DRTpv

import csv
import datetime
import time

num_per_day = 1000
num_per_minute = 5

all_ips = []
all_times = []
potential_robots_ips_1 = []
all_potential_robots_1 = {}
potential_robots_ips_2 = []

file_loc = input("Enter file directory: ")
file_name = input("Enter file name: ")
status = input("Would you like to know the status every 100000 rows? (slightly slower)(y/n) ")

t1 = time.time()
rows = 0
with open(file_loc+"/"+file_name) as csvfile:
    file = csv.reader(csvfile)
    for row in file:
        if row[0] != "ip":
            time_split = row[2].split(":")
            total_seconds = datetime.timedelta(hours = int(time_split[0]), minutes = int(time_split[1]), seconds = int(time_split[2])).total_seconds()
            all_ips.append(row[0])
            all_times.append(total_seconds)

            if row[0] in potential_robots_ips_1 or all_ips.count(row[0]) > num_per_day:
                if row[0] not in potential_robots_ips_1: #if it's not in potential_robots_ips_1, add current ip to all_potential_robots_1 and reloop to find the other 99 it missed
                    potential_robots_ips_1.append(row[0])
                    all_potential_robots_1.update({row[0]: [total_seconds]})
                    num = 0
                    for i in range(len(all_ips)-2, -1, -1):
                        if all_ips[i] == row[0]:
                            all_potential_robots_1[row[0]].insert(0, all_times[i])
                            num += 1
                        if num >= 99:
                            break
                else: #else, add the time to the dictionary
                    all_potential_robots_1[row[0]].append(total_seconds)
            if status == "y":
                rows += 1
                if rows%100000 == 0:
                    print(rows, "rows processed")
#check the more than 5 per minute condition for each ip address
for k, v in all_potential_robots_1.items():
    for i in range(len(v)-(num_per_minute+1)):
        if v[i+num_per_minute+1] - v[i] <= 60:
            potential_robots_ips_2.append(k)
            break
t2 = time.time()
print("DRTpv robots:", potential_robots_ips_1)
print("There are",len(potential_robots_ips_1),"DRTpv robots in",file_name)
print("Finished in",t2-t1,"seconds")