sed replace in-line a specific column number value at a specific line number

Question

I have a 5 columns csv file (space separated) like this :

username1 20130310 enabled 20130310 enabled
username2 20130310 enabled 20130321 disabled
username3 2013         


        

Answer 1

This might work for you (GNU sed & Bash):

lineNumber=4 newValue=xxxx sed -i ${lineNumber}'s/\S\+/'"${newValue}"'/4' file

However beware. If the newValue contains a / you will need to change the substitute command delimiter to something else e.g. s@\S\+@${newValue}@4

Answer 2

Here is one way:

$ sed '/^username4/{s/ [^ ]*/ anything/3}' file
username1 20130310 enabled 20130310 enabled
username2 20130310 enabled 20130321 disabled
username3 20130320 disabled 20130321 enabled
username4 20130310 disabled anything disabled

# store changes back to the file 
$ sed -i '/^username4/{s/ [^ ]*/ anything/3}' file

But avoiding awk because sed has the -i option isn't a good reason. awk is more suited to working with this kind of problem.

$ awk '$1=="username4"{$4="anything"}1' file
username1 20130310 enabled 20130310 enabled
username2 20130310 enabled 20130321 disabled
username3 20130320 disabled 20130321 enabled
username4 20130310 disabled anything disabled

# store changes back to the file
$ awk '$1=="username4"{$4="anything"}1' file > tmp && mv tmp file

With awk you can easily do field comparison and editing, using shell variable isn't a quoting nightmare and understanding scripts you wrote only yesterday isn't and issue unlike with sed:

$ linenumber=4

$ newvalue=anything 

$ awk 'NR==n{$4=a}1' n=$linenumber a=$newvalue file 
username1 20130310 enabled 20130310 enabled
username2 20130310 enabled 20130321 disabled
username3 20130320 disabled 20130321 enabled
username4 20130310 disabled anything disabled

$ awk 'NR==n{$4=a}1' n=$linenumber a=$newvalue file > tmp && mv tmp file