Data structure: insert, remove, contains, get random element, all at O(1)

Question

I was given this problem in an interview. How would you have answered?

Design a data structure that offers the following operations in O(1) time:

• inse

Answer 1

Though this is way old, but since there's no answer in C++, here's my two cents.

#include <vector>
#include <unordered_map>
#include <stdlib.h>

template <typename T> class bucket{
    int size;
    std::vector<T> v;
    std::unordered_map<T, int> m;
public:
    bucket(){
        size = 0;
        std::vector<T>* v = new std::vector<T>();
        std::unordered_map<T, int>* m = new std::unordered_map<T, int>();
    }
    void insert(const T& item){
        //prevent insertion of duplicates
        if(m.find(item) != m.end()){
            exit(-1);
        }
        v.push_back(item);
        m.emplace(item, size);
        size++;

    }
    void remove(const T& item){
        //exits if the item is not present in the list
        if(m[item] == -1){
            exit(-1);
        }else if(m.find(item) == m.end()){
            exit(-1);
        }

        int idx = m[item];
        m[v.back()] = idx;
        T itm = v[idx];
        v.insert(v.begin()+idx, v.back());
        v.erase(v.begin()+idx+1);
        v.insert(v.begin()+size, itm);
        v.erase(v.begin()+size);
        m[item] = -1;
        v.pop_back();
        size--;

    }

     T& getRandom(){
      int idx = rand()%size;
      return v[idx];

     }

     bool lookup(const T& item){
       if(m.find(item) == m.end()) return false;
       return true;

     }
    //method to check that remove has worked
    void print(){
        for(auto it = v.begin(); it != v.end(); it++){
            std::cout<<*it<<" ";
        }
    }
};

Here's a piece of client code to test the solution.

int main() {

    bucket<char>* b = new bucket<char>();
    b->insert('d');
    b->insert('k');
    b->insert('l');
    b->insert('h');
    b->insert('j');
    b->insert('z');
    b->insert('p');

    std::cout<<b->random()<<std::endl;
    b->print();
    std::cout<<std::endl;
    b->remove('h');
    b->print();

    return 0;
}

Answer 2

I agree with Anon. Except for the last requirement where getting a random element with equal fairness is required all other requirements can be addressed only using a single Hash based DS. I will choose HashSet for this in Java. The modulo of hash code of an element will give me the index no of the underlying array in O(1) time. I can use that for add, remove and contains operations.