Rolling Tweet Deletion

tweettrim.rb

#!/usr/bin/env ruby
require 'rubygems'
require 'twitter'
require 'json'
require 'faraday'

# things you must configure
TWITTER_USER = "you"

# get these from dev.twitter.com
CONSUMER_KEY = "-"
CONSUMER_SECRET = "-"
OAUTH_TOKEN = "aka access key"
OAUTH_TOKEN_SECRET = "aka access secret"

# things you might want to change
MAX_AGE_IN_DAYS = 10 # anything older than this is deleted / unfavorited
DELAY_BETWEEN_DELETES = 0.5 # in seconds
DELAY_BETWEEN_REQS = 45

# save tweets that have been massively favorited or retweeted??? nah. all is vanity.
POINT_THRESHOLD = 9999

# you don't need to mess with this
API_TWEET_MAX = 3200
TWEETS_PER_PAGE = 200 # api max is 250 or something... 200 seems like a nice round number
NUM_PAGES = (API_TWEET_MAX / TWEETS_PER_PAGE).to_i
MAX_AGE_IN_SECONDS = MAX_AGE_IN_DAYS*24*60*60
NOW_IN_SECONDS = Time.now

FIELDS = [:id,
        :created_at,
        :text,
        :retweet_count,
        :in_reply_to_screen_name,
        :in_reply_to_user_id,
        :in_reply_to_status_id]

# tweet methods

client = Twitter::REST::Client.new do |config|
  config.consumer_key        = CONSUMER_KEY
  config.consumer_secret     = CONSUMER_SECRET
  config.access_token        = OAUTH_TOKEN
  config.access_token_secret = OAUTH_TOKEN_SECRET
end

def delete_from_twitter(tweet,client)
  begin
    client.destroy_status(tweet.id)
  rescue StandardError => e
    puts e.inspect
    puts "Error deleting #{tweet.id} "
    # exit
  else
    puts "Deleted"
  end
end

# init twitter

error_count = 0
del_count = 0

timeline = []

begin
  tweet_max = client.user(TWITTER_USER).statuses_count < 3200 ? client.user(TWITTER_USER).statuses_count : API_TWEET_MAX
  oldest_page = (tweet_max / TWEETS_PER_PAGE).to_i # we go to this page -- the older tweets -- first
  puts "#{tweet_max} tweets available, Oldest page is #{oldest_page}"
rescue StandardError => e
  puts e
  puts "Error getting info about @#{TWITTER_USER}. Try again."
  exit
end

(1..oldest_page).each do |i|
  begin
    puts "Requesting tweets #{i*TWEETS_PER_PAGE}-#{(i+1)*TWEETS_PER_PAGE}"
    timeline = timeline.concat(client.user_timeline(TWITTER_USER,{:count=>TWEETS_PER_PAGE,:page=>i,:include_entities=>true,:include_rts=>true}))
  rescue
    error_count += 1
    if (error_count > 5) then
      puts "Too many errors! Try again later."
      exit
    else
      puts "Error getting tweets #{oldest_page*TWEETS_PER_PAGE}-#{tweet_max}, retrying after #{2**error_count} seconds..."
      sleep(2**error_count)
      retry
    end
  end
  sleep DELAY_BETWEEN_DELETES
end

puts "Got #{timeline.size} tweets"
puts ""

timeline.each do |tweet|
  tweet_age = NOW_IN_SECONDS - tweet.created_at

  if (tweet_age < MAX_AGE_IN_SECONDS) then
    puts "Ignored a tweet #{(tweet_age/(24*60*60)).round} days old"
  elsif (tweet.favorite_count + tweet.retweet_count >= POINT_THRESHOLD) and (! tweet.text.match(/^RT/)) then
    puts "Ignored a tweet with #{tweet.favorite_count} faves and #{tweet.retweet_count} RTs"
  else
    puts "Preparing to delete a tweet #{(tweet_age/(24*60*60)).round} days old"
    delete_from_twitter(tweet,client)
    del_count += 1
  end

  puts "  #{tweet.text}"
  puts ""

  sleep DELAY_BETWEEN_DELETES
end

puts "deleted #{del_count}"