File Coverage

blib/lib/Math/Round/Var.pm

Criterion	Covered	Total	%
statement	63	77	81.8
branch	15	26	57.6
condition	2	9	22.2
subroutine	12	14	85.7
pod	2	2	100.0
total	94	128	73.4

package Math::Round::Var;
$VERSION=v1.0.0;

use 5.006;
use strict;
use warnings;
use Carp;

=head1 NAME

Math::Round::Var - Variations on rounding.

=head1 SYNOPSIS

Simple decimal rounding:

  use Math::Round::Var;
  my $rnd = Math::Round::Var->new(0.01);
  # rounds to two decimal places:
  my $num = 399886.758673;
  $num = $rnd->round($num);
  print "$num\n"; # 399886.76

Now it all makes sense.

  #!/usr/bin/perl
  use Math::Round::Var;
  my $scheme = shift; # let user specify the rounding
  my $num1 = shift;
  my $rnd = Math::Round::Var->new($scheme);
  my $num2 = $rnd->round($num1);
  print "$num1 rounds to $num2 according to scheme: $scheme\n";

=head1 DESCRIPTION

This module gives you the ability to round numbers to either decimal or
fractional precision while encapsulating the rounding precision in an
object.  This allows scripts and modules to maintain multiple precision
values as objects.

It also implements flexible scheme parsing, so that your programs and
modules can offload the how-to-round decisions to this module.

=head1 AUTHOR

Eric L. Wilhelm @

http://scratchcomputing.com

=head1 COPYRIGHT

This module is copyright (C) 2003-2008 by Eric L. Wilhelm.

=head1 LICENSE

This module is distributed under the same terms as Perl.  See the Perl
source package for details.

You may use this software under one of the following licenses:

  (1) GNU General Public License
    (found at http://www.gnu.org/copyleft/gpl.html)
  (2) Artistic License
    (found at http://www.perl.com/pub/language/misc/Artistic.html)

=cut
########################################################################

=head1 Front-End Constructor

The Math::Round::Var->new() constructor only decides between the
sub-packages based on the format of your precision argument.

This is the extent of the purpose of the Math::Round::Var class.

=head2 new

  Math::Round::Var->new($precision);

=cut
sub new {
	my $caller = shift;
	my $precision = shift;
	# decide which to use
	my ($type, $count) = format_of($precision);
	if($type eq "fraction") {
		return(Math::Round::Var::Fraction->new(round_to => $precision));
	}
	elsif($type eq "decimal") {
		return(Math::Round::Var::Float->new(precision => $count));
	}
	elsif($type eq "fake") {
		return(Math::Round::Var::Fake->new());
	}
	else {
		die("$type is not a valid rounding type");
	}
} # end subroutine new definition
########################################################################

=head2 format_of

Returns "decimal" or "fraction" for $type based on the format of
$precision.  If $type is "decimal", then $count will be the number of
digits to use.

  my ($type, $count) = format_of($precision);

Valid formats should be any of the number formats which are used by
Perl.  Basically, the 'fraction' methods will work for anything (as long
as Perl can divide by it), but we would be wasting time if we only want
to round to a certain decimal place.

Fractional Formats:

Anything which does not reduce to a 'multiple of 10'.

  0.125
  0.00007
  2
  2.885

Decimal Formats:

Anything which can be expressed as 1.0e.

  0.0000001
  1.0e-10

Number-of-Digits (Decimal) Format:

Anything which matches the /^d\d+$/ pattern will be used as a 'digit
count'.

  d0
  d5
  d60 # bad idea, but valid

Fake Format:

Anything less than zero.

  -1
  -0.001

=cut
sub format_of {
	my ($prec) = @_;
	my $frac = "fraction";
	my $dec = "decimal";
	# short-circuit decimal:
	if($prec =~ m/^d(\d+)$/) {
		my $count = $1;
		return($dec, $count);
	}
	# we may want to round off like %0.0f
	unless($prec) {
		defined($prec) or carp("assuming round-to-integer");
		return($dec, 0);
	}
	if($prec < 0) {
		return('fake');
	}
	# if we want to round by numbers which are larger than 1, we must
	# use the fractional methods:
	if($prec >= 1) {
		return($frac, $prec);
	}
	# seems that the easiest way is to divide by a big number so that it
	# is guaranteed to be in exponential notation, then we simply have
	# to look at what comes before the 'e'
	my $num = $prec;
	# assumption is that this accomplishes the transform:
	$num /= 1e+4;
	if($num =~ m/^(.*?)e-(\d+)$/) {
		my ($n, $d) = ($1, $2);
		# print "number: $n\n", "digits: $d\n";
		if($n == 1) {
			return($dec, $d - 4);
		}
		else {
			return($frac, 0);
		}
	}
	else {
		croak("$prec tricks me ($num)");
	}

} # end subroutine format_of definition
########################################################################

package Math::Round::Var::Float;

use Carp;

=head1 Decimal-based rounding

=head2 new

Creates a new decimal-based rounding object.

  Math::Round::Var::Float->new(precision => 7);

The argument to precision is the number of digits to use in rounding.
This is used as part of a sprintf() format.

=cut
sub new {
	my $caller = shift;
	my $class = ref($caller) || $caller;
	my $self = {@_};
	my $p = $self->{precision};
	defined($p) or croak("must define 'precision'");
	($p =~ m/^\d+$/) or croak("precision must be an integer");
	bless($self, $class);
	return($self);
} # end subroutine new definition
########################################################################

=head2 round

  $number = $rounder->round($number);

=cut
sub round {
	my $self = shift;
	my $rnd = $self->{precision};
	my $number = shift;
	return(sprintf("%0.${rnd}f", $number));
} # end subroutine round definition
########################################################################

package Math::Round::Var::Fraction;

use Carp;

=head1 Fraction-based rounding.

=head2 new

  Math::Round::Var::Fraction->new();

=cut
sub new {
	my $caller = shift;
	my $class = ref($caller) || $caller;
	my $self = {@_};
	my $r = $self->{round_to};
	defined($r) or croak("must define 'round_to'");
	bless($self, $class);
	return($self);
} # end subroutine new definition
########################################################################

=head2 round

  $number = $rounder->round($number);

=cut
sub round {
	my $self = shift;
	my $rnd = $self->{round_to};
	my $number = shift;
	return(sprintf("%0.0f", $number / $rnd) * $rnd);
} # end subroutine round definition
########################################################################

package Math::Round::Var::Fake;


=head1 Fake rounding

This mode doesn't round at all.  This is useful when you need user-input
to be able to disable rounding without rewriting a lot of code.

=head2 new

  Math::Round::Var::Fake->new();

=cut
sub new {
	my $caller = shift;
	my $class = ref($caller) || $caller;
	my $self = {@_};
	bless($self, $class);
	return($self);
} # end subroutine new definition
########################################################################

=head2 round

  $fake->round();

=cut
sub round {
	my $self = shift;
	return(shift);
} # end subroutine round definition
########################################################################
1;