A sax parser for pod
my $h = XML::SAX::Writer->new();
my $p = Pod::SAX->new( Handler => $h );
$p->parse_uri('perlpodspec.pod');
Very simply, this module parses \s-1POD\s0 (or perl) files and turns the Plain Old Documentation into \s-1SAX\s0 events (which often you'll use to turn into \s-1XML\s0, but there are other uses as well).
The aim of this module is not round-tripping, so some things may be lost in the conversion. The aim is to be as standards compliant as possible, while giving you very simple access to the data.
The main motivation for this module though was simple standards compliance - all the Pod parsers out there seem to have their own unique way of doing things, and so my aim was to unify that and allow the flexibility that \s-1SAX\s0 gives me at the same time.
For an introduction to \s-1SAX\s0, please read XML::SAX::Intro.
One very important point to note is that just because this is a \s-1SAX\s0 module it doesn't mandate that the results are \s-1XML\s0. You could just as easily use this module to extract all filenames from a \s-1POD\s0 file, or extract custom =for/=begin sections. And because it uses standardised interfaces this is a lot simpler than working with any other \s-1POD\s0 parser out there, and the knowledge is transferrable.
To construct a parser simply call new(). It is customary to pass in the handler object that will receive the \s-1SAX\s0 events at this time, though you do not have to:
my $parser = Pod::SAX->new(Handler => $h);
You can re-use this parser object multiple times. It's possible to change the handler at a later date using \*(C`$parser-\*(C'set_handler()>. This and many other \s-1API\s0 calls are documented in XML::SAX::Base, which this module inherits from.
This method is an auto-detecting parser - it will try and figure out what you passed to it (a string, a file handle or a filename) and parse the data using the appropriate technique.
These are simply the non-detecting methods that parse() uses internally. Use these if you are paranoid about what you're parsing, and don't want the overhead of \s-1SAX\s0 trying to guess.
The \s-1XML\s0 format is intended to be simple and map fairly closely to the source \s-1POD\s0. The documentation here shows the \s-1POD\s0 marker and the tag that it maps to.
<pod> <!-- Pod::SAX v0.14, using POD::Parser v1.13 -->
The comment is automatically generated so that you can see what version of Pod::SAX was used in parsing this document. The closing \*(C`</pod>\*(C' tag is generated when the end of the \s-1POD\s0 is reached.
<head1>Text here</head1>
All head levels are supported.
Plain paragraphs are represented with:
<para>text</para>
Verbatim sections (i.e. when you indent the text) are represented with:
<verbatim>text</verbatim>
Pod::SAX automatically detects whether a list is itemized or ordered (i.e. whether it should have bullet points or numbers), and so =over/=back are represented by:
<itemizedlist> </itemizedlist>
and
<orderedlist> </orderedlist>
respectively. The indent value (as in \*(L"=over 4\*(R") is saved in the \*(C`indent_width\*(C' attribute, although for most purposes this can be ignored.
For both bulleted and numbered lists, the =item tag always maps to:
<listitem>text</listitem>
If a paragraph follows an =item tag (and occurs before any =back) then the paragraph is included immediately after the tag, so for example:
=item foo
Some text about foo
Maps to:
<listitem>foo <para>Some text about foo</para> </listitem>
And \*(L"=for foo\*(R" (the two are semantically equivalent in Pod)
<markup type="foo" ordinary_paragraph="0">text here</markup>
If the markup section is meant for ordinary processing (see the perlpodspec section on \*(L"About Data Paragraphs and \*(R"=begin/=end\*(L" Regions\*(R"), which means the type name begins with a colon as in:
=begin :biblio
Then the markup produced indicates that using:
<markup type="biblio" ordinary_paragraph="1"/>
And the parser will expand all interior Pod commands as it should.
Note: There is no special treatment of =begin html or =begin \s-1XML\s0 or any variant thereof. The contents of those markers will simply be treated as text, and it is up to the user of this module to parse that data as \s-1XML\s0 if they wish to do so.
Sequences in \s-1POD\s0 consist of the following:
L<> - Links E<> - Entities I<> - Italics B<> - Bold C<> - Code F<> - Filename S<> - Non breaking space X<> - Index marker Z<> - Null
Most sequences are simply converted to tags of the same name, case preserved:
<B>This is bold text</B> and also <I>some in italics</I>.
Special treatment is given to the L<> E<> and S<> tags only.
Links
Links in Pod are... funky.
Parsing links is really hard, so don't expect that I've got this right. Basically though you've got this mapping:
L<foo/bar> => <link type="pod" page="foo" section="bar">foo</link>
L<Some Foo|foo> => <link type="pod" page="foo" section="">Some Foo</link>
L<select(3)> => <link type="man" page="select(3)" section="">select(3)</link>
L<http://foo.com> => <xlink href="http://foo.com">http://foo.com</xlink>
And many variations thereof! Basically it should do the right thing
Entities
In \s-1POD\s0 an E<> marker defines an entity. In pod these are single characters only, and take either a text form, in which case they map to the standard \s-1HTML\s0 entities (e.g. ouml, Agrave etc), or a decimal number in which case they map to the unicode character at that code point.
In Pod::SAX entities are always converted to unicode and never generate any tags or markers in the data stream so it will be as though the entity was never there.
Non-Breaking Space
Non breaking space is simply achieved by changing all space within the S<> section into the unicode codepoint 160 - the non-breaking space character. Normally this is enough to do the right thing, but if you need to you can detect this with a regexp.
Matt Sergeant, [email protected]. Copyright AxKit.com Ltd 2002
There may be bugs in the unicode handling on perl 5.8, because it's just really hard to get things right on 5.8 when dealing with unicode. :-)
This is free software. You may use it and redistribute it under the same terms as Perl itself.