Unix (awk)

[malpa]

Hi

I would like to do this


awk ' BEGIN{FS=";"}
{ n[$1";"$2]++; l[$1","$2]+=$7 }
END{
for ( i in n) {
print i":"n";"l
} ' /input/file


but only with arrays in bash or sh or csh o ksh script. is it possible.??


Thanks malpa

 

[Annihilannic]

Not sure about csh, but as far as I know arrays in bash, sh and ksh can only be indexed by integers, so you have to approach it differently... try this perhaps:

tot=0
n=0
sort -t';' -k 1,1 -k 2,2 /input/file | while IFS=';' read a b c d e f g
do
        if [[ "$prev" != "" && "$a;$b" != "$prev" ]]
        then
                echo "$prev:$n;$tot"
                tot=0
                n=0
        fi
        ((n=n+1))
        ((tot=tot+$g))
        prev="$a;$b"
done
echo "$prev:$n;$tot"

Incidentally, I think there's an error on the second line of your awk script; shouldn't the index of l have a semi-colon too, rather than a colon?

      { n[$1";"$2]++; l[$1"[COLOR=red];[/color]"$2]+=$7 }

Annihilannic.

 

[feherke]

Hi

If you want associative arrays in [tt]bash[/tt], take a look at the libbash project. Its install package contains a library, hashstash. That may help you.

( Note, that I never used it so I know nothing more about it. I just suppose it can be quite slow as it is based on [tt]eval[/tt]. )

Feherke.

http://rootshell.be/~feherke/