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                                 // char may be signed, avoid output of
 156                                 // something like 0xffffffc2
 157                                 boost::uint32_t const b =
 158                                         *reinterpret_cast<unsigned char const *>(buf + i);
 159                                 lyxerr << " 0x" << b;
 160                         }
 161                         lyxerr << endl;
 162                         break;
 163                 case EINVAL:
 164                         lyxerr << "EINVAL An incomplete multibyte sequence"
 165                                 << " has been encountered in the input.\n"
 166                                 << "When converting from " << fromcode_
 167                                 << " to " << tocode_ << ".\n";
 168                         lyxerr << "Input:" << std::hex;
 169                         for (size_t i = 0; i < buflen; ++i) {
 170                                 // char may be signed, avoid output of
 171                                 // something like 0xffffffc2
 172                                 boost::uint32_t const b =
 173                                         *reinterpret_cast<unsigned char const *>(buf + i);
 174                                 lyxerr << " 0x" << b;
 175                         }
 176                         lyxerr << endl;
 177                         break;
 178                 default:
 179                         lyxerr << "\tSome other error: " << errno << endl;
 180                         break;
 181         }
 182         // We got an error so we close down the conversion engine
 183         if (iconv_close(pimpl_->cd) == -1) {
 184                 lyxerr << "Error returned from iconv_close("
 185                         << errno << ")" << endl;
 186         }
 187         pimpl_->cd = invalid_cd;
 188         return -1;
 189 }
 190
 191
 192 namespace {
 193
 194
 195 template<typename RetType, typename InType>
 196 std::vector<RetType>
 197 iconv_convert(IconvProcessor & processor,
 198               InType const * buf,
 199               size_t buflen)
 200 {
 201         if (buflen == 0)
 202                 return std::vector<RetType>();
 203
 204         char const * inbuf = reinterpret_cast<char const *>(buf);
 205         size_t inbytesleft = buflen * sizeof(InType);
 206
 207         size_t const outsize = 32768;
 208         static char out[outsize];
 209         char * outbuf = out;
 210
 211         int bytes = processor.convert(inbuf, inbytesleft, outbuf, outsize);
 212         if (bytes <= 0)
 213                 // Conversion failed
 214                 // FIXME Maybe throw an exception and handle that in the caller?
 215                 return std::vector<RetType>();
 216
 217         RetType const * tmp = reinterpret_cast<RetType const *>(out);
 218         return std::vector<RetType>(tmp, tmp + bytes / sizeof(RetType));
 219 }
 220
 221 } // anon namespace
 222
 223
 224 std::vector<lyx::char_type> utf8_to_ucs4(std::vector<char> const & utf8str)
 225 {
 226         if (utf8str.empty())
 227                 return std::vector<lyx::char_type>();
 228
 229         return utf8_to_ucs4(&utf8str[0], utf8str.size());
 230 }
 231
 232
 233 std::vector<lyx::char_type>
 234 utf8_to_ucs4(char const * utf8str, size_t ls)
 235 {
 236         static IconvProcessor processor(ucs4_codeset, "UTF-8");
 237         return iconv_convert<lyx::char_type>(processor, utf8str, ls);
 238 }
 239
 240
 241 std::vector<char_type>
 242 utf16_to_ucs4(unsigned short const * s, size_t ls)
 243 {
 244         static IconvProcessor processor(ucs4_codeset, utf16_codeset);
 245         return iconv_convert<char_type>(processor, s, ls);
 246 }
 247
 248
 249 std::vector<unsigned short>
 250 ucs4_to_utf16(char_type const * s, size_t ls)
 251 {
 252         static IconvProcessor processor(utf16_codeset, ucs4_codeset);
 253         return iconv_convert<unsigned short>(processor, s, ls);
 254 }
 255
 256
 257 std::vector<char>
 258 ucs4_to_utf8(lyx::char_type c)
 259 {
 260         static IconvProcessor processor("UTF-8", ucs4_codeset);
 261         return iconv_convert<char>(processor, &c, 1);
 262 }
 263
 264
 265 std::vector<char>
 266 ucs4_to_utf8(std::vector<lyx::char_type> const & ucs4str)
 267 {
 268         if (ucs4str.empty())
 269                 return std::vector<char>();
 270
 271         return ucs4_to_utf8(&ucs4str[0], ucs4str.size());
 272 }
 273
 274
 275 std::vector<char>
 276 ucs4_to_utf8(lyx::char_type const * ucs4str, size_t ls)
 277 {
 278         static IconvProcessor processor("UTF-8", ucs4_codeset);
 279         return iconv_convert<char>(processor, ucs4str, ls);
 280 }
 281
 282
 283 std::vector<lyx::char_type>
 284 eightbit_to_ucs4(char const * s, size_t ls, std::string const & encoding)
 285 {
 286         static std::map<std::string, IconvProcessor> processors;
 287         if (processors.find(encoding) == processors.end()) {
 288                 IconvProcessor processor(ucs4_codeset, encoding.c_str());
 289                 processors.insert(std::make_pair(encoding, processor));
 290         }
 291         return iconv_convert<char_type>(processors[encoding], s, ls);
 292 }
 293
 294
 295 std::vector<char>
 296 ucs4_to_eightbit(lyx::char_type const * ucs4str, size_t ls, std::string const & encoding)
 297 {
 298         static std::map<std::string, IconvProcessor> processors;
 299         if (processors.find(encoding) == processors.end()) {
 300                 IconvProcessor processor(encoding.c_str(), ucs4_codeset);
 301                 processors.insert(std::make_pair(encoding, processor));
 302         }
 303         return iconv_convert<char>(processors[encoding], ucs4str, ls);
 304 }
 305
 306 } // namespace lyx