LibXML adapter

Purpose

The LibXML adapter provides XML processing through the libxml-ruby library, offering excellent performance through native libxml2 bindings with full XPath 1.0 support.

Overview

libxml-ruby provides Ruby bindings to the libxml2 C library, offering similar performance characteristics to Nokogiri but as an alternative implementation. It’s ideal when you need native performance with full XML features.

Installation

Add to your Gemfile:

gem 'moxml'
gem 'libxml-ruby'

Install:

bundle install

Configuration 

# Explicit selection
context = Moxml.new
context.config.adapter = :libxml

# Global default
Moxml::Config.default_adapter = :libxml

Features

Full XPath 1.0 support

Complete XPath 1.0 implementation through libxml2:

doc = Moxml.new.parse(xml)

# All XPath features work
books = doc.xpath('//book[@price < 30]')
count = doc.xpath('count(//book)')
titles = doc.xpath('//book[position() < 3]/title')

# Namespace-aware queries
doc.xpath('//ns:element', 'ns' => 'http://example.org')

# Complex predicates and functions
doc.xpath('//book[author and price > 20]')
doc.xpath('//chapter[last()]')

Complete namespace support

Full namespace handling:

# Create namespaced elements
element.add_namespace('dc', 'http://purl.org/dc/elements/1.1/')

# Query with namespaces
results = doc.xpath('//dc:creator',
  'dc' => 'http://purl.org/dc/elements/1.1/')

# Namespace inheritance works correctly

All node types

Complete support for:

  • Elements with attributes

  • Text nodes

  • CDATA sections

  • Comments

  • Processing instructions

  • DOCTYPE declarations (with limitations)

  • XML declarations

Performance

  • Parsing speed: Very fast (native libxml2)

  • Serialization speed: Very fast

  • Memory usage: Excellent

  • XPath performance: Excellent (native)

Limitations

Minor limitations compared to other adapters:

DOCTYPE handling:

  • DOCTYPE parsing works correctly

  • DOCTYPE serialization is limited

  • Round-trip preservation of DOCTYPE may not work perfectly

Performance:

  • Serialization slightly slower than Ox in some cases

  • Still very competitive with other adapters

# DOCTYPE limitation example
xml_with_doctype = <<~XML
  <!DOCTYPE root SYSTEM "test.dtd">
  <root/>
XML

doc = Moxml.new.parse(xml_with_doctype)
# Parsing works fine

# But re-serialization may not preserve DOCTYPE perfectly
output = doc.to_xml
# DOCTYPE may be formatted differently or missing

Performance characteristics

Based on benchmarks:

Operation Performance Notes

Parse medium XML

~120 ips

Very fast native parsing

Serialize medium XML

~1,200 ips

Fast serialization

XPath queries

~50,000+ ips

Native libxml2 XPath

Memory usage

Excellent

Efficient memory management

Best use cases

Choose LibXML when:

  • You want an alternative to Nokogiri

  • Native C performance is important

  • Full XPath 1.0 support is required

  • Namespace handling is critical

  • You prefer libxml2 over libxml2-wrapped-by-Nokogiri

Avoid LibXML when:

  • Pure Ruby is required (use Oga)

  • DOCTYPE round-trip is essential

  • You need wider community/ecosystem (use Nokogiri)

Example usage

Basic operations 

require 'moxml'

# Configure LibXML adapter
context = Moxml.new
context.config.adapter = :libxml

xml = '<library><book id="1">Ruby Programming</book></library>'
doc = context.parse(xml)

# Query and modify
book = doc.at_xpath('//book[@id="1"]')
book.text = 'Advanced Ruby Programming'
book['edition'] = '2nd'

puts doc.to_xml(indent: 2)

Complex XPath 

# All XPath 1.0 features supported
expensive_books = doc.xpath('//book[price > 30]')
fiction_count = doc.xpath('count(//book[@category="fiction"])')
last_chapter = doc.at_xpath('//chapter[last()]')

Namespace operations 

xml = <<~XML
  <library xmlns:dc="http://purl.org/dc/elements/1.1/">
    <book>
      <dc:title>Programming</dc:title>
      <dc:creator>Smith</dc:creator>
    </book>
  </library>
XML

doc = Moxml.new.parse(xml)

# Query with namespaces
ns = { 'dc' => 'http://purl.org/dc/elements/1.1/' }
titles = doc.xpath('//dc:title', ns)
creators = doc.xpath('//dc:creator', ns)

puts "Title: #{titles.first.text}"
puts "Creator: #{creators.first.text}"

Comparison with Nokogiri

Both LibXML and Nokogiri use libxml2, but differ in their approach:

Aspect LibXML Nokogiri

Underlying library

libxml2 directly

libxml2 via wrapper

Community size

Smaller

Very large

Performance

Excellent

Excellent

Feature completeness

Full (except DOCTYPE)

Full

Pure Ruby option

No

No

Cross-platform

Good

Excellent

Documentation

Good

Extensive

References

See also