Shell Programming and Scripting

OK. If you've pre-verified the input file, you can strip my suggestion down to:

Code:

#!/bin/ksh
awk '/[^[:digit:]]/ { h[++hc] = $0; next}
{       o[hc,++oc[hc]] = $0}
END {   printf("%-3s%s\n", h[1], h[2])
        for(i = 1; i <= oc[1]; i++) printf("%s %s\n", o[1, i], o[2, i])
}' Input

and get the same output.

Code:

pr -t -s' ' -2 -l 7 input

Code:

awk '/[a-zA-Z]/{i=0}
{i++;A[i]=A[i]?A[i]"\t"$0:$0}
END{for(j=1;j<=i;j++){print A[j]}}' file

Code:

awk '/[^0-9]/{f=NR} f>1{print A[NR-f+1],$0; next} {A[NR]=$0}' file

You can just add

Code:

sed 's/ /\t/g'

to the end to replace the space to tab or you can edit the original code to be like this:

Code:

cat test | tr '\n' '\t' | awk ' { for (i=1; i<=NF/2; i++) { print $i, "\t", $(i+(NF/2)), "\n" } }'

I see people give you the codes but I think it will be more benefit to you if you're doing it yourself. So, instead of give you the code, I would rather show you one of many ways on how to do it. You can split the file into 2 files, column_a and column_b, then merge them back to a single file with two columns. You can do that first by "grep" for where column b starts with the "-n" option to show line number. Then with the combination of using "head" and "tail" with a "-number" based on your "grep" result (you cannot use that exact number, why?), you can split the file in to two files: column_a and column_b. You can then use the "paste" command to merger the two files in to one with 2 columns, separated by tab.

You can also add "pre" check stuffs to check the input file is correct such as the line where column b start is right in the middle of the file.

Have fun writing your codes :-)

Toiday

You're quite right about the need to understand the options available rather than just using code snippets. Meanwhile I'm still at the cut-and-paste stage of my scripting learning curve so thanks again to all those who replied to my original post.