Split a Gmail MBOX file into individual mail files

mbox.py

#!/usr/bin/env python3
import os, re, sys, time, datetime
from dateutil.parser import parse
from dateutil.tz import tzutc

def seg2hms( value ):
	return str(datetime.timedelta(seconds=value))

def datasize( value ):
	idx, prefix = 0, ["", "K", "M", "G", "T"]
	while value>1024:
		idx, value = idx+1, value/1024
	return "{:.2f} {}B".format(value, prefix[idx])

def print_info( mbox, msize, current, processed, total, start ):
	percent  = 100*processed//total if total!=0 else 0
	ellapsed = int(time.time()) - start
	eta      = int(ellapsed*(total-processed)/processed) if processed!=0 else 0
	print("\033[1;1H" + " "*60)
	print("  Mailbox         {} ({:d})".format(mbox, msize).ljust(60, " "))
	print("  Current email   {}".format(datasize(current)).ljust(60, " "))
	print("  Ellapsed / ETA  {} / {}".format(seg2hms(ellapsed), seg2hms(eta)).ljust(60, " "))
	print("  Processed       ({}) {:d}% of {}".format(datasize(processed), percent, datasize(total)).ljust(60, " "))
	print((" "*60 + "\n")*2)

def split_mbox( mbox ):
	idx, CURRENT, PROCESSED = 0, 0, 0
	TOTAL = os.path.getsize(mbox + ".mbox")
	START = int(time.time())
	if not os.path.isdir( mbox ):
		os.mkdir(mbox)
	with open(mbox + ".mbox", "r") as f:
		for line in f:
			if line.startswith("From "):
				idx, CURRENT = idx+1, 0
				open(mbox + "/" + ("{:05d}".format(idx)), "w").write(line)
			else:
				open(mbox + "/" + ("{:05d}".format(idx)), "a").write(line)
			CURRENT   += len(line)
			PROCESSED += len(line)
			print_info( mbox, idx, CURRENT, PROCESSED, TOTAL, START )

os.system("clear")
if __name__!="__main__" or len(sys.argv)!=2:
	exit()

if not os.path.exists(sys.argv[1] + ".mbox"):
	print("'{}' Not found! :/".format(sys.argv[1]))
	exit()
split_mbox( sys.argv[1] )

requirements.txt

requests==2.12.2
python-dateutil==2.6.0