regression_test/iconv/iconver.cc

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Copyright (c) 2000-2014 Ericsson Telecom AB
   3 // All rights reserved. This program and the accompanying materials
   4 // are made available under the terms of the Eclipse Public License v1.0
   5 // which accompanies this distribution, and is available at
   6 // http://www.eclipse.org/legal/epl-v10.html
   7 ///////////////////////////////////////////////////////////////////////////////
   8 #include <TTCN3.hh>
   9 #include <iconv.h>
  10 #include <errno.h>
  11 #include <vector>
  12
  13 #if defined(WIN32)
  14 #include <cygwin/version.h>
  15 #endif
  16
  17 #if defined(LINUX) || defined(INTERIX) || (CYGWIN_VERSION_DLL_MAJOR >= 1007)
  18 #define ICONV_INPUT_TYPE char*
  19 #else
  20 #define ICONV_INPUT_TYPE const char*
  21 #endif
  22
  23 static const universal_char ucs4_bom = { 0,0,0xFE,0xFF };
  24
  25 namespace converter
  26 {
  27   // Convert from octetstring in UTF-8 to universal charstring
  28   INTEGER o2u(const CHARSTRING& encoding, const OCTETSTRING& input, UNIVERSAL_CHARSTRING& output)
  29   {
  30     errno = 0;
  31     iconv_t t = iconv_open("UCS-4BE", (const char*)encoding); // to, from
  32     if ((iconv_t)-1 == t) {
  33       perror(encoding);
  34       return errno;
  35     }
  36     size_t inbytesleft = input.lengthof();
  37     ICONV_INPUT_TYPE inptr = (ICONV_INPUT_TYPE)(const unsigned char *)input;
  38
  39     // Scott Meyers, Effective STL, Item 16.
  40     // Don't even think of using Titan's vector like this!
  41     std::vector<universal_char> outbuf(inbytesleft + 1);
  42     // +1 for the BOM that Solaris iconv likes to put in at the front
  43     char* outptr = (char*)&outbuf.front();
  44     size_t const outbytes =  outbuf.size() * 4; // size is in universal_char
  45     size_t outbytesleft = outbytes; // will be modified by iconv
  46
  47     size_t rez = iconv(t, &inptr, &inbytesleft, &outptr, &outbytesleft);
  48     if (rez == (size_t)-1) {
  49       perror("*** iconv ***"); // oh dear
  50     }
  51     else {
  52       size_t is_bom = ( ucs4_bom == outbuf.front() );
  53       output = UNIVERSAL_CHARSTRING((outbytes - outbytesleft - is_bom) / 4, &outbuf[is_bom]);
  54       errno = 0; // Solaris iconv may set errno to nonzero even after successful conversion :-O
  55     }
  56     iconv_close(t);
  57
  58     return errno;
  59   }
  60
  61   // Convert from universal charstring to octetstring in UTF-8
  62   INTEGER u2o(const CHARSTRING& encoding, const UNIVERSAL_CHARSTRING& input, OCTETSTRING& output)
  63   {
  64     errno = 0;
  65     iconv_t t = iconv_open((const char*)encoding, "UCS-4BE"); // to, from
  66     if ((iconv_t)-1 == t) {
  67       perror(encoding);
  68       return errno;
  69     }
  70
  71     size_t inbytesleft = input.lengthof(); // in characters
  72     size_t to_be = inbytesleft * 6; // max 6 bytes in UTF-8 per character
  73     inbytesleft *= 4; // count in bytes
  74
  75     char* const inbuf = (char*)input.operator const universal_char*();
  76     ICONV_INPUT_TYPE inptr = inbuf; // will be modified by iconv
  77     std::vector<unsigned char> outbuf(to_be | !to_be);
  78     // to_be, or not to_be, that is the question.
  79     // This makes sure that the vector's size is at least 1.
  80     char* outptr = (char*)&outbuf.front();
  81
  82     size_t rez = iconv(t, &inptr, &inbytesleft, &outptr, &to_be);
  83     if (rez == (size_t)-1) {
  84       perror("*** iconv ***"); // oh dear
  85     }
  86     else {
  87       output = OCTETSTRING(outbuf.size() - (to_be | !to_be), &outbuf.front());
  88       errno = 0; // Solaris iconv may set errno to nonzero even after successful conversion :-O
  89     }
  90     iconv_close(t);
  91
  92     return errno;
  93   }
  94
  95 } // namespace
  96
  97 // Drag o2u and u2o out to the global namespace for compatibility with old names.
  98 // Ifdef-ing the namespace away will probably fail when this file is used
  99 // from projects where old naming is permanently disabled.
 100
 101 INTEGER o2u(const CHARSTRING& encoding, const OCTETSTRING& input, UNIVERSAL_CHARSTRING& output) {
 102   return converter::o2u(encoding, input, output);
 103 }
 104
 105 INTEGER u2o(const CHARSTRING& encoding, const UNIVERSAL_CHARSTRING& input, OCTETSTRING& output) {
 106   return converter::u2o(encoding, input, output);
 107 }
 108