NAME
   Smart::Dispatch - first-class switch statements

SYNOPSIS
    use Smart::Dispatch;
    my $given = dispatcher {
      match qr{ ^[A-J] }ix, dispatch { "Volume 1" };
      match qr{ ^[K-Z] }ix, dispatch { "Volume 2" };
      otherwise failover { Carp::croak "unexpected surname" };
    };
    my $surname = "Inkster";
    say $surname, " is in ", $dispatch->($surname), " of the phone book.";

DESCRIPTION
   People have been using dispatch tables for years. They work along the
   lines of:

    my $thing = get_foo_or_bar();

    my %dispatch = (
      foo   => sub { ... },
      bar   => sub { ... },
      );
    $dispatch{$thing}->();

   Dispatch tables are often more elegant than long groups of
   `if`/`elsif`/`else` statements, but they do have drawbacks. Consider how
   you'd change the example above to deal with $thing being not just "foo" or
   "bar", but adding all integers to the allowed values.

   Perl 5.10 introduced smart match and the `given` block. This allows stuff
   like:

    my $thing = get_foo_or_bar();

    given ($thing)
    {
      when ("foo") { ... }
      when ("bar") { ... }
      when (looks_like_number($_)) { ... }
    }

   The conditions in `when` clauses can be arbirarily complex tests, and
   default to comparisons using the smart match operator. This is far more
   flexible.

   `given` blocks do have some drawbacks over dispatch tables though. A
   dispatch table is a first class object - you can put a reference to it in
   a variable, and pass that reference as an argument to functions. You can
   check to see whether a dispatch table contains particular entries:

    if ($dispatch{"foo"})  # dispatch table can deal with $thing="foo"

   If passed a reference to an existing dispatch table, you can easily add
   entries to it, or remove entries from it.

   Smart::Dispatch is an attempt to combine some of the more useful features
   of `given` with dispatch tables.

 Building a Dispatch Table
   All the keywords used a build a dispatch table are lexical subs, which
   means that you can import them into a particular code block and they will
   not be available outside that block.

  `dispatcher { CODE }`
   A dispatch table is built using the `dispatcher` function which takes a
   single block argument. This block will typically consist of a number of
   `match` statements, though you can theoretically put anything you want
   inside it. (The code is run just once, when the dispatch table is being
   built, and is called in void context.)

    my $dispatch_table = dispatcher { ... };

   The return value is an Smart::Dispatch::Table object.

  `match $test, %args`
   The `match` function adds a single entry to the current dispatch table.
   The entry is a Smart::Dispatch::Match object.

   The $test argument is the trigger for dispatching to that particular entry
   in the table. It's like the contents of `when(...)` in a `given` block. It
   is used as the right hand argument to a smart match operation (see
   perlop), so it can be a string/numeric constant, `undef`, a `qr/.../`
   quoted regular expression, or a coderef, or an reference to an array
   containing any of the above. (There are other possibilities too, though
   they are somewhat obscure.)

   The hash of other arguments is passed to the constructor of
   Smart::Dispatch::Match.

  `dispatch { CODE }`
   This introduces the code to run when a match has been successful. It is
   used as follows:

    my $dispatch_table = dispatcher {
      match "foo", dispatch { "Monkey" };
      match "bar", dispatch { my $x = get_simian(); return $x };
    };

   Actually the above is just syntactic sugar for

    my $dispatch_table = dispatcher {
      match "foo", 'dispatch' => sub { "Monkey" };
      match "bar", 'dispatch' => sub { my $x = get_simian(); return $x };
    };

   So the only thing `dispatch` is doing is depositing a coderef into the
   %args hash of `match`.

  `value => $value`
   In the case of the "Monkey" bit above, it's actually a little wasteful to
   define a coderef (and run it when we do the dispatching later on) just to
   return a constant string, so in this case we can use the 'value' argument
   for `match`, to provide a slight optimization:

    my $dispatch_table = dispatcher {
      match "foo", value => "Monkey";
      match "bar", dispatch { my $x = get_simian(); return $x };
    };

   Note that `value` is not a function. It's just a named argument for
   `match`. Nothing much magic is going on.

  `match_using { CODE } %args`
   `match_using` is exactly like `match` but declared with a coderef
   prototype (see perlsub). That is, it just gives you syntactic sugar for
   the case where $test is a coderef. The following are equivalent:

   `match_using { $_ < 5 } dispatch { say "$_ is low" };`
   `match sub { $_ < 5 }, 'dispatch' => sub { say "$_ is low" };`

  `otherwise %args`
   `otherwise` is equivalent to `default` in `given` blocks, or `else` in
   `if` blocks. It matches all other cases, and must thus be the last match
   declared.

   Again this is really just syntactic sugar. The following are equivalent:

   `otherwise dispatch { undef };`
   `match sub { 1 }, 'is_unconditional' => 1, 'dispatch' => sub { undef };`

   Note that `otherwise` explicitly marks the match as an "unconditional"
   match. This allows Smart::Dispatch to complain if `otherwise` is not the
   last match in a dispatch table. And it helps when you try to combine
   multiple dispatch tables to know which is the "otherwise" match.

  `failover { CODE }`
   This is roughly the same as `dispatch`, but is intended for marking
   dispatches that can be regarded as failures:

    my $roman = dispatcher {
      match qr{\D}, failover { croak "non-numeric" };
      match [1..3], dispatch { "I" x $_ };
      match 4, value => 'IV';
      match [5..8], dispatch { 'V'.('I' x ($_-5)) };
      match 9, value => 'IX';
      match 10, value => 'X';
      otherwise failover { croak "out of range" };
    };

   In terms of actually dispatching from the dispatch table, failovers work
   exactly the same as any other dispatches. However, because the dispatch
   table knows which matches are successes and which are failures, this
   information can be queried.

   It should be no surprise by now that the `failover` function is just
   syntactic sugar, and the same effect can be achieved without it. The
   following are equivalent:

   `match $test, failover {...};`
   `match $test, 'is_failover' => 1, 'dispatch' => sub {...};`

 Using a Dispatch Table
   OK, so now you know how to build a dispatch table, but once we've got one,
   how can we use it?

   Dispatch tables, although they are not coderefs, overload `&{}`, which
   means they can be called like coderefs.

    my $biological_sex = dispatcher {
      match 'XX',         dispatch { 'Female' };
      match ['XY', 'YX'], dispatch { 'Male' };
      otherwise           failover { '????' };
    };

    my $sex_chromosomes = 'XY';
    say "I am a ", $biological_sex->($sex_chromosomes);

   The above will say "I am a Male".

   Note that the dispatch and failover subs here are pretty boring (we could
   have just used `<value`>), but any arbitrary Perl function is allowed.
   Perl functions of course accept argument lists. Any argument list passed
   into the dispatch table will be passed on to the dispatched function.

    my $biological_sex = dispatcher {
      match 'XX',
        dispatch { $_[1] eq 'fr' ? 'Femelle' : 'Female' };
      match ['XY', 'YX'],
        dispatch { $_[1] eq 'fr' ? 'Male' : 'Male' };
      otherwise
        failover { '????' };
    };

    my $sex_chromosomes = 'XX';
    say "I am a ", $biological_sex->($sex_chromosomes, 'en');
    say "Je suis ", $biological_sex->($sex_chromosomes, 'fr');

   Note that within `match_using`, `dispatch` and `failover` blocks, the
   value being matched is available in the variable $_. The following match
   demonstrates this:

    match_using { $_ < 5 } dispatch { say "$_ is low" }

   It is possible to check whether a dispatch table is able to handle a
   particular value.

    my $sex_chromosomes = 'AA';
    if ($biological_sex ~~ $sex_chromosomes)
    {
      say "Dispatch table cannot handle chromosomes $sex_chromosomes";
    }
    else
    {
      say $biological_sex->($sex_chromosomes);
    }

   This is where `failover` comes in. Failover matches are not considered
   when determining whether a dispatch table is capable of handling a value.

 Manipulating Dispatch Tables
   If you have an existing dispatch table, it's possible to add more entries
   to it. For this purpose, Smart::Dispatch overloads the `.=` and `+=`
   operators.

    my $more_sexes = dispatcher {
      match 'XYY',  dispatch { 'Supermale' };
      match 'XXX',  dispatch { 'Superfemale' };
    };
    $biological_sex .= $more_sexes;

   The difference between the two operators is the priority is which matches
   are tested.

    my $match1 = dispatcher {
      match 1, dispatch { 'One' };
    };

   We can add some more matches like this:

    $match1 .= dispatcher {
      match qr{^1}, dispatch { 'Leading one' };
    };

   When dispatching value "1", the result will still be "One", because the
   added matches have lower priority than the original ones.

   But if they are combined as:

    $match += dispatcher {
      match qr{^1}, dispatch { 'Leading one' };
    };

   Then when dispatching value "1", the result will be "Leading one" because
   the newer matches are given higher priority.

   It is also possible to use `.` and `+` in their non-assignment forms:

    my $enormous_match = $big_match . $large_match . $mighty_match;

   (Some future version may introduce the ability to do subtraction, but
   there are difficulties with this concept. For now, if you want to do
   subtraction, look at the internals of Smart::Dispatch::Table.)

   If one or both dispatch tables contain an unconditional match
   (`otherwise`), then these will be combined intelligently. The result will
   only have one unconditional match (the higher priority one).

 Import
   By default Smart::Dispatch exports the following functions:

   *   `dispatch`

   *   `dispatcher`

   *   `failover`

   *   `match`

   *   `match_using`

   *   `otherwise`

   It is possible to only import a subset of those:

    use Smart::Dispatch qw/dispatcher match otherwise/;

   As noted in the "Building a Dispatch Table" section, a minimal set of
   functions is just `dispatcher` and `match`. All the others are just
   syntactic sugar. If you just want those two, then you can do:

    use Smart::Dispatch qw/:tiny/;

   Smart::Dispatch uses Sub::Exporter which provides a dizzying array of cool
   options, such as:

    use Smart::Dispatch -all => { -prefix => 'sd_' };

   which imports all the symbols but prefixed with "sd_".

    use Smart::Dispatch
      qw/dispatcher dispatch match/,
      otherwise => { -as => 'last_resort' };

   which renames "otherwise" to "last_resort".

   If you've written subclasses of Smart::Dispatch::Table and
   Smart::Dispatch::Match and you want Smart::Dispatch to use your
   subclasses, then you can do this:

    use Smart::Dispatch
      qw/dispatcher dispatch match/,
      otherwise => { -as => 'last_resort' },
      class => {
        table => 'My::Dispatch::Table',
        match => 'My::Dispatch::Match',
        };

   Whatsmore, the `class` option can be set on a keyword-by-keyword basis for
   `match`, `match_using` and `otherwise`.

    use Smart::Dispatch
      qw/dispatcher dispatch match/,
      otherwise => {
        -as   => 'last_resort',
        class => 'My::Other::Match',
        },
      class => {
        table => 'My::Dispatch::Table',
        match => 'My::Dispatch::Match',
        };

 Constants
   *   `DEFAULT_MATCH_CLASS`

   *   `DEFAULT_TABLE_CLASS`

 Dispatch Table Internals
   See Smart::Dispatch::Table and Smart::Dispatch::Match.

   Note that this is an early release, so the internals are still likely to
   change somewhat between versions. The function-based API should be fairly
   steady though.

BUGS
   Please report any bugs to
   <http://rt.cpan.org/Dist/Display.html?Queue=Smart-Dispatch>.

SEE ALSO
   "Switch statements" in perlsyn; Acme::Given::Hash.

   <http://www.perlmonks.org/?node_id=954831>.

AUTHOR
   Toby Inkster <[email protected]>.

COPYRIGHT AND LICENCE
   This software is copyright (c) 2012 by Toby Inkster.

   This is free software; you can redistribute it and/or modify it under the
   same terms as the Perl 5 programming language system itself.

DISCLAIMER OF WARRANTIES
   THIS PACKAGE IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED
   WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
   MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE.