src/mathed/math_parser.C

   1 /**
   2  * \file math_parser.C
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author André Pönitz
   7  *
   8  * Full author contact details are available in file CREDITS.
   9  */
  10
  11 /*
  12
  13 If someone desperately needs partial "structures" (such as a few
  14 cells of an array inset or similar) (s)he could uses the
  15 following hack as starting point to write some macros:
  16
  17   \newif\ifcomment
  18   \commentfalse
  19   \ifcomment
  20           \def\makeamptab{\catcode`\&=4\relax}
  21           \def\makeampletter{\catcode`\&=11\relax}
  22     \def\b{\makeampletter\expandafter\makeamptab\bi}
  23     \long\def\bi#1\e{}
  24   \else
  25     \def\b{}\def\e{}
  26   \fi
  27
  28   ...
  29
  30   \[\begin{array}{ccc}
  31 1
  32 &
  33
  34   \end{array}\]
  35
  36 */
  37
  38
  39 #include <config.h>
  40
  41 #include "math_parser.h"
  42 #include "math_arrayinset.h"
  43 #include "math_braceinset.h"
  44 #include "math_charinset.h"
  45 #include "math_commentinset.h"
  46 #include "math_deliminset.h"
  47 #include "math_envinset.h"
  48 #include "math_factory.h"
  49 #include "math_kerninset.h"
  50 #include "math_macro.h"
  51 #include "math_macroarg.h"
  52 #include "math_macrotemplate.h"
  53 #include "math_parinset.h"
  54 #include "math_rootinset.h"
  55 #include "math_scriptinset.h"
  56 #include "math_sqrtinset.h"
  57 #include "math_support.h"
  58 #include "math_tabularinset.h"
  59
  60 //#include "insets/insetref.h"
  61 #include "ref_inset.h"
  62
  63 #include "lyxlex.h"
  64 #include "debug.h"
  65
  66 #include <sstream>
  67
  68 #ifndef CXX_GLOBAL_CSTD
  69 using std::atoi;
  70 #endif
  71 using std::endl;
  72 using std::fill;
  73
  74 using std::string;
  75 using std::ios;
  76 using std::istream;
  77 using std::istringstream;
  78 using std::ostream;
  79 using std::vector;
  80
  81
  82 //#define FILEDEBUG
  83
  84
  85 namespace {
  86
  87 MathInset::mode_type asMode(MathInset::mode_type oldmode, string const & str)
  88 {
  89         //lyxerr << "handling mode: '" << str << "'" << endl;
  90         if (str == "mathmode")
  91                 return MathInset::MATH_MODE;
  92         if (str == "textmode" || str == "forcetext")
  93                 return MathInset::TEXT_MODE;
  94         return oldmode;
  95 }
  96
  97
  98 bool stared(string const & s)
  99 {
 100         string::size_type const n = s.size();
 101         return n && s[n - 1] == '*';
 102 }
 103
 104
 105 // These are TeX's catcodes
 106 enum CatCode {
 107         catEscape,     // 0    backslash
 108         catBegin,      // 1    {
 109         catEnd,        // 2    }
 110         catMath,       // 3    $
 111         catAlign,      // 4    &
 112         catNewline,    // 5    ^^M
 113         catParameter,  // 6    #
 114         catSuper,      // 7    ^
 115         catSub,        // 8    _
 116         catIgnore,     // 9
 117         catSpace,      // 10   space
 118         catLetter,     // 11   a-zA-Z
 119         catOther,      // 12   none of the above
 120         catActive,     // 13   ~
 121         catComment,    // 14   %
 122         catInvalid     // 15   <delete>
 123 };
 124
 125 CatCode theCatcode[256];
 126
 127
 128 inline CatCode catcode(unsigned char c)
 129 {
 130         return theCatcode[c];
 131 }
 132
 133
 134 enum {
 135         FLAG_BRACE_LAST = 1 << 1,  //  last closing brace ends the parsing
 136         FLAG_RIGHT      = 1 << 2,  //  next \\right ends the parsing process
 137         FLAG_END        = 1 << 3,  //  next \\end ends the parsing process
 138         FLAG_BRACK_LAST = 1 << 4,  //  next closing bracket ends the parsing
 139         FLAG_TEXTMODE   = 1 << 5,  //  we are in a box
 140         FLAG_ITEM       = 1 << 6,  //  read a (possibly braced token)
 141         FLAG_LEAVE      = 1 << 7,  //  leave the loop at the end
 142         FLAG_SIMPLE     = 1 << 8,  //  next $ leaves the loop
 143         FLAG_EQUATION   = 1 << 9,  //  next \] leaves the loop
 144         FLAG_SIMPLE2    = 1 << 10, //  next \) leaves the loop
 145         FLAG_OPTION     = 1 << 11, //  read [...] style option
 146         FLAG_BRACED     = 1 << 12  //  read {...} style argument
 147 };
 148
 149
 150 //
 151 // Helper class for parsing
 152 //
 153
 154 class Token {
 155 public:
 156         ///
 157         Token() : cs_(), char_(0), cat_(catIgnore) {}
 158         ///
 159         Token(char c, CatCode cat) : cs_(), char_(c), cat_(cat) {}
 160         ///
 161         Token(string const & cs) : cs_(cs), char_(0), cat_(catIgnore) {}
 162
 163         ///
 164         string const & cs() const { return cs_; }
 165         ///
 166         CatCode cat() const { return cat_; }
 167         ///
 168         char character() const { return char_; }
 169         ///
 170         string asString() const { return cs_.size() ? cs_ : string(1, char_); }
 171
 172 private:
 173         ///
 174         string cs_;
 175         ///
 176         char char_;
 177         ///
 178         CatCode cat_;
 179 };
 180
 181 ostream & operator<<(ostream & os, Token const & t)
 182 {
 183         if (t.cs().size())
 184                 os << '\\' << t.cs();
 185         else if (t.cat() == catLetter)
 186                 os << t.character();
 187         else
 188                 os << '[' << t.character() << ',' << t.cat() << ']';
 189         return os;
 190 }
 191
 192
 193 class Parser {
 194 public:
 195         ///
 196         typedef  MathInset::mode_type mode_type;
 197
 198         ///
 199         Parser(LyXLex & lex);
 200         ///
 201         Parser(istream & is);
 202
 203         ///
 204         bool parse(MathAtom & at);
 205         ///
 206         void parse(MathArray & array, unsigned flags, mode_type mode);
 207         ///
 208         void parse1(MathGridInset & grid, unsigned flags, mode_type mode,
 209                 bool numbered);
 210         ///
 211         MathArray parse(unsigned flags, mode_type mode);
 212         ///
 213         int lineno() const { return lineno_; }
 214         ///
 215         void putback();
 216
 217 private:
 218         ///
 219         void parse2(MathAtom & at, unsigned flags, mode_type mode, bool numbered);
 220         /// get arg delimited by 'left' and 'right'
 221         string getArg(char left, char right);
 222         ///
 223         char getChar();
 224         ///
 225         void error(string const & msg);
 226         /// dump contents to screen
 227         void dump() const;
 228         ///
 229         void tokenize(istream & is);
 230         ///
 231         void tokenize(string const & s);
 232         ///
 233         void skipSpaceTokens(istream & is, char c);
 234         ///
 235         void push_back(Token const & t);
 236         ///
 237         void pop_back();
 238         ///
 239         Token const & prevToken() const;
 240         ///
 241         Token const & nextToken() const;
 242         ///
 243         Token const & getToken();
 244         /// skips spaces if any
 245         void skipSpaces();
 246         ///
 247         void lex(string const & s);
 248         ///
 249         bool good() const;
 250         ///
 251         string parse_verbatim_item();
 252         ///
 253         string parse_verbatim_option();
 254
 255         ///
 256         int lineno_;
 257         ///
 258         vector<Token> tokens_;
 259         ///
 260         unsigned pos_;
 261 };
 262
 263
 264 Parser::Parser(LyXLex & lexer)
 265         : lineno_(lexer.getLineNo()), pos_(0)
 266 {
 267         tokenize(lexer.getStream());
 268         lexer.eatLine();
 269 }
 270
 271
 272 Parser::Parser(istream & is)
 273         : lineno_(0), pos_(0)
 274 {
 275         tokenize(is);
 276 }
 277
 278
 279 void Parser::push_back(Token const & t)
 280 {
 281         tokens_.push_back(t);
 282 }
 283
 284
 285 void Parser::pop_back()
 286 {
 287         tokens_.pop_back();
 288 }
 289
 290
 291 Token const & Parser::prevToken() const
 292 {
 293         static const Token dummy;
 294         return pos_ > 0 ? tokens_[pos_ - 1] : dummy;
 295 }
 296
 297
 298 Token const & Parser::nextToken() const
 299 {
 300         static const Token dummy;
 301         return good() ? tokens_[pos_] : dummy;
 302 }
 303
 304
 305 Token const & Parser::getToken()
 306 {
 307         static const Token dummy;
 308         //lyxerr << "looking at token " << tokens_[pos_] << " pos: " << pos_ << endl;
 309         return good() ? tokens_[pos_++] : dummy;
 310 }
 311
 312
 313 void Parser::skipSpaces()
 314 {
 315         while (nextToken().cat() == catSpace || nextToken().cat() == catNewline)
 316                 getToken();
 317 }
 318
 319
 320 void Parser::putback()
 321 {
 322         --pos_;
 323 }
 324
 325
 326 bool Parser::good() const
 327 {
 328         return pos_ < tokens_.size();
 329 }
 330
 331
 332 char Parser::getChar()
 333 {
 334         if (!good())
 335                 error("The input stream is not well...");
 336         return tokens_[pos_++].character();
 337 }
 338
 339
 340 string Parser::getArg(char left, char right)
 341 {
 342         skipSpaces();
 343
 344         string result;
 345         char c = getChar();
 346
 347         if (c != left)
 348                 putback();
 349         else
 350                 while ((c = getChar()) != right && good())
 351                         result += c;
 352
 353         return result;
 354 }
 355
 356
 357 void Parser::skipSpaceTokens(istream & is, char c)
 358 {
 359         // skip trailing spaces
 360         while (catcode(c) == catSpace || catcode(c) == catNewline)
 361                 if (!is.get(c))
 362                         break;
 363         //lyxerr << "putting back: " << c << endl;
 364         is.putback(c);
 365 }
 366
 367
 368 void Parser::tokenize(istream & is)
 369 {
 370         // eat everything up to the next \end_inset or end of stream
 371         // and store it in s for further tokenization
 372         string s;
 373         char c;
 374         while (is.get(c)) {
 375                 s += c;
 376                 if (s.size() >= 10 && s.substr(s.size() - 10) == "\\end_inset") {
 377                         s = s.substr(0, s.size() - 10);
 378                         break;
 379                 }
 380         }
 381         // Remove the space after \end_inset
 382         if (is.get(c) && c != ' ')
 383                 is.unget();
 384
 385         // tokenize buffer
 386         tokenize(s);
 387 }
 388
 389
 390 void Parser::tokenize(string const & buffer)
 391 {
 392         istringstream is(buffer, ios::in | ios::binary);
 393
 394         char c;
 395         while (is.get(c)) {
 396                 //lyxerr << "reading c: " << c << endl;
 397
 398                 switch (catcode(c)) {
 399                         case catNewline: {
 400                                 ++lineno_;
 401                                 is.get(c);
 402                                 if (catcode(c) == catNewline)
 403                                         ; //push_back(Token("par"));
 404                                 else {
 405                                         push_back(Token('\n', catNewline));
 406                                         is.putback(c);
 407                                 }
 408                                 break;
 409                         }
 410
 411 /*
 412                         case catComment: {
 413                                 while (is.get(c) && catcode(c) != catNewline)
 414                                         ;
 415                                 ++lineno_;
 416                                 break;
 417                         }
 418 */
 419
 420                         case catEscape: {
 421                                 is.get(c);
 422                                 if (!is) {
 423                                         error("unexpected end of input");
 424                                 } else {
 425                                         string s(1, c);
 426                                         if (catcode(c) == catLetter) {
 427                                                 // collect letters
 428                                                 while (is.get(c) && catcode(c) == catLetter)
 429                                                         s += c;
 430                                                 skipSpaceTokens(is, c);
 431                                         }
 432                                         push_back(Token(s));
 433                                 }
 434                                 break;
 435                         }
 436
 437                         case catSuper:
 438                         case catSub: {
 439                                 push_back(Token(c, catcode(c)));
 440                                 is.get(c);
 441                                 skipSpaceTokens(is, c);
 442                                 break;
 443                         }
 444
 445                         case catIgnore: {
 446                                 lyxerr << "ignoring a char: " << int(c) << endl;
 447                                 break;
 448                         }
 449
 450                         default:
 451                                 push_back(Token(c, catcode(c)));
 452                 }
 453         }
 454
 455 #ifdef FILEDEBUG
 456         dump();
 457 #endif
 458 }
 459
 460
 461 void Parser::dump() const
 462 {
 463         lyxerr << "\nTokens: ";
 464         for (unsigned i = 0; i < tokens_.size(); ++i) {
 465                 if (i == pos_)
 466                         lyxerr << " <#> ";
 467                 lyxerr << tokens_[i];
 468         }
 469         lyxerr << " pos: " << pos_ << endl;
 470 }
 471
 472
 473 void Parser::error(string const & msg)
 474 {
 475         lyxerr << "Line ~" << lineno_ << ": Math parse error: " << msg << endl;
 476         dump();
 477         //exit(1);
 478 }
 479
 480
 481 bool Parser::parse(MathAtom & at)
 482 {
 483         skipSpaces();
 484         MathArray ar;
 485         parse(ar, false, MathInset::UNDECIDED_MODE);
 486         if (ar.size() != 1 || ar.front()->getType() == "none") {
 487                 lyxerr << "unusual contents found: " << ar << endl;
 488                 at = MathAtom(new MathParInset(ar));
 489                 //if (at->nargs() > 0)
 490                 //      at.nucleus()->cell(0) = ar;
 491                 //else
 492                 //      lyxerr << "unusual contents found: " << ar << endl;
 493                 return true;
 494         }
 495         at = ar[0];
 496         return true;
 497 }
 498
 499
 500 string Parser::parse_verbatim_option()
 501 {
 502         skipSpaces();
 503         string res;
 504         if (nextToken().character() == '[') {
 505                 Token t = getToken();
 506                 for (Token t = getToken(); t.character() != ']' && good(); t = getToken()) {
 507                         if (t.cat() == catBegin) {
 508                                 putback();
 509                                 res += '{' + parse_verbatim_item() + '}';
 510                         } else
 511                                 res += t.asString();
 512                 }
 513         }
 514         return res;
 515 }
 516
 517
 518 string Parser::parse_verbatim_item()
 519 {
 520         skipSpaces();
 521         string res;
 522         if (nextToken().cat() == catBegin) {
 523                 Token t = getToken();
 524                 for (Token t = getToken(); t.cat() != catEnd && good(); t = getToken()) {
 525                         if (t.cat() == catBegin) {
 526                                 putback();
 527                                 res += '{' + parse_verbatim_item() + '}';
 528                         }
 529                         else
 530                                 res += t.asString();
 531                 }
 532         }
 533         return res;
 534 }
 535
 536
 537 MathArray Parser::parse(unsigned flags, mode_type mode)
 538 {
 539         MathArray ar;
 540         parse(ar, flags, mode);
 541         return ar;
 542 }
 543
 544
 545 void Parser::parse(MathArray & array, unsigned flags, mode_type mode)
 546 {
 547         MathGridInset grid(1, 1);
 548         parse1(grid, flags, mode, false);
 549         array = grid.cell(0);
 550 }
 551
 552
 553 void Parser::parse2(MathAtom & at, const unsigned flags, const mode_type mode,
 554         const bool numbered)
 555 {
 556         parse1(*(at.nucleus()->asGridInset()), flags, mode, numbered);
 557 }
 558
 559
 560 void Parser::parse1(MathGridInset & grid, unsigned flags,
 561         const mode_type mode, const bool numbered)
 562 {
 563         int limits = 0;
 564         MathGridInset::row_type cellrow = 0;
 565         MathGridInset::col_type cellcol = 0;
 566         MathArray * cell = &grid.cell(grid.index(cellrow, cellcol));
 567
 568         if (grid.asHullInset())
 569                 grid.asHullInset()->numbered(cellrow, numbered);
 570
 571         //dump();
 572         //lyxerr << " flags: " << flags << endl;
 573         //lyxerr << " mode: " << mode  << endl;
 574         //lyxerr << "grid: " << grid << endl;
 575
 576         while (good()) {
 577                 Token const & t = getToken();
 578
 579 #ifdef FILEDEBUG
 580                 lyxerr << "t: " << t << " flags: " << flags << endl;
 581                 lyxerr << "mode: " << mode  << endl;
 582                 cell->dump();
 583                 lyxerr << endl;
 584 #endif
 585
 586                 if (flags & FLAG_ITEM) {
 587
 588                         if (t.cat() == catBegin) {
 589                                 // skip the brace and collect everything to the next matching
 590                                 // closing brace
 591                                 parse1(grid, FLAG_BRACE_LAST, mode, numbered);
 592                                 return;
 593                         }
 594
 595                         // handle only this single token, leave the loop if done
 596                         flags = FLAG_LEAVE;
 597                 }
 598
 599
 600                 if (flags & FLAG_BRACED) {
 601                         if (t.cat() == catSpace)
 602                                 continue;
 603
 604                         if (t.cat() != catBegin) {
 605                                 error("opening brace expected");
 606                                 return;
 607                         }
 608
 609                         // skip the brace and collect everything to the next matching
 610                         // closing brace
 611                         flags = FLAG_BRACE_LAST;
 612                 }
 613
 614
 615                 if (flags & FLAG_OPTION) {
 616                         if (t.cat() == catOther && t.character() == '[') {
 617                                 MathArray ar;
 618                                 parse(ar, FLAG_BRACK_LAST, mode);
 619                                 cell->append(ar);
 620                         } else {
 621                                 // no option found, put back token and we are done
 622                                 putback();
 623                         }
 624                         return;
 625                 }
 626
 627                 //
 628                 // cat codes
 629                 //
 630                 if (t.cat() == catMath) {
 631                         if (mode != MathInset::MATH_MODE) {
 632                                 // we are inside some text mode thingy, so opening new math is allowed
 633                                 Token const & n = getToken();
 634                                 if (n.cat() == catMath) {
 635                                         // TeX's $$...$$ syntax for displayed math
 636                                         cell->push_back(MathAtom(new MathHullInset("equation")));
 637                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 638                                         getToken(); // skip the second '$' token
 639                                 } else {
 640                                         // simple $...$  stuff
 641                                         putback();
 642                                         cell->push_back(MathAtom(new MathHullInset("simple")));
 643                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 644                                 }
 645                         }
 646
 647                         else if (flags & FLAG_SIMPLE) {
 648                                 // this is the end of the formula
 649                                 return;
 650                         }
 651
 652                         else {
 653                                 error("something strange in the parser");
 654                                 break;
 655                         }
 656                 }
 657
 658                 else if (t.cat() == catLetter)
 659                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 660
 661                 else if (t.cat() == catSpace && mode != MathInset::MATH_MODE) {
 662                         if (cell->empty() || cell->back()->getChar() != ' ')
 663                                 cell->push_back(MathAtom(new MathCharInset(t.character())));
 664                 }
 665
 666                 else if (t.cat() == catNewline && mode != MathInset::MATH_MODE) {
 667                         if (cell->empty() || cell->back()->getChar() != ' ')
 668                                 cell->push_back(MathAtom(new MathCharInset(' ')));
 669                 }
 670
 671                 else if (t.cat() == catParameter) {
 672                         Token const & n = getToken();
 673                         cell->push_back(MathAtom(new MathMacroArgument(n.character()-'0')));
 674                 }
 675
 676                 else if (t.cat() == catActive)
 677                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 678
 679                 else if (t.cat() == catBegin) {
 680                         MathArray ar;
 681                         parse(ar, FLAG_BRACE_LAST, mode);
 682                         // do not create a BraceInset if they were written by LyX
 683                         // this helps to keep the annoyance of  "a choose b"  to a minimum
 684                         if (ar.size() == 1 && ar[0]->extraBraces())
 685                                 cell->append(ar);
 686                         else
 687                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
 688                 }
 689
 690                 else if (t.cat() == catEnd) {
 691                         if (flags & FLAG_BRACE_LAST)
 692                                 return;
 693                         error("found '}' unexpectedly");
 694                         //BOOST_ASSERT(false);
 695                         //add(cell, '}', LM_TC_TEX);
 696                 }
 697
 698                 else if (t.cat() == catAlign) {
 699                         ++cellcol;
 700                         //lyxerr << " column now " << cellcol << " max: " << grid.ncols() << endl;
 701                         if (cellcol == grid.ncols()) {
 702                                 //lyxerr << "adding column " << cellcol << endl;
 703                                 grid.addCol(cellcol - 1);
 704                         }
 705                         cell = &grid.cell(grid.index(cellrow, cellcol));
 706                 }
 707
 708                 else if (t.cat() == catSuper || t.cat() == catSub) {
 709                         bool up = (t.cat() == catSuper);
 710                         // we need no new script inset if the last thing was a scriptinset,
 711                         // which has that script already not the same script already
 712                         if (!cell->size())
 713                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 714                         else if (cell->back()->asScriptInset() &&
 715                                         !cell->back()->asScriptInset()->has(up))
 716                                 cell->back().nucleus()->asScriptInset()->ensure(up);
 717                         else if (cell->back()->asScriptInset())
 718                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 719                         else
 720                                 cell->back() = MathAtom(new MathScriptInset(cell->back(), up));
 721                         MathScriptInset * p = cell->back().nucleus()->asScriptInset();
 722                         // special handling of {}-bases
 723                         // is this always correct?
 724                         // It appears that this is wrong (Dekel)
 725                         //if (p->nuc().size() == 1 && p->nuc().back()->asNestInset() &&
 726                         //    p->nuc().back()->extraBraces())
 727                         //      p->nuc() = p->nuc().back()->asNestInset()->cell(0);
 728                         parse(p->cell(p->idxOfScript(up)), FLAG_ITEM, mode);
 729                         if (limits) {
 730                                 p->limits(limits);
 731                                 limits = 0;
 732                         }
 733                 }
 734
 735                 else if (t.character() == ']' && (flags & FLAG_BRACK_LAST)) {
 736                         //lyxerr << "finished reading option" << endl;
 737                         return;
 738                 }
 739
 740                 else if (t.cat() == catOther)
 741                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 742
 743                 else if (t.cat() == catComment) {
 744                         string s;
 745                         while (good()) {
 746                                 Token const & t = getToken();
 747                                 if (t.cat() == catNewline)
 748                                         break;
 749                                 s += t.asString();
 750                         }
 751                         cell->push_back(MathAtom(new MathCommentInset(s)));
 752                         skipSpaces();
 753                 }
 754
 755                 //
 756                 // control sequences
 757                 //
 758
 759                 else if (t.cs() == "lyxlock") {
 760                         if (cell->size())
 761                                 cell->back().nucleus()->lock(true);
 762                 }
 763
 764                 else if (t.cs() == "def" ||
 765                         t.cs() == "newcommand" ||
 766                         t.cs() == "renewcommand")
 767                 {
 768                         string const type = t.cs();
 769                         string name;
 770                         int nargs = 0;
 771                         if (t.cs() == "def") {
 772                                 // get name
 773                                 name = getToken().cs();
 774
 775                                 // read parameter
 776                                 string pars;
 777                                 while (good() && nextToken().cat() != catBegin) {
 778                                         pars += getToken().cs();
 779                                         ++nargs;
 780                                 }
 781                                 nargs /= 2;
 782                                 //lyxerr << "read \\def parameter list '" << pars << "'" << endl;
 783
 784                         } else { // t.cs() == "newcommand" || t.cs() == "renewcommand"
 785
 786                                 if (getToken().cat() != catBegin) {
 787                                         error("'{' in \\newcommand expected (1) ");
 788                                         return;
 789                                 }
 790
 791                                 name = getToken().cs();
 792
 793                                 if (getToken().cat() != catEnd) {
 794                                         error("'}' in \\newcommand expected");
 795                                         return;
 796                                 }
 797
 798                                 string arg  = getArg('[', ']');
 799                                 if (!arg.empty())
 800                                         nargs = atoi(arg.c_str());
 801
 802                         }
 803
 804                         MathArray ar1;
 805                         parse(ar1, FLAG_ITEM, MathInset::UNDECIDED_MODE);
 806
 807                         // we cannot handle recursive stuff at all
 808                         //MathArray test;
 809                         //test.push_back(createMathInset(name));
 810                         //if (ar1.contains(test)) {
 811                         //      error("we cannot handle recursive macros at all.");
 812                         //      return;
 813                         //}
 814
 815                         // is a version for display attached?
 816                         skipSpaces();
 817                         MathArray ar2;
 818                         if (nextToken().cat() == catBegin)
 819                                 parse(ar2, FLAG_ITEM, MathInset::MATH_MODE);
 820
 821                         cell->push_back(MathAtom(new MathMacroTemplate(name, nargs, type,
 822                                 ar1, ar2)));
 823                 }
 824
 825                 else if (t.cs() == "(") {
 826                         cell->push_back(MathAtom(new MathHullInset("simple")));
 827                         parse2(cell->back(), FLAG_SIMPLE2, MathInset::MATH_MODE, false);
 828                 }
 829
 830                 else if (t.cs() == "[") {
 831                         cell->push_back(MathAtom(new MathHullInset("equation")));
 832                         parse2(cell->back(), FLAG_EQUATION, MathInset::MATH_MODE, false);
 833                 }
 834
 835                 else if (t.cs() == "protect")
 836                         // ignore \\protect, will hopefully be re-added during output
 837                         ;
 838
 839                 else if (t.cs() == "end") {
 840                         if (flags & FLAG_END) {
 841                                 // eat environment name
 842                                 //string const name =
 843                                 getArg('{', '}');
 844                                 // FIXME: check that we ended the correct environment
 845                                 return;
 846                         }
 847                         error("found 'end' unexpectedly");
 848                 }
 849
 850                 else if (t.cs() == ")") {
 851                         if (flags & FLAG_SIMPLE2)
 852                                 return;
 853                         error("found '\\)' unexpectedly");
 854                 }
 855
 856                 else if (t.cs() == "]") {
 857                         if (flags & FLAG_EQUATION)
 858                                 return;
 859                         error("found '\\]' unexpectedly");
 860                 }
 861
 862                 else if (t.cs() == "\\") {
 863                         grid.vcrskip(LyXLength(getArg('[', ']')), cellrow);
 864                         ++cellrow;
 865                         cellcol = 0;
 866                         if (cellrow == grid.nrows())
 867                                 grid.addRow(cellrow - 1);
 868                         if (grid.asHullInset())
 869                                 grid.asHullInset()->numbered(cellrow, numbered);
 870                         cell = &grid.cell(grid.index(cellrow, cellcol));
 871                 }
 872
 873 #if 0
 874                 else if (t.cs() == "multicolumn") {
 875                         // extract column count and insert dummy cells
 876                         MathArray count;
 877                         parse(count, FLAG_ITEM, mode);
 878                         int cols = 1;
 879                         if (!extractNumber(count, cols)) {
 880                                 lyxerr << " can't extract number of cells from " << count << endl;
 881                         }
 882                         // resize the table if necessary
 883                         for (int i = 0; i < cols; ++i) {
 884                                 ++cellcol;
 885                                 if (cellcol == grid.ncols()) {
 886                                         //lyxerr << "adding column " << cellcol << endl;
 887                                         grid.addCol(cellcol - 1);
 888                                 }
 889                                 cell = &grid.cell(grid.index(cellrow, cellcol));
 890                                 // mark this as dummy
 891                                 grid.cellinfo(grid.index(cellrow, cellcol)).dummy_ = true;
 892                         }
 893                         // the last cell is the real thng, not a dummy
 894                         grid.cellinfo(grid.index(cellrow, cellcol)).dummy_ = false;
 895
 896                         // read special alignment
 897                         MathArray align;
 898                         parse(align, FLAG_ITEM, mode);
 899                         //grid.cellinfo(grid.index(cellrow, cellcol)).align_ = extractString(align);
 900
 901                         // parse the remaining contents into the "real" cell
 902                         parse(*cell, FLAG_ITEM, mode);
 903                 }
 904 #endif
 905
 906                 else if (t.cs() == "limits")
 907                         limits = 1;
 908
 909                 else if (t.cs() == "nolimits")
 910                         limits = -1;
 911
 912                 else if (t.cs() == "nonumber") {
 913                         if (grid.asHullInset())
 914                                 grid.asHullInset()->numbered(cellrow, false);
 915                 }
 916
 917                 else if (t.cs() == "number") {
 918                         if (grid.asHullInset())
 919                                 grid.asHullInset()->numbered(cellrow, true);
 920                 }
 921
 922                 else if (t.cs() == "hline") {
 923                         grid.rowinfo(cellrow).lines_ ++;
 924                 }
 925
 926                 else if (t.cs() == "sqrt") {
 927                         MathArray ar;
 928                         parse(ar, FLAG_OPTION, mode);
 929                         if (ar.size()) {
 930                                 cell->push_back(MathAtom(new MathRootInset));
 931                                 cell->back().nucleus()->cell(0) = ar;
 932                                 parse(cell->back().nucleus()->cell(1), FLAG_ITEM, mode);
 933                         } else {
 934                                 cell->push_back(MathAtom(new MathSqrtInset));
 935                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
 936                         }
 937                 }
 938
 939                 else if (t.cs() == "xrightarrow" || t.cs() == "xleftarrow") {
 940                         cell->push_back(createMathInset(t.cs()));
 941                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
 942                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
 943                 }
 944
 945                 else if (t.cs() == "ref" || t.cs() == "prettyref" ||
 946                                 t.cs() == "pageref" || t.cs() == "vpageref" || t.cs() == "vref") {
 947                         cell->push_back(MathAtom(new RefInset(t.cs())));
 948                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
 949                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
 950                 }
 951
 952                 else if (t.cs() == "left") {
 953                         skipSpaces();
 954                         string l = getToken().asString();
 955                         MathArray ar;
 956                         parse(ar, FLAG_RIGHT, mode);
 957                         skipSpaces();
 958                         string r = getToken().asString();
 959                         cell->push_back(MathAtom(new MathDelimInset(l, r, ar)));
 960                 }
 961
 962                 else if (t.cs() == "right") {
 963                         if (flags & FLAG_RIGHT)
 964                                 return;
 965                         //lyxerr << "got so far: '" << cell << "'" << endl;
 966                         error("Unmatched right delimiter");
 967                         return;
 968                 }
 969
 970                 else if (t.cs() == "begin") {
 971                         string const name = getArg('{', '}');
 972
 973                         if (name == "array" || name == "subarray") {
 974                                 string const valign = parse_verbatim_option() + 'c';
 975                                 string const halign = parse_verbatim_item();
 976                                 cell->push_back(MathAtom(new MathArrayInset(name, valign[0], halign)));
 977                                 parse2(cell->back(), FLAG_END, mode, false);
 978                         }
 979
 980                         else if (name == "tabular") {
 981                                 string const valign = parse_verbatim_option() + 'c';
 982                                 string const halign = parse_verbatim_item();
 983                                 cell->push_back(MathAtom(new MathTabularInset(name, valign[0], halign)));
 984                                 parse2(cell->back(), FLAG_END, MathInset::TEXT_MODE, false);
 985                         }
 986
 987                         else if (name == "split" || name == "cases" ||
 988                                          name == "gathered" || name == "aligned" ||
 989                                    name == "alignedat") {
 990                                 cell->push_back(createMathInset(name));
 991                                 parse2(cell->back(), FLAG_END, mode, false);
 992                         }
 993
 994                         else if (name == "math") {
 995                                 cell->push_back(MathAtom(new MathHullInset("simple")));
 996                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, true);
 997                         }
 998
 999                         else if (name == "equation" || name == "equation*"
1000                                         || name == "displaymath") {
1001                                 cell->push_back(MathAtom(new MathHullInset("equation")));
1002                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, (name == "equation"));
1003                         }
1004
1005                         else if (name == "eqnarray" || name == "eqnarray*") {
1006                                 cell->push_back(MathAtom(new MathHullInset("eqnarray")));
1007                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1008                         }
1009
1010                         else if (name == "align" || name == "align*") {
1011                                 cell->push_back(MathAtom(new MathHullInset("align")));
1012                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1013                         }
1014
1015                         else if (name == "flalign" || name == "flalign*") {
1016                                 cell->push_back(MathAtom(new MathHullInset("flalign")));
1017                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1018                         }
1019
1020                         else if (name == "alignat" || name == "alignat*") {
1021                                 // ignore this for a while
1022                                 getArg('{', '}');
1023                                 cell->push_back(MathAtom(new MathHullInset("alignat")));
1024                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1025                         }
1026
1027                         else if (name == "xalignat" || name == "xalignat*") {
1028                                 // ignore this for a while
1029                                 getArg('{', '}');
1030                                 cell->push_back(MathAtom(new MathHullInset("xalignat")));
1031                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1032                         }
1033
1034                         else if (name == "xxalignat") {
1035                                 // ignore this for a while
1036                                 getArg('{', '}');
1037                                 cell->push_back(MathAtom(new MathHullInset("xxalignat")));
1038                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1039                         }
1040
1041                         else if (name == "multline" || name == "multline*") {
1042                                 cell->push_back(MathAtom(new MathHullInset("multline")));
1043                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1044                         }
1045
1046                         else if (name == "gather" || name == "gather*") {
1047                                 cell->push_back(MathAtom(new MathHullInset("gather")));
1048                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1049                         }
1050
1051                         else if (latexkeys const * l = in_word_set(name)) {
1052                                 if (l->inset == "matrix") {
1053                                         cell->push_back(createMathInset(name));
1054                                         parse2(cell->back(), FLAG_END, mode, false);
1055                                 }
1056                         }
1057
1058                         else {
1059                                 dump();
1060                                 lyxerr << "found unknown math environment '" << name << "'" << endl;
1061                                 // create generic environment inset
1062                                 cell->push_back(MathAtom(new MathEnvInset(name)));
1063                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1064                         }
1065                 }
1066
1067                 else if (t.cs() == "kern") {
1068 #ifdef WITH_WARNINGS
1069 #warning A hack...
1070 #endif
1071                         string s;
1072                         while (true) {
1073                                 Token const & t = getToken();
1074                                 if (!good()) {
1075                                         putback();
1076                                         break;
1077                                 }
1078                                 s += t.character();
1079                                 if (isValidLength(s))
1080                                         break;
1081                         }
1082                         cell->push_back(MathAtom(new MathKernInset(s)));
1083                 }
1084
1085                 else if (t.cs() == "label") {
1086                         string label = parse_verbatim_item();
1087                         MathArray ar;
1088                         asArray(label, ar);
1089                         if (grid.asHullInset()) {
1090                                 grid.asHullInset()->label(cellrow, label);
1091                         } else {
1092                                 cell->push_back(createMathInset(t.cs()));
1093                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
1094                         }
1095                 }
1096
1097                 else if (t.cs() == "choose" || t.cs() == "over" || t.cs() == "atop") {
1098                         MathAtom at = createMathInset(t.cs());
1099                         at.nucleus()->cell(0) = *cell;
1100                         cell->clear();
1101                         parse(at.nucleus()->cell(1), flags, mode);
1102                         cell->push_back(at);
1103                         return;
1104                 }
1105
1106                 else if (t.cs() == "color") {
1107                         MathAtom at = createMathInset(t.cs());
1108                         parse(at.nucleus()->cell(0), FLAG_ITEM, MathInset::TEXT_MODE);
1109                         parse(at.nucleus()->cell(1), flags, mode);
1110                         cell->push_back(at);
1111                         return;
1112                 }
1113
1114                 else if (t.cs() == "substack") {
1115                         cell->push_back(createMathInset(t.cs()));
1116                         parse2(cell->back(), FLAG_ITEM, mode, false);
1117                 }
1118
1119                 else if (t.cs() == "framebox" || t.cs() == "makebox") {
1120                         cell->push_back(createMathInset(t.cs()));
1121                         parse(cell->back().nucleus()->cell(0), FLAG_OPTION, MathInset::TEXT_MODE);
1122                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, MathInset::TEXT_MODE);
1123                         parse(cell->back().nucleus()->cell(2), FLAG_ITEM, MathInset::TEXT_MODE);
1124                 }
1125
1126 #if 0
1127                 else if (t.cs() == "infer") {
1128                         MathArray ar;
1129                         parse(ar, FLAG_OPTION, mode);
1130                         cell->push_back(createMathInset(t.cs()));
1131                         parse2(cell->back(), FLAG_ITEM, mode, false);
1132                 }
1133
1134                 // Disabled
1135                 else if (1 && t.cs() == "ar") {
1136                         auto_ptr<MathXYArrowInset> p(new MathXYArrowInset);
1137                         // try to read target
1138                         parse(p->cell(0), FLAG_OTPTION, mode);
1139                         // try to read label
1140                         if (nextToken().cat() == catSuper || nextToken().cat() == catSub) {
1141                                 p->up_ = nextToken().cat() == catSuper;
1142                                 getToken();
1143                                 parse(p->cell(1), FLAG_ITEM, mode);
1144                                 //lyxerr << "read label: " << p->cell(1) << endl;
1145                         }
1146
1147                         cell->push_back(MathAtom(p.release()));
1148                         //lyxerr << "read cell: " << cell << endl;
1149                 }
1150 #endif
1151
1152                 else if (t.cs().size()) {
1153                         latexkeys const * l = in_word_set(t.cs());
1154                         if (l) {
1155                                 if (l->inset == "font") {
1156                                         cell->push_back(createMathInset(t.cs()));
1157                                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, asMode(mode, l->extra));
1158                                 }
1159
1160                                 else if (l->inset == "oldfont") {
1161                                         cell->push_back(createMathInset(t.cs()));
1162                                         parse(cell->back().nucleus()->cell(0), flags, asMode(mode, l->extra));
1163                                         return;
1164                                 }
1165
1166                                 else if (l->inset == "style") {
1167                                         cell->push_back(createMathInset(t.cs()));
1168                                         parse(cell->back().nucleus()->cell(0), flags, mode);
1169                                         return;
1170                                 }
1171
1172                                 else {
1173                                         MathAtom at = createMathInset(t.cs());
1174                                         for (MathInset::idx_type i = 0; i < at->nargs(); ++i)
1175                                                 parse(at.nucleus()->cell(i), FLAG_ITEM, asMode(mode, l->extra));
1176                                         cell->push_back(at);
1177                                 }
1178                         }
1179
1180                         else {
1181                                 MathAtom at = createMathInset(t.cs());
1182                                 MathInset::mode_type m = mode;
1183                                 //if (m == MathInset::UNDECIDED_MODE)
1184                                 //lyxerr << "default creation: m1: " << m << endl;
1185                                 if (at->currentMode() != MathInset::UNDECIDED_MODE)
1186                                         m = at->currentMode();
1187                                 //lyxerr << "default creation: m2: " << m << endl;
1188                                 MathInset::idx_type start = 0;
1189                                 // this fails on \bigg[...\bigg]
1190                                 //MathArray opt;
1191                                 //parse(opt, FLAG_OPTION, MathInset::VERBATIM_MODE);
1192                                 //if (opt.size()) {
1193                                 //      start = 1;
1194                                 //      at.nucleus()->cell(0) = opt;
1195                                 //}
1196                                 for (MathInset::idx_type i = start; i < at->nargs(); ++i) {
1197                                         parse(at.nucleus()->cell(i), FLAG_ITEM, m);
1198                                         skipSpaces();
1199                                 }
1200                                 cell->push_back(at);
1201                         }
1202                 }
1203
1204
1205                 if (flags & FLAG_LEAVE) {
1206                         flags &= ~FLAG_LEAVE;
1207                         break;
1208                 }
1209         }
1210 }
1211
1212
1213
1214 } // anonymous namespace
1215
1216
1217 void mathed_parse_cell(MathArray & ar, string const & str)
1218 {
1219         istringstream is(str);
1220         mathed_parse_cell(ar, is);
1221 }
1222
1223
1224 void mathed_parse_cell(MathArray & ar, istream & is)
1225 {
1226         Parser(is).parse(ar, 0, MathInset::MATH_MODE);
1227 }
1228
1229
1230 bool mathed_parse_normal(MathAtom & t, string const & str)
1231 {
1232         istringstream is(str);
1233         return Parser(is).parse(t);
1234 }
1235
1236
1237 bool mathed_parse_normal(MathAtom & t, istream & is)
1238 {
1239         return Parser(is).parse(t);
1240 }
1241
1242
1243 bool mathed_parse_normal(MathAtom & t, LyXLex & lex)
1244 {
1245         return Parser(lex).parse(t);
1246 }
1247
1248
1249 void mathed_parse_normal(MathGridInset & grid, string const & str)
1250 {
1251         istringstream is(str);
1252         Parser(is).parse1(grid, 0, MathInset::MATH_MODE, false);
1253 }
1254
1255
1256 void initParser()
1257 {
1258         fill(theCatcode, theCatcode + 256, catOther);
1259         fill(theCatcode + 'a', theCatcode + 'z' + 1, catLetter);
1260         fill(theCatcode + 'A', theCatcode + 'Z' + 1, catLetter);
1261
1262         theCatcode[int('\\')] = catEscape;
1263         theCatcode[int('{')]  = catBegin;
1264         theCatcode[int('}')]  = catEnd;
1265         theCatcode[int('$')]  = catMath;
1266         theCatcode[int('&')]  = catAlign;
1267         theCatcode[int('\n')] = catNewline;
1268         theCatcode[int('#')]  = catParameter;
1269         theCatcode[int('^')]  = catSuper;
1270         theCatcode[int('_')]  = catSub;
1271         theCatcode[int(0x7f)] = catIgnore;
1272         theCatcode[int(' ')]  = catSpace;
1273         theCatcode[int('\t')] = catSpace;
1274         theCatcode[int('\r')] = catNewline;
1275         theCatcode[int('~')]  = catActive;
1276         theCatcode[int('%')]  = catComment;
1277 }