Racket code to find missing ranges

Question

I'm doing some toy problems in a variety of languages and wanted to do the following in Racket Lisp.

---

Given a sorted integer array nums, where the range of elements are in the inclusive range [lower, upper], return its missing ranges.

0, 1, 3, 50, 75

Lower: 0

Upper: 99

Expected Output:

["2", "4->49", "51->74", "76->99"]

---

As you can see on the linked page I had to take two whacks at it but eventually came up with the following. How does it look? Is it idiomatic? Could I have done certain things far better?

(define (missing-ranges nums lower upper)
 (in-generator
 (define (yield-missing-range current num)
 (yield (if (eq? num (add1 current))
 (~a current)
 (~a current "->" (sub1 num)))))
 (let recur ([remaining-nums nums]
 [current lower])
 (match remaining-nums
 [(cons num rest-nums) (if (< current num)
 (begin
 (yield-missing-range current num)
 (recur rest-nums num))
 (recur rest-nums (add1 current)))]
 ['() (cond [(<= current upper)
 (yield-missing-range current (add1 upper))])]))))

Answer 1

Your solution seems to me quite complex, with iterations, recursions, and the use of match.

Here is an alternative version with a simple linear recursion, where the result is given with lists and not with strings.

(define (missing-ranges nums lower upper)
 (cond ((> lower upper) '())
 ((null? nums) (list (list lower upper)))
 ((= lower (car nums)) (missing-ranges (cdr nums) (+ lower 1) upper))
 ((= (car nums) (+ 1 lower)) (cons (list lower) (missing-ranges nums (+ lower 1) upper)))
 (else (cons (list lower (- (car nums) 1)) (missing-ranges (cdr nums) (+ (car nums) 1) upper)))))
(missing-ranges '(0 1 3 50 75) 0 99)
'((2) (4 49) (51 74) (76 99))
(missing-ranges '(8) 0 99)
'((0 7) (9 99))
(missing-ranges '() 0 99)
'((0 99))
(missing-ranges '(3 9 88 99) 0 99)
'((0 2) (4 8) (10 87) (89 98))
(missing-ranges '(0 99) 0 99)
'((1 98))
(missing-ranges '(1 2 3) 1 3)
'()

• George Mauer: Yeah, the string thing was an original requirement (though I control the requirements). I suppose taking those output lists and stringifying could just be a separate step. I also wanted to use generators specifically, though I think that should still be possible with your version