ACE/ace/UTF16_Encoding_Converter.h

   1 // -*- C++ -*-
   2
   3 //=========================================================================
   4 /**
   5  * @file UTF16_Encoding_Converter.h
   6  *
   7  * This class contains declarations for methods that convert between
   8  * UTF-16 (both BE and LE) and UTF-8
   9  *
  10  * @author Chad Elliott <elliott_c@ociweb.com>
  11  */
  12 //=========================================================================
  13
  14 #ifndef ACE_UTF16_ENCODING_CONVERTER_H
  15 #define ACE_UTF16_ENCODING_CONVERTER_H
  16
  17 #include /**/ "ace/pre.h"
  18
  19 #include "ace/Encoding_Converter.h"
  20
  21 #if defined (ACE_USES_WCHAR)
  22 ACE_BEGIN_VERSIONED_NAMESPACE_DECL
  23
  24 /** Convert from UTF-16 to UTF-8 and from UTF-8 to UTF-16.
  25  * This class implements the ACE_Encoding_Converter interface.
  26  */
  27 class ACE_UTF16_Encoding_Converter: public ACE_Encoding_Converter
  28 {
  29 public:
  30   /// The swap parameter determines whether we need to swap byte order on
  31   /// the stream as each word is pulled off when converting to UTF-8.
  32   ACE_UTF16_Encoding_Converter (bool swap = false);
  33
  34   /// This is a do nothing destructor.
  35   virtual ~ACE_UTF16_Encoding_Converter ();
  36
  37   /// Convert the source from UTF-16 to UTF-8 and store it in the
  38   /// provided target buffer.
  39   virtual Result to_utf8 (const void* source,
  40                           size_t source_size,
  41                           ACE_Byte* target,
  42                           size_t target_size,
  43                           bool strict = true);
  44
  45   /// Convert the UTF-8 source into a UTF-16 encoding and store it
  46   /// in the provided target buffer.
  47   virtual Result from_utf8 (const ACE_Byte* source,
  48                             size_t source_size,
  49                             void* target,
  50                             size_t target_size,
  51                             bool strict = true);
  52
  53   /// This factory helper method determines if the source stream is UTF-16
  54   /// encoded.  If it is, allocate an ACE_UTF16_Encoding_Converter and
  55   /// return it.  The caller then owns the allocated object.
  56   static ACE_UTF16_Encoding_Converter* encoded (const ACE_Byte* source,
  57                                                 size_t source_size);
  58
  59 protected:
  60   /// Determines if the source buffer is legal UTF-8
  61   bool is_legal_utf8 (const ACE_Byte* source,
  62                       size_t length) const;
  63
  64   static ACE_UINT32 get_UNI_SUR_HIGH_START ();
  65   static ACE_UINT32 get_UNI_SUR_LOW_END ();
  66   static ACE_UINT32 get_UNI_REPLACEMENT_CHAR ();
  67   static const ACE_Byte* get_first_byte_mark ();
  68   static const ACE_Byte* get_trailing_bytes_for_utf8 ();
  69   static const ACE_UINT32* get_offsets_from_utf8 ();
  70
  71   bool swap_;
  72 };
  73
  74 ACE_END_VERSIONED_NAMESPACE_DECL
  75
  76 #if defined (__ACE_INLINE__)
  77 #include "ace/UTF16_Encoding_Converter.inl"
  78 #endif /* __ACE_INLINE__ */
  79
  80 #endif /* ACE_USES_WCHAR */
  81
  82 #include /**/ "ace/post.h"
  83
  84 #endif /* ACE_UTF16_ENCODING_CONVERTER_H */