Script to search through s3 for missing company, campaign, and/or email data

check_missing_emails.py

import argparse
import csv
import fnmatch
import multiprocessing
import os
import re
import sys


def search_logfiles_for_email(email):
    logfiles = get_email_sending_logfiles(args.logfile_path)
    print email
    for file_ in logfiles:
        f = open(file_, "r")
        for line in f:
            if re.search(email, line):
                print "***************************"
                print "match for: {}".format(email)
                print "***************************"
                print line


def get_emails_from_csv(csv_file):
    emails = []
    with open(csv_file, 'rb') as f:
        reader = csv.reader(f)
        for row in reader:
            emails.append(row[1])
    # First entry returned is just "Email"
    # Removing it to remove a lot of false positives
    emails.append('buehler315@gmail.com')
    return emails[1:]


def get_email_sending_logfiles(filepath, filetype='syslog'):
    matches = []
    for root, dirnames, filenames in os.walk(filepath):
       for filename in fnmatch.filter(filenames, filetype):
           matches.append(os.path.join(root, filename))
    return matches


def search_applogs_for_ids(campaign_id, company_id=None):
    logfiles = get_email_sending_logfiles(args.logfile_path, filetype='app.log')
    for file_ in logfiles:
        f = open(file_, "r")
        for line in f:
            if re.search(campaign_id, line):
                print "***************************"
                print "match for: {}".format(campaign_id)
                print "***************************"
                print line
                continue
            if company_id:
                if re.search(company_id, line):
                    print "***************************"
                    print "match for: {}".format(company_id)
                    print line


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument('--csv', dest="csv_file", required=True, help='CSV containing email data')
    parser.add_argument('--logfiles', dest="logfile_path", required=True, help='Log data location')
    parser.add_argument('--campaign-id', dest="campaign_id", help='The ID of the missing campaign')
    parser.add_argument('--company-id', dest="company_id", help='The ID of the company with the missing campaign')
    global args
    args = parser.parse_args()

    if args.campaign_id:
        if args.company_id:
            search_applogs_for_ids(args.campaign_id, args.company_id) 
        else:
            search_applogs_for_ids(args.campaign_id)
        sys.exit(0)

    emails = get_emails_from_csv(args.csv_file)

    print "Checking logfiles for email matches"
    cpu_count = multiprocessing.cpu_count()
    pool = multiprocessing.Pool(cpu_count)
    pool.map(search_logfiles_for_email, emails)

    pool.close()
    pool.join()


if __name__ == "__main__":
    main()