authnet_code_scraper.rb

#!/usr/bin/env ruby
# Author  => Jason Amster
# Email => jayamster@gmail.com
# Time.now => Thu Jul 16 17:33:50 -0400 2009
# Purpose => To get those friggin error codes off of Authorize.net and into somthing i can actually use

require 'rubygems'
require "activesupport"
require 'hpricot'
require 'open-uri'

def fetch_and_html_and_parse_to_arry(url='http://developer.authorize.net/guides/SIM/Transaction_Response/Response_Reason_Codes_and_Response_Reason_Text.htm')
  puts url
  html = open(url).read
  doc = Hpricot html
  elements = doc/"tr"
  output = []
  elements.each do |e|
    tds =  e/'td'
    row = tds.map do |td|
      el = td/'p/'
      el.to_s.strip
    end.inject([]){|arr, el| arr << el; arr}
    output << row unless row[0] == 'RESPONSE CODE'
  end
  output
end

def sanitize_anchor_links(yaml_string)
  yaml_string.gsub(/<a name=\"[0-9]\"><\/a>/, '')
end

def save_results(array, file_name='authnet_codes.yml')
  text = sanitize_anchor_links(array.to_yaml)
  f = File.new(File.dirname(__FILE__)+"/#{file_name}" , 'w')
  f.write(text)
  f.close
end

save_results(fetch_and_html_and_parse_to_arry)