[SOLVED] I have commandline want to embed in a script

markotitel · 05-03-2010, 08:11 PM

Hello,

Im having trouble inserting a command with output as shown

Quote:

[root@server test]# squidstat --host 10.20.64.14 --port 8080 | awk '/active IPs:/ {print $6}' | sed '$s/,$//'
11
[root@server test]#

Now i want thaat in a perl script i tried something like this

Quote:

#!/usr/bin/perl

open(PROCESS, "squidstat --host 10.20.64.14 --port 8080 | awk '/active IPs:/ {print $6}' | sed '$s/","$//' |");
$output = <PROCESS>;
close(PROCESS);
chomp($output);
print $output;

but it wont work

.
I have no programing experience so any help is welcome

Sergei Steshenko · 05-03-2010, 08:30 PM

Quote:

Originally Posted by markotitel

Hello,

Im having trouble inserting a command with output as shown

Now i want thaat in a perl script i tried something like this

but it wont work

.
I have no programing experience so any help is welcome

Start by putting

Code:

use strict;
use warnings;

just after

Code:

#!/usr/bin/perl

and make sure your script compiles and produces no warnings during run.

markotitel · 05-03-2010, 08:48 PM

Hello,

thanks for answer now I have this

Quote:

#!/usr/bin/perl
use strict;
use warnings;

open(PROCESS, "squidstat --host 10.20.64.14 --port 8080 | awk '/active IPs:/ {printf $6}' | sed '$s/,$//' |");
$output = <PROCESS>;
close(PROCESS);
chomp($output);
print $output;

And i get error like this

Quote:

Global symbol "$s" requires explicit package name at unix_processes.pl line 5.
Global symbol "$output" requires explicit package name at unix_processes.pl line 6.
Global symbol "$output" requires explicit package name at unix_processes.pl line 8.
Global symbol "$output" requires explicit package name at unix_processes.pl line 9.

Sergei Steshenko · 05-03-2010, 09:38 PM

Quote:

Originally Posted by markotitel

Hello,

thanks for answer now I have this

And i get error like this

So, fix the errors - read about global and lexical variables and about the way variables are interpolated in double quoted strings.

For example, start from visiting

http://perldoc.perl.org/search.html?q=lexical+variable
.

markotitel · 05-04-2010, 07:06 AM

Thanks ive got something like this that works maybe it is not by the rules but It serves my purposes

Quote:

#!/usr/bin/perl

my $sq = `squidstat --host 10.20.64.14 --port 8080`;
$sq = `echo "$sq" | awk '/active IPs:/'`;
$sq = `echo "$sq" | awk '\{printf \$6 \}'`;
$sq =~ s/,//;
chomp $sq;
print "users:$sq";

grail · 05-04-2010, 08:02 AM

I must say I am curious why it takes two awk and a sed to get what you require??
Maybe if you show what squidstat returns you we could tidy that up? (only if you want of course

)

PMP · 05-04-2010, 08:16 AM

Why do need awk inside Perl !!
Perl itself can do these stuff for you!!

grail · 05-04-2010, 08:34 AM

@PMP - and that is definitely another point ...

markotitel · 05-04-2010, 08:47 AM

Ok thank you for interest in my question, here is output of squidstat

Quote:

[root@server scripts]# squidstat --host 10.20.64.14 --port 8080

Active connections: 56, active IPs: 22, average speed: 302.6Kb/s

IP: 10.20.64.211 (size: 290Mb 14Kb, speed: 98.1Kb/s)
http://data44.sevenload.com/p2/slcom...vwllljinnh.flv

IP: 10.20.64.20 (size: 104Mb 472Kb, speed: 45.1Kb/s)
188.27.67.47:443
http://2.channel40.facebook.com/x/34...p_1661507833=0
http://1.channel40.facebook.com/x/32...p_1661507833=0
http://3.channel44.facebook.com/x/10...000179569334=4
http://0.channel40.facebook.com/x/15...p_1661507833=0

Output is a little biger but what do I need is ACTIVE IPS value, I want to graph that value in cacti.
If you can show me nicer and cleaner way please help me.

grail · 05-04-2010, 09:00 AM

Code:

$sq=`echo $sq | awk -F: '/active/{gsub(/^[ ]*|,.*/,"",$3);print $3}'`

But as PMP said, Perl is one of the masters of regex (not me unfortunately) and can probably do this cleaner all in the same language.

I hope this helps though

Sergei Steshenko · 05-04-2010, 09:36 AM

At all, when it's in Perl, no sed/awk/grep is needed.

grail · 05-04-2010, 10:09 AM

@Sergei - I agree, perhaps you would show us the equivalent in Perl (as I have said I am not so familiar)

Sergei Steshenko · 05-04-2010, 10:23 AM

Quote:

Originally Posted by grail

@Sergei - I agree, perhaps you would show us the equivalent in Perl (as I have said I am not so familiar)

For that I would need to learn 'awk' in the first place

.

...

If one needs to extract 22 from the line below:

Code:

Active connections: 56, active IPs: 22, average speed: 302.6Kb/s

, then in Perl one can write it this way:

Code:

my $stdout = `squidstat --host 10.20.64.14 --port 8080`;

if($stdout =~ m/\bActive\s+connections\:\s+\d+,\s+active\s+IPs\:\s+(\d+)/s)
  {
  my $number_of_active_connections = $1;
  warn "\$number_of_active_connections=$number_of_active_connections";
  }
else
  {
  warn "unexpected \$stdout=$stdout";
  }

- I haven't checked the code.

PMP · 05-04-2010, 01:12 PM

Quote:

Originally Posted by Sergei Steshenko

For that I would need to learn 'awk' in the first place

.

...

If one needs to extract 22 from the line below:

Code:

Active connections: 56, active IPs: 22, average speed: 302.6Kb/s

, then in Perl one can write it this way:

Code:

my $stdout = `squidstat --host 10.20.64.14 --port 8080`;

if($stdout =~ m/\bActive\s+connections\:\s+\d+,\s+active\s+IPs\:\s+(\d+)/s)
  {
  my $number_of_active_connections = $1;
  warn "\$number_of_active_connections=$number_of_active_connections";
  }
else
  {
  warn "unexpected \$stdout=$stdout";
  }

- I haven't checked the code.

Do we really need such a big pattern ?
only this should be sufficient, not tested though.

Code:

if ($stdout =~ /active\s+IPs\:\s+(\d+)/)
{
..
}

For this particular example line
Is it really required to use regex ?
I can see I can work with split only

Sergei Steshenko · 05-04-2010, 01:23 PM

Quote:

Originally Posted by PMP

Do we really need such a big pattern ?
only this should be sufficient, not tested though.

Code:

if ($stdout =~ /active\s+IPs\:\s+(\d+)/)
{
..
}

For this particular example line
Is it really required to use regex ?
I can see I can work with split only

First you need to identify the line, don't you ? So, how are you going to identify the line in the first place ?

Regarding the pattern - I wanted to identify the line as unambiguously as possible.