fix_multipart_messages_with_16_bit_udh.rb

class Jobs::IncomingMessage
  class << self
    def encode(content)
      content = content.gsub(/\215./) { |match| GSM_ESCAPED_CHARACTERS[match[1]] }
      content = Iconv.conv("UTF-8", "HP-ROMAN8", content)
      content.gsub!(EURO_TOKEN, "\342\202\254")
      content
    end

    def decode(content)
      content = content.gsub("\342\202\254", EURO_TOKEN)
      content = Iconv.conv("HP-ROMAN8", "UTF-8", content)
      GSM_ESCAPED_CHARACTERS.invert.each do |(character, replacement)|
        content.gsub!(character, "\215#{replacement.chr}")
      end
      content
    end
  end
end

def convert_message_into_message_part(subscriber, message, message_id, part, total_parts, content)
  message_part = MessagePart.new(
    :subscriber_id => subscriber.id,
    :message_id => message_id,
    :part => part,
    :total_parts => total_parts,
    :content => content,
    :contact_msisdn => message.contact_msisdn,
    :timestamp => message.timestamp,
    :sent => message.sent
  )

  puts "#{'%5d' % message_id} - saving part #{part} of #{total_parts}"
  if message_part.save
    message.destroy
  else
    puts "*** Error saving MessagePart: " + message_part.errors.full_messages.join("\n")
  end
end

Subscriber.all.each do |subscriber|
  messages_with_long_udh = subscriber.messages.where(:content => %r{^(\330\210\320|\330\210\321|\330\210\322|\330\210\323)})
  messages_with_long_udh.each do |message|
    decoded_content = Iconv.conv("UTF-16BE", "UTF-8", message.content)
    udl, iei, iedl, message_id, total_parts, part = decoded_content.slice!(0..6).unpack("CCCSCC")
    content = Iconv.conv("UTF-8", "UTF-16BE", decoded_content)
    convert_message_into_message_part(subscriber, message, message_id, part, total_parts, content)
  end

  messages_with_long_udh = subscriber.messages.where(:content => %r{^\006\010\040})
  messages_with_long_udh.each do |message|
    decoded_content = Jobs::IncomingMessage.decode(message.content)
    udl, iei, iedl, message_id, total_parts, part = decoded_content.slice!(0..6).unpack("CCCSCC")
    content = Jobs::IncomingMessage.encode(decoded_content)
    convert_message_into_message_part(subscriber, message, message_id, part, total_parts, content)
  end
end