delete-hipchat-history.rb

#!/usr/bin/env ruby

# To make it work:
# - change account details to yours ( obviously ;) )
# - install mechanize gem in version 1.0.0 ( gem install mechanize -v '1.0.0' )
# - look for !!HERE!! comments in code in case you'd like to change default behavior
# - for best results run this script few times as hipchat / ruby openssl don't play together that well and
#   there might be an exception thrown from time to time. Script should retry then but hey.. Nobody is perfect.
# .. and finally. Remember that other part should run this script as well to make history clean.

require 'rubygems'
require 'mechanize' # for making everything work
require 'chronic' # for parsing weird timestamps from hipchat

account = {
    :email    => "[email protected]",
    :password => "pswd",
}

agent = Mechanize.new
agent.redirect_ok = true
site = agent.post("https://www.hipchat.com/sign_in", account).uri.host

# Get our user id
member_id = agent.get("https://#{site}/account").
    search('img[alt="Your photo"]').first["src"].match(%r{photos/(\d+)})[1]

# Determine our Signup date and the actual date
registered_around = agent.get("https://#{site}/people/show/#{member_id}").
    search('li:contains("Signed up")').text.sub(/Signed up: /, '')
    
registered_on = Date.parse(Chronic.parse(registered_around).to_s)
# !!HERE!! 
# You can change behaviour to 'remove only messages since...' by changing registered_on 
# variable to specific date, like:
# registered_on = Date.parse("01-01-2014")

puts "Found registration date: #{registered_on}"
intervals = (registered_on..Date.today).map { |d| d.strftime("/%Y/%m/%d") }

# Identify all users ids
members = agent.get("https://#{site}/people").search("ul.members").children.search("a").
    map { |m| m.attributes["href"].value.to_s.split('/')[3] }.
    reject { |m| m =~ /#{member_id}/ }
    
# !!HERE!!
# Instead of removing all the private chats, with everyone in your organisation 
# you might define single user or user ids just like this:
# members = [ "666" , "777" ]

max_process = 20
# !!HERE!!
# I _don't_ recommend going above 20 as it makes hipchat devops cry and their servers suffer.
# Just leave default '20' and leave script for the night or weekend ;)

processing_pids = {}

# deleting all the history
members.each do |member|
  intervals.each do |interval|
    # Added some level of verbosity in checks.
    puts "Looking for chats with user #{member} on #{interval.gsub(/^\//, '')}"
    day_history = agent.get("https://#{site}/history/member/#{member.to_i}#{interval}")
    day_history.forms.each do |f|
      begin
        if f.action =~ /history\/member\/#{member}/
          puts ">> QUEUE >> #{processing_pids.size}"
          while processing_pids.size < max_process
            pid_starting = Process.fork {
              puts "Deleting.."
              f.submit
            }
            processing_pids[pid_starting] = "1"
          end
        end
        if processing_pids.size != 0
          if pid_done = Process.wait(0, Process::WNOHANG)
            if job_finished = processing_pids.delete(pid_done)
              puts "Deleted"
            end
          else
            sleep(2)
            # raise an exception to get a retry
            raise "Force retry"
          end
        end
      rescue
        puts "Retrying"
        retry
      end
    end
  end
end