Re: replace deprecated function ?

[B. T. Raven]

On 2/13/2018 15:14, Kaushal Modi wrote:
> On Tue, Feb 13, 2018 at 1:21 PM B. T. Raven <btraven@nihilo.net> wrote:
>
> > But I notice that query-replace-regexp-eval is no longer considered
> > kosher. How should this be rewritten for Emacs ver. 25? Especially I
> > would like a function that is more immediately understandable than what
> > I have now.
>
> Hello,
>
> First of all, I know that you asked for a replacement function for emacs
> 25.x, but I have an alternative that needs emacs 26.x (because `ucs-names'
> returns a hash instead of an alist in emacs 26).
>
> It was fun to come up with this elisp function. Emacs 26 is nearing
> release, so it might be a good time to try out its pretest version[1].
>
> If you are still interested.. here's the function:

Thanks, Kaushal. That is impressive but too advanced for me. I will save 
it in order to understand unicode better. It looks like the defun takes 
a huge data gulp at (let ((char-hash (ucs-names))). I know that hash 
tables are somehow related to crc's, sha256, signatures etc but I don't 
know how to use them. I will probably have to wait for a w32 binary of 
26.1 before I can try your defun.
btw, I don't want to operate on an entire region here because there are 
too many numbers scattered about the texts that should not be 
superscripted. It's more efficient for me to inspect them in context one 
by one. I would need a rectangle region or maybe even a gerrymandered 
one. ;-)

> (defun num-to-supnum (beg end)
>    "Replace digits with superscript digits in the selected region."
>    (interactive "r")
>    (save-restriction
>      (save-excursion
>        (narrow-to-region beg end)
>        (goto-char (point-min))
>        (let ((char-hash (ucs-names)))
>          (while (re-search-forward "[0-9]" nil :noerror)
>            (let* ((matched-char-name (get-char-code-property (string-to-char
> (match-string-no-properties 0)) 'name))
>                   (new-name (replace-regexp-in-string "DIGIT" "SUPERSCRIPT"
> matched-char-name nil :literal))
>                   (str (char-to-string (gethash new-name char-hash))))
>              (replace-match str)))))))
>
> It looks wordy, but simply does this:
>
> - First narrows the buffer to your selected region, so that replacements
> don't happen outside that.
> - Replacements happen only for characters 0 through 9 (if present).
> - Let's say "0" is found.. it then finds the 'name property for that char..
> which is "DIGIT ZERO".
> - Interestingly, the 'name property for  ⁰ (superscript zero) is..
> "SUPERSCRIPT ZERO". So I simply derive the name of the to-be-replaced-with
> string by replacing DIGIT with SUPERSCRIPT.
> - Then using "SUPERSCRIPT ZERO" as the key in the hash table, I find the
> character code for superscript zero, and replace "0" with "⁰".

All this could eventually be useful for stripping diacritics in order to 
search on the base glyphs.

> So, if I have:
>
>      123 abc 456 def 78 90
>
> and select just the "56 def 78" portion, and do M-x num-to-supnum, I end up
> with:
>
>      123 abc 4⁵⁶ def ⁷⁸ 90
>
> [1]: http://lists.gnu.org/r/emacs-devel/2018-01/msg00641.html

I've downloaded this and will try to build it on my Linux laptop but I 
don't know that OS well enough to get much done in Emacs on it. Are 
general Linux questions pertinent to getting the GUI and Emacs 
customized allowable in this ng?

Ed

>