Twitter Stats

fuckish.rb

require 'twitter'
require 'unicode_utils'

client = Twitter::Streaming::Client.new do |config|
  config.consumer_key        = "YOUR_CONSUMER_KEY"
  config.consumer_secret     = "YOUR_CONSUMER_SECRET"
  config.access_token        = "YOUR_ACCESS_TOKEN"
  config.access_token_secret = "YOUR_ACCESS_SECRET"
end

# You can fetch up to 3,200 tweets for a user, 200 at a time.
puts statuses_count = client.user.statuses_count

current_user_id = client.current_user.id

collection = []
max_id = nil

options = { count: 200, include_rts: true }

while true do
  options[:max_id] = max_id unless max_id.nil?
  response = client.user_timeline(current_user_id, options)
  collection += response

  if response.empty?
    collection.flatten
    break
  else
    max_id = response.last.id - 1
    next
  end
end

profanity_regexp = /\s* ( \S* (ху[йяюєї]|п[иі]зд|[єї]б[ауе]|бл[ьєя]|сук[аиоу]|йоб|курв)+ \S* ) \s*/ix

bad_words_collection = collection.select { |tweet| tweet.text =~ profanity_regexp }

words = Hash.new { |h,k| h[k] = 0 }

bad_words_collection.each do |tweet|
  words_arry = tweet.text.scan(profanity_regexp)
  word = words_arry.dig(0, 0)
  word = UnicodeUtils.downcase(word)
  word.gsub!(/[^[:alnum:]]/, '')
  words[word] += 1
end

(bad_words_collection.size.to_f / collection.size)*100

words_ordered = words.sort_by{|k,v| v}.reverse