Bug Tracker

The report category will depend on whether the issue is stemming from glossaries-accsupp (in which case, I'll change the report category to glossaries) or if it's coming from a modification made by glossaries-extra.

Would you be interested in testing the pending new version of glossaries-extra? If so, you can download the experimental 1.48b: glossaries-extra-1_48b.zip It's a working copy so the documented code is incorporated into the sty files. (Release versions have the comments stripped.) Some of the issues you have logged may now be fixed.

Hi Nicola,

Sure! I'll be happy to do some testing.
True, my range cannot be but somewhat limited, since I'm new to the package and am far from knowing my way around it.
But I'll certainly take a look in the cases I have reported, and also in the features I'm using in the project I'm working on.
I'll report back when I have done so (hopefully tomorrow, I had intended to work on an AUCTeX style file for glossaries, but I can rearrange that ;-).

Best regards,
Gustavo.

Hi Nicola,

reporting on v1.48, as requested.

Regarding this particular issue, there is some change of behavior I could not track. The original MWE of the report now copy pastes the following content:

Aaaa Bbbb Cccc (abc), abc, abc
Doctor, \376\377\000D\000o\000c\000t\000o\000r (\376\377\000D\000r\000.), \376\377\000D\000o\000c\000t\000o\000r (\376\377\000D\000r\000.)

Besides the issue specific comments, which I've made above and elsewhere, in general, I put the new version to test in my current working document. Flawless. I diffed the PDFs generated with both versions, and they are equal. For you to have an idea what to infer from this, it is a book-length document (~500 pages), and I'm using basically the abbreviations functionality from glossaries-extra. It is a subset from it, of course, but a decent one: different types of abbreviations, categories for formatting, including some foreign entries, period discarding, etc. They are not many entries (~10), but a lot of references (>1000).

Furthermore, I took a general look, of course, and can see a lot of new interesting things coming. If I understood the purpose of \GlsXtrIfInGlossary well, I'm already have some ideas for it. Also the manual has received a lot of care, which is great. All in all, I'm definitely looking forward to it. Thank you very, very much!! :-)

Thank you very much for your feedback. It's been very helpful.

The sequence \376\377\000D\000o\000c\000t\000o\000r translates to the characters FE FF 00 D 00 o 00 c 00 t 00 o 00 r. The first two (FE FF) are the byte order mark. This looks like the UTF-16 string for "Doctor".

The reason the long form is being used is because the default value for the short access is obtained from \glsdefaultshortaccess{long}{short}. This command is defined (in glossaries-accsup.sty) as:

\newcommand*{\glsdefaultshortaccess}[2]{#1}

If you want the short form instead, you need to redefine this:

\renewcommand*{\glsdefaultshortaccess}[2]{#2}

\renewcommand*{\glsdefaultshortaccess}[2]{}

The following uses \show to show the accessibility value:

\documentclass{article}

\usepackage{hyperref}
\usepackage[accsupp]{glossaries-extra}

\renewcommand*{\glsshortaccsupp}[2]{\def\tmp{#1}\show\tmp
 \glsaccessibility{ActualText}{#1}{#2}}

\newacronym{dr}{Dr}{Doctor}

\begin{document}
\gls{dr}, \gls{dr}.
\end{document}

> \tmp=macro:
->\376\377\000D\000o\000c\000t\000o\000r (\376\377\000D\000r).

\documentclass{article}

\usepackage{hyperref}
\usepackage{glossaries-accsupp}

\renewcommand*{\glsshortaccsupp}[2]{\def\tmp{#1}\show\tmp
 \glsaccessibility{ActualText}{#1}{#2}}

\newacronym{dr}{Dr}{Doctor}

\begin{document}
\gls{dr}, \gls{dr}.
\end{document}

> \tmp=macro:
->Doctor.

I've just noticed that glossaries-extra redefines \glsdefaultshortaccess to "long (short)". I think it would be better to revert this back to the original default in glossaries-accsupp. This may cause some backward compatibility issues, but it's more appropriate.

The default values for the accessibility fields are assigned from the original short and long values provided to \newabbreviation or \newacronym (not the values obtained after processing attributes such as insertdots). The assignment is performed using \pdfstringdef, if it has been defined, or with just \protected@edef otherwise.

It's \pdfstringdef that's converting the string to UTF-16BE. I've checked with the hyperref manual and that's the default setting, so the copy+paste problem may stem from your PDF viewer not supporting UTF-16BE. You can change the encoding, for example:

\hypersetup{pdfencoding=pdfdoc}

If you want to strip formatting commands, you can append a local change to \glsxtrassignactualsetup. For example:

\appto\glsxtrassignactualsetup{\letcs{\glstextup}{@firstofone}}

It's \pdfstringdef that's converting the string to UTF-16BE. I've checked with the hyperref manual and that's the default setting, so the copy+paste problem may stem from your PDF viewer not supporting UTF-16BE. You can change the encoding, for example:

\hypersetup{pdfencoding=pdfdoc}

but this has limited support.

And, indeed, you are right that without hardcoding \pdfstringdef into \@gls@assign@actual (as is done in glossaries-extra) we get correct results:

\documentclass{article}

\usepackage{hyperref}

\usepackage[
  accsupp,
  abbreviations,
  shortcuts=abbr,
  debug=showaccsupp,
]{glossaries-extra}

\makeglossaries

\renewcommand*{\glsshortaccsupp}[2]{\glsaccessibility[method=pdfstringdef]{ActualText}{#1}{#2}}

\makeatletter
\renewcommand{\@gls@assign@actual}{%
  \begingroup
  \glsxtrassignactualsetup
  \protected@edef\@gls@tmp{\endgroup
    \def\noexpand\@gls@actualshort{\glsxtrorgshort}%
    \def\noexpand\@gls@actuallong{\glsxtrorglong}%
    \def\noexpand\@gls@actualshortpl{\@gls@shortpl}%
    \def\noexpand\@gls@actuallongpl{\@gls@longpl}%
  }%
  \@gls@tmp
}
\makeatletter

\setabbreviationstyle[initialism]{long-short-sc}
\setabbreviationstyle{long-only-short-only}

\renewcommand*{\glsdefaultshortaccess}[2]{#2}

\pdfstringdefDisableCommands{\renewcommand*{\glstextup}{}}

\newabbreviation[
  category=initialism,
  shortaccess={ABC},
]{ABC}{abc}{Aaaa Bbbb Cccc}

\newabbreviation{Dr}{Dr.}{Doctor}

\begin{document}

\ab{ABC}, \ab{ABC}, \ab{ABC}

\ab{Dr}, \ab{Dr}, \ab{Dr}

\end{document}

Aaaa Bbbb Cccc (ABC), ABC, ABC
Doctor, Dr., Dr.

Still there is some mystery there, since I cannot tell why the difference between ABC and Dr

Nicola,

might I be granted a question?

You've called my attention to \glsdefaultshortaccess above which I had missed in my attempts to use glossaries-accsupp. I eventually gave up on using it for my current project because (for my use case) I felt it inappropriate that entries which did not have a shortaccess value were wrapped with ActualText, since the intended actual text was the same as the text itself, and this adds an unnecessary complexity. So I feel that I might have missed some handle which is already there to control this, just like I missed \glsdefaultshortaccess.

The question is, is there a way to use the accessibility support only for some entries, particularly only those which have an explicit shortaccess field set? In terms of the MWE, how to wrap only ABC, but not Dr?

(This is a question, not a request or report. So, feel to chose from "No", "It's complicated", "Possibly, but it's a homework for you" or whatever else in the same vein as an answer).

Ah, a small correction to a previous comment. method=pdfstringdef is not the default of accsupp. What the documentation says is "If package hyperref is loaded, then its \pdfstringdef is used". (As far as I can tell from the code, the default is escape.)

Still, one alternative would be to condition the definition of \gls@accessibility to the existence of \pdfstringdef, and use method=pdfstringdef in that case, instead of conditioning the definition of \@gls@assign@actual.

Hi Nicola,
thank you very much for the fix!
Best regards,
Gustavo.

Hi Gustavo,
thank you for your feedback. I'm sorry it took longer than anticipated as I had to get glossaries v4.50 and mfirstuc v2.08 out of the way first. Allow a few days for all the updates to make it into the TeX distributions.
Best regards
Nicola

Hi Nicola,
oh, please, don't be. Whatever your timing is, it is appreciated.
Just as v4.50 which came in today is very much so. :-)
Thank you!
Best regards,
Gustavo.