[accessibility] Current packages and methods of generating tagged PDF from LaTeX

Jason White jason at jasonjgw.net
Sun Jun 30 22:38:09 CEST 2019

As a naïve user of the tagpdf package, here are some initial comments.

If I use the "uncompress" option to \tagpdfsetup, with current TeX Live 2019 (on a Mac OS machine, though that probably doesn't matter), I get the following from lualtex:

! Undefined control sequence.
\key code > uftag/setup/compresslevel/0 ...sable: 

Omitting the uncompress option works around this, enabling PDF generation to succeed (obviously yielding compressed output - presumably in the PDF content streams).

It's easy to tag basic document structures (paragraphs, sections, headings, etc.). I've verified that this works. The short document that I tried to tag uses the acmart class. Predictably, none of the automatically generated text is tagged, including the bibliography. For that, I would need to develop my LaTeX programming skills beyond basic macro definitions, and redefine various macros. As the tagpdf documentation suggests, anything which is not tagged is omitted from what is presented to the screen reader.

The acmart class uses the Libertine fonts. I noticed in the output of Adobe Reader that letters occurring in ligatures were omitted, suggesting that there was an issue with the mapping of those glyphs to the correct Unicode sequences.

Unfortunately, I can't share the document yet, due to copyright considerations, but that situation may change. 

I think the tagpdf package is a very substantial step that demonstrates what can be achieved.

On 6/27/19, 09:33, "Ulrike Fischer" <fischer at troubleshooting-tex.de> wrote:

    > Thank you, Ulrike. I've started reading the documentation of your package.
    > What would be the best installation process to obtain your package and all
    > of its dependencies?
    Simply use the package manager of your texsystem. The main requirement
    is that it should be up-to-date so texlive 2019
    or a current miktex so that you get updates.
    > I can certainly run lualatex. A small sample document would be a useful
    > starting point.
    There are examples in the package. Your package manager will put them
    in the doc folder. The source of the documentation itself is also an
    > For this to become widely used, it seems to me that there will need to be
    > well documented and easily used macros that LaTeX package writers can
    > include in their packages to declare structural elements and delimit page
    > regions.
    Yes. Also the latex kernel will need hooks and changes to allow this.
    But to be able to make sensible suggestions here, more tests are
    needed. That's why we are interested in users trying out things and
    reporting back problem, suggestions, feature requests ...
    Also things are still bound to change. E.g.
    the pdf Association has only just published a practice guide
    and this means that I will have to change the way I tagged the table
    of contents in the documentation ;-(
    > I also wonder whether this could become a more robust pathway to
    > generating HTML as a byproduct.
    I doubt it, perhaps some parts can be reused but pdf has really rather
    special requirements. Beside this tex4ht is imho quite stable, mature and
    Ulrike Fischer

More information about the accessibility mailing list