GitHub repo: github.com/ohler55/ox

RubyGems repo: rubygems.org/gems/ox

Optimized XML (Ox), as the name implies was written to provide speed optimized XML handling. It was designed to be an alternative to Nokogiri in generic XML parsing and as an alternative to Marshal for Object serialization.

Nokogiri relies on libXml while Ox is self contained. Ox uses nothing other than standard C libraries so version issues with libXml are not an issue.

Marshal uses a binary format for serializing Objects. That binary format changes with releases making Marshal dumped Object incompatible between some versions. The use of a binary format make debugging message streams or file contents next to impossible unless the same version of Ruby and only Ruby is used for inspecting the serialize Object. Ox on the other hand uses human readable XML. Ox also includes options that allow strict, tolerant, or a mode that automatically defines missing classes.

It is possible to write an XML serialization gem with Nokogiri but writing such a package in Ruby results in a module significantly slower than Marshal. This is what triggered the start of Ox development.

Ox handles XML documents in two ways. It is a generic XML parser and writer as well as a fast Object / XML marshaller. Ox was written for speed as a replacement for Nokogiri and for Marshal.

As an XML parser it is 2 or more times faster than Nokogiri and as a generic XML writer it is as much as 20 times faster than Nokogiri. Of course different files may result in slightly different times.

As an Object serializer Ox is up to 6 times faster than the standard Ruby Marshal.dump() and up to 3 times faster than Marshal.load().

require 'ox'

class Sample
  attr_accessor :a, :b, :c

  def initialize(a, b, c)
    @a = a
    @b = b
    @c = c
  end
end

# Create Object
obj = Sample.new(1, "bee", ['x', :y, 7.0])
# Now dump the Object to an XML String.
xml = Ox.dump(obj)
# Convert the object back into a Sample Object.
obj2 = Ox.parse_obj(xml)

require 'ox'

doc = Ox::Document.new(:version => '1.0')

top = Ox::Element.new('top')
top[:name] = 'sample'
doc << top

mid = Ox::Element.new('middle')
mid[:name] = 'second'
top << mid

bot = Ox::Element.new('bottom')
bot[:name] = 'third'
mid << bot

xml = Ox.dump(doc)
puts xml
doc2 = Ox.parse(xml)
puts "Same? #{doc == doc2}"

The XML format used for Object encoding follows the structure of the Object. Each XML element is encoded so that the XML element name is a type indicator. Attributes of the element provide additional information such as the Class if relevant, the Object attribute name, and Object ID if necessary.

The type indicator map is:

• a => Array

• b => Base64

• c => Class

• f => Float

• g => Regexp

• h => Hash

• i => Fixnum

• j => Bignum

• l => Rational

• m => Symbol

• n => FalseClass

• o => Object

• p => Ref

• r => Range

• s => String

• t => Time

• u => Struct

• v => Complex

• x => Raw

• y => TrueClass

• z => NilClass

If the type is an Object, type ‘o’ then an attribute named ‘c’ should be set with the full Class name including the Module names. If the XML element represents an Object then a sub-elements is included for each attribute of the Object. An XML element attribute ‘a’ is set with a value that is the name of the Ruby Object attribute. In all cases, except for the Exception attribute hack the attribute names begin with an @ character. (Exception are strange in that the attributes of the Exception Class are not named with a @ suffix. A hack since it has to be done in C and can not be done through the interpreter.)

Values are encoded as the text portion of an element or in the sub-elements of the principle. For example, a Fixnum is encoded as:

<i>123</i>

An Array has sub-elements and is encoded similar to this example.

<a>
  <i>1</i>
  <s>abc</s>
</a>

A Hash is encoded with an even number of elements where the first element is the key and the second is the value. This is repeated for each entry in the Hash. An example is of { 1 => ‘one’, 2 => ‘two’ } encoding is:

<h>
  <i>1</i>
  <s>one</s>
  <i>2</i>
  <s>two</s>
</h>

Strings with characters not allowed in XML are base64 encoded amd will be converted back into a String when loaded.

Ox supports circular references where attributes of one Object can refer to an Object that refers back to the first Object. When this option is used an Object ID is added to each XML Object element as the value of the ‘a’ attribute.