|
|
|
|
|
#include <bits/stdc++.h>
|
|
|
|
|
|
using namespace std;
|
|
|
|
|
|
|
|
|
|
|
|
typedef pair<int, int> PII;
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
|
举栗子： 以3行5列为例，共3*5=15个格子
|
|
|
|
|
|
是4行*6列=24个点
|
|
|
|
|
|
推广公式化:(n+1)*(m+1)
|
|
|
|
|
|
*/
|
|
|
|
|
|
const int N = 510 * 510;
|
|
|
|
|
|
const int M = 4 * N; // 1个格子里面有4条边，分别为 a->b,b->a,c->d,d->c
|
|
|
|
|
|
const int INF = 0X3f3f3f3f;
|
|
|
|
|
|
|
|
|
|
|
|
int n, m;
|
|
|
|
|
|
int h[N], e[M], w[M], ne[M], idx;
|
|
|
|
|
|
int dist[N];
|
|
|
|
|
|
bool st[N];
|
|
|
|
|
|
|
|
|
|
|
|
void add(int a, int b, int c) {
|
|
|
|
|
|
e[idx] = b, w[idx] = c, ne[idx] = h[a], h[a] = idx++;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
int getNum(int x, int y) {
|
|
|
|
|
|
/*
|
|
|
|
|
|
* 按点看 ，为 0,1,2,3,4,...,m
|
|
|
|
|
|
* 按格子看，为 1,2,3,4,...,m
|
|
|
|
|
|
即按点看，每行m+1个点
|
|
|
|
|
|
推广公式化： 编号 = 行号(可以从0开始，也可以从1开始)*列的数量 + 列号
|
|
|
|
|
|
*/
|
|
|
|
|
|
return x * (m + 1) + y; //按点看，共m+1列
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
int dijkstra() {
|
|
|
|
|
|
//初始化最短距离
|
|
|
|
|
|
memset(dist, 0x3f, sizeof dist);
|
|
|
|
|
|
dist[0] = 0; // 0号点的最短距离是0
|
|
|
|
|
|
|
|
|
|
|
|
// greater<>()小顶层堆
|
|
|
|
|
|
priority_queue<PII, vector<PII>, greater<PII>> q;
|
|
|
|
|
|
q.emplace(0, 0); //注意:先是距离，后是序号！之所以这样设计，是因为小顶堆的排序是默认安排PII的第一维x进行由小到大
|
|
|
|
|
|
|
|
|
|
|
|
while (q.size()) {
|
|
|
|
|
|
auto t = q.top();
|
|
|
|
|
|
q.pop();
|
|
|
|
|
|
int u = t.second, distance = t.first;
|
|
|
|
|
|
//如果已经走过,不需要再尝试
|
|
|
|
|
|
if (st[u]) continue;
|
|
|
|
|
|
//标识走过
|
|
|
|
|
|
st[u] = true;
|
|
|
|
|
|
//枚举每个出边
|
|
|
|
|
|
for (int i = h[u]; ~i; i = ne[i]) {
|
|
|
|
|
|
int j = e[i];
|
|
|
|
|
|
if (dist[j] > distance + w[i]) { //在原距离distanc基础上，再走w[i]后可以到达j,如果这样走近，则更新dist[j]
|
|
|
|
|
|
dist[j] = distance + w[i];
|
|
|
|
|
|
q.emplace(dist[j], j); //将j加入队列，继续扩展
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
//计算最后一个坐标位置对应的节点，它离起点的最短距离是多少
|
|
|
|
|
|
return dist[getNum(n, m)];
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
signed main() {
|
|
|
|
|
|
//加快读入
|
|
|
|
|
|
cin.tie(0), ios::sync_with_stdio(false);
|
|
|
|
|
|
int T;
|
|
|
|
|
|
cin >> T;
|
|
|
|
|
|
while (T--) {
|
|
|
|
|
|
//还原是否走过的状态st=false
|
|
|
|
|
|
memset(st, false, sizeof st);
|
|
|
|
|
|
//多组测试数据，清空邻接表
|
|
|
|
|
|
memset(h, -1, sizeof h);
|
|
|
|
|
|
idx = 0;
|
|
|
|
|
|
|
|
|
|
|
|
// n行m列
|
|
|
|
|
|
cin >> n >> m;
|
|
|
|
|
|
//读入数据+建图
|
|
|
|
|
|
for (int i = 1; i <= n; i++) {
|
|
|
|
|
|
for (int j = 1; j <= m; j++) {
|
|
|
|
|
|
char op;
|
|
|
|
|
|
cin >> op;
|
|
|
|
|
|
//现在读入的是当前(i,j)坐标的格子中操作符
|
|
|
|
|
|
/*
|
|
|
|
|
|
点坐标(0,0) 点坐标(0,1) 点坐标(i-1,j-1) 点坐标(i-1,j)
|
|
|
|
|
|
格子坐标：(1,1) => 格子坐标：(i,j)
|
|
|
|
|
|
点坐标(1,0) 点坐标(1,1) 点坐标(i,j-1) 点坐标(i,j)
|
|
|
|
|
|
*/
|
|
|
|
|
|
int a = getNum(i - 1, j - 1);
|
|
|
|
|
|
int b = getNum(i, j);
|
|
|
|
|
|
int c = getNum(i, j - 1);
|
|
|
|
|
|
int d = getNum(i - 1, j);
|
|
|
|
|
|
|
|
|
|
|
|
// Q:为什么权值定义的是 /:1 \:0 呢？
|
|
|
|
|
|
// A:因为我们描述的是 a->b 的边:
|
|
|
|
|
|
// 如果a->b之间有边，就是左上角到右下角有边, \ , 权值为0.
|
|
|
|
|
|
// 如果a->b之间无边，就是右上角到左下角有边, / , 此时需要把它转过来，权值为1.
|
|
|
|
|
|
int w = op == '/' ? 1 : 0;
|
|
|
|
|
|
add(a, b, w), add(b, a, w); //无向图，双向建边
|
|
|
|
|
|
add(c, d, !w), add(d, c, !w); //取反也建边，模拟01边权的无向图
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
//求最短路
|
|
|
|
|
|
int res = dijkstra(); // dijkstra可以处理非负权边的图
|
|
|
|
|
|
|
|
|
|
|
|
if (res == INF)
|
|
|
|
|
|
puts("NO SOLUTION");
|
|
|
|
|
|
else
|
|
|
|
|
|
printf("%d\n", res);
|
|
|
|
|
|
}
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
}
|
