#!/usr/bin/python

import email
import gzip
import mailbox
import md5
import mimetypes
import os 
from datetime import datetime
from email.header import decode_header
from glob import glob

skip = ['multipart/alternative', 'multipart/related', 'multipart/mixed',
        'message/delivery-status', 'text/plain', 'text/html']

def decode(header):
  return decode_header(header)[0][0]

def detach(msg):
  # quick exit if we have seen this entry before
  id = md5.new(msg['message-id']).hexdigest()
  if os.path.exists(os.path.join('tally',id)): return

  # collect eligible attachments
  attachments = []
  for payload in msg.get_payload():
    if payload.get_content_type() in skip: continue
    if payload.get_content_type() == 'image/gif':
      if len(payload.get_payload(decode=True))<10240: continue
    attachments.append(payload)
  if len(attachments) == 0: return

  # determine output file name prefix
  prefix = ''
  if len(attachments) > 1:
    dirname = datetime(*email.utils.parsedate(msg['date'])[:7]).isoformat()
    prefix = dirname.replace(':','_').replace('-','_') + os.sep
    received = os.path.join('received',prefix)
    if not os.path.exists(received): os.mkdir(received)
    os.system('svn add ' + received)
  elif len(attachments) == 1:
    name=decode(attachments[0].get_filename())
    if not name: return
    if len(name)<16:
      prefix = decode(msg['from'])
      if prefix.startswith('"eFax"'):
        prefix = 'eFax'
      else:
        if prefix.find('<')>=0: prefix = prefix.split('<')[1]
        prefix = prefix.split('@')[0]
      prefix = prefix + '-'

  # determine commit message
  summary = "\n".join([
    'Subject: ' + decode(msg['subject']),
    'From: ' + decode(msg['from']),
    'Date: ' + msg['date'],
    'Message-Id: ' + msg['message-id'],
  ])

  # decode payloads and place add to svn
  for attachment in attachments:
    mime = attachment.get_content_type()
    if mime == 'application/octet-stream':
      mime = mimetypes.guess_type(decode(attachment.get_filename()))[0]
    name=decode(attachment.get_filename()).replace(' ','-')

    content = attachment.get_payload(decode=True)
    if content:
      file=os.path.join('received',prefix+name)
      fh=open(file,'w')
      fh.write(content)
      fh.close()

      os.system('svn add ' + file)
      os.system('svn propset svn:mime-type ' + mime + ' ' + file)

  if len(attachments)>1: file = os.path.join('received',prefix)

  tally = os.path.join('tally',id)
  fh=open(tally,'w')
  fh.write(summary + "\n")
  fh.close()
  os.system('svn commit --file ' + tally + ' ' + file)

if __name__ == "__main__":
  # determine starting point
  os.chdir(os.path.join(os.environ['HOME'], 'apmail'))
  latest = max([os.stat(file).st_mtime for file in glob('secretary/20*')])

  # download updates
  os.system('...elided...')

  # process updated mbox files
  for file in glob('secretary/20*'):
    if latest >= os.stat(file).st_mtime: continue

    # open gzipped/raw file
    if file.endswith('.gz'):
      fh=gzip.open(file)
    else:
      fh=open(file)

    # process each multipart message in the mailbox 
    for msg in iter(mailbox.UnixMailbox(fh, email.message_from_file)):
      if msg.is_multipart():
        detach(msg)