Delete specific columns from csv file maintaining same structure on output

Question

I would like to delete column 3 and keep the same structure on output file.

input file

12,10,10,10 10,1
12,23 1,45,6,7
11  2,33,45,1,2
1,2,34,5,6

I tried

awk -F, '!($3="")' file | awk -v OFS=',' '{$1=$1}1'
12,10,10,10,1
12,23,1,6,7
11,2,33,1,2
1,2,5,6

Output desired

12,10,10 10,1
12,23 1,6,7
11  2,33,1,2
1,2,5,6

Appreciate your help

from the duplicate, `cut -d, --complement -f3` would also be a nice solution for this case — Sundeep, Dec 14 '17 at 12:51

anubhava · Accepted Answer · 2017-12-14T15:27:45.770

3

Better to use sed here:

sed -E 's/^(([^,]*,){2})[^,]*,/\1/' file

12,10,10 10,1
12,23 1,6,7
11  2,33,1,2
1,2,5,6

Search regex:

^: Start
(: Start 1st capture group
- (: Start 2nd capture group
- [^,]*,: Match 0 or more non-comma characters followed by a comma
- ){2}: End 2nd capture group. {2} means match 2 pairs of above match
): End 1st capture group
[^,]*,: Match 0 or more non-comma characters followed by a comma

Replacement:

edited Dec 14 '17 at 15:27

answered Dec 14 '17 at 12:19

anubhava

anubhava, please can you explain how the code works ( the logic ), it can work to able to delete any column ?. In my input example i show only 5 columns but this can be more? .. Appreciate your comments.. – OXXO Dec 14 '17 at 15:14
I have added detailed explanation in my answer. Let me know if there is any further doubt. – anubhava Dec 14 '17 at 15:31
1

anuhava, works perfectly, many thanks for the explanation in the code – OXXO Dec 14 '17 at 19:59

score 2 · Answer 2 · answered Dec 14 '17 at 12:20

2

sed solution:

sed -E 's/^([^,]+,[^,]+,)[^,]+,/\1/' file

The output:

12,10,10 10,1
12,23 1,6,7
11  2,33,1,2
1,2,5,6

answered Dec 14 '17 at 12:20

RomanPerekhrest

score 2 · Answer 3 · answered Dec 14 '17 at 12:24

2

In awk:

$ awk 'BEGIN{FS=OFS=","}{for(i=3;i<NF;i++)$i=$(i+1);NF--}1' file
12,10,10 10,1
12,23 1,6,7
11  2,33,1,2
1,2,5,6

answered Dec 14 '17 at 12:24

James Brown

score 2 · Answer 4 · answered Dec 14 '17 at 12:26

Perl solution:

perl -l -aF/,/ -ne 'splice @F, 2, 1; print join ",", @F'

-l removes newlines from input and adds them to prints
-n reads the input line by line
-a splits each line into the @F array
-F specifies how to split (/,/ means on comma)
splice removes elements from arrays, in this case it removes 1 element at position 2.

4 Answers4