#!/usr/bin/env ruby

ARCHIVE_DIR = "/var/lib/mailman/archives/private/"
NEW_MESSAGE_IDS_FILE = "new_message_ids.yml"


require 'mbox'
require 'fileutils'
require 'digest/sha1'
require 'yaml'


# keep the original content for metadata and content
# (quick and ugly way)
class Mbox
class Mail
def self.parse (input, options = {})
	metadata = ''
	headers  = Mbox::Mail::Headers.new
	content  = ''

	next until input.eof? || (line = input.readline).match(options[:separator])

	return if !line || line.empty?

	# metadata parsing
	metadata << line
	until input.eof? || (line = input.readline).match(options[:separator])
		break unless line.match(/^>+/)

		metadata << line
	end
	metadata.chomp!

	# headers parsing
	current = ''
	begin
		break if line.strip.empty?

		current << line
	end until input.eof? || (line = input.readline).match(options[:separator])
	headers.parse(current)

	# content parsing
	current = ''
	until input.eof? || (line = input.readline).match(options[:separator])
		next if options[:headers_only]

		current << line
	end

	unless options[:headers_only]
		content = current.chomp
	end

	# put the last separator back in its place
	if !input.eof? && line
		input.seek(-line.length, IO::SEEK_CUR)
	end

	Mail.new(metadata, headers, content)
end
end
end

def fix_mbox(mbox, new_message_ids)
  filename = File.join(ARCHIVE_DIR, "#{mbox}.mbox", "#{mbox}.mbox")
  return unless File.exists? filename

  puts "Opening mbox '#{mbox}'"

  fixed_needed = false

  fixed_filename = filename + ".fixed"
  backup_filename = filename + ".bak"

  # only work with the original file
  if File.exists? backup_filename
    FileUtils.rm(filename)
    FileUtils.mv(backup_filename, filename)
  end

  File.open(fixed_filename, "w") do |fixed_mbox|
    Mbox.open(filename).each do |mail|
      mail_raw = mail.metadata.to_s + "\n" + mail.headers.to_s + "\n" + mail.content.to_s + "\n"
      if mail.headers['Message-Id'].nil?
        hash = Digest::SHA1.hexdigest(mail_raw)
        n_m_i = new_message_ids[hash] || generate_message_id(mail)
        mail.headers['Message-Id'] = n_m_i
        # recalculate mail_raw after change
        mail_raw = mail.metadata.to_s + "\n" + mail.headers.to_s + "\n" + mail.content.to_s + "\n"
        new_message_ids[hash] = n_m_i
        fixed_needed = true
      end
      fixed_mbox << mail_raw
    end
  end

  if fixed_needed
    # replace original file for import but keep original safe aside
    FileUtils.mv(filename, backup_filename)
    FileUtils.mv(fixed_filename, filename)
    puts "  FIXED"
  else
    # useless, no fix was done
    FileUtils.rm(fixed_filename)
  end
end

# inspired from ruby-mail library
def generate_message_id(mail)
  t = Time.now
  random_tag = sprintf('%x%x_%x%x%d%x',
    t.to_i, t.tv_usec,
    $$, Thread.current.object_id.abs, mail.object_id, rand(255))
  "<#{random_tag}@fixed.lists.ovirt.org.mail>"
end


new_message_ids = {}
if File.exists? NEW_MESSAGE_IDS_FILE
  new_message_ids = YAML.load_file(NEW_MESSAGE_IDS_FILE)
end

Dir.new(ARCHIVE_DIR).each do |dir|
  fix_mbox($1, new_message_ids) if dir =~ /^(.*)\.mbox$/
end

puts "new_message_ids: #{new_message_ids.size}"

File.open(NEW_MESSAGE_IDS_FILE, "w") do |file|
  file.write new_message_ids.to_yaml
end