_str, options);
 }
 
 /*
  * call-seq:
  *    parser.parse -> document
- * 
+ *
  * Parse the input XML and create an XML::Document with
  * it's content. If an error occurs, XML::Parser::ParseError
  * is thrown.
  */
-static VALUE
-rxml_html_parser_parse(VALUE self) {
+static VALUE rxml_html_parser_parse(VALUE self)
+{
   VALUE input = rb_ivar_get(self, INPUT_ATTR);
   htmlDocPtr xdoc;
 
@@ -90,25 +91,25 @@ rxml_html_parser_parse(VALUE self) {
     xdoc = rxml_html_parser_read_io(input);
   else
     rb_raise(rb_eArgError, "You must specify a parser data source");
-  
+
   if (!xdoc)
     rxml_raise(&xmlLastError);
 
   return rxml_document_wrap(xdoc);
 }
 
-// Rdoc needs to know 
+// Rdoc needs to know
 #ifdef RDOC_NEVER_DEFINED
-  mLibXML = rb_define_module("LibXML");
-  mXML = rb_define_module_under(mLibXML, "XML");
+mLibXML = rb_define_module("LibXML");
+mXML = rb_define_module_under(mLibXML, "XML");
 #endif
 
-void
-ruby_init_html_parser(void) {	
+void ruby_init_html_parser(void)
+{
   INPUT_ATTR = rb_intern("@input");
 
   cXMLHTMLParser = rb_define_class_under(mXML, "HTMLParser", rb_cObject);
- 
+
   /* Atributes */
   rb_define_attr(cXMLHTMLParser, "input", 1, 0);
 

data/ext/libxml/ruby_xml_input.c

@@ -8,54 +8,54 @@
 VALUE cXMLInput;
 
 /*
-* Document-class: LibXML::XML::Input
-*
-* Input is a helper class that defines a libxml data source.
-* Libxml can parse files, strings, io streams and documents
-* accessible via networking protocols such as http.
-* Be default, the ruby-libxml bindings expose parsing
-* files, io streams and strings.
-*
-* Generally you will not directly work with the input object,
-* but instead will use the various Document and Parser apis.
-* For example:
-*
-*   parser = XML::Parser.file('my_file') 
-*   parser = XML::Parser.string('<myxml/>') 
-*   parser = XML::Parser.io(File.open('my_file')) 
-*
-*  XML::HTMLParser, XML::Reader, XML::SaxParser and
-*  XML::Document work in the same way.
-*
-* LibXML converts all data sources to UTF8 internally before 
-* processing them.  By default, LibXML will determine a data
-* source's encoding using the algorithm described on its
-* website[* http://xmlsoft.org/encoding.html].
-*
-* However, its some cases it is possible to tell LibXML
-* the data source's encoding via the constants defined in
-* the Encoding module.
-*
-* Example 1:
-* 
-*   parser = XML::Parser.new
-*   parser.input.encoding = XML::Input::ISO_8859_1
-*   parser.io = File.open('some_file', 'rb')
-*   doc = parser.parse
-*
-* Example 2:
-* 
-*   parser = XML::HTMLParser.new
-*   parser.encoding = XML::Input::ISO_8859_1
-*   parser.file = "some_file"
-*   doc = parser.parse
-*
-* Example 3:
-* 
-*   document = XML::Document.new
-*   encoding_string =  XML::Input.encoding_to_s(XML::Encoding::ISO_8859_1)
-*   document.encoding = document
-*   doc << XML::Node.new */
+ * Document-class: LibXML::XML::Input
+ *
+ * Input is a helper class that defines a libxml data source.
+ * Libxml can parse files, strings, io streams and documents
+ * accessible via networking protocols such as http.
+ * Be default, the ruby-libxml bindings expose parsing
+ * files, io streams and strings.
+ *
+ * Generally you will not directly work with the input object,
+ * but instead will use the various Document and Parser apis.
+ * For example:
+ *
+ *   parser = XML::Parser.file('my_file')
+ *   parser = XML::Parser.string('<myxml/>')
+ *   parser = XML::Parser.io(File.open('my_file'))
+ *
+ *  XML::HTMLParser, XML::Reader, XML::SaxParser and
+ *  XML::Document work in the same way.
+ *
+ * LibXML converts all data sources to UTF8 internally before
+ * processing them.  By default, LibXML will determine a data
+ * source's encoding using the algorithm described on its
+ * website[* http://xmlsoft.org/encoding.html].
+ *
+ * However, its some cases it is possible to tell LibXML
+ * the data source's encoding via the constants defined in
+ * the Encoding module.
+ *
+ * Example 1:
+ *
+ *   parser = XML::Parser.new
+ *   parser.input.encoding = XML::Input::ISO_8859_1
+ *   parser.io = File.open('some_file', 'rb')
+ *   doc = parser.parse
+ *
+ * Example 2:
+ *
+ *   parser = XML::HTMLParser.new
+ *   parser.encoding = XML::Input::ISO_8859_1
+ *   parser.file = "some_file"
+ *   doc = parser.parse
+ *
+ * Example 3:
+ *
+ *   document = XML::Document.new
+ *   encoding_string =  XML::Input.encoding_to_s(XML::Encoding::ISO_8859_1)
+ *   document.encoding = document
+ *   doc << XML::Node.new */
 
 ID BASE_URL_ATTR;
 ID ENCODING_ATTR;
@@ -65,18 +65,18 @@ ID IO_ATTR;
 
 static ID READ_METHOD;
 
-
 /* This method is called by libxml when it wants to read
-   more data from a stream. We go with the duck typing
-   solution to support StringIO objects. */
-int rxml_read_callback(void *context, char *buffer, int len) {
-  VALUE io = (VALUE)context;
+ more data from a stream. We go with the duck typing
+ solution to support StringIO objects. */
+int rxml_read_callback(void *context, char *buffer, int len)
+{
+  VALUE io = (VALUE) context;
   VALUE string = rb_funcall(io, READ_METHOD, 1, INT2NUM(len));
   int size;
- 
-  if(string == Qnil)
+
+  if (string == Qnil)
     return 0;
-  
+
   size = RSTRING_LEN(string);
   memcpy(buffer, StringValuePtr(string), size);
 
@@ -86,89 +86,88 @@ int rxml_read_callback(void *context, char *buffer, int len) {
 /*
  * call-seq:
  *    Input.encoding_to_s(Input::ENCODING) -> "encoding"
- * 
+ *
  * Converts an encoding contstant defined on the XML::Input
  * class to its text representation.
  */
-VALUE
-rxml_input_encoding_to_s(VALUE self, VALUE encoding)
+VALUE rxml_input_encoding_to_s(VALUE self, VALUE encoding)
 {
   char* encodingStr = NULL;
 
   switch (NUM2INT(encoding))
   {
-    case XML_CHAR_ENCODING_ERROR:
-      encodingStr = "Error";
-      break;
-    case XML_CHAR_ENCODING_NONE:
-      encodingStr = "None";
-      break;
-    case XML_CHAR_ENCODING_UTF8:
-      encodingStr = "UTF-8";
-      break;
-    case XML_CHAR_ENCODING_UTF16LE:
-      encodingStr = "UTF-16LE";
-      break;
-    case XML_CHAR_ENCODING_UTF16BE:
-      encodingStr = "UTF-16BE";
-      break;
-    case XML_CHAR_ENCODING_UCS4LE:
-      encodingStr = "UCS-4LE";
-      break;
-    case XML_CHAR_ENCODING_UCS4BE:
-      encodingStr = "UCS-4BE";
-      break;
-    case XML_CHAR_ENCODING_EBCDIC:
-      encodingStr = "EBCDIC";
-      break;
-    case XML_CHAR_ENCODING_UCS4_2143:
-      encodingStr = "UCS-4";
-      break;
-    case XML_CHAR_ENCODING_UCS4_3412:
-      encodingStr = "UCS-4";
-      break;
-    case XML_CHAR_ENCODING_UCS2:
-      encodingStr = "UCS-2";
-      break;
-    case XML_CHAR_ENCODING_8859_1:
-      encodingStr = "ISO-8859-1";
-      break;
-    case XML_CHAR_ENCODING_8859_2:
-      encodingStr = "ISO-8859-2";
-      break;
-    case XML_CHAR_ENCODING_8859_3:
-      encodingStr = "ISO-8859-3";
-      break;
-    case XML_CHAR_ENCODING_8859_4:
-      encodingStr = "ISO-8859-4";
-      break;
-    case XML_CHAR_ENCODING_8859_5:
-      encodingStr = "ISO-8859-5";
-      break;
-    case XML_CHAR_ENCODING_8859_6:
-      encodingStr = "ISO-8859-6";
-      break;
-    case XML_CHAR_ENCODING_8859_7:
-      encodingStr = "ISO-8859-7";
-      break;
-    case XML_CHAR_ENCODING_8859_8:
-      encodingStr = "ISO-8859-8";
-      break;
-    case XML_CHAR_ENCODING_8859_9:
-      encodingStr = "ISO-8859-9";
-      break;
-    case XML_CHAR_ENCODING_2022_JP:
-      encodingStr = "ISO-2022-JP";
-      break;
-    case XML_CHAR_ENCODING_SHIFT_JIS:
-      encodingStr = "Shift_JIS";
-      break;
-    case XML_CHAR_ENCODING_EUC_JP:
-      encodingStr = "EUC-JP";
-      break;
-    case XML_CHAR_ENCODING_ASCII:
-      encodingStr = "ASCII";
-      break;
+  case XML_CHAR_ENCODING_ERROR:
+    encodingStr = "Error";
+    break;
+  case XML_CHAR_ENCODING_NONE:
+    encodingStr = "None";
+    break;
+  case XML_CHAR_ENCODING_UTF8:
+    encodingStr = "UTF-8";
+    break;
+  case XML_CHAR_ENCODING_UTF16LE:
+    encodingStr = "UTF-16LE";
+    break;
+  case XML_CHAR_ENCODING_UTF16BE:
+    encodingStr = "UTF-16BE";
+    break;
+  case XML_CHAR_ENCODING_UCS4LE:
+    encodingStr = "UCS-4LE";
+    break;
+  case XML_CHAR_ENCODING_UCS4BE:
+    encodingStr = "UCS-4BE";
+    break;
+  case XML_CHAR_ENCODING_EBCDIC:
+    encodingStr = "EBCDIC";
+    break;
+  case XML_CHAR_ENCODING_UCS4_2143:
+    encodingStr = "UCS-4";
+    break;
+  case XML_CHAR_ENCODING_UCS4_3412:
+    encodingStr = "UCS-4";
+    break;
+  case XML_CHAR_ENCODING_UCS2:
+    encodingStr = "UCS-2";
+    break;
+  case XML_CHAR_ENCODING_8859_1:
+    encodingStr = "ISO-8859-1";
+    break;
+  case XML_CHAR_ENCODING_8859_2:
+    encodingStr = "ISO-8859-2";
+    break;
+  case XML_CHAR_ENCODING_8859_3:
+    encodingStr = "ISO-8859-3";
+    break;
+  case XML_CHAR_ENCODING_8859_4:
+    encodingStr = "ISO-8859-4";
+    break;
+  case XML_CHAR_ENCODING_8859_5:
+    encodingStr = "ISO-8859-5";
+    break;
+  case XML_CHAR_ENCODING_8859_6:
+    encodingStr = "ISO-8859-6";
+    break;
+  case XML_CHAR_ENCODING_8859_7:
+    encodingStr = "ISO-8859-7";
+    break;
+  case XML_CHAR_ENCODING_8859_8:
+    encodingStr = "ISO-8859-8";
+    break;
+  case XML_CHAR_ENCODING_8859_9:
+    encodingStr = "ISO-8859-9";
+    break;
+  case XML_CHAR_ENCODING_2022_JP:
+    encodingStr = "ISO-2022-JP";
+    break;
+  case XML_CHAR_ENCODING_SHIFT_JIS:
+    encodingStr = "Shift_JIS";
+    break;
+  case XML_CHAR_ENCODING_EUC_JP:
+    encodingStr = "EUC-JP";
+    break;
+  case XML_CHAR_ENCODING_ASCII:
+    encodingStr = "ASCII";
+    break;
   }
 
   return rb_str_new2(encodingStr);
@@ -177,35 +176,34 @@ rxml_input_encoding_to_s(VALUE self, VALUE encoding)
 /*
  * call-seq:
  *    initialize -> LibXML::XML::Input instance
- * 
+ *
  * Initialize a new intput object.
  */
-static VALUE
-rxml_input_initialize(VALUE self) {
+static VALUE rxml_input_initialize(VALUE self)
+{
   rb_ivar_set(self, ENCODING_ATTR, INT2NUM(XML_CHAR_ENCODING_UTF8));
   return self;
 }
 
-
 /*
  * call-seq:
  *    input.FILE -> "FILE"
- * 
+ *
  * Obtain the FILE this parser will read from.
  */
-static VALUE
-rxml_input_file_get(VALUE self) {
+static VALUE rxml_input_file_get(VALUE self)
+{
   return rb_ivar_get(self, FILE_ATTR);
 }
 
 /*
  * call-seq:
  *    input.FILE = "FILE"
- * 
+ *
  * Set the FILE this parser will read from.
  */
-static VALUE
-rxml_input_file_set(VALUE self, VALUE FILE) {
+static VALUE rxml_input_file_set(VALUE self, VALUE FILE)
+{
   Check_Type(FILE, T_STRING);
   rb_ivar_set(self, FILE_ATTR, FILE);
   rb_ivar_set(self, IO_ATTR, Qnil);
@@ -213,27 +211,25 @@ rxml_input_file_set(VALUE self, VALUE FILE) {
   return self;
 }
 
-
 /*
  * call-seq:
  *    input.string -> "string"
- * 
+ *
  * Obtain the string this parser will read from.
  */
-static VALUE
-rxml_input_string_get(VALUE self) {
+static VALUE rxml_input_string_get(VALUE self)
+{
   return rb_ivar_get(self, STRING_ATTR);
 }
 
-
 /*
  * call-seq:
  *    input.string = "string"
- * 
+ *
  * Set the string this parser will read from.
  */
-static VALUE
-rxml_input_string_set(VALUE self, VALUE string) {
+static VALUE rxml_input_string_set(VALUE self, VALUE string)
+{
   Check_Type(string, T_STRING);
   rb_ivar_set(self, FILE_ATTR, Qnil);
   rb_ivar_set(self, IO_ATTR, Qnil);
@@ -244,44 +240,42 @@ rxml_input_string_set(VALUE self, VALUE string) {
 /*
  * call-seq:
  *    input.io -> IO
- * 
+ *
  * Obtain the IO instance this parser works with.
  */
-static VALUE
-rxml_input_io_get(VALUE self) {
+static VALUE rxml_input_io_get(VALUE self)
+{
   return rb_ivar_get(self, IO_ATTR);
 }
 
 /*
  * call-seq:
  *    input.io = IO
- * 
+ *
  * Set the IO instance this parser works with.
  */
-static VALUE
-rxml_input_io_set(VALUE self, VALUE io) {
+static VALUE rxml_input_io_set(VALUE self, VALUE io)
+{
   rb_ivar_set(self, FILE_ATTR, Qnil);
   rb_ivar_set(self, IO_ATTR, io);
   rb_ivar_set(self, STRING_ATTR, Qnil);
   return self;
 }
 
-// Rdoc needs to know 
+// Rdoc needs to know
 #ifdef RDOC_NEVER_DEFINED
-  mLibXML = rb_define_module("LibXML");
-  mXML = rb_define_module_under(mLibXML, "XML");
+mLibXML = rb_define_module("LibXML");
+mXML = rb_define_module_under(mLibXML, "XML");
 #endif
 
-
-
-// Rdoc needs to know 
+// Rdoc needs to know
 #ifdef RDOC_NEVER_DEFINED
-  mLibXML = rb_define_module("LibXML");
-  mXML = rb_define_module_under(mLibXML, "XML");
+mLibXML = rb_define_module("LibXML");
+mXML = rb_define_module_under(mLibXML, "XML");
 #endif
 
-void
-ruby_init_xml_input(void) {	
+void ruby_init_xml_input(void)
+{
   BASE_URL_ATTR = rb_intern("@base_url");
   ENCODING_ATTR = rb_intern("@encoding");
   FILE_ATTR = rb_intern("@file");
@@ -291,33 +285,34 @@ ruby_init_xml_input(void) {
   READ_METHOD = rb_intern("read");
 
   cXMLInput = rb_define_class_under(mXML, "Input", rb_cObject);
-  rb_define_singleton_method(cXMLInput, "encoding_to_s", rxml_input_encoding_to_s, 1);
+  rb_define_singleton_method(cXMLInput, "encoding_to_s",
+      rxml_input_encoding_to_s, 1);
 
   rb_define_const(cXMLInput, "UNDEFINED", INT2NUM(XPATH_UNDEFINED));
-  rb_define_const(cXMLInput, "ERROR", INT2NUM(XML_CHAR_ENCODING_ERROR));         /* No char encoding detected */
-  rb_define_const(cXMLInput, "NONE", INT2NUM(XML_CHAR_ENCODING_NONE));           /* No char encoding detected */
-  rb_define_const(cXMLInput, "UTF8", INT2NUM(XML_CHAR_ENCODING_UTF8));	          /* UTF-8 */
-  rb_define_const(cXMLInput, "UTF16LE", INT2NUM(XML_CHAR_ENCODING_UTF16LE));	    /* UTF-16 little endian */
-  rb_define_const(cXMLInput, "UTF16BE", INT2NUM(XML_CHAR_ENCODING_UTF16BE));	    /* UTF-16 big endian */
-  rb_define_const(cXMLInput, "UCS4LE", INT2NUM(XML_CHAR_ENCODING_UCS4LE));	      /* UCS-4 little endian */
-  rb_define_const(cXMLInput, "UCS4BE", INT2NUM(XML_CHAR_ENCODING_UCS4BE));	      /* UCS-4 big endian */
-  rb_define_const(cXMLInput, "EBCDIC", INT2NUM(XML_CHAR_ENCODING_EBCDIC));	      /* EBCDIC uh! */
+  rb_define_const(cXMLInput, "ERROR", INT2NUM(XML_CHAR_ENCODING_ERROR)); /* No char encoding detected */
+  rb_define_const(cXMLInput, "NONE", INT2NUM(XML_CHAR_ENCODING_NONE)); /* No char encoding detected */
+  rb_define_const(cXMLInput, "UTF8", INT2NUM(XML_CHAR_ENCODING_UTF8)); /* UTF-8 */
+  rb_define_const(cXMLInput, "UTF16LE", INT2NUM(XML_CHAR_ENCODING_UTF16LE)); /* UTF-16 little endian */
+  rb_define_const(cXMLInput, "UTF16BE", INT2NUM(XML_CHAR_ENCODING_UTF16BE)); /* UTF-16 big endian */
+  rb_define_const(cXMLInput, "UCS4LE", INT2NUM(XML_CHAR_ENCODING_UCS4LE)); /* UCS-4 little endian */
+  rb_define_const(cXMLInput, "UCS4BE", INT2NUM(XML_CHAR_ENCODING_UCS4BE)); /* UCS-4 big endian */
+  rb_define_const(cXMLInput, "EBCDIC", INT2NUM(XML_CHAR_ENCODING_EBCDIC)); /* EBCDIC uh! */
   rb_define_const(cXMLInput, "UCS4_2143", INT2NUM(XML_CHAR_ENCODING_UCS4_2143)); /* UCS-4 unusual ordering */
   rb_define_const(cXMLInput, "UCS4_3412", INT2NUM(XML_CHAR_ENCODING_UCS4_3412)); /* UCS-4 unusual ordering */
-  rb_define_const(cXMLInput, "UCS2", INT2NUM(XML_CHAR_ENCODING_UCS2));	          /* UCS-2 */
-  rb_define_const(cXMLInput, "ISO_8859_1", INT2NUM(XML_CHAR_ENCODING_8859_1));	      /* ISO-8859-1 ISO Latin 1 */
-  rb_define_const(cXMLInput, "ISO_8859_2", INT2NUM(XML_CHAR_ENCODING_8859_2));	      /* ISO-8859-2 ISO Latin 2 */
-  rb_define_const(cXMLInput, "ISO_8859_3", INT2NUM(XML_CHAR_ENCODING_8859_3));	      /* ISO-8859-3 */
-  rb_define_const(cXMLInput, "ISO_8859_4", INT2NUM(XML_CHAR_ENCODING_8859_4));	      /* ISO-8859-4 */
-  rb_define_const(cXMLInput, "ISO_8859_5", INT2NUM(XML_CHAR_ENCODING_8859_5));	      /* ISO-8859-5 */
-  rb_define_const(cXMLInput, "ISO_8859_6", INT2NUM(XML_CHAR_ENCODING_8859_6));	      /* ISO-8859-6 */
-  rb_define_const(cXMLInput, "ISO_8859_7", INT2NUM(XML_CHAR_ENCODING_8859_7));	      /* ISO-8859-7 */
-  rb_define_const(cXMLInput, "ISO_8859_8", INT2NUM(XML_CHAR_ENCODING_8859_8));	      /* ISO-8859-8 */
-  rb_define_const(cXMLInput, "ISO_8859_9", INT2NUM(XML_CHAR_ENCODING_8859_9));	      /* ISO-8859-9 */
-  rb_define_const(cXMLInput, "ISO_2022_JP", INT2NUM(XML_CHAR_ENCODING_2022_JP));     /* ISO-2022-JP */
+  rb_define_const(cXMLInput, "UCS2", INT2NUM(XML_CHAR_ENCODING_UCS2)); /* UCS-2 */
+  rb_define_const(cXMLInput, "ISO_8859_1", INT2NUM(XML_CHAR_ENCODING_8859_1)); /* ISO-8859-1 ISO Latin 1 */
+  rb_define_const(cXMLInput, "ISO_8859_2", INT2NUM(XML_CHAR_ENCODING_8859_2)); /* ISO-8859-2 ISO Latin 2 */
+  rb_define_const(cXMLInput, "ISO_8859_3", INT2NUM(XML_CHAR_ENCODING_8859_3)); /* ISO-8859-3 */
+  rb_define_const(cXMLInput, "ISO_8859_4", INT2NUM(XML_CHAR_ENCODING_8859_4)); /* ISO-8859-4 */
+  rb_define_const(cXMLInput, "ISO_8859_5", INT2NUM(XML_CHAR_ENCODING_8859_5)); /* ISO-8859-5 */
+  rb_define_const(cXMLInput, "ISO_8859_6", INT2NUM(XML_CHAR_ENCODING_8859_6)); /* ISO-8859-6 */
+  rb_define_const(cXMLInput, "ISO_8859_7", INT2NUM(XML_CHAR_ENCODING_8859_7)); /* ISO-8859-7 */
+  rb_define_const(cXMLInput, "ISO_8859_8", INT2NUM(XML_CHAR_ENCODING_8859_8)); /* ISO-8859-8 */
+  rb_define_const(cXMLInput, "ISO_8859_9", INT2NUM(XML_CHAR_ENCODING_8859_9)); /* ISO-8859-9 */
+  rb_define_const(cXMLInput, "ISO_2022_JP", INT2NUM(XML_CHAR_ENCODING_2022_JP)); /* ISO-2022-JP */
   rb_define_const(cXMLInput, "SHIFT_JIS", INT2NUM(XML_CHAR_ENCODING_SHIFT_JIS)); /* Shift_JIS */
-  rb_define_const(cXMLInput, "EUC_JP", INT2NUM(XML_CHAR_ENCODING_EUC_JP));       /* EUC-JP */
-  rb_define_const(cXMLInput, "ASCII", INT2NUM(XML_CHAR_ENCODING_ASCII));         /* pure ASCII */
+  rb_define_const(cXMLInput, "EUC_JP", INT2NUM(XML_CHAR_ENCODING_EUC_JP)); /* EUC-JP */
+  rb_define_const(cXMLInput, "ASCII", INT2NUM(XML_CHAR_ENCODING_ASCII)); /* pure ASCII */
 
   rb_define_attr(cXMLInput, "base_url", 1, 1);
   rb_define_attr(cXMLInput, "encoding", 1, 1);