NAME

Algorithm::DrillDown - Turns a long list into an easy-to-navigate tree

SYNOPSIS

my $drilldown = new Algorithm::DrillDown;
my $tree = $drilldown->generate(\@authors);
print Dump $tree;

sub myslicer {
  my($drilldown, $level, $string) = @_;
  $string =~ tr/0-9A-Za-z//cd;
  return substr lc $string, 0, $level;
}

my $drilldown_custom = new Algorithm::DrillDown
     slicer => \&myslicer,
     maxitems => 32,
     maxdepth => 8;

DESCRIPTION

This module was written because I kept tripping over lists of data that were too large and unwieldy to display directly and wanted a means of grouping them into a hierarchy that could be more easily navigated.

By way of specific example, I wanted a list of CPAN authors on a web page, but as there are over six thousand of them, putting them all on a single page was not sensible. The most obvious solution would be to just arbitrarily split after every few hundred authors and put a "more" button, but that's just icky. Wouldn't it be nicer if one could just click on "MO" (or "MOBILEART..MOXFYRE") then "MOOLI" than randomly-guess that it's probably around page eleven of twenty?

Similarly, I archive mailing lists in IMAP folders backed by a Maildir, but these folders get kind of big and unwieldy after a while and I'd like to automatically decide that an archive folder is getting too large and split it into multiple dated folders.

This object-wraps an otherwise stateless operation which takes an input list and produces a hash-of-array. The hash keys are the result of your slicer function, and the values are your input values that produced that result. The items will be placed into the least-specific grouping that will still ensure that the array is not larger than maxitems.

A specific example may help. The default slicer function is substr, so "least specific" means "shortest prefix". So if we start with something like:

my @list = qw ( AADLER AAKD AAKHTER AALLAN AANKHEN AANZLOVAR AAR AARDEN
AARDO AARE AARON AARONJJ AARONSCA AASSAD AAU AAYARS ABALAMA ABARCLAY
ABCDEFGH ABE ABELEW ABELTJE ABERGMAN ABERNDT ABEROHAM ABH ABHAS ABHIDHAR
ZTURK ZUMMO ZUQIF ZURAWSKI ZZCGUMK );

then a rather compressed/obfuscated invocation:

my $result = Algorithm::DrillDown->new(maxitems=>16)->generate(\@list);

And the end result is a structure like this:

my $result = {
    AA => [qw( AADLER, AAKD, AAKHTER, AALLAN, AANKHEN, AANZLOVAR, AAR,
               AARDEN, AARDO, AARE, AARON, AARONJJ, AARONSCA, AASSAD, AAU,
               AAYARS )],
    AB => [qw( ABALAMA, ABARCLAY, ABCDEFGH, ABE, ABELEW, ABELTJE, ABERGMAN,
               ABERNDT, ABEROHAM, ABH, ABHAS, ABHIDHAR )]
    Z => [qw( ZTURK, ZUMMO, ZUQIF, ZURAWSKI, ZZCGUMK )],
};

Note that the lists are not guaranteed to be in any particular order. (The hash is obviously unsorted too.)

FUNCTIONS

new

my $drilldown = new Algorithm::Drilldown;

sub myslicer {
  my(undef, $level, $string) = @_;
  $string =~ tr/0-9A-Za-z//cd;
  return substr lc $string, 0, $level;
}

my $drilldown_custom = new Algorithm::DrillDown
     slicer => \&myslicer,
     maxitems => 32,
     maxdepth => 8;

This creates a new Algorithm::DrillDown object. There are three possible parameters:

slicer

A reference to your slicer function, that is one that takes the Algorithm::DrillDown object, a level (an integer, starting at zero) and your input scalar (which does not have to be a string) and returns a string which summarises that input at the given level.

The default function is based on substr, and returns the string truncated to the same number of characters as the level.

maxitems

The maximum number of items to place in an output array. Note that an array may still exceed this value if maxlevel has been reached first.

maxlevel

The maximum level that will be passed to your slicer function.

generate

BUGS

This documentation isn't great.

Tests. We've heard of them.

Pathological data will cause more than maxitems items to appear in a bucket, and also this will cause it to appear at the maxdepth level. Since it works well enough for now, this is not being checked for. There are loads of possibly icky edge cases. (Note again the lack of tests.)

SEE ALSO

AUTHOR

All code and documentation by Peter Corlett <abuse@cabal.org.uk>.

COPYRIGHT

Copyright (C) 2007 Peter Corlett <abuse@cabal.org.uk>. All rights reserved.

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

SUPPORT / WARRANTY

This is free software. IT COMES WITHOUT WARRANTY OF ANY KIND.

cpanm Algorithm::DrillDown

perl -MCPAN -e shell
install Algorithm::DrillDown