How do I match comments where the start and the end character are the same?

Question

I have a programming language I want to make a major mode for where both the start and the end characters of a comment are the same. Specifically #.

#
This is a comment
#

As you can see it can match either multiple lines as well as single lines. Single lines are no problem, that can be done with this regex in a font-lock-defaults block: "#\\([^#]*?\\)#.

However the problem occurs on multiple line comments. Due to the start and the end being the same Emacs sometimes misjudges which part of the code is actually a comment, especially if one part of the comment is off-screen.

#
This is a comment and should always be recognized
#
This isn't a comment but is sometimes recognized if the top is off screen
#
This is a comment and is always recognized but sometimes isn't if the last block is wrongly detected as a comment
#

I've also tried this code set as the syntax-propertize-function

(defconst amanatsu-comment-regex
 "\\(#\\([^#]*?\\)#\\)" )
(defun amanatsu-syntax-propertize-function (start end)
  (save-excursion
    (goto-char start)
    (while (re-search-forward amanatsu-comment-regex end 'noerror)
      (let ((a (match-beginning 1))
            (b (match-end 1))
            (comment-fence (string-to-syntax "!")))
        (put-text-property a (1+ a) 'syntax-table comment-fence)
        (put-text-property (1- b) b 'syntax-table comment-fence)))))

but this occasionally suffers the same off-screen errors.

Answer 1

font-lock typically highlights chunks of your buffer at a time. Each line is a couple of lines long.

You can solve this in (at least) two ways:

1. Expand the region that font-lock highlights by defining your own font-lock-extend-region-functions.

2. Ensure that your syntax propertize function works even if start and end happens to be in the middle of a comment block.

To visually see how font-lock refontifies a buffer, you can use the package highlight-refontification.

An example of a package that use font-lock-extend-region-functions is highlight-doxygen.