#!/usr/bin/env ruby
# -*- ruby -*-
#
# Input:  
#  Output of analyze_mail with lines like this:
#  <date> \t <email>
#
# Output:
#  CSV with emails as columns and sorted dates as rows.
#
# Note:
#  This sucks, but it works well enough for now. I'm going to get back
#  to watching football.
#

require 'utils.rb'

def main(argv)
  dates2emails2counts = get_dates2emails2counts argv
  sorted = dates2emails2counts.sort {|a,b| a[0] <=> b[0]}

  # Gather total counts of emails and note the dates that should be filtered out
  emails2totals = {}
  min_date_amount = 500
  dates2exclude = {}
  sorted.each do |date,emails2counts|
    total = 0
    emails2counts.each do |email,count|
      cnt = emails2totals[email]
      cnt = 0 if !cnt
      total += cnt
      cnt += count
      emails2totals[email] = cnt
    end
    if total < min_date_amount
      dates2exclude[date] = true
    end
  end

  # Output the first row as emails, so long as that email has enough entries
  min_email_amount = 100
  print 'DATE'
  emails2totals.each do |email,total|
    next if total<min_email_amount
    print ',' + email
  end
  puts

  # Print out the rows.  Every row starts with a date and then
  # followed by the counts for every column (email)
  sorted.each do |date,emails2counts|
    next if dates2exclude[date]
    print date.gsub(/\//,':')
    emails2totals.each do |email,total|
      next if total<min_email_amount
      count = emails2counts[email]
      count = 0 if !count
      print ',' + count.to_s
    end
    puts
  end

end

main ARGV