#!/bin/ksh
PCF=".Product_Counts"
SCF=".Skew_Counts"
# Variable dictionary:
# cmd                           command string to be used to sort product and
#                               skew count files
# ec                            exit code
# i                             loop control
# ndp                           # of different products in top 3 products
# nds                           # of different skews in top 5 skews for a
#                               given top 3 product
# P                             current product name
# p["product"]                  # of times "product" appears in 1st field
# pcf                           sorted product count filename
# pl[p["product"],plc["product"]] list of skew values associated with "product"
# plc["product"]                # of skew values associated with "product"
# ppc                           previous product count
# psc                           previous skew count
# r1, r2                        return code from some function
# S                             current skew
# s["product","skew"]           # of times "skew" appears with "product"
# scf                           sorted skew count filename
awk -v pcf="$PCF" -v scf="$SCF" '
BEGIN { FS = OFS = "|"
}
{       # Process input data...
        # Increment # of times we have seen this product.
        p[$1]++
        if(!(($1, $2) in s))
                # Add a new skew for this product.
                pl[$1, ++plc[$1]] = $2
        # Increment # of times we have seen this skew with this product.
        s[$1, $2]++
}
END {   # Sort the product counts.
        cmd = "sort -t \"|\" -k2,2nr -k1,1 -o " pcf
        for(i in p) print i, p[i] | cmd
        close(cmd)
        ppc = 0
        # Set the sort command to be used to process the skew counts.
        cmd = "sort -t \"|\" -k2,2nr -k1,1 -o " scf
        # Process the top 3 products.
        while((r1 = (getline < pcf)) == 1) {
                # Increment count of top 3 products, but include more if the
                # number of hits is the same for later products
                if(++ndp > 3 && ppc != $2) break
                P = $1
                ppc = $2
                printf("%d hits for product: %s\n", ppc, P)
                # Sort the skew counts for this product.
                for(i = 1; i <= plc[P]; i++)
                        print pl[P, i], s[P, pl[P, i]] | cmd
                close(cmd)
                # Process the top 5 skews for this product.
                nds = 0
                while((r2 = (getline < scf)) == 1) {
                        # Increment count of top 5 skews, but include more if
                        # the number of hits is the same for later skews.
                        if(++nds > 5 && psc != $2) break
                        S = $1
                        psc = $2
                        printf("\t%d hits for skew: %s\n", psc, S)
                }
                if(r2 < 0) {
                        printf("Error reading top 5 skew list from \"%s\".\n",
                                scf)
                        ec = 1
                }
                close(scf)
        }
        if(r1 < 0) {
                printf("Error reading top 3 product list from "%s".\n", pcf)
                ec = 1
        }
        close(pcf)
        exit(ec + 0)
}' file
if [ $? -eq 0 ]
then    # awk completed successfully...
        # Remove sort output files.
        rm "$PCF" "$SCF"
        exit
fi
exit 1

19 hits for product: p4
	9 hits for skew: 98707
	4 hits for skew: 098
	3 hits for skew: 98708
	2 hits for skew: 1234
	1 hits for skew: 98706
13 hits for product: p3
	9 hits for skew: 234345
	3 hits for skew: 234
	1 hits for skew: 2343
8 hits for product: p6
	8 hits for skew: 23467