Email Checker

#!/usr/bin/python

import sys
import imaplib
import getpass
import email
import datetime
import random
import urllib2
import socket
import ssl
import re

password = open('password','r').read().strip()

urlregex = re.compile(ur'(?i)\b((?:https?://|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}/)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\))+(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:\'".,<>?\xab\xbb\u201c\u201d\u2018\u2019]))')

def process_mailbox(M,rate,linkrate):
  endrate = 0
  rv, data = M.search(None, "ALL")
  if rv != 'OK':
      print "No messages found!"
      return
  count = 0
  numEmails = len(data[0].split())
  for num in data[0].split()[::-1]:
      if random.random() < endrate:
          break
      if not random.random() < rate:
          continue

      count += 1

      rv, data = M.fetch(num, '(BODY[1])')
      if rv != 'OK':
          print "ERROR getting message", num
          return

      msg = email.message_from_string(data[0][1])
      msgstr = str(msg).decode('quopri').replace('\n',' ').replace('\r',' ')
      links = urlregex.findall(msgstr)

      visited = []
      for a in links:
        if random.random() < linkrate:
            try:
                site = urllib2.urlopen(a[0], timeout=1)
                if not 'html' in site.info().getheaders("Content-Type")[0]:
                    raise (ValueError('Not HTML'))
                html = site.read()
                visited.append(a[0])
            except (urllib2.HTTPError, ValueError, urllib2.URLError, socket.timeout, ssl.SSLError, IndexError):
                donothing = 0

      print 'MESSAGE: ' + msgstr[0:100]
      print 'LINKS: ' + str(len(links))
      print 'VISITED: ' + str(len(visited))

      rate -= 0.02
      endrate += 0.005
  return numEmails

MAIL_SERVERS = {'gmail': 'imap.gmail.com',
                'yahoo': 'imap.mail.yahoo.com',
                'aol': 'imap.aol.com',
                'hotmail' : 'imap-mail.outlook.com'}

emails = open('emails','r').read().strip().split('\n')
passwords = open('passwords','r').read().strip().split('\n')


for i in range(0,len(emails)):
    M = imaplib.IMAP4_SSL(MAIL_SERVERS['yahoo'], 993)
    try:
        M.login(emails[i], passwords[i])
    except imaplib.IMAP4.error:
        print "> LOGIN FAILED"

    rv, mailboxes = M.list()

    if rv != 'OK':
        print "> LOGIN FAILED"
        continue

    rv, data = M.select("INBOX")

    if rv == 'OK':
        print '> User: ' + str(i) + ' ' + emails[i]
        print '> Processing mailbox...'
        numEmails = process_mailbox(M,0.7,0.5)
        print '> Finished Processing'
        print '> ' + str(numEmails) + ' email' + ('s' if numEmails != 1 else '') + ' processed'
        print ''
        print ''
        M.close()
    else:
        print "> LOGIN FAILED"
        continue
    M.logout()