I would like to replace "CC" with "C" and "AA" with A" in a particular column of Tab delimited file (using awk probably).

  • My input data file abc.txt has this "20131008","0004","0568","98" I ran this command to replace the text nawk -F, -vOFS=',' '{gsub("0568","0808",$3); print }' abc.txt In the output the delimiter "," is replaced with a space. The output looks like this "20131008" "0004" "0568" "98". Can someone please help with this? I tried to replace the -vOFS but doesnt work. Aug 19 '14 at 1:14
awk -F'\t' -vOFS='\t' '{ gsub("CC", "C", $1) ; gsub("AA", "A", $1) ; print }'

Replace $1 with the column that you wish to modify.

  • In case the delimiter has to be kept, a -vOFS='\t' option may be added.
    – manatwork
    Jun 14 '12 at 10:18
  • @manatwork Good point, added.
    – Chris Down
    Jun 14 '12 at 10:20
  • fantastic! works as per usual!!!
    – alex
    Jun 14 '12 at 10:51
  • Some versions of awk don't have a gsub function. How would you do it then? Can sed be used instead?
    – rahmu
    Jun 14 '12 at 13:33
  • 1
    @rahmu - gsub() has been defined by POSIX for over 20 years, so my suggestion would be to get a version of awk that isn't downright awful ;-)
    – Chris Down
    Jun 14 '12 at 13:40

If you didn't have access to gsub(), but you do have access to split(), you could just create the equivalent thusly:

Given the input

AA      AA      CC      CC      AA      CC

the following awk script

    OFS = "\t";
    split("1 3 5", Fields);
    split("A C", Replacements);
    for (i in Fields) {
        for (j in Replacements) {
            Replace = Replacements[j];
            sub(Replace Replace, Replace, $Fields[i]);

would produce the desired results for fields 1, 3, and 5:

A       AA      C       CC      A       CC

Even without split() it's possible if you want to hard-code the Fields and Replacements arrays in the BEGIN block.


To find and replace in one field use this command:

sed 's/whatyouwanttofind/whatyouwanttoreplace/field#'

i.e. you echo this:

$echo -e "1are,2are,3are,4are\n5are,6are,7are,8are"



and you want to do this

$echo -e "1are,2are,3are,4are\n5are,6are,7are,8are" |  sed 's/are/arrr/2'



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.