regexworkshop.rb

require 'nokogiri'
require 'open-uri'
require 'colorize'
require 'ripl'
require 'yaml'

class RegexWorkshop

  def initialize()
    reset
  end

  def reset
    @ua_dict = {}
    @ua_test = {}
    @focus = nil
    @last_match_regex = nil
  end

  def focus() @focus end
  def focus=(browser) @focus = browser end
  def verbose() @verbose end
  def verbose=(v=true) @verbose = v end

  def ua_dict
    if focus.nil?
      @ua_dict
    else
      {focus => @ua_dict[focus]}
    end
  end

  def load(file)
    if /^http:\/\/(www\.)?useragentstring\.com/ =~ file
      doc = Nokogiri::HTML(open(file))
      ua_links = doc.xpath("//div[@id='liste']//*[self::h4 or self::a]")

      browser = nil
      ua_links.each do |node|
        if node.name== 'h4'
          # start new
          browser = node.content.strip
          @ua_dict[browser] = []
        elsif browser
          @ua_dict[browser] << node.content.strip
        end
      end
    elsif File.exists? file
      data = YAML.load_file(file)

      data.each do |k, v|
        @ua_dict[k] = [] if @ua_dict[k].nil?
        @ua_dict[k].concat v
      end
    else
      $stderr.puts "Cannot load file"
    end
  end

  def dump(file)
    File.open(file, 'w') do |out|
      YAML.dump(@ua_dict, out)
    end
  end

  def show(opts={:show_true=>true,:show_false=>true})
    browser    = opts[:browser] || focus
    show_true  = opts[:show_true]
    show_false = opts[:show_false]
    if browser.nil?
      expand_dict = @ua_test || {}
    else
      expand_dict = {browser => @ua_test[browser]}
    end
    expand_dict.each do |b, match_data|
      next if match_data.empty?
      puts b
      match_data[true].each do |ua|
        puts ua.green
      end if show_true
      match_data[false].each do |ua|
        puts ua.red
      end if show_false
      puts
    end
    nil
  end

  def match_last
    unless @last_match_regex.nil?
      match @last_match_regex
    end
    @last_match_regex
  end

  def match(regex, opts={:show_true=>true, :show_false=>true})
    @last_match_regex = regex
    ua_test = {}

    ua_dict.each do |b, uas|
      ua_test[b] = {true=>[], false=>[]}
      uas.each do |ua|
        ua_test[b][(regex =~ ua).nil?.!] << ua
      end
    end

    if verbose
      ua_test.each do |b, match_data|
        match_data[true].each do |ua|
          puts ua.green
        end if opts[:show_true]
        match_data[false].each do |ua|
          puts ua.red
        end if opts[:show_false]
      end
    else
      ua_test.each do |b, match_data|
        num_t = match_data[true].length
        num_f = match_data[false].length
        sum   = num_t + num_f
        next if sum.zero?
        pc = "%d/%d" % [num_t, sum]
        if num_t.zero?
          pc = pc.light_blue
        elsif num_f.zero?
          pc = pc.green
        else
          pc = pc.red
        end
        puts "#{b} (#{pc})"
      end
    end

    @ua_test = ua_test
    nil
  end

end

## RUN

regex_workshop = RegexWorkshop.new
ARGV.each do |url|
  regex_workshop.load url
end
Ripl.start :binding => regex_workshop.instance_eval { binding }