You are given a string $s$ consisting of $n$ lowercase Latin letters.

Let's define a substring as a contiguous subsegment of a string. For example, "acab" is a substring of "abacaba" (it starts in position $3$ and ends in position $6$), but "aa" or "d" aren't substrings of this string. So the substring of the string $s$ from position $l$ to position $r$ is $s[l; r] = s_l s_{l + 1} \dots s_r$.

You have to choose exactly one of the substrings of the given string and reverse it (i. e. make $s[l; r] = s_r s_{r - 1} \dots s_l$) to obtain a string that is less lexicographically. Note that it is not necessary to obtain the minimum possible string.

If it is impossible to reverse some substring of the given string to obtain a string that is less, print "NO". Otherwise print "YES" and any suitable substring.

String $x$ is lexicographically less than string $y$, if either $x$ is a prefix of $y$ (and $x \ne y$), or there exists such $i$ ($1 \le i \le min(|x|, |y|)$), that $x_i < y_i$, and for any $j$ ($1 \le j < i$) $x_j = y_j$. Here $|a|$ denotes the length of the string $a$. The lexicographic comparison of strings is implemented by operator < in modern programming languages​​.

Input

The first line of the input contains one integer $n$ ($2 \le n \le 3 \cdot 10^5$) — the length of $s$.

The second line of the input contains the string $s$ of length $n$ consisting only of lowercase Latin letters.

Output

If it is impossible to reverse some substring of the given string to obtain a string which is lexicographically less, print "NO". Otherwise print "YES" and two indices $l$ and $r$ ($1 \le l < r \le n$) denoting the substring you have to reverse. If there are multiple answers, you can print any.

见题面 Input 部分

见题面 Output 部分

In the first testcase the resulting string is "aacabba".