issue with awk

maddyfreaks · 08-29-2017, 12:33 PM

Hi Team
Am new to unix/linus

I am trying to use awk to pull certain cols but having trouble.

Below is the data in my faile.

Code:

IDNU      KEY_NAM    SIZE
--------------------------------
DB901     TAB_A      9.8 GB
DB890     TAB _A_1   1.1 GB
DB797     T   _A _1  0.1 GB

cat file.log| awk '{ print $1";"$2}'

am only getting TAB_A, TAB, T but missing others in column 2

i tried using sed 's/\s\s * / /g' but it only works for TAB _A_1 but not others

any other better idea to pull whole column 2

Turbocapitalist · 08-29-2017, 12:36 PM

Are the columns separated by tabs?

Code:

awk '{ print $1,$2; }' FS="\t" OFS=";" file.log

maddyfreaks · 08-29-2017, 12:44 PM

Nope they are not tab separated.

smallpond · 08-29-2017, 01:15 PM

I think you will just have to count columns.

Code:

awk '{c2=substr($0,11,10); print $1 ";" c2}' foo
IDNU;KEY_NAM   
--------------------------------;----------
DB901;TAB_A     
DB890;TAB _A_1  
DB797;T   _A _1

Turbocapitalist · 08-29-2017, 01:15 PM

You'll have to split or adjust the columns based on their absolute position then.

If you use sed first, then it could be like this:

Code:

sed -E -e 's/^(.{9}) (.{10}) /\1\t\2\t/; s/ +\t/\t/g;' file.txt | awk . . . FS="\t"

Or you could do it all in awk somewhat differently. Probably the substr() function would be of use there. Or if you have Gnu awk (gawk) you can use the FIELDWIDTHS variable.

Which version of awk do you have?

jlinkels · 08-29-2017, 02:03 PM

If you are using Gawk (and you should), there is a solution for exactly that:
https://www.gnu.org/software/gawk/ma...tant-Size.html

jlinkels

maddyfreaks · 08-29-2017, 03:16 PM

Thanks all. sed worked out for me

Turbocapitalist · 08-29-2017, 11:34 PM

No problem.

Can we still ask which version of awk you have? There is an all-awk solution or two also but the optimal way for that depends on which version you have.

allend · 08-30-2017, 03:19 AM

To just get column 2, why not 'cut -b11-20'?

Turbocapitalist · 08-30-2017, 05:30 AM

Quote:

Originally Posted by allend

To just get column 2, why not 'cut -b11-20'?

That would work well unless there is some constraint to use only awk. If gawk is available, then the following would be a solution:

Code:

gawk '{print $1, $2, $3;}' OFS=";" FIELDWIDTHS="10 11 10" inputfile.txt

But the FIELDWIDTHS variable is not available for other versions of awk.

MadeInGermany · 08-30-2017, 06:52 AM

Also consider unexpand to convert to tab-separated fields.

Code:

man unexpand

maddyfreaks · 08-30-2017, 12:29 PM

awk - 3.1.7

Turbocapitalist · 08-30-2017, 12:43 PM

Thanks. But is it mawk, gawk, nawk, or old awk? Only gawk has the FIELDWIDTHS built-in variable which can split columns based on width.

MadeInGermany · 09-06-2017, 11:04 AM

With unexpand and the input from post#1 you can convert space to TAB at certain positions,
then pipe to awk with a field separator set to TAB

Code:

unexpand -t 10,21 file.log | awk -F"\t" '{print $1,$2}'
IDNU KEY_NAM
-------------------------------- 
DB901 TAB_A
DB890 TAB _A_1
DB797 T   _A _1