On 2017-09-21 12:01, Wyatt Zacharias wrote:

So I've got a storage appliance that I'm trying to generate a growth chart on, but the output format of the statistics is not in a usable format.

The output looks like this, with each day of the week in a column, and then 6 rows of data underneath, and then it repeats.

-2- -3- -4- -5- -6- -7- -8- 5827.1 6865.6 3551.2 4649.5 4006.6 15803.5 10305.0 199.4 353.4 175.9 200.7 172.5 584.0 554.2 9.9x 4.4x 6.4x 5.9x 6.3x 5.6x 5.3x 3.0x 4.4x 3.2x 3.9x 3.7x 4.9x 3.5x 29.2x 19.4x 20.2x 23.2x 23.2x 27.1x 18.6x 96.6 94.9 95.0 95.7 95.7 96.3 94.6

-23- -24- -25- -26- -27- -28- -29- 2798.2 2235.3 2357.9 4701.3 9074.3 13796.5 11705.7 86.7 74.1 131.6 178.1 449.3 483.5 473.7 11.4x 12.0x 7.0x 8.1x 3.9x 6.3x 6.7x 2.8x 2.5x 2.6x 3.2x 5.2x 4.6x 3.7x 32.3x 30.2x 17.9x 26.4x 20.2x 28.5x 24.7x 96.9 96.7 94.4 96.2 95.0 96.5 96.0

-30- -31- -1- -2- -3- -4- -5- 2798.6 2274.5 2325.8 2472.1 2526.0 13955.3 11224.2 86.7 88.9 145.2 115.2 105.6 497.9 432.4 10.8x 10.1x 7.4x 8.1x 9.8x 6.2x 6.9x 3.0x 2.5x 2.2x 2.7x 2.4x 4.5x 3.8x 32.3x 25.6x 16.0x 21.5x 23.9x 28.0x 26.0x 96.9 96.1 93.8 95.3 95.8 96.4 96.1

Is there a way I can get awk to process 6 rows of each column at a time?

Not particularly... awk is a single line filter, doing multi-line leads to madness. But you can use the line number to gauge/tract state.

Crappy awk pseudo code...

BEGIN {state counter = 0 } ^- { Line starts with ^- it's a header line, set state counter to 0. } ^[0-9]+ & state counter == 0 { next line starts with a number and state counter is 0 so do first line thing, increment state counter } ^[0-9]+ & state counter == 1 { next line starts with a number and state counter is 1 so do second line thing, increment state counter } ... ^[0-9]+ & state counter == 5 { next line starts with a number and state counter is 5 so last line in the sequence, set counter to 0. } repeat END { output your fancy stats }

You can then use a data structure like an array or a 2d matrix to handle the data you output at the end.

(PS. Yes I know you can nest the stanzas but trying to keep the suggestion clear)

-- Sean

On 2017-09-21 Wyatt Zacharias wrote:

So I've got a storage appliance that I'm trying to generate a growth chart on, but the output format of the statistics is not in a usable format.

The output looks like this, with each day of the week in a column, and then 6 rows of data underneath, and then it repeats. -2- -3- -4- -5- -6- -7- -8- 5827.1 6865.6 3551.2 4649.5 4006.6 15803.5 10305.0 199.4 353.4 175.9 200.7 172.5 584.0 554.2

Like John said, this is perfect for perl. And perl is available on every system, even HPUX, and it doesn't even need to be a modern perl. You've probably already solved it, but I'll throw a perl solution out there for kicks:

# build a sane data structure of what you want while (<>) { next if /^\s*$/;

$redates='\s+-(\d+)-'x7; # build the regex @dates=/$redates/,$stanza++,next if /$redates/;

$renums='\s*([0-9.x]+)'x7; if (@nums=/$renums/) { foreach $i (0..6) { push(@{$table{sprintf "%02d%02d",$stanza,$dates[$i]}},$nums[$i]); } } }

# now do what you want on the data foreach $stanzadate (sort keys %table) { print "$stanzadate: ".join(' , ',@{$table{$stanzadate}})."\n";

# do something with it # foreach $entry (@{$table{$stanzadate}}) { $sum+=$entry... whatever } }

It was a little more complicated because the dates can (and do) repeat, so the hash key I made a combination key so it still sorts nicely.

I can also see a way to do it in plain bash with head and cut, and maybe paste? But my bash looping syntax skillz stink, so I leave that as an exercise.