awk question: grouping 3 columns

[fluxi]

Hi to all,
I am new to this forum. I have an awk problem:
The input data looks something like that

a1 b1 c1
a1 b2 c2
a2 b2 c3
a3 b4 c1
a1 b1 c10
a2 b5 c5
a3 b1 c4
a4 b2 c3
a1 b2 c2
a2 b2 c3
a1 b1 c1
a1 b2 c2
a1 b2 c2
a2 b2 c3
a2 b2 c3
b1 c1: 2
The output should look like that:
a1 b1 c1: 2 times
______c10:1 time
___b2 c2: 3 times
a2 b2 c3: 3 times
and so on.

It should count the groups for all $i.
I have allready done some easy awk-programming, but I can't get this one right.

Maybe you can help me
yours
Petra

[r-senior]

Is this a homework problem? What do you have so far?

If you are completely stuck, you might want to read up about associative arrays:

https://www.gnu.org/software/gawk/manual/html_node/Array-Basics.html#Array-Basics

[fluxi]

Hi,
no,no homework. These years are long gone ;-) Allas....
I have: awk -F" " '{if(k[$1])k[$1]=k[$1]":"$2 ":" $3; else k[$1]=$2;}END{for (j in k)print j, k[j];}' test

this gives:
a1 b1:b2:c2:b3:c10:b2:c2:b1:c1:b2:c2:b2:c2
a2 b2:b5:c5:b2:c3:b2:c3:b2:c3
a3 b4:b1:c4 but there are strange signs in the output... I can't copy past them. But they are not in the input file, but in the output. I tried to change the character encoding of the terminal, but they remain...
a4 b2

The first set is allways wrong. And I can't get the counting right ;-)


I now tried something slightly different:
awk -F":" '{coli[$1,$2,$3]++} END {for (i in coli) print i,":" col[i]}' test.txt |sort

This gives
a1 b1 c1 :2
a1 b2 c2 :4
a1 b3 c10 :1
a2 b2 c3 :4
a2 b5 c5 :1
a3 b1 c4 :1
a3 b4 c1 :1
a4 b2 c3 :1


Everything seems fine, but this seems too easy to me. I will have to test it on the real data tomorrow...
Yours Petra

[greenpeace]

Hi!

Maybe I'm oversimplifying it, but it almost seems like you might be able to do it by using the commands:


sort file.txt | uniq -c

Which will tell you the count of each unique line in file.txt

Would that work?

Cheers,
Gp

[r-senior]

I think this does it in awk, if you want control over the output:


awk '{++count[$0]} END {for (pattern in count) printf("%-12s: %d times\n", pattern, count[pattern])}'

For example:


$ cat input
a1 b1 c1
a1 b2 c2
a2 b2 c3
a3 b4 c1
a1 b1 c10
a2 b5 c5
a3 b1 c4
a4 b2 c3
a1 b2 c2
a2 b2 c3
a1 b1 c1
a1 b2 c2
a1 b2 c2
a2 b2 c3
a2 b2 c3
$ awk '{++count[$0]} END {for (pattern in count) printf("%-12s: %d times\n", pattern, count[pattern])}' input
a2 b5 c5 : 1 times
a3 b1 c4 : 1 times
a2 b2 c3 : 4 times
a1 b1 c10 : 1 times
a1 b2 c2 : 4 times
a1 b1 c1 : 2 times
a3 b4 c1 : 1 times
a4 b2 c3 : 1 times


So it makes an associative array 'count' using the whole input line as the key, incrementing the value in the associative array on each occurrence. Then the end pattern iterates over the array and prints the key and value.

EDIT: That's basically what you have in your second attempt. I just didn't read it properly.

[drmrgd]

If you're not fixed on using Awk for this, you could do it fairly easily in Perl (with the same exact strategy, I should point out):


#!/usr/bin/perl

use warnings;
use strict;

my %lineCount;

while(<>) {
chomp;
$lineCount{$_}++;
}

foreach my $key ( keys %lineCount ) {
my $count = $lineCount{$key};
print "$key: $count times\n";
}

Assuming your input file is called 'file.txt' just run it like so:


perl count.pl file.txt

[ofnuts]

Hi!

Maybe I'm oversimplifying it, but it almost seems like you might be able to do it by using the commands:


sort file.txt | uniq -c

Which will tell you the count of each unique line in file.txt

Would that work?

Cheers,
Gp
I was going to suggest the same...