When code highlight is applied for rmarkdown::html_document output, screen-reader users experience unneccessary difficulties in reading code blocks.

I am not going to duplicate the issue here so please kindly visit pandoc/issue#6352 that I have detailed there.

Fortunately, the fix will be implemented in the next pandoc version; however, I wanted to create backword solution towards previous versions <= v2.9.2.1.

To this end, I have attempted to write a simple lua filter, but that does not seem to be possible unless we consider postprocessing an html output.

Many blind people, including me, have already experienced this challenge caused by redundant codeblock ID information, and I think we need to come up with some universal solutions for Rmd-to-html output process given that many tutorials and online books have been produced by rmarkdown/pandoc under the hood.

Does anyone have a great solution?