[Orgmode] Drill mode for org-learn

[Paul Sexton]

I have nearly finished a basic "drill mode" for org-learn. It works a bit 
like Anki/Mnemosyne.

When invoked, it scans files for topics which either have the tag :question:,
or have the LEARN_DATA property (set by org-learn), and which are either 
unscheduled (ie new) or are not scheduled in the future.

It shuffles them, picks a random subset if 
'org-learn-maximum-items-per-drill-session' is set, and presents them one
by one. The user presses a key, then presses a key 0-5 to rate their 
recall. 'org-smart-reschedule' is called for every reviewed item at the
end of the test.

Example topics which it understands:

1. The body under the heading tagged :question: is shown, with the :back: 
(ie "back of the card") subsection hidden.

* Greeting Q1                                   :question:
What is your name? (informal)
** Answer                                           :back:
¿Cómo te llamas?

2. A card with both a front and a back. One side will be shown at random, 
with the other hidden.

* Noun                                          :question:
** Front                                           :front:
el gato
** Back                                             :back:
the cat

3. Cloze deletion. Text within SINGLE square brackets is obscured
with a special font.

* Grammar rule                                  :question:
To form an adverb from an adjective, add [-mente] to 
the [feminine] (gender) form of the adjective.

I have a slight problem however. I don't know how to "record" headings so I
can revisit them later. I need something like:

(org-map-entries
  (lambda () (push (org-save-this-heading) headings))
  "" files)

(dolist (heading headings)
   (jump-to-heading heading)
   ...)

Saving the buffer position does not work because each call to
org-smart-reschedule alters the contents of the buffer.

If/when I solve this problem, I should be able to release this code. 

Paul