NAME
Sub::PatternMatching - Functional languages' Pattern Matching for Perl subs
SYNOPSIS
use Sub::PatternMatching;
my $code_ref = patternmatch( pattern => sub {code}, ... );
*NamedRoutine = patternmatch( more_patterns => sub{code}, ... );
NamedRoutine( ...arguments to do pattern matching on... );
$code_ref->( ...arguments to do pattern matching on... );DESCRIPTION
Sub::PatternMatching implements "Pattern Matching," a programming idiom often found in functional languages like Haskell or OCaml. Pattern Matching refers to functions that do different things for different arguments. It is often referred to as polymorphism as well.
The syntax is currently a bit convoluted taking away much from the benefit of readability by replacing nested if-else constructs with pattern matching. This is supposed to be fixed in a later release, but maintaining backward-compatibility.
To construct a pattern matching subroutine, you assign the result of a call to the patternmatch() subroutine that is exported by the module to a scalar or symbol table entry. patternmatch() returns an anonymous function.
The arguments to patternmatch() are expected to be pairs of array or hash references and code references like this:
C<[ ...pattern... ] => sub { ...implementation... }>
or
C<{ ...pattern... } => sub { ...implementation... }>If the pattern applies to the function arguments, the corresponding code reference will be called with the original arguments.
PATTERNS
The patterns are evaluated in the order they were passed to the patternmatch() routine. The first that matches determines the code to execute. Pattern evaluation is implemented using the Params::Validate module. The syntax for a function with positional parameters is basically:
[{conditions for param1}, {conditions for param2}, ...] => sub {...},
[{conditions for param1}, {conditions for param2}, ...] => sub {...},
...The conditions are defined exactly as you would specify conditions for positional parameters of a subroutine using Params::Validate's validate_pos() routine. (So read the docs for that module right now.)
Similarily, patterns for named arguments are specified as follows:
{ paramname1 => {conditions for param1}, paramname2 =>... } => sub {...},
{ paramname1 => {conditions for param1}, paramname2 =>... } => sub {...},Here, of course, the order of the parameters doesn't matter because we're dealing with named parameters. Again, conditions are specified just as if you were dealing with Params::Validate directly.
Please have a look at the examples section below.
There is one exception to the rule of supplying [pattern] = sub{ implementation }> pairs: If, at the end of the list of pattern pairs, you supply a single code ref, that routine will be used as the default handler for unmatched parameters. By default, the patternmatched subroutine will throw a fatal error when unmatched parameters are passed in.
EXPORT
By default, the module exports the 'patternmatch' subroutine.
EXAMPLES
Example 1
If you just wanted to do different things for different numbers of arguments, you could, for example do something like the following: (incomplete code)
*function = pattermatch(
[ 1 ] => sub { ... actions for one argument ... },
[ 1, 1 ] => sub { ... actions for two arguments ... },
[ 1, 1, 1, 0 ] => sub { ... actions for three or four args ... },
...
[ (1) x $n ] => sub { ... actions for $n arguments ... },
...
[ (1) x @_ ] => sub { ... arbitrary number of arguments ... },
);Example 2
The following example implements a simple-minded data structure dumper that can deal with arbitrarily nested hashes, arrays, and scalars.
use strict;
use warnings;
use Sub::PatternMatching;
use Params::Validate qw/:all/;
my $simple_dumper;
$simple_dumper = patternmatch(
[{ type => HASHREF }] =>
sub {
"HASH {\n"
. join(",\n", map {
"$_: "
. $simple_dumper->($_[0]{$_})
} keys %{$_[0]}
)
. "\n}"
},
[{ type => ARRAYREF }] =>
sub {
"ARRAY [\n"
. join(",\n",
map { $simple_dumper->($_) } @{$_[0]}
)
. "\n]"
},
[{ type => SCALAR }] => sub {"SCALAR ($_[0])"},
);
print $simple_dumper->([{foo => 'bar'}, 'a', [1..3]]);Example 3
This, admittedly more involved, example demostrates a functional style set of algebraic operators and a routine that can derive simple algebraic formulae:
use strict;
use warnings;
use lib 'lib';
use Sub::PatternMatching 'patternmatch';
use Params::Validate qw/:all/;
no warnings 'once';
my $stringifier = sub {
my $obj = shift;
my $str = ref($obj);
if (@$obj) {
$str .= '( '
. join( ', ', map { ref($_) ? $_->stringify : "$_" } @$obj ) . ' )';
}
return $str;
};
my $printer = sub { print shift()->stringify, "\n" };
*Product::stringify = $stringifier;
*Quotient::stringify = $stringifier;
*Sum::stringify = $stringifier;
*Difference::stringify = $stringifier;
*Constant::stringify = $stringifier;
*X::stringify = $stringifier;
*Product::show = $printer;
*Quotient::show = $printer;
*Sum::show = $printer;
*Difference::show = $printer;
*Constant::show = $printer;
*X::show = $printer;
sub Product ($$) { bless [ @_[ 0, 1 ] ] => 'Product' }
sub Quotient ($$) { bless [ @_[ 0, 1 ] ] => 'Quotient' }
sub Sum ($$) { bless [ @_[ 0, 1 ] ] => 'Sum' }
sub Difference ($$) { bless [ @_[ 0, 1 ] ] => 'Difference' }
sub Constant ($) { bless [ $_[0] ] => 'Constant' }
sub X () { bless [ ] => 'X' }
*::derive = patternmatch(
[ { isa => 'Constant' } ] => sub { Constant 0 },
[ { isa => 'X' } ] => sub { Constant 1 },
[ { isa => 'Sum' } ]
=> sub {
my ( $l, $r ) = @{ $_[0] };
Sum( derive($l), derive($r) );
},
[ { isa => 'Difference' } ]
=> sub {
my ( $l, $r ) = @{ $_[0] };
Difference derive($l), derive($r);
},
[ { isa => 'Product' } ]
=> sub {
my ( $l, $r ) = @{ $_[0] };
Sum
Product( derive($l), $r ),
Product( derive($r), $l );
},
[ { isa => 'Quotient' } ]
=> sub {
my ( $l, $r ) = @{ $_[0] };
Quotient
Difference(
Product( derive($l), $r ),
Product( derive($r), $l )
),
Product( $r, $r );
},
);
my $function = Product Constant 5, X;
print "We'll derive this: ";
$function->show;
print "\nThe derivative of the above is computed to:\n";
derive($function)->show;CAVEATS
Functional languages' compilers usually optimize away the pattern matching overhead of evaluating the conditions for every call until a matching condition is found. This is mostly possible because of their static typing system which Perl proudly lacks. Therefore, using this module for pattern matching currently takes an O(n) performance hit for every call to the patternmatching function. n is the number of branches, sets of conditions. Note that if you would implement your function with a giant if-elsif-else construct, you would end up with O(n) as well.
AUTHOR
Copyright (C) 2004 Steffen Mueller <pattern-module at steffen-mueller dot net>
This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
SEE ALSO
Current versions of this module may be found on http://steffen-mueller.net or CPAN.
Please send your suggestions, inquiries, and feedback to pattern-module at steffen-mueller dot net. Bug reports should use RT or be mailed to bug-Sub-PatternMatching@rt.cpan.org