Shell Programming and Scripting

My problem is kind of wired:

I am trying to store a awk string in a variable, like


var="awk 'BEGIN{FS=OFS=\"|\";a=\"JAN,FEB,MAR,APR,MAY,JUN,JUL,AUG,SEP,OCT,NOV,DEC\"; split(a,m,\",\"); for(i=1;i<13;i++) n[m[i]]=i}{split($97,d,\"-\");$97=sprintf(\"%02d/%s/%s\",n[d[2]],d[1],d[3])} {split($6,d,\"-\");$6=sprintf(\"%02d/%s/%s\",n[d[2]],d[1],d[3])} {split($5,d,\"-\");$5=sprintf(\"%02d/%s/%s\",n[d[2]],d[1],d[3])} {print}' text.txt > result.txt"

this awk command is working fine ie if i run this awk it gives the desired output.
The awk command is:
awk 'BEGIN{FS=OFS="|";a="JAN,FEB,MAR,APR,MAY,JUN,JUL,AUG,SEP,OCT,NOV,DEC"; split(a,m,","); for(i=1;i<13;i++) n[m[i]]=i}{split($97,d,"-");$97=sprintf("%02d/%s/%s",n[d[2]],d[1],d[3])} {split($6,d,"-");$6=sprintf("%02d/%s/%s",n[d[2]],d[1],d[3])} {split($5,d,"-");$5=sprintf("%02d/%s/%s",n[d[2]],d[1],d[3])}{print}' text.txt > result.txt


But whenever i do $var. It gives the following error:
awk: cmd. line:1: 'BEGIN{FS=OFS="|";a="JAN,FEB,MAR,APR,MAY,JUN,JUL,AUG,SEP,OCT,NOV,DEC";
awk: cmd. line:1: ^ invalid char ''' in expression

It completely amuses me.
Any help would be welcome

Someone any help on this!

It doesn't amuse me that you didn't bother mentioning what does "whenever i do $var" mean? Maybe a bit of code "doing" $var can help?

folder='/home/xx/yy'
var=(`grep .csv ${folder}/aa/bb.txt | tr ':' ' '`)
for((i=1;i<${#var[*]};i++))
do
fun=( "{split(\$${var[$i]},d,\"-\");\$${var[$i]}=sprintf(\"%02d/%s/%s\",n[d[2]],d[1],d[3])}" )
final="$fun $final"
done
field=("awk 'BEGIN{FS=OFS=\"|\";a=\"JAN,FEB,MAR,APR,MAY,JUN,JUL,AUG,SEP,OCT,NOV,DEC\";split(a,m,\",\");
for(i=1;i<13;i++)
n[m[i]]=i}")
tot=("$field $final {print}' text.txt > result.txt")
$tot

this is the code.
The shell script generates the error, while the stored string in tot when executed separately works fine.

Sample of ${folder}/aa/bb.txt would be useful since you are trying to build an awk script dynamically and its near impossible to debug w/o the data.

ok..bb.txt will have data as:

filename:3:8:28

The numbers specify the fields that need to be changed.

Hence fun will have:
{split(\$28,d,\"-\");\$28=sprintf(\"%02d/%s/%s\",n[d[2]],d[1],d[3])}
and accordingly..

I would do something like this:

Code:

#!/usr/bin/bash

#file='/home/xx/yy/aa/bb.txt'
file=$PWD/bb.txt

{
cat <<AW'K'
awk 'BEGIN{FS=OFS="|";a="JAN,FEB,MAR,APR,MAY,JUN,JUL,AUG,SEP,OCT,NOV,DEC";
      split(a,m,","); for(i=1;i<13;i++) n[m[i]]=i;}{
AWK
grep .csv $file |cut -d":" -f2-|tr ':' ' ' |while read entry ; do
      for i in $entry ; do
cat <<AWK
split(\$$i,d,"-"); \$$i=sprintf("%02d/%s/%s",n[d[2]],d[1],d[3]);
AWK
done
done
echo "print }' text.txt > result.txt"
} > awk.sh
chmod 755 awk.sh

awk.sh
cat result.txt

This runs w/o errors. This will need to be modified since using test.txt doesn't make sense to me since you are looking up the .csv filename and the columns for date transformation from bb.txt so it would make sense to do that for each .csv.

I suppose I've given you enough to work it out yourself.