=head1 NAME Array::Each::Tutorial - POD giving various examples how to use Array::Each. =head1 VERSION This document refers to version 0.02 of Array::Each. =head1 SYNOPSIS man Array::Each man Array::Each::Tutorial or perldoc Array::Each perldoc Array::Each::Tutorial =head1 DESCRIPTION =head2 Overview This tutorial contains only POD, so don't do this: use Array::Each::Tutorial; # don't do this Rather, simply read the POD (as you are doing). But first, please read the docs for Array::Each, because the whole scoop is there. This tutorial is intended to augment those docs with examples showing situations where you might want to use Array::Each instead of other techniques. =head1 EXAMPLES =head2 Parallel Arrays vs. Using a Hash First of all, use a hash. It's almost always the best solution if you want to associate a "key" with a "value". And there are modules available that will let you do wonderful things with hashes, like keeping the keys sorted or keeping them in the order they were added. So given a hash, you might at some point want to do this: my %h = ( a=>1, b=>2, c=>3, d=>4, e=>5 ); while( my( $k, $v ) = each %h ) { # ... do something with $k and $v ... } On the other hand, if parallel arrays better implement your algorithm, then you may find you want to do something like this: my @k = qw( a b c d e ); my @v = qw( 1 2 3 4 5 ); for my $i ( 0 .. $#k ) { my( $k, $v ) = ( $k[$i], $v[$i] ); # ... do something with $k and $v (and maybe $i) ... } Using Array::Each, you could do the same thing this way: use Array::Each; my @k = qw( a b c d e ); my @v = qw( 1 2 3 4 5 ); my $obj = Array::Each->new( \@k, \@v ); while( my( $k, $v, $i ) = $obj->each ) { # ... do something with $k and $v (and maybe $i) ... } If you don't need C<$i> at all, you can leave it out, e.g., while( my( $k, $v ) = $obj->each ) { # ... do something with $k and $v ... } If you have more than two parallel arrays, include them all in the call to new() and add as many "capture" variables as you need, e.g., my @k = qw( a b c d e ); my @v = qw( 1 2 3 4 5 ); my @p = qw( - + ~ = : ); my $obj = Array::Each->new( \@k, \@v, \@p ); while( my( $k, $v, $p, $i ) = $obj->each ) { # ... do something with $k, $v, and $p (and maybe $i) ... } =head2 One Array vs. Using a Hash If you find that you can solve your problem with one array that holds pairs of data (or triplets, or quadruplets ...), you could do the following. Note that we're passing the B<< set=> >>, and B<< group=> >> named parameters to new(). # pairs my @a = ( a=>1, b=>2, c=>3, d=>4, e=>5 ); my $hash_like = Array::Each->new( set=>[\@a], group=>2 ); while( my( $k, $v, $i ) = $hash_like->each ) { # ... do something with $k and $v ... # note that $i is successively, 0, 2, 4, 6, 8 # (but see count below) } # triplets my @a = ( a=>1,'-', b=>2,'+', c=>3,'~', d=>4,'=', e=>5,':' ); my $tre = Array::Each->new( set=>[\@a], group=>3 ); while( my( $k, $v, $p, $i ) = $tre->each ) { # ... do something with $k, $v, and $p ... # note that $i is successively, 0, 3, 6, 9, 12 # (but see count below) } =head2 Groups of Elements from an Array Of course, you don't have to have a "hash-like" situation to want to iterate over multiple elements. For example, if you want to list elements five at a time, following are two options using splice: # destructive my @n = ( 1..100 ); while( my @a = splice( @n, 0, 5 ) ) { # ... do something with @a ... } # non-destructive my @n = ( 1..100 ); my @n2 = @n; # sacrificial copy while( my @a = splice( @n2, 0, 5 ) ) { # ... do something with @a ... } With Array::Each (also non-destructive): my @n = ( 1..100 ); my $obj = Array::Each->new( set=>[\@n], group=>5 ); while( my @a = $obj->each ) { my $i = pop @a; # because each returns index, too # ... do something with @a ... } Benchmarks show this to be considerably slower than using splice. However, if making a sacrificial copy is untenable, e.g., if you have a very large and/or tied array, then Array::Each may fit the bill better. =head1 SOME ATTRIBUTE COMBINATIONS This section shows how some attribute combinations will affect what each() returns. =head2 Iterator and Rewind Setting C<iterator> will determine the starting point of the very next iteration. Setting C<rewind> will determine where the rewind() method (and internal rewind operations) will rewind to. Setting them both to the same value will cause successive iterations to start at the same place as the first one. In the following hypothetical situation, the first array element has a special meaning, while the remaining ones are meaningful pairs. So we want to skip the first element when we print out the pairs by setting both C<iterator> and C<rewind> to 1. my @a = ( [d=>4], a=>1, b=>2, c=>3, d=>4 ); my $obj = Array::Each->new( set=>[\@a], iterator=>1, rewind=>1, group=>2 ); # iterator affects where this starts while( my( $k, $v ) = $obj->each ) { print "$k => $v\n"; } push @a, @{$a[0]} = ( e=>5 ); # rewind affects where this starts while( my( $k, $v ) = $obj->each ) { print "$k => $v\n"; } =head2 Group and Undef By default the C<bound> attribute is true. This means that iterations will stop when the end of the shortest (or only) array is reached. However, if C<group> is set greater than 1, and the size of the shortest array is not a multiple of the C<group> value, then the last iteration will "go beyond" the end of the shortest array. When this happens, each() will return the value of the C<undef> attribute for the "missing" array elements. By default, the C<undef> attribute is undefined. So each() will return perl's undef for "missing" or "non-existent" array elements. If this is acceptable, there is no need to set the C<undef> attribute. If not acceptable, setting C<undef> lets you specify what you want. The following example shows how you might generate a 3-column HTML table from the contents of an array. Note, C<count> is being set, so each() will return a line count instead of the (in this case useless) array index. my @a = ( 'a' .. 'm' ); my $obj = Array::Each->new( set=>[\@a], group=>3, undef=>' ', count=>1 ); print qq{<table border="1">\n}; while( my @row = $obj->each ) { printf "<tr> <td>%d.</td> ", pop @row; print map( "<td>$_</td> ", @row ), "</tr>\n"; } print "</table>\n"; =head2 Bound and Undef As stated above, C<bound> is true by default. If you set it to 0 (false), the iterations are no longer bound by the size of the shortest array. Instead they are bound by the size of the largest array (but see C<stop> to change that). So if you have more than one array, and they are not the same size, each() will return the value of C<undef> for the "missing" array elements. If the default value (perl's undef) is not acceptable, setting C<undef> lets you choose what you want. The following example shows how you might create a text table giving the sums of elements of three arrays. Because the arrays are different sizes, C<bound> is set to 0. Because we want to add "missing" elements to the totals, C<undef> is set to 0. Again, C<count> is set to give us a line count. my @a = ( [ 1..5 ], [ 1..8 ], [ 7..18 ], ); my $cols = @a; my $fmt = " %4d." . " %5d" x $cols . "\n"; my $div = ' 'x6 . ' -----' x $cols . "\n"; my $tot = ' 'x6 . " %5d" x $cols . "\n"; my @totals; my $obj = Array::Each->new( set=>[@a], bound=>0, undef=>0, count=>1 ); while( my @row = $obj->each ) { my $count = pop @row; printf $fmt, $count, @row; @totals = map { $totals[$_] += $row[$_] } ( 0 .. $#row ) } print $div; printf $tot, @totals; =head1 AUTHOR Brad Baxter, bbaxter@cpan.org =head1 COPYRIGHT Copyright (c) 2003-2004, Brad Baxter, All rights reserved. This module is free software. It may be used, redistributed and/or modified under the same terms as Perl itself. __________