1

As the title states, I am looking for a way to disable the use of special unicode symbols for mathematical letters (Mathematical Alphanumeric Symbols) in the PDF-output that I get when using the unicode-math package.

I use the unicode-math package for the following reasons:

  • Being able to use unicode-letters in the source text.
  • Being able to copy text and math from the generated PDF as text with all unicode symbols preserved.

The only part of unicode-math that I want to turn off is that alphanumeric letters like x and Γ generate different unicode-letters depending on the “font style” (symnormal, symbfit, symbb, etc). Some of the problems I have with this are:

  • Some screen readers read all math letters such as $A$ as "MATHEMATICAL ITALIC CAPITAL A", making the PDF less accessible for some.
  • It is difficult to search for mathematical symbols in the PDF.
  • It is harder to copy and paste mathematics as "pseudo-plaintext".

So really all I want is to have the same visual output, but map alphanumeric characters back to their ASCII/non-mathematical variants in the plain text stored in the PDF. I have some ideas, but I'm not knowledgeable enough about *TeX and unicode to actually know what is possible:

  • Are there some parameters that can be passed to the unicode-math package?
  • Is it possible to generate a cmap (similar to the cmap and mmap packages)?
  • Is there some other PDF feature that can be used (e.g. additional PDF metadata)?
Steven B. Segletes
  • 237,551
MyComputer
  • 161
  • Some of my prior answers seem related to this question of accessibility: https://tex.stackexchange.com/questions/233390/in-which-way-have-fake-spaces-made-it-to-actual-use/233397#233397, https://tex.stackexchange.com/questions/454944/is-there-screen-reader-software-or-a-built-in-method-that-supports-latex-equat/505462#505462, and https://tex.stackexchange.com/questions/352649/can-we-get-maths-in-pdfs-that-can-be-copy-pasted-and-screen-read/407019#407019 – Steven B. Segletes Apr 13 '21 at 10:11
  • well you can change the Tounicode value, see e.g. https://tex.stackexchange.com/a/420568/2388. But \symnormal, \symit are not styles, they are different symbols with different unicode code points. If you map them to the same unicode value screen reader will no longer "see" them as different. – Ulrike Fischer Apr 13 '21 at 10:16
  • Thanks for your replies! I had to stop using unicode-math, because of another problem, but I will try to take a look once I figure that out. – MyComputer Apr 15 '21 at 09:22
  • One option would be to wrap the letters in \textnormal, for example, \textnormal{\itshape x}. You’d of course want to write a shorter macro. – Davislor May 02 '21 at 15:26
  • Does this answer your question? Can we get maths in PDFs that can be copy-pasted and screen-read? – Elad Den May 03 '21 at 06:21

1 Answers1

4

This may not solve your problem, but I can offer some background on the Unicode assignment.

While recognizing that some people may wish to create texts using script or fraktur styles, the Unicode committee specifically stated that alternate fonts should be used for that purpose. This is not possible for such letters used in math, as illustrated in Unicode Technical Report #25, Unicode support for mathematics, in the section on "Semantic distinctions" (pp.6-7). (There is also a precedent in the content of the "letterlike symbols" block (2100-214F).)

The unicode-math package follows the Unicode directive. Equating either the symbols in the package or those in the Unicode "mathematical alphanumeric symbols" area is therefore considered, at the very least, a bad idea.

barbara beeton
  • 88,848