[SOLVED] simplifying awk command

froggins · 02-23-2011, 04:58 AM

Hello,
I have just started using linux but fast becoming a convert.

I have a text file with 8,130 rows and 106 columns arranged as follows:

1 2 3 4 1 2
3 4 1 2 3 4

I want to re-arrange it so that it looks like this:

1/2 3/4 1/2
3/4 1/2 3/4

I have so far got

awk '{print $1"/"$2" "$3"/"$4" "$5"/"$6" "$7"/"$8" "105"/"106}' > output.txt

But I thought this was rather inelegant and wondered if I can write a script that will repeat the same operation on adjacent columns.

Apologies if this is a simple question.

Thanks!

Kirsten

Snark1994 · 02-23-2011, 07:45 AM

I used 'sed'...:

Code:

sed "s/\([0-9]\) \([0-9]\)/\1\/\2/g" input.txt > output.txt

This replaces every occurrence of

Code:

\([0-9]\) \([0-9]\)

(ie. two digits separated by a space - the $ and $ mean we can recall the digits we matched later on)

with

Code:

\1\/\2

(ie. the two digits we matched before, now separated by a slash. The slash needs to be escaped to prevent sed interpreting it as the end of the pattern)

Hope this helps,

grail · 02-23-2011, 08:26 AM

I agree that sed is a more compact solution here, but would add that the use of -r alleviates all the escaping so it can be a little clearer:

Code:

sed -r 's@([0-9]+) ([0-9]+)@\1/\2@g' file

Edit: Just thought of a funky awk solution:

Code:

awk 'ORS=(NR % 2)?"\\":RT' RS="[ \n]" file

kurumi · 02-23-2011, 08:44 AM

Code:

$ awk '{for(i=2;i<=NF;i+=2)$i="/"$i}{gsub(/ \//,"/")}1' file

grail · 02-23-2011, 09:21 AM

Or even:

Code:

awk '$0 = gensub(/(.) (.)/,"\\1\\\\2","g")' file

Equivalent to sed

froggins · 02-24-2011, 05:33 AM

Dear all,

Wow. Thank you for all these options. I have learned a lot from trying them all.

Kirsten