[alexandria-devel] [RFC] collect-duplicates

[Andreas Fuchs]

Hi there,

Prompted by a discussion on #lisp, I've made the following
collect-duplicates function: Takes a sequence and a test, and returns
as two values a sequence of duplicates and a sequence of the duplicate
count (matching their sequence type to the input sequence's).

(defun collect-duplicates (sequence &key (test #'eql))
  ;; This will work only for TESTs that are valid hash tests:
  (let ((dupes (make-hash-table :test test))
        (known (make-hash-table :test test)))
    (map nil
         (lambda (elt)
           (if (gethash elt known)
               (incf (gethash elt dupes 1))
               (setf (gethash elt known) t)))
         sequence)
    (let ((duplicates (make-sequence (class-of sequence)
(hash-table-count dupes)))
          (counts (make-sequence (class-of sequence) (hash-table-count dupes)))
          (i 0))
      (maphash (lambda (k v)
                 (setf (elt duplicates i) k
                       (elt counts i) v)
                 (incf i))
               dupes)
      (values duplicates counts))))

Issues that I see with it:

 * Needs a :key argument.
 * :test should allow for non-hash-table tests, as well.
 * (class-of ...) is not really a good idea, as it doesn't preserve
the element type for arrays.

If I fixed these issues, would there be interest in having this
function in alexandria?

Cheers,
-- 
Andreas Fuchs, (http://|im:asf@|mailto:asf@)boinkor.net, antifuchs