Pod::Abstract (3pm)

 use Pod::Abstract;
 use Pod::Abstract::BuildNode qw(node);

 # Get all the first level headings, and put them in a verbatim block
 # at the start of the document
 my $pa = Pod::Abstract->load_filehandle(\*STDIN);
 my @headings = $pa->select('/head1@heading');
 my @headings_text = map { $_->pod } @headings;
 my $headings_node = node->verbatim(join "\n",@headings_text);

 $pa->unshift( node->cut );
 $pa->unshift( $headings_node );
 $pa->unshift( node->pod );

 print $pa->pod;

POD::Abstract provides a means to load a \s-1POD\s0 (or \s-1POD\s0 compatible) document without direct reference to it's syntax, and perform manipulations on the abstract syntax tree.

This can be used to support additional features for \s-1POD\s0, to format output, to compile into alternative formats, etc.

\s-1WHY\s0?

If you've ever asked yourself \*(L"What does Pod do for me?\*(R", this module is intended to answer that question.

While Pod looks like a simple format, the specification calls for a number of special cases to be handled, and that makes any software that works on Pod as text more complex than it needs to be.

In addition to this, Pod does not lend itself to a natural structured model. This makes it difficult to manipulate without damaging the validity of the document.

Pod::Abstract solves these problems by loading the document into a structured tree, and providing consistent traversal, searching, manpulation and re-serialisation. Pod related utilities are easy to write using Pod::Abstract.

The design goal of Pod::Abstract is to do the hard work for the programmer - the library should work for you, and as such it should be significantly easier than string mashing what you want out of a Pod document.

\s-1PROCESSING\s0 \s-1MODEL\s0

The intent with POD::Abstract is to provide a means to decorate a parse tree, rather than manipulate text, to allow other software to add features and functionality to \s-1POD\s0 based documenation systems.

If you wish to write modules that interact nicely with other POD::Abstract modules, then you should provide a POD::Abstract -> POD::Abstract translation. Leave any document element that your program is not interested in directly untouched in the parse tree, and if you have data that could be useful to other packages, decorate the parse tree with that data even if you don't see any direct way to use it in the output.

In this way, when you want one more feature for \s-1POD\s0, rather than write or fork a whole translator, a single inline \*(L"decorator\*(R" can be added.

The \*(C`paf\*(C' utility provides a good starting point, which also allows you to hook in to an existing filter/transform library. Simply add a \*(C`Pod::Abstract::Filter\*(C' class to the namespace and it should start working as a \*(C`paf\*(C' command.

\s-1EXAMPLE\s0

Suppose you are frustrated by the verbose list syntax used by regular \s-1POD\s0. You might reasonably want to define a simplified list format for your own use, except \s-1POD\s0 formatters won't support it.

With Pod::Abstract you can write an inline filter to convert:

* item 1 * item 2 * item 3

into:

=over

=item *

item 1

=item *

item 2

=item *

item 3

=back

This transformation can be simply performed on the document tree. If your formatter does not use Pod::Abstract, you can simply pipe out \s-1POD\s0 and use a regular formatter. If your formatter supports Pod::Abstract though, then you can feed in the syntax tree directly without having to re-serialise and parse the document.

In addition to this, because the source document is still valid Pod, you aren't breaking compatibility with regular perldoc just by making Pod::Abstract transformations.

\s-1POD\s0 \s-1SUPPORT\s0

Pod::Abstract aims to support all \s-1POD\s0 rules defined in perlpodspec (even the ones I don't like), except for those directly related to formatting output, or which cannot be implemented generically.

Pod::Abstract is comprised of:

• The parser, which loads a document tree for you. You should access this through \*(C`Pod::Abstract\*(C', not directly

• The document tree, which is the root node you are given by the parser. Calling pod on the root node should always give you back your original document. See Pod::Abstract::Node

• Pod::Abstract::Path, the node selection expression language. This is generally called by doing \*(C`$node->select(PATH_EXP)\*(C'. Pod::Abstract::Path is the most complex and powerful component of this module, and if you're not using it you should be. ;) This allows you to ask questions like: \*(L"In the first head1 that starts with \*(R"A\*(L", find me the head2 matching 'foo' with bold text somewhere in the preceding paragraph or heading\*(R" /head1[@heading=~{^A}](0)/head2[@heading=~{foo}i]<<head2 :paragraph[//:B] You probably don't need anything that complex, but it's there if you do.

• The node builder, Pod::Abstract::BuildNode

load_file

my $pa = Pod::Abstract->load_file( FILENAME );

Read the \s-1POD\s0 document in the named file. Returns the root node of the document.

load_filehandle

my $pa = Pod::Abstract->load_file( FH );

Load a \s-1POD\s0 document from the provided filehandle reference. Returns the root node of the document.

load_string

my $pa = Pod::Abstract->load_string( STRING );

Loads a \s-1POD\s0 document from a scalar string value. Returns the root node of the document.

Ben Lilburne <[email protected]>

Copyright (C) 2009 Ben Lilburne

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