#!/usr/bin/env python
# Given a maildrop log file, generate stats on how many messages it filed per
# day in folders matching regexps.

import sys, os, re, time, getopt

groups = [
	("spam", re.compile(r'[/\.]spam(|-learn)$')),
	("virus", re.compile(r'[/\.]virus$')),
	("other", None)
	]

def analyse():
	f = open(os.getenv("HOME") + "/.maildroplog")
	date = None
	days = {}
	for l in f.xreadlines():
		l = l.strip()
		if l == "":
			continue
		fs = l.split(": ", 1)
		if fs[0] == "Date":
			date = time.strptime(fs[1])
			date = (date[0], date[1], date[2], 0, 0, 0, date[6], date[7], 0)
		elif fs[0] == "File":
			fn = fs[1].split()[0]
			i = 0
			for (g, r) in groups:
				if r is None or r.search(fn) is not None:
					dh = days.setdefault(date, [0] * len(groups))
					dh[i] += 1
					break
				i += 1
	f.close()

	daylist = days.keys()
	daylist.sort()
	f = open("maildropstats.data", "w")
	for day in daylist[1:-1]:
		counts = days[day]
		f.write(time.strftime("%Y-%m-%d", day) + " " + " ".join(map(str, counts)) + "\n")
	f.close()

def plot(outputfile, format):
	if format == "png":
		terminal = "png truecolor size 800x600 small"
	else:
		terminal = "postscript color size 29.7cm,21cm"

	gp = os.popen("gnuplot", "w")
	gp.write("""
set terminal """ + terminal + """
set output '""" + outputfile + """'
set title 'Mail counts'
set xlabel 'Date'
set ylabel 'Number of messages per day'
set xdata time
set timefmt '%Y-%m-%d'
set format x '%Y-%m'
set data style lines
set key top left
plot """)
	i = 1
	sep = ""
	for (g, r) in groups:
		i += 1
		gp.write(sep + "'maildropstats.data' using 1:" + str(i) + " title '" + g + "'")
		sep = ", "
	gp.write("\n")
	gp.close()

def usage():
	print """maildropstats by Adam Sampson
Usage: maildropstats [OPTION...]

-r, --replot      Don't reanalyse; just plot with the existing data
-p, --png         Produce PNG output (default Postscript)
-o FILE           Write output file FILE (default maildropstats.ps/png)
-h, --help        Show usage

Report bugs to <ats@offog.org>."""

def main(args):
	try:
		opts, args = getopt.getopt(args, "rpho:", ["replot", "png", "help"])
	except getopt.GetoptError:
		usage()
		sys.exit(1)

	replot = False
	format = "ps"
	outputfile = None
	for (o, a) in opts:
		if o in ("-h", "--help"):
			usage()
			sys.exit(0)
		elif o in ("-r", "--replot"):
			replot = True
		elif o in ("-p", "--png"):
			format = "png"
		elif o == "-o":
			outputfile = a
	if outputfile is None:
		outputfile = "maildropstats." + format

	if not replot:
		analyse()
	plot(outputfile, format)

if __name__ == "__main__":
	main(sys.argv[1:])