Statistics on email activity.

email_stats.rb

#!/usr/bin/env ruby
# email_stats.rb

# ======================`README.md`========================
# 
# Email Stats
# ===========
# 
# Requirements
# ------------
# 
# You need **MongoDB** installed and the **bundler** gem.
# 
#     bundle install # installs gem dependencies 
# 
# It uses a database called email-stats in **MongoDB**.
# 
# Usage
# -----
# 
# You need to set your **IMAP\_USERNAME** and **IMAP\_PASSWORD** environment variable.
# 
# To download all your mail
# 
#     [email protected] IMAP_PASSWORD=password ruby email_stats.rb
# 
# To load up the environment to query with
# 
#     irb -r ./email_models.rb
# 
# Or you can use the **Rakefile**
# 
#     [email protected] IMAP_PASSWORD=password rake download
#     rake console
# 
# *Fancy!*
# 
# For some stats
# 
#     rake top_senders          # See who sends you the most emails
#     LIMIT=20 rake top_senders # See the top 20!
#     rake emails_from          # See a list of emails form someone.
#     rake reset                # It will delete your data!
# 
# 
# TODO
# ----
# 
# - multiple threads
# - homedir config. encrypt?
# - graph frequency on gnuplot
# 
# Gems Needed
# -----------
# 
# - mongo_mapper
# - bson_ext
# - colored
# 

# ======================`Rakefile`=========================
# 
# desc "open a console to script with the database with"
# task :console do
#   sh "bundle exec irb -r ./email_models.rb"
# end
# 
# desc "start downloading emails"
# task :download do
#   require "./email_stats.rb"
# end
# 
# task :env do
#   require './email_models.rb'
# end
# 
# desc "find the top senders"
# task :top_senders => [:env] do
#   limit = ENV["LIMIT"].to_i | 9 # The total amount of records to display
#   
#   hosts = {}
#   Email.all.each do |email|
#     hosts[email.address.host] = 0 unless hosts[email.address.host]
#     hosts[email.address.host] += 1
#   end
# 
#   result = hosts.sort {|a,b| -1*(a[1] <=> b[1])}
#   
#   limit = result.count if limit > result.count
#   result[0..limit].each do |r|
#     puts "#{r[1]}: \t#{r[0]}"
#   end
# end
# 
# desc "emails from someone"
# task :emails_from => [:env] do
#   print "From who: "
#   sender = STDIN.gets.chomp
#   mails = Address.where(:email => sender).first.emails
#   puts "Total: #{mails.count}"
#   mails.each do |mail|
#     puts "#{mail.time.strftime("%x %X")}: \t#{truncate(mail.subject)}"
#   end
# end
# 
# desc "reset database"
# task :reset => [:env] do
#   # To reset the database
#   Email.delete_all
#   Address.delete_all
# end
# 
# def truncate(message)
#   if message.size > 56
#     r = message[0..52] + "..."
#   else
#     r = message
#   end
#   r
# end
#
# =========================================================

# ======================`email_stats.rb`===================

require "rubygems"
require "bundler/setup"

# Mongo Mapper
require "rubygems"
require "bundler/setup"

require 'mongo_mapper'

MongoMapper.connection = Mongo::Connection.new("localhost", 27017)
MongoMapper.database = 'email-stats'

class Address
  include MongoMapper::Document
  safe
  
  key :name, String
  key :email, String
  
  def host
    email.split("@")[1]
  end
  def mail_name
    email.split("@")[0]
  end
  
  many :emails
end

class Email
  include MongoMapper::Document
  safe
  
  key :time, Time, :required => true
  key :subject, String
  key :message_id, String, :unique => true
  
  belongs_to :address
end


require 'colored'
require 'net/imap'

# Imap settings
# Change gmail to whatever address you want to use
imap = Net::IMAP.new("imap.gmail.com", 993, true)
imap.login ENV['IMAP_USERNAME'], ENV['IMAP_PASSWORD']

# See all mailboxes
# p imap.list("", "%/%")

# Gmail's All Mail inbox.
# Change accordingly depending on 
imap.examine('[Gmail]/All Mail')

# Get the search query from start or last email done
search_query = "ALL"
if Email.count > 0
  search_query = ["SINCE", (Email.all(:order => :time).last.time).strftime("%d-%b-%Y")]
  puts "Expect the first few records to be RED if you are resuming."
end

# - 86400 to get the day before

puts "Searching for: #{search_query}"

imap.search(search_query).each do |message_id|
  envelope = imap.fetch(message_id, "ENVELOPE")[0].attr["ENVELOPE"]
  
  # The sender of the email
  f = envelope.from[0]  
  address = Address.find_or_initialize_by_email([f.mailbox, f.host].join "@")
  if address.id
    address.name = f.name
    address.save
  end
  
  # Make a new email
  mail = address.emails.build
  # somehow some of my emails didn't have a date
  mail.time = Time.parse( envelope.date) if envelope.date
  mail.subject = envelope.subject
  mail.message_id = envelope.message_id
  
  env_name = "#{envelope.date}: \t#{envelope.subject[0..37] rescue "NO SUBJECT"}"
  
  # Save the new email
  if mail.save
    puts env_name.green
  else
    puts env_name.red
  end
  
end