[Raku PDF Project] / PDF::Tags

PDF-Tags-raku

A small DOM-like API for the creation of tagged PDF files for accessibility purposes.

This module enables PDF tagged content manipulation, construction, XPath queries and basic XML serialization.

See also PDF::Tags::Reader, which is designed to read content from existing tagged PDF files.

Synopsis

use PDF::Tags;
use PDF::Tags::Elem;

use PDF::API6;
use PDF::Page;
use PDF::XObject::Image;

my PDF::API6 $pdf .= new;
my PDF::Tags $tags .= create: :$pdf;
# create the document root
my PDF::Tags::Elem $doc = $tags.Document: :Lang<en-NZ>;

my PDF::Page $page = $pdf.add-page;
my $header-font = $page.core-font: :family<Helvetica>, :weight<bold>;
my $body-font = $page.core-font: :family<Helvetica>;

$page.graphics: -> $gfx {

    $doc.Header1: $gfx, {
        .say('Marked Level 1 Header',
             :font($header-font),
             :font-size(15),
             :position[50, 120]);
    };

    $doc.Paragraph: $gfx, {
        .say('Marked paragraph text', :position[50, 100], :font($body-font), :font-size(12));
    };

    # add a marked image
    my PDF::XObject::Image $img .= open: "t/images/lightbulb.gif";
    $doc.Figure: $gfx, $img, :Alt('Incandescent apparatus');

}

$pdf.save-as: "/tmp/synopsis.pdf"

Description

A tagged PDF contains additional markup information describing the logical document structure of PDF documents.

PDF tagging may assist PDF readers and other automated tools in reading PDF documents and locating content such as text and images.

This module provides a DOM like interface for creating and traversing PDF structure and content via tags. It also has an XPath like search capability. It is designed for use in conjunction with PDF::Class or PDF::API6.

Standard Tags

Elements may be constructed using their Tag name or Mnemonic, as listed below. For example:

$root.P: $gfx, { .say('Marked paragraph text') };

Can also be written as:

$root.Paragraph: $gfx, { .say('Marked paragraph text') };

Or as:

$root.add-kid(:name<P>).mark: $gfx, { .say('Marked paragraph text') };

"Grouping" elements:

"Block" elements:

Mnemonic | Tag | Description

"Table" elements:

"Inline" elements:

"Illustration" elements (should have Alt and/or ActualText set):

Non-structure tags:

Classes in this Distribution

• PDF::Tags - Tagged PDF root node
• PDF::Tags::Attr - A single node attribute
• PDF::Tags::Elem - Structure Tree descendant node
• PDF::Tags::Node - Abstract node
• PDF::Tags::Node::Parent - Abstract parent node
• PDF::Tags::Mark - Leaf content marker node
• PDF::Tags::Text - Text content node
• PDF::Tags::ObjRef - A reference to a PDF object (PDF::Annot, PDF::Field or PDF::XObject)
• PDF::Tags::XML-Writer - XML Serializer
• PDF::Tags::XPath - XPath evaluation context

Advanced Topics

Form and Image XObjects

In the simple case, both PDF::XObject::Form's and PDF::XObject::Image's are inserted and externally tagged as an atomic graphical element, typically tagged as Figure or Form:

my PDF::XObject::Image $img .= open: "t/images/lightbulb.gif";

my $figure = $doc.Figure: $gfx, $img, :position[50, 70], :Alt("A light-bulb");

A PDF::XObject::Form may be associated with a document fragment. The form can then be rendered, and the fragment inserted into the document, by repeatedly calling do on the fragment, as demonstrated below:

use PDF::Tags;
use PDF::Tags::Elem;
use PDF::Class;
use PDF::XObject::Form;
my PDF::Class $pdf .= new;
my PDF::Tags $tags .= create: :$pdf;
my PDF::Tags::Elem $doc = $tags.Document;

my PDF::Page $page = $pdf.add-page;
$page.graphics: -> $gfx {
   $doc.Header1: $gfx, {
        .say('Header text');
   }

    my PDF::XObject::Form $form = $page.xobject-form: :BBox[0, 0, 200, 50];
    my PDF::Tags::Elem $form-frag = $doc.fragment;

    $form.text: {
        my $font-size = 12;
        .text-position = [10, 38];
        $form-frag.Header2: $_, {
            .say: "Tagged XObject header";
        };
        my $p = $form-frag.Paragraph: $_, {
            .say: "Some sample tagged text";
        };
    }

    # multiple rendering of the form, and insertion into the structure tree
    $doc.do($gfx, $form-frag, :position[150, 70]);
    $doc.do($gfx, $form-frag, :position[150, 20]);
}

Links

Links are usually contained in a block element, such as a Paragraph. If the link is internal, it should be further enclosed in a Reference element.

Furthermore, if the link is in flowing text, such as a paragraph, the mark method may be needed to mark preceding text, the link, and following text.

Please see examples/link.raku which demonstrates adding an tagged internal link to a PDF.

Graphics Content Tags

As a rule, all content doesn't have to form part of the structure tree, but should be tagged to meet accessibility guidelines.

This sometimes requires tagging of incidental graphics. PDF::Content has a tag() method for this. The content is tagged, but does not appear in the structure tree.

Some of the commonly used content tags are:

Artifact

Artifact content forms part of the visual display, but does not belong in the structure Tree.

For example:

$gfx.tag: Artifact, {
    .say("Page $page-num", :$font, :position[ 250, 20 ]);
}

Clipped

A clipped region encompasses additional graphics that are being used as part of a clipping operation. The clipped area may include graphics that are part of the structure tree. For example:

use PDF::Class;
use PDF::Tags;
use PDF::Tags::Elem;
use PDF::Content::Tag :ContentTags, :ParagraphTags;

my PDF::Class $pdf .= new;
my PDF::Tags $tags .= create: :$pdf;
# create the document root
my PDF::Tags::Elem $doc = $tags.Document;

$pdf.add-page.graphics: {
    .tag: Clipped, {
        .Rectangle: 100, 100, 125, 20;
        .Clip;
        .EndPath;
        $doc.Paragraph: $_, {
            .say: 'Clip me', :position[98, 98];
        }
    }
}

The above example is setting up a clipping sequence. The clipped text is being inserted as a paragraph in the structure tree.

Span

This tag may be used either in the structure tree, or at the content level to defined attributes of a graphics sequence. Its usage is similar to the XHTML span tag.

$gfx.tag: Span: :Lang<es-MX>, {
    .say('Hasta la vista', :position[50, 80]);
}

Role Maps

The PDF standard allows user defined tags, which are mapped to standard tags via a role-map table. PDF::Tags supports this feature, via the create :role-map option. One declared, role-map entries, can be called as standard construction methods:

use PDF::Tags;
use PDF::Tags::Elem;
use PDF::Class;
use PDF::Page;

enum Roles ( :Book<Document>, :FootNote<Note> );
constant %role-map = Roles.enums.Hash;

my PDF::Class $pdf .= new;
my PDF::Tags $tags .= create: :$pdf, :%role-map;

my PDF::Tags::Elem $doc = $tags.Book;
my PDF::Page $page = $pdf.add-page;

$page.graphics: -> $gfx {

    $doc.Paragraph: $gfx, {
        .say: 'Some body text¹', :position[50, 150];
    };

    $doc.FootNote: $gfx, {
        .say: '¹With a foot-note', :position[50, 20];
    };

}

Content Continuation

Content sometimes needs to be continued across graphics elements, such as a paragraph that spans multiple pages. The mark method may be called repeatably to achieve this:

use PDF::Class;
use PDF::Tags;
use PDF::Tags::Elem;
use PDF::Page;
use PDF::Content::FontObj;

my PDF::Class $pdf .= new;
my PDF::Tags $tags .= create: :$pdf;
my PDF::Tags::Elem $doc = $tags.Document;
my PDF::Page $page = $pdf.add-page;
my PDF::Content::FontObj $font = $pdf.core-font: :family<Helvetica>;
my PDF::Tags::Elem $para = $doc.Paragraph;

$page.graphics: -> $gfx {

    $para.mark: $gfx, {
        .say('This paragraph starts on first page...',
             :$font,
             :font-size(15),
             :position[50, 600]);
    };

}

$page = $pdf.add-page;
$page.graphics: -> $gfx {

    $para.mark: $gfx, {
        .say('...and finishes on the second page',
             :$font,
             :font-size(15),
             :position[50, 600]);
    };

}

$pdf.save-as: "span.pdf";

Tagging and Metadata

If a PDF document is tagged for accessibility, additional metadata usually also needs to be set-up; typically Title, Author, Subject, Keywords and language. For example:

use PDF::API6;
use PDF::Info;
my PDF::API6 $pdf .= open: "/tmp/synopsis.pdf";

my PDF::Info $info = $pdf.Info //= {};
$info.Title = 'PDF::Tags README Synopsis';
$info.Subject = 'Sample Output PDF';
$info.Author = 'David Warring';
$info.Keywords = 'accessibility,pdf,tags';
$pdf.catalog.Lang = 'en';
$pdf.update;

Note that PDF::API6 automatically fills in Producer, Creator and CreationDate.

Verification

The pdf-tag-dump.raku script from the PDF::Tags::Reader module can be used to view the logical content of PDF files as XML, for example:

$ pdf-tag-dump.raku /tmp/synopsis.pdf

Produces

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE Document SYSTEM "http://pdf-raku.github.io/dtd/tagged-pdf.dtd">
<?xml-stylesheet type="text/css" href="https://pdf-raku.github.io/css/tagged-pdf.css"?>
<Document>
  <H1>
    Marked Level 1 Header
  </H1>
  <P>
    Marked paragraph text
  </P>
  <Figure BBox="0 0 19 19">
  </Figure>
</Document>

The XML output from pdf-tag-dump.raku includes an external DtD for basic validation purposes.

For example, it can be piped to xmllint, from the libxml2 package, to check the structure of the tags:

$ pdf-tag-dump.raku /tmp/synopsis.pdf | xmllint --noout --valid -

See Also

• PDF::Tags::Reader - for reading PDF files with existing tagged content

Further Work

• Type-casting of PDF::StructElem.A to roles; as per 14.8.5. Possibly belongs in PDF::Class, however slightly complicated by the need to apply role-mapping.

• Develop a tag/accessibility checker. A low-level sanity checker that a tagged PDF meets PDF association recommendations pdf-tag-checker.raku --ua. See https://www.pdfa.org/wp-content/uploads/2014/06/MatterhornProtocol_1-02.pdf and Wikipedia Clause 7 guidelines:

  • Complete tagging of "real content" in logical reading order
  • Tags must correctly represent the document's semantic structures (headings, lists, tables, etc.)
  • Problematic content is prohibited, including illogical headings, the use of color/contrast to convey information, inaccessible JavaScript, and more
  • Meaningful graphics must include alternative text descriptions
  • Security settings must allow assistive technology access to the content
  • Fonts must be embedded, and text mapped to Unicode

The PDF accessibility standard ISO 14289-1 cannot be distributed and needs to be purchased from ISO.

• Editing. Currently the API doesn't readily support editing tags into existing content. More work is also needed in the PDF::Content module to support content editing.