In sed processing, there are two buffers: [pattern space] and [hold space]

Sed execution process:

First read a line, remove the trailing newline character, store it in [pattern space], and execute the edit command. After processing, unless the – n parameter is added, the current [pattern space] will be printed out, and the line breaks that have been removed will be printed at the end. Leave [pattern space] blank. Then read the next line and process the next line.

The default output of SED: the content in [pattern space] is output to the standard output.

Common options:

The content in [pattern space] is not output to standard output: – n
By default, only one script can be executed, and multiple scripts can be executed: – e script, – expression = script

There can be multiple – e scripts
If there are too many scripts to execute, you can specify a script file: – F / path / to / sed_ scirpt

In the script file, there is an edit command for each line.
Support the use of extended regular expression, the default is basic regular expression: – R
Edit the original file directly: – I

Address delimitation:

1. Do not give address: process the full text

2. Single address

#: specify the line
/Pattern /: each row matched by this pattern

3. Address range

$: last line
#Beginning and end
#, + ා: start, and how many lines are added from the beginning
#, / pat1 /: start to, line to which pat1 matches
/Pat1 /, / pat2 /: the row that pat1 matches, and the row to which pat2 matches

4. Step by step~

1~2：1，3，5，7，9.。。 Rows (all odd rows)
2~2：2，4，6，8，10.。。 Rows (all even rows)

Edit command:

If there are multiple commands, separate them with semicolons.

Delete the content in [pattern space]: D

Delete the first to fifth lines of / etc / fstab.
```
# sed '1,5d' /etc/fstab
```

Print the content in [pattern space]: P

At first glance it looks like odd lines are printed, but actually even lines are printed, and odd lines are printed twice.
# sed '1~2p' /etc/fstab
After using - n to disable sed's default behavior, it is time to print odd lines
# sed -n '1~2p' /etc/fstab

Insert before line: – I / text supports multi line insertion using

# sed '3a \new line\nother line' /etc/fstab

#
# /etc/fstab
new line
other line
Add a comment before the UUID line:
# sed '/^UUID/i \#this is base for UUID' /etc/fstab
#this is base for UUID
UUID=3d3b316a-529e-484a-9895-e785fdde5365 /boot   xfs     defaults        0 0

Insert after line: – A / text supports multi line insertion using

# sed '3i \new line\nother line' /etc/fstab

#
new line
other line
# /etc/fstab

Replace line: – C \ \ text replace the matched line with text

Replace the line beginning with UUID with text.
```
# sed '/^UUID/c \#this is base for UUID' /etc/fstab
```
Save the matching line to the specified file: – w / path / to / save

Keep the lines that do not begin with “ා” to / TMP / fsnew
```
# sed -n '/^[^#]/w /tmp/fsnew' /etc/fstab
```
Insert the contents of the specified file after the matching line: – R / path / to / insert

After the third line of / etc / fstab, insert the contents of / TMP / TST.
```
# cat /tmp/tst
aaa
bbb
[[email protected] ~]# sed '3r /tmp/tst' /etc/fstab

#
# /etc/fstab
aaa
bbb
```

Add the line number on the previous line of the matched line:=

Add a line number to the line that begins with UUID

# sed '/^UUID/=' /etc/fstab

#
# /etc/fstab
# Created by anaconda on Fri Nov 29 16:44:28 2019
#
# Accessible filesystems, by reference, are maintained under '/dev/disk'
# See man pages fstab(5), findfs(8), mount(8) and/or blkid(8) for more info
#
/dev/mapper/centos-root /                       xfs     defaults        0 0
10
UUID=3d3b316a-529e-484a-9895-e785fdde5365 /boot     xfs     defaults        0 0

The matched line will not execute the following command; the unmatched line will execute the following command:!. Note: before processing the command.

Delete lines not beginning with ා:

# sed '/^#/!d' /etc/fstab
#
# /etc/fstab
# Created by anaconda on Fri Nov 29 16:44:28 2019
#
# Accessible filesystems, by reference, are maintained under '/dev/disk'
# See man pages fstab(5), findfs(8), mount(8) and/or blkid(8) for more info
#

Find replacement: S / to replace / replace with / replace tag. Its separator / can be specified by itself. The common ones are s @ @ @, S###

Replacement mark:
- Global substitution: G
- Save the successful replacement result to the file: w / path / to / save
- Display the lines with successful replacement: P

practice

1. Delete all white space characters at the beginning of all lines that start with white space characters in the test file

# cat test
   11
222
   333
\+: the number of times before matching is greater than or equal to 1
# sed '[email protected]^[[:space:]]\[email protected]@' test
11
222
333
# sed -n '[email protected]^[[:space:]]\[email protected]@p' test
11
333

2. Delete the “#” sign at the beginning of all lines beginning with “#” and all blank characters after “#” in the / etc / fstab file

# sed '[email protected]^#[[:space:]]*@@' /etc/fstab

3. Delete the “ා” sign at the beginning of the line and all blank characters after “#” in the / etc / fstab file, and delete all lines beginning with UUID

# sed -e '[email protected]^#[[:space:]]*@@' -e '/^UUID/d' /etc/fstab

4. Output an absolute path to the SED command and take out the directory part. Its behavior is similar to dirname

If there is / after the message, the matching fails.
# echo "/var/log/messages" | sed '[email protected][^/]\[email protected]@'
/var/log/
Even if there is / after message, it can match: not/

[^ /]: not/
\+: at least one
/\?: / has 0 or 1
# echo "/var/log/messages/" | sed '[email protected][^/]\+/\[email protected]@'
/var/log/
Even if there is / after message, it can match
Due to the use of - R (extended regular expression), + and? The preceding / can be omitted.
# echo "/var/log/messages/" | sed -r '[email protected][^/]+/[email protected]@'
/var/log/
[[email protected] ~]#

Advanced editing commands

Cover the contents in the [pattern space] space to the [hold space] space: H
Add the content in the [pattern space] space to the [hold space] space, and then delete the content in the [pattern space] space: H
Cover the contents in the [hold space] space to the [pattern space] space: G
Add the content in the [hold space] space to the [pattern space] space, and then delete the content in the [hold space] space: G
Exchange the contents in [hold space] [pattern space]: X
Put the next line of the matched line into [pattern space], and delete the matched line: n
Put the next line of the matched line into [pattern space], and do not delete the matched line: n
Delete the line in [pattern space]: D
Delete all lines in [pattern space] in multi line mode. (for example, if n is used, there are multiple lines in [pattern space]): D

1, display even lines:

Execution process: first read the first line into [pattern space], then the command n. N means to read the next line, delete the content in the current [pattern space], and then put the content of the next line into [pattern space]; the next command is p, P is to output the content in the current [pattern space] to the standard output, and all the second lines will be printed to the standard output. Then read line 3, the following command is n, read another line, and so on.

# sed -n 'n;p' /etc/fstab

2. Inverted text:

see: https://blog.csdn.net/itsenlin/article/details/21129405

Execution process: first read the first line into [pattern space], and find that the command in the first line is 1! G, so it is not processed; then, h, the content of [pattern space] will be covered to [hold space]; then, the command $! D will be executed because it is not the last line, so execute the command d to delete the contents in [pattern space].

Read the second line to [pattern space], and find that the command is 1! G. since it is not the first line, execute the G command and append the content of hold space] to [pattern space]. At this time, the contents of the second line and the first line are placed in [pattern space], and the second line is in front of the first line. Next is $! D. since it is not the last line, execute the D command to delete the contents in [pattern space].

# cat test
   11
222
   333
In the last line, D is not executed, and the content in [pattern space] is not used, so the flashback is printed to standard output.
# sed '1!G;h;$!d' ./test
   333
222
   11

You can also use the following command to achieve inversion.

Since the D command is not used, there is always content in [pattern space], so you must use – N. finally, use the P command to output the content in [pattern space] to the standard output.

# sed -n '1!G;h;$p' test

3. Take out the last line:

# sed '$!d' /etc/fstab

4. Take out the last two lines:

# sed '$!N;$!D' test

5. Delete all the original blank lines, and then add a blank line for all non blank lines

# cat test
   11
Empty line
Empty line
222
Empty line
Empty line
   333
444
  555
666
# sed '/^$/d;G' test
   11
Empty line
222
Empty line
   333
Empty line
444
Empty line
  555
Empty line
666
Empty line

6, display odd lines

# sed 'n;d' /etc/fstab

7. Add a blank line after each line

# sed 'G' /etc/fstab

8. Practice the usage of H and G

# cat t2
one
two
three
# sed 'H;g' t2
A blank line
one
A blank line
one
two
A blank line
one
two
three

Illustration: H is added to hold space,In hold space, there is a line break. So the first time you add one to hold space, the line before one is empty.

QQ group of C / C + + Learning mutual aid: 877684253

Introduction to sed command