Extract email sent times from a mailbox

date_extract.py

#!/usr/bin/env python

import collections, glob, os, re
from dateutil.parser import *

sentmail_mbox = glob.glob(os.path.expanduser('~/.thunderbird/*/ImapMail/*/*/Sent Mail'))
lines = []
for mbox in sentmail_mbox:
  lines.extend(open(mbox).read().split('\n'))

c = re.compile('^Date: (.*)00')
dates = [ c.search(l).group(1) for l in lines if c.search(l) ]
pdates = [parse(d) for d in dates]

keys = [(p.year,) for p in pdates]
kcounts = collections.defaultdict(int)
for k in keys:
  kcounts[k] += 1

mapping = collections.defaultdict(int)
for k, p in zip(keys, pdates):
  mapping[tuple(list(k) + [p.hour])] += 1.0 / kcounts[k]

for k, v in sorted(mapping.items()):
  print ",".join([str(kk) for kk in k]), ',', v