Sed of three swordsmen in text processing

1.sed working principle

sed namely Stream EDitor, and vi Different ,sed It's a line editor

Sed It's reading a line from a file or pipe , Deal with a line , Output one line ; Read one more line , One more line , One more line , until The last line . Whenever a row is processed , Store the currently processed rows in a temporary buffer , be called Mode space （Pattern

Space）, Then use sed Command processing buffer contents , After processing , Send the contents of the buffer to the screen . Then deal with a line , This is repeated , Until the end of the file . The design pattern of processing one line at a time makes sed A high performance ,sed When reading large files There will be no Caton . If you use vi Command to open dozens of M Hundreds of M The file of , It's obvious that there's a Caton , This is because vi The command to open a file is to load the file into memory at one time , Then open it again .Sed This situation is avoided , Line by line , It opens very fast , The execution speed is also very fast

2.sed Basic usage

Format ：

sed [option]... 'script;script;...' [inputfile...]

Common options

-n Do not output mode space content to screen , Do not print automatically

-e Multi point editing

-f FILE Read the edit script from the specified file

-r, -E Using extended regular expressions

-i.bak Back up the file and edit it in place

-s           Treat multiple files as separate files , Instead of a single continuous long file stream

# explain :

-ir I won't support it

-i -r Support

-ri   Support

-ni   Will empty the file

 script Format ：

' Address order

Address format ：

1. No address ： Process the full text

2. Single address ：

  #： Specified row ,$： The last line

  /pattern/： Every line that can be matched by this pattern

3. Address range ：

  #,#     # from # Go to the first place # That's ok ,3,6 From 3 Go to the first place 6 That's ok

  #,+#   # from # Row to +# That's ok ,3,+4 From 3 Go to the first place 7 That's ok

  /pat1/,/pat2/

  #,/pat/

  /pat/,#

4. Stepping ：~

    1~2 Odd line

    2~2 Even number line

command ：

p Print the contents of the current mode space , Append to default output

Ip Ignore case output

d Delete pattern space matching lines , And start the next cycle now

a [\]text Append text to the specified line , Support use \n Implement multiline appending

i [\]text Insert text before the line

c [\]text The replacement behavior is a single or multiple line of text

w file Save pattern matching lines to the specified file

r file After reading the text of the specified file to the matching line in the pattern space

= Print line numbers for lines in schema space

! Reverse matching row processing in pattern space

q           End or exit sed

 

Find overrides

s/pattern/string/ Modifier Search and replace , Other separators are supported , It can be in other forms ：[email protected]@@,s###

Replace modifier ：

g In line global substitution

p Show the lines that were replaced successfully

w   /PATH/FILE Save the replaced line to the file

I,i   Ignore case

 

Example : Get partition utilization

[[email protected] ~]#df | sed -En '/^\/dev\/sd/[email protected]* ([0-9]+)%.*@\[email protected]'
3
31
13

  Example ： take IP Address

[[email protected] ~]#ifconfig eth0 |sed -nr "2s/[^0-9]+([0-9.]+).*/\1/p"  
10.0.0.8
[[email protected] ~]#ifconfig eth0 | sed -En '2s/^[^0-9]+([0-9.]{7,15}).*/\1/p'
10.0.0.6
[[email protected] ~]#ifconfig eth0 | sed -rn '2s/^[^0-9]+([0-9.]+) .*$/\1/p'
10.0.0.8
[[email protected] ~]#ifconfig eth0 | sed -n '2s/^.*inet //p' | sed -n 's/ 
netmask.*//p'
10.0.0.8 
[[email protected] ~]#ifconfig eth0 | sed -n '2s/^.*inet //;s/ netmask.*//p'
10.0.0.8 
[[email protected] ~]#ifconfig eth0 | sed -rn '2s/(.*inet )([0-9].*)( 
netmask.*)/\2/p' 
10.0.0.8

3.sed Advanced usage  

sed In addition to pattern space , It also supports keeping space （Hold Space）, Use this space , The number in the pattern space can be

According to the , Temporarily save to hold space , So that subsequent processing , Achieve more powerful functions .

Common advanced commands

P Print mode space starts with \n Content , And append to the default output

h Overlay content from schema space to hold space

H Append content from pattern space to hold space

g Extract data from hold space and overlay into pattern space

G Extract content from hold space and append to pattern space

x Exchange content in schema space with content in retention space

n Read the next row of the matched row to cover the pattern space

N Read the next row of the matched row and append it to the pattern space

d Delete rows in schema space

D If the pattern space contains line breaks , Delete the text in the pattern space up to the first newline character , It doesn't read new input lines , And make

Restart the loop with the synthesized pattern space . If the pattern space does not contain line breaks , It's like sending out d Command to start a normal new cycle

Example : Print even lines

[[email protected] ~]#seq 10 | sed -n 'n;p'
2
4
6
8
10
[[email protected] ~]#seq 10 | sed -n '2~2p'
2
4
6
8
10
[[email protected] ~]#seq 10 | sed '1~2d'
2
4
6
8
10
[[email protected] ~]#seq 10 | sed -n '1~2!p'
2
4
6
8
10