I think something like this is what you want,

use Data::Dumper;
my %kid;
while (<DATA>) {
    chomp;
    my @line = split /,/;
    $kid{$line[1]}{$line[2]}{$line[3]}
        = { duration      => $line[4],
            function_name => $line[5] };
}
print Dumper(\%kid);
__DATA__
1,2,3,4,100,A
10,20,30,40,200,B
11,21,31,41,300,C
12,22,32,42,400,D
13,23,33,43,500,E
13,23,33,53,600,F
13,23,33,63,700,G
13,23,34,73,800,H
13,23,34,83,900,I
13,24,35,93,1000,J
13,24,36,103,1100,K
[download]

Deep hashes like this are not the easiest thing to work with. Care must be taken to check existence of a key at each level while looking things up. From the feel of this sample problem, it's possible that Graph will make a more comfortable representation of the data.

Hi Zaxo,

Many thanks for your time and effort deciphering my post and posting, what on the face of it, looks like an excellent reply. This looks like exactly what I want, I must give it a try. I kinda thought this problem would either be very hard/impossible to solve (very unlikely!)or have a concise elegant solution like you've posted.

I intend to have a play and may come back with some more questions, if thats ok?

Thanks
Pat

Your description is a little distracted and does not really make explicit how you want to access the data. Making a few assumptions, perhaps this is close to what you want. I am using a hash, of array refs, which contain one or more hash refs to cope with you duplicate '?' thinging. ie in the struct there is a key '13' that contains an array ref. This array ref contains a series of hashes that contain the data pertaining to the '13' == ? thingy(s). Call it a link list. Literally < 10 minutes in Perl, hours in C/C++. It will be fast but it will burn memory. If you have the memory no worries. If not 1GB =~ 3 hours coding time in business case terms.

This program was my first real foray into Perl also and the script (and my Perl knowledge) has been slowly evolving over the last 3 years.

If I understand you correctly, you are starting with a 2-D array (a flat table) such that in each row of 6 columns (0-5): column 0 is meaningless, columns 1-3 combine to make up a primary key, and columns 4 and 5 are the data of interest for each primary key.

If all rows are unique in terms of the tuples formed by columns 1-3, then you really only need a simple hash, using the concatenation of these three fields as the hash key -- e.g., building on Zaxo's example:

my %table;

while (<DATA>) {
    chomp;
    my @line = split /,/;
    my $key = join ",", @line[1..3];
    if ( exists( $table{$key} )) {
        warn "oops -- duplicate key $key... is there a problem with th
+at?\n";
        next;  # just in case there's a problem with that
    }
    $table{$key} = join ",", @line[4,5];
}
# later on, if you really need the individual columns from the
# hash key or the hash value, just use "split /,/" as needed

# <update>
# note that you can easily search for groups of hash keys:

my @root23keys = sort grep /^23,/, keys %table;
my @r23p34keys = sort grep /^23,34,/, keys %table;

# and so on... </update>


__DATA__
1,2,3,4,100,A
10,20,30,40,200,B
11,21,31,41,300,C
12,22,32,42,400,D
13,23,33,43,500,E
13,23,33,53,600,F
13,23,33,63,700,G
13,23,34,73,800,H
13,23,34,83,900,I
13,24,35,93,1000,J
13,24,36,103,1100,K
[download]

Hi graff,

Thanks for your time. Looked at and played with the code and it concisely and simply does what I want.

Thanks
Pat