# -*- coding: UTF-8 -*-
import os,sys
import gzip
#找出路徑下access_log.gz文件
files=[]
fileDir=sys.argv[1]
listFile = os.listdir(fileDir)
for line in listFile:
filepath = os.path.join(fileDir,line)
if os.path.isdir(filepath):
continue;#如果filepath是目錄
if (filepath.find("access_log")<0 or filepath.find(".gz")<0):
continue;
files.append(filepath)
#可能有影響的房型id列表
roomids=[]
f=open("roomIds.txt",'r')
for line in f:
roomids.append(line.strip())
f.close()
print roomids
for filename in files:
#計數器
count=0
#解析accesslog
readfile = gzip.GzipFile(filename)
for line in readfile:
if(line.find("http://www.****.com/order/user/orderEdit.jsp?")<0 or (line.find("MSIE 8.0")<0 and line.find("MSIE 9.0")<0)):
continue;
roomId=line[line.find("&roomId=")+8:line.find("&partial")]
if(roomId in roomids):
count=count+1
print line
readfile.close()
print "受影響的*****訪問次數爲",count