File Coverage

lib/Data/Processor.pm

Criterion	Covered	Total	%
statement	109	116	93.9
branch	30	42	71.4
condition	19	34	55.8
subroutine	23	24	95.8
pod	7	8	87.5
total	188	224	83.9

package Data::Processor;

use strict;
use 5.010_001;
our $VERSION = '1.0.9';

use Carp;
use Scalar::Util qw(blessed);
use Data::Processor::Error::Collection;
use Data::Processor::Validator;
use Data::Processor::Transformer;
use Data::Processor::Generator;
use Data::Processor::PodWriter;
use Data::Processor::ValidatorFactory;

=head1 NAME

Data::Processor - Transform Perl Data Structures, Validate Data against a Schema, Produce Data from a Schema, or produce documentation directly from information in the Schema.

=head1 SYNOPSIS

  use Data::Processor;
  my $schema = {
    section => {
        description => 'a section with a few members',
        error_msg   => 'cannot find "section" in config',
        members => {
            foo => {
                # value restriction either with a regex..
                value => qr{f.*},
                description => 'a string beginning with "f"'
            },
            bar => {
                # ..or with a validator callback.
                validator => sub {
                    my $self   = shift;
                    my $parent = shift;
                    # undef is "no-error" -> success.
                    no strict 'refs';
                    return undef
                        if $self->{value} == 42;
                }
            },
            wuu => {
                optional => 1
            }
        }
    }
  };

  my $p = Data::Processor->new($schema);

  my $data = {
    section => {
        foo => 'frobnicate',
        bar => 42,
        # "wuu" being optional, can be omitted..
    }
  };

  my $error_collection = $p->validate($data, verbose=>0);
  # no errors :-)

  # in case of errors:
  # ------------------
  # print each error on one line.
  say $error_collection;

  # same
  for my $e ($error_collection->as_array){
      say $e;
      # do more..
  }

=head1 DESCRIPTION

Data::Processor is a tool for transforming, verifying, and producing Perl data structures from / against a schema, defined as a Perl data structure.

=head1 METHODS

=head2 new

 my $processor = Data::Processor->new($schema);

optional parameters:
- indent: count of spaces to insert when printing in verbose mode. Default 4
- depth: level at which to start. Default is 0.
- verbose: Set to a true value to print messages during processing.

=cut
sub new{
    my $class  = shift;
    my $schema = shift;
    my %p     = @_;
    my $self = {
        schema      => $schema // {},
        errors      => Data::Processor::Error::Collection->new(),
        depth       => $p{depth}  // 0,
        indent      => $p{indent} // 4,
        parent_keys => ['root'],
        verbose     => $p{verbose} // undef,
    };
    bless ($self, $class);
    my $e = $self->validate_schema;
    if ($e->count > 0){
        croak "There is a problem with your schema:".join "\n", $e->as_array;
    }
    return $self;
}

=head2 validate
Validate the data against a schema. The schema either needs to be present
already or be passed as an argument.

 my $error_collection = $processor->validate($data, verbose=>0);
=cut
sub validate{
    my $self = shift;
    my $data = shift;
    my %p    = @_;

    $self->{validator}=Data::Processor::Validator->new(
        $self->{schema} // $p{schema},
        verbose     => $p{verbose} // $self->{verbose} // undef,
        errors      => $self->{errors},
        depth       => $self->{depth},
        indent      => $self->{indent},
        parent_keys => $self->{parent_keys},
    );
    return $self->{validator}->validate($data);
}

=head2 validate_schema

check that the schema is valid.
This method gets called upon creation of a new Data::Processor object.

 my $error_collection = $processor->validate_schema();

=cut

sub validate_schema {
    my $self = shift;
    my $vf = Data::Processor::ValidatorFactory->new;
    my $bool = $vf->rx(qr(^[01]$),'Expected 0 or 1');
    my $schemaSchema;
    $schemaSchema = {
        '.+' => {
            regex => 1,
            optional => 1,
            description => 'content description for the key',
            members => {
                description => {
                    description => 'the description of this content of this key',
                    optional => 1,
                    validator => $vf->rx(qr(^.+$),'expected a description string'),
                },
                example => {
                    description => 'an example value for this key',
                    optional => 1,
                    validator => $vf->rx(qr(^.+$),'expected an example string'),
                },
                no_descend_into => {
                    optional => 1,
                    description => 'do not check inside this node',
                },
                regex => {
                    description => 'should this key be treated as a regular expression?',
                    optional => 1,
                    validator => $bool
                },
                value => {
                    description => 'a regular expression describing the expected value',
                    optional => 1,
                    validator => sub {
                        ref shift eq 'Regexp' ? undef : 'expected a regular expression value (qr/.../)'
                    }
                },
                error_msg => {
                    description => 'an error message for the case that the value regexp does not match',
                    optional => 1,
                    validator => $vf->rx(qr(^.+$),'expected an error message string'),
                },
                optional => {
                    description => 'is this key optional ?',
                    optional => 1,
                    validator => $bool,
                },
                default => {
                    description => 'the default value for this key',
                    optional => 1
                },
                array => {
                    description => 'is the value of this key expected to be an array? In array mode, value and validator will be applied to each element of the array.',
                    optional => 1,
                    validator => $bool
                },
                allow_empty => {
                    description => 'allow empty entries in an array',
                    optional => 1,
                    validator => sub {
                        my ($value, $parent) = @_;
                        return 'allow_empty can only be set for array' if !$parent->{array};
                        return $bool->($value);
                    }
                },
                order => {
                    description => 'numeric value to specify the validation order',
                    optional => 1,
                    validator => $vf->rx(qr(^\d+$), 'expected an integer'),
                },
                members => {
                    description => 'what keys do I expect in a hash hanging off this key',
                    optional => 1,
                    validator => sub {
                        my $value = shift;
                        if (ref $value ne 'HASH'){
                            return "expected a hash"
                        }
                        my $subVal=Data::Processor::Validator->new($schemaSchema,%$self);
                        my $e = $subVal->validate($value);
                        return ( $e->count > 0 ? join("\n", $e->as_array) : undef);
                    }
                },
                validator => {
                    description => 'a callback which gets called with (value,section) to validate the value. If it returns anything, this is treated as an error message',
                    optional => 1,
                    validator => sub {
                        my $v = shift;
                        # "0" is a valid package, but is "false"
                        my $blessed = blessed $v;
                        if (defined $blessed){
                            $v->can('validate') && return undef;
                            return 'validator object must implement method "validate()"';
                        }
                        ref $v eq 'CODE' ? undef : 'expected a callback';
                    },
                    example => 'sub { my ($value,$section) = @_; return $value <= 1 ? "value must be > 1" : undef}'
                },
                transformer => {
                    description => 'a callback which gets called on the value with (value,section) to transform the date into a format suitable for further processing. This is called BEFORE the validator. Die with C<{msg=>"error"}> if there is a problem!',
                    optional => 1,
                    validator => sub {
                        ref shift eq 'CODE' ? undef : 'expected a callback'
                    }
                },
                'x-.+' => {
                    optional => 1,
                    regex => 1,
                    description => 'metadata'
                }
            }
        }
    };
    return Data::Processor::Validator->new($schemaSchema,%$self)->validate($self->{schema});
}

=head2 merge_schema

merges another schema into the schema (optionally at a specific node)

 my $error_collection = $processor->merge_schema($schema_2);

merging rules:
 - merging transformers will result in an error
 - merge checks if all merged elements match existing elements
 - non existing elements will be added from merging schema
 - validators from existing and merging schema get combined

=cut

sub merge_schema {
    my $self      = shift;
    my $schema    = shift;
    my $mergeNode = $self->{schema};

    for my $key (@{$_[0]}){
        exists $mergeNode->{$key} || ($mergeNode->{$key} = {});
        $mergeNode = $mergeNode->{$key};
    }

    my $mergeSubSchema;
    $mergeSubSchema = sub {
        my $subSchema      = shift;
        my $otherSubSchema = shift;

        my $checkKey = sub {
            my $elem  = shift;
            my $key   = shift;

            #nothing to do if key value is not defined
            return if !defined $otherSubSchema->{$elem}->{$key};

            if (!defined $subSchema->{$elem}->{$key}){
                $subSchema->{$elem}->{$key} = $otherSubSchema->{$elem}->{$key};
            }
            elsif ($subSchema->{$elem}->{$key} ne $otherSubSchema->{$elem}->{$key}){
                croak "merging element '$elem' : $key does not match";
            }
        };

        for my $elem (keys %$otherSubSchema){
            #copy whole sub schema if element does not yet exist or is empty
            if (!(exists $subSchema->{$elem} && %{$subSchema->{$elem}})){
                $subSchema->{$elem} = $otherSubSchema->{$elem};
                next;
            }

            #merge members subtree recursively
            if (exists $otherSubSchema->{$elem}->{members}) {
                exists $subSchema->{$elem}->{members} || ($subSchema->{$elem}->{members} = {});
                $mergeSubSchema->($subSchema->{$elem}->{members}, $otherSubSchema->{$elem}->{members});
            }

            #check elements
            for my $key (qw(description example default error_msg regex array value)){
                $checkKey->($elem, $key);
            }

            #special handler for transformer
            if ($otherSubSchema->{$elem}->{transformer}) {
                croak "merging element '$elem': merging conflicting transformers not allowed"
                    if $subSchema->{$elem}->{transformer}
                        && $subSchema->{$elem}->{transformer} != $otherSubSchema->{$elem}->{transformer};

                $subSchema->{$elem}->{transformer} = $otherSubSchema->{$elem}->{transformer};
            }

            #special handler for optional: set it mandatory if at least one is not optional
            delete $subSchema->{$elem}->{optional}
                if !($subSchema->{$elem}->{optional} && $otherSubSchema->{$elem}->{optional});

            #special handler for validator: combine validator subs
            if ($otherSubSchema->{$elem}->{validator}) {
                if ((my $validator = $subSchema->{$elem}->{validator})
                    && $subSchema->{$elem}->{validator} != $otherSubSchema->{$elem}->{validator}) {

                    $subSchema->{$elem}->{validator} = sub {
                        return $validator->(@_) // $otherSubSchema->{$elem}->{validator}->(@_);
                    };
                }
                else{
                    $subSchema->{$elem}->{validator}
                        = $otherSubSchema->{$elem}->{validator};
                }
            }
        }
    };

    $mergeSubSchema->($mergeNode, $schema);

    return $self->validate_schema;
}

=head2 schema

Returns the schema. Useful after schema merging.

=cut

sub schema{
    return shift->{schema};
}

=head2 transform_data

Transform one key in the data according to rules specified
as callbacks that themodule calls for you.
Transforms the data in-place.

 my $validator = Data::Processor::Validator->new($schema, data => $data)
 my $error_string = $processor->transform($key, $schema_key, $value);

This is not tremendously useful at the moment, especially because validate()
transforms during validation.

=cut
# XXX make this traverse a data tree and transform everything
# XXX across.
# XXX Before hacking something here, think about factoring traversal out of
# XXX D::P::Validator
sub transform_data{
    my $self = shift;
    my $key  = shift;
    my $schema_key = shift;
    my $val  = shift;

    return Data::Processor::Transformer->new()->transform($key, $schema_key, $val);
}

=head2 make_data

Writes a data template using the information found in the schema.

 my $data = $processor->make_data(data=>$data);

=cut
sub make_data{
    my $self = shift;
    my $entry_point = shift // $self->{schema};
    return Data::Processor::Generator::make_data_template($entry_point);
}

=head2 make_pod

Write descriptive pod from the schema.

 my $pod_string = $processor->make_pod();

=cut
sub pod_write{
    my $self = shift;
    return Data::Processor::PodWriter::pod_write(
        $self->{schema},
        "=head1 Schema Description\n\n"
    );
}

=head1 SCHEMA REFERENCE


=head2 Top-level keys and members

The schema is described by a nested hash. At the top level, and within a
members definition, the keys are the same as the structure you are
describing. So for example:

 my $schema = {
     coordinates => {
         members => {
             x => {
                 description => "the x coordinate",
             },
             y => {
                 description => "the y coordinate",
             },
         }
     }
 };


This schema describes a structure which might look like this:

 { coordinates => { x => 1, y => 2} }

Obviously this can be nested all the way down:

  my $schema = {
     house => {
        members => {
            bungalow => {
                members => {
                    rooms => {
                      #...
                    }
                }
            }
        }
     }
  };

=head2 array

To have a key point to an array of things, simply use the array key. So:

 my $schema = {
    houses => {
       array => 1,
    }
 };

Would describe a structure like:

 { houses => [] }

And of course you can nest within here so:

 my $schema = {
    houses => {
       array => 1,
       members => {
           name => {},
           windows => {
               array => 1,
           }
       },
    },
 };

Might describe:

 {
   houses => [
      { name => 'bob',
        windows => []},
      { name => 'harry',
        windows => []},
   ]
 }

=head2 description

The description key within a definition describes that value:

 my $schema = {
     x => { description => 'The x coordinate' },
 };

=head2 error_msg

The error_msg key can be set to provide extra context for when a value is not
found or fails the L test.

=head2 optional

Most values are required by default. To reverse this use the "optional" key:

 my $schema = {
     x => {
       optional => 1,
     },
     y => {
       # required
     },
 };

=head2 regex

B

If you set "regex" within a definition then it's key will be treated as a
regular expression.

 my $schema = {
    'color_.+' => {
       regex => 1
    },
 };
 my $data = { color_red => 'red', color_blue => 'blue'};
 Data::Processor->new($schema)->validate($data);

=head2 transformer

B

Transformer maps to a sub ref which will be passed the value and the containing
structure. Your return value provides the new value.

 my $schema = {
    x => {
        transformer => sub{
           my( $value, $section ) = @_;
           $value = $value + 1;
           return $value;
        }
    }
 };
 my $data = { x => 1 };
 my $p = Data::Processor->new($schema);
 my $val = Data::Processor::Validator->new( $schema, data => $data);
 $p->transform_data('x', 'x', $val);
 say $data->{x}; #will print 2

If you wish to provide an error from the transformer you should die with a
hash reference with a key of "msg" mapping to your error:


 my $schema = {
    x => {
         transformer => sub{
             die { msg => "SOMETHING IS WRONG" };
         }
    },
 };

 my $p = Data::Processor->new($schema);
 my $data = { x => 1 };
 my $val = Data::Processor::Validator->new( $schema, data => $data);
 my $error = $p->transform_data('x', 'x', $val);

 say $error; # will print: error transforming 'x': SOMETHING IS WRONG


The transformer is called before any validator, so:

 my $schema = {
    x => {
        transformer => sub{
           my( $value, $section ) = @_;
           return $value + 1;
        },
        validator => sub{
           my( $value ) = @_;
           if( $value < 2 ){
              return "too low"
           }
        },
    },
 };
 my $p = Data::Processor->new( $schema );
 my $data = { x => 1 };
 my $errors = $p->validate();
 say $errors->count; # will print 0
 say $data->{x}; # will print 2

=head2 value

B

To check a value against a regular expression you can use the I key
within a definition, mapped to a quoted regex:

 my $schema = {
     x => {
        value => qr{\d+}
     }
 };

=head2 validator

B

To conduct extensive checks you can use I and provide a
callback. Your sub will be passed the value and it's container. If you return
anything it will be regarded as an error message, so to indicate a valid
value you return nothing:

 my $schema = {
    bob => {
      validator => sub{
         my( $value, $section ) = @_;
         if( $value ne 'bob' ){
            return "Bob must equal bob!";
         }
         return;
      },
    },
 };
 my $p = Data::Processor->new($schema);
 # would validate:
 $p->validate({ bob => "bob" });
 # would fail:
 $p->validate({ bob => "harry"});

See also L

=head3 Validator objects

Validator may also be an object, in this case the object must implement a
"validate" method.

The "validate" method should return undef if valid, or an error message string if there is a problem.

 package FiveChecker;

 sub new {
     bless {}, shift();
 }

 sub validate{
     my( $self, $val ) = @_;
     $val == 5 or return "I wanted five!";
     return;
 }
 package main;

 my $checker = FiveChecker->new;
 my $schema = (
     five => (
         validator => $checker,
     ),
 );
 my $dp = Data::Processor->new($schema);
 $dp->validate({five => 6}); # fails
 $dp->validate({five => 5}); # passes

You can for example use MooseX::Types and Type::Tiny type constraints that are objects
which offer validate methods which work this way.

 use Types::Standard -all;

 # ... in schema ...
      foo => {
          validator => ArrayRef[Int],
          description => 'an arrayref of integers'
      },

=head1 AUTHOR

Matthias Bloch [e-mail]

=head1 COPYRIGHT

Copyright 2015- Matthias Bloch

=head1 LICENSE

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

=cut
1;
__END__