bug#20690: 25.0.50; query-replace: incorrect history when replacing the NUL character.

[Juri Linkov <juri@linkov.net>]

> ;; We do a query-replace to remove NUL characters and replace them by
> ;; an empty string:
>  M-% [query-replace]
>  C-q [quoted-insert]
>  0 <return> <return> [exit-minibuffer]
>  <return> [exit-minibuffer]
> ;; now we try to do it again, by relying on the history
>  M-% [query-replace]
>  <up> [previous-line-or-history-element]
>  <return> [exit-minibuffer]
> ;; At this point, emacs asks what to replace "two NUL chars" with,
> ;; instead of removing NULs like the previous query-replace did.
>
> I suggest the following changes :

Thank you for the fix!  I noticed that besides fixing the NUL character problem
your patch also changes the behaviour when there are no NUL characters, i.e.:

> +(defun query-replace--split-string (string)
> +  "Split string STRING at a character with property `separator'"
> +  (let* ((length (length string))
> +         (split-pos (text-property-any 0 length 'separator t string)))
> +    (if (not split-pos)
> +        string
           ======
This used to be (substring-no-properties string)

> -	(let* ((to (if (and (string-match separator from)
> -			    (get-text-property (match-beginning 0) 'separator from))
> -		       (substring-no-properties from (match-end 0))))
> -	       (from (if to (substring-no-properties from 0 (match-beginning 0))
> -		       (substring-no-properties from))))
                        =======================
Like you can see here above that stripes properties from the from string.