Read ini files without section names

I want to make a configuration file which hold some objects, like this (where of course none of the paramaters can be considered as a primary key)

param1=abc
param2=ghj

param1=bcd
param2=hjk
; always the sames parameters

This file could be read, lets say with Config::IniFiles, because it has a direct transcription into ini file, like this

[0]
param1=abc
param2=ghj

[1]
param1=bcd
param2=hjk

with, for example, something like

perl -pe 'if (m/^\s*$/ || !$section ) print "[", ($section++ || 0) , "]"'

And finish with

open my $fh, '<', "/path/to/config_file.ini" or die $!;
$cfg = Config::IniFiles->new( -file => $fh );
(...parse here the sections starting with 0.)

But, I here ask me some question about the thing becoming quite complex....

(A) Is There a way to transform the $fh, so that it is not required to execute the perl one-liner BEFORE reading the file sequentially? So, to transform the file during perl is actually reading it.

(B) Is there a module to read my wonderfull flat database? Or something approching? I let myslef said, that Gnu coreutils does this kind of flat file reading, but I cannot remember how.

Solution

You can create a simple subclass of Config::INI::Reader:

package MyReader;

use strict;
use warnings;

use base 'Config::INI::Reader';

sub new {
    my $class = shift;
    my $self = $class->SUPER::new( @_ );

    $self->{section} = 0;

    return $self;
}


sub starting_section { 0 };

sub can_ignore { 0 };

sub parse_section_header {
     my ( $self, $line ) = @_;

    return $line =~ /^\s*$/ ? ++$self->{section} : undef ;
}

1;

With your input this gives:

% perl -MMyReader -MData::Dumper -e 'print Dumper( MyReader->read_file("cfg") )'
$VAR1 = {
          '1' => {
                   'param2' => 'hjk',
                   'param1' => 'bcd'
                 },
          '0' => {
                   'param2' => 'ghj',
                   'param1' => 'abc'
                 }
        };