YAML::Active - Combine data and logic in YAML


version 1.100810


  use YAML::Active;
  my $data = Load(<<'EOYAML');
  pid: /YAML::Active::PID
  foo: bar
  include_test: /YAML::Active::Include
      filename: t/testperson.yaml
  ticket_no: /YAML::Active::Concat
    - '20010101.1234'
    - /YAML::Active::PID
    - /YAML::Active::Eval
      code: sub { sprintf "%04d", ++(our $cnt) }
    1: /Registry::YAML::Active::WritePerson
         personname: Foobar
         handle: AB123456-NICAT
    2: /Registry::YAML::Active::WritePerson
       person: /YAML::Active::Include
         filename: t/testperson.yaml


YAML is an intuitive way to describe nested data structures. This module extends YAML's capabilities so that it ceases to be a static data structure and become something more active, with data and logic combined. This makes the logic reusable since it is bound to the data structure. Without YAML::Active, you have to load the YAML data, then process it in some way. The logic describing which parts of the data have to be processed and how was separated from the data. Using YAML::Active, the description of how to process the data can be encapsulated in the data structure itself.

The way this works is to assign a transfer type to the YAML nodes you want to process. The transfer type refers to a Perl package which is expected to have a yaml_active() method which contains the logic; you can think of the array or hash structure below that node as the subroutine's arguments.

YAML::Active provides its own Load() and LoadFile() subroutines which work like the subroutines of the same name in YAML, except that they also traverse the whole data structure, recognizing packages named as transfer types that have a yaml_active() method and calling that method on the given node.

An example:

  some_string: /YAML::Active::Concat
    - foo
    - bar
    - baz

defines a hash key whose value is an active YAML element. When you call YAML::Active's Load() on that data, at some point the hash value is being encountered. The YAML::Active::Concat plugin (as a convenience also defined in the same file as YAML::Active) has a yaml_active() method which expects to be called on an array reference (that is, the thing blessed into YAML::Active::Concat is expected to be an array reference). The method in turn activates all of the array's elements and joins the results. So after loading the data structure, the result is equivalent to

  some_string: foobarbaz

Because YAML::Active::Concat also activates all of its arguments, you can nest activation logic:

  some_string /YAML::Active::Concat
    - foo
    - /YAML::Active::PID
    - /YAML::Active::Eval
      code: sub { sprintf "%04d", ++(our $cnt) }

This active YAML uses two more plugins, YAML::Active::PID and YAML::Active::Eval. YAML::Active::PID replaces its node with the current process's id. Note that even though this plugin doesn't need any arguments, we have to provide something - anything, in fact, whether it be an array reference or a hash reference, because YAML can bless only references. YAML::Active::Eval expects a hash reference with a code key whose value is the source code for an anonymous sub which the plugin calls and whose return value it uses to replace the activated node.

An activation plugin (that is, a package referred to by a node's transfer type) can have any name, but if that name contains the string YAML::Active, it is being required() if it doesn't already provide a yaml_active() method. This is merely a convenience so you don't have to use() or require() the packages beforehand and things work a bit more transparently. If you merely want to bless a node (that is, provide a transfer type) into a package that's not an activation plugin, be sure that the package name doesn't contain the string YAML::Active.



Like YAML's Load(), but activates the data structure after loading it and returns the activated data structure.


Like YAML's Load(), it doesn't activate the data structure.


Like YAML's Dump().


Dumps, then loads the data structure and returns the result.


Like YAML's LoadFile(), but activates the data structure after loading it and returns the activated data structure.


Expects a reference and recursively activates it, returning the resulting reference.

If it encounters an array, it calls array_activate() on the node and returns the result.

If it encounters a hash, it calls hash_activate() on the node and returns the result.

If it encounters a node that can be activated (i.e., that is blessed into a package that has a yaml_activate() method, it activates the node and returns the result. If the package name contains the string YAML::Active and it doesn't have a yaml_activate() method, node_activate() tries to require() the package (as a convenience). That is, if you want to write a plugin, you can either include the string YAML::Active somewhere in its package name, or use any other name but then you'd have to use() or require() it before activating some YAML.

Otherwise it just returns the node as it could be an unblessed scalar or a reference blessed into a package that's got nothing to do with activation.


Takes an array reference and activates every array element in turn, then returns a new array references containing the results. Null elements (that is, elements blessed into YAML::Active::NULL) are ignored.


Takes a hash reference and activates every value, then returns a new hash references containing the results (the hash keys are left alone). Keys with null values (that is, values blessed into YAML::Active::NULL) are ignored.


Checks that its argument is an array reference. If not, die()s reporting the caller.


Checks that its argument is a hash reference. If not, die()s reporting the caller.


Returns an empty hash reference blessed into the YAML::Active::NULL package. This function is used by side-effect-only plugins that don't want to have a trace of their existence left in the activated data structure. For an example see the YAML::Active::Print.


This is a constant with the value YAML::Active::NULL.


Dumps a data structure node. It is used by Dump(). It calls array_dump() and hash_dump() as necessary. For scalar nodes, if the node is a blessed reference and the package it is blessed into has a prepare_dump() function, that function is called. Then, if the package has a yaml_dump() function, that function is called as well. After the recursive dump is finished, each package that had a prepare_dump() function is checked for a finish_dump() function, which is called if it exists.


Dumps an array reference node. It is called by node_dump() and calls node_dump() itself on the array elements.


Dumps a hash reference node. It is called by node_dump() and calls node_dump() itself on the hash values.


Takes as arguments a node and optionally a phase. If the phase argument is given, return true for those nodes that have this phase requirements. If the phase argument is not given, return true for those nodes that don't have a phase requirements.


Nothing is exported by default, but you can request each of the subroutines individually or grouped by tags. The tags and their symbols are, in YAML notation:

    - Load
    - LoadFile
    - node_activate
    - array_activate
    - hash_activate
    - assert_arrayref
    - assert_hashref
    - yaml_NULL
    - NULL

There is also the all tag, which contains all of the above symbols.



Expects an array reference and joins the activated array elements, returning the joined string.

For an example, see the DESCRIPTION above.


Expects a hash reference with a code key. evals the activated hash value returns the result from executing the coderef (passing no arguments).


  - /YAML::Active::Eval
    code: sub { sprintf "%04d", ++(our $cnt) }


  - 1

At least, that's the answer the first time around.


Expects a hash reference with a filename key. Calls YAML::Active's LoadFile() on the activated filename. That is, the filename can itself use an activation plugin, and the file contents are activated as well.


  description: /YAML::Active::Include
    filename: description.yaml


  description: >
    The content of the included file goes here.

Returns the current process id.


  the_pid: /YAML::Active::PID

Result (for example):

  the_pid: 12345

Note that, although this plugin doesn't require any arguments, we have to give it either an array reference or a hash reference, because YAML can't bless something that's not a reference. The contents of the reference don't matter.


Expects an array reference and returns another array reference with the activated original elements in random order.


  data: /YAML::Active::Shuffle
        - 1
        - 2
        - 3
        - 4
        - 5

Result (for example):

    - 3
    - 5
    - 1
    - 2
    - 4

Expects an array reference and joins the activated array elements, printing the result and returning a null (i.e., a YAML::Active::NULL) node. That is, the node won't appear in the resulting activated data structure.


    - foo
    - /YAML::Active::Print
       - '# Hello, world!'
       - 'Goodbye, world!'
    - baz


    - foo
    - baz

and the string # Hello, world!Goodbye, world! is printed.


Replaces node values (scalars, array elements and hash values) with their lowercased value. Does not descend into deeper array references or hash references, but passes them through unaltered.


  data: /YAML::Active::uc
    - Hello
    - world and
    - one: GOoD
      two: byE
    - wOrLd!


    - HELLO
    - one: GOoD
      two: byE
    - WORLD!

Like YAML::Active::uc, but lowercases the values.


Suppose you want to write an activation plugin that takes a reference to an array of numbers and adds them.

By including the string YAML::Active in the package name we can let YAML::Active load the package when necessary. All we need to do is to provide a yaml_activate() method that does the work.

  package My::YAML::Active::Add;

  use YAML::Active qw/array_activate assert_arrayref/;

  sub yaml_activate {
      my $self = shift;
      my $result;
      $result += $_ for @{ array_activate($self) };
      return $result;

Now you can do:

  result: /My::YAML::Active::Add
    - 1
    - 2
    - 3
    - 7
    - 15

And the result would be:

  result: 28

This could be the beginning of a YAML-based stack machine or at least an RPN calculator...


See perlmodinstall for information and options on installing Perl modules.


No bugs have been reported.

Please report any bugs or feature requests through the web interface at


The latest version of this module is available from the Comprehensive Perl Archive Network (CPAN). Visit to find a CPAN site near you, or see

The development version lives at Instead of sending patches, please fork this project using the standard git and github infrastructure.


  Marcel Gruenauer <>


This software is copyright (c) 2003 by Marcel Gruenauer.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.