The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.
 / 
Decision-ParseTree-0.041
River stage zero No dependents
++
/ Decision::ParseTree

NAME

Decision::ParseTree - Replacing waterfall IF-ELSIF-ELSE blocks

VERSION

Version 0.041

SYNOPSIS

Death to long if-elsif-else blocks that are hard to maintain, and hard to explain to your manager. Heres an overly simplistic example:

OLD CODE

   if ( $obj->is_numeric ) {
      if ( $obj->is_positive ) {
         print 'Positive Number';
      } 
      elsif ( $obj->is_negative ) 
         print 'Negative Number';
      }
      else {
         print 'Looks like zero';
      }
   else {
      print 'Non-Numeric Value';
   }

NEW CODE

YAML Decision Tree

   ---
      - is_num : 
         0 : Non-Numeric Value
         1 : - is_pos :
               1 : Positive Number
             - is_neg :
               = : Looks like zero
               1 : Negative Number
   ...

Rules Object

   package Rules;
   use Scalar::Util;
  
   sub is_num { 
      my ( $self, $obj ) = @_;
      return (Scalar::Util::looks_like_number($obj->{value})) ? 1 : 0; 
   }

   sub is_pos { 
      my ( $self, $obj ) = @_;
      return ($obj->{value} > 0 ) ? 1 : 0; 
   }

   sub is_neg { 
      my ( $self, $obj ) = @_;
      return ($obj->{value} < 0 ) ? 1 : 0; 
   }

Goal Object to be passed thru the rules

   package Number;
   
   sub new {
      my ( $class, $value ) = @_
      my $self = { parse_path => [], 
                   value => $value };
      return bless $self, $class;
   }

Replacement to that if-else block

   use Decision::ParseTree q{ParseTree};

   my $rules = Rules->new;
   my $tree  = LoadFile('tree.yaml');
   
   print ParseTree( $tree, $rules, Number->new(10) ); # Positive Number
   print ParseTree( $tree, $rules, Number->new(-1) ); # Negative Number
   print ParseTree( $tree, $rules, Number->new(0)  ); # Looks like zero
   print ParseTree( $tree, $rules, Number->new('a')); # Non-Numeric Value

DESCRIPTION

YAML as a Decision tree

To make this all work we need a few parts:

  • A rules object: This will be a library of rules.

  • An object that will be passed thru the rules.

  • A YAML doc that outlines your decision tree.

Why YAML

So this all started as a way to make a decision tree thats easy to parse and easy to read for non-programmers. So to do this I looked to YAML, it's easy to read and easy to parse. Though make this work we have some hard and fast rules to follow for the tree construction:

  • RULES are a key value pair

    • the key is the method to run in the rules object

    • the value must be an arrayref or hashref

  • ARRAYS are a series of rules run in order

  • HASHES are a series of answers

  • SCALARS are endpoints

Why add more parts, why blow everything in to separate objects.

Sometimes you have to make things messy before they can get clean.

Theres a flexibility that comes with breaking things apart in to nice, neat little chunks. By separating the rule logic in to one place you can make very complex rules that do not gunk up your code. You pull the order of these rules in to another place as it's completely possible that you would want to tweak the order. And lastly you need to glue these separate things together, so you have an object that gets passed thru to make this all work. Tada!

Examples

It would be nice to whip up a big example here to show all the interesting bits, sadly I can't think of a good example. Ideas?

  • Selecting a tests to run for hardware

  • Building settings/configuration files on the fly for varried hardware.

  • Would any one like to use this to write up a GO AI engine? Chess?

FEATURES

  • tracking for free

    • If $obj->{parse_path} exists then every step that this obj takes thru the rules will be tracked. This path will be stored as an array ref, of hash refs.

         $obj = Number->new(10);
   ParseTree( $tree, $rules, $obj );
   # $obj->{parse_path} will now look like :
   #   [  { 'is_num' => 1 },
   #      { 'is_pos' => 1 },
   #   ]

    • If $obj->{parse_answer} exists then, when an answer is found, then it gets stored here as well as being returned.

        print $obj->{parse_answer}; # Positive Number

EXPORT OK-able

ParseTree is the only thing that can get exported, it's also the only thing in here, so export away.

FUNCTIONS

ParseTree($tree, $rules, $obj)

Runs $obj thru $tree, using $rules as the library of rules.

Returns the first endpoint that you run into as the answer.

CAVEATS / TODO

  • Currently $tree is expected to be a pre-parsed YAML File, This should change here soon to also accept a filename. Currently though it does not.

  • would like even more examples.

  • need to flush out the docs more.

AUTHOR

ben hengst, <notbenh at cpan.org>

BUGS

Please report any bugs or feature requests to bug-decision-parsetree at rt.cpan.org, or through the web interface at http://rt.cpan.org/NoAuth/ReportBug.html?Queue=Decision-ParseTree. I will be notified, and then you'll automatically be notified of progress on your bug as I make changes.

SUPPORT

You can find documentation for this module with the perldoc command.

    perldoc Decision::ParseTree

You can also look for information at:

ACKNOWLEDGEMENTS

COPYRIGHT & LICENSE

Copyright 2007 ben hengst, all rights reserved.

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