http_log.py

from argparse import ArgumentParser
from collections import defaultdict


if __name__ == '__main__':
    parser = ArgumentParser()
    parser.add_argument('input', type=str)

    args = parser.parse_args()

    count = 0
    ip_count = defaultdict(int)
    http_code_count = defaultdict(int)
    total_time_spent = 0
    with open(args.input, mode='r') as f:
        entry = f.readline()
        while entry:
            count += 1
            entry = entry.replace('\n', '')
            parts = entry.split()
            ip = parts[0]
            http_code = parts[8]
            response_time = int(parts[-1])
            ip_count[ip] = ip_count[ip] + 1
            http_code_count[http_code] = http_code_count[http_code] + 1
            total_time_spent += response_time
            # print(ip, http_code, response_time)
            entry = f.readline()

    print('number of entries', count)

    print('ip address', 'count')
    for ip, count in ip_count.items():
        print(ip, count)

    print('http code', 'count')
    for http_code, count in http_code_count.items():
        print(http_code, count)

    print('total time spent', total_time_spent)