ACE/ACEXML/common/Encoding.cpp

   1 // -*- C++ -*-
   2
   3 #include "ACEXML/common/Encoding.h"
   4 #include "ace/OS_NS_string.h"
   5
   6 const ACEXML_Char* ACEXML_Encoding::encoding_names_[8] = {
   7   ACE_TEXT ("UCS-4BE"),
   8   ACE_TEXT ("UCS-4LE"),
   9   ACE_TEXT ("UCS-4_2143"),
  10   ACE_TEXT ("UCS-4_3412"),
  11   ACE_TEXT ("UTF-16"),
  12   ACE_TEXT ("UTF-16"),
  13   ACE_TEXT ("UTF-8"),
  14   ACE_TEXT ("Unsupported Encoding")
  15 };
  16
  17 const ACEXML_UTF8 ACEXML_Encoding::byte_order_mark_[][4] = {
  18   { '\x00', '\x00', '\xFE', '\xFF' }, // UCS-4, big-endian  (1234 order)
  19   { '\xFF', '\xFE', '\x00', '\x00' }, // UCS-4, little-endian  (4321 order)
  20   { '\x00', '\x00', '\xFF', '\xFE' }, // UCS-4, unusual octet order (2143)
  21   { '\xFE', '\xFF', '\x00', '\x00' }, // UCS-4, unusual octet order (3412)
  22   { '\xFE', '\xFF', '\xFF', '\xFF' }, // UTF-16, big-endian (3 & 4 != 0)
  23   { '\xFF', '\xFE', '\xFF', '\xFF' }, // UTF-16, little-endian ( 3 & 4 != 0)
  24   { '\xEF', '\xBB', '\xBF', '\xFF' }  // UTF-8
  25 };
  26
  27 const ACEXML_UTF8 ACEXML_Encoding::magic_values_[][4] = {
  28   { '\x00', '\x00', '\x00', '\x3c' }, //
  29   { '\x3c', '\x00', '\x00', '\x00' }, // UCS-4 and variants
  30   { '\x00', '\x00', '\x3c', '\x00' }, //
  31   { '\x00', '\x3c', '\x00', '\x00' }, //
  32   { '\x00', '\x3c', '\x00', '\x3f' }, // UTF-16BE
  33   { '\x3c', '\x00', '\x3f', '\x00' }, // UTF-16LE
  34   { '\x3c', '\x3f', '\x78', '\x6d' }, // UTF-8
  35 };
  36
  37 const ACEXML_Char*
  38 ACEXML_Encoding::get_encoding (const char* input)
  39 {
  40   if ((ACE_OS::memcmp (&ACEXML_Encoding::byte_order_mark_[ACEXML_Encoding::UTF16BE][0], input, 2) == 0)
  41       && (input[2] != 0 || input[3] != 0)) // 3 & 4 should not be both zero
  42     return ACEXML_Encoding::encoding_names_[ACEXML_Encoding::UTF16BE];
  43   else if ((ACE_OS::memcmp (&ACEXML_Encoding::byte_order_mark_[ACEXML_Encoding::UTF16LE][0], input, 2) == 0)
  44     && (input[2] != 0 || input[3] != 0)) // 3 & 4 should not be both zero
  45     return ACEXML_Encoding::encoding_names_[ACEXML_Encoding::UTF16LE];
  46   else if (ACE_OS::memcmp (&ACEXML_Encoding::byte_order_mark_[ACEXML_Encoding::UTF8][0], input, 3) == 0)
  47     return ACEXML_Encoding::encoding_names_[ACEXML_Encoding::UTF8];
  48   else if (ACE_OS::memcmp (&ACEXML_Encoding::magic_values_[ACEXML_Encoding::UTF16BE][0], input, 4) == 0)
  49     return ACEXML_Encoding::encoding_names_[ACEXML_Encoding::UTF16BE];
  50   else if (ACE_OS::memcmp (&ACEXML_Encoding::magic_values_[ACEXML_Encoding::UTF16LE][0], input, 4) == 0)
  51     return ACEXML_Encoding::encoding_names_[ACEXML_Encoding::UTF16LE];
  52   else if (ACE_OS::memcmp (&ACEXML_Encoding::magic_values_[ACEXML_Encoding::UTF8][0], input, 4) == 0)
  53     return ACEXML_Encoding::encoding_names_[ACEXML_Encoding::UTF8];
  54   else
  55     {
  56       // ACE_ERROR ((LM_ERROR, "Unknown encoding. Assuming UTF-8\n"));
  57       return ACEXML_Encoding::encoding_names_[ACEXML_Encoding::UTF8];
  58     }
  59 }