UTF-8 编码验证

//本地测试没有问题，但是AC=30%，望大佬改进下
//leetcode 393
#include <iostream>
#include <string>
#include <vector>
#include <algorithm>
#include <limits.h>
#include <stdlib.h>

using namespace std;
//字符串转二进制
string stringToBinary(string temp) {
    string res = "";
    int num = stoi(temp);
    for (int i = 7; i >= 0; i--) {
        res += ((num >> i) & 1) + '0';
    }
    return res;
}

void utf8Coding(string s) {
    vector<string> vec;
    string temp;
    string str_bin;
    int len = s.size();
    for (int i = 0; i < len; i++) {
        if (s[i] != ',') {
            temp += s[i];
        }
        else {
            str_bin = stringToBinary(temp);
            vec.push_back(str_bin);
            temp = "";
        }
    }
    str_bin = stringToBinary(temp);
    vec.push_back(str_bin);
    string res = "true";
    int cnt = 0;
    int vec_len = vec.size();
    int pos = 0; //记录1字节字符在原字符串中出现的位置
    bool flag = false;
    //判断为几字节字符，并统计个数
    for (int i = 0; i < vec_len; i++) {
        if (vec[i][0] == '0') {
            pos = i;
            continue;
        }
        //只有前4位有效
        for (int j = 0; j < 4; j++) {
            if (vec[i][j] == '1') {
                cnt++;
            }
            else {
                flag = true;
                break;
            }
        }
        if (flag) {
            break;
        }
    }
    //分两部分判断，第一部分是1字节字符前边的
    for (int i = 0; i < pos; i++) {
        if (vec[i][0] == '1' && vec[i][1] == '0') {
            continue;
        }
        else {
            res = "false";
        }
    }
    //1字节字符后边的
    for (int i = pos + 2; i < cnt; i++) {
        if (vec[i][0] == '1' && vec[i][1] == '0') {
            continue;
        }
        else {
            res = "false";
        }
    }
    cout << res << endl;
}

int main(){
    string s;
    while(cin>>s) {
        utf8Coding(s);
    }
    return 0;
}

import java.util.*;

public class Main {

    public static void main(String[] args) {
        Scanner sc = new Scanner(System.in);

        while (sc.hasNext()) {

            String[] str = sc.next().split(",");
            short[] arr = new short[str.length];
            for (int i = 0; i < str.length; i++) {
                short n = Short.valueOf(str[i]);
                arr[i] = (short)(n & 0xFF);
            }

            boolean illegal = true;
            for (int i = 0; i < arr.length; i++) {
                short b = arr[i];
                if (b >= 0 && b < 128) {
                    continue;
                } else if (b >= 192 && b < 224) {
                    if (i + 1 >= arr.length) {
                        illegal = false;
                        break;
                    }

                    illegal = arr[i + 1] >= 128 && arr[i + 1] < 192;
                    if (!illegal)
                        break;

                    i += 1;
                } else if (b >= 224 && b < 240) {
                    if (i + 2 >= arr.length) {
                        illegal = false;
                        break;
                    }

                    illegal = arr[i + 1] >= 128 && arr[i + 1] < 192 && arr[i + 2] >= 128 && arr[i + 2] < 192;
                    if (!illegal)
                        break;

                    i += 2;
                } else if (b >= 240 && b < 248) {
                    if (i + 3 >= arr.length) {
                        illegal = false;
                        break;
                    }

                    illegal = arr[i + 1] >= 128 && arr[i + 1] < 192 && arr[i + 2] >= 128 && arr[i + 2] < 192 &&
                            arr[i + 3] >= 128 && arr[i + 3] < 192;

                    if (!illegal)
                        break;

                    i += 3;
                } else {
                    illegal = false;
                    break;
                }
            }

            System.out.println(illegal);
        }
    }
}