Anonymous Monk has asked for the wisdom of the Perl Monks concerning the following question:

hi,
Anyone to have idea how can I create a such a structure that act like a 4way-HASH i.e. I can query back and fourth 4-way meta-keys . Example :
the four key-types are : mac, ip, file, name So now I want to be able to say : 
print $hash{$name1}{ip};#name->ip conversation

print $hash{$ip5}{name};#ip->name

print $hash{$mac10}{file};#mac->filename

print $hash{$mac}{ip};#mac -> ip

[download]
etc.....did u get the idea ... The problem is that it is useless and time consuming to populate such a HASH structure..
I can make a hash based only on one key-type, say "name" and for the rest to use foreach cycles to find a match, but this also seems not as good to me...

ANY IDEA

Replies are listed 'Best First'.
Re: compound structure? 4way-hash
by Aristotle (Chancellor) on Aug 26, 2002 at 12:51 UTC
    That isn't directly possible. You can store the data in an array of hashes and use four hashes with references to the entries. 
    my @address;
my %by_name = map { ($_->{name}, $_) } @address;
my %by_file = map { ($_->{file}, $_) } @address;
my %by_mac = map { ($_->{mac}, $_) } @address;
my %by_ip = map { ($_->{ip}, $_) } @address;

print $by_ip{"1.2.3.4"}{name}, $by_name{foobar}, "\n";

    [download]
    The drawback of course is that whenever the @address data changes you have to update the index hashes accordingly.

    Makeshifts last the longest.

[reply]
[d/l]
Re: compound structure? 4way-hash
by fglock (Vicar) on Aug 26, 2002 at 12:39 UTC

    You can use a subroutine:

    get_info(mac=>'0494944545', want=>'file');

    If you really need a hash, transform the subroutine into a hash using "tie".

    $hash{mac_04994994_file};

    (hint) Use split("_") to separate the words. It is much easier than simulating a hash-of-hashes.

[reply]
[d/l]
[select]
Re: compound structure? 4way-hash
by Anonymous Monk on Aug 26, 2002 at 13:39 UTC
    I'm parsing a dhcpd.conf file and expect the number of entires to reach probably 1000 at long turn.. At the moment I don't look so much at speed, but will need it may be !!!
    At the moment there is only one case when I will resolv-all entiries (mac->name)...

    my code up to the moment is :
    ......

sub parse {
 my $self = shift;
  #parse the classes
 my @clases = $$self{_data} =~ /class\s+?"(.+?)"\s+?\{.*?=\s*(.+?)\s*;
+/gs;
 while ($#clases > 0) {
    $$self{_clases}{pop @clases} = pop @clases;
 }
 
 my @hosts = $$self{_data} =~ /
    host\s*(.+?)\s*\{.*?
     filename\s*"(.+?)"\s*;.*?
     ethernet\s*(.+?)\s*;.*?
     fixed-address\s*(.+?)\s*;
 /xgs; 
 while ($#hosts > 0) {
    my $h = shift @hosts;
   ($$self{_hosts}{$h}{file}, $$self{_hosts}{$h}{mac}, 
     $$self{_hosts}{$h}{ip})  = splice @hosts,0,3;
 }
}

sub name2 {
 my ($self, $name, $to) = @_;
 $$self{_hosts}{$name}{$to}
}

1;

    [download]
    One thing on which i would like to hear your comment is the while cycles ... did u see "my $h", if i try this : 
       (my $h, $$self{_hosts}{$h}{file}, $$self{_hosts}{$h}{mac}, 
     $$self{_hosts}{$h}{ip})  = splice @hosts,0,4;

    [download]
    it wont work... not a big deal but it is interesting to know why ?
[reply]
[d/l]
[select]
      How often is this script running? "1000 lines" doesn't say much. If you run it once a day, 1000 lines is a pretty miniscule burden.

      Makeshifts last the longest.

[reply]
Re: compound structure? 4way-hash
by BrowserUk (Patriarch) on Aug 26, 2002 at 14:06 UTC

    To be really generic, you'd need to add combinations algorithm to generate the 2 from n indices at the top, but 2 from 4 is easier to do by hand. 

    #! perl -w
use strict;
use Data::Dumper;

# if the number of fields was to vary you could generate the combinati
+ons using an algorithm.
my @sets = (
    [1,2,3],     # 1st (index 0) field can be combined with these othe
+r fields.
    [0,2,3],    # 2nd with these...
    [0,1,3],
    [0,1,2]
);

my %hash;

my @labels = split/\s+/, <DATA>;    # read the labels from the first l
+ine of the DATA...

while (<DATA>) {
    chomp;
    my @fields = split;
    
    for my $index (0 .. $#sets) {
        
        $hash{ $labels[$index] }{ $fields[$_] } =
            $fields[$index] 
                for (@{$sets[$index]}[0 .. 2]);
    }
}
print Dumper(\%hash);

__DATA__
mac                ip        name        file
ABCDEFABCDEF10    10.0.0.1    www.foo1.com    afilename1
ABCDEFABCDEF02    10.0.0.2    www.foo2.com    afilename2
ABCDEFABCDEF03    10.0.0.3    www.foo3.com    afilename3
ABCDEFABCDEF04    10.0.0.4    www.foo4.com    afilename4
ABCDEFABCDEF05    10.0.0.5    www.foo5.com    afilename5

    [download]

    When run produces the following output.

    Read more... (5 kB)
[reply]
[d/l]
[select]
Re: compound structure? 4way-hash
by derby (Abbot) on Aug 26, 2002 at 12:32 UTC
    Hmmm ... RDBMS qualities without the database overhead, perhaps some of the "lite" DBD drivers would be helpful: DBD::SQLite or DBD::CSV.

    -derby

[reply]

Back to Seekers of Perl Wisdom