escape_pres.rb

require 'cgi'
require 'nokogiri'

def escape_pres(text)
  doc = Nokogiri::HTML(text)

  pres = doc.css('pre').select do |pre| 
    # Select only PREs that don't have PREs as parents
    # Slow but works
    tag = pre.parent
    while !tag.is_a?(Nokogiri::XML::Document) && tag.name!='pre'
      tag = tag.parent
    end
    tag.is_a?(Nokogiri::XML::Document)
  end

  pres.each do |pre|
    pre.children = pre.children.map {|node|
      if node.name=='code'
        node.children = CGI.escapeHTML(node.children.to_html)
        node.to_html
      else
        CGI.escapeHTML(node.to_html)
      end
    }.join('')
  end

  doc.to_html
end

escape_pres_spec.rb

load 'escape_pres.rb'

describe 'The escape_pres function' do
  it 'should escape tags within PRE tag' do
    result = escape_pres('<pre><br></pre>')
    result.should include '&lt;br&gt;'
  end

  it 'should not escape tags outside of PRE tags' do
    result = escape_pres('<pre></pre><br>')
    result.should include '<br>'
  end

  it 'should escape PRE tags within PRE tags' do
    result = escape_pres('<pre><pre></pre></pre>')
    result.should include '&lt;pre&gt;'
  end

  it 'should not escape CODE tags within PRE tags' do
    result = escape_pres('<pre><code><br></code></pre>')
    result.should include '<code>'
    result.should include '&lt;br&gt;'
  end
end