emailparser/parse

#!/usr/bin/python

import os, argparse, hashlib, email, email.header, email.utils, glob, sqlite3, time

parser = argparse.ArgumentParser(description='Parses emails into an SQLite database and an attachment folder.')

parser.add_argument('-p', dest='pattern', action='store', default='*',
                   help='glob pattern (including path) that has to be matched for a file to be parsed')
                   
parser.add_argument('-d', dest='database', action='store', default='emails.db',
                   help='path of the database that should be used to store the emails (will be created if it does not exist yet)')
                   
parser.add_argument('-a', dest='attachment_dir', action='store', default='attachments',
                   help='path where attachments should be stored (will be created if it does not exist yet)')
                   
parser.add_argument('-f', '--forced', dest='forced', action='store_true',
                   help='force insertion into database, even if entries already exist')

args = parser.parse_args()
options = vars(args)

def getheader(header_text, default="ascii"):
	headers = email.header.decode_header(header_text)
	try:
		header_sections = [unicode(text, charset or default) for text, charset in headers]
	except UnicodeDecodeError:
		header_sections = [unicode(text, 'windows-1252') for text, charset in headers]
	except LookupError:
		return u""
	
	return u"".join(header_sections)
	
def find_submessages(message):
	if message.is_multipart():
		return [find_submessages(part) for part in message.get_payload()]
	else:
		return message

def flatten(x):
	# http://kogs-www.informatik.uni-hamburg.de/~meine/python_tricks
	result = []
	for el in x:
		if hasattr(el, "__iter__") and not isinstance(el, basestring):
			result.extend(flatten(el))
		else:
			result.append(el)
	return result

def get_charset(part):
	charset = None
	
	if part.get_content_charset():
		charset = part.get_content_charset()
	elif part.get_charset():
		charset = part.get_charset()
	
	if charset is None or charset == "default" or charset.startswith("us-ascii"):
		return "ascii"
	else:
		return charset

# Connect to database
database = sqlite3.connect(options['database'])
cursor = database.cursor()

try:
	# Create attachment directory first
	os.makedirs(options['attachment_dir'])
except OSError:
	pass

try:
	# Try to create emails table
	cursor.execute("CREATE TABLE emails (`message_id`, `from`, `to`, `subject`, `date`, `body`, `html`, `hash`)")
except sqlite3.OperationalError:
	pass
	
try:
	# Try to create attachments table
	cursor.execute("CREATE TABLE attachments (`message_hash`, `filename`, `type`, `hash`, `size`)")
except sqlite3.OperationalError:
	pass

# Select all files matching the given pattern
file_list = glob.glob(options['pattern'])

finished = 0

for email_file in file_list:
	# To save time when updating a database, let's first check whether the filename is already present in the database as a hash.
	# There is no need to check for the format here, since if the filename is not a valid hash, it simply won't match anything.
	sha1_hash = os.path.splitext(os.path.split(email_file)[1])[0]
	
	if options['forced'] == False:
		cursor.execute("SELECT * FROM emails WHERE `Hash` = ?", (sha1_hash,))
		if len(cursor.fetchall()) > 0:
			print "Skipping %s, already exists in the database according to filename." % sha1_hash
			continue
	
	message = email.message_from_file(open(email_file, 'r'))
	
	if message['message-id'] is None:
		print "%s is not a valid e-mail file." % email_file
	else:
		if 'subject' not in message or message['subject'] is None:
			subject = ""
		else:
			subject = message['subject']
		
		textbody = ""
		htmlbody = ""
		attachment_list = []
		
		sha1_hash = hashlib.sha1("%s/%s/%s/%s" % (message['from'], message['to'], message['message-id'], subject)).hexdigest()
		
		if options['forced'] == False:
			cursor.execute("SELECT * FROM emails WHERE `Hash` = ?", (sha1_hash,))
			if len(cursor.fetchall()) > 0:
				print "Skipping %s, already exists in the database according to message hash." % sha1_hash
				continue
			
		message_parts = [find_submessages(message)]
		message_parts = flatten(message_parts)
		
		for part in message_parts:
			if part.get_filename() is None:
				# Part of the message
				if part.get_content_type() == "text/plain":
					if textbody == "":
						try:
							textbody = part.get_payload(decode=True).decode(get_charset(part))
						except UnicodeDecodeError:
							# This part is probably in windows-1252 encoding
							try:
								textbody = part.get_payload(decode=True).decode('windows-1252')
							except UnicodeDecodeError:
								# Ok, we really have no clue how to decode this, we'll just skip it...
								continue
						except LookupError:
							pass
				elif part.get_content_type() == "text/html" or part.get_content_type == "text/xhtml+xml" or part.get_content_type == "application/xhtml+xml":
					if htmlbody == "":
						try:
							htmlbody = part.get_payload(decode=True).decode(get_charset(part))
						except UnicodeDecodeError:
							# This part is probably in windows-1252 encoding
							try:
								htmlbody = part.get_payload(decode=True).decode('windows-1252')
							except UnicodeDecodeError:
								# Ok, we really have no clue how to decode this, we'll just skip it...
								continue
						except LookupError:
							pass
				else:
					# Technically this is supposed to be part of the message body, but we have no idea what format it is in...
					print "WARNING: Unknown message format encountered in %s, message may be incomplete." % sha1_hash
			else:
				# Attachment
				attachment_data = part.get_payload(decode=True)
				attachment_sha1 = hashlib.sha1(attachment_data).hexdigest()
				attachment_filename = part.get_filename()
				attachment_type = part.get_content_type()
				attachment_extension = os.path.splitext(attachment_filename)[1]
				attachment_size = len(attachment_data)
				attachment_destination = "%s/%s%s" % (options['attachment_dir'], attachment_sha1, attachment_extension)
				attachment_list.append((attachment_filename, attachment_type, attachment_sha1, attachment_size))
				
				attachment_file = open(attachment_destination, "w")
				attachment_file.write(attachment_data)
				attachment_file.close()
		
		try:
			timestamp = int(time.mktime(email.utils.parsedate(message['date'])))
		except TypeError:
			timestamp = 0
			print "WARNING: Failed to determine unix timestamp for %s." % sha1_hash
	
		new_row = (getheader(message['message-id']), getheader(message['from']), getheader(message['to']), getheader(subject), timestamp, textbody, htmlbody, sha1_hash)
		cursor.execute("INSERT INTO emails VALUES (?, ?, ?, ?, ?, ?, ?, ?)", new_row)
		print "Successfully parsed and inserted e-mail with SHA1 hash %s." % sha1_hash
		
		if len(attachment_list) > 0:
			inserted = 0
			
			for attachment in attachment_list:
				if options['forced'] == False:
					cursor.execute("SELECT * FROM attachments WHERE `hash` = ? AND `message_hash` = ?", (attachment[2], sha1_hash))
					if len(cursor.fetchall()) > 0:
						print "Skipping attachment %s, already exists in the database." % attachment[2]
						continue
					
				new_row = (sha1_hash, attachment[0], attachment[1], attachment[2], attachment[3])
				cursor.execute("INSERT INTO attachments VALUES (?, ?, ?, ?, ?)", new_row)
				inserted += 1
			
			if inserted > 0:
				print "Successfully inserted %d attachment(s) for %s." % (inserted, sha1_hash)
				
	finished += 1
	
	if finished % 100 == 0:
		database.commit()
		print "%d e-mails done, commited changes to database." % finished

database.commit()
print "Changes successfully committed to database, all done."