codeforce 754 C. Vladik and chat 安排人说话 （经典DP题）

Recently Vladik discovered a new entertainment — coding bots for social networks. He would like to use machine learning in his bots so now he want to prepare some learning data for them.

At first, he need to download t chats. Vladik coded a script which should have downloaded the chats, however, something went wrong. In particular, some of the messages have no information of their sender. It is known that if a person sends several messages in a row, they all are merged into a single message. It means that there could not be two or more messages in a row with the same sender. Moreover, a sender never mention himself in his messages.

Vladik wants to recover senders of all the messages so that each two neighboring messages will have different senders and no sender will mention himself in his messages.

He has no idea of how to do this, and asks you for help. Help Vladik to recover senders in each of the chats!

Input

The first line contains single integer t (1 ≤ t ≤ 10) — the number of chats. The t chats follow. Each chat is given in the following format.

The first line of each chat description contains single integer n (1 ≤ n ≤ 100) — the number of users in the chat.

The next line contains n space-separated distinct usernames. Each username consists of lowercase and uppercase English letters and digits. The usernames can't start with a digit. Two usernames are different even if they differ only with letters' case. The length of username is positive and doesn't exceed 10 characters.

The next line contains single integer m (1 ≤ m ≤ 100) — the number of messages in the chat. The next m line contain the messages in the following formats, one per line:

• <username>:<text> — the format of a message with known sender. The username should appear in the list of usernames of the chat.
• <?>:<text> — the format of a message with unknown sender.

The text of a message can consist of lowercase and uppercase English letter, digits, characters '.' (dot), ',' (comma), '!'(exclamation mark), '?' (question mark) and ' ' (space). The text doesn't contain trailing spaces. The length of the text is positive and doesn't exceed 100 characters.

We say that a text mention a user if his username appears in the text as a word. In other words, the username appears in a such a position that the two characters before and after its appearance either do not exist or are not English letters or digits. For example, the text "Vasya, masha13 and Kate!" can mention users "Vasya", "masha13", "and" and "Kate", but not "masha".

It is guaranteed that in each chat no known sender mention himself in his messages and there are no two neighboring messages with the same known sender.

Output

Print the information about the t chats in the following format:

If it is not possible to recover senders, print single line "Impossible" for this chat. Otherwise print m messages in the following format:

<username>:<text>

If there are multiple answers, print any of them.

Examples

input

1
2
Vladik netman
2
?: Hello, Vladik!
?: Hi

output

netman: Hello, Vladik!
Vladik: Hi

input

1
2
netman vladik
3
netman:how are you?
?:wrong message
vladik:im fine

output

Impossible

input

2
3
netman vladik Fedosik
2
?: users are netman, vladik, Fedosik
vladik: something wrong with this chat
4
netman tigerrrrr banany2001 klinchuh
4
?: tigerrrrr, banany2001, klinchuh, my favourite team ever, are you ready?
klinchuh: yes, coach!
?: yes, netman
banany2001: yes of course.

output

Impossible
netman: tigerrrrr, banany2001, klinchuh, my favourite team ever, are you ready?
klinchuh: yes, coach!
tigerrrrr: yes, netman
banany2001: yes of course.

题目大意：

       有n个人和m句话，有些话的说话人不明，要求是每个人不能连着说两句话，每句话不能提到自己，看看能不能将这些话的说话人都找出来，答案可能不固定。

解题思路：

       根据题目意思将问题简化，建立dp[i][j]意思为第 j 句话可能是 i 说的，首先明确说话人的话，将已知的说话人作为唯一的可能对象，其他的话只要不是提到的人都作为可能对象。接下来其实就是dfs找一条结果树，只要满足相邻两句话不是一个人说的就行了，但是直接爆搜的复杂度有点大，可以使用记忆化搜索，记录所有可能解。最后输出就好了。

我的解题思路

说话的顺序就是一个天然的序， 安排第i个人说话的时候，有两个限制条件， 1. 不能与前一个说话的人相同， 2. 不能出现在后面说话的句子里

那么DP 思想就很显然了。。。。。 F【i,j】 表示第i句话是否为第j个人说的。

转移方程也就so eazy了 ， 我们可以通过枚举前一句话是由谁讲的，进行更新状态！！！！ GOOD 好题啊艹

昂神代码来一发 233333

#include <bits/stdc++.h>
using namespace std;
const int N = 105;

int n , m;
map<string , int> h;
string name[N] , str[N] , re[N];
bool f[N][N];
int g[N][N];

void print(int x , int y) {
    if (x == 0) {
        return;
    }
    print(x - 1 , g[x][y]);
    cout << name[y] << re[x - 1] << endl;
}

void work() {
    scanf("%d" , &n);
    h.clear();       //名字编号 
    for (int i = 0 ; i < n ; ++ i) {
        cin >> name[i];
        h[name[i]] = i;
    }
    scanf("%d\n" , &m);
    memset(f , 0 , sizeof(f));      // f【i，j】 = 1 表示 第i句话是第j个人说的 ........  
    memset(g , 0 , sizeof(g));
    f[0][n] = 1;
    for (int i = 0 ; i < m ; ++ i) {
        getline(cin , str[i]);
        re[i].clear();
    }
    for (int i = 0 ; i < m ; ++ i) {
        string usr , buf = str[i];
        int p = 0;
        while (p < buf.size() && buf[p] != ':') {
            usr += buf[p ++];
        }
        int q = p;
        while (q < buf.size()) {
            re[i] += buf[q ++];
        }
        set<int> mention;
        while (p < buf.size()) {
            string word;
            while (p < buf.size() && isalnum(buf[p])) {
                word += buf[p ++];
            }
            if (h.count(word)) {             //存在该名单中 
                mention.insert(h[word]);     // 第i句化不能由  mention 这些人 来讲 
            }
            if (p < buf.size()) ++ p;
        }
        if (usr == "?") {                    //该名单暂定时 
            for (int j = 0 ; j <= n ; ++ j) {  // j<=n 考虑第一个人这种特殊情况   ，遍历第i句话可以由哪些人说 
                if (!f[i][j]) continue;        // 前一句话话如由j这个人说  ..  进行更新 
                for (int k = 0 ; k < n ; ++ k) {      //则接下来只要这两个人不相邻，且不在黑名单中 就可以更新 
                    if (mention.count(k) || k == j) continue;
                    f[i + 1][k] = 1;         
                    g[i + 1][k] = j;          // 记录该状态的转移 
                }
            }
        }
		else {
            if (!h.count(usr)) {     // 不存在这个人 
                puts("Impossible");
                return;
            }
            int id = h[usr];          //第id个人 
            if (!mention.count(id)) {       //第id个人不在黑名单内 
                for (int j = 0 ; j <= n ; ++ j) {     //更新当前这句话是由 id 这个讲的 
                    if (f[i][j] && id != j) {
                        f[i + 1][id] = 1;
                        g[i + 1][id] = j;
                    }
                }
            }
        }
    }
    int x = -1;
    for (int i = 0 ; i < n ; ++ i) {
        if (f[m][i]) {
            x = i;
        }
    }
    if (x == -1) {
        puts("Impossible");
    } else {
        print(m , x);
    }
}

int main() {
    int T;
    scanf("%d" , &T);
    while (T --) {
        work();
    }
}