awk pairwise differences

Question

I run into this problem a lot with data, if I have some integers, I frequently want to look at differences between adjacent integers, I usually solve this in Ruby or Python, but I think such a thing could be done in awk, and I'd prefer that.

I have a table of data such as:

201309,694
201310,699
201311,700
201312,705
201401,713
201402,740

And I would like to find the adjacent differences, that is:

201310-201309,699-694
201311-201310,700-699
201312-201311,705-700
201401-201312,713-715
201402-201401,740-713

I found the function getline in the awk manpage, but I haven't been able to use it successfully.

Answer 1

This one-liner produces the output with the minus signs:

awk -F, '{if (NR>1) {print $1 "-" a "," $2 "-" b} a=$1 ; b=$2}' numbers
201310-201309,699-694
201311-201310,700-699
201312-201311,705-700
201401-201312,713-705
201402-201401,740-713

This version shows the actual differences:

awk -F, '{if (NR>1) {print $1-a, $2-b } a=$1 ; b=$2}' numbers
1 5
1 1
1 5
89 8
1 27

In both cases, the awk program begins with an option to set the field separator to a comma (-F,). Then, for each line in the file, an if statement is executed: if we are past the first line ((NR>1) then the differences are printed. The next two commands (a=$1 ; b=$2) update the variables that store the last values.

Answer 2

No need for getline. Just keep track of the columns in the AWK script, e.g.

#!/usr/bin/awk -f

BEGIN {
    FS="," # We separate on the comma character
}

{
    if(NR > 0) { ## NR is the current line number
            print $1 "-" old[1] "," $2 "-" old[2]
    }
    for(field = 1; field <= 2; field++) {
            old[field] = $(field)
    }
}

I assume you want to post "thisline_field1-lastline_field1,thisline_field2-lastline_field2" as a literal string; if you want to subtract, it would be print $1 - old[1] "-" $2 - old[2]