Create Accessible PDF files with iText

Question

Recently I downloaded a trial license of iText. I try to achieve the following goals:

Create PDF documents from WORD, PP, Excel etc.
Manipulating existing PDF files.
(Most Important) - Files MUST be created / converted as full accessible for disables according to the WCAG 502 requirements (PDF/UA)

I tried the following code: (C#)

    LicenseKey.LoadLicenseFile(@"D:\Development\itextkey-0.xml");
    PdfDocument pdfDoc = new PdfDocument(new PdfReader(SRC), new PdfWriter(DEST, new WriterProperties().SetPdfVersion(PdfVersion.PDF_1_7)));
    pdfDoc.SetTagged();
    pdfDoc.GetCatalog().SetLang(new PdfString("HE-IL"));
    pdfDoc.GetCatalog().SetViewerPreferences(
            new PdfViewerPreferences().SetDisplayDocTitle(true));
    PdfDocumentInfo info = pdfDoc.GetDocumentInfo();
    info.SetTitle("iText7 PDF/UA example");
    pdfDoc.Close();

But yet, after checking at Acrobat Reader the output file marked as "Not Tagged" PDF file.

Please advise how I should use iText to achieve my goals.

Tagging requires human intelligence. You can't expect software (in general) to correctly tag a document without human interaction. I don't see you introducing the tags anywhere in your code. I don't see you providing any font programs to substitute fonts that weren't embedded. It's only normal that your code sample doesn't create a properly tagged PDF file. Tell whoever gave you this assignment that the requirement can't be met, and respectfully decline the opportunity. If you find software that claims it can be done "out of the box": don't trust it! — Bruno Lowagie, Nov 06 '17 at 13:43
Also, please don't expect new PdfReader(SRC) to work if the SRC path resolves into a word document. I mean, I would love the feature. But it's not going to happen. — Joris Schellekens, Nov 06 '17 at 14:07

score 2 · Accepted Answer · edited Nov 06 '17 at 14:10

Can't be done.

Let me give you the easiest proof:
Suppose the input document contains an image of two cats fighting over a ball of yarn.

pdf/UA requires you to insert sensible alternative text for your imagines.
There is currently no system available that is able to provide a sensible caption for any random image you throw at it.

Not to mention that whatever system comes up with a caption for images, would have to linked to a perfect translation service. Since most image recognition services are in English, and this might not be the language you are writing documents in. Which also implies you need a system that is capable of detecting the language you are writing in.

We've now added 3 insanely hard problems, simply to be able to handle images:

tagging
translation
language detection

Now imagine the other kind of fun stuff, like

tables
graphs and charts
..

Furthermore, PDF/UA requires fonts to be embedded. What if you are faced with a PDF that uses fonts that aren't embedded. Do you have access to font programs that can be used to substitute those fonts?

In your snippet, you use PdfReader, and you provide a path to a file SRC. You need to convert Word, PPT, and other files, but iText doesn't convert Word, PPT, etc to PDF. PdfReader only accepts PDF files (as the name indicates).

`There is currently no system available that is able to provide a sensible caption for any random image you throw at it.` - In comes machine learning. In case future readers are interested - https://github.com/tensorflow/models/tree/master/research/im2txt — Pramesh Bajracharya, Apr 24 '19 at 10:02

Create Accessible PDF files with iText

1 Answers1