HTML::HTML5::Outline(3) implementation of the HTML5 Outline algorithm

SYNOPSIS


use JSON;
use HTML::HTML5::Outline;

my $html = <<'HTML';
<!doctype html>
<h1>Hello</h1>
<h2>World</h2>
<h1>Good Morning</h1>
<h2>Vietnam</h2>
HTML

my $outline = HTML::HTML5::Outline->new($html);
print to_json($outline->to_hashref, {pretty=>1,canonical=>1});

DESCRIPTION

This is an implementation of the HTML5 Outline algorithm, as per <http://www.w3.org/TR/html5/sections.html#outlines>.

The module can output a JSON-friendly hashref, or an RDF model.

Constructor

"HTML::HTML5::Outline->new($html, %options)"

Construct a new outline. $html is the HTML to generate an outline from, either as an HTML or XHTML string, or as an XML::LibXML::Document object.

Options:

  • default_language - default language to assume text is in when no lang/xml:lang attribute is available. e.g. 'en-gb'.
  • element_subjects - rather advanced feature that doesn't bear explaining. See USE WITH RDF::RDFA::PARSER for an example.
  • microformats - support "<ul class="xoxo">", "<ol class="xoxo">" and "<whatever class="figure">" as sectioning elements (like "<section>", "<figure>", etc). Boolean, defaults to false.
  • parser - 'html' (default) or 'xml' - choose the parser to use for XHTML/HTML. If the constructor is passed an XML::LibXML::Document, this is ignored.
  • suppress_collections - allows rdf:List stuff to be suppressed from RDF output. RDF output - especially in Turtle format - looks somewhat nicer without them, but if you care about the order of headings and sections, then you'll want them. Boolean, defaults to false.
  • uri - the document URI for resolving relative URI references. Only really used by the RDF output.

Object Methods

  • "to_hashref"

    Returns data as a nested hashref/arrayref structure. Dump it as JSON and you'll figure out the format pretty easily.

  • "to_rdf"

    Returns data as a n RDF::Trine::Model. Requires RDF::Trine to be installed. Otherwise this method won't exist.

  • "primary_outlinee"

    Returns a HTML::HTML5::Outline::Outlinee element representing the outline for the page.

Class Methods

  • "has_rdf"

    Indicates whether the "to_rdf" object method exists.

USE WITH RDF::RDFA::PARSER

This module produces RDF data where many of the resources described are HTML elements. RDFa data typically does not, but RDF::RDFa::Parser does also support some extensions to RDFa which do (e.g. support for the "cite" and "role" attributes). It's useful to combine the RDF data from each, and RDF::RDFa::Parser 1.093 and upwards contains a few shims to make this possible.

Without further ado...

        use HTML::HTML5::Outline;
        use RDF::RDFa::Parser 1.093;
        use RDF::TrineShortcuts;
        my $rdfa = RDF::RDFa::Parser->new(
                $html_source,
                $base_url,
                RDF::RDFa::Parser::Config->new(
                        'html5', '1.1',
                        role_attr     => 1,
                        cite_attr     => 1,
                        longdesc_attr => 1,
                        ),
                )->consume;
        
        my $outline = HTML::HTML5::Outline->new(
                $rdfa->dom,
                uri              => $rdfa->uri,
                element_subjects => $rdfa->element_subjects,
                );
        
        # Merging two graphs is pretty complicated in RDF::Trine
        # but a little easier with RDF::TrineShortcuts...
        my $combined = rdf_parse();
        rdf_parse($rdfa->graph,     model => $combined);
        rdf_parse($outline->to_rdf, model => $combined);
        
        my $NS = {
                dc    => 'http://purl.org/dc/terms/',
                o     => 'http://ontologi.es/outline#',
                type  => 'http://purl.org/dc/dcmitype/',
                xs    => 'http://www.w3.org/2001/XMLSchema#',
                xhv   => 'http://www.w3.org/1999/xhtml/vocab#',
                };
        
        print rdf_string($combined => 'Turtle', namespaces => $NS);

AUTHOR

Toby Inkster, <[email protected]>

ACKNOWLEDGEMENTS

This module is a fork of the document structure parser from Swignition <http://buzzword.org.uk/swignition/>.

That in turn includes the following credits: thanks to Ryan King and Geoffrey Sneddon for pointing me towards [the HTML5] algorithm. I also used Geoffrey's python implementation as a crib sheet to help me figure out what was supposed to happen when the HTML5 spec was ambiguous.

COPYRIGHT AND LICENCE

Copyright (C) 2008-2011 by Toby Inkster

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