unicode_iterator.hpp

C++的一个好库。。。现在很流行

      dereference()const
   {
      if(m_current == 4)
         extract_current();
      return m_values[m_current];
   }
   bool equal(const u32_to_u8_iterator& that)const
   {
      if(m_position == that.m_position)
      {
         // either the m_current's must be equal, or one must be 0 and 
         // the other 4: which means neither must have bits 1 or 2 set:
         return (m_current == that.m_current)
            || (((m_current | that.m_current) & 3) == 0);
      }
      return false;
   }
   void increment()
   {
      // if we have a pending read then read now, so that we know whether
      // to skip a position, or move to a low-surrogate:
      if(m_current == 4)
      {
         // pending read:
         extract_current();
      }
      // move to the next surrogate position:
      ++m_current;
      // if we've reached the end skip a position:
      if(m_values[m_current] == 0)
      {
         m_current = 4;
         ++m_position;
      }
   }
   void decrement()
   {
      if((m_current & 3) == 0)
      {
         --m_position;
         extract_current();
         m_current = 3;
         while(m_current && (m_values[m_current] == 0))
            --m_current;
      }
      else
         --m_current;
   }
   BaseIterator base()const
   {
      return m_position;
   }
   // construct:
   u32_to_u8_iterator() : m_position(), m_current(0)
   {
      m_values[0] = 0;
      m_values[1] = 0;
      m_values[2] = 0;
      m_values[3] = 0;
      m_values[4] = 0;
   }
   u32_to_u8_iterator(BaseIterator b) : m_position(b), m_current(4)
   {
      m_values[0] = 0;
      m_values[1] = 0;
      m_values[2] = 0;
      m_values[3] = 0;
      m_values[4] = 0;
   }
private:

   void extract_current()const
   {
      boost::uint32_t c = *m_position;
      if(c > 0x10FFFFu)
         detail::invalid_utf32_code_point(c);
      if(c < 0x80u)
      {
         m_values[0] = static_cast<unsigned char>(c);
         m_values[1] = static_cast<unsigned char>(0u);
         m_values[2] = static_cast<unsigned char>(0u);
         m_values[3] = static_cast<unsigned char>(0u);
      }
      else if(c < 0x800u)
      {
         m_values[0] = static_cast<unsigned char>(0xC0u + (c >> 6));
         m_values[1] = static_cast<unsigned char>(0x80u + (c & 0x3Fu));
         m_values[2] = static_cast<unsigned char>(0u);
         m_values[3] = static_cast<unsigned char>(0u);
      }
      else if(c < 0x10000u)
      {
         m_values[0] = static_cast<unsigned char>(0xE0u + (c >> 12));
         m_values[1] = static_cast<unsigned char>(0x80u + ((c >> 6) & 0x3Fu));
         m_values[2] = static_cast<unsigned char>(0x80u + (c & 0x3Fu));
         m_values[3] = static_cast<unsigned char>(0u);
      }
      else
      {
         m_values[0] = static_cast<unsigned char>(0xF0u + (c >> 18));
         m_values[1] = static_cast<unsigned char>(0x80u + ((c >> 12) & 0x3Fu));
         m_values[2] = static_cast<unsigned char>(0x80u + ((c >> 6) & 0x3Fu));
         m_values[3] = static_cast<unsigned char>(0x80u + (c & 0x3Fu));
      }
      m_current= 0;
   }
   BaseIterator m_position;
   mutable U8Type m_values[5];
   mutable unsigned m_current;
};

template <class BaseIterator, class U32Type = ::boost::uint32_t>
class u8_to_u32_iterator
   : public boost::iterator_facade<u8_to_u32_iterator<BaseIterator, U32Type>, U32Type, std::bidirectional_iterator_tag, const U32Type>
{
   typedef boost::iterator_facade<u8_to_u32_iterator<BaseIterator, U32Type>, U32Type, std::bidirectional_iterator_tag, const U32Type> base_type;
   // special values for pending iterator reads:
   BOOST_STATIC_CONSTANT(U32Type, pending_read = 0xffffffffu);

#if !defined(BOOST_NO_STD_ITERATOR_TRAITS) && !defined(BOOST_NO_TEMPLATE_PARTIAL_SPECIALIZATION)
   typedef typename std::iterator_traits<BaseIterator>::value_type base_value_type;

   BOOST_STATIC_ASSERT(sizeof(base_value_type)*CHAR_BIT == 8);
   BOOST_STATIC_ASSERT(sizeof(U32Type)*CHAR_BIT == 32);
#endif

public:
   typename base_type::reference
      dereference()const
   {
      if(m_value == pending_read)
         extract_current();
      return m_value;
   }
   bool equal(const u8_to_u32_iterator& that)const
   {
      return m_position == that.m_position;
   }
   void increment()
   {
      // skip high surrogate first if there is one:
      unsigned c = detail::utf8_byte_count(*m_position);
      std::advance(m_position, c);
      m_value = pending_read;
   }
   void decrement()
   {
      // Keep backtracking until we don't have a trailing character:
      unsigned count = 0;
      while((*--m_position & 0xC0u) == 0x80u) ++count;
      // now check that the sequence was valid:
      if(count != detail::utf8_trailing_byte_count(*m_position))
         invalid_sequnce();
      m_value = pending_read;
   }
   BaseIterator base()const
   {
      return m_position;
   }
   // construct:
   u8_to_u32_iterator() : m_position()
   {
      m_value = pending_read;
   }
   u8_to_u32_iterator(BaseIterator b) : m_position(b)
   {
      m_value = pending_read;
   }
private:
   static void invalid_sequnce()
   {
      std::out_of_range e("Invalid UTF-8 sequence encountered while trying to encode UTF-32 character");
      boost::throw_exception(e);
   }
   void extract_current()const
   {
      m_value = static_cast<U32Type>(static_cast< ::boost::uint8_t>(*m_position));
      // we must not have a continuation character:
      if((m_value & 0xC0u) == 0x80u)
         invalid_sequnce();
      // see how many extra byts we have:
      unsigned extra = detail::utf8_trailing_byte_count(*m_position);
      // extract the extra bits, 6 from each extra byte:
      BaseIterator next(m_position);
      for(unsigned c = 0; c < extra; ++c)
      {
         ++next;
         m_value <<= 6;
         m_value += static_cast<boost::uint8_t>(*next) & 0x3Fu;
      }
      // we now need to remove a few of the leftmost bits, but how many depends
      // upon how many extra bytes we've extracted:
      static const boost::uint32_t masks[4] = 
      {
         0x7Fu,
         0x7FFu,
         0xFFFFu,
         0x1FFFFFu,
      };
      m_value &= masks[extra];
      // check the result:
      if(m_value > static_cast<U32Type>(0x10FFFFu))
         invalid_sequnce();
   }
   BaseIterator m_position;
   mutable U32Type m_value;
};

template <class BaseIterator>
class utf16_output_iterator
{
public:
   typedef void                                   difference_type;
   typedef void                                   value_type;
   typedef boost::uint32_t*                       pointer;
   typedef boost::uint32_t&                       reference;
   typedef std::output_iterator_tag               iterator_category;

   utf16_output_iterator(const BaseIterator& b)
      : m_position(b){}
   utf16_output_iterator(const utf16_output_iterator& that)
      : m_position(that.m_position){}
   utf16_output_iterator& operator=(const utf16_output_iterator& that)
   {
      m_position = that.m_position;
      return *this;
   }
   const utf16_output_iterator& operator*()const
   {
      return *this;
   }
   void operator=(boost::uint32_t val)const
   {
      push(val);
   }
   utf16_output_iterator& operator++()
   {
      return *this;
   }
   utf16_output_iterator& operator++(int)
   {
      return *this;
   }
   BaseIterator base()const
   {
      return m_position;
   }
private:
   void push(boost::uint32_t v)const
   {
      if(v >= 0x10000u)
      {
         // begin by checking for a code point out of range:
         if(v > 0x10FFFFu)
            detail::invalid_utf32_code_point(v);
         // split into two surrogates:
         *m_position++ = static_cast<boost::uint16_t>(v >> 10) + detail::high_surrogate_base;
         *m_position++ = static_cast<boost::uint16_t>(v & detail::ten_bit_mask) + detail::low_surrogate_base;
      }
      else
      {
         // 16-bit code point:
         // value must not be a surrogate:
         if(detail::is_surrogate(v))
            detail::invalid_utf32_code_point(v);
         *m_position++ = static_cast<boost::uint16_t>(v);
      }
   }
   mutable BaseIterator m_position;
};

template <class BaseIterator>
class utf8_output_iterator
{
public:
   typedef void                                   difference_type;
   typedef void                                   value_type;
   typedef boost::uint32_t*                       pointer;
   typedef boost::uint32_t&                       reference;
   typedef std::output_iterator_tag               iterator_category;

   utf8_output_iterator(const BaseIterator& b)
      : m_position(b){}
   utf8_output_iterator(const utf8_output_iterator& that)
      : m_position(that.m_position){}
   utf8_output_iterator& operator=(const utf8_output_iterator& that)
   {
      m_position = that.m_position;
      return *this;
   }
   const utf8_output_iterator& operator*()const
   {
      return *this;
   }
   void operator=(boost::uint32_t val)const
   {
      push(val);
   }
   utf8_output_iterator& operator++()
   {
      return *this;
   }
   utf8_output_iterator& operator++(int)
   {
      return *this;
   }
   BaseIterator base()const
   {
      return m_position;
   }
private:
   void push(boost::uint32_t c)const
   {
      if(c > 0x10FFFFu)
         detail::invalid_utf32_code_point(c);
      if(c < 0x80u)
      {
         *m_position++ = static_cast<unsigned char>(c);
      }
      else if(c < 0x800u)
      {
         *m_position++ = static_cast<unsigned char>(0xC0u + (c >> 6));
         *m_position++ = static_cast<unsigned char>(0x80u + (c & 0x3Fu));
      }
      else if(c < 0x10000u)
      {
         *m_position++ = static_cast<unsigned char>(0xE0u + (c >> 12));
         *m_position++ = static_cast<unsigned char>(0x80u + ((c >> 6) & 0x3Fu));
         *m_position++ = static_cast<unsigned char>(0x80u + (c & 0x3Fu));
      }
      else
      {
         *m_position++ = static_cast<unsigned char>(0xF0u + (c >> 18));
         *m_position++ = static_cast<unsigned char>(0x80u + ((c >> 12) & 0x3Fu));
         *m_position++ = static_cast<unsigned char>(0x80u + ((c >> 6) & 0x3Fu));
         *m_position++ = static_cast<unsigned char>(0x80u + (c & 0x3Fu));
      }
   }
   mutable BaseIterator m_position;
};

} // namespace boost

#endif // BOOST_REGEX_UNICODE_ITERATOR_HPP