# # This file is part of Language::Befunge. # Copyright (c) 2001-2008 Jerome Quelin, all rights reserved. # # This program is free software; you can redistribute it and/or modify # it under the same terms as Perl itself. # # package Language::Befunge::Vector; use strict; use warnings; use integer; use Carp; use overload '=' => \©, '+' => \&_add, '-' => \&_substract, 'neg' => \&_invert, '+=' => \&_add_inplace, '-=' => \&_substract_inplace, '<=>' => \&_compare, '""' => \&as_string; # try to load speed-up LBV eval 'use Language::Befunge::Vector::XS'; if ( defined $Language::Befunge::Vector::XS::VERSION ) { my $xsversion = $Language::Befunge::Vector::XS::VERSION; my @subs = qw[ new new_zeroes copy as_string get_dims get_component get_all_components clear set_component bounds_check _add _substract _invert _add_inplace _substract_inplace _compare ]; foreach my $sub ( @subs ) { no strict 'refs'; no warnings 'redefine'; my $lbvxs_sub = "Language::Befunge::Vector::XS::$sub"; *$sub = \&$lbvxs_sub; } # LBV::XS 1.1.0 adds rasterize() @subs = qw[ rasterize _xs_rasterize_ptr ]; if($xsversion gt "1.0.0") { # import the XS functions from LBVXS no strict 'refs'; no warnings 'redefine'; foreach my $sub (@subs) { my $lbvxs_sub = "Language::Befunge::Vector::XS::$sub"; *$sub = \&$lbvxs_sub; } } else { # export the pure-perl functions to LBVXS no strict 'refs'; no warnings 'redefine'; foreach my $sub (@subs) { my $lbvxs_sub = "Language::Befunge::Vector::XS::$sub"; *$lbvxs_sub = \&$sub; } } } # -- CONSTRUCTORS # # my $vec = LB::Vector->new( $x [, $y, ...] ); # # Create a new vector. The arguments are the actual vector data; one # integer per dimension. # sub new { my $pkg = shift; # sanity checks my $usage = "Usage: $pkg->new(\$x, ...)"; croak $usage unless scalar(@_) > 0; # regular LBV object my $self = [@_]; bless $self, $pkg; return $self; } # # my $vec = LB::Vector->new_zeroes($dims); # # Create a new vector of dimension $dims, set to the origin (all # zeroes). LBV->new_zeroes(2) is exactly equivalent to LBV->new(0, 0). # sub new_zeroes { my ($pkg, $dims) = @_; # sanity checks my $usage = "Usage: $pkg->new_zeroes(\$dimensions)"; croak $usage unless defined $dims; croak $usage unless $dims > 0; # regular LBV object my $self = [ (0) x $dims ]; bless $self, $pkg; return $self; } # # my $vec = $v->copy; # # Return a new LBV object, which has the same dimensions and coordinates # as $v. # sub copy { my $vec = shift; return bless [@$vec], ref $vec; } # -- PUBLIC METHODS #- accessors # # my $str = $vec->as_string; # my $str = "$vec"; # # Return the stringified form of $vec. For instance, a Befunge vector # might look like "(1,2)". # sub as_string { my $self = shift; return "(" . join(",",@$self) . ")"; } # # my $dims = $vec->get_dims; # # Return the number of dimensions, an integer. # sub get_dims { my $self = shift; return scalar(@$self); } # # my $val = $vec->get_component($dim); # # Get the value for dimension $dim. # sub get_component { my ($self, $dim) = @_; croak "No such dimension $dim!" unless $dim >= 0 && $self->get_dims > $dim; return $self->[$dim]; } # # my @vals = $vec->get_all_components; # # Get the values for all dimensions, in order from 0..N. # sub get_all_components { my ($self) = @_; return @$self; } # - mutators # # $vec->clear; # # Set the vector back to the origin, all 0's. # sub clear { my ($self) = @_; @$self = (0) x $self->get_dims; } # # $vec->set_component($dim, $value); # # Set the value for dimension $dim to $value. # sub set_component { my ($self, $dim, $val) = @_; croak "No such dimension $dim!" unless $dim >= 0 && $self->get_dims > $dim; $self->[$dim] = $val; } #- other methods # # my $is_within = $vec->bounds_check($begin, $end); # # Check whether $vec is within the box defined by $begin and $end. # Return 1 if vector is contained within the box, and 0 otherwise. # sub bounds_check { my ($vchk, $begin, $end) = @_; croak "uneven dimensions in bounds check!" unless $vchk->get_dims == $begin->get_dims; croak "uneven dimensions in bounds check!" unless $vchk->get_dims == $end->get_dims; for (my $d = 0; $d < $vchk->get_dims; $d++) { return 0 if $vchk->get_component($d) < $begin->get_component($d); return 0 if $vchk->get_component($d) > $end->get_component($d); } return 1; } # # $vec = $vec->rasterize($min, $max); # # Return the next vector in raster order, or undef if the hypercube space # has been fully covered. To enumerate the entire storage area, the caller # should call rasterize on the storage area's "min" value the first time, # and keep looping while the return value is defined. To enumerate a # smaller rectangle, the caller should pass in the min and max vectors # describing the rectangle, and keep looping while the return value is # defined. # sub rasterize { my ($v, $min, $max) = @_; return undef unless $v->bounds_check($min, $max); $v = $v->copy; my $nd = $v->get_dims(); for my $d (0..$nd-1) { if($v->get_component($d) >= $max->get_component($d)) { # wrap to the next highest dimension, continue loop $v->set_component($d, $min->get_component($d)); } else { # still have farther to go in this dimension. $v->set_component($d, $v->get_component($d) + 1); return $v; } } # ran out of dimensions! return undef; } # -- PRIVATE METHODS #- math ops # # my $vec = $v1->_add($v2); # my $vec = $v1 + $v2; # # Return a new LBV object, which is the result of $v1 plus $v2. # sub _add { my ($v1, $v2) = @_; my $nd = scalar @$v1; croak "uneven dimensions in vector addition!" unless $nd == scalar @$v2; return ref($v1)->new(map { $$v1[$_] + $$v2[$_] } (0..$nd-1)); } # # my $vec = $v1->_substract($v2); # my $vec = $v1 - $v2; # # Return a new LBV object, which is the result of $v1 minus $v2. # sub _substract { my ($v1, $v2) = @_; my $nd = scalar @$v1; croak "uneven dimensions in vector subtraction!" unless $nd == scalar @$v2; return ref($v1)->new(map { $$v1[$_] - $$v2[$_] } (0..$nd-1)); } # # my $v2 = $v1->_invert; # my $v2 = -$v1; # # Subtract $v1 from the origin. Effectively, gives the inverse of the # original vector. The new vector is the same distance from the origin, # in the opposite direction. # sub _invert { my ($v1) = @_; my $nd = scalar @$v1; return ref($v1)->new(map { -$_ } (@$v1)); } #- inplace math ops # # $v1->_add_inplace($v2); # $v1 += $v2; # # sub _add_inplace { my ($v1, $v2) = @_; my $nd = scalar @$v1; croak "uneven dimensions in vector addition!" unless $nd == scalar @$v2; map { $$v1[$_] += $$v2[$_] } (0..$nd-1); return $v1; } # # $v1->_substract_inplace($v2); # $v1 -= $v2; # # Substract $v2 to $v1, and stores the result back into $v1. # sub _substract_inplace { my ($v1, $v2) = @_; my $nd = scalar @$v1; croak "uneven dimensions in vector substraction!" unless $nd == scalar @$v2; map { $$v1[$_] -= $$v2[$_] } (0..$nd-1); return $v1; } #- comparison # # my $bool = $v1->_compare($v2); # my $bool = $v1 <=> $v2; # # Check whether the vectors both point at the same spot. Return 0 if they # do, 1 if they don't. # sub _compare { my ($v1, $v2) = @_; my $nd = scalar @$v1; croak "uneven dimensions in bounds check!" unless $nd == scalar @$v2; for (my $d = 0; $d < $nd; $d++) { return 1 if $$v1[$d] != $$v2[$d]; } return 0; } #- other private methods # # my $ptr = $v1->_xs_rasterize_ptr(); # # Get a pointer to the C "rasterize" function. Returns undef if LBVXS is not # loaded. This is useful for external XS modules, because calling the C # function directly is faster. # # The prototype of the C rasterize function is: # # AV *rasterize(AV *vec_array, AV *min_array, AV *max_array); # # It operates just like the perl rasterize function, and returns NULL when the # end of the loop has been reached. # sub _xs_rasterize_ptr { return undef } 1; __END__ =head1 NAME Language::Befunge::Vector - an opaque, N-dimensional vector class. =head1 SYNOPSIS my $v1 = Language::Befunge::Vector->new($x, $y, ...); my $v2 = Language::Befunge::Vector->new_zeroes($dims); =head1 DESCRIPTION This class abstracts normal vector manipulation. It lets you pass around one argument to your functions, rather than N arguments, one per dimension. This means much of your code doesn't have to care how many dimensions you're working with. You can do vector arithmetic, test for equality, or even stringify the vector to a string like I<"(1,2,3)">. =head1 CONSTRUCTORS =head2 my $vec = LB::Vector->new( $x [, $y, ...] ) Create a new vector. The arguments are the actual vector data; one integer per dimension. =head2 my $vec = LB::Vector->new_zeroes($dims); Create a new vector of dimension C<$dims>, set to the origin (all zeroes). C<< LBV->new_zeroes(2) >> is exactly equivalent to B<< LBV->new(0,0) >>. =head2 my $vec = $v->copy; Return a new LBV object, which has the same dimensions and coordinates as $v. =head1 PUBLIC METHODS =head2 my $str = $vec->as_string; Return the stringified form of C<$vec>. For instance, a Befunge vector might look like C<(1,2)>. This method is also applied to stringification, ie when one forces string context (C<"$vec">). =head2 my $dims = $vec->get_dims; Return the number of dimensions, an integer. =head2 my $val = $vec->get_component($dim); Get the value for dimension C<$dim>. =head2 my @vals = $vec->get_all_components; Get the values for all dimensions, in order from 0..N. =head2 $vec->clear; Set the vector back to the origin, all 0's. =head2 $vec->set_component($dim, $value); Set the value for dimension C<$dim> to C<$value>. =head2 my $is_within = $vec->bounds_check($begin, $end); Check whether C<$vec> is within the box defined by C<$begin> and C<$end>. Return 1 if vector is contained within the box, and 0 otherwise. =head2 $vec->rasterize($min, $max); Return the next vector in raster order, or undef if the hypercube space has been fully covered. To enumerate the entire storage area, the caller should call rasterize on the storage area's "min" value the first time, and keep looping while the return value is defined. To enumerate a smaller rectangle, the caller should pass in the min and max vectors describing the rectangle, and keep looping while the return value is defined. =head1 MATHEMATICAL OPERATIONS =head2 Standard operations One can do some maths on the vectors. Addition and substraction work as expected: my $v = $v1 + $v2; my $v = $v1 - $v2; Either operation return a new LBV object, which is the result of C<$v1> plus / minus C<$v2>. The inversion is also supported: my $v2 = -$v1; will subtracts C<$v1> from the origin, and effectively, gives the inverse of the original vector. The new vector is the same distance from the origin, in the opposite direction. =head2 Inplace operations LBV objects also supports inplace mathematical operations: $v1 += $v2; $v1 -= $v2; effectively adds / substracts C<$v2> to / from C<$v1>, and stores the result back into C<$v1>. =head2 Comparison Finally, LBV objects can be tested for equality, ie whether two vectors both point at the same spot. print "same" if $v1 == $v2; print "differ" if $v1 != $v2; =head1 PRIVATE METHODS =head2 _xs_rasterize_ptr my $ptr = $v1->_xs_rasterize_ptr(); Get a pointer to the C "rasterize" function. Returns undef if LBVXS is not loaded. This is useful for external XS modules, to allow them to call the C function directly for additional speed. The prototype of the C rasterize function is: AV *rasterize(AV *vec_array, AV *min_array, AV *max_array); It operates just like the perl rasterize function, and returns NULL when the end of the loop has been reached. =head1 SEE ALSO L =head1 AUTHOR Jerome Quelin, Ejquelin@cpan.orgE Development is discussed on Elanguage-befunge@mongueurs.netE =head1 COPYRIGHT & LICENSE Copyright (c) 2001-2008 Jerome Quelin, all rights reserved. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself. =cut