Shulou(Shulou.com)06/01 Report--

How to understand the Linux basic command Chinese text stream editing sed command, in view of this problem, this article introduces the corresponding analysis and solution in detail, hoping to help more partners who want to solve this problem to find a more simple and feasible method.

Unlike vim, sed is a non-interactive text editor. At the same time, it is character stream oriented, and each line of data is output after sed processing.

Sed [OPTION]... [script] [file]...

Sed works like this: first, initialize two data buffer pattern spaces and hold spaces Sed reads one line of input (from standard input or file), removes the ending newline character (\ n) and places it in the pattern space, and then executes the 'sed command' against the string in the pattern space. Each command can have an address associated with it, and the address can be regarded as a condition. Only when the condition is established, the relevant command is executed. After all executable commands are processed, the string that is still in mode space is appended with a newline character and printed out; then the next line of input is read to do the same processing until the active exit (Q) or the end of the input.

Address

The address can be in the following form

1. Number indicates the line number

2. First~step means starting from the first line, every step line

3. $represents a * line (note that it represents the end of the line when it appears in the regular expression)

4. / regexp/ means to match the regular expression regexp (see this article for regular expressions)

5.\% regexp% means that the matching regular expression regexp,% can be replaced with any other single character. (for cases where regexp contains slashes /)

6. / regexp/I is case-insensitive when matching the regular expression regexp

/ regexp/M enables regular multiline mode so that $matches not only the end of the line, but also the position before n or r, and ^ matches not only the beginning of the line, but also the position after n or r. At this point, you can use (\ `) to match the beginning of the pattern space and (\') to match the end of the pattern space.

You can also use a comma to separate two addresses to represent a range

Indicates that it starts from matching * addresses to the end of the second address or file. If the second address is a regular expression, the second address matching will not be performed on * address matching lines; if the second address is a line number, but less than or equal to * address matching line numbers, only one line (* address matching lines) will be matched.

In this case, the regular expression regexp starts to match on the line *. 0 can be used for * addresses only if the second address is a regular expression.

9. Addr1,+n means to match the address addr1 and the n lines that follow it.

10. Addr1,~n starts with the matching address addr1 and ends with a multiple line of n.

If no address is given, all lines match; append characters to the address or address range! Indicates that all lines that do not match will be processed only if the address is reversed.

Option

-n every processed string is printed out by default, and this option turns off this default behavior. Only the string that is acted on by the command p will be output.

-f file means to read sed commands from file

-I means to modify in place. When you apply this option, sed creates a temporary file and outputs the processing results to this file, and when you are finished, overwrites the temporary file to the original file.

-r means to use extended regular expressions

Command

P represents print mode space content, usually used with the option-n

[root@centos7 ~] # seq 5 1 2 3 4 5 [root@centos7 ~] # output only from the second line to the fourth line [root@centos7 ~] # seq 5 | sed-n '2Jing 4p' 23 4 [root@centos7 ~] #

D delete the contents of the schema space and process the next line of input immediately.

# Delete * * one line [root@centos7 ~] # seq 5 | sed'$d' 1 2 3 4 [root@centos7 ~] #

Q exit immediately, no longer process any commands and inputs (only a single address is accepted)

[root@centos7 ~] # seq 5 | sed'/ 3amp q' 1 2 3 [root@centos7 ~] #

N if you do not use the option-n, after outputting the contents of the mode space, read the next line of input and overwrite the current mode space contents. If there are no more input lines, sed exits execution.

[root@centos7 ~] # seq 9 | sed-n'nposition p' 2468 [root@centos7 ~] # Note that multiple commands are separated by semicolons

S/regexp/replacement/flag means to replace the part of the pattern space that matches the regular expression regexp with replacement. Here the symbol / can be replaced with any single character.

[root@centos7 ~] # echo "hello123world" | sed's / [0-9]\ + /, / 'hello,world # Note that + needs to be escaped here, but not if you use the option-r

In replacement

1.\ n (n is a number in 1-9) represents a grouping (.) in a regular expression. Citation of

[root@centos7 ~] # echo "hello123world" | sed-r's / [Amurz] + ([0-9] +) [Amurz] + /\ 1Mather '123 [root@centos7 ~] # echo "hello123world" | sed-r's / ([Amurz] +) [0-9] + ([Amurz] +) /\ 1,\ 2Universe' hello,world

2. & represents all parts of the schema space that match regexp

[root@centos7 ~] # echo "hello123world" | sed-r's / [0-9] + /: &: / 'hello:123:world

3.\ L convert the following characters to lowercase until\ U or\ E appears

4.\ l convert the next character to lowercase

5.\ U convert the following characters to uppercase until\ L or\ E appears

6.\ u convert the next character to uppercase

7.\ e stop case conversion starting with\ L or\ U

[root@centos7 ~] # echo "hello123world" | sed-r's / ^ ([Amurz] +) [0-9] + ([Amurz] +) $/\ U\ 1\ E,\ u\ 2gamma 'HELLO,World [root@centos7 ~] #

Flag

1. The number n means to replace the nth match.

[root@centos7 ~] # head-1 / etc/passwd root:x:0:0:root:/root:/bin/bash # the fifth part separated by replacing colons is empty [root@centos7 ~] # head-1 / etc/passwd | sed's / [^:]\ +: / / 5' root:x:0:0:/root:/bin/bash

2. G means global replacement

[root@centos7 ~] # echo "hello123world" | sed's HELLO123WORLD. /\ U &\ E hello123world. 'Hello123world [root@centos7 ~] # [root@centos7 ~] # echo "hello123world" | sed's HELLO123WORLD. /\ U &\ E HELLO123WORLD # when the numbers n and g are used at the same time Indicates that the replacement starts from the nth match until the * * match [root@centos7] # head-1 / etc/passwd | sed's / [^:]\ +: / / 4g 'root:x:0:/bin/bash/

3. P means that if the replacement is successful, the contents of the mode space are printed.

4. W file means that if the replacement is successful, the contents of the mode space are output to the file file.

5, I, and I indicate that case is not sensitive when matching regexp.

[root@centos7 ~] # echo 'HELLO123world' | sed-r's / [a Meiz] + / / Ig' 123 [root@centos7 ~] #

6, M and m indicate that regular multiline mode is enabled (as mentioned earlier). (give an example when talking about order N)

[root@centos7 ~] # echo hello | sed'le root@centos7 'hLEEo [root@centos7 ~] #

A text means to output mode space content followed by additional output text content

[root@centos7 ~] # seq 3 | sed'1 hello' 2a hello' 1 hello 2 hello 3 [root@centos7 ~] #

I text means to output text content before outputting schema space content

[root@centos7 ~] # seq 3 | sed'$ihello' 1 2 hello 3 [root@centos7 ~] #

C text means to delete the pattern space content that matches the address or address range, and output the text content. If it is a single address, each matching line is output, and if it is an address range, it is output only once.

[root@centos7 ~] # seq 5 | sed'1 hello 3 chello' hello 4 5 [root@centos7 ~] # seq 5 | sed'/ ^ [^ 3-4] / chello' hello hello 3 4 hello

= indicates that the current input line number is printed

[root@centos7 ~] # seq 100 | sed-n'$= '100 [root@centos7 ~] # seq 100 | sed-n' / ^ 10\ | ^ 20 Universe'10 20 100 [root@centos7 ~] # escaped | represents logic or

R file indicates that the contents of the file are read and output after the current mode space content is output

[root@centos7 ~] # cat file hello world [root@centos7 ~] # seq 3 | sed'1 file' 1 hello world 2 hello world 3 [root@centos7 ~] #

W file indicates that the contents of the mode space are output to file

N reads a line into the pattern space, then appends the next line to the pattern space (where the content in the pattern space is shaped like line1\ nline2). If there is no next line, sed exits.

[root@centos7 ~] # seq 10 | sed-n'Ntens / / p' 1 2 3 4 5 6 7 7 8 9 10 [root@centos7] # s command m flag for example [root@centos7 ~] # seq 3 | sed'Nutters / ^ 2 / xxx/' 1 2 3 [root@centos7 ~] # seq 3 | sed'Ninvestors / ^ 2 / xxx/m' 1 xxx 3 [root@centos7 ~] # seq 3 | sed'N' Seq 1 $/ xxx/' 1 2 3 [root@centos7 ~] # seq 3 | sed'NignentsAccord 1 $/ xxx/M' xxx 2 3

D if there is no new line in the pattern space (such as the new line generated by command N), it has the same effect as command d; if the new line is included, the contents of the * line are deleted and the rest of the pattern space is restarted. (note: commands after D will be ignored)

[root@centos7 ~] # seq 5 | sed 'NutterD' 5 [root@centos7 ~] # seq 5 | sed 'NutterD' 4 5

P print the contents of * lines in the mode space

[root@centos7 ~] # seq 10 | sed-n 'NtertP' 1357.9 [root@centos7 ~] # seq 10 | sed-n'Nentantor P' 147 # Note the different [root@centos7 ~] # seq 10 in the output of another way of writing | sed-n '1room3P' 1 4 7 10

G replace the content in the schema space with the content in the hold space

[root@centos7 ~] # seq 5 | sed-n'gtern *

G appends a newline character to the pattern space, and then appends the contents of the holding space to the newline character. (at this point, the content in the schema space is shaped like PATTERN\ nHOLD)

[root@centos7 ~] # seq 5 | sed'Gutterss /\ nUniverse xxUnix '1xx 2xx 3xx 4xx 5xx

H replace the content in the hold space with the content in the schema space (note that the content in the schema space is not cleared at this time)

[root@centos7 ~] # seq 5 | sed-n'hash Gentlemen /\ nUnix root@centos7 seq p '1xx1 2xx2 3xx3 4xx4 5xx5 [root@centos7 ~] # 1xx1xx1 2xx2xx2 3xx3xx3 4xx4xx4 5xx5xx5 | sed-n'hterGente Gentleman Gentleman /\ nCharacterGp' 1xx1xx1 2xx2xx2 3xx3xx3 4xx4xx4 5xx5xx5

H appends a newline character to the reserved space, and then appends the contents of the pattern space to the newline character. (keep the content in the space shaped like HOLD\ nPATTERN at this time)

[root@centos7 ~] # seq 3 | sed-n'Hutter Genders /\ nUniverse xxAGPP '1xxxx1 2xxxx1xx2 3xxxx1xx2xx3 [root@centos7 ~] #

X swap the contents of mode space and hold space

[root@centos7 ~] # seq 9 | sed-n'1! {x Ting N}; s /\ nAccord p' 3 25 47 69 # in {...} is the command group

Label specifies the label location for the branch command (address matching is not allowed)

B label unconditionally jumps to the label branch, if label is omitted, then jumps to the end of the entire command (that is, to start the next read)

# if you delete the comment section of the xml file (the middle part is the comment, you can have multiple lines) sed'/! {Nbomba} D} 'server.xml # means to execute N all the time until the match, and delete the content in the pattern space after matching to->. For example, in the configuration file of nagios, there are many fields of define host {...}, as follows: define host {use windows-server host_name serverA hostgroups 060202 alias 060202 contact_groups yu address 192.168.1.1} # now you need to delete the segment with ip address 192.168.1.1 It can be like this: sed-I'/ define host/ {: a N _ file /! ba;/192\ .168\ .1 / d} 'note the difference between this and the previous example

T label jumps to label only after a successful s replacement command is executed. If label is omitted, jump to the end of the entire command (that is, start the next read)

# such as row-column conversion [root@centos7 ~] # seq 10 | sed': a person who has changed the MAC address 78A35114F798 to the format 78:A3:51:14:F7:98 [root@centos7 temp] # echo '78A35114F798' | sed-r': sed /\ B\ w {2}\ bbank Ta' 78:A3:51:14:F7:98 [root@centos7 temp] # # where\ b means to match the word boundary, and\ B means to match any other character that is not the word boundary. Of course, it can also be implemented in other ways: [root@centos7 temp] # echo '78A35114F798' | sed-r's sed.\ BAccording to the word boundary,\ B means matching the word boundary, and\ B means matching any other character that is not the word boundary.

T label jumps to label after one input as long as the replacement command is not successfully executed, and if label is omitted, jump to the end of the entire command (that is, start the next read)

Z means to clear the contents of the pattern space, which has the same effect as sswab. Hammer /, but is more effective.

More examples

1. Delete the previous and next rows of matching rows

# for example, the input data is the output of the command seq 10 (of course, it can also be any other file content) # it is required to delete the previous line and the last line that match 5. [root@centos7 temp] # seq 10 | sed-n'$! Nittn /\ n5 / {sram. *\ nAccording to pash Nacid}; Pash D' 1 2 3 5 7 8 9 10

2. Merge odd and even rows

# the input data is the output of the command seq 11, and the odd and even numbers are required to be placed on the same line respectively # output * * line `1 3 5 7 9 11 `, the second line `2 4 6 8 10` [root@centos7 ~] # seq 11 | sed-nr'$! Nitt2neighbors / ([^\ n] +)\ n ((. +)\ n)? (. +)\ n (. +) /\ 4\ 1\ n\ 5\ 3hand h $p' 1 3 5 7 9 11 2 4 6 8 10 [root@centos7 ~] #

3. Merge multiple files

# content of text a.txt: 01 12510101 4001 02 12310001 4002 03 12550101 4003 04 12610001 4004 05 12810001 4005 06 12310001 4006 07 12710001 4007 08 12310001 4008 09 12810101 4009 10 12510101 4010 11 12310001 4011 12610001 4012 13 12310001 4013 # text b.txt: a 12410101 2006-02-15 2009-01-31 2009-01-31 4002 B 12310001 2006-08-31 2008-08-29 4001 C 12610001 2008-05-23 2008-05-22 4002 D 12810001 1992-12-10 1993310304001 E 12660001 text b.txt content: a 12410101 2006-02-15 2009-01-31 4002 B 12310001 2006-08-31 4001 C 12610001 2008-05-23 2008-05-22 4002 D 12810001 1992-12-10 1993Uni6304001 E 12660001 06ax 01 4005 # requires that the second column in the a.txt content is the same as the second column in the b.txt And append two corresponding date columns in b.txt. # such as: 02 12310001 4002 2006-08-31 2008-08-29 sed-rn'/ ^ [01] / ba;H;:a;G;s/ ^ ((.) (.) (. *) ([^\ n] +)). *\ 3 (([^] *) {2}). * /\ 1\ 5NR==FNR p 'b.txt a.txt # of course, if you use awk to deal with it, the solution is easier to understand: awk 'NR==FNR {a [$2] = $3FS$4 Next} {if ($2 in a) print $0 in a [$2]} 'b.txt a.txt

To deepen your understanding of the various command features of sed, analyze these three examples yourself.

The combination of various commands, coupled with the power of regular expressions, allows sed to handle all the problems that can be evaluated. However, due to the poor readability of the code, it is difficult to understand. Sed is usually used as a text editor to stream the text non-interactively. If you understand the meaning of the above commands and use them skillfully, you will find the power of sed.

This is the answer to the question on how to understand Linux basic commands Chinese text flow editing sed commands. I hope the above content can be of some help to you. If you still have a lot of doubts to be solved, you can follow the industry information channel for more related knowledge.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.