bug#12717: 24.2.50; [PATCH] `imenu--split-submenus' incorrectly distinguishes submenus

bug#12717: 24.2.50; [PATCH] `imenu--split-submenus' incorrectly distinguishes submenus

[Drew Adams]

Imenu allows the use of something it calls "special" menu elements,
which have this form: (INDEX-NAME POSITION FUNCTION ARGUMENTS...)
 
`imenu--split-submenus' needs to distinguish submenus from ordinary menu
elements, including from special menu elements.  Currently this is
bugged.
 
The current code for `imenu--split-submenus' does this:
 
(defun imenu--split-submenus (alist)
  "..."
  (mapcar (lambda (elt)
            (if (and (consp elt) (stringp (car elt)) (listp (cdr elt)))
                (imenu--split-menu (cdr elt) (car elt))
              elt))
   alist))
 
The `if' condition should instead test whether the element is a submenu.
We have a function that does that: `imenu--subalist-p'.  I believe this
is the correct code:
 
(defun imenu--split-submenus (alist)
  "..."
  (mapcar (lambda (elt)
            (if (imenu--subalist-p elt)
                (imenu--split-menu (cdr elt) (car elt))
              elt))
   alist))

In GNU Emacs 24.2.50.1 (i386-mingw-nt5.1.2600)
 of 2012-10-22 on DANI-PC
Bzr revision: 110618 monnier@iro.umontreal.ca-20121022132928-232zm0fecassmhfb
Windowing system distributor `Microsoft Corp.', version 5.1.2600
Configured using:
 `configure --with-gcc (4.7) --no-opt --enable-checking --cflags
 -I../../libs/libxpm-3.5.8/include -I../../libs/libxpm-3.5.8/src
 -I../../libs/libpng-1.4.10 -I../../libs/zlib-1.2.6
 -I../../libs/giflib-4.1.4-1/include -I../../libs/jpeg-6b-4/include
 -I../../libs/tiff-3.8.2-1/include
 -I../../libs/libxml2-2.7.8-w32-bin/include/libxml2
 -I../../libs/gnutls-3.0.16/include
 -I../../libs/libiconv-1.14-2-mingw32-dev/include'
 

bug#12717: 24.2.50; [PATCH] `imenu--split-submenus' incorrectly distinguishes submenus

[Stefan Monnier]

> (defun imenu--split-submenus (alist)
>   "..."
>   (mapcar (lambda (elt)
>             (if (and (consp elt) (stringp (car elt)) (listp (cdr elt)))
>                 (imenu--split-menu (cdr elt) (car elt))
>               elt))
>    alist))
 
> The `if' condition should instead test whether the element is a submenu.
> We have a function that does that: `imenu--subalist-p'.  I believe this
> is the correct code:
 
> (defun imenu--split-submenus (alist)
>   "..."
>   (mapcar (lambda (elt)
>             (if (imenu--subalist-p elt)
>                 (imenu--split-menu (cdr elt) (car elt))
>               elt))
>    alist))

This looks eminently reasonable, except that I don't understand the code
enough to be sure (e.g. the current test includes (consp elt) whereas
imenu--split-menu starts right off by calling cdr).  Do you happen to
have some kind of test case?


        Stefan

bug#12717: 24.2.50; [PATCH] `imenu--split-submenus' incorrectly distinguishes submenus

[Drew Adams]

> This looks eminently reasonable, except that I don't 
> understand the code enough to be sure (e.g. the current test
> includes (consp elt) whereas imenu--split-menu starts right
> off by calling cdr).  Do you happen to have some kind of test
> case?

I ran into the bug when trying to make use of so-called "special" menu items.

I add two general toggle items to the menus, along with the general item
*Rescan*.  In doing that I discovered the bug.

This is the redefinition of `imenu--make-index-alist' that I use.  It is here
that I add two menu items: one to toggle sorting on/off, and one to toggle
case-sensitivity of sorting when sorting by name.

;; REPLACE ORIGINAL in `imenu.el'.
;;
;; 1. Respect `ignore-comments-flag', if defined:
;;    use `with-comments-hidden'.
;; 2. Add Imenu+ toggle commands to menu.
;;
(defun imenu--make-index-alist (&optional noerror)
  "Create an index alist for the definitions in the current buffer.
This works by using the hook function `imenu-create-index-function'.
Report an error if the list is empty unless NOERROR is supplied and
non-nil.

If `ignore-comments-flag' is defined and non-nil, then respect it,
ignoring hidden comments.

See `imenu--index-alist' for the format of the index alist."
  (or (and imenu--index-alist
           (or (not imenu-auto-rescan)
               (and imenu-auto-rescan
                    (> (buffer-size) imenu-auto-rescan-maxout))))
      ;; Get the index; truncate if necessary
      (progn (setq imenu--index-alist
                   (save-excursion
                     (save-restriction
                       (widen)
                       (if (and (> emacs-major-version 20)
                                (require 'hide-comnt nil t))
                           (let ((search-invisible  nil))
                             (with-comments-hidden
                                 (point-min) (point-max)
                                 (funcall imenu-create-index-function)))
                         (funcall imenu-create-index-function)))))
             (imenu--truncate-items imenu--index-alist)))
  (or imenu--index-alist
      noerror
      (user-error
       "No items suitable for an index found in this buffer"))
  (or imenu--index-alist  (setq imenu--index-alist  (list nil)))
  (cons imenu--rescan-item              ; `*Rescan*'.
        (cons '("Toggle Case-Sensitive Name-Sort"
                IGNORE
                (lambda (&rest _ignore)
                  (imenup-toggle-case-sensitive-sorting)))
              (cons '("Toggle Sorting"
                      IGNORE
                      (lambda (&rest _ignore)
                        (imenup-toggle-sort)))
                    imenu--index-alist))))

Clearly, adding general menu items this way is a make-do hack.  I don't use the
POSITION, and the functions invoked don't use any of the args.

And I don't know how, using this kind of menu creation, to, for example, make
the case-sensitivity toggle item inactive when the sort method is not by name.
But I didn't find a better way to add ordinary, non-indexing menu items.

Anyway, that's how I ran into the bug.

The full code is here, FWIW:
http://www.emacswiki.org/emacs-en/download/imenu%2b.el

bug#12717: 24.2.50; [PATCH] `imenu--split-submenus' incorrectly distinguishes submenus

[Drew Adams]

Sorry, I was confused in my reply.  This wasn't directly about
`imenu--make-index-alist', but about `imenu--split-submenus'.  I made the fix
for that at the same time, for the same reason (to add the toggle menu items).

`imenu--subalist-p' does this:
(defun imenu--subalist-p (item)
  (and (consp (cdr item)) (listp (cadr item))
       (not (eq (car (cadr item)) 'lambda))))

I don't think it is necessary to check that the elements of the ALIST arg to
`imenu--split-submenus' are conses.  But if you want to be sure that the `cdr'
in `imenu--subalist-p' doesn't cause a problem, then either add a `consp' test
before `imenu--sublist-p' or change the `cdr' to `cdr-safe'.

I'm pretty sure that the ALIST arg will have only cons elements, but I guess
there is no guarantee.

My guess is that the original test (and (consp elt) (stringp (car elt)) (listp
(cdr elt))) was just a feeble attempt to check for a sublmenu.

Anyway, now you know as much as I about the code, and can decide how you want to
fix it.

The point of the bug report is that the current definition ends up trying to
handle a "special" menu item as if it were a submenu: it is a consp with a
string car and listp cdr, but it is not a submenu.  It ends up choking.

bug#12717: 24.2.50; [PATCH] `imenu--split-submenus' incorrectly distinguishes submenus

[Stefan Monnier]

> This looks eminently reasonable, except that I don't understand the code
> enough to be sure (e.g. the current test includes (consp elt) whereas
> imenu--split-menu starts right off by calling cdr).  Do you happen to
> have some kind of test case?

Well, I installed it in trunk, it looks safe enough there,


        Stefan