main.rb

def top_words(n)
  words.tally.sort_by(&:last).reverse.map(&:first)[0..n]
end

def words
  alice = File.read("alice.txt")
  alice.split(/\s+/)
end

def ngram(n)
  words.lazy.each_with_index.map do |word, index|
    [word, words[index+1]]
  end
end

def ngram_frequency(n)
  ngram(n).tally.sort_by(&:last).reverse
end

# {["Alice", "said"] => 10, ["Alice", "did"] => 2}

def markov(root: "Alice", length: 10, bigram_freq: ngram_frequency(2))
  chain = [root]
  length.times do
    root_bigrams = bigram_freq.select do |bigram, _|
      bigram.first == chain.last
    end
    total_freq = root_bigrams.sum(&:last)
    # [[Alice, said], 10], ...]
    followers = []
    root_bigrams.each do |ngram, freq|
      followers += freq.times.inject([]) { |memo| memo << ngram.last }
    end
    chain << followers[rand(total_freq)]
  end
  chain.join(" ")
end

puts markov(root: "The")
puts markov(root: "Alice")