import os,time def getip(filepath):     ip2city={}     file=open(filepath,'r')     lines=file.readlines()     file.close()     for line in lines:         ip=line.split(' ')[0]         city=line.split(' ')[1]         haship=hashm(ip)         if haship in ip2city:             pass         else:             ip2city[haship]=city     print('Hash done!')     return ip2city def hashm(ip):     iplist=ip.split('.')     ip=int(iplist[0])*4+int(iplist[1])*2+int(iplist[2])     return ip def getcityfromip(filepath,ipandcity):     outputstr=[]     for file in os.listdir(filepath):         file_handler=open(filepath+'\'+file,'r')         line=file_handler.readline()         while line:             ip=hashm(line.rstrip())             if ip in ipandcity:                 outputstr.append(line.rstrip()+'    '+ipandcity[ip])             line=file_handler.readline()         file_handler.close()         outfile_handler=open(filepath+'\'+file.split('.')[0]+'_out.txt','a+')         outfile_handler.writelines(outputstr)         outfile_handler.close()         print(file.split('.')[0]+'_out.txt'+'done!')         def splitfile(filepath):     file=open(filepath,'r')     block_size=8000000     filecount=1     temp=[]     count=0     line=file.readline()     while line or temp:         if count==block_size:             wfile=open('D:\ipfile\file_'+str(filecount)+'.txt','a+')             wfile.writelines(temp)             temp=[]             count=0             wfile.close()             filecount+=1             print('Split'+str(filecount)+' done!')         else:             count+=1             temp.append(line)             line=file.readline()     file.close()     return os.path.join('D:\'+'ipfile') if __name__ == '__main__':     start=time.clock()     filepath='D:\ip.txt'     ippath='D:\citys.txt'     ip2city=getip(ippath)     splitfilepath=splitfile(filepath)     getcityfromip('D:\'+'ipfile',ip2city)     end=time.clock()     print(end-start)

#Generate 100 millon ip import random import time def generateIpAdd(file,num):     ip=[]     file=open(file,'a+')     for i in range(num):         ipAdd='192.168.'+str(random.randint(0,255))+'.'+str(random.randint(0,255))         ip.append(ipAdd+'n')     file.writelines(ip)     file.close() if __name__=='__main__':     start=time.clock()     for i in range(10000):         generateIpAdd('D:ip.txt',10000)     end=time.clock()     print(end-start)