[SOLVED] how to drop the last few substrings in a string?

michaelk · 09-06-2022, 09:49 PM

You never answered the questions on spacing, is it fixed, how many, is it variable etc.
Change OFS as desired. $nbr is the number of fields to ouput i.e x=6 will output 1-6.

Code:

awk -v x=$nbr 'BEGIN {OFS="  "} {NF =  x} {print $0 }' input.file

I don't know of an easy way to drop one field with awk. But then again this is a new requirement? cut would be better but does not handle multiple field separator.

syg00 · 09-06-2022, 11:57 PM

Quote:

Originally Posted by michaelk

$nbr is the number of fields to ouput

Not from my example. The OP indicates the number of fields to be deleted from the end is known, and that's what I coded for. Likewise grail.

Substringing would work but looks to be messy with awk in this case - sed is likely more robust here. Maybe just use a bash case for the various $nbr values.

MadeInGermany · 09-07-2022, 01:29 AM

Yet untested:

Code:

nbr=3
sed "s/\([[:space:]]*[^[:space:]]\{1,\}\)\{$nbr\}[[:space:]]*$//" inputfile

grail · 09-07-2022, 04:47 AM

Thnaks syg00 for the follow up (thought I had mis-read)

I think the question becomes, is the whitespace important and if so, what is it going into that needs it kept that way??

Side note :- I see a number of people have mentioned a cut solution. I for one would be curious to see this in action with different sized spacing in the input?

allend · 09-08-2022, 12:46 AM

Quote:

Side note :- I see a number of people have mentioned a cut solution. I for one would be curious to see this in action with different sized spacing in the input?

I take your point, cut is not the tool with different sized spacing (which was not apparent when I initially posted), but for giggles.

Code:

sed 's/[[:space:]]\([^[:space:]]\)/ \t\1/g' inputfile | rev | cut -f 1-3 --complement | rev | tr -d "\t"