I wish to include some non-ascii characters in the documentation for an R package, specifically the diacritic ă.
I am using roxygen2, so must not include any non-ASCII characters in the .R file from which the documentation will be generated. Moreover, the non-ASCII characters may not be available to the software that compiles a PDF manual, so I wish to specify a safe fallback for the PDF documentation.
The obvious approach seems to be to conditionally display HTML entities, i.e. \ifelse{html}{ă}{a}, but the entities are evidently escaped at some point: the .rd file reads ă where I expect an ă.
Bonus points for answers that can be applied to diacritics included from REFERENCES.bib via the \insertRef macro...
Per Stéphane Laurent's comment, the simple case can be solved by wrapping the desired HTML entities in \out{}, i.e.
\ifelse{html}{\out{ă}}{a}
For bonus points, it's possible to provide the appropriate PDF character with an ASCII fallback to be used if this is not available with
\ifelse{html}{\out{ă}}{\v{a}}{a}
It would be nice to think that when diacritics occur in references, it would be possible to create two bibtex entries, one using only ascii characters:
\insertCite{\ifelse{html}{RefWithAccents}{RefAscii}}
But this triggers a server error...