Hi Arash,
>>>>> Arash Esbati <[email protected]> writes:
> AUCTeX relies for filling on the function `fill-delete-newlines' (see
> `LaTeX-fill-region-as-para-do'). So this is what we get from Emacs.
> Try this snippet in a .tex file:
> % - Start
> } Test
> % - Start
> } Test
> % - Start
> } Test
> Now eval this function and run it in your .tex file:
> (defun fill-delete-newlines-test ()
> (let (s e)
> (save-excursion
> (goto-char (point-min))
> (while (search-forward "% - Start" nil t)
> (forward-line)
> (setq s (point))
> (setq e (line-end-position))
> (fill-delete-newlines s e nil nil nil)))))
> I see:
> % - Start
> }Test
> % - Start
> } Test
> % - Start
> }Test
Thanks for your analysis. I reached `canonically-space-region' tracking
down the issue.
If I change its last part as follows (emacs 30.2), one space is retained
between "}" and "Test" for me.
----------------------------------------------------------------------
(defun canonically-space-region (beg end)
"Remove extra spaces between words in region.
Leave one space between words, two at end of sentences or after colons
\(depending on values of `sentence-end-double-space', `colon-double-space',
and `sentence-end-without-period').
Remove indentation from each line."
(interactive "*r")
;; Ideally, we'd want to scan the text from the end, so that changes to
;; text don't affect the boundary, but the regexp we match against does
;; not match as eagerly when matching backward, so we instead use
;; a marker.
(unless (markerp end) (setq end (copy-marker end t)))
(let ((end-spc-re (concat "\\(" (sentence-end) "\\) *\\| +")))
(save-excursion
(goto-char beg)
;; Nuke tabs; they get screwed up in a fill.
;; This is quick, but loses when a tab follows the end of a sentence.
;; Actually, it is difficult to tell that from "Mr.\tSmith".
;; Blame the typist.
(subst-char-in-region beg end ?\t ?\s)
(while (and (< (point) end)
(re-search-forward end-spc-re end t))
(delete-region
(cond
;; `sentence-end' matched and did not match all spaces.
;; I.e. it only matched the number of spaces it needs: drop the rest.
((and (match-end 1) (> (match-end 0) (match-end 1))) (match-end 1))
;; `sentence-end' matched but with nothing left. Either that means
;; nothing should be removed, or it means it's the "old-style"
;; sentence-end which matches all it can. Keep only 2 spaces.
;; We probably don't even need to check `sentence-end-double-space'.
((match-end 1)
(min (match-end 0)
(+ (if sentence-end-double-space 2 1)
(save-excursion (goto-char (match-end 0))
(skip-chars-backward " ")
(point)))))
(t ;; It's not an end of sentence.
(+ (match-beginning 0)
;; Determine number of spaces to leave:
(save-excursion
(skip-chars-backward " ]})\"'")
(cond ((and sentence-end-double-space
(or (memq (preceding-char) '(?. ?? ?!))
(and sentence-end-without-period
(= (char-syntax (preceding-char)) ?w)))) 2)
((and colon-double-space
(= (preceding-char) ?:)) 2)
; ((char-equal (preceding-char) ?\n) 0)
((char-equal (preceding-char) ?\n) 1) ; CHANGED
(t 1))))))
(match-end 0))))))
----------------------------------------------------------------------
I haven't understood the former part of its `end-spc-re', namely
(concat "\\(" (sentence-end) "\\) *"), but in myopinion, it isn't
appropriate for the rest part, namely " +", to delete the entire
matched string; we have to leave one space there.
Ikumi Keita
#StandWithUkraine #StopWarInUkraine
#Gaza #StopGenocide #CeasefireNOW
#IProtestAgainstTrumpAndNetanyahu'sAttackOnIran,WhichViolatesInternationalLaw
_______________________________________________
bug-auctex mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/bug-auctex
