为何向vector插入内容后stringstream出现乱码？如何修复？

问题背景 #

当前代码中，注释掉mapLayerCollision[row].push_back(std::stoi(value.str()));时，所有打印的stringstream内容正常；但取消注释后，哪怕是插入前的打印语句（如std::cout << "Value AFTER: " << value.str() << std::endl;）也会出现乱码，且乱码语句位于插入代码之前。

核心代码 #

void someFunc {
    std::stringstream value;
    std::vector<std::vector<int>> mapLayerCollision;
    int row = 0;

    for(int i = 0; i < gid_list.length(); i++) {
        if(gid_list[i] == ',') {
            value.str("");
            int j = 1;
            while (gid_list[i-j] != ',' and i - j > 0) {
                std::cout << gid_list[i-j] << " | ";
                j++;
                std::cout << gid_list[i-j] << std::endl;
            }
            j--;
            value.str("");
            std::cout << "j: " << j << std::endl;
            while (j > 0) {
                std::cout << "Inserting: " << gid_list[i-j] << std::endl;
                value << gid_list[i-j];
                std::cout << "Value AFTER: " << value.str() << std::endl;
                j--;
            }
            std::cout << "Putting into vector: " << std::stoi(value.str()) << std::endl;
            //mapLayerCollision[row].push_back(std::stoi(value.str()));
            value.str("");
        }
        
        else if(gid_list[i] == '\n') {
            value.str("");
            int j = 1;
            while (gid_list[i-j] != ',') {
                j++;
            }
            j--;
            value.str("");
            while (j > 0) {
                value << gid_list[i-j];
                j--;
            }
            std::cout << "Putting into vector: " << value.str() << std::endl;
            //mapLayerCollision[row].push_back(std::stoi(value.str()));
            value.str("");
            row++;
        }
    }
}

gid_list内容 #

79,454,454,454,454,454,454,454,454,454,454,454,454,454,454,454,454,454,454,454,454,454,454,455,70,
274,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
274,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
274,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
274,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,195,210,0,0,0,0,0,259,
274,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,274,0,0,0,0,0,259,
274,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,323,338,0,0,0,0,0,259,
274,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,387,402,0,0,0,0,0,259,
274,0,0,346,0,0,0,350,0,0,0,0,0,0,0,0,0,451,466,0,0,0,0,0,259,
274,0,0,410,0,0,0,414,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
274,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
338,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
402,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
338,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,195,210,0,0,0,0,0,0,259,
402,0,0,0,0,195,198,199,200,210,0,0,0,0,0,0,451,466,0,0,0,0,0,0,259,
274,0,0,0,0,387,264,264,264,402,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
274,0,0,0,0,451,455,456,457,466,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
274,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,259,
212,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,216,
73,198,198,198,198,198,198,198,198,198,198,198,198,198,198,198,198,198,199,200,201,199,200,201,76

原因分析 #

1. vector越界访问：mapLayerCollision是二维vector，但初始化时为空，mapLayerCollision[row]在row=0时会访问不存在的元素，属于未定义行为。
2. 未定义行为的连锁破坏：未定义行为会导致程序出现任何诡异结果，包括看似不相关的代码异常。越界访问会破坏内存，可能覆盖stringstream内部的缓冲区数据，或干扰程序执行流程，最终导致打印乱码。
3. 注释时正常的原因：注释插入vector的代码后，没有触发越界访问的未定义行为，内存未被破坏，stringstream的内容和打印自然正常。

修复方案 #

核心是确保访问二维vector的行之前，该行已被创建，同时修正循环逻辑的错误：

修正后的代码 #

void someFunc {
    std::stringstream value;
    std::vector<std::vector<int>> mapLayerCollision;
    int row = 0;
    // 提前创建第一行
    mapLayerCollision.emplace_back();

    for(int i = 0; i < gid_list.length(); i++) {
        if(gid_list[i] == ',') {
            value.str("");
            int j = 1;
            // 修正循环条件：避免跳过第一个元素
            while (i - j >= 0 && gid_list[i-j] != ',') {
                j++;
            }
            j--;
            value.str("");
            std::cout << "j: " << j << std::endl;
            while (j > 0) {
                value << gid_list[i-j];
                j--;
            }
            int num = std::stoi(value.str());
            std::cout << "Putting into vector: " << num << std::endl;
            mapLayerCollision[row].push_back(num);
            value.str("");
        }
        
        else if(gid_list[i] == '\n') {
            value.str("");
            int j = 1;
            while (i - j >= 0 && gid_list[i-j] != ',') {
                j++;
            }
            j--;
            value.str("");
            while (j > 0) {
                value << gid_list[i-j];
                j--;
            }
            int num = std::stoi(value.str());
            std::cout << "Putting into vector: " << num << std::endl;
            mapLayerCollision[row].push_back(num);
            value.str("");
            // 创建新行用于下一组数据
            mapLayerCollision.emplace_back();
            row++;
        }
    }
    // 移除最后可能存在的空行
    if(!mapLayerCollision.empty() && mapLayerCollision.back().empty()) {
        mapLayerCollision.pop_back();
    }
}

关键优化点 #

• 修正了处理逗号的循环条件，避免跳过第一个元素；
• 提前创建vector的行，确保mapLayerCollision[row]访问合法；
• 将value.str()转成int后存储，避免多次调用std::stoi，提升效率同时减少重复解析的风险。

内容的提问来源于stack exchange，提问作者DinoNuggies