Wrong highlighting when multi-line string is closed just before an escaped quote

[ronisbr]

Hi!

The highlighting engine does not understand that a multi-line comment section (""") was ended if it ends just after an escaped quote:

If I add a space between \" and """, then everything works.

[non-Jedi]

This is the same problem as #15 which I tried to solve with #114. Unfortunately, that PR has strange transient behavior. Fontifying a buffer from scratch works fine but if you edit a triple-quoted string all text after the triple-quoted string is fontified a string until you reopen the buffer. Any help on fixing it would be appreciated. :)

[ronisbr]

Hi @non-Jedi

Unfortunately, I have close to zero experience in elisp to help you :( However, python has a similar syntax. They solved it by defining the following expression for the delimiters:

      (string-delimiter . ,(rx (and
                                ;; Match even number of backslashes.
                                (or (not (any ?\\ ?\' ?\")) point
                                    ;; Quotes might be preceded by a escaped quote.
                                    (and (or (not (any ?\\)) point) ?\\
                                         (* ?\\ ?\\) (any ?\' ?\")))
                                (* ?\\ ?\\)
                                ;; Match single or triple quotes of any kind.
                                (group (or  "\"" "\"\"\"" "'" "'''"))))))

[ronisbr]

Actually, using this delimiter instead of the group in Julia mode seems to work:

(defconst julia-syntax-propertize-function
  (syntax-propertize-rules
   ;; triple-quoted strings are a single string rather than 3
   ((rx 
     (and
      ;; Match even number of backslashes.
      (or (not (any ?\\ ?\' ?\")) point
       ;; Quotes might be preceded by a escaped quote.
       (and (or (not (any ?\\)) point) ?\\
	(* ?\\ ?\\) (any ?\' ?\")))
      (* ?\\ ?\\)
      ;; Match single or triple quotes of any kind.
      (group (or  "\"" "\"\"\"" "'")))
     (and
      ;; Match even number of backslashes.
      (or (not (any ?\\ ?\' ?\")) point
       ;; Quotes might be preceded by a escaped quote.
       (and (or (not (any ?\\)) point) ?\\
	(* ?\\ ?\\) (any ?\' ?\")))
      (* ?\\ ?\\)
      ;; Match single or triple quotes of any kind.
      (group (or  "\"" "\"\"\"" "'")))
     (and
      ;; Match even number of backslashes.
      (or (not (any ?\\ ?\' ?\")) point
       ;; Quotes might be preceded by a escaped quote.
       (and (or (not (any ?\\)) point) ?\\
	(* ?\\ ?\\) (any ?\' ?\")))
      (* ?\\ ?\\)
      ;; Match single or triple quotes of any kind.
      (group (or  "\"" "\"\"\"" "'"))))
	;; First " starts a string if not already inside a string (or comment)
    (1 (let ((ppss (save-excursion (syntax-ppss (match-beginning 0)))))
         (unless (or (nth 3 ppss) (nth 4 ppss))
           (string-to-syntax "|"))))
    ;; Last " ends a string if already inside a string
    (2 (and (nth 3 (save-excursion (syntax-ppss (match-beginning 0))))
            (string-to-syntax "|"))))
   ;; backslash acts as an operator if it's not inside a string
   ("\\\\"
    (0 (unless (nth 3 (save-excursion (syntax-ppss (match-beginning 0))))
         (string-to-syntax "."))))
   (julia-char-regex
    (1 "\"")                    ; Treat ' as a string delimiter.
    (2 ".")                     ; Don't highlight anything between.
    (3 "\"")))) ; Treat the last " in """ as a string delimiter.

However, I will not dare to submit a PR because I know you guys can write a much more cleaner code than I did :D Up to now, I have found no regressions so far and this does not have the transient behavior mentioned by @non-Jedi .

EDIT: Julia does not use ''' as in Python.

[ronisbr]

There is a problem with my regex. It does fix \"""". However, in Julia, the following string is valid:

"""
""
"""

Thus, we must have something that, if the string is started by """, then it must end by """. Question: should I submit a PR to fix the reported problem? Notice that this behavior also exists in current version. So there is not a regression.

Nerver mind, there will be a regression.