Idiomatic handling of region boundaries while modifying its content

Question

I want to write a function to indent SQL embedded in a markdown file.

It seems to make sense to use the region for this.

However, the looping form I am using is:

(defun rr/sql-indent-region (beginning end)
  (interactive "r")
  (while (< (point) end)
     ...
     (forward-word 1)

The ... is adding \n characters after SELECT, FROM, GROUP BY etc, so that the block I started with no longer ends at END.

As I write this, I guess one option would be to copy the region to a temporary buffer, and use (< (point) (point-max)). Is that the best way to proceed?

Answer 1

There are several common approaches you could take. I don't think there's any single "best" option in general.

Using a temporary buffer is entirely reasonable. Buffers are cheap to create. insert-buffer-substring is useful here.

You can also just use narrowing in the current buffer, in which case (point-max) will be the end of the narrowed region, just like in the temporary buffer. C-hig (elisp)Narrowing. Note also eobp.

Finally you could use a marker for the end position, which will dynamically adjust itself when the text before the marker is modified. C-hig (elisp)Markers

In your case I would suggest using narrowing, but experimenting with all three options could be useful.

https://github.com/alex-hhh/emacs-sql-indent may be of interest to you as well.