all messages for Emacs-related lists mirrored at yhetil.org
 help / color / mirror / code / Atom feed
* bug#9183: js-mode highlights quotes inside regexp character sets as string
@ 2011-07-27 19:27 Štěpán Němec
  2011-08-05 16:54 ` Stefan Monnier
  0 siblings, 1 reply; 6+ messages in thread
From: Štěpán Němec @ 2011-07-27 19:27 UTC (permalink / raw)
  To: 9183

The following is valid JavaScript:

/[']/.test(str)

js-mode erroneously highlights everything following `'' as a string.
Same problem with `"'.

-- 
Štěpán





^ permalink raw reply	[flat|nested] 6+ messages in thread

* bug#9183: js-mode highlights quotes inside regexp character sets as string
  2011-07-27 19:27 bug#9183: js-mode highlights quotes inside regexp character sets as string Štěpán Němec
@ 2011-08-05 16:54 ` Stefan Monnier
  2011-08-06 11:34   ` Štěpán Němec
  0 siblings, 1 reply; 6+ messages in thread
From: Stefan Monnier @ 2011-08-05 16:54 UTC (permalink / raw)
  To: Štěpán Němec; +Cc: 9183-done

> The following is valid JavaScript:
> /[']/.test(str)

> js-mode erroneously highlights everything following `'' as a string.
> Same problem with `"'.

I installed the patch below which should fix this case and a few others.
But I don't know how to solve it completely because I don't know
Javascript enough to be sure exactly how to distinguish a /-division
from a /-regexp from a /-comment-starter in all cases:
- is something like

    x = 1 + /a/.test("foo");

  valid?  If so, is there a list of infix operators somewhere?

- I'm giving up on handling a /regexp/ that immediately follows an
  implicit semicolon.


        Stefan


=== modified file 'lisp/progmodes/js.el'
--- lisp/progmodes/js.el	2011-06-30 16:22:33 +0000
+++ lisp/progmodes/js.el	2011-08-05 16:52:20 +0000
@@ -1658,15 +1658,19 @@
 ;; below.
 (eval-and-compile
   (defconst js--regexp-literal
-    "[=(,:]\\(?:\\s-\\|\n\\)*\\(/\\)\\(?:\\\\.\\|[^/*\\]\\)\\(?:\\\\.\\|[^/\\]\\)*\\(/\\)"
+    (concat
+     ;; We want to match regular expressions only at the beginning of
+     ;; expressions.
+     ;; FIXME: Should we also allow /regexp/ after infix operators such as +,
+     ;; /, -, *, >, ...?
+     "\\(?:\\`\\|[=([{,:;]\\)\\(?:\\s-\\|\n\\)*"
+     "\\(/\\)\\(?:\\\\.\\|[^/*\\]\\)\\(?:\\\\.\\|[^/\\]\\)*\\(/\\)")
   "Regexp matching a JavaScript regular expression literal.
 Match groups 1 and 2 are the characters forming the beginning and
 end of the literal."))
 
 (defconst js-syntax-propertize-function
   (syntax-propertize-rules
-   ;; We want to match regular expressions only at the beginning of
-   ;; expressions.
    (js--regexp-literal (1 "\"") (2 "\""))))
 
 ;;; Indentation






^ permalink raw reply	[flat|nested] 6+ messages in thread

* bug#9183: js-mode highlights quotes inside regexp character sets as string
  2011-08-05 16:54 ` Stefan Monnier
@ 2011-08-06 11:34   ` Štěpán Němec
  2011-08-12 15:48     ` Stefan Monnier
  0 siblings, 1 reply; 6+ messages in thread
From: Štěpán Němec @ 2011-08-06 11:34 UTC (permalink / raw)
  To: 9183

reopen 9183
thanks

On Fri, 05 Aug 2011 18:54:45 +0200
Stefan Monnier wrote:

>> The following is valid JavaScript:
>> /[']/.test(str)
>
>> js-mode erroneously highlights everything following `'' as a string.
>> Same problem with `"'.
>
> I installed the patch below which should fix this case and a few
> others.

Thanks, but it doesn't seem to fix anything, just mess things up in a
somewhat different way.

E.g., with the following in a JS buffer:

    /[']/.test(str);
    /ahoj/.test(str);
    /[']/.test(str);
    /[']/.test(str);

All of the first /[']/ is highlighted as a string, and everything from
the second single quote up to the third single quote is highlighted as a
string.

(Just in case, I downloaded the fixed js.el and loaded that with my
oldish Emacs 24 (started as emacs -Q), but I don't think that should
matter, right?)

> But I don't know how to solve it completely because I don't know
> Javascript enough to be sure exactly how to distinguish a /-division
> from a /-regexp from a /-comment-starter in all cases:
> - is something like
>
>     x = 1 + /a/.test("foo");
>
>   valid?

Yes, it's valid, although I suspect it would be considered bad style.

(BTW, in Debian there is a spidermonkey-bin package, which has minimal
dependencies and provides a `js' command, giving you a REPL.)

> If so, is there a list of infix operators somewhere?

I'd have to search for it just as you might have... There's this, for
instance:
https://developer.mozilla.org/en/JavaScript/Reference/Operators

-- 
Štěpán





^ permalink raw reply	[flat|nested] 6+ messages in thread

* bug#9183: js-mode highlights quotes inside regexp character sets as string
  2011-08-06 11:34   ` Štěpán Němec
@ 2011-08-12 15:48     ` Stefan Monnier
  2011-08-12 16:47       ` Daniel Colascione
  2011-08-12 17:25       ` Štěpán Němec
  0 siblings, 2 replies; 6+ messages in thread
From: Stefan Monnier @ 2011-08-12 15:48 UTC (permalink / raw)
  To: Štěpán Němec; +Cc: 9183

> Thanks, but it doesn't seem to fix anything, just mess things up in a
> somewhat different way.

It did fix the problem to some extent, but indeed it suffered from
a "multiline" problem (partly inherited from the previous code).

I've installed the patch below which should fix the problem, tho it may
introduce others.  Please try it out.

> (Just in case, I downloaded the fixed js.el and loaded that with my
> oldish Emacs 24 (started as emacs -Q), but I don't think that should
> matter, right?)

That should work OK, yes.  It won't work with Emacs-23 not with an ancient
Emacs-24, but oldish should be fine ;-)

>> But I don't know how to solve it completely because I don't know
>> Javascript enough to be sure exactly how to distinguish a /-division
>> from a /-regexp from a /-comment-starter in all cases:
>> - is something like
>> x = 1 + /a/.test("foo");
>> valid?
> Yes, it's valid, although I suspect it would be considered bad style.

Hmm, that makes it more interesting.
Especially if you ignore semantics and focus on syntax where it would
seem that "x = /foo/ / /bar/;" would then be valid, tho meaningless).

> (BTW, in Debian there is a spidermonkey-bin package, which has minimal
> dependencies and provides a `js' command, giving you a REPL.)

Thanks.

>> If so, is there a list of infix operators somewhere?
> I'd have to search for it just as you might have... There's this, for
> instance:
> https://developer.mozilla.org/en/JavaScript/Reference/Operators

Thanks, I'll keep it for next time I'm bored.


        Stefan


=== modified file 'lisp/ChangeLog'
--- lisp/ChangeLog	2011-08-12 13:49:45 +0000
+++ lisp/ChangeLog	2011-08-12 15:30:59 +0000
@@ -1,5 +1,13 @@
 2011-08-12  Stefan Monnier  <monnier@iro.umontreal.ca>
 
+	* progmodes/js.el (js-syntax-propertize, js-syntax-propertize-regexp):
+	New function.
+	(js--regexp-literal, js-syntax-propertize-function): Remove.
+	(js-mode): Use js-syntax-propertize to handle multilines (bug#9183).
+	(js-mode-map): Don't rebind electric keys.
+	(js-insert-and-indent): Remove.
+	(js-mode): Setup electric-layout and electric-indent instead.
+
 	* epa-file.el (epa-file-select-keys): Revert to nil default (bug#9280).
 
 2011-08-12  Daiki Ueno  <ueno@unixuser.org>

=== modified file 'lisp/progmodes/js.el'
--- lisp/progmodes/js.el	2011-08-05 19:53:46 +0000
+++ lisp/progmodes/js.el	2011-08-12 15:29:28 +0000
@@ -1653,25 +1636,35 @@
                                    js--font-lock-keywords-3)
   "Font lock keywords for `js-mode'.  See `font-lock-keywords'.")
 
-;; XXX: Javascript can continue a regexp literal across lines so long
-;; as the newline is escaped with \. Account for that in the regexp
-;; below.
-(eval-and-compile
-  (defconst js--regexp-literal
-    (concat
-     ;; We want to match regular expressions only at the beginning of
-     ;; expressions.
-     ;; FIXME: Should we also allow /regexp/ after infix operators such as +,
-     ;; /, -, *, >, ...?
-     "\\(?:\\`\\|[=([{,:;]\\)\\(?:\\s-\\|\n\\)*"
-     "\\(/\\)\\(?:\\\\.\\|[^/*\\]\\)\\(?:\\\\.\\|[^/\\]\\)*\\(/\\)")
-  "Regexp matching a JavaScript regular expression literal.
-Match groups 1 and 2 are the characters forming the beginning and
-end of the literal."))
-
-(defconst js-syntax-propertize-function
+(defun js-syntax-propertize-regexp (end)
+  (when (eq (nth 3 (syntax-ppss)) ?/)
+    ;; A /.../ regexp.
+    (when (re-search-forward "\\(?:\\=\\|[^\\]\\)\\(?:\\\\\\\\\\)*/" end 'move)
+      (put-text-property (1- (point)) (point)
+                         'syntax-table (string-to-syntax "\"/")))))
+
+(defun js-syntax-propertize (start end)
+  ;; Javascript allows immediate regular expression objects, written /.../.
+  (goto-char start)
+  (js-syntax-propertize-regexp end)
+  (funcall
   (syntax-propertize-rules
-   (js--regexp-literal (1 "\"") (2 "\""))))
+    ;; Distinguish /-division from /-regexp chars (and from /-comment-starter).
+    ("\\(?:^\\|[=([{,:;]\\)\\(?:[ \t]\\)*\\(/\\)[^/*]"
+     (1 (ignore
+	 (forward-char -1)
+         (when (or (not (memq (char-after (match-beginning 0)) '(?\s ?\t)))
+                   ;; If the / is at the beginning of line, we have to check
+                   ;; the end of the previous text.
+                   (save-excursion
+                     (goto-char (match-beginning 0))
+                     (forward-comment (- (point)))
+                     (memq (char-before)
+                           (eval-when-compile (append "=({[,:;" '(nil))))))
+           (put-text-property (match-beginning 1) (match-end 1)
+                              'syntax-table (string-to-syntax "\"/"))
+           (js-syntax-propertize-regexp end))))))
+   (point) end))
 
 ;;; Indentation
 
@@ -3302,7 +3295,7 @@
   (set (make-local-variable 'font-lock-defaults)
        (list js--font-lock-keywords))
   (set (make-local-variable 'syntax-propertize-function)
-       js-syntax-propertize-function)
+       #'js-syntax-propertize)
 
   (set (make-local-variable 'parse-sexp-ignore-comments) t)
   (set (make-local-variable 'parse-sexp-lookup-properties) t)





^ permalink raw reply	[flat|nested] 6+ messages in thread

* bug#9183: js-mode highlights quotes inside regexp character sets as string
  2011-08-12 15:48     ` Stefan Monnier
@ 2011-08-12 16:47       ` Daniel Colascione
  2011-08-12 17:25       ` Štěpán Němec
  1 sibling, 0 replies; 6+ messages in thread
From: Daniel Colascione @ 2011-08-12 16:47 UTC (permalink / raw)
  To: Stefan Monnier; +Cc: Štěpán Němec, 9183

[-- Attachment #1: Type: text/plain, Size: 545 bytes --]

On 8/12/2011 8:48 AM, Stefan Monnier wrote:
>> Thanks, but it doesn't seem to fix anything, just mess things up in a
>> somewhat different way.
> 
> It did fix the problem to some extent, but indeed it suffered from
> a "multiline" problem (partly inherited from the previous code).
> 
> I've installed the patch below which should fix the problem, tho it may
> introduce others.  Please try it out.

I haven't had a chance to try it yet, but the patch below is actually
pretty clever. I'll have to add its approach to my toolbox.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 258 bytes --]

^ permalink raw reply	[flat|nested] 6+ messages in thread

* bug#9183: js-mode highlights quotes inside regexp character sets as string
  2011-08-12 15:48     ` Stefan Monnier
  2011-08-12 16:47       ` Daniel Colascione
@ 2011-08-12 17:25       ` Štěpán Němec
  1 sibling, 0 replies; 6+ messages in thread
From: Štěpán Němec @ 2011-08-12 17:25 UTC (permalink / raw)
  To: Stefan Monnier; +Cc: 9183

On Fri, 12 Aug 2011 17:48:11 +0200
Stefan Monnier wrote:

> I've installed the patch below which should fix the problem, tho it may
> introduce others.  Please try it out.

That seems to work fine, thank you!

> Hmm, that makes it more interesting.
> Especially if you ignore semantics and focus on syntax where it would
> seem that "x = /foo/ / /bar/;" would then be valid, tho meaningless).

Heh. And indeed spidermonkey does accept that, too. You can apparently
even divide string literals: "a" / "b" => NaN

-- 
Štěpán





^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2011-08-12 17:25 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-07-27 19:27 bug#9183: js-mode highlights quotes inside regexp character sets as string Štěpán Němec
2011-08-05 16:54 ` Stefan Monnier
2011-08-06 11:34   ` Štěpán Němec
2011-08-12 15:48     ` Stefan Monnier
2011-08-12 16:47       ` Daniel Colascione
2011-08-12 17:25       ` Štěpán Němec

Code repositories for project(s) associated with this external index

	https://git.savannah.gnu.org/cgit/emacs.git
	https://git.savannah.gnu.org/cgit/emacs/org-mode.git

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.