Email

read_mbox.rb

require 'kconv'
require 'mail'  # gem install mail
require 'csv'

def parse_message(message)
	m = Mail.new(message)
	o = {:id => m.message_id, :from => m.from, :to => m.to, :subject => m.subject, :reply => m.in_reply_to, :date => m.date, :ref => m.references}		
	p o
	o
end

def read_mbox(path,max_count=100000)
	f = open(path)

	message = nil

	res = []
	count = 0
	while (line = f.gets)
		begin
			m = line.match(/\AFrom /)
		rescue
			m = nil
		end
		if m
			count += 1
			res << parse_message(message) if (message)
			print '.' if count % 100 == 0
			puts " #{count} processed." if count % 1000 == 0
			if count >= max_count
				break
			end
			message = ''
		else
			begin
				s = line.sub(/^\>From/, 'From')
			rescue
				s = line
			end
			message << s
		end
	end

	f.close

	Hash[res.map{|m| [m[:id], m]}]

end

folder = "/path/to/mbox"
user = '****'
sent = read_mbox("#{folder}/#{user}/Sent Mail.mbox/mbox", 100000)
all = read_mbox("#{folder}/#{user}/All Mail.mbox/mbox", 100000)

count = Hash.new
count[:reply] = 0
count[:not_reply] = 0
count[:error] = 0
res = []

CSV.open("result.csv", "wb") do |csv|
	sent.each{|k,m|
		if not m[:reply]   # Not a reply
	#		p ['not a reply', m]
			count[:not_reply] += 1
			next
		end

		if not all[m[:reply]]
		#	p ['Missing replied message.', m[:id], m[:from], m[:subject], m[:reply]]
			count[:error] += 1
		else
			d1 = all[m[:reply]][:date]
			d2 = m[:date]
			if d1 and d2
				csv << [d2.strftime('%s').to_i - d1.strftime('%s').to_i,d1.to_s,m[:to],m[:subject]]
				count[:reply] += 1
			else
				p [d1,d2]
				puts 'Date is missing.'
				count[:error] += 1
			end
		end
	}
end
p count