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 "Buffer.h"
  18 #include "BufferList.h"
  19 #include "InsetIterator.h"
  20 #include "LaTeXFeatures.h"
  21 #include "Lexer.h"
  22 #include "LyXRC.h"
  23
  24 #include "support/debug.h"
  25 #include "support/FileName.h"
  26 #include "support/lstrings.h"
  27 #include "support/textutils.h"
  28 #include "support/unicode.h"
  29
  30 #include <boost/cstdint.hpp>
  31
  32 #include <sstream>
  33
  34 using namespace std;
  35 using namespace lyx::support;
  36
  37 namespace lyx {
  38
  39 Encodings encodings;
  40
  41 Encodings::MathCommandSet Encodings::mathcmd;
  42 Encodings::TextCommandSet Encodings::textcmd;
  43 Encodings::MathSymbolSet  Encodings::mathsym;
  44
  45 namespace {
  46
  47 char_type arabic_table[172][4] = {
  48         {0xfe80, 0xfe80, 0xfe80, 0xfe80}, // 0x0621 = hamza
  49         {0xfe81, 0xfe82, 0xfe81, 0xfe82}, // 0x0622 = ligature madda on alef
  50         {0xfe83, 0xfe84, 0xfe83, 0xfe84}, // 0x0623 = ligature hamza on alef
  51         {0xfe85, 0xfe86, 0xfe85, 0xfe86}, // 0x0624 = ligature hamza on waw
  52         {0xfe87, 0xfe88, 0xfe87, 0xfe88}, // 0x0625 = ligature hamza under alef
  53         {0xfe89, 0xfe8a, 0xfe8b, 0xfe8c}, // 0x0626 = ligature hamza on ya
  54         {0xfe8d, 0xfe8e, 0xfe8d, 0xfe8e}, // 0x0627 = alef
  55         {0xfe8f, 0xfe90, 0xfe91, 0xfe92}, // 0x0628 = baa
  56         {0xfe93, 0xfe94, 0xfe93, 0xfe94}, // 0x0629 = taa marbuta
  57         {0xfe95, 0xfe96, 0xfe97, 0xfe98}, // 0x062a = taa
  58         {0xfe99, 0xfe9a, 0xfe9b, 0xfe9c}, // 0x062b = thaa
  59         {0xfe9d, 0xfe9e, 0xfe9f, 0xfea0}, // 0x062c = jeem
  60         {0xfea1, 0xfea2, 0xfea3, 0xfea4}, // 0x062d = haa
  61         {0xfea5, 0xfea6, 0xfea7, 0xfea8}, // 0x062e = khaa
  62         {0xfea9, 0xfeaa, 0xfea9, 0xfeaa}, // 0x062f = dal
  63
  64         {0xfeab, 0xfeac, 0xfeab, 0xfeac}, // 0x0630 = thal
  65         {0xfead, 0xfeae, 0xfead, 0xfeae}, // 0x0631 = ra
  66         {0xfeaf, 0xfeb0, 0xfeaf, 0xfeb0}, // 0x0632 = zain
  67         {0xfeb1, 0xfeb2, 0xfeb3, 0xfeb4}, // 0x0633 = seen
  68         {0xfeb5, 0xfeb6, 0xfeb7, 0xfeb8}, // 0x0634 = sheen
  69         {0xfeb9, 0xfeba, 0xfebb, 0xfebc}, // 0x0635 = sad
  70         {0xfebd, 0xfebe, 0xfebf, 0xfec0}, // 0x0636 = dad
  71         {0xfec1, 0xfec2, 0xfec3, 0xfec4}, // 0x0637 = tah
  72         {0xfec5, 0xfec6, 0xfec7, 0xfec8}, // 0x0638 = zah
  73         {0xfec9, 0xfeca, 0xfecb, 0xfecc}, // 0x0639 = ain
  74         {0xfecd, 0xfece, 0xfecf, 0xfed0}, // 0x063a = ghain
  75         {0, 0, 0, 0}, // 0x063b
  76         {0, 0, 0, 0}, // 0x063c
  77         {0, 0, 0, 0}, // 0x063d
  78         {0, 0, 0, 0}, // 0x063e
  79         {0, 0, 0, 0}, // 0x063f
  80
  81         {0, 0, 0, 0}, // 0x0640
  82         {0xfed1, 0xfed2, 0xfed3, 0xfed4}, // 0x0641 = fa
  83         {0xfed5, 0xfed6, 0xfed7, 0xfed8}, // 0x0642 = qaf
  84         {0xfed9, 0xfeda, 0xfedb, 0xfedc}, // 0x0643 = kaf
  85         {0xfedd, 0xfede, 0xfedf, 0xfee0}, // 0x0644 = lam
  86         {0xfee1, 0xfee2, 0xfee3, 0xfee4}, // 0x0645 = meem
  87         {0xfee5, 0xfee6, 0xfee7, 0xfee8}, // 0x0646 = noon
  88         {0xfee9, 0xfeea, 0xfeeb, 0xfeec}, // 0x0647 = ha
  89         {0xfeed, 0xfeee, 0xfeed, 0xfeee}, // 0x0648 = waw
  90         {0xfeef, 0xfef0, 0xfeef, 0xfef0}, // 0x0649 = alef maksura
  91         {0xfef1, 0xfef2, 0xfef3, 0xfef4}, // 0x064a = ya
  92         {0x065b, 0x065b, 0x065b, 0x065b}, // 0x064b = fathatan
  93         {0x065c, 0x065c, 0x065c, 0x065c}, // 0x064c = dammatan
  94         {0x064d, 0x064d, 0x064d, 0x064d}, // 0x064d = kasratan
  95         {0x064e, 0x064e, 0x064e, 0x064e}, // 0x064e = fatha
  96         {0x064f, 0x064f, 0x064f, 0x064f}, // 0x064f = damma
  97
  98         {0x0650, 0x0650, 0x0650, 0x0650}, // 0x0650 = kasra
  99         {0x0651, 0x0651, 0x0651, 0x0651}, // 0x0651 = shadda
 100         {0x0652, 0x0652, 0x0652, 0x0652}, // 0x0652 = sukun
 101
 102         {0, 0, 0, 0}, // 0x0653
 103         {0, 0, 0, 0}, // 0x0654
 104         {0, 0, 0, 0}, // 0x0655
 105         {0, 0, 0, 0}, // 0x0656
 106         {0, 0, 0, 0}, // 0x0657
 107         {0, 0, 0, 0}, // 0x0658
 108         {0, 0, 0, 0}, // 0x0659
 109         {0, 0, 0, 0}, // 0x065a
 110         {0, 0, 0, 0}, // 0x065b
 111         {0, 0, 0, 0}, // 0x065c
 112         {0, 0, 0, 0}, // 0x065d
 113         {0, 0, 0, 0}, // 0x065e
 114         {0, 0, 0, 0}, // 0x065f
 115         {0, 0, 0, 0}, // 0x0660
 116         {0, 0, 0, 0}, // 0x0661
 117         {0, 0, 0, 0}, // 0x0662
 118         {0, 0, 0, 0}, // 0x0663
 119         {0, 0, 0, 0}, // 0x0664
 120         {0, 0, 0, 0}, // 0x0665
 121         {0, 0, 0, 0}, // 0x0666
 122         {0, 0, 0, 0}, // 0x0667
 123         {0, 0, 0, 0}, // 0x0668
 124         {0, 0, 0, 0}, // 0x0669
 125         {0, 0, 0, 0}, // 0x066a
 126         {0, 0, 0, 0}, // 0x066b
 127         {0, 0, 0, 0}, // 0x066c
 128         {0, 0, 0, 0}, // 0x066d
 129         {0, 0, 0, 0}, // 0x066e
 130         {0, 0, 0, 0}, // 0x066f
 131         {0, 0, 0, 0}, // 0x0670
 132         {0, 0, 0, 0}, // 0x0671
 133         {0, 0, 0, 0}, // 0x0672
 134         {0, 0, 0, 0}, // 0x0673
 135         {0, 0, 0, 0}, // 0x0674
 136         {0, 0, 0, 0}, // 0x0675
 137         {0, 0, 0, 0}, // 0x0676
 138         {0, 0, 0, 0}, // 0x0677
 139         {0, 0, 0, 0}, // 0x0678
 140         {0, 0, 0, 0}, // 0x0679
 141         {0, 0, 0, 0}, // 0x067a
 142         {0, 0, 0, 0}, // 0x067b
 143         {0, 0, 0, 0}, // 0x067c
 144         {0, 0, 0, 0}, // 0x067d
 145         {0xfb56, 0xfb57, 0xfb58, 0xfb59}, // 0x067e = peh
 146         {0, 0, 0, 0}, // 0x067f
 147         {0, 0, 0, 0}, // 0x0680
 148         {0, 0, 0, 0}, // 0x0681
 149         {0, 0, 0, 0}, // 0x0682
 150         {0, 0, 0, 0}, // 0x0683
 151         {0, 0, 0, 0}, // 0x0684
 152         {0, 0, 0, 0}, // 0x0685
 153         {0xfb7a, 0xfb7b, 0xfb7c, 0xfb7d}, // 0x0686 = tcheh
 154         {0, 0, 0, 0}, // 0x0687
 155         {0, 0, 0, 0}, // 0x0688
 156         {0, 0, 0, 0}, // 0x0689
 157         {0, 0, 0, 0}, // 0x068a
 158         {0, 0, 0, 0}, // 0x068b
 159         {0, 0, 0, 0}, // 0x068c
 160         {0, 0, 0, 0}, // 0x068d
 161         {0, 0, 0, 0}, // 0x068e
 162         {0, 0, 0, 0}, // 0x068f
 163         {0, 0, 0, 0}, // 0x0690
 164         {0, 0, 0, 0}, // 0x0691
 165         {0, 0, 0, 0}, // 0x0692
 166         {0, 0, 0, 0}, // 0x0693
 167         {0, 0, 0, 0}, // 0x0694
 168         {0, 0, 0, 0}, // 0x0695
 169         {0, 0, 0, 0}, // 0x0696
 170         {0, 0, 0, 0}, // 0x0697
 171         {0xfb8a, 0xfb8b, 0xfb8a, 0xfb8b}, // 0x0698 = jeh
 172         {0, 0, 0, 0}, // 0x0699
 173         {0, 0, 0, 0}, // 0x069a
 174         {0, 0, 0, 0}, // 0x069b
 175         {0, 0, 0, 0}, // 0x069c
 176         {0, 0, 0, 0}, // 0x069d
 177         {0, 0, 0, 0}, // 0x069e
 178         {0, 0, 0, 0}, // 0x069f
 179         {0, 0, 0, 0}, // 0x06a0
 180         {0, 0, 0, 0}, // 0x06a1
 181         {0, 0, 0, 0}, // 0x06a2
 182         {0, 0, 0, 0}, // 0x06a3
 183         {0, 0, 0, 0}, // 0x06a4
 184         {0, 0, 0, 0}, // 0x06a5
 185         {0, 0, 0, 0}, // 0x06a6
 186         {0, 0, 0, 0}, // 0x06a7
 187         {0, 0, 0, 0}, // 0x06a8
 188         {0xfb8e, 0xfb8f, 0xfb90, 0xfb91}, // 0x06a9 = farsi kaf
 189         {0, 0, 0, 0}, // 0x06aa
 190         {0, 0, 0, 0}, // 0x06ab
 191         {0, 0, 0, 0}, // 0x06ac
 192         {0, 0, 0, 0}, // 0x06ad
 193         {0, 0, 0, 0}, // 0x06ae
 194         {0xfb92, 0xfb93, 0xfb94, 0xfb95}, // 0x06af = gaf
 195         {0, 0, 0, 0}, // 0x06b0
 196         {0, 0, 0, 0}, // 0x06b1
 197         {0, 0, 0, 0}, // 0x06b2
 198         {0, 0, 0, 0}, // 0x06b3
 199         {0, 0, 0, 0}, // 0x06b4
 200         {0, 0, 0, 0}, // 0x06b5
 201         {0, 0, 0, 0}, // 0x06b6
 202         {0, 0, 0, 0}, // 0x06b7
 203         {0, 0, 0, 0}, // 0x06b8
 204         {0, 0, 0, 0}, // 0x06b9
 205         {0, 0, 0, 0}, // 0x06ba
 206         {0, 0, 0, 0}, // 0x06bb
 207         {0, 0, 0, 0}, // 0x06bc
 208         {0, 0, 0, 0}, // 0x06bd
 209         {0, 0, 0, 0}, // 0x06be
 210         {0, 0, 0, 0}, // 0x06bf
 211         {0, 0, 0, 0}, // 0x06c0
 212         {0, 0, 0, 0}, // 0x06c1
 213         {0, 0, 0, 0}, // 0x06c2
 214         {0, 0, 0, 0}, // 0x06c3
 215         {0, 0, 0, 0}, // 0x06c4
 216         {0, 0, 0, 0}, // 0x06c5
 217         {0, 0, 0, 0}, // 0x06c6
 218         {0, 0, 0, 0}, // 0x06c7
 219         {0, 0, 0, 0}, // 0x06c8
 220         {0, 0, 0, 0}, // 0x06c9
 221         {0, 0, 0, 0}, // 0x06ca
 222         {0, 0, 0, 0}, // 0x06cb
 223         {0xfbfc, 0xfbfd, 0xfbfe, 0xfbff} // 0x06cc = farsi yeh
 224 };
 225
 226
 227 char_type const arabic_start = 0x0621;
 228 char_type const arabic_end = 0x06cc;
 229
 230
 231 /// Information about a single UCS4 character
 232 struct CharInfo {
 233         /// LaTeX command (text mode) for this character
 234         docstring textcommand;
 235         /// LaTeX command (math mode) for this character
 236         docstring mathcommand;
 237         /// Needed LaTeX preamble (or feature) for text mode
 238         string textpreamble;
 239         /// Needed LaTeX preamble (or feature) for math mode
 240         string mathpreamble;
 241         /// Is this a combining character?
 242         bool combining;
 243         /// Is \c textpreamble a feature known by LaTeXFeatures, or a raw LaTeX
 244         /// command?
 245         bool textfeature;
 246         /// Is \c mathpreamble a feature known by LaTeXFeatures, or a raw LaTeX
 247         /// command?
 248         bool mathfeature;
 249         /// Always force the LaTeX command, even if the encoding contains
 250         /// this character?
 251         bool force;
 252 };
 253
 254
 255 typedef map<char_type, CharInfo> CharInfoMap;
 256 CharInfoMap unicodesymbols;
 257
 258 typedef std::set<char_type> CharSet;
 259 CharSet forced;
 260
 261 typedef std::set<char_type> MathAlphaSet;
 262 MathAlphaSet mathalpha;
 263
 264
 265 /// The highest code point in UCS4 encoding (1<<20 + 1<<16)
 266 char_type const max_ucs4 = 0x110000;
 267
 268 } // namespace anon
 269
 270
 271 EncodingException::EncodingException(char_type c)
 272         : failed_char(c), par_id(0), pos(0)
 273 {
 274 }
 275
 276
 277 const char * EncodingException::what() const throw()
 278 {
 279         return "Could not find LaTeX command for a character";
 280 }
 281
 282
 283 Encoding::Encoding(string const & n, string const & l, string const & g,
 284                    string const & i, bool f, Encoding::Package p)
 285         : name_(n), latexName_(l), guiName_(g), iconvName_(i), fixedwidth_(f), package_(p)
 286 {
 287         if (n == "ascii") {
 288                 // ASCII can encode 128 code points and nothing else
 289                 start_encodable_ = 128;
 290                 complete_ = true;
 291         } else if (i == "UTF-8") {
 292                 // UTF8 can encode all UCS4 code points
 293                 start_encodable_ = max_ucs4;
 294                 complete_ = true;
 295         } else {
 296                 complete_ = false;
 297         }
 298 }
 299
 300
 301 void Encoding::init() const
 302 {
 303         if (complete_)
 304                 return;
 305
 306         start_encodable_ = 0;
 307         // temporarily switch off lyxerr, since we will generate iconv errors
 308         lyxerr.disable();
 309         if (fixedwidth_) {
 310                 // We do not need to check all UCS4 code points, it is enough
 311                 // if we check all 256 code points of this encoding.
 312                 for (unsigned short j = 0; j < 256; ++j) {
 313                         char const c = char(j);
 314                         vector<char_type> const ucs4 = eightbit_to_ucs4(&c, 1, iconvName_);
 315                         if (ucs4.size() != 1)
 316                                 continue;
 317                         char_type const uc = ucs4[0];
 318                         CharInfoMap::const_iterator const it = unicodesymbols.find(uc);
 319                         if (it == unicodesymbols.end() || !it->second.force)
 320                                 encodable_.insert(uc);
 321                 }
 322         } else {
 323                 // We do not know how many code points this encoding has, and
 324                 // they do not have a direct representation as a single byte,
 325                 // therefore we need to check all UCS4 code points.
 326                 // This is expensive!
 327                 for (char_type c = 0; c < max_ucs4; ++c) {
 328                         vector<char> const eightbit = ucs4_to_eightbit(&c, 1, iconvName_);
 329                         if (!eightbit.empty()) {
 330                                 CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 331                                 if (it == unicodesymbols.end() || !it->second.force)
 332                                         encodable_.insert(c);
 333                         }
 334                 }
 335         }
 336         lyxerr.enable();
 337         CharSet::iterator it = encodable_.find(start_encodable_);
 338         while (it != encodable_.end()) {
 339                 encodable_.erase(it);
 340                 ++start_encodable_;
 341                 it = encodable_.find(start_encodable_);
 342         }
 343         complete_ = true;
 344 }
 345
 346
 347 docstring Encoding::latexChar(char_type c, bool for_mathed) const
 348 {
 349         // assure the used encoding is properly initialized
 350         init();
 351
 352         if (iconvName_ == "UTF-8" && package_ == none)
 353                 return docstring(1, c);
 354         if (c < start_encodable_ && !encodings.isForced(c))
 355                 return docstring(1, c);
 356         if (encodable_.find(c) != encodable_.end())
 357                 return docstring(1, c);
 358         if (for_mathed)
 359                 return docstring();
 360
 361         // c cannot (or should not) be encoded in this encoding
 362         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 363         if (it == unicodesymbols.end())
 364                 throw EncodingException(c);
 365         // at least one of mathcommand and textcommand is nonempty
 366         if (it->second.textcommand.empty())
 367                 return "\\ensuremath{" + it->second.mathcommand + '}';
 368         return it->second.textcommand;
 369 }
 370
 371
 372 vector<char_type> Encoding::symbolsList() const
 373 {
 374         // assure the used encoding is properly initialized
 375         init();
 376
 377         // first all encodable characters
 378         vector<char_type> symbols(encodable_.begin(), encodable_.end());
 379         // add those below start_encodable_
 380         for (char_type c = 0; c < start_encodable_; ++c)
 381                 symbols.push_back(c);
 382         // now the ones from the unicodesymbols file
 383         CharInfoMap::const_iterator const end = unicodesymbols.end();
 384         CharInfoMap::const_iterator it = unicodesymbols.begin();
 385         for (; it != end; ++it)
 386                 symbols.push_back(it->first);
 387         return symbols;
 388 }
 389
 390
 391 bool Encodings::latexMathChar(char_type c, bool mathmode,
 392                         Encoding const * encoding, docstring & command)
 393 {
 394         if (encoding)
 395                 command = encoding->latexChar(c, true);
 396
 397         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 398         if (it == unicodesymbols.end()) {
 399                 if (!encoding || command.empty())
 400                         throw EncodingException(c);
 401                 if (mathmode)
 402                         addMathSym(c);
 403                 return false;
 404         }
 405         // at least one of mathcommand and textcommand is nonempty
 406         bool use_math = (mathmode && !it->second.mathcommand.empty()) ||
 407                         (!mathmode && it->second.textcommand.empty());
 408         if (use_math) {
 409                 command = it->second.mathcommand;
 410                 addMathCmd(c);
 411         } else {
 412                 if (!encoding || command.empty()) {
 413                         command = it->second.textcommand;
 414                         addTextCmd(c);
 415                 }
 416                 if (mathmode)
 417                         addMathSym(c);
 418         }
 419         return use_math;
 420 }
 421
 422
 423 char_type Encodings::fromLaTeXCommand(docstring const & cmd, bool & combining)
 424 {
 425         CharInfoMap::const_iterator const end = unicodesymbols.end();
 426         CharInfoMap::const_iterator it = unicodesymbols.begin();
 427         for (combining = false; it != end; ++it) {
 428                 docstring const math = it->second.mathcommand;
 429                 docstring const text = it->second.textcommand;
 430                 if (math == cmd || text == cmd) {
 431                         combining = it->second.combining;
 432                         return it->first;
 433                 }
 434         }
 435         return 0;
 436 }
 437
 438
 439 docstring Encodings::fromLaTeXCommand(docstring const & cmd, docstring & rem,
 440                                       int cmdtype)
 441 {
 442         bool const mathmode = cmdtype & MATH_CMD;
 443         bool const textmode = cmdtype & TEXT_CMD;
 444         docstring symbols;
 445         size_t i = 0;
 446         size_t const cmdend = cmd.size();
 447         CharInfoMap::const_iterator const uniend = unicodesymbols.end();
 448         for (size_t j = 0; j < cmdend; ++j) {
 449                 // Also get the char after a backslash
 450                 if (j + 1 < cmdend && cmd[j] == '\\')
 451                         ++j;
 452                 // If a macro argument follows, get it, too
 453                 if (j + 1 < cmdend && cmd[j + 1] == '{') {
 454                         size_t k = j + 1;
 455                         int count = 1;
 456                         while (k < cmdend && count && k != docstring::npos) {
 457                                 k = cmd.find_first_of(from_ascii("{}"), k + 1);
 458                                 if (cmd[k] == '{')
 459                                         ++count;
 460                                 else
 461                                         --count;
 462                         }
 463                         if (k != docstring::npos)
 464                                 j = k;
 465                 }
 466                 // Start with this substring and try augmenting it when it is
 467                 // the prefix of some command in the unicodesymbols file
 468                 docstring const subcmd = cmd.substr(i, j - i + 1);
 469
 470                 CharInfoMap::const_iterator it = unicodesymbols.begin();
 471                 size_t unicmd_size = 0;
 472                 char_type c = 0;
 473                 for (; it != uniend; ++it) {
 474                         docstring const math = mathmode ? it->second.mathcommand
 475                                                         : docstring();
 476                         docstring const text = textmode ? it->second.textcommand
 477                                                         : docstring();
 478                         size_t cur_size = max(math.size(), text.size());
 479                         // The current math or text unicode command cannot
 480                         // match, or we already matched a longer one
 481                         if (cur_size < subcmd.size() || cur_size <= unicmd_size)
 482                                 continue;
 483
 484                         docstring tmp = subcmd;
 485                         size_t k = j;
 486                         while (prefixIs(math, tmp) || prefixIs(text, tmp)) {
 487                                 ++k;
 488                                 if (k >= cmdend || cur_size <= tmp.size())
 489                                         break;
 490                                 tmp += cmd[k];
 491                         }
 492                         // No match
 493                         if (k == j)
 494                                 continue;
 495
 496                         // The last added char caused a mismatch, because
 497                         // we didn't exhaust the chars in cmd and didn't
 498                         // exceed the maximum size of the current unicmd
 499                         if (k < cmdend && cur_size > tmp.size())
 500                                 tmp.resize(tmp.size() - 1);
 501
 502                         // If this is an exact match, we found a (longer)
 503                         // matching entry in the unicodesymbols file.
 504                         // If the entry doesn't start with '\', we take note
 505                         // of the match and continue (this is not a ultimate
 506                         // acceptance, as some other entry may match a longer
 507                         // portion of the cmd string). However, if the entry
 508                         // does start with '\', we accept the match only if
 509                         // this is a valid macro, i.e., either it is a single
 510                         // (nonletter) char macro, or nothing else follows,
 511                         // or what follows is a nonletter char, or the last
 512                         // character is a }.
 513                         if ((math == tmp || text == tmp)
 514                             && (tmp[0] != '\\'
 515                                    || (tmp.size() == 2 && !isAlphaASCII(tmp[1]))
 516                                    || k == cmdend
 517                                    || !isAlphaASCII(cmd[k])
 518                                    || tmp[tmp.size() - 1] == '}')
 519                                  ) {
 520                                 c = it->first;
 521                                 j = k - 1;
 522                                 i = j + 1;
 523                                 unicmd_size = cur_size;
 524                         }
 525                 }
 526                 if (unicmd_size)
 527                         symbols += c;
 528                 else if (j + 1 == cmdend)
 529                         // No luck. Return what remains
 530                         rem = cmd.substr(i);
 531         }
 532         return symbols;
 533 }
 534
 535
 536 void Encodings::initUnicodeMath(Buffer const & buffer, bool clear_sets)
 537 {
 538 #ifdef TEX2LYX
 539         // The code below is not needed in tex2lyx and requires additional stuff
 540         (void)buffer;
 541         (void)clear_sets;
 542 #else
 543         if (clear_sets) {
 544                 mathcmd.clear();
 545                 textcmd.clear();
 546                 mathsym.clear();
 547         }
 548
 549         // Check master
 550         Inset & inset = buffer.inset();
 551         InsetIterator it = inset_iterator_begin(inset);
 552         InsetIterator const end = inset_iterator_end(inset);
 553         for (; it != end; ++it)
 554                 it->initUnicodeMath();
 555
 556         // Check children
 557         BufferList::iterator bit = theBufferList().begin();
 558         BufferList::iterator const bend = theBufferList().end();
 559         for (; bit != bend; ++bit)
 560                 if (buffer.isChild(*bit))
 561                         initUnicodeMath(**bit, false);
 562 #endif
 563 }
 564
 565
 566 void Encodings::validate(char_type c, LaTeXFeatures & features, bool for_mathed)
 567 {
 568 #ifdef TEX2LYX
 569         // The code below is not needed in tex2lyx and requires additional stuff
 570         (void)c;
 571         (void)features;
 572         (void)for_mathed;
 573 #else
 574         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 575         if (it != unicodesymbols.end()) {
 576                 // In mathed, c could be used both in textmode and mathmode
 577                 bool const use_math = (for_mathed && isMathCmd(c)) ||
 578                                       (!for_mathed && it->second.textcommand.empty());
 579                 bool const use_text = (for_mathed && isTextCmd(c)) ||
 580                                       (!for_mathed && !it->second.textcommand.empty());
 581                 if (use_math) {
 582                         if (!it->second.mathpreamble.empty()) {
 583                                 if (it->second.mathfeature) {
 584                                         string feats = it->second.mathpreamble;
 585                                         while (!feats.empty()) {
 586                                                 string feat;
 587                                                 feats = split(feats, feat, ',');
 588                                                 features.require(feat);
 589                                         }
 590                                 } else
 591                                         features.addPreambleSnippet(it->second.mathpreamble);
 592                         }
 593                 }
 594                 if (use_text) {
 595                         if (!it->second.textpreamble.empty()) {
 596                                 if (it->second.textfeature) {
 597                                         string feats = it->second.textpreamble;
 598                                         while (!feats.empty()) {
 599                                                 string feat;
 600                                                 feats = split(feats, feat, ',');
 601                                                 features.require(feat);
 602                                         }
 603                                 } else
 604                                         features.addPreambleSnippet(it->second.textpreamble);
 605                         }
 606                 }
 607         }
 608         if (for_mathed && isMathSym(c)) {
 609                 features.require("amstext");
 610                 features.require("lyxmathsym");
 611         }
 612 #endif
 613 }
 614
 615
 616 bool Encodings::isHebrewComposeChar(char_type c)
 617 {
 618         return c <= 0x05c2 && c >= 0x05b0 && c != 0x05be && c != 0x05c0;
 619 }
 620
 621
 622 // Special Arabic letters are ones that do not get connected from left
 623 // they are hamza, alef_madda, alef_hamza, waw_hamza, alef_hamza_under,
 624 // alef, tah_marbota, dal, thal, rah, zai, wow, alef_maksoura
 625
 626 bool Encodings::isArabicSpecialChar(char_type c)
 627 {
 628         return (c >= 0x0621 && c <= 0x0625) || (c >= 0x0630 && c <= 0x0632)
 629                 || c == 0x0627 || c == 0x0629 || c == 0x062f || c == 0x0648
 630                 || c == 0x0649 || c == 0x0698;
 631 }
 632
 633
 634 bool Encodings::isArabicComposeChar(char_type c)
 635 {
 636         return c >= 0x064b && c <= 0x0652;
 637 }
 638
 639
 640 bool Encodings::isArabicChar(char_type c)
 641 {
 642         return c >= arabic_start && c <= arabic_end
 643                 && arabic_table[c-arabic_start][0];
 644 }
 645
 646
 647 char_type Encodings::transformChar(char_type c, Encodings::LetterForm form)
 648 {
 649         return isArabicChar(c) ? arabic_table[c-arabic_start][form] : c;
 650 }
 651
 652
 653 bool Encodings::isCombiningChar(char_type c)
 654 {
 655         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 656         if (it != unicodesymbols.end())
 657                 return it->second.combining;
 658         return false;
 659 }
 660
 661
 662 bool Encodings::isKnownScriptChar(char_type const c, string & preamble)
 663 {
 664         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 665
 666         if (it == unicodesymbols.end())
 667                 return false;
 668
 669         if (it->second.textpreamble != "textgreek" && it->second.textpreamble != "textcyr")
 670                 return false;
 671
 672         if (preamble.empty()) {
 673                 preamble = it->second.textpreamble;
 674                 return true;
 675         }
 676         return it->second.textpreamble == preamble;
 677 }
 678
 679
 680 bool Encodings::isForced(char_type c)
 681 {
 682         return (!forced.empty() && forced.find(c) != forced.end());
 683 }
 684
 685
 686 bool Encodings::isMathAlpha(char_type c)
 687 {
 688         return mathalpha.count(c);
 689 }
 690
 691
 692 Encoding const * Encodings::fromLyXName(string const & name) const
 693 {
 694         EncodingList::const_iterator const it = encodinglist.find(name);
 695         return it != encodinglist.end() ? &it->second : 0;
 696 }
 697
 698
 699 Encoding const * Encodings::fromLaTeXName(string const & n) const
 700 {
 701         string name = n;
 702         // FIXME: if we have to test for too many of these synonyms,
 703         // we should instead extend the format of lib/encodings
 704         if (n == "ansinew")
 705                 name = "cp1252";
 706
 707         // We don't use find_if because it makes copies of the pairs in
 708         // the map.
 709         // This linear search is OK since we don't have many encodings.
 710         // Users could even optimize it by putting the encodings they use
 711         // most at the top of lib/encodings.
 712         EncodingList::const_iterator const end = encodinglist.end();
 713         for (EncodingList::const_iterator it = encodinglist.begin(); it != end; ++it)
 714                 if (it->second.latexName() == name)
 715                         return &it->second;
 716         return 0;
 717 }
 718
 719
 720 Encodings::Encodings()
 721 {
 722 }
 723
 724
 725 void Encodings::read(FileName const & encfile, FileName const & symbolsfile)
 726 {
 727         // We must read the symbolsfile first, because the Encoding
 728         // constructor depends on it.
 729         Lexer symbolslex;
 730         symbolslex.setFile(symbolsfile);
 731         bool getNextToken = true;
 732         while (symbolslex.isOK()) {
 733                 char_type symbol;
 734                 CharInfo info;
 735                 string flags;
 736
 737                 if (getNextToken) {
 738                         if (!symbolslex.next(true))
 739                                 break;
 740                 } else
 741                         getNextToken = true;
 742
 743                 istringstream is(symbolslex.getString());
 744                 // reading symbol directly does not work if
 745                 // char_type == wchar_t.
 746                 boost::uint32_t tmp;
 747                 if(!(is >> hex >> tmp))
 748                         break;
 749                 symbol = tmp;
 750
 751                 if (!symbolslex.next(true))
 752                         break;
 753                 info.textcommand = symbolslex.getDocString();
 754                 if (!symbolslex.next(true))
 755                         break;
 756                 info.textpreamble = symbolslex.getString();
 757                 if (!symbolslex.next(true))
 758                         break;
 759                 flags = symbolslex.getString();
 760
 761                 info.combining = false;
 762                 info.textfeature = false;
 763                 info.force = false;
 764                 while (!flags.empty()) {
 765                         string flag;
 766                         flags = split(flags, flag, ',');
 767                         if (flag == "combining") {
 768                                 info.combining = true;
 769                         } else if (flag == "force") {
 770                                 info.force = true;
 771                                 forced.insert(symbol);
 772                         } else if (flag == "mathalpha") {
 773                                 mathalpha.insert(symbol);
 774                         } else {
 775                                 lyxerr << "Ignoring unknown flag `" << flag
 776                                        << "' for symbol `0x"
 777                                        << hex << symbol << dec
 778                                        << "'." << endl;
 779                         }
 780                 }
 781                 // mathcommand and mathpreamble have been added for 1.6.0.
 782                 // make them optional so that old files still work.
 783                 int const lineno = symbolslex.lineNumber();
 784                 bool breakout = false;
 785                 if (symbolslex.next(true)) {
 786                         if (symbolslex.lineNumber() != lineno) {
 787                                 // line in old format without mathcommand and mathpreamble
 788                                 getNextToken = false;
 789                         } else {
 790                                 info.mathcommand = symbolslex.getDocString();
 791                                 if (symbolslex.next(true)) {
 792                                         if (symbolslex.lineNumber() != lineno) {
 793                                                 // line in new format with mathcommand only
 794                                                 getNextToken = false;
 795                                         } else {
 796                                                 // line in new format with mathcommand and mathpreamble
 797                                                 info.mathpreamble = symbolslex.getString();
 798                                         }
 799                                 } else
 800                                         breakout = true;
 801                         }
 802                 } else {
 803                         breakout = true;
 804                 }
 805
 806                 if (!info.textpreamble.empty())
 807                         info.textfeature = info.textpreamble[0] != '\\';
 808                 if (!info.mathpreamble.empty())
 809                         info.mathfeature = info.mathpreamble[0] != '\\';
 810
 811                 LYXERR(Debug::INFO, "Read unicode symbol " << symbol << " '"
 812                         << to_utf8(info.textcommand) << "' '" << info.textpreamble
 813                         << "' " << info.combining << ' ' << info.textfeature
 814                         << " '" << to_utf8(info.mathcommand) << "' '"
 815                         << info.mathpreamble << "' " << info.mathfeature);
 816
 817                 // we assume that at least one command is nonempty when using unicodesymbols
 818                 if (!info.textcommand.empty() || !info.mathcommand.empty())
 819                         unicodesymbols[symbol] = info;
 820
 821                 if (breakout)
 822                         break;
 823         }
 824
 825         // Now read the encodings
 826         enum {
 827                 et_encoding = 1,
 828                 et_end,
 829         };
 830
 831         LexerKeyword encodingtags[] = {
 832                 { "encoding", et_encoding },
 833                 { "end", et_end }
 834         };
 835
 836         Lexer lex(encodingtags);
 837         lex.setFile(encfile);
 838         lex.setContext("Encodings::read");
 839         while (lex.isOK()) {
 840                 switch (lex.lex()) {
 841                 case et_encoding:
 842                 {
 843                         lex.next();
 844                         string const name = lex.getString();
 845                         lex.next();
 846                         string const latexname = lex.getString();
 847                         lex.next();
 848                         string const guiname = lex.getString();
 849                         lex.next();
 850                         string const iconvname = lex.getString();
 851                         lex.next();
 852                         string const width = lex.getString();
 853                         bool fixedwidth = false;
 854                         if (width == "fixed")
 855                                 fixedwidth = true;
 856                         else if (width == "variable")
 857                                 fixedwidth = false;
 858                         else
 859                                 lex.printError("Unknown width");
 860
 861                         lex.next();
 862                         string const p = lex.getString();
 863                         Encoding::Package package = Encoding::none;
 864                         if (p == "none")
 865                                 package = Encoding::none;
 866                         else if (p == "inputenc")
 867                                 package = Encoding::inputenc;
 868                         else if (p == "CJK")
 869                                 package = Encoding::CJK;
 870                         else if (p == "japanese")
 871                                 package = Encoding::japanese;
 872                         else
 873                                 lex.printError("Unknown package");
 874
 875                         LYXERR(Debug::INFO, "Reading encoding " << name);
 876                         encodinglist[name] = Encoding(name, latexname,
 877                                 guiname, iconvname, fixedwidth, package);
 878
 879                         if (lex.lex() != et_end)
 880                                 lex.printError("Missing end");
 881                         break;
 882                 }
 883                 case et_end:
 884                         lex.printError("Misplaced end");
 885                         break;
 886                 case Lexer::LEX_FEOF:
 887                         break;
 888                 default:
 889                         lex.printError("Unknown tag");
 890                         break;
 891                 }
 892         }
 893 }
 894
 895
 896 } // namespace lyx