#!/usr/bin/python
import sys
import os
from collections import Counter
from datetime import datetime

##### start arg part
import argparse
parser = argparse.ArgumentParser(description='This does things to an Apache log file')
parser.add_argument('filename', help='is your filename correct?') # required=True is implied here
my_arg = parser.parse_args()
##### end arg part

##### initialize stuff
ip_list = [] # for the list of IPs
ms_list = list() # for the list of load times
cnt = int() # initialize counting iterations at 0...
http_code = list() # initialize list
ms_sum = long(0)
##### initialize stuff

with open(my_arg.filename) as fp:
    my_lines = fp.readline() # reads first line
    print("First date and time: %s" % my_lines.strip().split()[3].lstrip('[')) # this is the first date and time
    while my_lines: # continue reading until EOF...
        awk_like = my_lines.strip().split() # whitespace delimited list per line
        ip_list.append(awk_like[0]) # add field 0 to the list of visitors... - make this into a dict
        #ms_sum += long(awk_like[-1]) # this doesn't correct the problem at all
        if awk_like[-1].isdigit():
            ms_list.append(float(awk_like[-1])) # add field -1 to list of load times - make this a set
    if awk_like[9].startswith(('4','5'))
            http_code.append(awk_like[9])
        cnt += 1 # counting the iterations - not the best approach but it works
        my_lines = fp.readline() # point to the next line...
        if my_lines == "": # check for last line in a file - checking for EOF actually here
            fp.seek(-2, os.SEEK_END)
            while fp.read(1) != b"\n":
                fp.seek(-2, os.SEEK_CUR)
            last = fp.readline()
            print("Last date and time: %s" % last.strip().split()[3].lstrip('[')) # this is the first date and time

print(all(isinstance(x, float) for x in ms_list)) # this looks like it's printing all floats...wtf??!?
print("Number of lines parsed: %s" % cnt) # print number of lines parsed - works I guess?
print("Most common visitor: %s" % Counter(ip_list).most_common(1)[0][0]) # find most common in list
print("Total length of load list: %s" % len(ms_list))
#print("ms_sum is", ms_sum) # nada
avg_load_time = (sum(ms_list)-len(ms_list))/len(ms_list) # getting the average load time
print("Average load time: %sms" % avg_load_time) # printing the average load time
max_load_time = max(ms_list)-1 # this is sloppy - shouldn't have to subtact 1
min_load_time = min(ms_list)-1 # this is sloppy - shouldn't have to subtract 1
total_load_time = sum(ms_list)-len(ms_list) # this is sloppy - can't figure out why sum is wrong (everything is a float!)
print("Min load time: %sms" % min_load_time)
print("Max load time: %sms" % max_load_time) # this doesn't work correctly - it's already an int though
print("Total load time: %sms" % total_load_time)