前天开发的同事让我帮忙分析下 nginx访问日志,他只要访问日志里面的4个值 ，我看了下nginx的日志格式，下面是其中一段

124.227.66.162 - - [25/Jan/2010:13:42:07 +0800] "POST /design/game.php HTTP/1.1" "uid=355288&cuid=355287×tamp=1264484517&check=68230e418e28a9d05b8cf1e2f7cbf392&action=plantInfo" 200 1019 "http://www.ime.com/design/flash/main.swf?v=439/`DYNAMIC`/1" "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)" -
124.240.39.49 - - [25/Jan/2010:13:42:07 +0800] "POST /design/game.php HTTP/1.1" "cid=2&lid=4&oid=2&action=researchLayer&cuid=496990×tamp=1264398138&check=b50cd4ade18c0797df24cb1a8828ae18" 200 219 "http://www.ime.com/design/flash/main.swf?v=439/`DYNAMIC`/1" "Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.0; SLCC1; .NET CLR 2.0.50727; .NET CLR 3.5.21022; .NET CLR 3.5.30729; .NET CLR 3.0.30618)" -
121.236.118.126 - - [25/Jan/2010:13:42:07 +0800] "POST /design/game.php HTTP/1.1" "check=8ec1521fc3df9c03d83af9a4d933dbb0&cuid=509590×tamp=1264398703&oid=2&action=oreInfo" 200 261 "http://www.ime.com/design/flash/main.swf?v=439/`DYNAMIC`/1" "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; QQDownload 1.7; .NET CLR 2.0.50727)" -

同事让我帮忙取 IP地址 时间 还有 cuid= 和 action= 的值

看上去好乱，但是还是有规律的，里面好多行没有 action 和cuid，我先把他过滤掉

awk '/action/{print $0}' access.log > action.log

因为 如果有action 就肯定会有cuid 所以只过滤一个action就好了
现在的所有行都有 cuid 和 action了
好了，我再来改一改格式，让他看起来更清晰一些

awk -F "[ '&''[']" '{print $1"\t"$5"\t"$10"\t"$11"\t"$12"\t"$13"\t"$14"\t"$15}' action.log > newlog
这样比较麻烦，不过确实能让他更清晰一些，下面是得到的结果

117.83.131.36    25/Jan/2010:14:31:34    "uid=438824    cuid=511252    timestamp=1264401079    check=fbb9ad922f01888e6c0757d117bf304e    action=plantInfo"    200
221.9.32.181    25/Jan/2010:14:31:34    "cuid=506517    action=plantInfo    timestamp=1264401075    check=01661377f346538eba790e856dd3713a    uid=539860"    200
221.178.128.146    25/Jan/2010:14:31:34    "timestamp=1264401105    check=7d5e41feeb3ae0482e1fe990f27ddc67    cuid=303367    display=1    action=plantInfuid=303367"
124.131.80.68    25/Jan/2010:14:31:34    "cuid=393678    timestamp=1264401093    action=checkResearchLayer    check=2f2cc50cc99aa9e05f02b6f6a47cbef6"200    765
125.107.199.28    25/Jan/2010:14:31:34    "timestamp=1264401094    oid=4    uid=350003    action=oreInfo    check=5d835e252b841c86da041b8b63b4b67e    cuid=356549"
111.167.145.209    25/Jan/2010:14:31:34    "action=plantInfo    cuid=154228    timestamp=1264401094    check=5d835e252b841c86da041b8b63b4b67e    uid=372981"    200

看到这里我有点发愁了，因为cuid 和 action 所在的列不是固定的，用简单的AWK过滤不行，需要借助AWK的循环和判断了，而这方面我没有做过于是就在群里发了求助信息，这时候有两个朋友 给了我回复一个是 辉太郎 另一个是 jeremy.zhang
他们的方案也不同，一个是用perl 脚本，另一个是直接用awk
先说说 用perl吧，其实perl我也不太懂，直接把他写的脚本贴上来

#!/usr/bin/perl -w
open(MYFILE,"/mnt/disk/newlog") || die "$!";
while()
     {
            $str = $_;
                if ($str =~ m/(.*?)\[/s)
                        {
                              $var1 =  $1;
                                 print  $var1;
                                     }
                if ($str =~ m/\[(.*?)\"/s)
                       {
                              $var4 = $1;
                                 print $var4;
                                    }
                if ($str =~ m/cuid=(\d+)/s)
                          {
                                  $var2 = $1;
                                    print "cuid=",$var2,"\t";
                                         }
               if ($str =~ m/action=(\w+)/s)
                           {
                                   $var3 = $1;
                                    print  "action=",$var3,"\n";
                                        }
                        }

/mnt/disk/newlog 这个是我刚才过滤出来的文件，执行的时候用perl 执行
perl 1.sh > newlog1
但是这条我执行后格式出了一点小偏差

124.197.61.124  25/Jan/2010:14:42:17    cuid=430334     action=plantInfo
54955 124.79.7.236    25/Jan/2010:14:42:17    cuid=318701     action=petsInfo
54956 122.230.66.90   25/Jan/2010:14:42:17    cuid=223422     action=compQuest
54957 113.128.147.225 25/Jan/2010:14:42:17    cuid=362043     action=plantInfo
54958 220.184.20.99   25/Jan/2010:14:42:17    cuid=484582     action=wordInfo
54959 222.161.49.201  25/Jan/2010:14:42:17    cuid=304167     218.95.48.90    25/Jan/2010:14:42:17    cuid=476480     action=plantInfo
54960 218.106.242.20  25/Jan/2010:14:42:17    cuid=501942     action=oreInfo
54961 221.137.223.58  25/Jan/2010:14:42:17    cuid=445595     action=takeQuest
54962 124.126.155.202 25/Jan/2010:14:42:17    cuid=0  action=initData
54963 113.224.227.68  25/Jan/2010:14:42:17    cuid=529218     action=editName
54964 121.4.66.146    25/Jan/2010:14:42:17    cuid=187626     action=researchLayer
54965 220.190.82.170  25/Jan/2010:14:42:17    cuid=62789      action=steal
54966 218.5.38.250    25/Jan/2010:14:42:17    cuid=456212     124.90.203.86   25/Jan/2010:14:42:17    cuid=492016     action=oreInfo

但是总体来讲还是可以接受的

下面看看 awk 命令，

第一步 awk '/action/{print $0}' access.log >tmp.log 过滤出包含action的行

第二步
awk '{print $1"\t"$4"\t"$9}' tmp.log > action.log 

 将没用的列去掉
第三部
过滤并输出 IP 时间 cuid= action=

awk -F"[ '['\"'&''=']+" '{printf $1"\t"$2"\t";for(i=3;i<=NF;i++){if($i=="cuid" || $i=="action")printf "%s",$i"="$(i+1)"\t"};printf "\n"}' action.log > cuid_action.log

下面是最终的结果

202.113.30.144        25/Jan/2010:13:42:07        cuid=181188    action=compound    
124.227.66.162        25/Jan/2010:13:42:07        cuid=355287    action=plantInfo    
124.240.39.49        25/Jan/2010:13:42:07        action=researchLayer    cuid=496990    
121.236.118.126        25/Jan/2010:13:42:07        cuid=509590    action=oreInfo    
113.139.18.82        25/Jan/2010:13:42:07        cuid=512461    action=oreInfo    
222.184.232.183        25/Jan/2010:13:42:07        cuid=520595    action=oreInfo    
218.59.80.95        25/Jan/2010:13:42:07        cuid=293339    action=questInfo    
221.6.38.37        25/Jan/2010:13:42:07        action=plantInfo    cuid=518015    
125.39.143.96        25/Jan/2010:13:42:07        cuid=133987    action=pkResult    
119.180.17.218        25/Jan/2010:13:42:07        cuid=452667    action=wordInfo

其实上面这三步可以合并成一步但是分开来弄更清晰一些
大家可以通过修改上面这些命令来 定制过滤自己需要的字段，希望对大家有所帮助