package Attribute::Params::Validate; use strict; use warnings; # ok to use cause Attribute::Handlers needs 5.6.0+ as well use attributes; use Attribute::Handlers; # this will all be re-exported use Params::Validate qw(:all); require Exporter; use vars qw($VERSION); our @ISA = qw(Exporter); my %tags = ( types => [ qw( SCALAR ARRAYREF HASHREF CODEREF GLOB GLOBREF SCALARREF HANDLE UNDEF OBJECT ) ], ); our %EXPORT_TAGS = ( 'all' => [ qw( validation_options ), map { @{ $tags{$_} } } keys %tags ], %tags, ); our @EXPORT_OK = ( @{ $EXPORT_TAGS{all} }, 'validation_options' ); $VERSION = sprintf '%2d.%02d', q$Revision: 1.7 $ =~ /(\d+)\.(\d+)/; sub UNIVERSAL::Validate : ATTR(CODE, INIT) { _wrap_sub('named', @_); } sub UNIVERSAL::ValidatePos : ATTR(CODE, INIT) { _wrap_sub('positional', @_); } sub _wrap_sub { my ($type, $package, $symbol, $referent, $attr, $params) = @_; my @p = ref $params ? @{ $params } : $params; my $subname = $package . '::' . *{$symbol}{NAME}; my %attributes = map { $_ => 1 } attributes::get($referent); my $is_method = $attributes{method}; { no warnings 'redefine'; no strict 'refs'; # An unholy mixture of closure and eval. This is done so that # the code to automatically create the relevant scalars from # the hash of params can create the scalars in the proper # place lexically. my $code = <<"EOF"; sub { package $package; EOF $code .= " my \$object = shift;\n" if $is_method; if ($type eq 'named') { $params = {@p}; $code .= " Params::Validate::validate(\@_, \$params);\n"; } else { $code .= " Params::Validate::validate_pos(\@_, \@p);\n"; } $code .= " unshift \@_, \$object if \$object;\n" if $is_method; $code .= <<"EOF"; \$referent->(\@_); } EOF my $sub = eval $code; die $@ if $@; *{$subname} = $sub; } } 1; =head1 NAME Attribute::Params::Validate - Validate method/function parameters using attributes =head1 SYNOPSIS use Attribute::Params::Validate qw(:all); # takes named params (hash or hashref) # foo is mandatory, bar is optional sub foo : Validate( foo => 1, bar => 0 ) { ... } # takes positional params # first two are mandatory, third is optional sub bar : ValidatePos( 1, 1, 0 ) { ... } # for some reason Perl insists that the entire attribute be on one line sub foo2 : Validate( foo => { type => ARRAYREF }, bar => { can => [ 'print', 'flush', 'frobnicate' ] }, baz => { type => SCALAR, callbacks => { 'numbers only' => sub { shift() =~ /^\d+$/ }, 'less than 90' => sub { shift() < 90 } } } ) { ... } # note that this is marked as a method. This is very important! sub baz : Validate( foo => { type => ARRAYREF }, bar => { isa => 'Frobnicator' } ) method { ... } =head1 DESCRIPTION The Attribute::Params::Validate module allows you to validate method or function call parameters just like Params::Validate does. However, this module allows you to specify your validation spec as an attribute, rather than by calling the C routine. Please see Params::Validate for more information on how you can specify what validation is performed. =head2 EXPORT This module exports everthing that Params::Validate does except for the C and C subroutines. =head2 ATTRIBUTES =over 4 =item * Validate This attribute corresponse to the C subroutine in Params::Validate. =item * ValidatePos This attribute corresponse to the C subroutine in Params::Validate. =back =head2 OO If you are using this module to mark B for validation, as opposed to subroutines, it is crucial that you mark these methods with the C<:method> attribute, as well as the C or C attribute. If you do not do this, then the object or class used in the method call will be passed to the validation routines, which is probably not what you want. =head2 CAVEATS You B put all the arguments to the C or C attribute on a single line, or Perl will complain. =head1 SEE ALSO Params::Validate =head1 AUTHOR Dave Rolsky, =cut