But this does not work this way -- it seems it just blindly consumes char before and after / and converts if this fraction has a glyph in the font. Also, what y

But this does not work this way -- it seems it just blindly consumes char before and after / and converts if this fraction has a glyph in the font. Also, what you suggest is hard to implement since there are numerous localisation-dependent options what a whitespace/punctuation is and where to search it.