一段不规则文本的截取(页 1) - Shell编程 - Linux伊甸园论坛

linux120 发表于 2007-10-11 11:46

一段不规则文本的截取

我有一个日志文件,如下:
[code] 10/11/2007 09:22:03 FX_FILESENT    2 file(s) sent to Destination
                              dm_reportD (Session ID = 1)

10/11/2007 09:22:19 FX_FILESENT    2 file(s) sent to Destination
                              dm_reportD (Session ID = 1)

10/11/2007 09:22:34 FX_FILESENT    2 file(s) sent to Destination
                              dm_reportD (Session ID = 1)

** 10/11/2007 09:22:46 FM_STAGFILE    Found 1 stagnant primary billing
                              file(s): Destination 'SXC2BP', oldest stagnant
                              file polled at 'Mon Oct 01 14:53:36 2007'.

** 10/11/2007 09:22:47 FM_STAGFILE    Found 22505 stagnant primary billing
                              file(s): Destination 'dm_reportD', oldest
                              stagnant file polled at 'Mon Oct 08 08:19:09
                              2007'.

10/11/2007 09:22:50 FX_FILESENT    2 file(s) sent to Destination
                              dm_reportD (Session ID = 1)[/code]
我需要将以**开始的那部分内容取出来,作下一步的处理,大家有什么好的办法么?

linux120 发表于 2007-10-11 11:49

其实我现在有实现,只是觉得效率不高,思路如下:
先找出**开始的那部分行号,然后用awk一次往下取一行,取到为空行的时候,就结束这一部分的读取.
[code]grep -n "^\*\\*" ${tmpLog} |awk -F\: '{print $1 }' >${tmpNum}

cat ${tmpNum} |while read num
do
i=${num}
Line=Null

while [ "${Line}" != "" ]
do
      Line=`awk 'NR=="'$i'"'  ${tmpLog} `
      #Debug
      #echo "Line=[$Line] and i = [$i]"
      echo "${Line}" >> ${sendFile}

      i=`expr $i + 1 `
      if [ $? -ne 0 ];then
            echo "序号计数器失败,请检查!!" >>${sendFile}
      fi
done

done[/code]

大家帮忙想想有没有更好的办法,最好能用awk ,或sed一行命令匹配出来.

[[i] 本帖最后由 linux120 于 2007-10-11 11:51 编辑 [/i]]

dearvoid 发表于 2007-10-11 12:16

只用 awk 就够了:
[quote][font=Courier][color=black][/color][color=blue]-(dearvoid@LinuxEden:Forum)-(~/tmp)-
[$$=1282 $?=0][/color] [color=red]; cat file
[/color][color=black] 1
** 2
** 3
4
[/color][color=blue]-(dearvoid@LinuxEden:Forum)-(~/tmp)-
[$$=1282 $?=0][/color] [color=red]; awk '/^\*\*/' file | while read line; do echo "$line"; done
[/color][color=black]** 2
** 3
[/color][color=blue]-(dearvoid@LinuxEden:Forum)-(~/tmp)-
[$$=1282 $?=0][/color] [color=red]; bye
[/color][color=black][/color][/font][/quote]

linux120 发表于 2007-10-11 12:29

呵呵,但是**下面还有一部分的,它不是一行,是不确定的几行组成的,以空行作结束符,我要获取整段

TUDOU01 发表于 2007-10-11 15:17

用awk编写的：
/[0-9][0-9]\/[0-9][0-9]\/200[0-9]/{if(/^\*/)
                                 {PD=1
                                    print$0;}
                                 else
                                 PD=0
                              }
!/[0-9][0-9]\/[0-9][0-9]\/200[0-9]/{if(PD==1) print $0}

040410615 发表于 2007-10-11 15:31

[root@localhost shell]# sed -n "/^\*\*/,/^$/p"  ./diay
** 10/11/2007 09:22:46 FM_STAGFILE    Found 1 stagnant primary billing
                              file(s): Destination 'SXC2BP', oldest stagnant
                              file polled at 'Mon Oct 01 14:53:36 2007'.

** 10/11/2007 09:22:47 FM_STAGFILE    Found 22505 stagnant primary billing
                              file(s): Destination 'dm_reportD', oldest
                              stagnant file polled at 'Mon Oct 08 08:19:09
                              2007'.
一条sed就可以解决了啊

linux120 发表于 2007-10-11 15:42

:victory:
好,不错,这就是我想要的结果,哈哈

linux120 发表于 2007-10-11 15:45

sed和awk太强大了,不过有时候用的不熟练,可以通过其它的方式来调整一下,条条大路通罗马啦....:lol

TUDOU01 发表于 2007-10-11 16:27

好聪明的方法！ok!

linux120 发表于 2007-10-11 16:42

其实oracle的日志里,有很多这种不规则的信息,用sed处理一下,告警就很方便了

li-jiahuan 发表于 2007-10-12 13:37

[code]
perl -00 -ne ' print if /^\Q**/ ' your_file
[/code]

等dearvoid出awk版:lol

li-jiahuan 发表于 2007-10-12 21:13

[quote]原帖由 [i]040410615[/i] 于 2007-10-11 15:31 发表 [url=http://linuxeden.com/forum/redirect.php?goto=findpost&pid=588998&ptid=159625][img]http://linuxeden.com/forum/images/common/back.gif[/img][/url]
[root@localhost shell]# sed -n "/^\*\*/,/^$/p"  ./diay
** 10/11/2007 09:22:46 FM_STAGFILE    Found 1 stagnant primary billing
                              file(s): Destination 'SXC2BP', o ... [/quote]

sed 的贪婪匹配
让这条记录正好可以‘正确’显示。。。

如果那两段带星的记录不是正好挨在一起
估计会有问题：）

err7dd 发表于 2007-11-14 22:01

awk '/\*\*/,/^$/ {print}' ./file
这样也可以吧

ArchieYao 发表于 2007-12-9 15:37

我也来一个,刚学awk,欢迎拍砖:loveliness:
awk '{RS="\n\n"}/^\*\*/{print}' logfile #思路:以两个连续的换行符作为awk的RS (记录分隔符)

页: [1]

Linux伊甸园论坛's Archiver

一段不规则文本的截取