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/gettext.h"
  26 #include "support/FileName.h"
  27 #include "support/lstrings.h"
  28 #include "support/textutils.h"
  29 #include "support/unicode.h"
  30
  31 #include <boost/cstdint.hpp>
  32
  33 #include <sstream>
  34
  35 using namespace std;
  36 using namespace lyx::support;
  37
  38 namespace lyx {
  39
  40 Encodings encodings;
  41
  42 Encodings::MathCommandSet Encodings::mathcmd;
  43 Encodings::TextCommandSet Encodings::textcmd;
  44 Encodings::MathSymbolSet  Encodings::mathsym;
  45
  46 namespace {
  47
  48 char_type arabic_table[172][4] = {
  49         {0xfe80, 0xfe80, 0xfe80, 0xfe80}, // 0x0621 = hamza
  50         {0xfe81, 0xfe82, 0xfe81, 0xfe82}, // 0x0622 = ligature madda on alef
  51         {0xfe83, 0xfe84, 0xfe83, 0xfe84}, // 0x0623 = ligature hamza on alef
  52         {0xfe85, 0xfe86, 0xfe85, 0xfe86}, // 0x0624 = ligature hamza on waw
  53         {0xfe87, 0xfe88, 0xfe87, 0xfe88}, // 0x0625 = ligature hamza under alef
  54         {0xfe89, 0xfe8a, 0xfe8b, 0xfe8c}, // 0x0626 = ligature hamza on ya
  55         {0xfe8d, 0xfe8e, 0xfe8d, 0xfe8e}, // 0x0627 = alef
  56         {0xfe8f, 0xfe90, 0xfe91, 0xfe92}, // 0x0628 = baa
  57         {0xfe93, 0xfe94, 0xfe93, 0xfe94}, // 0x0629 = taa marbuta
  58         {0xfe95, 0xfe96, 0xfe97, 0xfe98}, // 0x062a = taa
  59         {0xfe99, 0xfe9a, 0xfe9b, 0xfe9c}, // 0x062b = thaa
  60         {0xfe9d, 0xfe9e, 0xfe9f, 0xfea0}, // 0x062c = jeem
  61         {0xfea1, 0xfea2, 0xfea3, 0xfea4}, // 0x062d = haa
  62         {0xfea5, 0xfea6, 0xfea7, 0xfea8}, // 0x062e = khaa
  63         {0xfea9, 0xfeaa, 0xfea9, 0xfeaa}, // 0x062f = dal
  64
  65         {0xfeab, 0xfeac, 0xfeab, 0xfeac}, // 0x0630 = thal
  66         {0xfead, 0xfeae, 0xfead, 0xfeae}, // 0x0631 = ra
  67         {0xfeaf, 0xfeb0, 0xfeaf, 0xfeb0}, // 0x0632 = zain
  68         {0xfeb1, 0xfeb2, 0xfeb3, 0xfeb4}, // 0x0633 = seen
  69         {0xfeb5, 0xfeb6, 0xfeb7, 0xfeb8}, // 0x0634 = sheen
  70         {0xfeb9, 0xfeba, 0xfebb, 0xfebc}, // 0x0635 = sad
  71         {0xfebd, 0xfebe, 0xfebf, 0xfec0}, // 0x0636 = dad
  72         {0xfec1, 0xfec2, 0xfec3, 0xfec4}, // 0x0637 = tah
  73         {0xfec5, 0xfec6, 0xfec7, 0xfec8}, // 0x0638 = zah
  74         {0xfec9, 0xfeca, 0xfecb, 0xfecc}, // 0x0639 = ain
  75         {0xfecd, 0xfece, 0xfecf, 0xfed0}, // 0x063a = ghain
  76         {0, 0, 0, 0}, // 0x063b
  77         {0, 0, 0, 0}, // 0x063c
  78         {0, 0, 0, 0}, // 0x063d
  79         {0, 0, 0, 0}, // 0x063e
  80         {0, 0, 0, 0}, // 0x063f
  81
  82         {0, 0, 0, 0}, // 0x0640
  83         {0xfed1, 0xfed2, 0xfed3, 0xfed4}, // 0x0641 = fa
  84         {0xfed5, 0xfed6, 0xfed7, 0xfed8}, // 0x0642 = qaf
  85         {0xfed9, 0xfeda, 0xfedb, 0xfedc}, // 0x0643 = kaf
  86         {0xfedd, 0xfede, 0xfedf, 0xfee0}, // 0x0644 = lam
  87         {0xfee1, 0xfee2, 0xfee3, 0xfee4}, // 0x0645 = meem
  88         {0xfee5, 0xfee6, 0xfee7, 0xfee8}, // 0x0646 = noon
  89         {0xfee9, 0xfeea, 0xfeeb, 0xfeec}, // 0x0647 = ha
  90         {0xfeed, 0xfeee, 0xfeed, 0xfeee}, // 0x0648 = waw
  91         {0xfeef, 0xfef0, 0xfeef, 0xfef0}, // 0x0649 = alef maksura
  92         {0xfef1, 0xfef2, 0xfef3, 0xfef4}, // 0x064a = ya
  93         {0x065b, 0x065b, 0x065b, 0x065b}, // 0x064b = fathatan
  94         {0x065c, 0x065c, 0x065c, 0x065c}, // 0x064c = dammatan
  95         {0x064d, 0x064d, 0x064d, 0x064d}, // 0x064d = kasratan
  96         {0x064e, 0x064e, 0x064e, 0x064e}, // 0x064e = fatha
  97         {0x064f, 0x064f, 0x064f, 0x064f}, // 0x064f = damma
  98
  99         {0x0650, 0x0650, 0x0650, 0x0650}, // 0x0650 = kasra
 100         {0x0651, 0x0651, 0x0651, 0x0651}, // 0x0651 = shadda
 101         {0x0652, 0x0652, 0x0652, 0x0652}, // 0x0652 = sukun
 102
 103         {0, 0, 0, 0}, // 0x0653
 104         {0, 0, 0, 0}, // 0x0654
 105         {0, 0, 0, 0}, // 0x0655
 106         {0, 0, 0, 0}, // 0x0656
 107         {0, 0, 0, 0}, // 0x0657
 108         {0, 0, 0, 0}, // 0x0658
 109         {0, 0, 0, 0}, // 0x0659
 110         {0, 0, 0, 0}, // 0x065a
 111         {0, 0, 0, 0}, // 0x065b
 112         {0, 0, 0, 0}, // 0x065c
 113         {0, 0, 0, 0}, // 0x065d
 114         {0, 0, 0, 0}, // 0x065e
 115         {0, 0, 0, 0}, // 0x065f
 116         {0, 0, 0, 0}, // 0x0660
 117         {0, 0, 0, 0}, // 0x0661
 118         {0, 0, 0, 0}, // 0x0662
 119         {0, 0, 0, 0}, // 0x0663
 120         {0, 0, 0, 0}, // 0x0664
 121         {0, 0, 0, 0}, // 0x0665
 122         {0, 0, 0, 0}, // 0x0666
 123         {0, 0, 0, 0}, // 0x0667
 124         {0, 0, 0, 0}, // 0x0668
 125         {0, 0, 0, 0}, // 0x0669
 126         {0, 0, 0, 0}, // 0x066a
 127         {0, 0, 0, 0}, // 0x066b
 128         {0, 0, 0, 0}, // 0x066c
 129         {0, 0, 0, 0}, // 0x066d
 130         {0, 0, 0, 0}, // 0x066e
 131         {0, 0, 0, 0}, // 0x066f
 132         {0, 0, 0, 0}, // 0x0670
 133         {0, 0, 0, 0}, // 0x0671
 134         {0, 0, 0, 0}, // 0x0672
 135         {0, 0, 0, 0}, // 0x0673
 136         {0, 0, 0, 0}, // 0x0674
 137         {0, 0, 0, 0}, // 0x0675
 138         {0, 0, 0, 0}, // 0x0676
 139         {0, 0, 0, 0}, // 0x0677
 140         {0, 0, 0, 0}, // 0x0678
 141         {0, 0, 0, 0}, // 0x0679
 142         {0, 0, 0, 0}, // 0x067a
 143         {0, 0, 0, 0}, // 0x067b
 144         {0, 0, 0, 0}, // 0x067c
 145         {0, 0, 0, 0}, // 0x067d
 146         {0xfb56, 0xfb57, 0xfb58, 0xfb59}, // 0x067e = peh
 147         {0, 0, 0, 0}, // 0x067f
 148         {0, 0, 0, 0}, // 0x0680
 149         {0, 0, 0, 0}, // 0x0681
 150         {0, 0, 0, 0}, // 0x0682
 151         {0, 0, 0, 0}, // 0x0683
 152         {0, 0, 0, 0}, // 0x0684
 153         {0, 0, 0, 0}, // 0x0685
 154         {0xfb7a, 0xfb7b, 0xfb7c, 0xfb7d}, // 0x0686 = tcheh
 155         {0, 0, 0, 0}, // 0x0687
 156         {0, 0, 0, 0}, // 0x0688
 157         {0, 0, 0, 0}, // 0x0689
 158         {0, 0, 0, 0}, // 0x068a
 159         {0, 0, 0, 0}, // 0x068b
 160         {0, 0, 0, 0}, // 0x068c
 161         {0, 0, 0, 0}, // 0x068d
 162         {0, 0, 0, 0}, // 0x068e
 163         {0, 0, 0, 0}, // 0x068f
 164         {0, 0, 0, 0}, // 0x0690
 165         {0, 0, 0, 0}, // 0x0691
 166         {0, 0, 0, 0}, // 0x0692
 167         {0, 0, 0, 0}, // 0x0693
 168         {0, 0, 0, 0}, // 0x0694
 169         {0, 0, 0, 0}, // 0x0695
 170         {0, 0, 0, 0}, // 0x0696
 171         {0, 0, 0, 0}, // 0x0697
 172         {0xfb8a, 0xfb8b, 0xfb8a, 0xfb8b}, // 0x0698 = jeh
 173         {0, 0, 0, 0}, // 0x0699
 174         {0, 0, 0, 0}, // 0x069a
 175         {0, 0, 0, 0}, // 0x069b
 176         {0, 0, 0, 0}, // 0x069c
 177         {0, 0, 0, 0}, // 0x069d
 178         {0, 0, 0, 0}, // 0x069e
 179         {0, 0, 0, 0}, // 0x069f
 180         {0, 0, 0, 0}, // 0x06a0
 181         {0, 0, 0, 0}, // 0x06a1
 182         {0, 0, 0, 0}, // 0x06a2
 183         {0, 0, 0, 0}, // 0x06a3
 184         {0, 0, 0, 0}, // 0x06a4
 185         {0, 0, 0, 0}, // 0x06a5
 186         {0, 0, 0, 0}, // 0x06a6
 187         {0, 0, 0, 0}, // 0x06a7
 188         {0, 0, 0, 0}, // 0x06a8
 189         {0xfb8e, 0xfb8f, 0xfb90, 0xfb91}, // 0x06a9 = farsi kaf
 190         {0, 0, 0, 0}, // 0x06aa
 191         {0, 0, 0, 0}, // 0x06ab
 192         {0, 0, 0, 0}, // 0x06ac
 193         {0, 0, 0, 0}, // 0x06ad
 194         {0, 0, 0, 0}, // 0x06ae
 195         {0xfb92, 0xfb93, 0xfb94, 0xfb95}, // 0x06af = gaf
 196         {0, 0, 0, 0}, // 0x06b0
 197         {0, 0, 0, 0}, // 0x06b1
 198         {0, 0, 0, 0}, // 0x06b2
 199         {0, 0, 0, 0}, // 0x06b3
 200         {0, 0, 0, 0}, // 0x06b4
 201         {0, 0, 0, 0}, // 0x06b5
 202         {0, 0, 0, 0}, // 0x06b6
 203         {0, 0, 0, 0}, // 0x06b7
 204         {0, 0, 0, 0}, // 0x06b8
 205         {0, 0, 0, 0}, // 0x06b9
 206         {0, 0, 0, 0}, // 0x06ba
 207         {0, 0, 0, 0}, // 0x06bb
 208         {0, 0, 0, 0}, // 0x06bc
 209         {0, 0, 0, 0}, // 0x06bd
 210         {0, 0, 0, 0}, // 0x06be
 211         {0, 0, 0, 0}, // 0x06bf
 212         {0, 0, 0, 0}, // 0x06c0
 213         {0, 0, 0, 0}, // 0x06c1
 214         {0, 0, 0, 0}, // 0x06c2
 215         {0, 0, 0, 0}, // 0x06c3
 216         {0, 0, 0, 0}, // 0x06c4
 217         {0, 0, 0, 0}, // 0x06c5
 218         {0, 0, 0, 0}, // 0x06c6
 219         {0, 0, 0, 0}, // 0x06c7
 220         {0, 0, 0, 0}, // 0x06c8
 221         {0, 0, 0, 0}, // 0x06c9
 222         {0, 0, 0, 0}, // 0x06ca
 223         {0, 0, 0, 0}, // 0x06cb
 224         {0xfbfc, 0xfbfd, 0xfbfe, 0xfbff} // 0x06cc = farsi yeh
 225 };
 226
 227
 228 char_type const arabic_start = 0x0621;
 229 char_type const arabic_end = 0x06cc;
 230
 231
 232 enum CharInfoFlags {
 233         ///
 234         CharInfoCombining = 1,
 235         ///
 236         CharInfoTextFeature = 2,
 237         ///
 238         CharInfoMathFeature = 4,
 239         ///
 240         CharInfoForce = 8,
 241         ///
 242         CharInfoTextNoTermination = 16,
 243         ///
 244         CharInfoMathNoTermination = 32,
 245 };
 246
 247 /// Information about a single UCS4 character
 248 struct CharInfo {
 249         /// LaTeX command (text mode) for this character
 250         docstring textcommand;
 251         /// LaTeX command (math mode) for this character
 252         docstring mathcommand;
 253         /// Needed LaTeX preamble (or feature) for text mode
 254         string textpreamble;
 255         /// Needed LaTeX preamble (or feature) for math mode
 256         string mathpreamble;
 257         /// Is this a combining character?
 258         bool combining() const { return flags & CharInfoCombining ? true : false; }
 259         /// Is \c textpreamble a feature known by LaTeXFeatures, or a raw LaTeX
 260         /// command?
 261         bool textfeature() const { return flags & CharInfoTextFeature ? true : false; }
 262         /// Is \c mathpreamble a feature known by LaTeXFeatures, or a raw LaTeX
 263         /// command?
 264         bool mathfeature() const { return flags & CharInfoMathFeature ? true : false; }
 265         /// Always force the LaTeX command, even if the encoding contains
 266         /// this character?
 267         bool force() const { return flags & CharInfoForce ? true : false; }
 268         /// TIPA shortcut
 269         string tipashortcut;
 270         /// \c textcommand needs no termination (such as {} or space).
 271         bool textnotermination() const { return flags & CharInfoTextNoTermination ? true : false; }
 272         /// \c mathcommand needs no termination (such as {} or space).
 273         bool mathnotermination() const { return flags & CharInfoMathNoTermination ? true : false; }
 274         ///
 275         unsigned int flags;
 276 };
 277
 278
 279 typedef map<char_type, CharInfo> CharInfoMap;
 280 CharInfoMap unicodesymbols;
 281
 282 typedef std::set<char_type> CharSet;
 283 CharSet forced;
 284
 285 typedef std::set<char_type> MathAlphaSet;
 286 MathAlphaSet mathalpha;
 287
 288
 289 /// The highest code point in UCS4 encoding (1<<20 + 1<<16)
 290 char_type const max_ucs4 = 0x110000;
 291
 292 } // namespace anon
 293
 294
 295 EncodingException::EncodingException(char_type c)
 296         : failed_char(c), par_id(0), pos(0)
 297 {
 298 }
 299
 300
 301 const char * EncodingException::what() const throw()
 302 {
 303         return "Could not find LaTeX command for a character";
 304 }
 305
 306
 307 Encoding::Encoding(string const & n, string const & l, string const & g,
 308                    string const & i, bool f, bool u, Encoding::Package p)
 309         : name_(n), latexName_(l), guiName_(g), iconvName_(i), fixedwidth_(f),
 310           unsafe_(u), package_(p)
 311 {
 312         if (n == "ascii") {
 313                 // ASCII can encode 128 code points and nothing else
 314                 start_encodable_ = 128;
 315                 complete_ = true;
 316         } else if (i == "UTF-8") {
 317                 // UTF8 can encode all UCS4 code points
 318                 start_encodable_ = max_ucs4;
 319                 complete_ = true;
 320         } else {
 321                 complete_ = false;
 322         }
 323 }
 324
 325
 326 void Encoding::init() const
 327 {
 328         if (complete_)
 329                 return;
 330
 331         start_encodable_ = 0;
 332         // temporarily switch off lyxerr, since we will generate iconv errors
 333         lyxerr.disable();
 334         if (fixedwidth_) {
 335                 // We do not need to check all UCS4 code points, it is enough
 336                 // if we check all 256 code points of this encoding.
 337                 for (unsigned short j = 0; j < 256; ++j) {
 338                         char const c = char(j);
 339                         vector<char_type> const ucs4 = eightbit_to_ucs4(&c, 1, iconvName_);
 340                         if (ucs4.size() != 1)
 341                                 continue;
 342                         char_type const uc = ucs4[0];
 343                         CharInfoMap::const_iterator const it = unicodesymbols.find(uc);
 344                         if (it == unicodesymbols.end() || !it->second.force())
 345                                 encodable_.insert(uc);
 346                 }
 347         } else {
 348                 // We do not know how many code points this encoding has, and
 349                 // they do not have a direct representation as a single byte,
 350                 // therefore we need to check all UCS4 code points.
 351                 // This is expensive!
 352                 for (char_type c = 0; c < max_ucs4; ++c) {
 353                         vector<char> const eightbit = ucs4_to_eightbit(&c, 1, iconvName_);
 354                         if (!eightbit.empty()) {
 355                                 CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 356                                 if (it == unicodesymbols.end() || !it->second.force())
 357                                         encodable_.insert(c);
 358                         }
 359                 }
 360         }
 361         lyxerr.enable();
 362         CharSet::iterator it = encodable_.find(start_encodable_);
 363         while (it != encodable_.end()) {
 364                 encodable_.erase(it);
 365                 ++start_encodable_;
 366                 it = encodable_.find(start_encodable_);
 367         }
 368         complete_ = true;
 369 }
 370
 371
 372 bool Encoding::encodable(char_type c) const
 373 {
 374         // assure the used encoding is properly initialized
 375         init();
 376
 377         if (iconvName_ == "UTF-8" && package_ == none)
 378                 return true;
 379         if (c < start_encodable_ && !encodings.isForced(c))
 380                 return true;
 381         if (encodable_.find(c) != encodable_.end())
 382                 return true;
 383         return false;
 384 }
 385
 386
 387 pair<docstring, bool> Encoding::latexChar(char_type c) const
 388 {
 389         if (encodable(c))
 390                 return make_pair(docstring(1, c), false);
 391
 392         // c cannot (or should not) be encoded in this encoding
 393         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 394         if (it == unicodesymbols.end())
 395                 throw EncodingException(c);
 396         // at least one of mathcommand and textcommand is nonempty
 397         if (it->second.textcommand.empty())
 398                 return make_pair(
 399                         "\\ensuremath{" + it->second.mathcommand + '}', false);
 400         return make_pair(it->second.textcommand, !it->second.textnotermination());
 401 }
 402
 403
 404 pair<docstring, docstring> Encoding::latexString(docstring const input, bool dryrun) const
 405 {
 406         docstring result;
 407         docstring uncodable;
 408         bool terminate = false;
 409         for (size_t n = 0; n < input.size(); ++n) {
 410                 try {
 411                         char_type const c = input[n];
 412                         pair<docstring, bool> latex_char = latexChar(c);
 413                         docstring const latex = latex_char.first;
 414                         if (terminate && !prefixIs(latex, '\\')
 415                             && !prefixIs(latex, '{')
 416                             && !prefixIs(latex, '}')) {
 417                                         // Prevent eating of a following
 418                                         // space or command corruption by
 419                                         // following characters
 420                                         if (latex == " ")
 421                                                 result += "{}";
 422                                         else
 423                                                 result += " ";
 424                                 }
 425                         result += latex;
 426                         terminate = latex_char.second;
 427                 } catch (EncodingException & /* e */) {
 428                         LYXERR0("Uncodable character in latexString!");
 429                         if (dryrun) {
 430                                 result += "<" + _("LyX Warning: ")
 431                                            + _("uncodable character") + " '";
 432                                 result += docstring(1, input[n]);
 433                                 result += "'>";
 434                         } else
 435                                 uncodable += input[n];
 436                 }
 437         }
 438         return make_pair(result, uncodable);
 439 }
 440
 441
 442 vector<char_type> Encoding::symbolsList() const
 443 {
 444         // assure the used encoding is properly initialized
 445         init();
 446
 447         // first all encodable characters
 448         vector<char_type> symbols(encodable_.begin(), encodable_.end());
 449         // add those below start_encodable_
 450         for (char_type c = 0; c < start_encodable_; ++c)
 451                 symbols.push_back(c);
 452         // now the ones from the unicodesymbols file
 453         CharInfoMap::const_iterator const end = unicodesymbols.end();
 454         CharInfoMap::const_iterator it = unicodesymbols.begin();
 455         for (; it != end; ++it)
 456                 symbols.push_back(it->first);
 457         return symbols;
 458 }
 459
 460
 461 bool Encodings::latexMathChar(char_type c, bool mathmode,
 462                         Encoding const * encoding, docstring & command,
 463                         bool & needsTermination)
 464 {
 465         command = empty_docstring();
 466         if (encoding)
 467                 if (encoding->encodable(c))
 468                         command = docstring(1, c);
 469         needsTermination = false;
 470
 471         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 472         if (it == unicodesymbols.end()) {
 473                 if (!encoding || command.empty())
 474                         throw EncodingException(c);
 475                 if (mathmode)
 476                         addMathSym(c);
 477                 return false;
 478         }
 479         // at least one of mathcommand and textcommand is nonempty
 480         bool use_math = (mathmode && !it->second.mathcommand.empty()) ||
 481                         (!mathmode && it->second.textcommand.empty());
 482         if (use_math) {
 483                 command = it->second.mathcommand;
 484                 needsTermination = !it->second.mathnotermination();
 485                 addMathCmd(c);
 486         } else {
 487                 if (!encoding || command.empty()) {
 488                         command = it->second.textcommand;
 489                         needsTermination = !it->second.textnotermination();
 490                         addTextCmd(c);
 491                 }
 492                 if (mathmode)
 493                         addMathSym(c);
 494         }
 495         return use_math;
 496 }
 497
 498
 499 char_type Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype,
 500                 bool & combining, bool & needsTermination, set<string> * req)
 501 {
 502         CharInfoMap::const_iterator const end = unicodesymbols.end();
 503         CharInfoMap::const_iterator it = unicodesymbols.begin();
 504         for (combining = false; it != end; ++it) {
 505                 docstring const math = it->second.mathcommand;
 506                 docstring const text = it->second.textcommand;
 507                 if ((cmdtype & MATH_CMD) && math == cmd) {
 508                         combining = it->second.combining();
 509                         needsTermination = !it->second.mathnotermination();
 510                         if (req && it->second.mathfeature() &&
 511                             !it->second.mathpreamble.empty())
 512                                 req->insert(it->second.mathpreamble);
 513                         return it->first;
 514                 }
 515                 if ((cmdtype & TEXT_CMD) && text == cmd) {
 516                         combining = it->second.combining();
 517                         needsTermination = !it->second.textnotermination();
 518                         if (req && it->second.textfeature() &&
 519                             !it->second.textpreamble.empty())
 520                                 req->insert(it->second.textpreamble);
 521                         return it->first;
 522                 }
 523         }
 524         needsTermination = false;
 525         return 0;
 526 }
 527
 528
 529 docstring Encodings::fromLaTeXCommand(docstring const & cmd, int cmdtype,
 530                 bool & needsTermination, docstring & rem, set<string> * req)
 531 {
 532         needsTermination = false;
 533         rem = empty_docstring();
 534         bool const mathmode = cmdtype & MATH_CMD;
 535         bool const textmode = cmdtype & TEXT_CMD;
 536         docstring symbols;
 537         size_t i = 0;
 538         size_t const cmdend = cmd.size();
 539         CharInfoMap::const_iterator const uniend = unicodesymbols.end();
 540         for (size_t j = 0; j < cmdend; ++j) {
 541                 // Also get the char after a backslash
 542                 if (j + 1 < cmdend && cmd[j] == '\\')
 543                         ++j;
 544                 // If a macro argument follows, get it, too
 545                 if (j + 1 < cmdend && cmd[j + 1] == '{') {
 546                         size_t k = j + 1;
 547                         int count = 1;
 548                         while (k < cmdend && count && k != docstring::npos) {
 549                                 k = cmd.find_first_of(from_ascii("{}"), k + 1);
 550                                 if (cmd[k] == '{')
 551                                         ++count;
 552                                 else
 553                                         --count;
 554                         }
 555                         if (k != docstring::npos)
 556                                 j = k;
 557                 }
 558                 // Start with this substring and try augmenting it when it is
 559                 // the prefix of some command in the unicodesymbols file
 560                 docstring const subcmd = cmd.substr(i, j - i + 1);
 561
 562                 CharInfoMap::const_iterator it = unicodesymbols.begin();
 563                 size_t unicmd_size = 0;
 564                 char_type c = 0;
 565                 for (; it != uniend; ++it) {
 566                         docstring const math = mathmode ? it->second.mathcommand
 567                                                         : docstring();
 568                         docstring const text = textmode ? it->second.textcommand
 569                                                         : docstring();
 570                         size_t cur_size = max(math.size(), text.size());
 571                         // The current math or text unicode command cannot
 572                         // match, or we already matched a longer one
 573                         if (cur_size < subcmd.size() || cur_size <= unicmd_size)
 574                                 continue;
 575
 576                         docstring tmp = subcmd;
 577                         size_t k = j;
 578                         while (prefixIs(math, tmp) || prefixIs(text, tmp)) {
 579                                 ++k;
 580                                 if (k >= cmdend || cur_size <= tmp.size())
 581                                         break;
 582                                 tmp += cmd[k];
 583                         }
 584                         // No match
 585                         if (k == j)
 586                                 continue;
 587
 588                         // The last added char caused a mismatch, because
 589                         // we didn't exhaust the chars in cmd and didn't
 590                         // exceed the maximum size of the current unicmd
 591                         if (k < cmdend && cur_size > tmp.size())
 592                                 tmp.resize(tmp.size() - 1);
 593
 594                         // If this is an exact match, we found a (longer)
 595                         // matching entry in the unicodesymbols file.
 596                         // If the entry doesn't start with '\', we take note
 597                         // of the match and continue (this is not a ultimate
 598                         // acceptance, as some other entry may match a longer
 599                         // portion of the cmd string). However, if the entry
 600                         // does start with '\', we accept the match only if
 601                         // this is a valid macro, i.e., either it is a single
 602                         // (nonletter) char macro, or nothing else follows,
 603                         // or what follows is a nonletter char, or the last
 604                         // character is a }.
 605                         if ((math == tmp || text == tmp)
 606                             && (tmp[0] != '\\'
 607                                    || (tmp.size() == 2 && !isAlphaASCII(tmp[1]))
 608                                    || k == cmdend
 609                                    || !isAlphaASCII(cmd[k])
 610                                    || tmp[tmp.size() - 1] == '}')
 611                                  ) {
 612                                 c = it->first;
 613                                 j = k - 1;
 614                                 i = j + 1;
 615                                 unicmd_size = cur_size;
 616                                 if (math == tmp)
 617                                         needsTermination = !it->second.mathnotermination();
 618                                 else
 619                                         needsTermination = !it->second.textnotermination();
 620                                 if (req) {
 621                                         if (math == tmp && it->second.mathfeature() &&
 622                                             !it->second.mathpreamble.empty())
 623                                                 req->insert(it->second.mathpreamble);
 624                                         if (text == tmp && it->second.textfeature() &&
 625                                             !it->second.textpreamble.empty())
 626                                                 req->insert(it->second.textpreamble);
 627                                 }
 628                         }
 629                 }
 630                 if (unicmd_size)
 631                         symbols += c;
 632                 else if (j + 1 == cmdend) {
 633                         // No luck. Return what remains
 634                         rem = cmd.substr(i);
 635                         if (needsTermination && !rem.empty()) {
 636                                 if (rem.substr(0, 2) == "{}") {
 637                                         rem = rem.substr(2);
 638                                         needsTermination = false;
 639                                 } else if (rem[0] == ' ') {
 640                                         needsTermination = false;
 641                                         // LaTeX would swallow all spaces
 642                                         rem = ltrim(rem);
 643                                 }
 644                         }
 645                 }
 646         }
 647         return symbols;
 648 }
 649
 650
 651 void Encodings::initUnicodeMath(Buffer const & buffer, bool for_master)
 652 {
 653 #ifdef TEX2LYX
 654         // The code below is not needed in tex2lyx and requires additional stuff
 655         (void)buffer;
 656         (void)for_master;
 657 #else
 658         if (for_master) {
 659                 mathcmd.clear();
 660                 textcmd.clear();
 661                 mathsym.clear();
 662         }
 663
 664         // Check this buffer
 665         Inset & inset = buffer.inset();
 666         InsetIterator it = inset_iterator_begin(inset);
 667         InsetIterator const end = inset_iterator_end(inset);
 668         for (; it != end; ++it)
 669                 it->initUnicodeMath();
 670
 671         if (!for_master)
 672                 return;
 673
 674         // Check children
 675         ListOfBuffers blist = buffer.getDescendents();
 676         ListOfBuffers::const_iterator bit = blist.begin();
 677         ListOfBuffers::const_iterator const bend = blist.end();
 678         for (; bit != bend; ++bit)
 679                 initUnicodeMath(**bit, false);
 680 #endif
 681 }
 682
 683
 684 void Encodings::validate(char_type c, LaTeXFeatures & features, bool for_mathed)
 685 {
 686 #ifdef TEX2LYX
 687         // The code below is not needed in tex2lyx and requires additional stuff
 688         (void)c;
 689         (void)features;
 690         (void)for_mathed;
 691 #else
 692         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 693         if (it != unicodesymbols.end()) {
 694                 // In mathed, c could be used both in textmode and mathmode
 695                 bool const math_mode = for_mathed && isMathCmd(c);
 696                 bool const use_math = math_mode ||
 697                                       (!for_mathed && it->second.textcommand.empty());
 698                 bool const use_text = (for_mathed && isTextCmd(c)) ||
 699                                       (!for_mathed && !it->second.textcommand.empty());
 700                 bool const plain_utf8 = (features.runparams().encoding->name() == "utf8-plain");
 701                 bool const unicode_math = (features.isRequired("unicode-math")
 702                         && features.isAvailable("unicode-math"));
 703                 // with utf8-plain, we only load packages when in mathed (see #7766)
 704                 // and if we do not use unicode-math
 705                 if ((math_mode && !unicode_math)
 706                      || (use_math && !plain_utf8)) {
 707                         if (!it->second.mathpreamble.empty()) {
 708                                 if (it->second.mathfeature()) {
 709                                         string feats = it->second.mathpreamble;
 710                                         while (!feats.empty()) {
 711                                                 string feat;
 712                                                 feats = split(feats, feat, ',');
 713                                                 features.require(feat);
 714                                         }
 715                                 } else
 716                                         features.addPreambleSnippet(it->second.mathpreamble);
 717                         }
 718                 }
 719                 // with utf8-plain, we do not load packages (see #7766)
 720                 if (use_text && !plain_utf8) {
 721                         if (!it->second.textpreamble.empty()) {
 722                                 if (it->second.textfeature()) {
 723                                         string feats = it->second.textpreamble;
 724                                         while (!feats.empty()) {
 725                                                 string feat;
 726                                                 feats = split(feats, feat, ',');
 727                                                 features.require(feat);
 728                                         }
 729                                 } else
 730                                         features.addPreambleSnippet(it->second.textpreamble);
 731                         }
 732                 }
 733         }
 734         if (for_mathed && isMathSym(c)) {
 735                 features.require("amstext");
 736                 features.require("lyxmathsym");
 737         }
 738 #endif
 739 }
 740
 741
 742 bool Encodings::isHebrewComposeChar(char_type c)
 743 {
 744         return c <= 0x05c2 && c >= 0x05b0 && c != 0x05be && c != 0x05c0;
 745 }
 746
 747
 748 // Special Arabic letters are ones that do not get connected from left
 749 // they are hamza, alef_madda, alef_hamza, waw_hamza, alef_hamza_under,
 750 // alef, tah_marbota, dal, thal, rah, zai, wow, alef_maksoura
 751
 752 bool Encodings::isArabicSpecialChar(char_type c)
 753 {
 754         return (c >= 0x0621 && c <= 0x0625) || (c >= 0x0630 && c <= 0x0632)
 755                 || c == 0x0627 || c == 0x0629 || c == 0x062f || c == 0x0648
 756                 || c == 0x0649 || c == 0x0698;
 757 }
 758
 759
 760 bool Encodings::isArabicComposeChar(char_type c)
 761 {
 762         return c >= 0x064b && c <= 0x0652;
 763 }
 764
 765
 766 bool Encodings::isArabicChar(char_type c)
 767 {
 768         return c >= arabic_start && c <= arabic_end
 769                 && arabic_table[c-arabic_start][0];
 770 }
 771
 772
 773 char_type Encodings::transformChar(char_type c, Encodings::LetterForm form)
 774 {
 775         return isArabicChar(c) ? arabic_table[c-arabic_start][form] : c;
 776 }
 777
 778
 779 bool Encodings::isCombiningChar(char_type c)
 780 {
 781         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 782         if (it != unicodesymbols.end())
 783                 return it->second.combining();
 784         return false;
 785 }
 786
 787
 788 string const Encodings::TIPAShortcut(char_type c)
 789 {
 790         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 791         if (it != unicodesymbols.end())
 792                 return it->second.tipashortcut;
 793         return string();
 794 }
 795
 796
 797 bool Encodings::isKnownScriptChar(char_type const c, string & preamble)
 798 {
 799         CharInfoMap::const_iterator const it = unicodesymbols.find(c);
 800
 801         if (it == unicodesymbols.end())
 802                 return false;
 803
 804         if (it->second.textpreamble != "textgreek" && it->second.textpreamble != "textcyr")
 805                 return false;
 806
 807         if (preamble.empty()) {
 808                 preamble = it->second.textpreamble;
 809                 return true;
 810         }
 811         return it->second.textpreamble == preamble;
 812 }
 813
 814
 815 bool Encodings::isForced(char_type c)
 816 {
 817         return (!forced.empty() && forced.find(c) != forced.end());
 818 }
 819
 820
 821 bool Encodings::isMathAlpha(char_type c)
 822 {
 823         return mathalpha.count(c);
 824 }
 825
 826
 827 Encoding const *
 828 Encodings::fromLyXName(string const & name, bool allowUnsafe) const
 829 {
 830         EncodingList::const_iterator const it = encodinglist.find(name);
 831         if (!allowUnsafe && it->second.unsafe())
 832                 return 0;
 833         return it != encodinglist.end() ? &it->second : 0;
 834 }
 835
 836
 837 Encoding const *
 838 Encodings::fromLaTeXName(string const & n, bool allowUnsafe) const
 839 {
 840         string name = n;
 841         // FIXME: if we have to test for too many of these synonyms,
 842         // we should instead extend the format of lib/encodings
 843         if (n == "ansinew")
 844                 name = "cp1252";
 845
 846         // We don't use find_if because it makes copies of the pairs in
 847         // the map.
 848         // This linear search is OK since we don't have many encodings.
 849         // Users could even optimize it by putting the encodings they use
 850         // most at the top of lib/encodings.
 851         EncodingList::const_iterator const end = encodinglist.end();
 852         for (EncodingList::const_iterator it = encodinglist.begin(); it != end; ++it)
 853                 if (it->second.latexName() == name) {
 854                         if (!allowUnsafe && it->second.unsafe())
 855                                 return 0;
 856                         return &it->second;
 857                 }
 858         return 0;
 859 }
 860
 861
 862 Encodings::Encodings()
 863 {}
 864
 865
 866 void Encodings::read(FileName const & encfile, FileName const & symbolsfile)
 867 {
 868         // We must read the symbolsfile first, because the Encoding
 869         // constructor depends on it.
 870         Lexer symbolslex;
 871         symbolslex.setFile(symbolsfile);
 872         bool getNextToken = true;
 873         while (symbolslex.isOK()) {
 874                 char_type symbol;
 875                 CharInfo info;
 876                 string flags;
 877
 878                 if (getNextToken) {
 879                         if (!symbolslex.next(true))
 880                                 break;
 881                 } else
 882                         getNextToken = true;
 883
 884                 istringstream is(symbolslex.getString());
 885                 // reading symbol directly does not work if
 886                 // char_type == wchar_t.
 887                 boost::uint32_t tmp;
 888                 if(!(is >> hex >> tmp))
 889                         break;
 890                 symbol = tmp;
 891
 892                 if (!symbolslex.next(true))
 893                         break;
 894                 info.textcommand = symbolslex.getDocString();
 895                 if (!symbolslex.next(true))
 896                         break;
 897                 info.textpreamble = symbolslex.getString();
 898                 if (!symbolslex.next(true))
 899                         break;
 900                 flags = symbolslex.getString();
 901
 902                 info.flags = 0;
 903                 if (suffixIs(info.textcommand, '}'))
 904                         info.flags |= CharInfoTextNoTermination;
 905                 if (suffixIs(info.mathcommand, '}'))
 906                         info.flags |= CharInfoMathNoTermination;
 907                 while (!flags.empty()) {
 908                         string flag;
 909                         flags = split(flags, flag, ',');
 910                         if (flag == "combining") {
 911                                 info.flags |= CharInfoCombining;
 912                         } else if (flag == "force") {
 913                                 info.flags |= CharInfoForce;
 914                                 forced.insert(symbol);
 915                         } else if (flag == "mathalpha") {
 916                                 mathalpha.insert(symbol);
 917                         } else if (flag == "notermination=text") {
 918                                 info.flags |= CharInfoTextNoTermination;
 919                         } else if (flag == "notermination=math") {
 920                                 info.flags |= CharInfoMathNoTermination;
 921                         } else if (flag == "notermination=both") {
 922                                 info.flags |= CharInfoTextNoTermination;
 923                                 info.flags |= CharInfoMathNoTermination;
 924                         } else if (flag == "notermination=none") {
 925                                 info.flags &= ~CharInfoTextNoTermination;
 926                                 info.flags &= ~CharInfoMathNoTermination;
 927                         } else if (contains(flag, "tipashortcut=")) {
 928                                 info.tipashortcut = split(flag, '=');
 929                         } else {
 930                                 lyxerr << "Ignoring unknown flag `" << flag
 931                                        << "' for symbol `0x"
 932                                        << hex << symbol << dec
 933                                        << "'." << endl;
 934                         }
 935                 }
 936                 // mathcommand and mathpreamble have been added for 1.6.0.
 937                 // make them optional so that old files still work.
 938                 int const lineno = symbolslex.lineNumber();
 939                 bool breakout = false;
 940                 if (symbolslex.next(true)) {
 941                         if (symbolslex.lineNumber() != lineno) {
 942                                 // line in old format without mathcommand and mathpreamble
 943                                 getNextToken = false;
 944                         } else {
 945                                 info.mathcommand = symbolslex.getDocString();
 946                                 if (symbolslex.next(true)) {
 947                                         if (symbolslex.lineNumber() != lineno) {
 948                                                 // line in new format with mathcommand only
 949                                                 getNextToken = false;
 950                                         } else {
 951                                                 // line in new format with mathcommand and mathpreamble
 952                                                 info.mathpreamble = symbolslex.getString();
 953                                         }
 954                                 } else
 955                                         breakout = true;
 956                         }
 957                 } else {
 958                         breakout = true;
 959                 }
 960
 961                 // backward compatibility
 962                 if (info.mathpreamble == "esintoramsmath")
 963                         info.mathpreamble = "esint|amsmath";
 964
 965                 if (!info.textpreamble.empty())
 966                         if (info.textpreamble[0] != '\\')
 967                                 info.flags |= CharInfoTextFeature;
 968                 if (!info.mathpreamble.empty())
 969                         if (info.mathpreamble[0] != '\\')
 970                                 info.flags |= CharInfoMathFeature;
 971
 972                 LYXERR(Debug::INFO, "Read unicode symbol " << symbol << " '"
 973                         << to_utf8(info.textcommand) << "' '" << info.textpreamble
 974                         << " '" << info.textfeature() << ' ' << info.textnotermination()
 975                         << ' ' << to_utf8(info.mathcommand) << "' '" << info.mathpreamble
 976                         << "' " << info.mathfeature() << ' ' << info.mathnotermination()
 977                         << ' ' << info.combining() << ' ' << info.force());
 978
 979                 // we assume that at least one command is nonempty when using unicodesymbols
 980                 if (!info.textcommand.empty() || !info.mathcommand.empty())
 981                         unicodesymbols[symbol] = info;
 982
 983                 if (breakout)
 984                         break;
 985         }
 986
 987         // Now read the encodings
 988         enum {
 989                 et_encoding = 1,
 990                 et_end
 991         };
 992
 993         LexerKeyword encodingtags[] = {
 994                 { "encoding", et_encoding },
 995                 { "end", et_end }
 996         };
 997
 998         Lexer lex(encodingtags);
 999         lex.setFile(encfile);
1000         lex.setContext("Encodings::read");
1001         while (lex.isOK()) {
1002                 switch (lex.lex()) {
1003                 case et_encoding:
1004                 {
1005                         lex.next();
1006                         string const name = lex.getString();
1007                         lex.next();
1008                         string const latexname = lex.getString();
1009                         lex.next();
1010                         string const guiname = lex.getString();
1011                         lex.next();
1012                         string const iconvname = lex.getString();
1013                         lex.next();
1014                         string const width = lex.getString();
1015                         bool fixedwidth = false;
1016                         bool unsafe = false;
1017                         if (width == "fixed")
1018                                 fixedwidth = true;
1019                         else if (width == "variable")
1020                                 fixedwidth = false;
1021                         else if (width == "variableunsafe") {
1022                                 fixedwidth = false;
1023                                 unsafe = true;
1024                         }
1025                         else
1026                                 lex.printError("Unknown width");
1027
1028                         lex.next();
1029                         string const p = lex.getString();
1030                         Encoding::Package package = Encoding::none;
1031                         if (p == "none")
1032                                 package = Encoding::none;
1033                         else if (p == "inputenc")
1034                                 package = Encoding::inputenc;
1035                         else if (p == "CJK")
1036                                 package = Encoding::CJK;
1037                         else if (p == "japanese")
1038                                 package = Encoding::japanese;
1039                         else
1040                                 lex.printError("Unknown package");
1041
1042                         LYXERR(Debug::INFO, "Reading encoding " << name);
1043                         encodinglist[name] = Encoding(name, latexname,
1044                                 guiname, iconvname, fixedwidth, unsafe,
1045                                 package);
1046
1047                         if (lex.lex() != et_end)
1048                                 lex.printError("Missing end");
1049                         break;
1050                 }
1051                 case et_end:
1052                         lex.printError("Misplaced end");
1053                         break;
1054                 case Lexer::LEX_FEOF:
1055                         break;
1056                 default:
1057                         lex.printError("Unknown tag");
1058                         break;
1059                 }
1060         }
1061 }
1062
1063
1064 } // namespace lyx