server.rb

# YADTTMCG
## Yet another Donald Trump Tweet Markov Chain Generator

require 'sinatra'
require 'open-uri'
require 'nokogiri'

def next_word ary
  ary[rand(ary.length).to_i][1]
end

def fetch_url
  Nokogiri::HTML(open("https://twitter.com/realdonaldtrump"))
end

def scrub_strategy(doc)
  doc.css('.tweet-text')
    .text
    .gsub(/@\w+/, " ")
    .gsub(/#\w+/, " ")
    .tr('"', ' ')
    .gsub(/\ +/, " ")
    .gsub('\n', "\n")
    .gsub(/pic\.twitter\.com\/\w+/, "")
    .gsub('!', "! ")
end

def stems_strategy(file)
  pseudo_stems = file.each_cons(3).to_a.shuffle
  pseudo_stems << file.each_cons(4).to_a.shuffle
  pseudo_stems << file.each_cons(5).to_a.shuffle
  pseudo_stems.shuffle.group_by { |word_pair| word_pair[0] }
end

def take_tweet(enum)
  output = ""
  140.times do
    output << enum.next + " " rescue enum.rewind
  end
  output.chars.take(140).join
end

def make_enum(stems)
  Enumerator.new do |e|
    word = stems.first.first
    while word
      e << word
      word = next_word(stems[word] || stems.first)
    end
  end
end

get '/' do
  doc = fetch_url
  file = scrub_strategy(doc).split(" ")
  stems = stems_strategy(file)
  enum = make_enum(stems)
  take_tweet(enum)
end