frequently used words picker

topwords.rb

# encoding: UTF-8
require "open-uri"
module Enumerable
  def take_by(nth)
    sort_by { |elem| yield elem }.slice(0...nth)
  end
end

class WordDictionary
  include Enumerable
  attr_reader :name, :words

  def initialize(input, name= 'none', inner_call=false)
    input = input_to_string(input, inner_call)
    @words = input.downcase.scan(/[a-z']+/)
    @freq_dic = @words.inject(Hash.new(0)) { |dic, word| dic[word] += 1 ; dic }
    @name = name
  end

  def each
    @freq_dic.each { |elem| yield elem }
  end

  def top_by_frequency(nth, &blk)
    take_by_value(nth, lambda { |v| -v }, &blk)
  end

  def bottom_by_frequency(nth, &blk)
    take_by_value(nth, lambda { |v| v }, &blk)
  end

  def top_by_length(nth, &blk)
    list = take_by_key(nth, lambda { |word| -word.length }, &blk)
    list.map { |word, freq| [word, freq, word.length] }
  end

  def select(regexp)
    text = @freq_dic.select { |word, freq| word =~ regexp }
                    .select { |word, freq| block_given? ? yield(freq) : freq }
                    .map { |word, freq| "#{word} " * freq }
                    .join(" ")
    WordDictionary.new(text, @name, true)
  end

  def to_s
    @freq_dic.to_s
  end

  def size
    @freq_dic.length
  end

  def +(other)
    arithmetics(:+, other)
  end

  def -(other)
    arithmetics(:-, other)
  end

  def &(other)
    arithmetics(:&, other)
  end

  def |(other)
    arithmetics(:|, other)
  end

  def uniq_words(nth, *base)
    base.inject(self) { |_self, b| _self - b.select(/./) { |freq| freq >= 10 }  }
        .top_by_frequency(nth)
  end
  
  protected :words
  private
  def input_to_string(input, inner_call)
    case input
    when /^http/
      begin
        open(input) { |f| return f.read }
      rescue Exception => e
        puts e
        exit
      end
    when String
      begin
        File.open(input, "r:utf-8") { |f| return f.read }
      rescue
        STDERR.puts "Argument has assumed as a text string" unless inner_call
        input
      end
    when ARGF.class
      input.read
    else
      raise "Wrong argument. ARGF, file or string are acceptable."
    end
  end

  def take_by_value(nth, sort_opt, &blk)
    val = lambda { |key, val| val }
    take_by_key_or_val(nth, sort_opt, val, &blk)
  end

  def take_by_key(nth, sort_opt, &blk)
    key = lambda { |key, val| key }
    take_by_key_or_val(nth, sort_opt, key, &blk)
  end

  def take_by_key_or_val(nth, sort_opt, by)
    @freq_dic.select { |key, val| block_given? ? yield(val) : val }
             .take_by(nth) { |key, val| sort_opt[by[key, val]] }
  end

  def arithmetics(op, other)
    result = (@words.send op, other.words).join(" ")
    WordDictionary.new(result, '', true)
  end
end

def pretty_print(data)
  max_stars = 60
  max_value = data.max_by { |word, freq| freq }.slice(1)
  data.each do |word, freq|
    stars = "*" * (max_stars * (freq/max_value.to_f)).ceil
    printf "%5d:%-5s %s\n", freq, word, stars
  end
end


if $0 == __FILE__
  base = WordDictionary.new('./public/base.txt')
  alice = WordDictionary.new('./public/alice.txt', :alice)
  jp_history = WordDictionary.new('./public/japanese_history.txt')
  p alice.uniq_words(40, base)
  p jp_history.uniq_words(40, base)
end

topwords_shoes.rb

require "topwords"
WIDTH = 510
WORD_POS = 23
WORD_WIDTH = 60
BAR_POS = WORD_POS + WORD_WIDTH
BAR_WIDTH = 350
MARGIN = 2
WORD_COLOR = "#253"
BG_COLOR = "#eed"

Shoes.app :width => WIDTH, :height => 650 do
  background BG_COLOR
  stack :height => 30 do
    background WORD_COLOR
    flow do
      caption "Frequently Used Words", :stroke => BG_COLOR

      button 'Show' do
        pretty_print
      end

      @fpara = para 'select file', :stroke => BG_COLOR

      button 'Open' do
        filename = ask_open_file
        @data = WordDictionary.new(filename).top_by_frequency(20)
        @fpara.replace File.basename(filename)
      end
    end
  end

  @body = stack :margin_top => 10 do
    def pretty_print
      max_value = @data.max { |a, b| a[1] <=> b[1] }.slice(1)
      @body.clear do
        @data.each_with_index do |(word, val), nth|
          rel_val = val/max_value.to_f
          flow :width => WIDTH, :height => 30 do
            para "#{nth+1}:", :stroke => WORD_COLOR
            para strong(word), :width => WORD_WIDTH, :left => WORD_POS,
                               :align => 'left', :stroke => WORD_COLOR

            bar = progress :width => BAR_WIDTH, :left => BAR_POS + MARGIN

            freq = para ""
            anim = animate 24 do |i|
              int = i/100.0
              bar.fraction = int * rel_val
              if int >=1 
                anim.stop
                freq.replace val, :left => BAR_POS + BAR_WIDTH + MARGIN,
                                  :stroke => WORD_COLOR
              end
            end
          end
        end
      end
    end
  end
end

words_in_books.rb

require "rubygems"
require "sinatra"
require "topwords"

BOOKS = {:alice => ['public/alice.txt', "Alice's Adventures in Wonderland"],
         :pride => ['public/pride_prejudice.txt', "Pride and Prejudice"],
         :japan => ['public/japanese_history.txt', "A History of the Japanese People"],
         :romeo => ['public/romeo_juliet.txt', "The Tragedy of Romeo and Juliet"],
         :science => ['public/outline_of_science.txt', "Outline of Science"]}

BASE = ['public/base.txt', 'base']

get '/' do
  init
  @result = [@alice, @pride, @japan, @romeo, @science].map do |obj|
    get_data(obj)
  end
  erb :index
end

post '/' do
  begin
    @input = params[:input]
    dic = WordDictionary.new(@input)
    @yours = get_data(dic).slice(1..-1)
    erb :show
  rescue
    redirect '/'
  end
end

helpers do
  def init
    @files = {}
    BOOKS.each do |name, (file, title)|
      instance_variable_set("@#{name}", WordDictionary.new(file, title) )
      @files[title] = File.basename(file)
    end
  end
  
  def get_data(obj)
    base = WordDictionary.new(BASE[0], BASE[1])
    [obj.name, obj.size, obj.top_by_frequency(30), obj.top_by_length(10), obj.uniq_words(20, base)]
  end
end

__END__
@@layout
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN"
  "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
<html>
  <head>
    <meta http-equiv="Content-type" content="text/html; charset=utf-8">
    <title>WORDS in Books</title>
    <style type="text/css" media="screen">
      * { margin: 0; padding: 0;}
      #side {float: right; padding:15px 10px 15px 10px; margin:5px 0 60em 3px; background:#471;}
      #input {margin-top:20px; background:#aa7;}
      #home_link {font-size:14pt; clear:left; margin:40px 0 0 20px;}
      
      body {background:#222;}
      a { text-decoration:none;}
      a:link, a:visited {color:#ffc;}
      a#home {color:#6b1;}
      h1, #title { color:#6b1; margin:7px 0 3px 7px;}
      h3.book_title { background:#606; color:#fc3; margin:5px 3px 3px 6px; padding:3px 10px 3px 10px; border-bottom:#300 solid 2px; border-right:#300 solid 2px; word-break: break-all;}
      h3.book_title:hover {background:#93f;}
      a:hover {color:#f6f;}
      div.obj { width:350px; float:left; padding-bottom:5px;}
      div.category { font-size:12pt; color:#f6f; font-weight:bolder; padding:3px 0 0 20px;}
      div.item { font-size:14pt; color:#ffc; padding-left:30px;}
      span#num { color:#ffc; margin-left:10px;}
    </style>
  </head>
  <body id="main">
    <div id="banner">
      <h1><a href='/' id='title'>WORDS in Books</a></h1>
    </div>
    <div id="side">
      <a href="http://www.gutenberg.org/wiki/Main_Page">Gutenberg</a><br/>
      <a href="http://www.sinatrarb.com/">Sinatra</a><br/>
      <a href="http://www8.plala.or.jp/abridge/ruby191class.html">Ruby 1.9.1 Methods List</a><br/>
      <div class='input_form'>
        <form action="/" method="POST" accept-charset="utf-8">
          <textarea name="input" id="input" rows="20">英文または英文の置かれているURLを入力してください</textarea>
          <p id='bottom'><input type="submit" value="Post"></p>
        </form>
      </div>
    </div>
    <div id="main">
      <%= yield %>      
    </div>
  </body>
</html>

@@ index
<% @result.each do |obj| %>
<div class='obj'>
  <h3 class='book_title'><a href=<%= @files[obj[0]] %>><%= obj.shift %></a></h3>
  <div class='category'>Total Words in Kind : <span id='num'><%= obj.shift %></span></div>
  <div class='category'>Most Used Words</div>
  <div class='item'>
    <% obj.shift.each do |word, freq| %>
      <%= word %>
    <% end %>
  </div>
  <div class='category'>Longest Words</div>
  <div class= 'item'>
    <% obj.shift.each do |word, freq, len| %>
      <%= word %>
    <% end %>
  </div>
  <div class='category'>Characteristic Words</div>
  <div class='item'>
    <% obj.shift.each do |word, freq| %>
      <%= word %>  
    <% end %>
  </div>
</div>
<% end %>
  
@@ show
<div class='obj'>
  <h3 class='book_title'><%= @input.slice(0...70) %></h3>
  <div class='category'>Total Words in Kind : <%= @yours.shift %></div>
  <div class='category'>Most Used Words</div>
  <div class='item'>
    <% @yours.shift.each do |word, freq| %>
      <%= word %>
    <% end %>
  </div>
  <div class='category'>Longest Words</div>
  <div class= 'item'>
    <% @yours.shift.each do |word, freq, len| %>
      <%= word %>
    <% end %>
  </div>
  <div class='category'>Characteristic Words</div>
  <div class='item'>
    <% @yours.shift.each do |word, freq| %>
      <%= word %>  
    <% end %>
  </div>
</div>
<br/><br/>
<div id='home_link'><a href='/' id='home'>Home</a></div>