LintCode 960: First Unique Number in Data Stream II (doubly linkedlist好题)

960 · First Unique Number in Data Stream II
Algorithms
Medium

Description
We need to implement a data structure named DataStream. There are two methods required to be implemented:

void add(number) // add a new number
int firstUnique() // return first unique number

You can assume that there must be at least one unique number in the stream when calling the firstUnique.

Example
Example 1:

Input:
add(1)
add(2)
firstUnique()
add(1)
firstUnique()
Output:
[1,2]
Example 2:

Input:
add(1)
add(2)
add(3)
add(4)
add(5)
firstUnique()
add(1)
firstUnique()
add(2)
firstUnique()
add(3)
firstUnique()
add(4)
firstUnique()
add(5)
add(6)
firstUnique()
Output:
[1,2,3,4,5,6]
Tags
Related Problems

209
First Unique Character in a String
Easy

646
First Position Unique Character
Easy

685
First Unique Number in Data Stream
Medium

解法1：
我的做法是基于doubly linked list，同时维护两个map，一个是<num, freq>，一个是<num, Node *>。注意doubley linked list有一个dummyHead, 一个dummyTail。每次都插到head的位置，这样只需要返回dummyTail前面一个元素的值就可以了。注意每次add num的时候，如果发现freq[num]>1了，需要把num对应的Node 节点删除。每次操作应该都是O(1)。

struct Node {
    int val;
    Node *prev, *next;
    Node(int v) : val(v), prev(NULL), next(NULL) {}
};

class DataStream {
public:
    DataStream(){
        dummyHead = new Node(0);
        dummyTail = new Node(0);
        dummyHead->next = dummyTail;
        dummyTail->prev = dummyHead;
    }
    
    /**
     * @param num: next number in stream
     * @return: nothing
     */
    void add(int num) {
        freq[num]++;
        //delete repeat node from doubly linked list
        if (freq[num] > 1 && mp.find(num) != mp.end()) {
            Node *deleteNode = mp[num];
            deleteNode->prev->next = deleteNode->next;
            deleteNode->next->prev = deleteNode->prev;
            delete(deleteNode);
            deleteNode = nullptr;
            mp.erase(num);
        }
        //always add to head
        if (freq[num] == 1) {
            Node *node = new Node(num);
            node->next = dummyHead->next;
            if (dummyHead->next) dummyHead->next->prev = node;
            node->prev = dummyHead;
            dummyHead->next = node;
            mp[num] = node;
        }
    }

    /**
     * @return: the first unique number in stream
     */
    int firstUnique() {
        //if (dummyHead->next == dummyTail) return -1; //但题目保证有至少一个unique，所以这行不需要
        return dummyTail->prev->val;
    }
private:
    unordered_map<int, Node *> mp;
    unordered_map<int, int> freq; //<int, freq>
    Node *dummyHead, *dummyTail;
};

解法2：用queue和unordered_map。
注意: 虽然firstUnique()里面有while()，其均摊时间复杂度仍然是O(1)。

class DataStream {
public:
    DataStream(){
    }
    
    /**
     * @param num: next number in stream
     * @return: nothing
     */
    void add(int num) {
        q.push(num);
        freq[num]++;
    }

    /**
     * @return: the first unique number in stream
     */
    int firstUnique() {
        while(!q.empty() && freq[q.front()] > 1) {
            q.pop();
        }
        if (q.empty()) return -1;
        return q.front();
    }
private:
    queue<int> q;
    unordered_map<int, int> freq; //<int, freq>
};