How can I make syntax-propertize skip part of the buffer?

Question

In SPARQL, comments starts with a # and follows to the end of the line. The problem is that URI's, which is a big part of SPARQL, can also contain #, but then they are not the start of a comment. Like so <http://example.com/foo#bar>. URI's are always surrounded by <> and never have any spaces in them.

Normally, we would write a syntax-table like this

(defvar sparql-mode-syntax-table
  (modify-syntax-entry ?# "<" syntax-table)
  (modify-syntax-entry ?\n ">" syntax-table))

but this would include the # of a URI. Now when the buffer is font-locked, or we try to move using the defined syntax, from the part from the URI that contains a # is now considered a comment.

How can I make syntax-propertize skip marking # in URI's as beginning of comments?

Answer 1

The way to make syntax-propertize skip part of a buffer is to use syntax-propertize-function.

(defvar syntax-mode-syntax-table
  (let ((syntax-table (make-syntax-table)))
    (modify-syntax-entry ?\n ">")
    syntax-table))

(defvar sparql-syntax-propertize-function
  (syntax-propertize-rules
    ("<\\S-*>" (0 "@"))
    ("#" (0 "<"))))

Since syntax-propertize-rules only propertizes a part of the buffer once and it matches from top to bottom, this will set the syntax-property for a URI (<\\S-*>, a string without space within <>) to inherit and therefore skip any # inside the URI. If it is outside it matches the second rule and is marked as a comment. The syntax table takes care of the comment ending.