awk和sed （十二）

【教程主题】：4.awk和sed

【主要内容】

【1】awk

AWK是贝尔实验室1977年搞出来的文本出现神器.之所以叫AWK是因为其取了三位创始人 Alfred Aho，Peter Weinberger, 和 Brian Kernighan 的Family Name的首字符。要学AWK，就得提一提AWK的一本相当经典的书《The AWK Programming Language》，它在豆瓣上的评分是9.4分！在亚马逊上居然卖1022.30元。

Awk和sed 第二版

我从netstat命令中提取了如下信息作为用例：

$ cat netstat.txt

Proto Recv-Q Send-Q Local-Address Foreign-Address State

tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN

tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN

tcp 0 0 127.0.0.1:9000 0.0.0.0:* LISTEN

tcp 0 0 coolshell.cn:80 124.205.5.146:18245 TIME_WAIT

tcp 0 0 coolshell.cn:80 61.140.101.185:37538 FIN_WAIT2

tcp 0 0 coolshell.cn:80 110.194.134.189:1032 ESTABLISHED

tcp 0 0 coolshell.cn:80 123.169.124.111:49809 ESTABLISHED

tcp 0 0 coolshell.cn:80 116.234.127.77:11502 FIN_WAIT2

tcp 0 0 coolshell.cn:80 123.169.124.111:49829 ESTABLISHED

tcp 0 0 coolshell.cn:80 183.60.215.36:36970 TIME_WAIT

tcp 0 4166 coolshell.cn:80 61.148.242.38:30901 ESTABLISHED

tcp 0 1 coolshell.cn:80 124.152.181.209:26825 FIN_WAIT1

tcp 0 0 coolshell.cn:80 110.194.134.189:4796 ESTABLISHED

tcp 0 0 coolshell.cn:80 183.60.212.163:51082 TIME_WAIT

tcp 0 1 coolshell.cn:80 208.115.113.92:50601 LAST_ACK

tcp 0 0 coolshell.cn:80 123.169.124.111:49840 ESTABLISHED

tcp 0 0 coolshell.cn:80 117.136.20.85:50025 FIN_WAIT2

tcp 0 0 :::22 :::* LISTEN

下面是最简单最常用的awk示例，其输出第1列和第4例，

· 其中单引号中的被大括号括着的就是awk的语句，注意，其只能被单引号包含。

· 其中的$1..$n表示第几例。注：$0表示整个行。

$ awk '{print $1, $4}' netstat.txt

Proto Local-Address

tcp 0.0.0.0:3306

tcp 0.0.0.0:80

tcp 127.0.0.1:9000

tcp coolshell.cn:80

tcp :::22

我们再来看看awk的格式化输出，和C语言的printf没什么两样：

$ awk'{printf "%-8s %-8s %-8s %-18s %-22s %-15s ",$1,$2,$3,$4,$5,$6}'netstat.txt

Proto Recv-Q Send-Q Local-Address Foreign-Address State

tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN

tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN

tcp 0 0 127.0.0.1:9000 0.0.0.0:* LISTEN

tcp 0 0 coolshell.cn:80 124.205.5.146:18245 TIME_WAIT

tcp 0 0 coolshell.cn:80 61.140.101.185:37538 FIN_WAIT2

tcp 0 0 coolshell.cn:80 110.194.134.189:1032 ESTABLISHED

tcp 0 0 coolshell.cn:80 123.169.124.111:49809 ESTABLISHED

tcp 0 0 coolshell.cn:80 116.234.127.77:11502 FIN_WAIT2

tcp 0 0 coolshell.cn:80 123.169.124.111:49829 ESTABLISHED

tcp 0 0 coolshell.cn:80 183.60.215.36:36970 TIME_WAIT

tcp 0 4166 coolshell.cn:80 61.148.242.38:30901 ESTABLISHED

tcp 0 1 coolshell.cn:80 124.152.181.209:26825 FIN_WAIT1

tcp 0 0 coolshell.cn:80 110.194.134.189:4796 ESTABLISHED

tcp 0 0 coolshell.cn:80 183.60.212.163:51082 TIME_WAIT

tcp 0 1 coolshell.cn:80 208.115.113.92:50601 LAST_ACK

tcp 0 0 coolshell.cn:80 123.169.124.111:49840 ESTABLISHED

tcp 0 0 coolshell.cn:80 117.136.20.85:50025 FIN_WAIT2

tcp 0 0 :::22 :::* LISTEN

过滤记录

我们再来看看如何过滤记录（下面过滤条件为：第三列的值为0 && 第6列的值为LISTEN）

$ awk '$3==0 && $6=="LISTEN" ' netstat.txt

tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN

tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN

tcp 0 0 127.0.0.1:9000 0.0.0.0:* LISTEN

tcp 0 0 :::22 :::* LISTEN

其中的“==”为比较运算符。其他比较运算符：!=, <, < >=, < p>

我们来看看各种过滤记录的方式：

$ awk' $3>0 {print $0}'netstat.txt

Proto Recv-Q Send-Q Local-Address Foreign-Address State

tcp 0 4166 coolshell.cn:80 61.148.242.38:30901 ESTABLISHED

tcp 0 1 coolshell.cn:80 124.152.181.209:26825 FIN_WAIT1

tcp 0 1 coolshell.cn:80 208.115.113.92:50601 LAST_ACK

如果我们需要表头的话，我们可以引入内建变量NR：

$ awk '$3==0 && $6=="LISTEN" || NR==1 ' netstat.txt

Proto Recv-Q Send-Q Local-Address Foreign-Address State

tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN

tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN

tcp 0 0 127.0.0.1:9000 0.0.0.0:* LISTEN

tcp 0 0 :::22 :::* LISTEN

再加上格式化输出：

$ awk'$3==0 && $6=="LISTEN" || NR==1 {printf "%-20s %-20s %s ",$4,$5,$6}'netstat.txt

Local-Address Foreign-Address State

0.0.0.0:3306 0.0.0.0:* LISTEN

0.0.0.0:80 0.0.0.0:* LISTEN

127.0.0.1:9000 0.0.0.0:* LISTEN

:::22 :::* LISTEN

内建变量

说到了内建变量，我们可以来看看awk的一些内建变量：

$0	当前记录（这个变量中存放着整个行的内容）
$1~$n	当前记录的第n个字段，字段间由FS分隔
FS	输入字段分隔符默认是空格或Tab
NF	当前记录中的字段个数，就是有多少列
NR	已经读出的记录数，就是行号，从1开始，如果有多个文件话，这个值也是不断累加中。
FNR	当前记录数，与NR不同的是，这个值会是各个文件自己的行号
RS	输入的记录分隔符，默认为换行符
OFS	输出字段分隔符，默认也是空格
ORS	输出的记录分隔符，默认为换行符
FILENAME	当前输入文件的名字

怎么使用呢，比如：我们如果要输出行号：

$ awk'$3==0 && $6=="ESTABLISHED" || NR==1 {printf "%02s %s %-20s %-20s %s ",NR, FNR, $4,$5,$6}'netstat.txt

01 1 Local-Address Foreign-Address State

07 7 coolshell.cn:80 110.194.134.189:1032 ESTABLISHED

08 8 coolshell.cn:80 123.169.124.111:49809 ESTABLISHED

10 10 coolshell.cn:80 123.169.124.111:49829 ESTABLISHED

14 14 coolshell.cn:80 110.194.134.189:4796 ESTABLISHED

17 17 coolshell.cn:80 123.169.124.111:49840 ESTABLISHED

指定分隔符

$ awk 'BEGIN{FS=":"} {print $1,$3,$6}' /etc/passwd

root 0 /root

bin 1 /bin

daemon 2 /sbin

adm 3 /var/adm

lp 4 /var/spool/lpd

sync5 /sbin

shutdown6 /sbin

halt 7 /sbin

上面的命令也等价于：（-F的意思就是指定分隔符）

$ awk-F: '{print $1,$3,$6}'/etc/passwd

注：如果你要指定多个分隔符，你可以这样来：

awk-F '[;:]'

再来看一个以作为分隔符输出的例子（下面使用了/etc/passwd文件，这个文件是以:分隔的）：

$ awk-F: '{print $1,$3,$6}'OFS=" "/etc/passwd

root 0 /root

bin 1 /bin

daemon 2 /sbin

adm 3 /var/adm

lp 4 /var/spool/lpd

sync5 /sbin

字符串匹配

我们再来看几个字符串匹配的示例：

$ awk '$6 ~ /FIN/ || NR==1 {print NR,$4,$5,$6}' OFS=" " netstat.txt

1 Local-Address Foreign-Address State

6 coolshell.cn:80 61.140.101.185:37538 FIN_WAIT2

9 coolshell.cn:80 116.234.127.77:11502 FIN_WAIT2

13 coolshell.cn:80 124.152.181.209:26825 FIN_WAIT1

18 coolshell.cn:80 117.136.20.85:50025 FIN_WAIT2

$ $ awk '$6 ~ /WAIT/ || NR==1 {print NR,$4,$5,$6}' OFS=" " netstat.txt

1 Local-Address Foreign-Address State

5 coolshell.cn:80 124.205.5.146:18245 TIME_WAIT

6 coolshell.cn:80 61.140.101.185:37538 FIN_WAIT2

9 coolshell.cn:80 116.234.127.77:11502 FIN_WAIT2

11 coolshell.cn:80 183.60.215.36:36970 TIME_WAIT

13 coolshell.cn:80 124.152.181.209:26825 FIN_WAIT1

15 coolshell.cn:80 183.60.212.163:51082 TIME_WAIT

18 coolshell.cn:80 117.136.20.85:50025 FIN_WAIT2

上面的第一个示例匹配FIN状态，第二个示例匹配WAIT字样的状态。其实 ~ 表示模式开始。/ /中是模式。这就是一个正则表达式的匹配。

其实awk可以像grep一样的去匹配第一行，就像这样：

$ awk '/LISTEN/' netstat.txt

tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN

tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN

tcp 0 0 127.0.0.1:9000 0.0.0.0:* LISTEN

tcp 0 0 :::22 :::* LISTEN

我们可以使用 “/FIN|TIME/” 来匹配 FIN 或者 TIME :

$ awk'$6 ~ /FIN|TIME/ || NR==1 {print NR,$4,$5,$6}'OFS=" "netstat.txt

1 Local-Address Foreign-Address State

5 coolshell.cn:80 124.205.5.146:18245 TIME_WAIT

6 coolshell.cn:80 61.140.101.185:37538 FIN_WAIT2

9 coolshell.cn:80 116.234.127.77:11502 FIN_WAIT2

11 coolshell.cn:80 183.60.215.36:36970 TIME_WAIT

13 coolshell.cn:80 124.152.181.209:26825 FIN_WAIT1

15 coolshell.cn:80 183.60.212.163:51082 TIME_WAIT

18 coolshell.cn:80 117.136.20.85:50025 FIN_WAIT2

再来看看模式取反的例子：

$ awk'$6 !~ /WAIT/ || NR==1 {print NR,$4,$5,$6}'OFS=" "netstat.txt

1 Local-Address Foreign-Address State

2 0.0.0.0:3306 0.0.0.0:* LISTEN

3 0.0.0.0:80 0.0.0.0:* LISTEN

4 127.0.0.1:9000 0.0.0.0:* LISTEN

7 coolshell.cn:80 110.194.134.189:1032 ESTABLISHED

8 coolshell.cn:80 123.169.124.111:49809 ESTABLISHED

10 coolshell.cn:80 123.169.124.111:49829 ESTABLISHED

12 coolshell.cn:80 61.148.242.38:30901 ESTABLISHED

14 coolshell.cn:80 110.194.134.189:4796 ESTABLISHED

16 coolshell.cn:80 208.115.113.92:50601 LAST_ACK

17 coolshell.cn:80 123.169.124.111:49840 ESTABLISHED

19 :::22 :::* LISTEN

或是：

Awk '!/WAIT/' netstat.txt

折分文件

awk拆分文件很简单，使用重定向就好了。下面这个例子，是按第6例分隔文件，相当的简单（其中的NR!=1表示不处理表头）。

$ awk 'NR!=1{print > $6}' netstat.txt

$ ls

ESTABLISHED FIN_WAIT1 FIN_WAIT2 LAST_ACK LISTEN netstat.txt TIME_WAIT

$ cat ESTABLISHED

tcp 0 0 coolshell.cn:80 110.194.134.189:1032 ESTABLISHED

tcp 0 0 coolshell.cn:80 123.169.124.111:49809 ESTABLISHED

tcp 0 0 coolshell.cn:80 123.169.124.111:49829 ESTABLISHED

tcp 0 4166 coolshell.cn:80 61.148.242.38:30901 ESTABLISHED

tcp 0 0 coolshell.cn:80 110.194.134.189:4796 ESTABLISHED

tcp 0 0 coolshell.cn:80 123.169.124.111:49840 ESTABLISHED

$ cat FIN_WAIT1

tcp 0 1 coolshell.cn:80 124.152.181.209:26825 FIN_WAIT1

$ cat FIN_WAIT2

tcp 0 0 coolshell.cn:80 61.140.101.185:37538 FIN_WAIT2

tcp 0 0 coolshell.cn:80 116.234.127.77:11502 FIN_WAIT2

tcp 0 0 coolshell.cn:80 117.136.20.85:50025 FIN_WAIT2

$ cat LAST_ACK

tcp 0 1 coolshell.cn:80 208.115.113.92:50601 LAST_ACK

$ cat LISTEN

tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN

tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN

tcp 0 0 127.0.0.1:9000 0.0.0.0:* LISTEN

tcp 0 0 :::22 :::* LISTEN

$ cat TIME_WAIT

tcp 0 0 coolshell.cn:80 124.205.5.146:18245 TIME_WAIT

tcp 0 0 coolshell.cn:80 183.60.215.36:36970 TIME_WAIT

tcp 0 0 coolshell.cn:80 183.60.212.163:51082 TIME_WAIT

你也可以把指定的列输出到文件：

awk 'NR!=1{print $4,$5 < $6}'netstat.txt

再复杂一点：（注意其中的if-else-if语句，可见awk其实是个脚本解释器）

$ awk 'NR!=1{if($6 ~ /TIME|ESTABLISHED/) print > "1.txt";

else if($6 ~ /LISTEN/) print > "2.txt";

else print > "3.txt" }' netstat.txt

$ ls?.txt

1.txt 2.txt 3.txt

$ cat1.txt

tcp 0 0 coolshell.cn:80 124.205.5.146:18245 TIME_WAIT

tcp 0 0 coolshell.cn:80 110.194.134.189:1032 ESTABLISHED

tcp 0 0 coolshell.cn:80 123.169.124.111:49809 ESTABLISHED

tcp 0 0 coolshell.cn:80 123.169.124.111:49829 ESTABLISHED

tcp 0 0 coolshell.cn:80 183.60.215.36:36970 TIME_WAIT

tcp 0 4166 coolshell.cn:80 61.148.242.38:30901 ESTABLISHED

tcp 0 0 coolshell.cn:80 110.194.134.189:4796 ESTABLISHED

tcp 0 0 coolshell.cn:80 183.60.212.163:51082 TIME_WAIT

tcp 0 0 coolshell.cn:80 123.169.124.111:49840 ESTABLISHED

$ cat2.txt

tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN

tcp 0 0 0.0.0.0:80 0.0.0.0:* LISTEN

tcp 0 0 127.0.0.1:9000 0.0.0.0:* LISTEN

tcp 0 0 :::22 :::* LISTEN

$ cat3.txt

tcp 0 0 coolshell.cn:80 61.140.101.185:37538 FIN_WAIT2

tcp 0 0 coolshell.cn:80 116.234.127.77:11502 FIN_WAIT2

tcp 0 1 coolshell.cn:80 124.152.181.209:26825 FIN_WAIT1

tcp 0 1 coolshell.cn:80 208.115.113.92:50601 LAST_ACK

tcp 0 0 coolshell.cn:80 117.136.20.85:50025 FIN_WAIT2

统计

下面的命令计算所有的C文件，CPP文件和H文件的文件大小总和。

$ ls-l *.cpp *.c *.h | awk'{sum+=$5} END {print sum}'

2511401

我们再来看一个统计各个connection状态的用法：（我们可以看到一些编程的影子了，大家都是程序员我就不解释了。注意其中的数组的用法）

$ awk 'NR!=1{a[$6]++;} END {for (i in a) print i ", " a[i];}' netstat.txt

TIME_WAIT, 3

FIN_WAIT1, 1

ESTABLISHED, 6

FIN_WAIT2, 3

LAST_ACK, 1

LISTEN, 4

再来看看统计每个用户的进程的占了多少内存（注：sum的RSS那一列）

$ ps aux | awk 'NR!=1{a[$1]+=$6;} END { for(i in a) print i ", " a[i]"KB";}'

dbus, 540KB

mysql, 99928KB

www, 3264924KB

root, 63644KB

hchen, 6020KB

awk脚本

在上面我们可以看到一个END关键字。END的意思是“处理完所有的行的标识”，即然说到了END就有必要介绍一下BEGIN，这两个关键字意味着执行前和执行后的意思，语法如下：

· BEGIN{ 这里面放的是执行前的语句 }

· END {这里面放的是处理完所有的行后要执行的语句 }

· {这里面放的是处理每一行时要执行的语句}

为了说清楚这个事，我们来看看下面的示例：

假设有这么一个文件（学生成绩表）：

$ catscore.txt

Marry 2143 78 84 77

Jack 2321 66 78 45

Tom 2122 48 77 71

Mike 2537 87 97 95

Bob 2415 40 57 62

我们的awk脚本如下（我没有写有命令行上是因为命令行上不易读，另外也在介绍另一种用法）：

$ cat cal.awk

#!/bin/awk -f

#运行前

BEGIN {

math = 0

english = 0

computer = 0

printf"NAME NO. MATH ENGLISH COMPUTER TOTAL "

printf"--------------------------------------------- "

}

#运行中

{

math+=$3

english+=$4

computer+=$5

printf"%-6s %-6s %4d %8d %8d %8d ", $1, $2, $3,$4,$5, $3+$4+$5

}

#运行后

END {

printf"--------------------------------------------- "

printf" TOTAL:%10d %8d %8d ", math, english, computer

printf"AVERAGE:%10.2f %8.2f %8.2f ", math/NR, english/NR, computer/NR

}

我们来看一下执行结果：（也可以这样运行 ./cal.awk score.txt）

$ awk-f cal.awk score.txt

NAME NO. MATH ENGLISH COMPUTER TOTAL

---------------------------------------------

Marry 2143 78 84 77 239

Jack 2321 66 78 45 189

Tom 2122 48 77 71 196

Mike 2537 87 97 95 279

Bob 2415 40 57 62 159

---------------------------------------------

TOTAL: 319 393 350

AVERAGE: 63.80 78.60 70.00

环境变量

即然说到了脚本，我们来看看怎么和环境变量交互：（使用-v参数和ENVIRON，使用ENVIRON的环境变量需要export）

$ x=5

$ y=10

$ export y

$ echo$x $y

5 10

$ awk -v val=$x '{print $1, $2, $3, $4+val, $5+ENVIRON["y"]}' OFS=" " score.txt

Marry 2143 78 89 87

Jack 2321 66 83 55

Tom 2122 48 82 81

Mike 2537 87 102 105

Bob 2415 40 62 72

最后，我们再来看几个小例子：

#从file文件中找出长度大于80的行

awk 'length>80' file

#按连接数查看客户端IP

netstat -ntu | awk'{print $5}'| cut-d: -f1 | sort| uniq-c | sort-nr

#打印99乘法表

Seq 9 | sed'H;g'| awk-vRS='''{for(i=1;i< NFiprintfdxd dsquot i NR iNR i="=NR?" ":" ")}'

关于其中的一些知识点可以参看gawk的手册：

· 内建变量，参看：http://www.gnu.org/software/gawk/manual/gawk.html#Built_002din-Variables

· 流控方面，参看：http://www.gnu.org/software/gawk/manual/gawk.html#Statements

· 内建函数，参看：http://www.gnu.org/software/gawk/manual/gawk.html#Built_002din

· 正则表达式，参看：http://www.gnu.org/software/gawk/manual/gawk.html#Regexp

【2】sed

sed全名叫stream editor，流编辑器，用程序的方式来编辑文本，相当的hacker啊。sed基本上就是玩正则模式匹配，所以，玩sed的人，正则表达式一般都比较强。

同样，本篇文章不会说sed的全部东西，你可以参看sed的手册

用s命令替换

我使用下面的这段文本做演示：

$ cat pets.txt

This is my cat

my cat's name is betty

This is my dog

my dog's name is frank

This is my fish

my fish's name is george

This is my goat

my goat's name is adam

把其中的my字符串替换成Hao Chen’s，下面的语句应该很好理解（s表示替换命令，/my/表示匹配my，/Hao Chen’s/表示把匹配替换成Hao Chen’s，/g 表示一行上的替换所有的匹配）：

$ sed "s/my/Hao Chen's/g" pets.txt

This is Hao Chen's cat

Hao Chen's cat's name is betty

This is Hao Chen's dog

Hao Chen's dog's name is frank

This is Hao Chen's fish

Hao Chen's fish's name is george

This is Hao Chen's goat

Hao Chen's goat's name is adam

注意：如果你要使用单引号，那么你没办法通过’这样来转义，就有双引号就可以了，在双引号内可以用”来转义。

再注意：上面的sed并没有对文件的内容改变，只是把处理过后的内容输出，如果你要写回文件，你可以使用重定向，如：

$ sed"s/my/Hao Chen's/g"pets.txt > hao_pets.txt

或使用 -i 参数直接修改文件内容：

$ sed -i "s/my/Hao Chen's/g" pets.txt

在每一行最前面加点东西：

$ sed 's/^/#/g' pets.txt

#This is my cat

# my cat's name is betty

#This is my dog

# my dog's name is frank

#This is my fish

# my fish's name is george

#This is my goat

# my goat's name is adam

在每一行最后面加点东西：

$ sed 's/$/ --- /g' pets.txt

This is my cat---

my cat's name is betty ---

This is my dog ---

my dog's name is frank ---

This is my fish ---

my fish's name is george ---

This is my goat ---

my goat's name is adam ---

顺手介绍一下正则表达式的一些最基本的东西：

· ^ 表示一行的开头。如：/^#/ 以#开头的匹配。

· $ 表示一行的结尾。如：/}$/ 以}结尾的匹配。

· < 表示词首。如 <abc 表示以 abc 为首的詞。

· > 表示词尾。如 abc> 表示以 abc 結尾的詞。

· . 表示任何单个字符。

· * 表示某个字符出现了0次或多次。

· [ ] 字符集合。如：[abc]表示匹配a或b或c，还有[a-zA-Z]表示匹配所有的26个字符。如果其中有^表示反，如[^a]表示非a的字符

正规则表达式是一些很牛的事，比如我们要去掉某html中的tags：

html.txt

< code>b<This</>b< is what < code>spanstyle="text-decoration: underline;"<I</>span< meant. Understand?

看看我们的sed命令

# 如果你这样搞的话，就会有问题

$ sed 's/< >//g' html.txt

Understand?

# 要解决上面的那个问题，就得像下面这样。

# 其中的'[^<]' 指定了除了<的字符重复0次或多次。

$ sed 's/<[^>]*>//g' html.txt

This is what I meant. Understand?

我们再来看看指定需要替换的内容：

$ sed"3s/my/your/g"pets.txt

This is my cat

my cat's name is betty

This is your dog

my dog's name is frank

This is my fish

my fish's name is george

This is my goat

my goat's name is adam

下面的命令只替换第3到第6行的文本。

$ sed "3,6s/my/your/g"pets.txt

This is my cat

my cat's name is betty

This is your dog

your dog's name is frank

This is your fish

your fish's name is george

This is my goat

my goat's name is adam

$ cat my.txt

This is my cat, my cat's name is betty

This is my dog, my dog's name is frank

This is my fish, my fish's name is george

This is my goat, my goat's name is adam

只替换每一行的第一个s：

$ sed 's/s/S/1' my.txt

ThiS is my cat, my cat's name is betty

ThiS is my dog, my dog's name is frank

ThiS is my fish, my fish's name is george

ThiS is my goat, my goat's name is adam

只替换每一行的第二个s：

$ sed's/s/S/2'my.txt

This iS my cat, my cat's name is betty

This iS my dog, my dog's name is frank

This iS my fish, my fish's name is george

This iS my goat, my goat's name is adam

只替换第一行的第3个以后的s：

$ sed's/s/S/3g'my.txt

This is my cat, my cat'S name iS betty

This is my dog, my dog'S name iS frank

This is my fiSh, my fiSh'S name iS george

This is my goat, my goat'S name iS adam

多个匹配

如果我们需要一次替换多个模式，可参看下面的示例：（第一个模式把第一行到第三行的my替换成your，第二个则把第3行以后的This替换成了That）

$ sed '1,3s/my/your/g; 3,$s/This/That/g'my.txt

This is your cat, your cat's name is betty

This is your dog, your dog's name is frank

That is your fish, your fish's name is george

That is my goat, my goat's name is adam

上面的命令等价于：（注：下面使用的是sed的-e命令行参数）

Sed -e '1,3s/my/your/g' -e '3,$s/This/That/g' my.txt

我们可以使用&来当做被匹配的变量，然后可以在基本左右加点东西。如下所示：

$ sed 's/my/[&]/g' my.txt

This is [my] cat, [my] cat's name is betty

This is [my] dog, [my] dog's name is frank

This is [my] fish, [my] fish's name is george

This is [my] goat, [my] goat's name is adam

圆括号匹配

使用圆括号匹配的示例：（圆括号括起来的正则表达式所匹配的字符串会可以当成变量来使用，sed中使用的是1,2…）

$ sed 's/This is my ([^,]*),.*is (.*)/1:2/g' my.txt

cat:betty

dog:frank

fish:george

goat:adam

上面这个例子中的正则表达式有点复杂，解开如下（去掉转义字符）：

正则为：This is my ([^,]*),.*is (.*)
匹配为：This is my (cat),……….is (betty)

然后：1就是cat，2就是betty

sed的命令

让我们回到最一开始的例子pets.txt，让我们来看几个命令：

N命令

先来看N命令 —— 把下一行的内容纳入当成缓冲区做匹配。

下面的的示例会把原文本中的偶数行纳入奇数行匹配，而s只匹配并替换一次，所以，就成了下面的结果：

$ sed'N;s/my/your/'pets.txt

This is your cat

my cat's name is betty

This is your dog

my dog's name is frank

This is your fish

my fish's name is george

This is your goat

my goat's name is adam

也就是说，原来的文件成了：

This is my cat my cat's name is betty

This is my dog my dog's name is frank

This is my fish my fish's name is george

This is my goat my goat's name is adam

这样一来，下面的例子你就明白了，

$ sed'N;s/ /,/'pets.txt

This is my cat, my cat's name is betty

This is my dog, my dog's name is frank

This is my fish, my fish's name is george

This is my goat, my goat's name is adam

a命令和i命令

a命令就是append， i命令就是insert，它们是用来添加行的。如：

# 其中的1i表明，其要在第1行前插入一行（insert）

$ sed "1 i This is my monkey, my monkey's name is wukong"my.txt

This is my monkey, my monkey's name is wukong

This is my cat, my cat's name is betty

This is my dog, my dog's name is frank

This is my fish, my fish's name is george

This is my goat, my goat's name is adam

# 其中的1a表明，其要在最后一行后追加一行（append）

$ sed"$ a This is my monkey, my monkey's name is wukong"my.txt

This is my cat, my cat's name is betty

This is my monkey, my monkey's name is wukong

This is my dog, my dog's name is frank

This is my fish, my fish's name is george

This is my goat, my goat's name is adam

我们可以运用匹配来添加文本：

# 注意其中的/fish/a，这意思是匹配到/fish/后就追加一行

$ sed"/fish/a This is my monkey, my monkey's name is wukong"my.txt

This is my cat, my cat's name is betty

This is my dog, my dog's name is frank

This is my fish, my fish's name is george

This is my monkey, my monkey's name is wukong

This is my goat, my goat's name is adam

下面这个例子是对每一行都挺插入：

$ sed"/my/a ----"my.txt

This is my cat, my cat's name is betty

----

This is my dog, my dog's name is frank

----

This is my fish, my fish's name is george

----

This is my goat, my goat's name is adam

----

c命令

c 命令是替换匹配行

$ sed "2 c This is my monkey, my monkey's name is wukong" my.txt

This is my cat, my cat's name is betty

This is my monkey, my monkey's name is wukong

This is my fish, my fish's name is george

This is my goat, my goat's name is adam

$ sed"/fish/c This is my monkey, my monkey's name is wukong"my.txt

This is my cat, my cat's name is betty

This is my dog, my dog's name is frank

This is my monkey, my monkey's name is wukong

This is my goat, my goat's name is adam

d命令

删除匹配行

$ sed'/fish/d'my.txt

This is my cat, my cat's name is betty

This is my dog, my dog's name is frank

This is my goat, my goat's name is adam

$ sed'2d'my.txt

This is my cat, my cat's name is betty

This is my fish, my fish's name is george

This is my goat, my goat's name is adam

$ sed'2,$d'my.txt

This is my cat, my cat's name is betty

p命令

打印命令

你可以把这个命令当成grep式的命令

# 匹配fish并输出，可以看到fish的那一行被打了两遍，

# 这是因为sed处理时会把处理的信息输出

$ sed'/fish/p'my.txt

This is my cat, my cat's name is betty

This is my dog, my dog's name is frank

This is my fish, my fish's name is george

This is my goat, my goat's name is adam

# 使用n参数就好了

$ sed -n '/fish/p'my.txt

This is my fish, my fish's name is george

# 从一个模式到另一个模式

$ sed -n '/dog/,/fish/p'my.txt

This is my dog, my dog's name is frank

This is my fish, my fish's name is george

#从第一行打印到匹配fish成功的那一行

$ sed-n '1,/fish/p'my.txt

This is my cat, my cat's name is betty

This is my dog, my dog's name is frank

This is my fish, my fish's name is george