[字符编码]windwos下使用libiconv转换编码格式(一)
我在工作中常用的转换类型就是Unicode和GBK、GB2312、GB18030的转换。其他编码格式遇到再说。
libiconv.lib是用自己编译libiconv-1.15版本源码生成的,当然下面的代码只是一个demo,仅仅是展示效果,代码中出现的文件需要读者自己准备。
在测试的过程中,也要注意,有些古文中的汉字在GB2312字符集中是没有的,这可能导致转换失败,因为我喜欢用出师表来测试,无意中发现的,GBK和GB18030包含的汉字更全一些,也可以去网上查询,字符集中是否包含此汉字。
#include
#include
#include
#include
#include "iconv.h" //包函libiconv库头文件
//导入libiconv库
#pragma comment(lib,"libiconv.lib")
bool readfile(const std::string& _filepath, std::string& _filecontent)
{
bool res = false;
std::ifstream file(_filepath);
if (!file.is_open()) { // 检查文件是否成功打开
std::cerr << "无法打开文件" << _filepath << std::endl;
}
else {
std::string line;
while (std::getline(file, line)) { // 逐行读取文件内容
_filecontent += line;
}
res = true;
}
file.close(); // 关闭文件
return res;
}
//字符串转换为二进制
void TransStringToBinary(const std::string & _str) {
std::string temp = _str;
for (int i = 0; i < temp.size(); ++i) {
std::bitset<8> binary(temp[i]);
std::cout << binary <<" ";
}
std::cout << std::endl;
}
//使用 libiconv 进行
int TransCore(const char * _pdesc, const char * _psrc, const char * _pstrin, size_t ilen, char * _pstrout, size_t * _polen)
{
const char ** ppin = &_pstrin;
char ** ppout = &_pstrout;
iconv_t cd = iconv_open(_pdesc, _psrc);
if (cd == (iconv_t)-1){
return -1;
}
memset(_pstrout, 0, *_polen);
int res = iconv(cd, ppin, &ilen, ppout, _polen);
std::cout << "iconv res = " << res << std::endl;
iconv_close(cd);
return res;
}
/*
desc 目标编码字符串
src 源编码字符串
_strin 转换前内容
_strout 转换后内容
*/
bool TransEncodeFormat(const char* _desc, const char* _src, const std::string & _strin, std::string & _strout) {
bool res = false;
if (_desc == nullptr || _src == nullptr || _strin.empty()) {
std::cout << "入参不符合要求" << std::endl;
return res;
}
size_t inlen = _strin.length();
#ifdef LOG
std::cout << "需要转换的内容 : [" << _strin << "]" << std::endl;
std::cout << "需要转换的字节数 : [" << inlen << "]" << std::endl;
#endif
size_t outlen = inlen * 10;
char* tempout = new char[outlen];
if (TransCore(_desc, _src, _strin.c_str(), inlen, tempout, &outlen) == 0 && tempout != nullptr) {
res = true;
}
#ifdef LOG
std::cout << "转换后的内容 : [" << tempout << "]" << std::endl;
#endif
std::string temp(tempout);
_strout = tempout;
delete[] tempout;
tempout = nullptr;
return res;
}
int main(int argc, char* argv[])
{
std::cout << "UTF8 -> GBK GB18030 GB2312" << std::endl;
{//utf-8 转换到 GBK
std::string filecontent;
std::string transcontent;
std::string utf8filepath = "./test-file/utf-8.txt";
readfile(utf8filepath, filecontent);
//TransStringToBinary(filecontent);
TransEncodeFormat("GBK", "UTF-8", filecontent, transcontent);
std::cout << "UTF-8["< GBK[" << transcontent <<"]" << std::endl;
}
{//GBK 转换到 GB18030格式
std::string filecontent;
std::string transcontent;
std::string utf8filepath = "./test-file/utf-8.txt";
readfile(utf8filepath, filecontent);
//TransStringToBinary(filecontent);
TransEncodeFormat("GB18030", "UTF-8", filecontent, transcontent);
std::cout << "UTF-8[" << filecontent << "] -> GB18030[" << transcontent << "]" << std::endl;
}
{//GBK 转换到 GB2312格式
std::string filecontent;
std::string transcontent;
std::string utf8filepath = "./test-file/utf-8.txt";
readfile(utf8filepath, filecontent);
//TransStringToBinary(filecontent);
TransEncodeFormat("GB2312", "UTF-8", filecontent, transcontent);
std::cout << "UTF-8[" << filecontent << "] -> GB2312[" << transcontent << "]" << std::endl;
}
std::cout << "\n\n\n" << std::endl;
std::cout << "GBK GB18030 GB2312 -> UTF-8" << std::endl;
{//GBK 转换到 utf-8
std::string filecontent;
std::string transcontent;
std::string gbkfilepath = "./test-file/gbk.txt";
readfile(gbkfilepath, filecontent);
TransEncodeFormat("UTF-8", "GBK", filecontent, transcontent);
std::cout << "GBK[" << filecontent << "] -> UTF-8[" << transcontent << "]" << std::endl;
}
{//GB2312 转换到utf-8
std::string filecontent;
std::string transcontent;
std::string gb2312filepath = "./test-file/gb2312.txt";
readfile(gb2312filepath, filecontent);
TransEncodeFormat("UTF-8", "GB2312", filecontent, transcontent);
std::cout << "GB2312[" << filecontent << "] -> UTF-8[" << transcontent << "]" << std::endl;
}
{//GB18030 转换到utf-8
std::string filecontent;
std::string transcontent;
std::string gb18030filepath = "./test-file/gb18030.txt";
readfile(gb18030filepath, filecontent);
TransEncodeFormat("UTF-8", "GB18030", filecontent, transcontent);
std::cout << "GB18030[" << filecontent << "] -> UTF-8[" << transcontent << "]" << std::endl;
}
return 0;
} TransCore函数是从网上抄的,自己又整理了一下,这个函数是核心,是关键。
其实我自己也仿照人家写了一个,但是有问题,转换成功了,但是获取不到转换内容,还是自己学艺不精啊,我想这和指针方面有关系,以后还得研究研究【TDOO】
int covertex(const char* desc, const char* src, const std::string& _strin, std::string& _strout)
{
std::cout << "======= covertex start =========" << std::endl;
size_t inlen = _strin.length();
size_t outlen = 255;
const char* tempin = _strin.c_str();
const char** pin = &tempin;
char* tempout = new char[outlen];
memset(tempout, '\0', outlen);
char** pout = &tempout;
iconv_t cd = iconv_open(desc, src);
if (cd == (iconv_t)-1)
{
std::cout <<"iconv_open error" << std::endl;
return -1;
}
int ret = iconv(cd, pin, &inlen, pout, &outlen);
std::cout << "iconv ret = " << ret << std::endl;
std::cout << "tempout 地址 = " << &tempout << std::endl;
std::cout << "tempout = " << tempout << std::endl;
std::cout << "*tempout = " << *tempout << std::endl;
std::cout << "pout = " << pout << std::endl;
std::cout << "*pout = " << *pout << std::endl;
std::cout << "**pout = " << **pout << std::endl;
if (tempout != nullptr) {
//delete[] tempout; 一执行就崩溃了
}
else {
std::cout << "tempout == nullptr" << std::endl;
}
iconv_close(cd);
std::cout << "======= covertex end =========" << std::endl;
return 0;
}
/*
UTF8 -> GBK GB18030 GB2312
======= covertex start =========
iconv ret = 0
tempout 地址 = 0079FDA4
tempout =
*tempout =
pout = 0079FDA4
*pout =
**pout =
======= covertex end =========
UTF-8[澶у濂斤紝"1234567890" ABCDEFGHIJKLMNOPQRSTUVWXYZ|abcdefghijklmnopqrstuvwxyz ] -> GBK[]
*/