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 lyx {
  28
  29 #ifdef WORDS_BIGENDIAN
  30         char const * ucs4_codeset = "UCS-4BE";
  31         char const * ucs2_codeset = "UCS-2BE";
  32 #else
  33         char const * ucs4_codeset = "UCS-4LE";
  34         char const * ucs2_codeset = "UCS-2LE";
  35 #endif
  36
  37 static const iconv_t invalid_cd = (iconv_t)(-1);
  38
  39
  40 struct IconvProcessor::Private {
  41         Private(): cd(invalid_cd) {}
  42         ~Private()
  43         {
  44                 if (cd != invalid_cd) {
  45                         if (iconv_close(cd) == -1) {
  46                                 lyxerr << "Error returned from iconv_close("
  47                                        << errno << ")" << endl;
  48                         }
  49                 }
  50         }
  51         iconv_t cd;
  52 };
  53
  54
  55 IconvProcessor::IconvProcessor(char const * tocode,
  56                 char const * fromcode): tocode_(tocode), fromcode_(fromcode),
  57                 pimpl_(new IconvProcessor::Private)
  58 {
  59 }
  60
  61
  62 IconvProcessor::IconvProcessor(IconvProcessor const & other)
  63         : tocode_(other.tocode_), fromcode_(other.fromcode_),
  64           pimpl_(new IconvProcessor::Private)
  65 {
  66 }
  67
  68
  69 IconvProcessor & IconvProcessor::operator=(IconvProcessor const & other)
  70 {
  71         if (&other == this)
  72                 return *this;
  73         tocode_ = other.tocode_;
  74         fromcode_ = other.fromcode_;
  75         pimpl_.reset(new Private);
  76         return *this;
  77 }
  78
  79
  80 IconvProcessor::~IconvProcessor() {}
  81
  82
  83 bool IconvProcessor::init()
  84 {
  85         if (pimpl_->cd != invalid_cd)
  86                 return true;
  87
  88         pimpl_->cd = iconv_open(tocode_.c_str(), fromcode_.c_str());
  89         if (pimpl_->cd != invalid_cd)
  90                 return true;
  91
  92         lyxerr << "Error returned from iconv_open" << endl;
  93         switch (errno) {
  94                 case EINVAL:
  95                         lyxerr << "EINVAL The conversion from " << fromcode_
  96                                 << " to " << tocode_
  97                                 << " is not supported by the implementation."
  98                                 << endl;
  99                         break;
 100                 default:
 101                         lyxerr << "\tSome other error: " << errno << endl;
 102                         break;
 103         }
 104         return false;
 105 }
 106
 107
 108 int IconvProcessor::convert(char const * buf, size_t buflen,
 109                 char * outbuf, size_t maxoutsize)
 110 {
 111         if (buflen == 0)
 112                 return 0;
 113
 114         if (pimpl_->cd == invalid_cd) {
 115                 if (!init())
 116                         return -1;
 117         }
 118
 119         char ICONV_CONST * inbuf = const_cast<char ICONV_CONST *>(buf);
 120         size_t inbytesleft = buflen;
 121         size_t outbytesleft = maxoutsize;
 122
 123         int res = iconv(pimpl_->cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
 124
 125         //lyxerr << std::dec;
 126         //lyxerr << "Inbytesleft: " << inbytesleft << endl;
 127         //lyxerr << "Outbytesleft: " << outbytesleft << endl;
 128
 129         if (res != -1)
 130                 // Everything went well.
 131                 return maxoutsize - outbytesleft;
 132
 133         // There are some errors in the conversion
 134         lyxerr << "Error returned from iconv" << endl;
 135         switch (errno) {
 136                 case E2BIG:
 137                         lyxerr << "E2BIG  There is not sufficient room at *outbuf." << endl;
 138                         break;
 139                 case EILSEQ:
 140                         lyxerr << "EILSEQ An invalid multibyte sequence"
 141                                 << " has been encountered in the input.\n"
 142                                 << "When converting from " << fromcode_
 143                                 << " to " << tocode_ << ".\n";
 144                         lyxerr << "Input: " << std::hex;
 145                         for (size_t i = 0; i < buflen; ++i) {
 146                                 boost::uint32_t const b = buf[i];
 147                                 lyxerr << "0x" << b << " ";
 148                         }
 149                         lyxerr << endl;
 150                         break;
 151                 case EINVAL:
 152                         lyxerr << "EINVAL An incomplete multibyte sequence"
 153                                 << " has been encountered in the input.\n"
 154                                 << "When converting from " << fromcode_
 155                                 << " to " << tocode_ << ".\n";
 156                         lyxerr << "Input: " << std::hex;
 157                         for (size_t i = 0; i < buflen; ++i) {
 158                                 boost::uint32_t const b = buf[i];
 159                                 lyxerr << "0x" << b << " ";
 160                         }
 161                         lyxerr << endl;
 162                         break;
 163                 default:
 164                         lyxerr << "\tSome other error: " << errno << endl;
 165                         break;
 166         }
 167         // We got an error so we close down the conversion engine
 168         if (iconv_close(pimpl_->cd) == -1) {
 169                 lyxerr << "Error returned from iconv_close("
 170                         << errno << ")" << endl;
 171         }
 172         pimpl_->cd = invalid_cd;
 173         return -1;
 174 }
 175
 176
 177 namespace {
 178
 179
 180 template<typename RetType, typename InType>
 181 std::vector<RetType>
 182 iconv_convert(IconvProcessor & processor,
 183               InType const * buf,
 184               size_t buflen)
 185 {
 186         if (buflen == 0)
 187                 return std::vector<RetType>();
 188
 189         char const * inbuf = reinterpret_cast<char const *>(buf);
 190         size_t inbytesleft = buflen * sizeof(InType);
 191
 192         size_t const outsize = 32768;
 193         static char out[outsize];
 194         char * outbuf = out;
 195
 196         int bytes = processor.convert(inbuf, inbytesleft, outbuf, outsize);
 197
 198         RetType const * tmp = reinterpret_cast<RetType const *>(out);
 199         return std::vector<RetType>(tmp, tmp + bytes / sizeof(RetType));
 200 }
 201
 202 } // anon namespace
 203
 204
 205 std::vector<lyx::char_type> utf8_to_ucs4(std::vector<char> const & utf8str)
 206 {
 207         if (utf8str.empty())
 208                 return std::vector<lyx::char_type>();
 209
 210         return utf8_to_ucs4(&utf8str[0], utf8str.size());
 211 }
 212
 213
 214 std::vector<lyx::char_type>
 215 utf8_to_ucs4(char const * utf8str, size_t ls)
 216 {
 217         static IconvProcessor processor(ucs4_codeset, "UTF-8");
 218         return iconv_convert<lyx::char_type>(processor, utf8str, ls);
 219 }
 220
 221
 222 lyx::char_type
 223 ucs2_to_ucs4(unsigned short c)
 224 {
 225         return ucs2_to_ucs4(&c, 1)[0];
 226 }
 227
 228
 229 std::vector<lyx::char_type>
 230 ucs2_to_ucs4(std::vector<unsigned short> const & ucs2str)
 231 {
 232         if (ucs2str.empty())
 233                 return std::vector<lyx::char_type>();
 234
 235         return ucs2_to_ucs4(&ucs2str[0], ucs2str.size());
 236 }
 237
 238
 239 std::vector<lyx::char_type>
 240 ucs2_to_ucs4(unsigned short const * ucs2str, size_t ls)
 241 {
 242         static IconvProcessor processor(ucs4_codeset, ucs2_codeset);
 243         return iconv_convert<lyx::char_type>(processor, ucs2str, ls);
 244 }
 245
 246
 247 unsigned short
 248 ucs4_to_ucs2(lyx::char_type c)
 249 {
 250         return ucs4_to_ucs2(&c, 1)[0];
 251 }
 252
 253
 254 std::vector<unsigned short>
 255 ucs4_to_ucs2(std::vector<lyx::char_type> const & ucs4str)
 256 {
 257         if (ucs4str.empty())
 258                 return std::vector<unsigned short>();
 259
 260         return ucs4_to_ucs2(&ucs4str[0], ucs4str.size());
 261 }
 262
 263
 264 std::vector<unsigned short>
 265 ucs4_to_ucs2(lyx::char_type const * s, size_t ls)
 266 {
 267         static IconvProcessor processor(ucs2_codeset, ucs4_codeset);
 268         return iconv_convert<unsigned short>(processor, s, ls);
 269 }
 270
 271
 272 std::vector<char>
 273 ucs4_to_utf8(lyx::char_type c)
 274 {
 275         static IconvProcessor processor("UTF-8", ucs4_codeset);
 276         return iconv_convert<char>(processor, &c, 1);
 277 }
 278
 279
 280 std::vector<char>
 281 ucs4_to_utf8(std::vector<lyx::char_type> const & ucs4str)
 282 {
 283         if (ucs4str.empty())
 284                 return std::vector<char>();
 285
 286         return ucs4_to_utf8(&ucs4str[0], ucs4str.size());
 287 }
 288
 289
 290 std::vector<char>
 291 ucs4_to_utf8(lyx::char_type const * ucs4str, size_t ls)
 292 {
 293         static IconvProcessor processor("UTF-8", ucs4_codeset);
 294         return iconv_convert<char>(processor, ucs4str, ls);
 295 }
 296
 297
 298 std::vector<lyx::char_type>
 299 eightbit_to_ucs4(char const * s, size_t ls, std::string const & encoding)
 300 {
 301         static std::map<std::string, IconvProcessor> processors;
 302         if (processors.find(encoding) == processors.end()) {
 303                 IconvProcessor processor(ucs4_codeset, encoding.c_str());
 304                 processors.insert(std::make_pair(encoding, processor));
 305         }
 306         return iconv_convert<char_type>(processors[encoding], s, ls);
 307 }
 308
 309
 310 std::vector<char>
 311 ucs4_to_eightbit(lyx::char_type const * ucs4str, size_t ls, std::string const & encoding)
 312 {
 313         static std::map<std::string, IconvProcessor> processors;
 314         if (processors.find(encoding) == processors.end()) {
 315                 IconvProcessor processor(encoding.c_str(), ucs4_codeset);
 316                 processors.insert(std::make_pair(encoding, processor));
 317         }
 318         return iconv_convert<char>(processors[encoding], ucs4str, ls);
 319 }
 320
 321 } // namespace lyx