Shell Programming and Scripting

I am am one of these people that it isn't good enough just to say, "Here, try this...". it is important for me to understand how and why something works (or doesn't work.)

All that being said, I am trying to parse out a command that we use that was handed down to me by someone I can no longer ask questions from, and I **think** I have most of it, but not quite.

The output of this routine prints a list of amount of space used by each subdirectory, sorted by size, low to high.

The command/routine looks like so:
(and it works like a charm, btw)

Code:

du -h | perl -e 'sub h{%h=(K=>10,M=>20,G=>30);($n,$u)=shift=~/([0-9.]+)(\D)/;
return $n*2**$h{$u}}print sort{h($a)<=>h($b)}<>;'

From what I can muddle out, this routine:
1. Runs the "du -h" command
2. Pipes the results into a subroutine executed by perl
3. Splits each line of the output into two pieces of data each (size and subdirectory) and puts that into a hash
3. Does some kind of math on the size of each element in the hash (the "**" throws me off, and I can't locate info on what that means)
4. And prints the result with subdirectory, sorted by size, ascending

Do I have this even close to being right?

Thanks in advance for any feedback.

The ** is an exponential. 2 ** 10 is 2 to the 10th power, or 1024.

It's looking up the values of K, M, and G in the hash and getting 10, 20, 30 to calculate the sizes of 2^10, 2^20, and 2^30 for kilobytes, megabytes, and gigabytes respectively. Each line gets pushed into an array, and finally, the giant array is sorted, and printed.

Pretty close -- but I have no idea why they made it so hard on themselves. If you don't want to bother translating K/M/G values, don't ask for them in the first place.

Code:

# smallest to largest
du | sort -n
# largest to smallest
du | sort -rn

I see that sort of thing a lot. People will script ls -l all the time, then use a three-deep pipe chain to strip out everything but the filename... emulating plain ls with no flags.

Thank you!!!

That was driving me BANANAS!!

I like having the K/M/G displayed. Makes it pretty tidy.

Thanks again!

I can see that, but why sort after? It means doing convert, un-convert, sort, re-convert. Why not sort the machine-readable, and print the human-readable?

Code:

du -B 1| sort -n | 
        awk '{ N=1; while((N<4) && (($1+0) > 1024)) { N++ ;  $1 /= 1024 }
        $1=sprintf("%d%s", $1, substr(" KMG", N,1)); } 1' OFS="\t" -

It divides by 1024 until it's either less than 1024, or big enough it's run out of letters. It picks space, K, M, or G depending on the number of times it divided.