mail-grouper one off script

email-date-grouper.rb

require "tempfile"
require "fileutils"
require "csv"
require "time"

# -*- coding: utf-8 -*-

rents = "/Users/seanjw/Library/Thunderbird/Profiles/gdg1f8hw.default/ImapMail/imap.googlemail.com/Parents"
output = "/Users/seanjw/Dropbox/Programming-Languages/Ruby/rents-email-dates.txt"
odd_output = "/Users/seanjw/Dropbox/Programming-Languages/Ruby/odd-rents-dates.txt"

input = File.open(rents, "r")
dirty_output= File.open(output, "r+")
odd = File.open(odd_output, "w")

count = 0
prefix = /^Date: /
input.each do |line|
  if line =~ prefix
    dirty_output.puts line.split(prefix).last
    count += 1
  end
end

input.close
dirty_output.close
puts "TOTAL: #{count}"

dirty_output= File.open(output, "r")
temp = Tempfile.new("good_data")
odd_count = 0
standard = /\d{1,2} [a-zA-z]{3} \d{4}/
dirty_output.each do |line|
  if line !~ standard
    odd.puts line
    odd_count += 1
  else
    temp.puts line
  end
end

dirty_output.close
odd.close

FileUtils.mv(temp.path, output)
puts "ODD: #{odd_count}"
temp.close

pre_csv = File.open(output)


def parseEmailDateLine(line)
  arr = line.split
  day = arr[0].chomp(",")
  time = arr[4].split(":").map{|s| s.to_i }
  offset = arr[5].length == 5 ? arr[5].insert(3,":") : "+00:00"
  row = []
  time = Time.new(arr[3], arr[2], arr[1], time[0], time[1], time[3], offset)
  local = time.getlocal
  row << local
  row << arr[0].chomp(",")
  row << [arr[1], arr[2], arr[3]].join(" ")
  row << local.hour
  row << local.day
  row << time.year
end

target_csv = CSV.open("rents-email.csv", "w") do |target|
  pre_csv.each do |line|
    line_csv = parseEmailDateLine(line)
    target << line_csv
  end
end