You are given an n × m rectangular table consisting of lower case English letters. In one operation you can completely remove one column from the table. The remaining parts are combined forming a new table. For example, after removing the second column from the table
abcd edfg hijk
we obtain the table:
acd efg hjk
A table is called good if its rows are ordered from top to bottom lexicographically, i.e. each row is lexicographically no larger than the following one. Determine the minimum number of operations of removing a column needed to make a given table good.
Input
The first line contains two integers — n and m (1 ≤ n, m ≤ 100).
Next n lines contain m small English letters each — the characters of the table.
Output
Print a single number — the minimum number of columns that you need to remove in order to make the table good.
Examples
Input
1 10 codeforces
Output
0
Input
4 4 case care test code
Output
2
Input
5 4 code forc esco defo rces
Output
4
Note
In the first sample the table is already good.
In the second sample you may remove the first and third column.
In the third sample you have to remove all the columns (note that the table where all rows are empty is considered good by definition).
Let strings s and t have equal length. Then, s is lexicographically larger than t if they are not equal and the character following the largest common prefix of s and t(the prefix may be empty) in s is alphabetically larger than the corresponding character of t.
题目大意:给出了n*m的矩阵题目要求删除最少的列使任意的第i行的字典序大于第i-1行的字典序。
解题思路:首先应该每列判断是否是递增的,如果该列不是递增的就要删除。如果是递增的就标记。最后如果判断到某列的不合法的,就往前找是否存在严格大于的。
#include<iostream>
#include<cstdio>
#include<cstring>
#include<queue>
#include<set>
#include<algorithm>
#include<map>
#include<cmath>
#define rep(i,j,k) for(int i=j;i<=k;i++)
#define inf 0x3f3f3f3f
#define LL long long
#define fir first
#define sec second
#define lson rt<<1
#define rson rt<<1|1
#define sq(x) (x)*(x)
#define sca(x) scanf("%d",&x)
#define pb(x) push_back(x)
using namespace std;
#define N 10005
#define maxn 2000005
char s[105][105];
int pre[105];
int n,m;
bool Find(int now,int p,int c)
{
//cout<<"Find:"<<now<<" "<<p<<" "<<c<<endl;
if(now==0) return false;
if(s[c][now]>s[c-1][now])return true;
else return Find(p,pre[p],c);
}
bool dfs(int now,int p,int c,int f)
{
//cout<<"dfs:"<<now<<" "<<p<<" "<<c<<endl;
if(c==n+1)return true;
if(p==0)
{
if(s[c][now]>=s[c-1][now])return dfs(now,p,c+1,0);
else return false;
}
else
{
if(s[c][now]>=s[c-1][now]) return dfs(now,p,c+1,0);
else
{
if(Find(p,pre[p],c)) return dfs(now,p,c+1,0);
else return false;
}
}
}
int get_pre(int now)
{
for(int i=now;i>=1;i--)
{
if(pre[i]!=0)return pre[i];
}
}
int main()
{
scanf("%d%d",&n,&m);
rep(i,1,n)
{
scanf("%s",s[i]+1);
}
memset(pre,0,sizeof(pre));
int ans=0;
for(int i=1;i<=m;i++)
{
if(dfs(i,pre[i],2,0))pre[i+1]=i;
else
{
ans++;
pre[i+1]=get_pre(i);
}
}
cout<<ans<<endl;
}