• Introduction
• Creating a PDF/A Conforming Document
• Best Practices

PDF/A Conformance

Introduction

The PDF/A specifications ISO 19005-1 through 19005-4 define restrictions on the PDF format to enable its use as archival format. Many governments, libraries, news papers, etc. need such a format to permanently store PDF documents.

The benefit of using PDF/A for this is that it restricts the use of certain PDF functionality which would make the PDF file dependent on external resources or special viewers. For example, all fonts have to be embedded and multimedia/Javascript content is forbidden.

While PDF/A-1 is based on PDF 1.4, PDF/A-2 and PDF/A-3 are based on PDF 1.7. The newest iteration PDF/A-4 is based on PDF 2.0.

Each PDF/A specification comes with a set of conformance levels to which a PDF file may conform:

PDF/A-1

The first iteration of PDF/A supports the levels ‘B’ (basic conformance) and ‘A’ (accessible conformance). Level ‘A’ is a superset of ‘B’ and requires PDF files to be tagged.

PDF/A-2

The second iteration of PDF/A supports the same conformance levels as PDF/A-1. Additionally, it adds the level ‘U’ which is between ‘A’ and ‘B’ and requires that all text has Unicode mappings available.

PDF/A-3

This is the same as PDF/A-2 but allows embedding any kind of file (whereas PDF/A-2 only allows embedding PDF/A conforming files).

PDF/A-4

This specification differs a bit from the former because it abandons the ‘B’, ‘U’ and ‘A’ conformance levels. A file may be conforming to PDF/A-4 or to the new conformance levels ‘F’ (PDF/A-4 with embedded files) or ‘E’ (PDF/A-4 for engineering).

HexaPDF supports the creation of PDF files conforming to the levels 2b, 2u, 3b, and 3u. Once tagged PDF is implemented, HexaPDF will eventually support levels 2a and 3a.

Creating a PDF/A Conforming Document

Though HexaPDF supports the creation of PDF/A files, the author needs to make sure they don’t use anything that is forbidden in them.

To tell HexaPDF that a document should be conforming to PDF/A, the :pdfa task needs to be invoked, with the optional level argument specifying the conformance level:

document.task(:pdfa, level: '3b')

The task ensures, among other things, that the basic additions to a standard PDF to make it PDF/A compatible are done.

Here is a partial list of requirements that need to be followed:

• The trailer ID must be set. Note that this is automatically done by HexaPDF when running the validation task before writing.

• No encryption may be used.

• Streams may not use external stream data.

• The LZW filter may not be used.

• Integers must be between -231 und 231-1.

• Strings may not be longer than 32,767 bytes.

• q/Q pairs (save_graphics_state and restore_graphics_state) may not be nested more than 28 levels.

• HexaPDF adds an sRGB output intent if no output intent has been set up. This takes care of using RGB and gray colors. If a document contains CMYK colors, the author needs to make sure to add an appropriate output intent for CMYK and RGB (if used).

• All fonts must be embedded. This means the standard 14 PDF fonts cannot be used! HexaPDF’s PDF/A task enforces this by removing the font loader for the standard 14 PDF fonts.

• The 3D, Sound, Screen, and Movie annotations may not be used.

• All annotations must have the print flag set and the hidden/invisible/toggle_no_view/no_view flags unset.

• All text annotations must have the no_zoom/no_rotate flags set.

• All annotations need an appearance dictionary, except zero-width/zero-height ones or popups.

• The Launch, Sound, Movie, ResetForm, ImportData, Hide, SetOCGState, Rendition, Trans, GoTo3DView and JavaScript actions may not be used.

• Fields, widgets, pages, and the catalog may not use the /AA entry.

• The /Name entry must be set for all optional content configuration dictionaries. HexaPDF sets this entry for the default optional content configuration dictionary.

• All optional content groups must be in the /Order array of the optional content configuration dictionary if it exists.

• The /AS entry of an optional content configuration dictionary may not be used

• The /AlternatePresentations in a document’s name dictionary may not be sued.

• The /PresSteps entry in any page dictionary may not be used.

• The /Requirements entry on the document’s catalog may not be used.

To ensure that a document created with HexaPDF is PDF/A compliant, a PDF/A validation tool like veraPDF can be used.

An PDF/A file created by HexaPDF that follows these rules is available as an example.

Best Practices

• Call the PDF/A task as early as possible to avoid any problems:

  HexaPDF::Composer.create('sample.pdf') do |composer|
    composer.task(:pdfa)
    ...
  end
  

• Use the ‘font.map’ configuration option to define substitute fonts for the standard 14 PDF fonts in case some code explicitly refers to them:

  composer.document.config['font.map'] = {
    'Times' => {
      none: 'Inter-Regular.ttf',
      bold: 'Inter-Bold.ttf',
      italic: 'Inter-Italic.ttf',
      bold_italic: 'Inter-BoldItalic.ttf',
    },
    'Helvetica' => {
      none: 'OpenSans-Regular.ttf',
      bold: 'OpenSans-Bold.ttf',
      italic: 'OpenSans-Italic.ttf',
      bold_italic: 'OpenSans-BoldItalic.ttf',
    },
    'ZapfDingbats' => {
      none: 'ZapfDingbats Regular.ttf'
    }
  }