Change string with awk

Question

I am trying to do a single line replacement in a file with awk like

changing this:

e1 is (on) e2 is (off) 

to:

 e1 is (on) e2 is (on) 

use command:

 awk '/e2/{gsub(/off/, "on")};{print}' ~/Documents/Prueba > ~/Documents/Prueba 

this makes a replacement, but the file ends up empty!

Answer 1

Another answer using another tool (sed and the -i flag (in place))

 sed -i '/e2/ s/off/on/' ~/Documents/Prueba 

Answer 2

Your awk is correct, however you are redirected to the same file as your original. This causes the source file to be overwritten before it is read. You will need to redirect the output to another file.

 awk '/e2/{gsub(/off/, "on")};{print}' ~/Documents/Prueba > ~/Documents/Prueba.new 

Rename Prueba.new after that, if necessary.

Answer 3

You can also use cat to read the file first, then use pipe to redirect to stdout, then read with awk from stdin:

 cat ~/Documents/Prueba | awk '/e2/{gsub(/off/, "on")};{print}' - > ~/Documents/Prueba 

I think dash is optional, since you are only reading stdin.

Some interesting documentation: https://www.gnu.org/software/gawk/manual/html_node/Naming-Standard-Input.html

Answer 4

As explained in other answers and in the question “ Why reading and writing the same file using I / O redirection leads to an empty file on Unix? ”, Shell redirection destroys your input file before it is read.

To solve this problem without explicitly accessing temporary files, review the sponge command from moreutils .

 awk '/e2/{gsub(/off/, "on")};{print}' ~/Documents/Prueba | sponge ~/Documents/Prueba 

Alternatively, if GNU awk is installed on your system, you can use the placement extension .

 gawk -i inplace '/e2/{gsub(/off/, "on")};{print}' ~/Documents/Prueba 

Answer 5

You cannot redirect to the same file as the input file. Choose a different file name.

> will delete your file first.