Display top 10 words, % and also a histogram distribution.

word_histogram.rb

#!/usr/bin/ruby
#encoding: UTF-8

require 'terminal-table'
require 'histogram/array'

# Usage:
# echo "a e i o u" | word_histogram

top_words = 10
contents = ''
ARGF.each { |line| contents << line }

# == count occurrence of all words ==
words_histogram_arr = []
words_histogram = Hash.new(0)
words = contents.split(/\W+/)
words.each { |word| words_histogram[word] += 1 if word.length > 3 }
words_histogram.each { |word, count| words_histogram_arr << {:word => word, :count => count} }
sort_by_count = words_histogram_arr.sort { |a, b| a[:count] <=> b[:count] }.reverse


# == create table with top words ==
total_words = words.length
options = {:title => "Top #{top_words} Words", :headings => ['Word', 'Count', '%']}
table = Terminal::Table.new(options) do |t|
  sort_by_count[0 .. top_words].each { |item|
    percentage = (item[:count].to_i*100.0)/total_words.to_i
    t.add_row [item[:word], item[:count], "#{sprintf('%.2f', percentage)}%"]
  }
  t.add_separator
  t.add_row ['total words ', total_words, '100%']
end
puts table

# == create histogram ==
sort_by_length = words_histogram_arr.sort { |a, b| a[:word].length <=> b[:word].length }.reverse
max_num_bars = 100
total_words = words.length
options = {
    :title => 'Word Histogram',
    :headings => %w(count length percentage word),
    :style => {:border_x => '', :border_i => '', :border_y => ''}
}
table = Terminal::Table.new(options) do |t|
  sort_by_length.each { |item|
    percentage = (item[:count].to_i*100.0)/total_words.to_i
    length = sprintf('%.2f', (percentage * max_num_bars * 1.0 / 100))
    next if length.to_i == 0
    bars = Array.new(length.to_i, '#').join('')
    t.add_row [
                  item[:count].to_s.rjust(2), #count
                  item[:word].length,  # length
                  sprintf('%.2f', percentage), # percentage
                  item[:word],
                  bars
              ]
  }
end
puts table