Python 2.x script for downloading, archiving and removing emails from an IMAP folder

download-archive-and-delete-from-imap-folder.py

"""
download, store and delete old messages from a folder on an IMAP account.

kind of crude in that you have to hard-code values that could be taken as
command-line options, and who knows what else is wrong with this as my
first Python effort of any consequence.
"""
# to do: make python3 compatible? process command line options instead of hard-coding

import imaplib, email, mailbox, re, os.path, logging, sys, time

log = logging
log.basicConfig(filename="mail-archive.log",format='%(asctime)s %(levelname)s: %(message)s',datefmt="%Y-%m-%d %I:%M:%S %p",level=logging.DEBUG)


# cutoff year. anything older than 01-Jan of 'until_year' gets downloaded and deleted.
until_year = "2017"

# IMAP connection parameters
hostname = 'your_imap_host'
user     = 'your_username'
password = 'your_password'

# name of folder to prune
folder_name = 'folder-to-prune'

# path to mbox files
mbox_path = "/path/to/your/email/archive"

# regexp for parsing year from email date header
pattern = re.compile(r'\b201[\d]\b')

log.info('starting. cutoff is '+until_year)

connection = imaplib.IMAP4(hostname)
connection.login(user, password)

type, data = connection.select(folder_name);
num_messages = data[0]

log.info("{} messages in the damn box".format(num_messages))

type, message_numbers_string = connection.search(None, 'ALL')
# print "{} is the response we got".format(type)

nums = message_numbers_string[0].split();

mbox = None
deleted = 0

# http://stackoverflow.com/questions/29432981/saving-imap-messages-with-python-mailbox-module

for n in nums:

    response = None
    try:
    	response, msg = connection.fetch(n,'(RFC822)')
    except Exception as e:
    	log.warning("caught exception fetching message {}, will try again".format(n))
    	for i in range(1,6):
    		# try again
    		log.info("sleeping for a few seconds")
    		print ("exception caught, trying again...")
    		time.sleep(5)
    		try:
    			response, msg = connection.fetch(n,'(RFC822)')
    			if (response == "OK"):
    				log.info("retry #{} worked, moving on".format(i))
    				break
    		except:
    			log.warning("shit failed again at retry#".format(i))
    			
    if not response:

    	print("shit is not working: {}".format(e.message))
    	print ("reluctantly moving on")

    # "Returned data are tuples of message part envelope and data"

    data = msg[0][1]
    for response_part in msg:

        if  isinstance(response_part, tuple):
            msg_obj = email.message_from_string(response_part[1])
            date_header = msg_obj['date']            

            year = pattern.search(date_header).group()

            sys.stdout.write('year: {}, processing: {} of {}\r'.format(year, n, num_messages, ))
            sys.stdout.flush()
            
            if year >= until_year:
                    # log.info("{} >= cutoff year {}, skipping".format(year, until_year))
                    continue
            path = mbox_path + "/" + year
            if not os.path.isfile(path) :
            	log.info("creating mbox: "+year)
                f = open(path,'w')
                f.close()
                
            # is there a mailbox instance?
            if not mbox:                
                mbox = mailbox.mbox(path)
                
            # is it the right one?
            if year not in mbox._path:
                # no, time for a new mbox instance
                # log.info("{} is not for {}, instantiating a new mbox".format(mbox._path,year))
                mbox.close()
                mbox = mailbox.mbox(path)
                    
            mbox.add(data)
            connection.store(n, '+FLAGS', '\\Deleted')
            deleted += 1

	
log.info("deleted {} messages".format(deleted))
log.info("expunging messages")
connection.expunge()
log.info("closing connection, logging out")
connection.close()
connection.logout()
log.info("done")
exit(0)