Contents

#!/usr/bin/env ruby

require 'rbconfig'
require 'tidy'
require 'nokogiri'

require 'strelka/cms/pagefilter' unless defined?( Strelka::CMS::PageFilter )

# A filter that cleans up broken HTML using libtidy.
class Strelka::CMS::PageFilter::Cleanup < Strelka::CMS::PageFilter

	# Options to pass to libtidy
	TIDY_OPTIONS = {
		:show_warnings     => true,
		:indent            => true,
		:indent_attributes => false,
		:indent_spaces     => 4,
		:vertical_space    => true,
		:tab_size          => 4,
		:wrap_attributes   => true,
		:wrap              => 100,
		:output_xhtml      => true,
		:char_encoding     => 'utf8'
	  }


	### Process the +page+'s source with the filter and return the altered content.
	def process( source, page )
		Tidy.open( TIDY_OPTIONS ) do |tidy|
			xml = tidy.clean( source )

			errors = tidy.errors
			self.log.error( errors.join('; ') ) unless errors.empty?

			return xml
		end
	end

end # class Strelka::CMS::Page::CleanupFilter

Version data entries

6 entries across 6 versions & 1 rubygems

Version	Path
strelka-cms-0.3.0	lib/strelka/cms/pagefilter/cleanup.rb
strelka-cms-0.2.0	lib/strelka/cms/pagefilter/cleanup.rb
strelka-cms-0.1.0	lib/strelka/cms/pagefilter/cleanup.rb
strelka-cms-0.0.1	lib/strelka/cms/pagefilter/cleanup.rb
strelka-cms-0.0.1.pre.19	lib/strelka/cms/pagefilter/cleanup.rb
strelka-cms-0.0.1.pre.15	lib/strelka/cms/pagefilter/cleanup.rb