ohyecloudy's programming notes

Unicode 문자 개수 구하기.

UTF-16 문자 개수 구하기.

int		GetUtf16GlyphCount(const wstring& utf16)
{
	static const wchar_t	HIGH_SURROGATE_MIN = L'\xD800';
	static const wchar_t	HIGH_SURROGATE_MAX = L'\xDBFF';
	static const wchar_t	LOW_SURROGATE_MIN = L'\xDC00';
	static const wchar_t	LOW_SURROGATE_MAX = L'\xDFFF';

	bool	bHighSurrogate = false;
	int		count = 0;
	for( wstring::const_iterator iter = utf16.begin(); iter != utf16.end(); ++iter )
	{
		const wchar_t& ch = *iter;

		if( HIGH_SURROGATE_MIN <= ch && ch <= HIGH_SURROGATE_MAX )
		{
			assert( !bHighSurrogate );
			bHighSurrogate = true;
		}
		else if( LOW_SURROGATE_MIN <= ch && ch <= LOW_SURROGATE_MAX )
		{
			assert( bHighSurrogate );
			bHighSurrogate = false;

			++count;
			assert( count < numeric_limits<int>::max() );
		}
		else
		{
			assert( !bHighSurrogate );
			++count;
		}
	}

	return count;
}

Continue reading »