xml2html.rb

#!/usr/bin/env ruby -wKU
# Convert XML to HTML
# 
# Created By Ed Botzum
#
# This script uses the EDM XSL Stylesheet to convert XMLs to HTMLs.
require 'nokogiri'
require 'csv'

xsl_template = Nokogiri::XSLT(File.open('Ikm837Tran.xsl'))

if ARGV.count < 2
  puts 'Please specify the data directory with --data_dir'
  exit
end

puts 'Converting XML to HTML'

index_path = ARGV[1]
header_created = false
new = CSV.open( File.join( index_path, 'index_fixed.txt' ), 'w', { :col_sep => ';', :force_quotes => true } )

CSV.foreach( File.join( index_path, 'index.txt' ), { :headers => true, :col_sep => ';' } ) do |row|
  unless header_created
    new << row.headers
    header_created =  true
  end

  file = File.join index_path, row['file_name']
  if file.downcase.include? 'xml'
    puts "  Converting #{file} to html"
    document = Nokogiri::XML File.read(file)
    html = xsl_template.transform document
    html_file = File.new file.downcase.gsub( 'xml', 'html' ), 'w'
    html_file << html
    html_file.close
    row['file_name'] = row['file_name'].downcase.gsub( 'xml', 'html' )
  end

  new << row
end

new.close
puts 'Done Converting XML to HTML'