Hett’s Problem 1.28

August 9, 2011

Over at PrologSite, Werner Hett gives a list of Ninety-Nine Prolog Problems “to give you the opportunity to practice your skills in logic programming.” Today’s exercise is number 1.28 on the list. Hett gives the problem two stars, meaning that he expects a skilled Prolog programmer to spend thirty to ninety minutes to solve it. Here is Hett’s statement of the problem:

1.28 (**) Sorting a list of lists according to length of sublists
a) We suppose that a list (InList) contains elements that are lists themselves. The objective is to sort the elements of InList according to their length. E.g. short lists first, longer lists later, or vice versa.

?- lsort([[a,b,c],[d,e],[f,g,h],[d,e],[i,j,k,l],[m,n],[o]],L).
L = [[o], [d, e], [d, e], [m, n], [a, b, c], [f, g, h], [i, j, k, l]]

b) Again, we suppose that a list (InList) contains elements that are lists themselves. But this time the objective is to sort the elements of InList according to their length frequency; i.e. in the default, where sorting is done ascendingly, lists with rare lengths are placed first, others with a more frequent length come later.

?- lfsort([[a,b,c],[d,e],[f,g,h],[d,e],[i,j,k,l],[m,n],[o]],L).
L = [[i, j, k, l], [o], [a, b, c], [f, g, h], [d, e], [d, e], [m, n]]

Note that in the above example, the first two lists in the result L have length 4 and 1, both lengths appear just once. The third and forth list have length 3; there are two list of this length. And finally, the last three lists have length 2. This is the most frequent length.

Your task is to solve Hett’s problem 1.28; you need not use Prolog. Be sure to follow Hett’s advice:

Your goal should be to find the most elegant solution of the given problems. Efficiency is important, but logical clarity is even more crucial.

When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.


Pages: 1 2

16 Responses to “Hett’s Problem 1.28”

  1. F. Carr said
    def hett1_28a(lsts):
        return sorted(lsts, key=len)
    from collections import defaultdict
    def hett1_28b(lsts):
        freq = defaultdict(int)
        for l in lsts:
            freq[len(l)] += 1
        return sorted(lsts, key=lambda l:freq[len(l)])
  2. Mike said

    Python 3 (Should work in 2.7 also).
    Basically the same as Carr’s above.

    from collections import Counter
    from functools import partial
    lsort = partial(sorted, key=len)
    lfsort = lambda r:sorted(r, key=lambda v:Counter(map(len,r))[len(v)])
  3. […] today’s Programming Praxis, our goal is to sort a list of lists by length and by length frequency. […]

  4. My Haskell solution (see http://bonsaicode.wordpress.com/2011/08/09/programming-praxis-hett%E2%80%99s-problem-1-28/ for a version with tests and comments):

    import qualified Data.List.Key as K
    byLength :: [[a]] -> [[a]]
    byLength = K.sort length
    byLengthFreq :: [[a]] -> [[a]]
    byLengthFreq = concat . byLength . K.group length . byLength
  5. Graham said

    Since there’s already a couple Python solutions, here’s my try in Common Lisp:

    (defun lsort (xss)
      (stable-sort (copy-list xss) #'< :key #'list-length))
    (defun lfsort (xss)
      (let ((freq (make-hash-table)))
          (dolist (xs xss) (incf (gethash (list-length xs) freq 0)))
          (stable-sort (copy-list xss) #'<
                :key #'(lambda (xs) (gethash (list-length xs) freq))))))

    I used stable-sort to make sure order was preserved between elements
    that are the same by the sort’s comparison. My lfsort basically uses the
    same logic as the Python solutions above. I tried to wrap it all up in a loop,
    but was unable to get things returned properly; if there are any CL gurus out there who
    could take the time to improve my lfsort, I’d appreciate the help!

  6. slabounty said

    A ruby version …

    list = [%w[a, b, c], %w[d], %w[e, f], %w[g, h, i, j, k], %w[l], %w[m, n, o]]
    list_sort_length = list.sort {|a, b| a.length <=> b.length}
    p list_sort_length
    hist = Hash.new{|h, k| h[k] = []}
    list.each { |l| hist[l.length] << l }
    list_sort_hist = []
    hist.sort {|a,b| a.length <=> b.length}.each { |key, value| value.each {|e| list_sort_hist << e } }
    p list_sort_hist
  7. Mike said

    Forget my function for lfsort shown above. I don’t know what I was thinking–it recreates the Counter() for every element in the input list. O(n^2) or something like that.

    A better lfsort is a variation on F. Carr’s:

    from collections import defaultdict
    from operator import concat
    from functools import reduce
    def lfsort(seq):
    	d = defaultdict(list)
    	for lst in seq:
    	return reduce(concat, lsort(d.values()))

    Or if you like one-liners, you could do it thus:

    lfsort = lambda s:reduce(concat,lsort([list(v) for _,v in groupby(lsort(s),len)]))

    Which, I noticed, is the same as Remco’s much more elegant Haskell code.

  8. Axio said
    (defparameter *l* '((a b c) (d e) (f g h) (d e) (i j k l) (m n) (o)))
    (defun f1 (l)
    &nbsp;&nbsp;(sort l #'(lambda (x y)
    &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(&lt;= (list-length x)
    &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(list-length y)))))
    (defun f2 (l)
    &nbsp;&nbsp;(let ((h (make-hash-table :test #'equal)))
    &nbsp;&nbsp;&nbsp;&nbsp;(loop for i in l
    &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;do (incf (gethash i h 0))
    &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;finally (let* ((r '()))
    &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(maphash (lambda (k v) (push (cons v k) r)) h)
    &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;(return (mapcar #'cdr (sort r #'&lt; :key #'car)))))))
  9. Axio said

    I mixed HTML and sourcecode option in the above post, and trying to repost didn’t show anything… I try again…
    @Graham: maybe something like this:

    (defparameter *l* '((a b c) (d e) (f g h) (d e) (i j k l) (m n) (o)))
    (defun f1 (l)
      (sort l #'(lambda (x y)
                  (<= (list-length x)
                      (list-length y)))))
    (defun f2 (l)
      (let ((h (make-hash-table :test #'equal)))
        (loop for i in l
              do (incf (gethash i h 0))
              finally (let* ((r '()))
                        (maphash (lambda (k v) (push (cons v k) r)) h)
                        (return (mapcar #'cdr (sort r #'< :key #'car)))))))
  10. Axio said

    Oh, f2 didn’t do what was asked for…
    Correct answer is

    (defun f2 (l)
      (let ((h (make-hash-table)))
        (loop for i in l
              do (incf (gethash (list-length i) h 0))
              finally (return  (sort l #'< :key #'(lambda (x) (gethash (list-length x) h)))))))

    but loop is in fact quite useless here, and the code amounts to what you wrote…

  11. Graham said

    Thanks @Axio! I was trying to make it all self-contained, with the hash
    table one of the local “for” variables in the loop, but could never get it
    working. Perhaps I’m overly fond of the loop macro, and shouldn’t get
    so fixated on using it for every problem :-)

  12. Boyo said

    Perl 5 solution, making use of the Schwartzian transform.

    sub lsort {
            map  { $_->[0] }
            sort { $a->[1] cmp $b->[1] }
            map  { [$_, scalar @$_] }
    sub lfsort {
        my %freq_of;
        foreach my $sublist (@_) {
            $freq_of{scalar @$sublist}++;
            map  { $_->[0] }
            sort { $a->[1] cmp $b->[1] }
            map  { [$_, $freq_of{scalar @$_}] }

    And a link to codepad with the original list sorted.

  13. Here’s my clojure solution. More details at my blog.

    ;;Hett's Problem 1.28
    (def test-list [[:a :b :c] [:d :e] [:f :g :h] [:d :e] [:i :j :k :l] [:m :n] [:o]])
    (defn sort-by-length [some-list]
      (sort-by count some-list))
    (defn sort-by-length-freq [some-list]
      (let [freq-map (frequencies (map count test-list))]
        (sort-by (fn [l] (freq-map (count l))) some-list)))
  14. #lang racket
    (define (lsort lists)
      (sort lists < #:key length))
    (define (lfsort lists)
      (let ((freq (for/fold ((h (hash))) ((l lists))
                    (dict-update h (length l) add1 0))))
        (sort lists < #:key (λ (l) (dict-ref freq (length l))))))
  15. wu said

    i did this in perl 5 also, with a minor variation on the lfsort posted by boyo:

    sub lfsort {
    my $length_freq;

    return map { $_->[1] }
    sort { $length_freq->{ $a->[0] } $length_freq->{ $b->[0] } }
    map { [ scalar @$_, $_ ] }
    map { $length_freq->{ scalar @$_ }++; $_ } @_


  16. wu said

    oops, forgot to mention that the perl5 solution does not require schwartzian transformation on the lsort:

    sub lsort {
    return sort { scalar @{$a} scalar @{$b} } @_;

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: