问题
I am trying to write a procedure that takes a a symbol and a list and inserts the symbol at every possible position inside the given list (thus generating a list of lists). I have coded the following definitions, which I must implement:
1
(define (insert-at pos elmt lst)
(if (empty? lst) (list elmt)
(if (= 1 pos)
(cons elmt lst)
(cons (first lst)
(insert-at (- pos 1) elmt (rest lst))))))
2
(define (generate-all-pos start end)
(if (= start end)
(list end)
(cons start (generate-all-pos (+ start 1) end))))
1 takes a position in a list (number), a symbol and the list itself and inserts the symbol at the requested position. 2 takes a start and a target position (numbers) and creates a sorted list with numbers from start to target. So far I have got this:
(define (insert-everywhere sym los)
(cond
[(empty? los) (list sym)]
[else (cons (insert-at (first (generate-all-pos (first los)
(first (foldl cons empty los)))) sym los) (insert-everywhere sym (rest los)))
]
)
)
Which results in
> (insert-everywhere 'r '(1 2 3))
(list (list 'r 1 2 3) (list 2 'r 3) (list 3 'r) 'r)
so I actually managed to move the 'r' around. I'm kind of puzzled about preserving the preceding members of the list. Maybe I'm missing something very simple but I've stared and poked at the code for quite some time and this is the cleanest result I've had so far. Any help would be appreciated.
回答1:
The insert-everywhere procedure is overly complicated, a simple map will do the trick. Try this:
(define (insert-everywhere sym los)
(map (lambda (i)
(insert-at i sym los))
(generate-all-pos 1 (add1 (length los)))))
Also notice that in Racket there exists a procedure called range, so you don't need to implement your own generate-all-pos:
(define (insert-everywhere sym los)
(map (lambda (i)
(insert-at i sym los))
(range 1 (+ 2 (length los)))))
Either way, it works as expected:
(insert-everywhere 'r '(1 2 3))
=> '((r 1 2 3) (1 r 2 3) (1 2 r 3) (1 2 3 r))
回答2:
Óscar López's answer shows how you can do this in terms of the procedures that you've already defined. I'd like to point out a way to do this that recurses down the input list. It uses an auxiliary function called revappend (I've taken the name from Common Lisp's revappend). revappend takes a list and a tail, and efficiently returns the same thing that (append (reverse list) tail) would.
(define (revappend list tail)
(if (null? list)
tail
(revappend (rest list)
(list* (first list) tail))))
> (revappend '(3 2 1) '(4 5 6))
'(1 2 3 4 5 6)
The reason that we're interested in such a function is that as we recurse down the input list, we can build up a list of the elements we've already seen, but it's in reverse order. That is, as we walk down (1 2 3 4 5), it's easy to have:
rhead tail (revappend rhead (list* item tail))
----------- ----------- -----------------------------------
() (1 2 3 4 5) (r 1 2 3 4 5)
(1) (2 3 4 5) (1 r 2 3 4 5)
(2 1) (3 4 5) (1 2 r 3 4 5)
(3 2 1) (4 5) (1 2 3 r 4 5)
(4 3 2 1) (5) (1 2 3 4 r 5)
(5 4 3 2 1) () (1 2 3 4 5 r)
In each of these cases, (revappend rhead (list* item tail)) returns a list with item inserted in one of the positions. Thus, we can define insert-everywhere in terms of rhead and tail, and revappend, if we build up the results list in reverse order, and reverse it at the end of the loop.
(define (insert-everywhere item list)
(let ie ((tail list)
(rhead '())
(results '()))
(if (null? tail)
(reverse (list* (revappend rhead (list* item tail)) results))
(ie (rest tail)
(list* (first tail) rhead)
(list* (revappend rhead (list* item tail)) results)))))
(insert-everywhere 'r '(1 2 3))
;=> '((r 1 2 3) (1 r 2 3) (1 2 r 3) (1 2 3 r))
What's interesting about this is that the sublists all share the same tail structure. That is, the sublists share the structure as indicated in the following “diagram.”
;=> '((r 1 2 3) (1 r 2 3) (1 2 r 3) (1 2 3 r))
; ----- +++ o
; +++ o
; o
来源：https://stackoverflow.com/questions/20125784/insert-everywhere-procedure