Hello! Summary:
The variables opening-re and closing-re in org-element--parse-generic-emphasis are hard-coded, and this is not what is expected by the user who has modified the value of org-emphasis-regexp-components. Background and details: One convenient trick to apply emphasis to parts of a word is to separate that part with zero-width non-breaking spaces. This allows us to draw attention to puns like "the ex_cite_ment one feels when one's work is getting cited" by underlining the "cite" part of the word "excitement".[1] Org doesn't support this out of the box but by adding the zero-width non-breaking space to org-emphasis-regexp-components we can get this to work. However! Once we export that Org document, the partial emphasis no longer carries through. This is because ox uses org-element, which defines its own emphasis regex in org-element--parse-generic-emphasis. (The variables are opening-re and closing-re.) Maybe it's reasonable for org-element to do its own thing and not rely on Org mode buffer content, but this particular thing should be easier to tailor. The way the code is written now, we can't even advise around the function to customise which characters are allowed around emphasised text – the only workaround I can think of is to literally copy-and-paste the entire function and add the characters desired to the corresponding regexen. I noticed this breakage at some point in the last few years, but only now was I bothered enough to investigate. It seems like this was a semi-deliberate choice made in late 2021, commit 2bbac43 with title "Remove dependency on ’org-emph-re’". [1]: Sorry about the awful example. My actual usecase is a little contrived and would distract from understanding the bug itself. Emacs : GNU Emacs 29.4 (build 1, x86_64-redhat-linux-gnu, X toolkit, cairo version 1.18.0, Xaw3d scroll bars) of 2024-07-16 Package: Org mode version 9.8-pre (release_9.7.10-127-g07dd3b @ /home/kqr/.emacs.d/straight/build/org/)
