bug#54804: 29.0.50; zap-to-char: case sensitive for upper-case letter

[Eli Zaretskii]

> From: Tino Calancha <tino.calancha@gmail.com>
> Cc: 54804@debbugs.gnu.org,  uyennhi.qm@gmail.com
> Date: Sat, 16 Apr 2022 12:58:06 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > but why does the implementation have to be so complicated?
> > Isn't this just about turning off case-fold-search while searching for
> > the character?  What am I missing?
> 
> The patch is moving the shared code in zap-to/zap-up-to in a
> helper function.  This has 2 motivations:
> - reduce the code duplication.
> - define these two related functions close each other.
> - separate the logic of the function to calculate the region to kill.
> 
> If you prefer, we can keep the functions in their current
> locations (simple.el/misc.el) with a patch like this one:

What I'd prefer is to have a single function (in subr.el) that
determines whether a character is an upper- or lower-case, and then
use that function in a simple condition in these two commands.

The function to check whether a character is upper-case doesn't have
to make a string from the character and then use the "heavy artillery"
of string-match-p, it could instead use something like

   (characterp (get-char-code-property CHAR 'uppercase))

(But beware of the situation where the Unicode tables are not yet
available during bootstrap -- in those cases the function should IMO
punt and return nil no matter what the character is, or maybe support
just the ASCII characters.  To test whether the 'uppercase table is
available, see if unicode-property-table-internal returns non-nil.)

> +  ;; Avoid "obsolete" warnings for translation-table-for-input.
> +  (with-no-warnings
> +    (if (char-table-p translation-table-for-input)
> +     (setq char (or (aref translation-table-for-input char) char))))

translation-table-for-input is obsolete for a reason, so adding it to
a new function is not something we want to do.  If anything, we should
remove it from the old function.