utf8_iterator.h

class Utf8Iterator
{
public:
    Utf8Iterator(const char* data, size_t size): mData(data), mSize(size), mPos(0) {}
    ~Utf8Iterator() {}

    bool HasNext() const
    {
        return mPos < mSize;
    }

    size_t Next()
    {
        if (! HasNext()) return mSize;

        mPos += GetWidth();
        return mPos;
    }

    size_t Next(size_t utf8CharCount)
    {
        for (size_t i = 0; i < utf8CharCount && HasNext(); ++ i)
        {
            Next();
        }
        return mPos;
    }
  
    size_t GetWidth() const
    {
        if (! HasNext()) return 0;

        if (Is4ByteUtf8CodePoint())
        {
            return 4;
        }
        else if (Is3ByteUtf8CodePoint())
        {
            return 3;
        }
        else if (Is2ByteUtf8CodePoint())
        {
            return 2;
        }
        else
        {
            return 1;
        }
    }

    const char* GetCodePoint() const
    {
        return mData + mPos;
    }

    std::string GetChar() const
    {
        return std::string(GetCodePoint(), GetWidth());
    }

private:
    bool Is1ByteUtf8CodePoint() const { return 0x00 == (0x80 & *GetCodePoint()); }
    bool Is2ByteUtf8CodePoint() const { return 0xc0 == (0xe0 & *GetCodePoint()); }
    bool Is3ByteUtf8CodePoint() const { return 0xe0 == (0xf0 & *GetCodePoint()); }
    bool Is4ByteUtf8CodePoint() const { return 0xf0 == (0xf8 & *GetCodePoint()); }

private:
    const char* mData;
    size_t mSize;
    size_t mPos;
};