Shell Programming and Scripting

Thank you! That finally explains it.

I'm guessing the 27.46 is a typo then?

Sorry, that output should be 24.5625 -81.7708 27.16 1
not
24.5625 -81.7708 27.46 1

You can sort everything by value and just let awk decide which belongs to which:

Code:

REF="test/190005.pnt"

sort -k3 -n test/{1900..2016}05.pnt |
        awk '
        # Read the values you want to rank from the first file
        # This trick works by NR and FNR only being the same while reading
        # the first file, not the second.
        NR==FNR {  A[$1,$2]=$3 ; next }
        # Read everything sorted by value, counting order per prefix
        # as it goes.  When a matching value is found, print its order.
        ++C[$1,$2] && A[$1,$2] == $3 { print $1, $2, C[$1,$2] }
' $REF -

Thanks Corona. The scripts works as expected.

However there is an added intricacy that I need help with.

For ties - I need the script to rank that tied value the same.

For instance, If there is a rank of 1, 2, 3 but 4, 5, and 6 are tied. The ranks would be 1, 2, 3, 4, 4, 4, 7,....

To add to the complexity, I need to rank at both tails of the range. That way if the tie is on the upper half of the ranks, it would assume the higher value.

For instance, if there is a rank of 100, 99, 98, but 97 and 96 are tied. The ranks would be 100, 99, 98, 97, 97, 95,.....

In order to do this the total number of ranks would need to be divided by 2 to find the midpoint. Any ties less than the midpoint would assume the lower rank. While any ties more than the midpoint would assume the higher rank.

The total number of ranks could be found in the number of files in the sort statement - In this case it would be

Code:

({1900..2016} + 1) / 2 = 59

Is this something that can be done in awk?

So you want it to trawl through all the ties and pick the "tie" which has the most repeated values in common?

The output that I currently have is great, I just need to add a layer of complexity to it. In that I need all the ties to be ranked as a tie.

This snipit does just that

Code:

awk '$3 != prev { rank = NR }{ print $0, "   "rank; prev = $3 }'

However, what I also need the script to do is to rank both ends of the values.

For instance

Code:

VALUE  |    RANK
100           1
200           2
200           2
300           4
400           5
500           6
600           8
600           8
700           9
800          10

Notice how the ties in the top half (100-400) default to the lower rank, while the ties in the bottom half (500-800) default to the higher rank.

What if changing one tie breaks another? Which wins?