Type::Params - Params::Validate-like parameter validation using Type::Tiny type constraints and coercions
use v5.10;
use strict;
use warnings;
use Type::Params qw( compile );
use Types::Standard qw( slurpy Str ArrayRef Num );
sub deposit_monies
{
state $check = compile( Str, Str, slurpy ArrayRef[Num] );
my ($sort_code, $account_number, $monies) = $check->(@_);
my $account = Local::BankAccount->new($sort_code, $account_number);
$account->deposit($_) for @$monies;
}
deposit_monies("12-34-56", "11223344", 1.2, 3, 99.99);
This module is covered by the
Type-Tiny stability policy.
Type::Params uses the Type::Tiny manpage constraints to validate the parameters to a
sub. It takes the slightly unorthodox approach of separating validation
into two stages:
-
Compiling the parameter specification into a coderef; then
-
Using the coderef to validate parameters.
The first stage is slow (it might take a couple of milliseconds), but you
only need to do it the first time the sub is called. The second stage is
fast; according to my benchmarks faster even than the XS version of
the Params::Validate manpage.
If you're using a modern version of Perl, you can use the state keyword
which was a feature added to Perl in 5.10. If you're stuck on Perl 5.8, the
example from the SYNOPSIS could be rewritten as:
my $deposit_monies_check;
sub deposit_monies
{
$deposit_monies_check ||= compile( Str, Str, slurpy ArrayRef[Num] );
my ($sort_code, $account_number, $monies) = $deposit_monies_check->(@_);
...;
}
Not quite as neat, but not awful either.
There's a shortcut reducing it to one step:
use Type::Params qw( validate );
sub deposit_monies
{
my ($sort_code, $account_number, $monies) =
validate( \@_, Str, Str, slurpy ArrayRef[Num] );
...;
}
Type::Params has a few tricks up its sleeve to make sure performance doesn't
suffer too much with the shortcut, but it's never going to be as fast as the
two stage compile/execute.
This module offers one-stage (``validate'') and two-stage (``compile'' then
``check'') variants of parameter checking for you to use. Performance with
the two-stage variant will always beat the one stage variant — I
cannot think of many reasons you'd want to use the one-stage version.
# One-stage, positional parameters
my @args = validate(\@_, @spec);
# Two-stage, positional parameters
state $check = compile(@spec);
my @args = $check->(@_);
# One-stage, named parameters
my $args = validate_named(\@_, @spec);
# Two-stage, named parameters
state $check = compile_named(@spec);
my $args = $check->(@_);
Use compile and compile_named, not validate and validate_named.
The @spec is where most of the magic happens.
The generalized form of specifications for positional parameters is:
@spec = (
\%general_opts,
$type_for_arg_1, \%opts_for_arg_1,
$type_for_arg_2, \%opts_for_arg_2,
$type_for_arg_3, \%opts_for_arg_3,
...,
slurpy($slurpy_type),
);
And for named parameters:
@spec = (
\%general_opts,
foo => $type_for_foo, \%opts_for_foo,
bar => $type_for_bar, \%opts_for_bar,
baz => $type_for_baz, \%opts_for_baz,
...,
slurpy($slurpy_type),
);
Option hashrefs can simply be omitted if you don't need to specify any
particular options.
The slurpy function is exported by the Types::Standard manpage. It may be
omitted if not needed.
Currently supported general options are:
- want_source => Bool
-
Instead of returning a coderef, return Perl source code string. Handy
for debugging.
- want_details => Bool
-
Instead of returning a coderef, return a hashref of stuff including the
coderef. This is mostly for people extending Type::Params and I won't go
into too many details about what else this hashref contains.
- class => ClassName
-
Named parameters only. > The check coderef will, instead of returning
a simple hashref, call
$class->new($hashref) and return a proper
object.
- constructor => Str
-
Named parameters only. > Specify an alternative method name instead
of
new for the class option described above.
- class => Tuple[ClassName, Str]
-
Named parameters only. > Given a class name and constructor name pair,
the check coderef will, instead of returning a simple hashref, call
$class->$constructor($hashref) and return a proper object. Shortcut
for declaring both the class and constructor options at once.
- bless => ClassName
-
Named parameters only. > Bypass the constructor entirely and directly
bless the hashref.
- description => Str
-
Description of the coderef that will show up in stack traces. Defaults to
``parameter validation for X'' where X is the caller sub name.
- subname => Str
-
If you wish to use the default description, but need to change the sub name,
use this.
- caller_level => Int
-
If you wish to use the default description, but need to change the caller
level for detecting the sub name, use this.
The types for each parameter may be any the Type::Tiny manpage type constraint, or
anything that Type::Tiny knows how to coerce into a Type::Tiny type
constraint, such as a MooseX::Types type constraint or a coderef.
The Optional parameterizable type constraint from the Types::Standard manpage
may be used to indicate optional parameters.
# Positional parameters
state $check = compile(Int, Optional[Int], Optional[Int]);
my ($foo, $bar, $baz) = $check->(@_); # $bar and $baz are optional
# Named parameters
state $check = compile(
foo => Int,
bar => Optional[Int],
baz => Optional[Int],
);
my $args = $check->(@_); # $args->{bar} and $args->{baz} are optional
As a special case, the numbers 0 and 1 may be used as shortcuts for
Optional[Any] and Any.
# Positional parameters
state $check = compile(1, 0, 0);
my ($foo, $bar, $baz) = $check->(@_); # $bar and $baz are optional
# Named parameters
state $check = compile_named(foo => 1, bar => 0, baz => 0);
my $args = $check->(@_); # $args->{bar} and $args->{baz} are optional
If you're using positional parameters, then required parameters must
precede any optional ones.
Specifications may include a single slurpy parameter which should have
a type constraint derived from ArrayRef or HashRef. (Any is
also allowed, which is interpreted as ArrayRef in the case of positional
parameters, and HashRef in the case of named parameters.)
If a slurpy parameter is provided in the specification, the $check
coderef will slurp up any remaining arguments from @_ (after
required and optional parameters have been removed), validate it against
the given slurpy type, and return it as a single arrayref/hashref.
For example:
sub xyz {
state $check = compile(Int, Int, slurpy ArrayRef[Int]);
my ($foo, $bar, $baz) = $check->(@_);
}
xyz(1..5); # $foo = 1
# $bar = 2
# $baz = [ 3, 4, 5 ]
A specification have one or zero slurpy parameters. If there is a slurpy
parameter, it must be the final one.
Note that having a slurpy parameter will slightly slow down $check
because it means that $check can't just check @_ and return
it unaltered if it's valid — it needs to build a new array to return.
Type coercions are automatically applied for all types that have
coercions.
my $RoundedInt = Int->plus_coercions(Num, q{ int($_) });
state $check = compile($RoundedInt, $RoundedInt);
my ($foo, $bar) = $check->(@_);
# if @_ is (1.1, 2.2), then $foo is 1 and $bar is 2.
Coercions carry over into structured types such as ArrayRef automatically:
sub delete_articles
{
state $check = compile( Object, slurpy ArrayRef[$RoundedInt] );
my ($db, $articles) = $check->(@_);
$db->select_article($_)->delete for @$articles;
}
# delete articles 1, 2 and 3
delete_articles($my_db, 1.1, 2.2, 3.3);
That's a the Types::Standard manpage feature rather than something specific to
Type::Params.
Note that having any coercions in a specification, even if they're not
used in a particular check, will slightly slow down $check
because it means that $check can't just check @_ and return
it unaltered if it's valid — it needs to build a new array to return.
The type constraint for a parameter may be followed by a hashref of
options for it.
The following options are supported:
- optional => Bool
-
This is an alternative way of indicating that a parameter is optional.
state $check = compile_named(
foo => Int,
bar => Int, { optional => 1 },
baz => Optional[Int],
);
The two are not exactly equivalent. If you were to set bar to a
non-integer, it would throw an exception about the Int type constraint
being violated. If baz were a non-integer, the exception would mention
the Optional[Int] type constraint instead.
- default => CodeRef|Ref|Str|Undef
-
A default may be provided for a parameter.
state $check = compile_named(
foo => Int,
bar => Int, { default => "666" },
baz => Int, { default => "999" },
);
Supported defaults are any strings (including numerical ones), undef,
and empty hashrefs and arrayrefs. Non-empty hashrefs and arrayrefs are
not allowed as defaults >.
Alternatively, you may provide a coderef to generate a default value:
state $check = compile_named(
foo => Int,
bar => Int, { default => sub { 6 * 111 } },
baz => Int, { default => sub { 9 * 111 } },
);
That coderef may generate any value, including non-empty arrayrefs and
non-empty hashrefs. For undef, simple strings, numbers, and empty
structures, avoiding using a coderef will make your parameter processing
faster.
The default will be validated against the type constraint, and
potentially coerced.
Defaults are not supported for slurpy parameters.
Note that having any defaults in a specification, even if they're not
used in a particular check, will slightly slow down $check
because it means that $check can't just check @_ and return
it unaltered if it's valid — it needs to build a new array to return.
Type::Params can export a multisig function that compiles multiple
alternative signatures into one, and uses the first one that works:
state $check = multisig(
[ Int, ArrayRef ],
[ HashRef, Num ],
[ CodeRef ],
);
my ($int, $arrayref) = $check->( 1, [] ); # okay
my ($hashref, $num) = $check->( {}, 1.1 ); # okay
my ($code) = $check->( sub { 1 } ); # okay
$check->( sub { 1 }, 1.1 ); # throws an exception
Coercions, slurpy parameters, etc still work.
The magic global ${^TYPE_PARAMS_MULTISIG} is set to the index of
the first signature which succeeded.
The present implementation involves compiling each signature independently,
and trying them each (in their given order!) in an eval block. The only
slightly intelligent part is that it checks if scalar(@_) fits into
the signature properly (taking into account optional and slurpy parameters),
and skips evals which couldn't possibly succeed.
It's also possible to list coderefs as alternatives in multisig:
state $check = multisig(
[ Int, ArrayRef ],
sub { ... },
[ HashRef, Num ],
[ CodeRef ],
compile_named( needle => Value, haystack => Ref ),
);
The coderef is expected to die if that alternative should be abandoned (and
the next alternative tried), or return the list of accepted parameters. Here's
a full example:
sub get_from {
state $check = multisig(
[ Int, ArrayRef ],
[ Str, HashRef ],
sub {
my ($meth, $obj);
die unless is_Object($obj);
die unless $obj->can($meth);
return ($meth, $obj);
},
);
my ($needle, $haystack) = $check->(@_);
for (${^TYPE_PARAMS_MULTISIG) {
return $haystack->[$needle] if $_ == 0;
return $haystack->{$needle} if $_ == 1;
return $haystack->$needle if $_ == 2;
}
}
get_from(0, \@array); # returns $array[0]
get_from('foo', \%hash); # returns $hash{foo}
get_from('foo', $obj); # returns $obj->foo
Here's a quick example function:
sub add_contact_to_database {
state $check = compile_named(
dbh => Object,
id => Int,
name => Str,
);
my $arg = $check->(@_);
my $sth = $arg->{db}->prepare('INSERT INTO contacts VALUES (?, ?)');
$sth->execute($arg->{id}, $arg->{name});
}
Looks simple, right? Did you spot that it will always die with an error
message Can't call method ``prepare'' on an undefined value >?
This is because we defined a parameter called 'dbh' but later tried to
refer to it as $arg{db}. Here, Perl gives us a pretty clear
error, but sometimes the failures will be far more subtle. Wouldn't it
be nice if instead we could do this?
sub add_contact_to_database {
state $check = compile_named_oo(
dbh => Object,
id => Int,
name => Str,
);
my $arg = $check->(@_);
my $sth = $arg->dbh->prepare('INSERT INTO contacts VALUES (?, ?)');
$sth->execute($arg->id, $arg->name);
}
If we tried to call $arg->db, it would fail because there was
no such method.
Well, that's exactly what compile_named_oo does.
As well as giving you nice protection against mistyped parameter names,
It also looks kinda pretty, I think. Hash lookups are a little faster
than method calls, of course (though Type::Params creates the methods
using the Class::XSAccessor manpage if it's installed, so they're still pretty
fast).
An optional parameter foo will also get a nifty $arg->has_foo
predicate method. Yay!
compile_named_oo gives you some extra options for parameters.
sub add_contact_to_database {
state $check = compile_named_oo(
dbh => Object,
id => Int, { default => '0', getter => 'identifier' },
name => Str, { optional => 1, predicate => 'has_name' },
);
my $arg = $check->(@_);
my $sth = $arg->dbh->prepare('INSERT INTO contacts VALUES (?, ?)');
$sth->execute($arg->identifier, $arg->name) if $arg->has_name;
}
The getter option lets you choose the method name for getting the
argument value. The predicate option lets you choose the method name
for checking the existence of an argument.
By setting an explicit predicate method name, you can force a predicate
method to be generated for non-optional arguments.
The objects returned by compile_named_oo are blessed into lightweight
classes which have been generated on the fly. Don't expect the names of
the classes to be stable or predictable. It's probably a bad idea to be
checking can, isa, or DOES on any of these objects. If you're
doing that, you've missed the point of them.
They don't have any constructor (new method). The $check
coderef effectively is the constructor.
This can be faked using positional parameters and a slurpy dictionary.
state $check = compile(
Int,
slurpy Dict[
foo => Int,
bar => Optional[Int],
baz => Optional[Int],
],
);
@_ = (42, foo => 21); # ok
@_ = (42, foo => 21, bar => 84); # ok
@_ = (42, foo => 21, bar => 10.5); # not ok
@_ = (42, foo => 21, quux => 84); # not ok
Some people like to shift off the invocant before running type checks:
sub my_method {
my $self = shift;
state $check = compile_named(
haystack => ArrayRef,
needle => Int,
);
my $arg = $check->(@_);
return $arg->{haystack}[ $self->base_index + $arg->{needle} ];
}
$object->my_method(haystack => \@somelist, needle => 42);
If you're using positional parameters, there's really no harm in including
the invocant in the check:
sub my_method {
state $check = compile(Object, ArrayRef, Int);
my ($self, $arr, $ix) = $check->(@_);
return $arr->[ $self->base_index + $ix ];
}
$object->my_method(\@somelist, 42);
Some methods will be designed to be called as class methods rather than
instance methods. Remember to use ClassName instead of Object in
those cases.
Type::Params exports an additional keyword Invocant on request. This
gives you a type constraint which accepts classnames and blessed
objects.
use Type::Params qw( compile Invocant );
sub my_method {
state $check = compile(Invocant, ArrayRef, Int);
my ($self_or_class, $arr, $ix) = $check->(@_);
return $arr->[ $ix ];
}
If you give compile a type constraint which has coercions, then
$check will always coerce >. It cannot be switched off.
Luckily, Type::Tiny gives you a very easy way to create a type
constraint without coercions from one that has coercions:
state $check = compile(
$RoundedInt->no_coercions,
$RoundedInt->minus_coercions(Num),
);
That's a Type::Tiny feature rather than a Type::Params feature though.
Type::Tiny provides an easy shortcut for adding coercions to
a type constraint:
# We want an arrayref, but accept a hashref and coerce it
state $check => compile(
ArrayRef->plus_coercions( HashRef, sub { [sort values %$_] } ),
);
You may further constrain a parameter using where:
state $check = compile(
Int->where('$_ % 2 == 0'), # even numbers only
);
This is also a Type::Tiny feature rather than a Type::Params feature.
This works:
sub print_coloured {
state $check = compile(
Str,
Str, { default => "black" },
);
my ($text, $colour) = $check->(@_);
...;
}
But so does this (and it might benchmark a little faster):
sub print_coloured {
state $check = compile(
Str,
Str, { optional => 1 },
);
my ($text, $colour) = $check->(@_);
$colour = "black" if @_ < 2;
...;
}
Just because Type::Params now supports defaults, doesn't mean you can't
do it the old-fashioned way. The latter is more flexible. In the example,
we've used if @_ < 2, but we could instead have done something like:
$colour ||= "black";
Which would have defaulted $colour to ``black'' if it were the empty
string.
- PERL_TYPE_PARAMS_XS
-
Affects the building of accessors for
compile_named_oo. If set to true,
will use the Class::XSAccessor manpage. If set to false, will use pure Perl. If this
environment variable does not exist, will use the Class::XSAccessor manpage if it
is available.
the Type::Params manpage is not really a drop-in replacement for the Params::Validate manpage;
the API differs far too much to claim that. Yet it performs a similar task,
so it makes sense to compare them.
the Params::ValidationCompiler manpage does basically the same thing as
the Type::Params manpage.
-
Params::ValidationCompiler and Type::Params are likely to perform fairly
similarly. In most cases, recent versions of Type::Params seem to be
slightly faster, but except in very trivial cases, you're unlikely to
notice the speed difference. Speed probably shouldn't be a factor when
choosing between them.
-
Type::Params's syntax is more compact:
state $check = compile(Object, Optional[Int], slurpy ArrayRef);
Versus:
state $check = validation_for(
params => [
{ type => Object },
{ type => Int, optional => 1 },
{ type => ArrayRef, slurpy => 1 },
],
);
-
the Params::ValidationCompiler manpage probably has slightly better exceptions.
Please report any bugs to
http://rt.cpan.org/Dist/Display.html.
the Type::Tiny manpage, the Type::Coercion manpage, the Types::Standard manpage.
Toby Inkster <tobyink@cpan.org>.
This software is copyright (c) 2013-2014, 2017-2019 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.
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.
|