Simple log file poller for Tor HS logs.

monitor.py

# -*- coding: utf-8 -*-
"""Polls the Tor HS log file and records all INTRODUCTION
cells seen to a database.
"""
import re
import time
import datetime


def parse_introduce_from_log(log_line):
    match = re.match(
        '^(?P<timestamp>[A-Za-z0-9:. ]*) \[(?P<log_level>[a-z]*)\] Received ' +
        'INTRODUCE2 cell for service "(?P<service_id>[a-z2-7]*)" on circ ' +
        '(?P<circuit_id>\d*).$', log_line)
    if match:
        return match.groupdict()
    else:
        return None


def parse_log_timestamp(timestamp):
    timestamp = datetime.datetime.strptime(timestamp, '%b %d %H:%M:%S.%f')
    # No year in the log file, default to the current year.
    return timestamp.replace(year=datetime.datetime.now().year)


def main():
    import argparse
    from database import db_session
    from models import INTRODUCERequest
    from database import init_db
    init_db()

    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("-f", "--log-file", type=argparse.FileType('r'),
                        default='/var/log/tor/notices.log',
                        help="Log file for the Tor hidden service process "
                        "(default: %(default)s)")
    parser.add_argument("-w", "--wait", type=int, default=5,
                        help="How often to poll the log file for updates "
                        "(default: %(default)s seconds)")
    parser.add_argument("-v", "--verbose", action='store_true')
    args = parser.parse_args()

    log_file = args.log_file
    request_count = 0

    # Read to the end of the current log file
    log_file.read()

    # Begin poll the log file
    print 'Starting polling'
    try:
        while True:
            # Read in any new log entries line-by-line
            while True:
                log_line = log_file.readline()
                if not log_line:
                    break

                parsed_request = parse_introduce_from_log(log_line)
                if parsed_request:
                    # Got an INTRODUCE cell, store it in the DB.
                    logged_request = INTRODUCERequest(
                        service_id=parsed_request.get('service_id'),
                        timestamp=parse_log_timestamp(
                            parsed_request.get('timestamp'))
                        )
                    db_session.add(logged_request)
                    db_session.commit()

                    print '{}: Got INTRODUCE cell for service {}'.format(
                        logged_request.timestamp,
                        logged_request.service_id
                        )
                    request_count += 1
            time.sleep(args.wait)
    except KeyboardInterrupt:
        print '\nFinishing. Received {} INTRODUCE cells.'.format(request_count)

if __name__ == "__main__":
    main()