grab_time_interval.py

# Author: Chenfeng Nie
# levoncf@gmail.com

import os
import re
import glob


#path = glob.glob('/Users/levonnie/Desktop/CIP_Result/10vs10/output/*')
###########################
#  Malware:
#           #/Users/levonpro/Desktop/CIP/Output_Malware_Apps/*
#  Non-Malware:
#           #/Users/levonpro/Desktop/CIP/Output_Non_Malware_Apps/*
#########################

path = glob.glob('/Users/levonpro/Desktop/CIP/Output_Non_Malware_Apps/*')


for folder in path:
    name = folder + '/*.txt'
    txtfile = (glob.glob(name))
    #print(txtfile)
    for filename in sorted(txtfile):
        #print(filename)
                #open the "data.txt" (named data_file in a scope)

        with open(filename) as data_file:
            #read the text from the data_file into ping_data
            #ping_data = data_file.read()
            #found_data = re.findall('time=(.*)ms', ping_data)
            processed_data = []
            for ping_data in data_file:
                    #replace timeout with 500.000 ms
                if "Request timeout for icmp_seq" in ping_data:
                    processed_data.append('500.000')
                elif 'time=' in ping_data:
                    find_ping = re.findall('time=(.*)ms', ping_data)
                    processed_data.append(find_ping)
            #print(processed_data)
            data_file.close()
            # grab filename in each txt file
            i = filename.split('/')[-1].split('.txt')[0]
            #print(i)
            #print("the number is: " + i )
            with open(folder + "/Processed_%s.csv" % i, "w") as found_file:
                for pattern in processed_data:
                    string = str(pattern).replace("['", '')
                    new_string = str(string).replace("']", '')
                    #print(new_string)
                    found_file.write(new_string + '\n')
                    print("Writing : " + "Processed_%s.csv" % i + " in " + folder)