python 正则表达式分析nginx日志

有个需求要分析nginx日志，也懒得去研究logstach之类的开源工具，干脆直接写一个脚本，自己根据需求来实现：

先看日志格式：我们跟别人的不太一样，所以没办法了：

12.195.166.35 [10/May/2015:14:38:09 +0800] “list.xxxx.com” “GET /new/10:00/9.html?cat=0,0&sort=price_asc HTTP/1.0” 200 42164 “http://list.zhonghuasuan.com/new/10:00/8.html?cat=0,0&sort=price_asc” “Mozilla/5.0 (Linux; U; Android 4.4.2; zh-CN; H60-L02 Build/HDH60-L02) AppleWebKit/534.30 (KHTML, like Gecko) Version/4.0 UCBrowser/10.4.0.558 U3/0.8.0 Mobile Safari/534.30”

上面是我的日志格式：

脚本如下：

#!/usr/bin/env python#-*- coding:utf-8 –*-#Author:xiaoluo#QQ:942729042#date:2015:05:12import reimport syslog = sys.argv[1]ip = r"?P<ip>[/d.]*"date = r"?P<date>/d+"month = r"?P<month>/w+"year = r"?P<year>/d+"log_time = r"?P<time>/S+"timezone = r"""?P<timezone>                 [^/"]*         """name = r"""?P<name>/"            [^/"]*/"             """method = r"?P<method>/S+"request = r"?P<request>/S+"protocol = r"?P<protocol>/S+"status = r"?P<status>/d+"bodyBytesSent = r"?P<bodyBytesSent>/d+"refer = r"""?P<refer>/"             [^/"]*/"             """userAgent=r"""?P<userAgent>                .*               """#f = open('access1.log','r')#for logline in f.readlines():p = re.compile(r"(%s)/ /[(%s)/(%s)/(%s)/:(%s)/ (%s)/ (%s)/ (%s)/ (%s)/ (%s)/ (%s)/ (%s)/ (%s)/ (%s)" %(ip, date, month, year, log_time,timezone,name,method,request,protocol,status,bodyBytesSent,refer,userAgent), re.VERBOSE)def getcode():    codedic={}    f = open(log,'r')    for logline in f.readlines():         matchs = p.match(logline)         if matchs !=None:             allGroups =matchs.groups()             status= allGroups[10]             codedic[status]=codedic.get(status,0) +1    return codedic    f.close()def getIP():    f = open(log,'r')     IPdic={}    for logline in f.readlines():        matchs = p.match(logline)        if matchs !=None:            allGroups =matchs.groups()            IP=allGroups[0]             IPdic[IP] = IPdic.get(IP,0) +1    IPdic=sorted(IPdic.iteritems(),key=lambda c:c[1],reverse=True)    IPdic=IPdic[0:21:1]    return IPdic    f.close()def getURL():    f = open(log,'r')    URLdic={}    for logline in f.readlines():        matchs = p.match(logline)        if matchs !=None:            allGroups =matchs.groups()            urlname = allGroups[6]             URLdic[urlname] = URLdic.get(urlname,0) +1    URLdic=sorted(URLdic.iteritems(),key=lambda c:c[1],reverse=True)    URLdic=URLdic[0:21:1]               return URLdicdef getpv():    f = open(log,'r')    pvdic={}    for logline in f.readlines():        matchs = p.match(logline)        if matchs !=None:           allGroups =matchs.groups()           timezone=allGroups[4]           time = timezone.split(':')           minute = time[0]+":"+time[1]           pvdic[minute]=pvdic.get(minute,0) +1    pvdic=sorted(pvdic.iteritems(),key=lambda c:c[1],reverse=True)    pvdic=pvdic[0:21:1]    return pvdicif __name__=='__main__':    print "网站监控状况检查状态码"    print getcode()     print "网站访问量最高的20个IP地址"    print getIP()    print "网站访问最多的20个站点名"    print getURL()    print getpv()

这里要指出的是。我当初是给正则匹配的时候单独封装一个函数的，这样就省去了下面每个函数要打开之前都要单独打开一遍文件，但是我return的时候只能用列表的形式返回，结果列表太大把我的内存耗光了，我的是32G的内存，15G的日志。

效果：