perl: process strange log file

ezekieldas · 03-25-2012, 01:23 AM

A snippet from a strange looking logfile. There may be thousands of lines like this...

Code:

2012-01-20 22:06:35 [quassi-f] LOG Start foo=a37f882dea96 op=go stat=starting file="/tmp/f823234.dat" size=23424
2012-01-20 22:06:54 [quassi] LOG Snd ll i/v/n=3221/3221/0 fio i/x/w=2/2340/000 fio a/b/c=1/1/1 af x/y/z=23/13/666 bl=9
2012-01-20 22:06:54 [quassi] LOG Snd FF oD v/d/n/p=234/5356/2324/34600

I'm concerned only about this in the 2nd line: "x/y/z=23/13/666" I want to do math with the values of x and z (in this case 23 and 666. If I can just get an array going that looks like (23,666,x,z,x,z...)

I must do this in perl (one of my weakest languages) and I'm having trouble getting that section. I've tried something like this but I'm sure I have the split() part all wrong. If you could help with that I think it'd get me going with the rest.

Code:

open(LOGFILE);
foreach $line (<LOGFILE>) {
 ($my_x, $my_z) = split(/x\/y\/z=/, $line)

  ...

}
close(LOGFILE);

Σigma · 03-25-2012, 02:13 AM

Try

Code:

my ($x, $z) = $line =~ m|x/y/z=(\d+)/\d+/(\d+)|;

ezekieldas · 03-25-2012, 03:38 PM

Thanks Σigma --that works. Can you (or someone) please explain the match used with pipes? I looked around for this and typically I see it used only like so: m/foo/

Also, I had to strip out empty lines, etc. The resulting code is:

Code:

open(LOGFILE, "/tmp/alog");

for $line (<LOGFILE>) {
 if ($line =~ /x\/y\/z/) {
  my ($tbytes, $ttime) = $line =~ m|x/y/z=(\d+)/\d+/(\d+)|;
  chomp $line;
  push(@bytesandtime, $tbytes, $ttime);
 }
}

Σigma · 03-25-2012, 04:53 PM

“the // default delimiters for a match can be changed to arbitrary delimiters by putting an 'm' out front” - http://perldoc.perl.org/perlrequick.html

Basically, by sticking an “m” in front you can delimit your regex with near anything you feel like. In this case “|” is easier because then you don't have to escape the “/”s.