ninxit.blog

#include <stdio.h>
#include <string.h>
 
int strlen_utf8( const char *buff )
{
	if( buff == NULL ) return 0;
 
	int count = 0;
	int pos = 0;
	int max_bytes = strlen( buff );
 
	// BOM 読み飛ばし
	if( max_bytes >= 3 )
	{
		if( static_cast<unsigned char>( buff[0] ) == 0xEF &&
			static_cast<unsigned char>( buff[1] ) == 0xBB &&
			static_cast<unsigned char>( buff[2] ) == 0xBF )
		{
			pos += 3;
		}
	}
 
	while( pos < max_bytes )
	{
		++count; // 文字数カウント
 
		if( ( buff[pos] & 0x80 ) == 0 )
		{
			++pos; // 1バイト文字
		}
		else
		{
			for( char tmp = buff[pos] & 0xfc; (tmp & 0x80); tmp = tmp << 1 )
			{
				++pos; // 複数バイト文字
			}
		}
	}
	return count;
}
 
int main()
{
	char *str = "あ過ｻタな";
	int cnt = strlen_utf8( str );
 
	printf( "String: %s\n", str );
	printf( "Count: %d\n", cnt );
 
	return 0;
}

String: あ過ｻタな
Count: 5