Simple REXML-SAX stream listener example

naics_synonym_listener.rb

require 'rubygems'
require 'rexml/document'
require 'rexml/streamlistener'
include REXML

class NaicsSynonymListener
  include StreamListener

  def initialize(f)
    @outfile = File.open(f, 'w')
    @sql = <<SQL
insert into business_type_syn (business_type_syn_id, business_type_id, locale, description)
  values (business_type_syn.nextval,
            'EN',
            (select business_type_id from business_type where naics_code = '$naics$'),
            '$desc_en$');
insert into business_type_syn (business_type_syn_id, business_type_id, locale, description)
  values (business_type_syn.nextval,
            'FR',
            (select business_type_id from business_type where naics_code = '$naics$'),
            '$desc_fr$');
SQL
    @count = 0
  end


  def tag_start(name, attrs)
    case name
    when "AlphaIndex_x002F_IndexAlpha"
      @naics = @desc_en = @desc_fr = nil
      @count +=1
    end
  end

  def converted_text
    @text.gsub(/\'/, "''")
  end

  def tag_end(name)
    case name
    when "AlphaIndex_x002F_IndexAlpha"
      outfile.puts sql.gsub(/\$naics\$/, @naics).gsub(/\$desc_en\$/, @desc_en).gsub(/\$desc_fr\$/, @desc_fr)
      outfile.puts "\ncommit;\n\n" if @count % 100 == 0
    when "NAICS2007Coding"
      @naics = @text
    when "ENGLISH_DESCRIPTION"
      @desc_en = converted_text
    when "FRENCH_DESCRIPTION"
      @desc_fr = converted_text
    end
  end

  def text(text)
    @text = text
  end
end

if __FILE__ == $0
  listener = NaicsSynonymListener.new( '../data/apply_naics_synonym_load.sql')
  parser = Parsers::StreamParser.new(File.new("../data/naics_synonym.xml"), listener)
  parser.parse
end