TDD Word Analytics - Given a long string or set of paragraphs, perform some analysis that provides the following: Number of each word Number of each letter Number of each symbol (any non-letter and non-digit character, excluding white space) Top three most common words Top three most common letters

tdd_word_analytics.rb

class WordAnalytics
  def initialize(stringed_words)
    @stringed_words = stringed_words
  end

  def separate_words
    @words = []
    split_words = []
    split_words = @stringed_words.split(' ')
    split_words.each do |word|
      @words << word.downcase.gsub(/\W/, '')
    end

    @words
  end

  def count_words
    @word_count = {}
    @words.each do |word|
      if @word_count.keys.include?(word)
        @word_count[word] += 1
      else
        @word_count[word] = 1
      end
    end

    @word_count
  end

  def count_letters
    @letter_count = {}
    @stringed_words.gsub(/\W/, '').split('').each do |letter|
      letter.downcase!
      if @letter_count.keys.include?(letter)
        @letter_count[letter] += 1
      else
        @letter_count[letter] = 1
      end
    end

    @letter_count
  end

  def count_symbols
    symbol_count = {}
    @stringed_words.gsub(/\w/, '').split('').each do |symbol|
      if symbol_count.keys.include?(symbol)
        symbol_count[symbol] += 1
      else
        symbol_count[symbol] = 1 unless symbol == " "
      end
    end

    symbol_count
  end

  def common_words
    top_three_words = {}
    sorted_words = @word_count.sort_by {|key,value| value}.reverse
    sorted_words = sorted_words[0..2]
    sorted_words.each do |index|
      top_three_words[index[0]] = index[1]
    end

    top_three_words
  end

  def common_letters
    top_three_letters = {}
    sorted_letters = @letter_count.sort_by {|key,value| value}.reverse
    sorted_letters = sorted_letters[0..2]
    sorted_letters.each do |index|
      top_three_letters[index[0]] = index[1]
    end

    top_three_letters
  end

  def display_results
    separate_words
    puts "The number of each word is #{count_words}"
    puts "The number of each letter is #{count_letters}"
    puts "The number of each symbol is #{count_symbols}"
    puts "The top three most common words are #{common_words}"
    puts "The top three most common letters are #{common_letters}"
  end

end

word_analytics_spec.rb

require 'rspec'
require_relative 'word_analytics'

describe WordAnalytics do

  it 'should separate words from symbols' do
    word_analytics = WordAnalytics.new("How, how, now brown cow! cow cow")
    expect(word_analytics.separate_words).to eql(["how", "how", "now", "brown", "cow", "cow", "cow"])
  end

  it 'should count words' do
    word_analytics = WordAnalytics.new("How, how, now brown cow! cow cow")
    word_analytics.separate_words
    expect(word_analytics.count_words).to eql({"how"=>2, "now"=>1, "brown"=>1, "cow"=>3})
  end

  it 'should count letters' do
    word_analytics = WordAnalytics.new("How, how, now brown cow! cow cow")
    expect(word_analytics.count_letters).to eql({"h"=>2, "o"=>7, "w"=>7, "n"=>2, "b"=>1, "r"=>1, "c"=>3})
  end

  it 'should count symbols without including any non-letter and non-digit character or white space' do
    word_analytics = WordAnalytics.new("How, how, now brown cow! cow cow")
    expect(word_analytics.count_symbols).to eql({","=>2, "!"=>1})
  end

  it 'should provide the three most common words' do
    word_analytics = WordAnalytics.new("How, how, now brown cow! cow cow")
    word_analytics.separate_words
    word_analytics.count_words
    expect(word_analytics.common_words).to eql({"cow"=>3, "how"=>2, "now"=>1})
  end

  it 'should provide the three most common letters' do
    word_analytics = WordAnalytics.new("How, how, now brown cow! cow cow")
    word_analytics.count_letters
    expect(word_analytics.common_letters).to eql({"w"=>7, "o"=>7, "c"=>3})
  end
end