Data::Validator - Rule based validator on type constraint system

This document describes Data::Validator version 0.06.

    use 5.10.0;
    use Data::Validator;

    # for functions
    sub get {
        state $rule = Data::Validator->new(
            uri        => { isa => 'Str', xor => [qw(schema host path_query)] },

            schema     => { isa => 'Str', default => 'http' },
            host       => { isa => 'Str' },
            path_query => { isa => 'Str', default => '/' },

            method     => { isa => 'Str', default => 'GET' },
        );

        my $args = $rule->validate(@_);
        # ...
    }
    get( uri => 'http://example.com/' );

    # for methods
    sub method {
        state $rule = Data::Validator->new(
            foo => 'Str',
        )->with('Method');

        my($self, $args) = $rule->validate(@_);
        # ...
    }
    Foo->method( foo => 'bar' );


    # using sequenced parameters
    sub seq {
        state $rule = Data::Validator->new(
            foo => 'Str',
        )->with('Sequenced');

        my $args = $rule->validate(@_);
        # ...
    }
    seq( 'bar' );          # seq() will get { foo => 'bar' }
    seq({ foo => 'bar' }); # named style are available!


    # using Method and Sequenced together
    sub seq_method {
        state $rule = Data::Validator->new(
            foo => 'Str',
        )->with( 'Method', 'Sequenced');

        my($self, $args) = $rule->validate(@_);
        # ...
    }
    Foo->seq_method( 'bar' ); # seq() will get { foo => 'bar' }

This is yet another validation library, based on Smart::Args but less smart.

This is under development. Any API will change without notice.

Concepts¶

Natural as Perl code

I love Smart::Args because it is really stylish, but it does not seem Perl-ish.

Thus, I have designed Data::Validator in more Perl-ish way with full of Smart::Args functionality.

Basics on type constraint system

Moose's type constraint system is awesome, and so is Mouse's. In fact, Mouse's type constraints are much faster than Moose's so that you need not hesitate to check types.

Thus, I have made Data::Validator on Mouse's type constraint system.

Pure Perl

Although I do not hesitate to depend on XS modules, some people think that XS modules are hard to install.

Thus, I have written Data::Validator in pure Perl and selected dependent modules which work in pure Perl.

Performance

Validators should be as fast as possible because they matter only for illegal inputs. Otherwise, one would want something like no validation option.

This is much faster than Params::Validate, which has an XS backend, though.

Data::Validator->new( $arg_name => $rule [, ...]) :Validator¶

Creates a validation rule. You should cache the rules for performance.

Attributes for $rule are as follows:

isa => $type : Str|Object

The type of the rule, which can be a Mouse type constraint name, a class name, or a type constraint object of either Mouse or Moose (i.e. it's duck-typed).

does => $role : Str|Object

The type of the rule, which can be a Mouse type constraint name, a role name, or a type constraint object of either Mouse or Moose (i.e. it's duck-typed).

Note that you cannot use it with the isa attribute.

coerce => $should_coercion : Bool

If false, the rule does not try to coerce when the validation fails. Default to true.

default=> $value : Any | CodeRef

The default value for the argument. If it is a CODE reference, it is called in scalar context as $default->($validator, $rule, $args) and its return value is used as a default value.

Because arguments are validated in the order of definitions, default callbacks can rely on the previously-filled values:

    my $v = Data::Validator->new(
        foo => { default => 99 },
        bar => { default => sub {
            my($validator, $this_rule, $args) = @_;
            return $args->{foo} + 1;
        } },
    );
    $v->validate();          # bar is 100
    $v->validate(foo => 42); # bar is 43

Unlike Moose/Mouse's default, any references are allowed, but note that they are statically allocated.

optional => $value : Bool

If true, users can omit the argument. Default to false.

xor => $exclusives : ArrayRef

Exclusive arguments, which users cannot pass together.

documentation => $doc : Str

Descriptions of the argument.

This is not yet used anywhere.

$validator->find_rule($name :Str)¶

Finds the rule named $name. Provided for error handling.

$validator->with(@extentions) :Validator¶

Applies @extentions to $validator and returns itself.

See "EXTENTIONS" for details.

$validator->validate(@args) :HashRef¶

Validates @args and returns a restricted HASH reference.

Restricted hashes are hashes which do not allow to access non-existing keys, so you must check a key exists in the hash before fetching its values.

There are extentions which changes behaviours of validate().

Method¶

Takes the first argument as an invocant (i.e. class or object instance), and returns it as the first value:

    my($invocant, $args) = $rule->validate(@_);

Sequenced¶

Deals with arguments in sequenced style, where users should pass arguments by the order of argument rules, instead of by-name.

Note that if the last argument is a HASH reference, it is regarded as named-style arguments.

AllowExtra¶

Regards unknown arguments as extra arguments, and returns them as a list of name-value pairs:

    my($args, %extra) = $rule->validate(@_);

NoThrow¶

Does not throw errors. Instead, it provides validators with the errors attribute:

    my $args = $v->validate(@_); # it never throws errors
    if($v->has_errors) {
        my $errors = $v->clear_errors;
        foreach my $e(@{$errors}) {
            # $e has 'type', 'message' and 'name'
            print $e->{message}, "\n";
        }
    }

Croak¶

Does not report stack backtraces on errors, i.e. uses croak() instead of confess() to throw errors.

All complex software has bugs lurking in it, and this module is no exception. If you find a bug please either email me, or add the bug to cpan-RT.