src/support/unicode.C

   1 /**
   2  * \file unicode.C
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author Lars Gullik Bjønnes
   7  *
   8  * Full author contact details are available in file CREDITS.
   9  *
  10  * A collection of unicode conversion functions, using iconv.
  11  */
  12
  13 #include <config.h>
  14
  15 #include "unicode.h"
  16
  17 #include "debug.h"
  18
  19 #include <iconv.h>
  20
  21 #include <cerrno>
  22 #include <iomanip>
  23 #include <map>
  24
  25 using std::endl;
  26
  27 namespace {
  28
  29 #ifdef WORDS_BIGENDIAN
  30         char const * utf16_codeset = "UTF16-BE";
  31 #else
  32         char const * utf16_codeset = "UTF16-LE";
  33 #endif
  34
  35 }
  36
  37
  38 namespace lyx {
  39
  40 #ifdef WORDS_BIGENDIAN
  41         char const * ucs4_codeset = "UCS-4BE";
  42 #else
  43         char const * ucs4_codeset = "UCS-4LE";
  44 #endif
  45
  46 static const iconv_t invalid_cd = (iconv_t)(-1);
  47
  48
  49 struct IconvProcessor::Private {
  50         Private(): cd(invalid_cd) {}
  51         ~Private()
  52         {
  53                 if (cd != invalid_cd) {
  54                         if (iconv_close(cd) == -1) {
  55                                 lyxerr << "Error returned from iconv_close("
  56                                        << errno << ")" << endl;
  57                         }
  58                 }
  59         }
  60         iconv_t cd;
  61 };
  62
  63
  64 IconvProcessor::IconvProcessor(char const * tocode,
  65                 char const * fromcode): tocode_(tocode), fromcode_(fromcode),
  66                 pimpl_(new IconvProcessor::Private)
  67 {
  68 }
  69
  70
  71 IconvProcessor::IconvProcessor(IconvProcessor const & other)
  72         : tocode_(other.tocode_), fromcode_(other.fromcode_),
  73           pimpl_(new IconvProcessor::Private)
  74 {
  75 }
  76
  77
  78 IconvProcessor & IconvProcessor::operator=(IconvProcessor const & other)
  79 {
  80         if (&other == this)
  81                 return *this;
  82         tocode_ = other.tocode_;
  83         fromcode_ = other.fromcode_;
  84         pimpl_.reset(new Private);
  85         return *this;
  86 }
  87
  88
  89 IconvProcessor::~IconvProcessor() {}
  90
  91
  92 bool IconvProcessor::init()
  93 {
  94         if (pimpl_->cd != invalid_cd)
  95                 return true;
  96
  97         pimpl_->cd = iconv_open(tocode_.c_str(), fromcode_.c_str());
  98         if (pimpl_->cd != invalid_cd)
  99                 return true;
 100
 101         lyxerr << "Error returned from iconv_open" << endl;
 102         switch (errno) {
 103                 case EINVAL:
 104                         lyxerr << "EINVAL The conversion from " << fromcode_
 105                                 << " to " << tocode_
 106                                 << " is not supported by the implementation."
 107                                 << endl;
 108                         break;
 109                 default:
 110                         lyxerr << "\tSome other error: " << errno << endl;
 111                         break;
 112         }
 113         return false;
 114 }
 115
 116
 117 int IconvProcessor::convert(char const * buf, size_t buflen,
 118                 char * outbuf, size_t maxoutsize)
 119 {
 120         if (buflen == 0)
 121                 return 0;
 122
 123         if (pimpl_->cd == invalid_cd) {
 124                 if (!init())
 125                         return -1;
 126         }
 127
 128         char ICONV_CONST * inbuf = const_cast<char ICONV_CONST *>(buf);
 129         size_t inbytesleft = buflen;
 130         size_t outbytesleft = maxoutsize;
 131
 132         int res = iconv(pimpl_->cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
 133
 134         //lyxerr << std::dec;
 135         //lyxerr << "Inbytesleft: " << inbytesleft << endl;
 136         //lyxerr << "Outbytesleft: " << outbytesleft << endl;
 137
 138         if (res != -1)
 139                 // Everything went well.
 140                 return maxoutsize - outbytesleft;
 141
 142         // There are some errors in the conversion
 143         lyxerr << "Error returned from iconv" << endl;
 144         switch (errno) {
 145                 case E2BIG:
 146                         lyxerr << "E2BIG  There is not sufficient room at *outbuf." << endl;
 147                         break;
 148                 case EILSEQ:
 149                         lyxerr << "EILSEQ An invalid multibyte sequence"
 150                                 << " has been encountered in the input.\n"
 151                                 << "When converting from " << fromcode_
 152                                 << " to " << tocode_ << ".\n";
 153                         lyxerr << "Input: " << std::hex;
 154                         for (size_t i = 0; i < buflen; ++i) {
 155                                 boost::uint32_t const b = buf[i];
 156                                 lyxerr << "0x" << b << " ";
 157                         }
 158                         lyxerr << endl;
 159                         break;
 160                 case EINVAL:
 161                         lyxerr << "EINVAL An incomplete multibyte sequence"
 162                                 << " has been encountered in the input.\n"
 163                                 << "When converting from " << fromcode_
 164                                 << " to " << tocode_ << ".\n";
 165                         lyxerr << "Input: " << std::hex;
 166                         for (size_t i = 0; i < buflen; ++i) {
 167                                 boost::uint32_t const b = buf[i];
 168                                 lyxerr << "0x" << b << " ";
 169                         }
 170                         lyxerr << endl;
 171                         break;
 172                 default:
 173                         lyxerr << "\tSome other error: " << errno << endl;
 174                         break;
 175         }
 176         // We got an error so we close down the conversion engine
 177         if (iconv_close(pimpl_->cd) == -1) {
 178                 lyxerr << "Error returned from iconv_close("
 179                         << errno << ")" << endl;
 180         }
 181         pimpl_->cd = invalid_cd;
 182         return -1;
 183 }
 184
 185
 186 namespace {
 187
 188
 189 template<typename RetType, typename InType>
 190 std::vector<RetType>
 191 iconv_convert(IconvProcessor & processor,
 192               InType const * buf,
 193               size_t buflen)
 194 {
 195         if (buflen == 0)
 196                 return std::vector<RetType>();
 197
 198         char const * inbuf = reinterpret_cast<char const *>(buf);
 199         size_t inbytesleft = buflen * sizeof(InType);
 200
 201         size_t const outsize = 32768;
 202         static char out[outsize];
 203         char * outbuf = out;
 204
 205         int bytes = processor.convert(inbuf, inbytesleft, outbuf, outsize);
 206
 207         RetType const * tmp = reinterpret_cast<RetType const *>(out);
 208         return std::vector<RetType>(tmp, tmp + bytes / sizeof(RetType));
 209 }
 210
 211 } // anon namespace
 212
 213
 214 std::vector<lyx::char_type> utf8_to_ucs4(std::vector<char> const & utf8str)
 215 {
 216         if (utf8str.empty())
 217                 return std::vector<lyx::char_type>();
 218
 219         return utf8_to_ucs4(&utf8str[0], utf8str.size());
 220 }
 221
 222
 223 std::vector<lyx::char_type>
 224 utf8_to_ucs4(char const * utf8str, size_t ls)
 225 {
 226         static IconvProcessor processor(ucs4_codeset, "UTF-8");
 227         return iconv_convert<lyx::char_type>(processor, utf8str, ls);
 228 }
 229
 230
 231 std::vector<char_type>
 232 utf16_to_ucs4(unsigned short const * s, size_t ls)
 233 {
 234         static IconvProcessor processor(ucs4_codeset, utf16_codeset);
 235         return iconv_convert<char_type>(processor, s, ls);
 236 }
 237
 238
 239 std::vector<unsigned short>
 240 ucs4_to_utf16(char_type const * s, size_t ls)
 241 {
 242         static IconvProcessor processor(utf16_codeset, ucs4_codeset);
 243         return iconv_convert<unsigned short>(processor, s, ls);
 244 }
 245
 246
 247 std::vector<char>
 248 ucs4_to_utf8(lyx::char_type c)
 249 {
 250         static IconvProcessor processor("UTF-8", ucs4_codeset);
 251         return iconv_convert<char>(processor, &c, 1);
 252 }
 253
 254
 255 std::vector<char>
 256 ucs4_to_utf8(std::vector<lyx::char_type> const & ucs4str)
 257 {
 258         if (ucs4str.empty())
 259                 return std::vector<char>();
 260
 261         return ucs4_to_utf8(&ucs4str[0], ucs4str.size());
 262 }
 263
 264
 265 std::vector<char>
 266 ucs4_to_utf8(lyx::char_type const * ucs4str, size_t ls)
 267 {
 268         static IconvProcessor processor("UTF-8", ucs4_codeset);
 269         return iconv_convert<char>(processor, ucs4str, ls);
 270 }
 271
 272
 273 std::vector<lyx::char_type>
 274 eightbit_to_ucs4(char const * s, size_t ls, std::string const & encoding)
 275 {
 276         static std::map<std::string, IconvProcessor> processors;
 277         if (processors.find(encoding) == processors.end()) {
 278                 IconvProcessor processor(ucs4_codeset, encoding.c_str());
 279                 processors.insert(std::make_pair(encoding, processor));
 280         }
 281         return iconv_convert<char_type>(processors[encoding], s, ls);
 282 }
 283
 284
 285 std::vector<char>
 286 ucs4_to_eightbit(lyx::char_type const * ucs4str, size_t ls, std::string const & encoding)
 287 {
 288         static std::map<std::string, IconvProcessor> processors;
 289         if (processors.find(encoding) == processors.end()) {
 290                 IconvProcessor processor(encoding.c_str(), ucs4_codeset);
 291                 processors.insert(std::make_pair(encoding, processor));
 292         }
 293         return iconv_convert<char>(processors[encoding], ucs4str, ls);
 294 }
 295
 296 } // namespace lyx