python datahanle.py

#!/usr/bin/env python
# -*- conding:utf-8 -*-
import os,redata =[]
asp = []
aspx = []
php = []
jsp = []
mdb = []
dirx = []def file_name():for files in os.walk("data"):#print(files[2]) #当前路径下文件passreturn files[2]
def datas(dir):with open("%s"%dir,"r",encoding="gb18030") as f:for i in f.readlines():data.append(i)
dirs =file_name()
for i in dirs:datas("data/%s"%(i))
print("一共有%s条路径"%(len(data)))
data = list(set(data))
print("去重后一共还有%s条路径"%(len(data)))count_asp = 0
count_aspx = 0
count_jsp = 0
count_php = 0
count_dirx = 0
count_mdb = 0data = sorted(data) #对列表排序for i in data:#rule = re.compile(r".*?asp",re.I)res = re.search('.*?\.asp',i,flags=re.I)if res == None:res = re.search('.*?\.mdb', i, flags=re.I)if res == None:res = re.search('.*?\.php', i, flags=re.I)if res == None:res = re.search('.*?\.jsp', i, flags=re.I)if res == None:dirx.append(i)count_dirx +=1else:jsp.append(i)count_jsp +=1else:php.append(i)count_php +=1else:mdb.append(i)count_mdb += 1else:res = re.search('.*?\.aspx', i, flags=re.I)if res == None:asp.append(i)count_asp+=1else:aspx.append(i)count_aspx +=1print("asp:%s aspx:%s php:%s jsp:%s dir:%s mdb:%s"%(count_asp,count_aspx,count_php,count_jsp,count_dirx,count_mdb))
with open('nwe_asp.txt','a',encoding='utf-8') as f:for i in asp:f.write(i)with open('nwe_aspx.txt','a',encoding='utf-8') as f:for i in aspx:f.write(i)with open('nwe_php.txt','a',encoding='utf-8') as f:for i in php:f.write(i)with open('nwe_jsp.txt','a',encoding='utf-8') as f:for i in jsp:f.write(i)with open('nwe_mdb.txt','a',encoding='utf-8') as f:for i in mdb:f.write(i)with open('nwe_dir.txt','a',encoding='utf-8') as f:for i in dirx:f.write(i)