使用c++11标准库转换字符编码

最新推荐文章于 2024-04-24 07:00:00 发布

转载最新推荐文章于 2024-04-24 07:00:00 发布 · 1.6k 阅读

文章标签：

#C++ #转换 #string #wstring

C/C++ 专栏收录该内容

12 篇文章

订阅专栏

本文提供了一系列用于在不同字符集（如UTF8和宽字符）间进行转换的C++函数实现，包括从宽字符到多字节字符的转换、从多字节字符到宽字符的转换以及UTF8和宽字符之间的相互转换。特别地，文中还提到了对于特定的UTF8字符（例如0X2C0XA0）进行特殊处理的方法。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

原帖地址：http://www.cnblogs.com/LinuxHunter/archive/2013/01/06/2848293.html

#include <stdio.h>
#include <locale>
#include <codecvt>

const std::string ws2s( const std::wstring& src )
{
	std::locale sys_locale("");

	const wchar_t* data_from = src.c_str();
	const wchar_t* data_from_end = src.c_str() + src.size();
	const wchar_t* data_from_next = 0;

	int wchar_size = 4;
	char* data_to = new char[(src.size() + 1) * wchar_size];
	char* data_to_end = data_to + (src.size() + 1) * wchar_size;
	char* data_to_next = 0;

	memset( data_to, 0, (src.size() + 1) * wchar_size );

	typedef std::codecvt<wchar_t, char, mbstate_t> convert_facet;
	mbstate_t out_state = 0;
	auto result = std::use_facet<convert_facet>(sys_locale).out(
		out_state, data_from, data_from_end, data_from_next,
		data_to, data_to_end, data_to_next );
	if( result == convert_facet::ok )
	{
		std::string dst = data_to;
		delete[] data_to;
		return dst;
	}
	else
	{
		printf( "convert error!\n" );
		delete[] data_to;
		return std::string("");
	}
}

const std::wstring s2ws( const std::string& src )
{
	std::locale sys_locale("");
		
	const char* data_from = src.c_str();
	const char* data_from_end = src.c_str() + src.size();
	const char* data_from_next = 0;

	wchar_t* data_to = new wchar_t[src.size() + 1];
	wchar_t* data_to_end = data_to + src.size() + 1;
	wchar_t* data_to_next = 0;

	wmemset( data_to, 0, src.size() + 1 );

	typedef std::codecvt<wchar_t, char, mbstate_t> convert_facet;
	mbstate_t in_state = 0;
	auto result = std::use_facet<convert_facet>(sys_locale).in(
		in_state, data_from, data_from_end, data_from_next,
		data_to, data_to_end, data_to_next );
	if( result == convert_facet::ok )
	{
		std::wstring dst = data_to;
		delete[] data_to;
		return dst;
	}
	else
	{
		printf( "convert error!\n" );
		delete[] data_to;
		return std::wstring(L"");
	}
}

const std::string ws2utf8( const std::wstring& src )
{
	std::wstring_convert<std::codecvt_utf8<wchar_t>> conv;
	return conv.to_bytes( src );
}

const std::wstring utf8_2_ws( const std::string& src )
{
	std::wstring_convert<std::codecvt_utf8<wchar_t> > conv;
	return conv.from_bytes( src );
}

遇到0X2C 0XA0的UTF8字符编码进行特殊处理，转换成普通空格。