src/Encoding.cpp

   1 /**
   2  * \file Encoding.cpp
   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 "Lexer.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
 127 /// The highest code point in UCS4 encoding (1<<20 + 1<<16)
 128 char_type const max_ucs4 = 0x110000;
 129
 130 } // namespace anon
 131
 132
 133 Encoding::Encoding(string const & n, string const & l, string const & i,
 134                    bool f, Encoding::Package p)
 135         : Name_(n), LatexName_(l), iconvName_(i), fixedwidth_(f), package_(p)
 136 {
 137         if (n == "ascii") {
 138                 // ASCII can encode 128 code points and nothing else
 139                 start_encodable_ = 128;
 140                 complete_ = true;
 141         } else if (i == "UTF-8") {
 142                 // UTF8 can encode all UCS4 code points
 143                 start_encodable_ = max_ucs4;
 144                 complete_ = true;
 145         } else {
 146                 complete_ = false;
 147         }
 148 }
 149
 150
 151 void Encoding::init() const
 152 {
 153         start_encodable_ = 0;
 154         // temporarily switch off lyxerr, since we will generate iconv errors
 155         lyxerr.disable();
 156         if (fixedwidth_) {
 157                 // We do not need to check all UCS4 code points, it is enough
 158                 // if we check all 256 code points of this encoding.
 159                 for (unsigned short j = 0; j < 256; ++j) {
 160                         char const c = j;
 161                         std::vector<char_type> const ucs4 = eightbit_to_ucs4(&c, 1, iconvName_);
 162                         if (ucs4.size() == 1) {
 163                                 char_type const c = ucs4[0];
 164                                 CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 165                                 if (it == unicodesymbols.end() || !it->second.force)
 166                                         encodable_.insert(c);
 167                         }
 168                 }
 169                 } else {
 170                 // We do not know how many code points this encoding has, and
 171                 // they do not have a direct representation as a single byte,
 172                 // therefore we need to check all UCS4 code points.
 173                 // This is expensive!
 174                 for (char_type c = 0; c < max_ucs4; ++c) {
 175                         std::vector<char> const eightbit = ucs4_to_eightbit(&c, 1, iconvName_);
 176                         if (!eightbit.empty()) {
 177                                 CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 178                                 if (it == unicodesymbols.end() || !it->second.force)
 179                                         encodable_.insert(c);
 180                         }
 181                 }
 182         }
 183         lyxerr.enable();
 184         CharSet::iterator it = encodable_.find(start_encodable_);
 185         while (it != encodable_.end()) {
 186                 encodable_.erase(it);
 187                 ++start_encodable_;
 188                 it = encodable_.find(start_encodable_);
 189         }
 190         complete_ = true;
 191 }
 192
 193
 194 docstring const Encoding::latexChar(char_type c) const
 195 {
 196         // assure the used encoding is properly initialized
 197         if (!complete_)
 198                 init();
 199         BOOST_ASSERT(complete_);
 200
 201         if (c < start_encodable_)
 202                 return docstring(1, c);
 203         if (encodable_.find(c) == encodable_.end()) {
 204                 // c cannot be encoded in this encoding
 205                 CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 206                 if (it == unicodesymbols.end())
 207                         lyxerr << "Could not find LaTeX command for character 0x"
 208                                << std::hex << c << std::dec
 209                                << ".\nLaTeX export will fail."
 210                                << endl;
 211                 else
 212                         return it->second.command;
 213         }
 214         return docstring(1, c);
 215 }
 216
 217
 218 void Encodings::validate(char_type c, LaTeXFeatures & features)
 219 {
 220         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 221         if (it != unicodesymbols.end() && !it->second.preamble.empty()) {
 222                 if (it->second.feature)
 223                         features.require(it->second.preamble);
 224                 else
 225                         features.addPreambleSnippet(it->second.preamble);
 226         }
 227 }
 228
 229
 230 bool Encodings::isComposeChar_hebrew(char_type c)
 231 {
 232         return c <= 0x05c2 && c >= 0x05b0 &&
 233                c != 0x05be && c != 0x05c0;
 234 }
 235
 236
 237 // Special Arabic letters are ones that do not get connected from left
 238 // they are hamza, alef_madda, alef_hamza, waw_hamza, alef_hamza_under,
 239 // alef, tah_marbota, dal, thal, rah, zai, wow, alef_maksoura
 240
 241 bool Encodings::is_arabic_special(char_type c)
 242 {
 243         return (c >= 0x0621 && c <= 0x0625) ||
 244                 c == 0x0627 || c == 0x0629  ||
 245                 c == 0x062f || c == 0x0648  ||
 246                (c >= 0x0630 && c <= 0x0632) ||
 247                 c == 0x0649;
 248 }
 249
 250
 251 bool Encodings::isComposeChar_arabic(char_type c)
 252 {
 253         return c >= 0x064b && c <= 0x0652;
 254 }
 255
 256
 257 bool Encodings::is_arabic(char_type c)
 258 {
 259         return c >= arabic_start && c <= arabic_end &&
 260                arabic_table[c-arabic_start][0];
 261 }
 262
 263
 264 char_type Encodings::transformChar(char_type c,
 265                                       Encodings::Letter_Form form)
 266 {
 267         if (!is_arabic(c))
 268                 return c;
 269
 270         return arabic_table[c-arabic_start][form];
 271 }
 272
 273
 274 bool Encodings::isCombiningChar(char_type c)
 275 {
 276         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 277         if (it != unicodesymbols.end())
 278                 return it->second.combining;
 279         return false;
 280 }
 281
 282
 283 Encoding const * Encodings::getFromLyXName(string const & name) const
 284 {
 285         EncodingList::const_iterator it = encodinglist.find(name);
 286         if (it != encodinglist.end())
 287                 return &it->second;
 288         else
 289                 return 0;
 290 }
 291
 292
 293 Encoding const * Encodings::getFromLaTeXName(string const & name) const
 294 {
 295         // We don't use std::find_if because it makes copies of the pairs in
 296         // the map.
 297         // This linear search is OK since we don't have many encodings.
 298         // Users could even optimize it by putting the encodings they use
 299         // most at the top of lib/encodings.
 300         EncodingList::const_iterator const end = encodinglist.end();
 301         for (EncodingList::const_iterator it = encodinglist.begin(); it != end; ++it)
 302                 if (it->second.latexName() == name)
 303                         return &it->second;
 304         return 0;
 305 }
 306
 307
 308 Encodings::Encodings()
 309 {
 310 }
 311
 312
 313 void Encodings::read(FileName const & encfile, FileName const & symbolsfile)
 314 {
 315         // We must read the symbolsfile first, because the Encoding
 316         // constructor depends on it.
 317         Lexer symbolslex(0, 0);
 318         symbolslex.setFile(symbolsfile);
 319         while (symbolslex.isOK()) {
 320                 char_type symbol;
 321                 CharInfo info;
 322                 string flags;
 323
 324                 if (symbolslex.next(true)) {
 325                         std::istringstream is(symbolslex.getString());
 326                         // reading symbol directly does not work if
 327                         // char_type == std::wchar_t.
 328                         boost::uint32_t tmp;
 329                         if(!(is >> std::hex >> tmp))
 330                                 break;
 331                         symbol = tmp;
 332                 } else
 333                         break;
 334                 if (symbolslex.next(true))
 335                         info.command = symbolslex.getDocString();
 336                 else
 337                         break;
 338                 if (symbolslex.next(true))
 339                         info.preamble = symbolslex.getString();
 340                 else
 341                         break;
 342                 if (symbolslex.next(true))
 343                         flags = symbolslex.getString();
 344                 else
 345                         break;
 346
 347                 info.combining = false;
 348                 info.feature = false;
 349                 info.force = false;
 350                 while (!flags.empty()) {
 351                         string flag;
 352                         flags = support::split(flags, flag, ',');
 353                         if (flag == "combining")
 354                                 info.combining = true;
 355                         else if (flag == "force")
 356                                 info.force = true;
 357                         else
 358                                 lyxerr << "Ignoring unknown flag `" << flag
 359                                        << "' for symbol `0x"
 360                                        << std::hex << symbol << std::dec
 361                                        << "'." << endl;
 362                 }
 363
 364                 if (!info.preamble.empty())
 365                         info.feature = info.preamble[0] != '\\';
 366
 367                 LYXERR(Debug::INFO)
 368                         << "Read unicode symbol " << symbol << " '"
 369                         << to_utf8(info.command) << "' '" << info.preamble
 370                         << "' " << info.combining << ' ' << info.feature
 371                         << endl;
 372                 unicodesymbols[symbol] = info;
 373         }
 374
 375         // Now read the encodings
 376         enum Encodingtags {
 377                 et_encoding = 1,
 378                 et_end,
 379                 et_last
 380         };
 381
 382         struct keyword_item encodingtags[et_last - 1] = {
 383                 { "encoding", et_encoding },
 384                 { "end", et_end }
 385         };
 386
 387         Lexer lex(encodingtags, et_last - 1);
 388         lex.setFile(encfile);
 389         while (lex.isOK()) {
 390                 switch (lex.lex()) {
 391                 case et_encoding:
 392                 {
 393                         lex.next();
 394                         string const name = lex.getString();
 395                         lex.next();
 396                         string const latexname = lex.getString();
 397                         lex.next();
 398                         string const iconvname = lex.getString();
 399                         lex.next();
 400                         string const width = lex.getString();
 401                         bool fixedwidth;
 402                         if (width == "fixed")
 403                                 fixedwidth = true;
 404                         else if (width == "variable")
 405                                 fixedwidth = false;
 406                         else
 407                                 lex.printError("Encodings::read: "
 408                                                "Unknown width: `$$Token'");
 409                         lex.next();
 410                         string const p = lex.getString();
 411                         Encoding::Package package;
 412                         if (p == "none")
 413                                 package = Encoding::none;
 414                         else if (p == "inputenc")
 415                                 package = Encoding::inputenc;
 416                         else if (p == "CJK")
 417                                 package = Encoding::CJK;
 418                         else
 419                                 lex.printError("Encodings::read: "
 420                                                "Unknown package: `$$Token'");
 421                         LYXERR(Debug::INFO) << "Reading encoding " << name << endl;
 422                         encodinglist[name] = Encoding(name, latexname,
 423                                                       iconvname, fixedwidth,
 424                                                       package);
 425                         if (lex.lex() != et_end)
 426                                 lex.printError("Encodings::read: "
 427                                                "missing end");
 428                         break;
 429                 }
 430                 case et_end:
 431                         lex.printError("Encodings::read: Misplaced end");
 432                         break;
 433                 case Lexer::LEX_FEOF:
 434                         break;
 435                 default:
 436                         lex.printError("Encodings::read: "
 437                                        "Unknown tag: `$$Token'");
 438                         break;
 439                 }
 440         }
 441 }
 442
 443
 444 } // namespace lyx