检测字符串的编码

#include <iostream>
#include <string>
#include <iconv.h>

// 函数用于尝试检测字符串的编码
std::string guessEncoding(const std::string& input) {
    // 尝试转换为几种常见编码
    const char* encodings[] = {"utf-8", "iso-8859-1", "gbk", "big5", "euc-jp"};
    iconv_t cd;

    for (const auto& encoding : encodings) {
        // 初始化转换描述符
        cd = iconv_open("utf-8", encoding);
        if (cd == (iconv_t)-1) {
            std::cerr << "Error initializing iconv: " << std::strerror(errno) << std::endl;
            continue;
        }

        // 尝试转换
        char output[1024] = {0};
        size_t inLen = input.length();
        char* in = (char*)input.c_str();
        size_t outLen = sizeof(output);
        char* out = output;

        if (iconv(cd, &in, &inLen, &out, &outLen) != (size_t)-1) {
            // 转换成功
            iconv_close(cd);
            return encoding;
        } else {
            // 转换失败，关闭转换描述符
            iconv_close(cd);
        }
    }

    return ""; // 未能确定编码
}

int main() {
    std::string sampleText = "你好，世界！"; // 示例文本

    std::string detectedEncoding = guessEncoding(sampleText);

    std::cout << "Detected Encoding: " << detectedEncoding << std::endl;

    return 0;
}