python 实现nginx/apache 日志格式的统计脚本
# !/usr/bin/env python# -*- coding:utf8 -*-
import sys
import re
import time
import os
def main():
months ={
"jan": 1, "Feb": 2, "Mar": 3, "Apr": 4, "May": 5, "Jun": 6,
"Jul": 07, "Aug": 8, "Sep": 9, "Oct": 10, "Nov": 11, "Dec": 12}
error = '请输入正确的时间格式,例如 2016-08-03 10:00:00'
start = raw_input("请输入要查询的开始时间段如:2016-08-03 10:00:00")
#start = '2016-08-03 10:37:00'
pattern = re.compile("^201-(0|1)-(|3)\s(|2)::$")
i_start = re.match(pattern, start)
if i_start == None:
print error
sys.exit(0)
end = raw_input("请输入要查询的结束时间段如:2016-08-03 10:01:00")
#end = '2016-08-03 10:38:00'
i_end = re.match(pattern, end)
if i_end == None:
print error
sys.exit(0)
print "time format is true !"
## 获取格式的时间的定时间戳 ##
start_stamp = time.mktime(time.strptime(start, "%Y-%m-%d %H:%M:%S"))
end_stamp = time.mktime(time.strptime(end, "%Y-%m-%d %H:%M:%S"))
## 处理文件 ###
log_abpath = raw_input("请输入log文件的绝对路径 例如:/var/log/apache2/access.log")
filepath = os.path.exists(log_abpath)
if not filepath:
raise "对不起该目录文件不存在！！"
sys.exit(0)
print "文件存在！！"
f = open(log_abpath, 'r')
#f = open("/var/log/apache2/access.log", "r")
logs = f.readlines()
position = 0
t_logs = []
for log in logs:
position += 1
log = log.split(" ")
log.insert(0, position)
t_logs.append(log)
for t_log in t_logs:
#print type(t_log)
tim = str(t_log)
mon = tim
#print mon
#print tim
#print type(tim)
#print months
tim_f = tim.replace(mon, str(months))
#print tim_f
d = time.strptime(tim_f, "%d/%m/%Y:%H:%M:%S")
stamp = time.mktime(d)
t_log.insert(1, stamp)
# print t_logs
count = 0
for t_log in t_logs:
if t_log > start_stamp < end_stamp:
print "该时间段内的请求日志出现在,"+str(t_log)+"行"
count += 1
if count != 0:
print "该段时间内一共出现了%d次请求"%(count)
else:
print "对不起，我已经努力帮你找了,可是找到该段时间的日志信息"
if __name__ == '__main__':
main()
页:
[1]