globalscape/python/ex_parser_slow.py

# TODO output to a file, logging to a file, better way to pass a file name into open

import time
import platform

full_list = False
results_value = 20
version = platform.architecture()
print(version[0])
start_time = time.time()

fname = "C:/Users/jbranan/Desktop/u_ex201209.log"

fileReadlist = []

with open(fname, encoding='utf8') as file_object:
    fileReadlist = file_object.readlines()

user_list = []
result_list = []

filename = fname.rsplit('/')
fn = filename[-1]

print(f'Looking for usernames in {fn}...')

for line in fileReadlist:
    if '#' in line:
        continue
    else:
        try:
            date, userinfo, otherdata = line.split(' -', 2)
            userinfo = userinfo.lstrip()
            user, userdata = userinfo.split(' [')
            user_list.append(user)
        except ValueError:
            continue

print("--- %s seconds ---" % (time.time() - start_time))

count_users_time = time.time()
user_cache = []
result_list_dict = {}

print('Building username appearance totals...')

for item in user_list:
    if item not in user_cache:
        user_cache.append(item)
        item_total = user_list.count(item)
        if item not in result_list:
            result_list_dict[item] = item_total

print("--- %s seconds ---" % (time.time() - count_users_time))
sort_time = time.time()

print('Sorting username appearance totals...')
results = sorted(result_list_dict.items(), key=lambda x: x[1], reverse=True)
print("--- %s seconds ---" % (time.time() - sort_time))

results_time = time.time()
print('Listing username appearance totals...')
if full_list:
    for k, v in results:
        print(f'User: {k} - Appearances:{v}')
else:
    for k, v in results[:results_value]:
        print(f'User: {k} - Appearances:{v}')

print("--- %s seconds ---" % (time.time() - results_time))