zoukankan      html  css  js  c++  java
  • 使用c++11标准库转换字符编码

     
    #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
  • 相关阅读:
    联通手机号停机保号了,想恢复要短信验证码登陆但是无法接收短信验证码怎么办
    记卖饭让我先吃
    POJ 3658 Artificial Lake
    POJ 3662 Telephone Lines (dijstra+二分)
    CodeForces 748C Santa Claus and Robot
    CodeForces 748B Santa Claus and Keyboard Check
    POJ 3659 Cell Phone Network(树形dp树的最小点支配集)
    【JZOJ 5455】拆网线 【树形DP】
    【JZOJ 5455】拆网线 【树形DP】
    【JZOJ 5455】拆网线 【树形DP】
  • 原文地址:https://www.cnblogs.com/LinuxHunter/p/2848293.html
Copyright © 2011-2022 走看看