src/encoding.C

   1 /**
   2  * \file encoding.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  * \author Jean-Marc Lasgouttes
   8  * \author Dekel Tsur
   9  *
  10  * Full author contact details are available in file CREDITS.
  11  */
  12
  13 #include <config.h>
  14
  15 #include "encoding.h"
  16
  17 #include "debug.h"
  18 #include "LaTeXFeatures.h"
  19 #include "lyxlex.h"
  20 #include "lyxrc.h"
  21
  22 #include "support/filename.h"
  23 #include "support/lstrings.h"
  24 #include "support/unicode.h"
  25
  26 #include <sstream>
  27
  28
  29 namespace lyx {
  30
  31 using support::FileName;
  32
  33 #ifndef CXX_GLOBAL_CSTD
  34 using std::strtol;
  35 #endif
  36
  37 using std::endl;
  38 using std::string;
  39
  40
  41 Encodings encodings;
  42
  43 namespace {
  44
  45 char_type arabic_table[50][4] = {
  46         {0xfe80, 0xfe80, 0xfe80, 0xfe80}, // 0x0621 = hamza
  47         {0xfe81, 0xfe82, 0xfe81, 0xfe82}, // 0x0622 = ligature madda on alef
  48         {0xfe83, 0xfe84, 0xfe83, 0xfe84}, // 0x0623 = ligature hamza on alef
  49         {0xfe85, 0xfe86, 0xfe85, 0xfe86}, // 0x0624 = ligature hamza on waw
  50         {0xfe87, 0xfe88, 0xfe87, 0xfe88}, // 0x0625 = ligature hamza under alef
  51         {0xfe89, 0xfe8a, 0xfe8b, 0xfe8c}, // 0x0626 = ligature hamza on ya
  52         {0xfe8d, 0xfe8e, 0xfe8d, 0xfe8e}, // 0x0627 = alef
  53         {0xfe8f, 0xfe90, 0xfe91, 0xfe92}, // 0x0628 = baa
  54         {0xfe93, 0xfe94, 0xfe93, 0xfe94}, // 0x0629 = taa marbuta
  55         {0xfe95, 0xfe96, 0xfe97, 0xfe98}, // 0x062a = taa
  56         {0xfe99, 0xfe9a, 0xfe9b, 0xfe9c}, // 0x062b = thaa
  57         {0xfe9d, 0xfe9e, 0xfe9f, 0xfea0}, // 0x062c = jeem
  58         {0xfea1, 0xfea2, 0xfea3, 0xfea4}, // 0x062d = haa
  59         {0xfea5, 0xfea6, 0xfea7, 0xfea8}, // 0x062e = khaa
  60         {0xfea9, 0xfeaa, 0xfea9, 0xfeaa}, // 0x062f = dal
  61
  62         {0xfeab, 0xfeac, 0xfeab, 0xfeac}, // 0x0630 = thal
  63         {0xfead, 0xfeae, 0xfead, 0xfeae}, // 0x0631 = ra
  64         {0xfeaf, 0xfeb0, 0xfeaf, 0xfeb0}, // 0x0632 = zain
  65         {0xfeb1, 0xfeb2, 0xfeb3, 0xfeb4}, // 0x0633 = seen
  66         {0xfeb5, 0xfeb6, 0xfeb7, 0xfeb8}, // 0x0634 = sheen
  67         {0xfeb9, 0xfeba, 0xfebb, 0xfebc}, // 0x0635 = sad
  68         {0xfebd, 0xfebe, 0xfebf, 0xfec0}, // 0x0636 = dad
  69         {0xfec1, 0xfec2, 0xfec3, 0xfec4}, // 0x0637 = tah
  70         {0xfec5, 0xfec6, 0xfec7, 0xfec8}, // 0x0638 = zah
  71         {0xfec9, 0xfeca, 0xfecb, 0xfecc}, // 0x0639 = ain
  72         {0xfecd, 0xfece, 0xfecf, 0xfed0}, // 0x063a = ghain
  73         {0, 0, 0, 0}, // 0x063b
  74         {0, 0, 0, 0}, // 0x063c
  75         {0, 0, 0, 0}, // 0x063d
  76         {0, 0, 0, 0}, // 0x063e
  77         {0, 0, 0, 0}, // 0x063f
  78
  79         {0, 0, 0, 0}, // 0x0640
  80         {0xfed1, 0xfed2, 0xfed3, 0xfed4}, // 0x0641 = fa
  81         {0xfed5, 0xfed6, 0xfed7, 0xfed8}, // 0x0642 = qaf
  82         {0xfed9, 0xfeda, 0xfedb, 0xfedc}, // 0x0643 = kaf
  83         {0xfedd, 0xfede, 0xfedf, 0xfee0}, // 0x0644 = lam
  84         {0xfee1, 0xfee2, 0xfee3, 0xfee4}, // 0x0645 = meem
  85         {0xfee5, 0xfee6, 0xfee7, 0xfee8}, // 0x0646 = noon
  86         {0xfee9, 0xfeea, 0xfeeb, 0xfeec}, // 0x0647 = ha
  87         {0xfeed, 0xfeee, 0xfeed, 0xfeee}, // 0x0648 = waw
  88         {0xfeef, 0xfef0, 0xfeef, 0xfef0}, // 0x0649 = alef maksura
  89         {0xfef1, 0xfef2, 0xfef3, 0xfef4}, // 0x064a = ya
  90         {0x065b, 0x065b, 0x065b, 0x065b}, // 0x064b = fathatan
  91         {0x065c, 0x065c, 0x065c, 0x065c}, // 0x064c = dammatan
  92         {0x064d, 0x064d, 0x064d, 0x064d}, // 0x064d = kasratan
  93         {0x064e, 0x064e, 0x064e, 0x064e}, // 0x064e = fatha
  94         {0x064f, 0x064f, 0x064f, 0x064f}, // 0x064f = damma
  95
  96         {0x0650, 0x0650, 0x0650, 0x0650}, // 0x0650 = kasra
  97         {0x0651, 0x0651, 0x0651, 0x0651}, // 0x0651 = shadda
  98         {0x0652, 0x0652, 0x0652, 0x0652}, // 0x0652 = sukun
  99 };
 100
 101
 102 char_type const arabic_start = 0x0621;
 103 char_type const arabic_end = 0x0652;
 104
 105
 106 /// Information about a single UCS4 character
 107 struct CharInfo {
 108         /// LaTeX command for this character
 109         docstring command;
 110         /// Needed LaTeX preamble (or feature)
 111         string preamble;
 112         /// Is this a combining character?
 113         bool combining;
 114         /// Is \c preamble a feature known by LaTeXFeatures, or a raw LaTeX
 115         /// command?
 116         bool feature;
 117         /// Always force the LaTeX command, even if the encoding contains
 118         /// this character?
 119         bool force;
 120 };
 121
 122
 123 typedef std::map<char_type, CharInfo> CharInfoMap;
 124 CharInfoMap unicodesymbols;
 125
 126 } // namespace anon
 127
 128
 129 Encoding::Encoding(string const & n, string const & l, string const & i)
 130         : Name_(n), LatexName_(l), iconvName_(i)
 131 {
 132         if (n == "ascii")
 133                 // ASCII can encode 128 code points and nothing else
 134                 start_encodable_ = 128;
 135         else if (i == "UTF-8")
 136                 // UTF8 can encode all 1<<20 + 1<<16 UCS4 code points
 137                 start_encodable_ = 0x110000;
 138         else {
 139                 start_encodable_ = 0;
 140                 // temporarily switch off lyxerr, since we will generate iconv errors
 141                 lyxerr.disable();
 142                 for (unsigned short j = 0; j < 256; ++j) {
 143                         char const c = j;
 144                         std::vector<char_type> const ucs4 = eightbit_to_ucs4(&c, 1, i);
 145                         if (ucs4.size() == 1) {
 146                                 char_type const c = ucs4[0];
 147                                 CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 148                                 if (it == unicodesymbols.end() || !it->second.force)
 149                                         encodable_.insert(c);
 150                         }
 151                 }
 152                 lyxerr.enable();
 153                 CharSet::iterator it = encodable_.find(start_encodable_);
 154                 while (it != encodable_.end()) {
 155                         encodable_.erase(it);
 156                         ++start_encodable_;
 157                         it = encodable_.find(start_encodable_);
 158                 }
 159         }
 160 }
 161
 162
 163 docstring const Encoding::latexChar(char_type c) const
 164 {
 165         if (c < start_encodable_)
 166                 return docstring(1, c);
 167         if (encodable_.find(c) == encodable_.end()) {
 168                 // c cannot be encoded in this encoding
 169                 CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 170                 if (it == unicodesymbols.end())
 171                         lyxerr << "Could not find LaTeX command for character 0x"
 172                                << std::hex << c << std::dec
 173                                << ".\nLaTeX export will fail."
 174                                << endl;
 175                 else
 176                         return it->second.command;
 177         }
 178         return docstring(1, c);
 179 }
 180
 181
 182 void Encoding::validate(char_type c, LaTeXFeatures & features) const
 183 {
 184         // Add the preamble stuff even if c can be encoded in this encoding,
 185         // since the inputenc package only maps the code point c to a command,
 186         // it does not make this command available.
 187         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 188         if (it != unicodesymbols.end() && !it->second.preamble.empty()) {
 189                 if (it->second.feature)
 190                         features.require(it->second.preamble);
 191                 else
 192                         features.addPreambleSnippet(it->second.preamble);
 193         }
 194 }
 195
 196
 197 bool Encodings::isComposeChar_hebrew(char_type c)
 198 {
 199         return c <= 0x05c2 && c >= 0x05b0 &&
 200                c != 0x05be && c != 0x05c0;
 201 }
 202
 203
 204 // Special Arabic letters are ones that do not get connected from left
 205 // they are hamza, alef_madda, alef_hamza, waw_hamza, alef_hamza_under,
 206 // alef, tah_marbota, dal, thal, rah, zai, wow, alef_maksoura
 207
 208 bool Encodings::is_arabic_special(char_type c)
 209 {
 210         return (c >= 0x0621 && c <= 0x0625) ||
 211                 c == 0x0627 || c == 0x0629  ||
 212                 c == 0x062f || c == 0x0648  ||
 213                (c >= 0x0630 && c <= 0x0632) ||
 214                 c == 0x0649;
 215 }
 216
 217
 218 bool Encodings::isComposeChar_arabic(char_type c)
 219 {
 220         return c >= 0x064b && c <= 0x0652;
 221 }
 222
 223
 224 bool Encodings::is_arabic(char_type c)
 225 {
 226         return c >= arabic_start && c <= arabic_end &&
 227                arabic_table[c-arabic_start][0];
 228 }
 229
 230
 231 char_type Encodings::transformChar(char_type c,
 232                                       Encodings::Letter_Form form)
 233 {
 234         if (!is_arabic(c))
 235                 return c;
 236
 237         return arabic_table[c-arabic_start][form];
 238 }
 239
 240
 241 bool Encodings::isCombiningChar(char_type c)
 242 {
 243         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 244         if (it != unicodesymbols.end())
 245                 return it->second.combining;
 246         return false;
 247 }
 248
 249
 250 Encoding const * Encodings::getFromLyXName(string const & name) const
 251 {
 252         EncodingList::const_iterator it = encodinglist.find(name);
 253         if (it != encodinglist.end())
 254                 return &it->second;
 255         else
 256                 return 0;
 257 }
 258
 259
 260 Encoding const * Encodings::getFromLaTeXName(string const & name) const
 261 {
 262         // We don't use std::find_if because it makes copies of the pairs in
 263         // the map.
 264         // This linear search is OK since we don't have many encodings.
 265         // Users could even optimize it by putting the encodings they use
 266         // most at the top of lib/encodings.
 267         EncodingList::const_iterator const end = encodinglist.end();
 268         for (EncodingList::const_iterator it = encodinglist.begin(); it != end; ++it)
 269                 if (it->second.latexName() == name)
 270                         return &it->second;
 271         return 0;
 272 }
 273
 274
 275 Encodings::Encodings()
 276 {
 277 }
 278
 279
 280 void Encodings::read(FileName const & encfile, FileName const & symbolsfile)
 281 {
 282         // We must read the symbolsfile first, because the Encoding
 283         // constructor depends on it.
 284         LyXLex symbolslex(0, 0);
 285         symbolslex.setFile(symbolsfile);
 286         while (symbolslex.isOK()) {
 287                 char_type symbol;
 288                 CharInfo info;
 289                 string flags;
 290
 291                 if (symbolslex.next(true)) {
 292                         std::istringstream is(symbolslex.getString());
 293                         // reading symbol directly does not work if
 294                         // char_type == std::wchar_t.
 295                         boost::uint32_t tmp;
 296                         if(!(is >> std::hex >> tmp))
 297                                 break;
 298                         symbol = tmp;
 299                 } else
 300                         break;
 301                 if (symbolslex.next(true))
 302                         info.command = symbolslex.getDocString();
 303                 else
 304                         break;
 305                 if (symbolslex.next(true))
 306                         info.preamble = symbolslex.getString();
 307                 else
 308                         break;
 309                 if (symbolslex.next(true))
 310                         flags = symbolslex.getString();
 311                 else
 312                         break;
 313
 314                 info.combining = false;
 315                 info.feature = false;
 316                 info.force = false;
 317                 while (!flags.empty()) {
 318                         string flag;
 319                         flags = support::split(flags, flag, ',');
 320                         if (flag == "combining")
 321                                 info.combining = true;
 322                         else if (flag == "force")
 323                                 info.force = true;
 324                         else
 325                                 lyxerr << "Ignoring unknown flag `" << flag
 326                                        << "' for symbol `0x"
 327                                        << std::hex << symbol << std::dec
 328                                        << "'." << endl;
 329                 }
 330
 331                 if (!info.preamble.empty())
 332                         info.feature = info.preamble[0] != '\\';
 333
 334                 LYXERR(Debug::INFO)
 335                         << "Read unicode symbol " << symbol << " '"
 336                         << to_utf8(info.command) << "' '" << info.preamble
 337                         << "' " << info.combining << ' ' << info.feature
 338                         << endl;
 339                 unicodesymbols[symbol] = info;
 340         }
 341
 342         // Now read the encodings
 343         enum Encodingtags {
 344                 et_encoding = 1,
 345                 et_end,
 346                 et_last
 347         };
 348
 349         struct keyword_item encodingtags[et_last - 1] = {
 350                 { "encoding", et_encoding },
 351                 { "end", et_end }
 352         };
 353
 354         LyXLex lex(encodingtags, et_last - 1);
 355         lex.setFile(encfile);
 356         while (lex.isOK()) {
 357                 switch (lex.lex()) {
 358                 case et_encoding:
 359                 {
 360                         lex.next();
 361                         string const name = lex.getString();
 362                         lex.next();
 363                         string const latexname = lex.getString();
 364                         lex.next();
 365                         string const iconvname = lex.getString();
 366                         LYXERR(Debug::INFO) << "Reading encoding " << name << endl;
 367                         encodinglist[name] = Encoding(name, latexname, iconvname);
 368                         if (lex.lex() != et_end)
 369                                 lex.printError("Encodings::read: "
 370                                                "missing end");
 371                         break;
 372                 }
 373                 case et_end:
 374                         lex.printError("Encodings::read: Misplaced end");
 375                         break;
 376                 case LyXLex::LEX_FEOF:
 377                         break;
 378                 default:
 379                         lex.printError("Encodings::read: "
 380                                        "Unknown tag: `$$Token'");
 381                         break;
 382                 }
 383         }
 384 }
 385
 386
 387 } // namespace lyx