package Set::IntSpan::Fast; use warnings; use strict; use Carp; use Data::Types qw(is_int); use List::Util qw(min max); =head1 NAME Set::IntSpan::Fast - Fast handling of sets containing integer spans. =head1 VERSION This document describes Set::IntSpan::Fast version 1.11 =cut use vars qw( $VERSION ); $VERSION = '1.11'; =head1 SYNOPSIS use Set::IntSpan::Fast; my $set = Set::IntSpan::Fast->new(); $set->add(1, 3, 5, 7, 9); $set->add_range(100, 1_000_000); print $set->as_string(), "\n"; # prints 1,3,5,7,9,100-1000000 =head1 DESCRIPTION C represents sets of integers. It is optimised for sets that contain contiguous runs of values. 1-1000, 2000-10000 # Efficiently handled Sets that don't have this characteristic may still be represented but some of the performance and storage space advantages will be lost. Consider using bit vectors if your set does not typically contain clusters of values. Sets may be infinite - assuming you're prepared to accept that infinity is actually no more than a fairly large integer. Specifically the constants C and C are defined to be -(2^31-1) and (2^31-2) respectively. To create an infinite set invert an empty one: my $inf = Set::IntSpan::Fast->new()->complement(); Sets need only be bounded in one direction - for example this is the set of all positive integers (assuming you accept the slightly feeble definition of infinity we're using): my $pos_int = Set::IntSpan::Fast->new(); $pos_int->add_range(1, $pos_int->POSITIVE_INFINITY); =head2 Set representation The internal representation used is extremely simple: a set is represented as a list of integers. Integers in even numbered positions (0, 2, 4 etc) represent the start of a run of numbers while those in odd numbered positions represent the ends of runs. As an example the set (1, 3-7, 9, 11, 12) would be represented internally as (1, 2, 3, 8, 11, 13). =head2 Comparision with Set::IntSpan The C module represents sets of integers as a number of inclusive ranges, for example '1-10,19-23,45-48'. Because many of its operations involve linear searches of the list of ranges its overall performance tends to be proportional to the number of distinct ranges. This is fine for small sets but suffers compared to other possible set representations (bit vectors, hash keys) when the number of ranges grows large. This module also represents sets as ranges of values but stores those ranges in order and uses a binary search for many internal operations so that overall performance tends towards O log N where N is the number of ranges. =head1 INTERFACE =cut use constant POSITIVE_INFINITY => 2**31 - 2; use constant NEGATIVE_INFINITY => -2**31 + 100; =head2 C Create a new set. Any arguments will be processed by a call to C: my $set = Set::IntSpan::Fast->new( '1, 3, 5, 10-100' ); Because C handles multiple arguments this will work: my @nums = ( 1, 2, 3, 4, 5 ); my $set = Set::IntSpan::Fast->new( @nums ); Bear in mind though that this validates each element of the array is it would if you called C so for large sets it will be slightly more efficient to create an empty set and then call C. =cut sub new { my $class = shift; my $self = bless [], $class; $self->add_from_string( @_ ) if @_; return $self; } =head2 C Complement the set. Because our notion of infinity is actually disappointingly finite inverting a finite set results in another finite set. For example inverting the empty set makes it contain all the integers between C and C inclusive. As noted above C and C are actually just big integers. =cut sub invert { my $self = shift; if ( $self->is_empty() ) { # Empty set @$self = ( NEGATIVE_INFINITY, POSITIVE_INFINITY ); } else { # Either add or remove infinity from each end. The net # effect is always an even number of additions and deletions if ( $self->[0] == NEGATIVE_INFINITY ) { shift @{$self}; } else { unshift @{$self}, NEGATIVE_INFINITY; } if ( $self->[-1] == POSITIVE_INFINITY ) { pop @{$self}; } else { push @{$self}, POSITIVE_INFINITY; } } } =head2 C Return an identical copy of the set. my $new_set = $set->copy(); =cut sub copy { my $self = shift; my $copy = Set::IntSpan::Fast->new(); @$copy = @$self; return $copy; } =head2 C Add the specified integers to the set. Any number of arguments may be specified in any order. All arguments must be integers between C and C inclusive. =cut sub add { my $self = shift; $self->add_range( _list_to_ranges( @_ ) ); } =head2 C Remove the specified integers from the set. It is not an error to remove non-members. Any number of arguments may be specified. =cut sub remove { my $self = shift; $self->remove_range( _list_to_ranges( @_ ) ); } =head2 C Add the inclusive range of integers to the set. Multiple ranges may be specified: $set->add_range(1, 10, 20, 22, 15, 17); Each pair of arguments constitute a range. The second argument in each pair must be greater than or equal to the first. =cut sub add_range { my $self = shift; _iterate_ranges( @_, sub { my ( $from, $to ) = @_; my $fpos = $self->_find_pos( $from ); my $tpos = $self->_find_pos( $to + 1, $fpos ); $from = $self->[ --$fpos ] if ( $fpos & 1 ); $to = $self->[ $tpos++ ] if ( $tpos & 1 ); splice @$self, $fpos, $tpos - $fpos, ( $from, $to ); } ); } =head2 C Add items to a set from a string representation, of the same form as C. Multiple strings may be supplied: $set->add_from_string( '1-10, 30-40', '100-200' ); is equivalent to $set->add_from_string( '1-10, 30-40, 100-200' ); By default items are separated by ',' and ranges delimited by '-'. You may select different punctuation like this: $set->add_from_string( { sep => ';', range => ':' }, '1;3;5;7:11;19:27' ); When supplying an options hash in this way the C and C option may be either a regular expression or a literal string. $set->add_from_string( { sep => qr/:+/, range => qr/[.]+/ }, '1::3::5:7...11:19..27' ); And embedded whitespace in the string will be ignored. =cut sub add_from_string { my $self = shift; my $ctl = {}; my $match_number = qr/\s* (-?\d+) \s*/x; my $match_single = qr/^ $match_number $/x; my $match_range; my @to_add = (); # Iterate args. Default punctuation spec prepended. for my $el ( { sep => qr/,/, range => qr/-/, }, @_ ) { # Allow parsing options to be set. if ( 'HASH' eq ref $el ) { %$ctl = ( %$ctl, %$el ); for ( values %$ctl ) { $_ = quotemeta( $_ ) unless ref $_ eq 'Regexp'; } $match_range = qr/^ $match_number $ctl->{range} $match_number $/x; } else { for my $part ( split $ctl->{sep}, $el ) { if ( my ( $start, $end ) = ( $part =~ $match_range ) ) { push @to_add, $start, $end; } elsif ( my ( $el ) = ( $part =~ $match_single ) ) { push @to_add, $el, $el; } else { croak "Invalid range string" unless $part =~ $match_single; } } } } $self->add_range( @to_add ); } =head2 C Remove the inclusive range of integers from the set. Multiple ranges may be specified: $set->remove_range(1, 10, 20, 22, 15, 17); Each pair of arguments constitute a range. The second argument in each pair must be greater than or equal to the first. =cut sub remove_range { my $self = shift; $self->invert(); $self->add_range( @_ ); $self->invert(); } =head2 C Remove items to a set from a string representation, of the same form as C. As with C the punctuation characters may be specified. =cut sub remove_from_string { my $self = shift; $self->invert(); $self->add_from_string( @_ ); $self->invert(); } =head2 C Merge the members of the supplied sets into this set. Any number of sets may be supplied as arguments. =cut sub merge { my $self = shift; for my $other ( @_ ) { my $iter = $other->iterate_runs(); while ( my ( $from, $to ) = $iter->() ) { $self->add_range( $from, $to ); } } } =head2 Operators =head3 C Returns a new set that is the complement of this set. See the comments about our definition of infinity above. =cut sub compliment { croak "That's very kind of you - but I expect you meant complement()"; } sub complement { my $new = shift->copy(); $new->invert(); return $new; } =head3 C Return a new set that is the union of this set and all of the supplied sets. May be called either as a method: $un = $set->union( $other_set ); or as a function: $un = Set::IntSpan::Fast::union( $set1, $set2, $set3 ); =cut sub union { my $new = Set::IntSpan::Fast->new(); $new->merge( @_ ); return $new; } =head3 C Return a new set that is the intersection of this set and all the supplied sets. May be called either as a method: $in = $set->intersection( $other_set ); or as a function: $in = Set::IntSpan::Fast::intersection( $set1, $set2, $set3 ); =cut sub intersection { my $new = Set::IntSpan::Fast->new(); $new->merge( map { $_->complement() } @_ ); $new->invert(); return $new; } =head3 C Return a new set that contains all of the members that are in this set or the supplied set but not both. Can actually handle more than two sets in which case it returns a set that contains all the members that are in some of the sets but not all of the sets. Can be called as a method or a function. =cut sub xor { return intersection( union( @_ ), intersection( @_ )->complement() ); } =head3 C Return a set containing all the elements that are in this set but not the supplied set. =cut sub diff { my $first = shift; return intersection( $first, union( @_ )->complement() ); } =head2 Tests =head3 C Return true if the set is empty. =cut sub is_empty { my $self = shift; return @$self == 0; } =head3 C Return true if the specified number is contained in the set. =cut *contains = *contains_all; =head3 C Return true if the set contains any of the specified numbers. =cut sub contains_any { my $self = shift; for my $i ( @_ ) { my $pos = $self->_find_pos( $i + 1 ); return 1 if $pos & 1; } return; } =head3 C Return true if the set contains all of the specified numbers. =cut sub contains_all { my $self = shift; for my $i ( @_ ) { my $pos = $self->_find_pos( $i + 1 ); return unless $pos & 1; } return 1; } =head3 C Return true if all the numbers in the range C<$low> to C<$high> (inclusive) are in the set. =cut sub contains_all_range { my ( $self, $lo, $hi ) = @_; croak "Range limits must be in ascending order" if $lo > $hi; my $pos = $self->_find_pos( $lo + 1 ); return ( $pos & 1 ) && $hi < $self->[$pos]; } =head3 C Returns the number of members in the set. If a clipping range is supplied return the count of members that fall within that inclusive range. =cut sub cardinality { my $self = shift; my $card = 0; my $iter = $self->iterate_runs( @_ ); while ( my ( $from, $to ) = $iter->() ) { $card += $to - $from + 1; } return $card; } =head3 C Returns true if this set is a superset of the supplied set. A set is always a superset of itself, or in other words $set->superset( $set ) returns true. =cut sub superset { return subset( reverse( @_ ) ); } =head3 C Returns true if this set is a subset of the supplied set. A set is always a subset of itself, or in other words $set->subset( $set ) returns true. =cut sub subset { my $self = shift; my $other = shift || croak "I need two sets to compare"; return $self->equals( $self->intersection( $other ) ); } =head3 C Returns true if this set is identical to the supplied set. =cut sub equals { return unless @_; # Array of array refs my @edges = @_; my $medge = scalar( @edges ) - 1; POS: for ( my $pos = 0;; $pos++ ) { my $v = $edges[0]->[$pos]; if ( defined( $v ) ) { for ( @edges[ 1 .. $medge ] ) { my $vv = $_->[$pos]; return unless defined( $vv ) && $vv == $v; } } else { for ( @edges[ 1 .. $medge ] ) { return if defined $_->[$pos]; } } last POS unless defined( $v ); } return 1; } =head2 Getting set contents =cut =head3 C Return an array containing all the members of the set in ascending order. =cut sub as_array { my $self = shift; my @ar = (); my $iter = $self->iterate_runs(); while ( my ( $from, $to ) = $iter->() ) { push @ar, ( $from .. $to ); } return @ar; } =head3 C Return a string representation of the set. my $set = Set::IntSpan::Fast->new(); $set->add(1, 3, 5, 7, 9); $set->add_range(100, 1_000_000); print $set->as_string(), "\n"; # prints 1,3,5,7,9,100-1000000 You may optionally supply a hash containing C and C options: print $set->as_string({ sep => ';', range => '*' ), "\n"; # prints 1;3;5;7;9;100*1000000 =cut sub as_string { my $self = shift; my $ctl = { sep => ',', range => '-' }; %$ctl = ( %$ctl, %{ $_[0] } ) if @_; my $iter = $self->iterate_runs(); my @runs = (); while ( my ( $from, $to ) = $iter->() ) { push @runs, $from == $to ? $from : join( $ctl->{range}, $from, $to ); } return join( $ctl->{sep}, @runs ); } =head3 C Returns an iterator that returns each run of integers in the set in ascending order. To iterate all the members of the set do something like this: my $iter = $set->iterate_runs(); while (my ( $from, $to ) = $iter->()) { for my $member ($from .. $to) { print "$member\n"; } } If a clipping range is specified only those members that fall within the range will be returned. =cut sub iterate_runs { my $self = shift; if ( @_ ) { # Clipped iterator my ( $clip_lo, $clip_hi ) = @_; my $pos = $self->_find_pos( $clip_lo ) & ~1; my $limit = ( $self->_find_pos( $clip_hi + 1, $pos ) + 1 ) & ~1; return sub { TRY: { return if $pos >= $limit; my @r = ( $self->[$pos], $self->[ $pos + 1 ] - 1 ); $pos += 2; # Catch some edge cases redo TRY if $r[1] < $clip_lo; return if $r[0] > $clip_hi; # Clip to range $r[0] = $clip_lo if $r[0] < $clip_lo; $r[1] = $clip_hi if $r[1] > $clip_hi; return @r; } }; } else { # Unclipped iterator my $pos = 0; my $limit = scalar( @$self ); return sub { return if $pos >= $limit; my @r = ( $self->[$pos], $self->[ $pos + 1 ] - 1 ); $pos += 2; return @r; }; } } sub _list_to_ranges { my @list = sort { $a <=> $b } @_; my @ranges = (); my $count = scalar( @list ); my $pos = 0; while ( $pos < $count ) { my $end = $pos + 1; $end++ while $end < $count && $list[$end] <= $list[ $end - 1 ] + 1; push @ranges, ( $list[$pos], $list[ $end - 1 ] ); $pos = $end; } return @ranges; } # Return the index of the first element >= the supplied value. If the # supplied value is larger than any element in the list the returned # value will be equal to the size of the list. sub _find_pos { my $self = shift; my $val = shift; my $low = shift || 0; my $high = scalar( @$self ); while ( $low < $high ) { my $mid = int( ( $low + $high ) / 2 ); if ( $val < $self->[$mid] ) { $high = $mid; } elsif ( $val > $self->[$mid] ) { $low = $mid + 1; } else { return $mid; } } return $low; } sub _iterate_ranges { my $cb = pop @_; my $count = scalar( @_ ); croak "Range list must have an even number of elements" if ( $count % 2 ) != 0; for ( my $p = 0; $p < $count; $p += 2 ) { my ( $from, $to ) = ( $_[$p], $_[ $p + 1 ] ); croak "Range limits must be integers" unless is_int( $from ) && is_int( $to ); croak "Range limits must be in ascending order" unless $from <= $to; croak "Value out of range" unless $from >= NEGATIVE_INFINITY && $to <= POSITIVE_INFINITY; # Internally we store inclusive/exclusive ranges to # simplify comparisons, hence '$to + 1' $cb->( $from, $to + 1 ); } } 1; __END__ =head2 Constants The constants C and C are exposed. As noted above these are infinitely smaller than infinity but they're the best we've got. They're not exported into the caller's namespace so if you want to use them you'll have to use their fully qualified names: $set->add_range(1, Set::IntSpan::Fast::POSITIVE_INFINITY); =head1 DIAGNOSTICS =head3 C<< Range list must have an even number of elements >> The lists of ranges passed to C and C consist of a number of pairs of integers each of which specify the start and end of a range. =head3 C<< Range limits must be integers >> You may only add integers to sets. =head3 C<< Range limits must be in ascending order >> When specifying a range in a call to C or C the range bounds must be in ascending order. Multiple ranges don't need to be in any particular order. =head3 C<< Value out of range >> Sets may only contain values in the range C to C inclusive. =head3 C<< That's very kind of you - but I expect you meant complement() >> The method that complements a set is called C. =head3 C<< I need two sets to compare >> C and C need two sets to compare. They may be called either as a function: $ss = Set::IntSpan::Fast::subset( $s1, $s2 ) or as a method: $ss = $s1->subset( $s2 ); =head3 C<< Invalid Range String >> The range string must only contain a comma separated list of ranges, with a hyphen used as the range limit separator. e.g. "1,5,8-12,15-29". =head1 CONFIGURATION AND ENVIRONMENT =for author to fill in: A full explanation of any configuration system(s) used by the module, including the names and locations of any configuration files, and the meaning of any environment variables or properties that can be set. These descriptions must also include details of any configuration language used. Set::IntSpan::Fast requires no configuration files or environment variables. =head1 DEPENDENCIES Data::Types List::Util =head1 INCOMPATIBILITIES Although this module was conceived as a replacement for C it isn't a drop-in replacement. =head1 BUGS AND LIMITATIONS No bugs have been reported. Please report any bugs or feature requests to C, or through the web interface at L. =head1 AUTHOR Andy Armstrong C<< >> =head1 CREDITS K. J. Cheetham L<< http://www.shadowcatsystems.co.uk/ >> for add_from_string, remove_from_string. I butchered his code so any errors are mine. =head1 LICENCE AND COPYRIGHT Copyright (c) 2006, Andy Armstrong C<< >>. All rights reserved. This module is free software; you can redistribute it and/or modify it under the same terms as Perl itself. See L. =head1 DISCLAIMER OF WARRANTY BECAUSE THIS SOFTWARE IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE SOFTWARE, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE SOFTWARE "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE SOFTWARE IS WITH YOU. SHOULD THE SOFTWARE PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR, OR CORRECTION. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE SOFTWARE AS PERMITTED BY THE ABOVE LICENCE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE SOFTWARE (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE SOFTWARE TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.