Reverse characters of each word in a sentence

Question

Reverse characters of each word in a sentence. For eg:

My name is alex

changes to

yM eman si xela

I thought of the normal O(n) time algorithm of using two pointer to point to either end of the word and reverse it.

But in the below site

http://www.businessinsider.com/8-mind-bending-interview-questions-that-google-asks-its-engineers-2012-7?op=1

(Refer to ans of ques 2) it is give that converting it to linked list and repetitively applying reversal of linked list for individual word is better. I found the following solution for the same program on Hackerearth:

http://learn.hackerearth.com/question/317/reverse-characters-of-each-word-in-a-sentence/

This solution takes O(n) time and O(n) space. The solution I suggested takes O(n) time O(1) space. How is the second one better?

Following is the code from Hackerearth:

public node stringReverseChars(node ll){
    if(ll == null || ll.next == null)
        return ll;
    node tmp = ll;
    node head = null, prev = null;
    while(tmp != null){
        while(tmp != null && tmp.data == ' '){
            if(head == null)
                head = tmp;
            prev = tmp;
            tmp = tmp.next;
        }
        if(tmp == null)
            break;
        node curr = tmp;
        while(tmp.next != null && tmp.next.data != ' '){
            tmp = tmp.next;
        }
        node np = tmp.next;
        tmp.next = null;
        node rev = reverseLL(curr);
        if(prev != null)
            prev.next = rev;
        prev = curr;
        curr.next = np;
        if(head == null)
            head = rev;
        tmp = np;
    }
    return head;
}

Answer 1

I'm pretty skeptical that those other approaches are better. They have worse memory usage (Θ(n) versus O(1)) and worse locality of reference (they use linked lists rather than arrays). I don't see anything wrong with your solution; in fact, I think it's the standard way to do this.

Hope this helps!