gistfile1.py

"""Cool histograms from your gmail account.

Remember to turn on IMAP, and if you're using two-steps verification
(you should) you'll have to request a password just for this (security
"tab" of your account configuration).
"""

from collections import Counter
from itertools import izip
import imaplib
import getpass
import matplotlib.pyplot as plt
from email.utils import mktime_tz, parsedate_tz
from datetime import datetime
import pytz


def _plotby_helper(iterable, xlim=None, width=0.7):
    x, y = izip(*Counter(iterable).iteritems())
    _plot_helper(x, y, xlim, width)

def _plot_helper(x, y, xlim=None, width=20):
    fig = plt.figure()
    ax = fig.add_subplot(111)
    if xlim is not None:
        ax.set_xlim(xlim)
    ax.bar(x, y, width=width)
    fig.autofmt_xdate()
    plt.show()

class EmailStats(object):
    def __init__(self, mailboxname, email, timezone="Europe/Madrid"):
        """timezone is the timezone you're usually in, so that the
        messages will appear at your right local time"""
        self.mailboxname = mailboxname
        self.email = email
        self.dtimes = None
        self.data = None
        self.imap = None
        self._connect()
        self._fetch_dates()
        self._disconnect()
        self.parse_dates(timezone)

    def _connect(self):
        self.imap = imaplib.IMAP4_SSL("imap.gmail.com")
        password = getpass.getpass("Enter your password: ")
        self.imap.login(self.email, password)

    def _disconnect(self):
        self.imap.logout()

    def _fetch_dates(self):
        self.imap.select(mailbox=self.mailboxname, readonly=True)
        print 'searching...'
        resp, items = self.imap.search(None, 'All')
        email_ids = ','.join(items[0].split())
        email_id_list = [int(ID) for ID in items[0].split()]
        print len(email_id_list), 'messages in ', self.mailboxname
        print 'fetching headers...'
        resp, self.data = self.imap.fetch(email_ids,
                                          "(BODY[HEADER.FIELDS (DATE)])")
        print str(len(self.data) / 2), 'headers fetched from', self.mailboxname

    def parse_dates(self, timezone):
        print 'reformatting dates...'
        baddates = [self.data[2*i][1][6:] for i in range(len(self.data) // 2)]
        print 'converting dates...'
        self.dtimes = [datetime.fromtimestamp(
            mktime_tz(parsedate_tz(i)),
            pytz.utc).astimezone(pytz.timezone(timezone)) for i in baddates]

    def plot_all(self):
        months, counts = izip(*((datetime(*(k + (1,))), v) for
                                k, v in Counter((i.year, i.month) for
                                                i in self.dtimes).iteritems()))
        _plot_helper(months, counts)

    def plotby_hours(self):
        _plotby_helper((i.hour for i in self.dtimes), [0, 24])

    def plotby_minutes(self):
        _plotby_helper((i.minute for i in self.dtimes), [0, 60])

    def plotby_weekday(self):
        _plotby_helper((i.weekday() for i in self.dtimes), [0, 7])

    
if __name__ == '__main__':
    # Or "[Gmail]/Sent" or labels "my_label" or...
    mailbox_name = '[Gmail]/All'
    email = '<your-email>@gmail.com'
    st = EmailStats(mailbox_name, email)
    st.plot_all()