Perl Programming help

tuxtutorials · 03-30-2009, 06:54 AM

Hello, linuxquestion users. I have the following Perl problem I have been banging away at and have been stuck with for a long time. The issue is with a some basic search and replacement within a file.

Here is the code I got so far. First I tried opening the file then I slurped it in for experimentation. My goal was to patter match "flushing-" store it into an array then loop through the file and replace "Windows" in Windows-service with the stored "location" number.

Code:

$IN = "$INFILE";
$filename = "services.cfg";


#open ($IN, "$filename") || die "Error reading file $!";

#while (<$IN>) {
        #if (/(Windows-/) {
        #$match = "$1";
        #}

open my $file_handle, 'services.cfg' || die "file can't be slurped:\n$!";

local $/;   # Set input to "slurp" mode.
my @string = <$file_handle>;


foreach (@string) {
if (@string =~ /(Windows-)/) {
$match = "$1";
}

print $match;

}



close $file_handle;

Here is what my data looks like originally:

Code:


define service {
        service_description             Win_Term-Service_Check
        use                             Windows-Service
        host_name                       kentucky-srvboxen2
        check_command                   check_win_trm_svcs
}


define service {
        service_description             Check_NRPE_disk_D:
        use                             Windows-Service
        host_name                       alabama-srv0022
        check_command                   Check_NRPE_disk_D:
}


define service {
        service_description             Check_NRPE_Win_CPULOAD
        use                             Windows-Service
        host_name                       flushing-srv001
        check_command                   Check_NRPE_Win_CPULOAD
}

This is the service file for or Nagios.cfg. I basically need to strip out the location such as flushing, alabama, kentucky and replace the "Windows-service" with "location-service." My final Services.cfg file with look like so:

Code:

define service {
        service_description             Check_NRPE_Win_CPULOAD
        use                             flushing-Service
        host_name                       flushing-srv001
        check_command                   Check_NRPE_Win_CPULOAD
}

Is this best done in Perl or should other utilities such as sed be used?

Thanks for all the help

Sergei Steshenko · 03-30-2009, 07:12 AM

There is too much confusion in your code. For example:

Code:

local $/;   # Set input to "slurp" mode.
my @string = <$file_handle>;

Reread info on $/ and on how data from file handles is read.

And in Perl you can do almost anything and much more than in sed/awk.

And this:

Code:

if (@string =~ /(Windows-)/)

is also wrong - reread info on matching, regular expressions, replacement, array vs scalar context.

ghostdog74 · 03-30-2009, 08:47 AM

awk

Code:

awk '
/use/{
   getline line
   m=split(line,a," ")
   n=split(a[2],b,"-")
   gsub("Windows",b[1])
   print ; print line;next 
}1'  config.cfg

Sergei Steshenko · 03-30-2009, 09:53 AM

Quote:

Originally Posted by ghostdog74

awk

Code:

awk '
/use/{
   getline line
   m=split(line,a," ")
   n=split(a[2],b,"-")
   gsub("Windows",b[1])
   print ; print line;next 
}1'  config.cfg

And if the/an OP makes mistakes in awk code, you'll give him/her a working solution in Python/Perl :-) ?

I'm talking about giving fish or teaching to fish.

tuxtutorials · 03-30-2009, 10:18 AM

Thanks for the help guys let me stumble around with it some more and I will try to post back with a updated solution.

Telemachos · 03-30-2009, 11:29 AM

I will give you one tip. I don't think you really want slurp mode. That is, you don't want to save the entire file to a single scalar. That would make it very unwieldy to work with. Consider reading by paragraphs and then splitting on newlines, as a contrast:

Code:

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

open my $fh, 'services.cfg' or die "Can't open 'services.cfg': $!";

my @records;

{
  local $/ = "";    # Set for paragraph reading
  while (<$fh>) {
    my @records = split /\n/, $_;
    # Insert code here to get the place name from $record[3]
    # Insert code here to put the item found from $record[3] into $record[2]
    foreach my $line (@records) {
      print "$line\n";
    }
    print "\n";
  }
}

This gives output like the following:

Code:

hektor ~/practice $ perl fixer
define service {
        service_description             Win_Term-Service_Check
        use                             Windows-Service
        host_name                       kentucky-srvboxen2
        check_command                   check_win_trm_svcs
}

define service {
        service_description             Check_NRPE_disk_D:
        use                             Windows-Service
        host_name                       alabama-srv0022
        check_command                   Check_NRPE_disk_D:
}

define service {
        service_description             Check_NRPE_Win_CPULOAD
        use                             Windows-Service
        host_name                       flushing-srv001
        check_command                   Check_NRPE_Win_CPULOAD
}

Since you know the structure of the records this well, you can work on the two items you care about easily now. You know that the line with the name you need is in $record[3] and that the one you want to edit is in $record[2].

Telemachos · 03-30-2009, 12:40 PM

Double post. Sorry.

ghostdog74 · 03-30-2009, 07:22 PM

Quote:

Originally Posted by Sergei Steshenko

I'm talking about giving fish or teaching to fish.

well, can i trust you to give this advice to any other forummers who post answers in full code next time?

Oh, also i would also expect you to be a role model and only give hints to answers next time for every post you make?