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_macrotemplate.h"
  52 #include "math_parboxinset.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 "support/std_sstream.h"
  65 #include "debug.h"
  66
  67
  68 using std::istream;
  69 using std::ostream;
  70 using std::ios;
  71 using std::endl;
  72 using std::fill;
  73 using std::vector;
  74 using std::atoi;
  75
  76
  77 //#define FILEDEBUG
  78
  79
  80 namespace {
  81
  82 MathInset::mode_type asMode(MathInset::mode_type oldmode, string const & str)
  83 {
  84         //lyxerr << "handling mode: '" << str << "'" << endl;
  85         if (str == "mathmode")
  86                 return MathInset::MATH_MODE;
  87         if (str == "textmode" || str == "forcetext")
  88                 return MathInset::TEXT_MODE;
  89         return oldmode;
  90 }
  91
  92
  93 bool stared(string const & s)
  94 {
  95         string::size_type const n = s.size();
  96         return n && s[n - 1] == '*';
  97 }
  98
  99
 100 // These are TeX's catcodes
 101 enum CatCode {
 102         catEscape,     // 0    backslash
 103         catBegin,      // 1    {
 104         catEnd,        // 2    }
 105         catMath,       // 3    $
 106         catAlign,      // 4    &
 107         catNewline,    // 5    ^^M
 108         catParameter,  // 6    #
 109         catSuper,      // 7    ^
 110         catSub,        // 8    _
 111         catIgnore,     // 9
 112         catSpace,      // 10   space
 113         catLetter,     // 11   a-zA-Z
 114         catOther,      // 12   none of the above
 115         catActive,     // 13   ~
 116         catComment,    // 14   %
 117         catInvalid     // 15   <delete>
 118 };
 119
 120 CatCode theCatcode[256];
 121
 122
 123 inline CatCode catcode(unsigned char c)
 124 {
 125         return theCatcode[c];
 126 }
 127
 128
 129 enum {
 130         FLAG_BRACE_LAST = 1 << 1,  //  last closing brace ends the parsing
 131         FLAG_RIGHT      = 1 << 2,  //  next \\right ends the parsing process
 132         FLAG_END        = 1 << 3,  //  next \\end ends the parsing process
 133         FLAG_BRACK_LAST = 1 << 4,  //  next closing bracket ends the parsing
 134         FLAG_TEXTMODE   = 1 << 5,  //  we are in a box
 135         FLAG_ITEM       = 1 << 6,  //  read a (possibly braced token)
 136         FLAG_LEAVE      = 1 << 7,  //  leave the loop at the end
 137         FLAG_SIMPLE     = 1 << 8,  //  next $ leaves the loop
 138         FLAG_EQUATION   = 1 << 9,  //  next \] leaves the loop
 139         FLAG_SIMPLE2    = 1 << 10, //  next \) leaves the loop
 140         FLAG_OPTION     = 1 << 11, //  read [...] style option
 141         FLAG_BRACED     = 1 << 12  //  read {...} style argument
 142 };
 143
 144
 145 void catInit()
 146 {
 147         fill(theCatcode, theCatcode + 256, catOther);
 148         fill(theCatcode + 'a', theCatcode + 'z' + 1, catLetter);
 149         fill(theCatcode + 'A', theCatcode + 'Z' + 1, catLetter);
 150
 151         theCatcode[int('\\')] = catEscape;
 152         theCatcode[int('{')]  = catBegin;
 153         theCatcode[int('}')]  = catEnd;
 154         theCatcode[int('$')]  = catMath;
 155         theCatcode[int('&')]  = catAlign;
 156         theCatcode[int('\n')] = catNewline;
 157         theCatcode[int('#')]  = catParameter;
 158         theCatcode[int('^')]  = catSuper;
 159         theCatcode[int('_')]  = catSub;
 160         theCatcode[int(0x7f)] = catIgnore;
 161         theCatcode[int(' ')]  = catSpace;
 162         theCatcode[int('\t')] = catSpace;
 163         theCatcode[int('\r')] = catNewline;
 164         theCatcode[int('~')]  = catActive;
 165         theCatcode[int('%')]  = catComment;
 166 }
 167
 168
 169
 170 //
 171 // Helper class for parsing
 172 //
 173
 174 class Token {
 175 public:
 176         ///
 177         Token() : cs_(), char_(0), cat_(catIgnore) {}
 178         ///
 179         Token(char c, CatCode cat) : cs_(), char_(c), cat_(cat) {}
 180         ///
 181         Token(string const & cs) : cs_(cs), char_(0), cat_(catIgnore) {}
 182
 183         ///
 184         string const & cs() const { return cs_; }
 185         ///
 186         CatCode cat() const { return cat_; }
 187         ///
 188         char character() const { return char_; }
 189         ///
 190         string asString() const { return cs_.size() ? cs_ : string(1, char_); }
 191
 192 private:
 193         ///
 194         string cs_;
 195         ///
 196         char char_;
 197         ///
 198         CatCode cat_;
 199 };
 200
 201 ostream & operator<<(ostream & os, Token const & t)
 202 {
 203         if (t.cs().size())
 204                 os << '\\' << t.cs();
 205         else if (t.cat() == catLetter)
 206                 os << t.character();
 207         else
 208                 os << '[' << t.character() << ',' << t.cat() << ']';
 209         return os;
 210 }
 211
 212
 213 class Parser {
 214
 215 public:
 216         ///
 217         typedef  MathInset::mode_type mode_type;
 218
 219         ///
 220         Parser(LyXLex & lex);
 221         ///
 222         Parser(istream & is);
 223
 224         ///
 225         bool parse(MathAtom & at);
 226         ///
 227         void parse(MathArray & array, unsigned flags, mode_type mode);
 228         ///
 229         void parse1(MathGridInset & grid, unsigned flags, mode_type mode,
 230                 bool numbered);
 231         ///
 232         MathArray parse(unsigned flags, mode_type mode);
 233         ///
 234         int lineno() const { return lineno_; }
 235         ///
 236         void putback();
 237
 238 private:
 239         ///
 240         void parse2(MathAtom & at, unsigned flags, mode_type mode, bool numbered);
 241         /// get arg delimited by 'left' and 'right'
 242         string getArg(char left, char right);
 243         ///
 244         char getChar();
 245         ///
 246         void error(string const & msg);
 247         /// dump contents to screen
 248         void dump() const;
 249         ///
 250         void tokenize(istream & is);
 251         ///
 252         void tokenize(string const & s);
 253         ///
 254         void skipSpaceTokens(istream & is, char c);
 255         ///
 256         void push_back(Token const & t);
 257         ///
 258         void pop_back();
 259         ///
 260         Token const & prevToken() const;
 261         ///
 262         Token const & nextToken() const;
 263         ///
 264         Token const & getToken();
 265         /// skips spaces if any
 266         void skipSpaces();
 267         ///
 268         void lex(string const & s);
 269         ///
 270         bool good() const;
 271         ///
 272         string parse_verbatim_item();
 273         ///
 274         string parse_verbatim_option();
 275
 276         ///
 277         int lineno_;
 278         ///
 279         vector<Token> tokens_;
 280         ///
 281         unsigned pos_;
 282 };
 283
 284
 285 Parser::Parser(LyXLex & lexer)
 286         : lineno_(lexer.getLineNo()), pos_(0)
 287 {
 288         tokenize(lexer.getStream());
 289         lexer.eatLine();
 290 }
 291
 292
 293 Parser::Parser(istream & is)
 294         : lineno_(0), pos_(0)
 295 {
 296         tokenize(is);
 297 }
 298
 299
 300 void Parser::push_back(Token const & t)
 301 {
 302         tokens_.push_back(t);
 303 }
 304
 305
 306 void Parser::pop_back()
 307 {
 308         tokens_.pop_back();
 309 }
 310
 311
 312 Token const & Parser::prevToken() const
 313 {
 314         static const Token dummy;
 315         return pos_ > 0 ? tokens_[pos_ - 1] : dummy;
 316 }
 317
 318
 319 Token const & Parser::nextToken() const
 320 {
 321         static const Token dummy;
 322         return good() ? tokens_[pos_] : dummy;
 323 }
 324
 325
 326 Token const & Parser::getToken()
 327 {
 328         static const Token dummy;
 329         //lyxerr << "looking at token " << tokens_[pos_] << " pos: " << pos_ << endl;
 330         return good() ? tokens_[pos_++] : dummy;
 331 }
 332
 333
 334 void Parser::skipSpaces()
 335 {
 336         while (nextToken().cat() == catSpace || nextToken().cat() == catNewline)
 337                 getToken();
 338 }
 339
 340
 341 void Parser::putback()
 342 {
 343         --pos_;
 344 }
 345
 346
 347 bool Parser::good() const
 348 {
 349         return pos_ < tokens_.size();
 350 }
 351
 352
 353 char Parser::getChar()
 354 {
 355         if (!good())
 356                 error("The input stream is not well...");
 357         return tokens_[pos_++].character();
 358 }
 359
 360
 361 string Parser::getArg(char left, char right)
 362 {
 363         skipSpaces();
 364
 365         string result;
 366         char c = getChar();
 367
 368         if (c != left)
 369                 putback();
 370         else
 371                 while ((c = getChar()) != right && good())
 372                         result += c;
 373
 374         return result;
 375 }
 376
 377
 378 void Parser::skipSpaceTokens(istream & is, char c)
 379 {
 380         // skip trailing spaces
 381         while (catcode(c) == catSpace || catcode(c) == catNewline)
 382                 if (!is.get(c))
 383                         break;
 384         //lyxerr << "putting back: " << c << endl;
 385         is.putback(c);
 386 }
 387
 388
 389 void Parser::tokenize(istream & is)
 390 {
 391         // eat everything up to the next \end_inset or end of stream
 392         // and store it in s for further tokenization
 393         string s;
 394         char c;
 395         while (is.get(c)) {
 396                 s += c;
 397                 if (s.size() >= 10 && s.substr(s.size() - 10) == "\\end_inset") {
 398                         s = s.substr(0, s.size() - 10);
 399                         break;
 400                 }
 401         }
 402         // Remove the space after \end_inset
 403         if (is.get(c) && c != ' ')
 404                 is.unget();
 405
 406         // tokenize buffer
 407         tokenize(s);
 408 }
 409
 410
 411 void Parser::tokenize(string const & buffer)
 412 {
 413         static bool init_done = false;
 414
 415         if (!init_done) {
 416                 catInit();
 417                 init_done = true;
 418         }
 419
 420         istringstream is(buffer.c_str(), ios::in | ios::binary);
 421
 422         char c;
 423         while (is.get(c)) {
 424                 //lyxerr << "reading c: " << c << endl;
 425
 426                 switch (catcode(c)) {
 427                         case catNewline: {
 428                                 ++lineno_;
 429                                 is.get(c);
 430                                 if (catcode(c) == catNewline)
 431                                         ; //push_back(Token("par"));
 432                                 else {
 433                                         push_back(Token('\n', catNewline));
 434                                         is.putback(c);
 435                                 }
 436                                 break;
 437                         }
 438
 439 /*
 440                         case catComment: {
 441                                 while (is.get(c) && catcode(c) != catNewline)
 442                                         ;
 443                                 ++lineno_;
 444                                 break;
 445                         }
 446 */
 447
 448                         case catEscape: {
 449                                 is.get(c);
 450                                 if (!is) {
 451                                         error("unexpected end of input");
 452                                 } else {
 453                                         string s(1, c);
 454                                         if (catcode(c) == catLetter) {
 455                                                 // collect letters
 456                                                 while (is.get(c) && catcode(c) == catLetter)
 457                                                         s += c;
 458                                                 skipSpaceTokens(is, c);
 459                                         }
 460                                         push_back(Token(s));
 461                                 }
 462                                 break;
 463                         }
 464
 465                         case catSuper:
 466                         case catSub: {
 467                                 push_back(Token(c, catcode(c)));
 468                                 is.get(c);
 469                                 skipSpaceTokens(is, c);
 470                                 break;
 471                         }
 472
 473                         case catIgnore: {
 474                                 lyxerr << "ignoring a char: " << int(c) << endl;
 475                                 break;
 476                         }
 477
 478                         default:
 479                                 push_back(Token(c, catcode(c)));
 480                 }
 481         }
 482
 483 #ifdef FILEDEBUG
 484         dump();
 485 #endif
 486 }
 487
 488
 489 void Parser::dump() const
 490 {
 491         lyxerr << "\nTokens: ";
 492         for (unsigned i = 0; i < tokens_.size(); ++i) {
 493                 if (i == pos_)
 494                         lyxerr << " <#> ";
 495                 lyxerr << tokens_[i];
 496         }
 497         lyxerr << " pos: " << pos_ << endl;
 498 }
 499
 500
 501 void Parser::error(string const & msg)
 502 {
 503         lyxerr << "Line ~" << lineno_ << ": Math parse error: " << msg << endl;
 504         dump();
 505         //exit(1);
 506 }
 507
 508
 509 bool Parser::parse(MathAtom & at)
 510 {
 511         skipSpaces();
 512         MathArray ar;
 513         parse(ar, false, MathInset::UNDECIDED_MODE);
 514         if (ar.size() != 1 || ar.front()->getType() == "none") {
 515                 lyxerr << "unusual contents found: " << ar << endl;
 516                 at = MathAtom(new MathParInset(ar));
 517                 //if (at->nargs() > 0)
 518                 //      at.nucleus()->cell(0) = ar;
 519                 //else
 520                 //      lyxerr << "unusual contents found: " << ar << endl;
 521                 return true;
 522         }
 523         at = ar[0];
 524         return true;
 525 }
 526
 527
 528 string Parser::parse_verbatim_option()
 529 {
 530         skipSpaces();
 531         string res;
 532         if (nextToken().character() == '[') {
 533                 Token t = getToken();
 534                 for (Token t = getToken(); t.character() != ']' && good(); t = getToken()) {
 535                         if (t.cat() == catBegin) {
 536                                 putback();
 537                                 res += '{' + parse_verbatim_item() + '}';
 538                         } else
 539                                 res += t.asString();
 540                 }
 541         }
 542         return res;
 543 }
 544
 545
 546 string Parser::parse_verbatim_item()
 547 {
 548         skipSpaces();
 549         string res;
 550         if (nextToken().cat() == catBegin) {
 551                 Token t = getToken();
 552                 for (Token t = getToken(); t.cat() != catEnd && good(); t = getToken()) {
 553                         if (t.cat() == catBegin) {
 554                                 putback();
 555                                 res += '{' + parse_verbatim_item() + '}';
 556                         }
 557                         else
 558                                 res += t.asString();
 559                 }
 560         }
 561         return res;
 562 }
 563
 564
 565 MathArray Parser::parse(unsigned flags, mode_type mode)
 566 {
 567         MathArray ar;
 568         parse(ar, flags, mode);
 569         return ar;
 570 }
 571
 572
 573 void Parser::parse(MathArray & array, unsigned flags, mode_type mode)
 574 {
 575         MathGridInset grid(1, 1);
 576         parse1(grid, flags, mode, false);
 577         array = grid.cell(0);
 578 }
 579
 580
 581 void Parser::parse2(MathAtom & at, const unsigned flags, const mode_type mode,
 582         const bool numbered)
 583 {
 584         parse1(*(at.nucleus()->asGridInset()), flags, mode, numbered);
 585 }
 586
 587
 588 void Parser::parse1(MathGridInset & grid, unsigned flags,
 589         const mode_type mode, const bool numbered)
 590 {
 591         int limits = 0;
 592         MathGridInset::row_type cellrow = 0;
 593         MathGridInset::col_type cellcol = 0;
 594         MathArray * cell = &grid.cell(grid.index(cellrow, cellcol));
 595
 596         if (grid.asHullInset())
 597                 grid.asHullInset()->numbered(cellrow, numbered);
 598
 599         //dump();
 600         //lyxerr << " flags: " << flags << endl;
 601         //lyxerr << " mode: " << mode  << endl;
 602         //lyxerr << "grid: " << grid << endl;
 603
 604         while (good()) {
 605                 Token const & t = getToken();
 606
 607 #ifdef FILEDEBUG
 608                 lyxerr << "t: " << t << " flags: " << flags << endl;
 609                 lyxerr << "mode: " << mode  << endl;
 610                 cell->dump();
 611                 lyxerr << endl;
 612 #endif
 613
 614                 if (flags & FLAG_ITEM) {
 615
 616                 if (t.cat() == catBegin) {
 617                                 // skip the brace and collect everything to the next matching
 618                                 // closing brace
 619                                 parse1(grid, FLAG_BRACE_LAST, mode, numbered);
 620                                 return;
 621                         }
 622
 623                         // handle only this single token, leave the loop if done
 624                         flags = FLAG_LEAVE;
 625                 }
 626
 627
 628                 if (flags & FLAG_BRACED) {
 629                         if (t.cat() == catSpace)
 630                                 continue;
 631
 632                         if (t.cat() != catBegin) {
 633                                 error("opening brace expected");
 634                                 return;
 635                         }
 636
 637                         // skip the brace and collect everything to the next matching
 638                         // closing brace
 639                         flags = FLAG_BRACE_LAST;
 640                 }
 641
 642
 643                 if (flags & FLAG_OPTION) {
 644                         if (t.cat() == catOther && t.character() == '[') {
 645                                 MathArray ar;
 646                                 parse(ar, FLAG_BRACK_LAST, mode);
 647                                 cell->append(ar);
 648                         } else {
 649                                 // no option found, put back token and we are done
 650                                 putback();
 651                         }
 652                         return;
 653                 }
 654
 655                 //
 656                 // cat codes
 657                 //
 658                 if (t.cat() == catMath) {
 659                         if (mode != MathInset::MATH_MODE) {
 660                                 // we are inside some text mode thingy, so opening new math is allowed
 661                                 Token const & n = getToken();
 662                                 if (n.cat() == catMath) {
 663                                         // TeX's $$...$$ syntax for displayed math
 664                                         cell->push_back(MathAtom(new MathHullInset("equation")));
 665                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 666                                         getToken(); // skip the second '$' token
 667                                 } else {
 668                                         // simple $...$  stuff
 669                                         putback();
 670                                         cell->push_back(MathAtom(new MathHullInset("simple")));
 671                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 672                                 }
 673                         }
 674
 675                         else if (flags & FLAG_SIMPLE) {
 676                                 // this is the end of the formula
 677                                 return;
 678                         }
 679
 680                         else {
 681                                 error("something strange in the parser");
 682                                 break;
 683                         }
 684                 }
 685
 686                 else if (t.cat() == catLetter)
 687                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 688
 689                 else if (t.cat() == catSpace && mode != MathInset::MATH_MODE) {
 690                         if (cell->empty() || cell->back()->getChar() != ' ')
 691                                 cell->push_back(MathAtom(new MathCharInset(t.character())));
 692                 }
 693
 694                 else if (t.cat() == catNewline && mode != MathInset::MATH_MODE) {
 695                         if (cell->empty() || cell->back()->getChar() != ' ')
 696                                 cell->push_back(MathAtom(new MathCharInset(' ')));
 697                 }
 698
 699                 else if (t.cat() == catParameter) {
 700                         Token const & n = getToken();
 701                         cell->push_back(MathAtom(new MathMacroArgument(n.character()-'0')));
 702                 }
 703
 704                 else if (t.cat() == catActive)
 705                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 706
 707                 else if (t.cat() == catBegin) {
 708                         MathArray ar;
 709                         parse(ar, FLAG_BRACE_LAST, mode);
 710                         // do not create a BraceInset if they were written by LyX
 711                         // this helps to keep the annoyance of  "a choose b"  to a minimum
 712                         if (ar.size() == 1 && ar[0]->extraBraces())
 713                                 cell->append(ar);
 714                         else
 715                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
 716                 }
 717
 718                 else if (t.cat() == catEnd) {
 719                         if (flags & FLAG_BRACE_LAST)
 720                                 return;
 721                         error("found '}' unexpectedly");
 722                         //lyx::Assert(0);
 723                         //add(cell, '}', LM_TC_TEX);
 724                 }
 725
 726                 else if (t.cat() == catAlign) {
 727                         ++cellcol;
 728                         //lyxerr << " column now " << cellcol << " max: " << grid.ncols() << endl;
 729                         if (cellcol == grid.ncols()) {
 730                                 //lyxerr << "adding column " << cellcol << endl;
 731                                 grid.addCol(cellcol - 1);
 732                         }
 733                         cell = &grid.cell(grid.index(cellrow, cellcol));
 734                 }
 735
 736                 else if (t.cat() == catSuper || t.cat() == catSub) {
 737                         bool up = (t.cat() == catSuper);
 738                         // we need no new script inset if the last thing was a scriptinset,
 739                         // which has that script already not the same script already
 740                         if (!cell->size())
 741                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 742                         else if (cell->back()->asScriptInset() &&
 743                                         !cell->back()->asScriptInset()->has(up))
 744                                 cell->back().nucleus()->asScriptInset()->ensure(up);
 745                         else if (cell->back()->asScriptInset())
 746                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 747                         else
 748                                 cell->back() = MathAtom(new MathScriptInset(cell->back(), up));
 749                         MathScriptInset * p = cell->back().nucleus()->asScriptInset();
 750                         // special handling of {}-bases
 751                         // is this always correct?
 752                         // It appears that this is wrong (Dekel)
 753                         //if (p->nuc().size() == 1 && p->nuc().back()->asNestInset() &&
 754                         //    p->nuc().back()->extraBraces())
 755                         //      p->nuc() = p->nuc().back()->asNestInset()->cell(0);
 756                         parse(p->cell(up), FLAG_ITEM, mode);
 757                         if (limits) {
 758                                 p->limits(limits);
 759                                 limits = 0;
 760                         }
 761                 }
 762
 763                 else if (t.character() == ']' && (flags & FLAG_BRACK_LAST)) {
 764                         //lyxerr << "finished reading option" << endl;
 765                         return;
 766                 }
 767
 768                 else if (t.cat() == catOther)
 769                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 770
 771                 else if (t.cat() == catComment) {
 772                         string s;
 773                         while (good()) {
 774                                 Token const & t = getToken();
 775                                 if (t.cat() == catNewline)
 776                                         break;
 777                                 s += t.asString();
 778                         }
 779                         cell->push_back(MathAtom(new MathCommentInset(s)));
 780                         skipSpaces();
 781                 }
 782
 783                 //
 784                 // control sequences
 785                 //
 786
 787                 else if (t.cs() == "lyxlock") {
 788                         if (cell->size())
 789                                 cell->back().nucleus()->lock(true);
 790                 }
 791
 792                 else if (t.cs() == "def" ||
 793                         t.cs() == "newcommand" ||
 794                         t.cs() == "renewcommand")
 795                 {
 796                         string const type = t.cs();
 797                         string name;
 798                         int nargs = 0;
 799                         if (t.cs() == "def") {
 800                                 // get name
 801                                 name = getToken().cs();
 802
 803                                 // read parameter
 804                                 string pars;
 805                                 while (good() && nextToken().cat() != catBegin) {
 806                                         pars += getToken().cs();
 807                                         ++nargs;
 808                                 }
 809                                 nargs /= 2;
 810                                 //lyxerr << "read \\def parameter list '" << pars << "'" << endl;
 811
 812                         } else { // t.cs() == "newcommand" || t.cs() == "renewcommand"
 813
 814                                 if (getToken().cat() != catBegin) {
 815                                         error("'{' in \\newcommand expected (1) ");
 816                                         return;
 817                                 }
 818
 819                                 name = getToken().cs();
 820
 821                                 if (getToken().cat() != catEnd) {
 822                                         error("'}' in \\newcommand expected");
 823                                         return;
 824                                 }
 825
 826                                 string arg  = getArg('[', ']');
 827                                 if (!arg.empty())
 828                                         nargs = atoi(arg.c_str());
 829
 830                         }
 831
 832                         MathArray ar1;
 833                         parse(ar1, FLAG_ITEM, MathInset::UNDECIDED_MODE);
 834
 835                         // we cannot handle recursive stuff at all
 836                         //MathArray test;
 837                         //test.push_back(createMathInset(name));
 838                         //if (ar1.contains(test)) {
 839                         //      error("we cannot handle recursive macros at all.");
 840                         //      return;
 841                         //}
 842
 843                         // is a version for display attached?
 844                         skipSpaces();
 845                         MathArray ar2;
 846                         if (nextToken().cat() == catBegin)
 847                                 parse(ar2, FLAG_ITEM, MathInset::MATH_MODE);
 848
 849                         cell->push_back(MathAtom(new MathMacroTemplate(name, nargs, type,
 850                                 ar1, ar2)));
 851                 }
 852
 853                 else if (t.cs() == "(") {
 854                         cell->push_back(MathAtom(new MathHullInset("simple")));
 855                         parse2(cell->back(), FLAG_SIMPLE2, MathInset::MATH_MODE, false);
 856                 }
 857
 858                 else if (t.cs() == "[") {
 859                         cell->push_back(MathAtom(new MathHullInset("equation")));
 860                         parse2(cell->back(), FLAG_EQUATION, MathInset::MATH_MODE, false);
 861                 }
 862
 863                 else if (t.cs() == "protect")
 864                         // ignore \\protect, will hopefully be re-added during output
 865                         ;
 866
 867                 else if (t.cs() == "end") {
 868                         if (flags & FLAG_END) {
 869                                 // eat environment name
 870                                 //string const name =
 871                                 getArg('{', '}');
 872                                 // FIXME: check that we ended the correct environment
 873                                 return;
 874                         }
 875                         error("found 'end' unexpectedly");
 876                 }
 877
 878                 else if (t.cs() == ")") {
 879                         if (flags & FLAG_SIMPLE2)
 880                                 return;
 881                         error("found '\\)' unexpectedly");
 882                 }
 883
 884                 else if (t.cs() == "]") {
 885                         if (flags & FLAG_EQUATION)
 886                                 return;
 887                         error("found '\\]' unexpectedly");
 888                 }
 889
 890                 else if (t.cs() == "\\") {
 891                         grid.vcrskip(LyXLength(getArg('[', ']')), cellrow);
 892                         ++cellrow;
 893                         cellcol = 0;
 894                         if (cellrow == grid.nrows())
 895                                 grid.addRow(cellrow - 1);
 896                         if (grid.asHullInset())
 897                                 grid.asHullInset()->numbered(cellrow, numbered);
 898                         cell = &grid.cell(grid.index(cellrow, cellcol));
 899                 }
 900
 901 #if 0
 902                 else if (t.cs() == "multicolumn") {
 903                         // extract column count and insert dummy cells
 904                         MathArray count;
 905                         parse(count, FLAG_ITEM, mode);
 906                         int cols = 1;
 907                         if (!extractNumber(count, cols)) {
 908                                 lyxerr << " can't extract number of cells from " << count << endl;
 909                         }
 910                         // resize the table if necessary
 911                         for (int i = 0; i < cols; ++i) {
 912                                 ++cellcol;
 913                                 if (cellcol == grid.ncols()) {
 914                                         //lyxerr << "adding column " << cellcol << endl;
 915                                         grid.addCol(cellcol - 1);
 916                                 }
 917                                 cell = &grid.cell(grid.index(cellrow, cellcol));
 918                                 // mark this as dummy
 919                                 grid.cellinfo(grid.index(cellrow, cellcol)).dummy_ = true;
 920                         }
 921                         // the last cell is the real thng, not a dummy
 922                         grid.cellinfo(grid.index(cellrow, cellcol)).dummy_ = false;
 923
 924                         // read special alignment
 925                         MathArray align;
 926                         parse(align, FLAG_ITEM, mode);
 927                         //grid.cellinfo(grid.index(cellrow, cellcol)).align_ = extractString(align);
 928
 929                         // parse the remaining contents into the "real" cell
 930                         parse(*cell, FLAG_ITEM, mode);
 931                 }
 932 #endif
 933
 934                 else if (t.cs() == "limits")
 935                         limits = 1;
 936
 937                 else if (t.cs() == "nolimits")
 938                         limits = -1;
 939
 940                 else if (t.cs() == "nonumber") {
 941                         if (grid.asHullInset())
 942                                 grid.asHullInset()->numbered(cellrow, false);
 943                 }
 944
 945                 else if (t.cs() == "number") {
 946                         if (grid.asHullInset())
 947                                 grid.asHullInset()->numbered(cellrow, true);
 948                 }
 949
 950                 else if (t.cs() == "hline") {
 951                         grid.rowinfo(cellrow).lines_ ++;
 952                 }
 953
 954                 else if (t.cs() == "sqrt") {
 955                         MathArray ar;
 956                         parse(ar, FLAG_OPTION, mode);
 957                         if (ar.size()) {
 958                                 cell->push_back(MathAtom(new MathRootInset));
 959                                 cell->back().nucleus()->cell(0) = ar;
 960                                 parse(cell->back().nucleus()->cell(1), FLAG_ITEM, mode);
 961                         } else {
 962                                 cell->push_back(MathAtom(new MathSqrtInset));
 963                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
 964                         }
 965                 }
 966
 967                 else if (t.cs() == "xrightarrow" || t.cs() == "xleftarrow") {
 968                         cell->push_back(createMathInset(t.cs()));
 969                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
 970                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
 971                 }
 972
 973                 else if (t.cs() == "ref" || t.cs() == "prettyref" ||
 974                                 t.cs() == "pageref" || t.cs() == "vpageref" || t.cs() == "vref") {
 975                         cell->push_back(MathAtom(new RefInset(t.cs())));
 976                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
 977                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
 978                 }
 979
 980                 else if (t.cs() == "left") {
 981                         skipSpaces();
 982                         string l = getToken().asString();
 983                         MathArray ar;
 984                         parse(ar, FLAG_RIGHT, mode);
 985                         skipSpaces();
 986                         string r = getToken().asString();
 987                         cell->push_back(MathAtom(new MathDelimInset(l, r, ar)));
 988                 }
 989
 990                 else if (t.cs() == "right") {
 991                         if (flags & FLAG_RIGHT)
 992                                 return;
 993                         //lyxerr << "got so far: '" << cell << "'" << endl;
 994                         error("Unmatched right delimiter");
 995                         return;
 996                 }
 997
 998                 else if (t.cs() == "begin") {
 999                         string const name = getArg('{', '}');
1000
1001                         if (name == "array" || name == "subarray") {
1002                                 string const valign = parse_verbatim_option() + 'c';
1003                                 string const halign = parse_verbatim_item();
1004                                 cell->push_back(MathAtom(new MathArrayInset(name, valign[0], halign)));
1005                                 parse2(cell->back(), FLAG_END, mode, false);
1006                         }
1007
1008                         else if (name == "tabular") {
1009                                 string const valign = parse_verbatim_option() + 'c';
1010                                 string const halign = parse_verbatim_item();
1011                                 cell->push_back(MathAtom(new MathTabularInset(name, valign[0], halign)));
1012                                 parse2(cell->back(), FLAG_END, MathInset::TEXT_MODE, false);
1013                         }
1014
1015                         else if (name == "split" || name == "cases" ||
1016                                          name == "gathered" || name == "aligned") {
1017                                 cell->push_back(createMathInset(name));
1018                                 parse2(cell->back(), FLAG_END, mode, false);
1019                         }
1020
1021                         else if (name == "math") {
1022                                 cell->push_back(MathAtom(new MathHullInset("simple")));
1023                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, true);
1024                         }
1025
1026                         else if (name == "equation" || name == "equation*"
1027                                         || name == "displaymath") {
1028                                 cell->push_back(MathAtom(new MathHullInset("equation")));
1029                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, (name == "equation"));
1030                         }
1031
1032                         else if (name == "eqnarray" || name == "eqnarray*") {
1033                                 cell->push_back(MathAtom(new MathHullInset("eqnarray")));
1034                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1035                         }
1036
1037                         else if (name == "align" || name == "align*") {
1038                                 cell->push_back(MathAtom(new MathHullInset("align")));
1039                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1040                         }
1041
1042                         else if (name == "flalign" || name == "flalign*") {
1043                                 cell->push_back(MathAtom(new MathHullInset("flalign")));
1044                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1045                         }
1046
1047                         else if (name == "alignat" || name == "alignat*") {
1048                                 // ignore this for a while
1049                                 getArg('{', '}');
1050                                 cell->push_back(MathAtom(new MathHullInset("alignat")));
1051                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1052                         }
1053
1054                         else if (name == "xalignat" || name == "xalignat*") {
1055                                 // ignore this for a while
1056                                 getArg('{', '}');
1057                                 cell->push_back(MathAtom(new MathHullInset("xalignat")));
1058                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1059                         }
1060
1061                         else if (name == "xxalignat") {
1062                                 // ignore this for a while
1063                                 getArg('{', '}');
1064                                 cell->push_back(MathAtom(new MathHullInset("xxalignat")));
1065                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1066                         }
1067
1068                         else if (name == "multline" || name == "multline*") {
1069                                 cell->push_back(MathAtom(new MathHullInset("multline")));
1070                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1071                         }
1072
1073                         else if (name == "gather" || name == "gather*") {
1074                                 cell->push_back(MathAtom(new MathHullInset("gather")));
1075                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1076                         }
1077
1078                         else if (latexkeys const * l = in_word_set(name)) {
1079                                 if (l->inset == "matrix") {
1080                                         cell->push_back(createMathInset(name));
1081                                         parse2(cell->back(), FLAG_END, mode, false);
1082                                 }
1083                         }
1084
1085                         else {
1086                                 dump();
1087                                 lyxerr << "found unknown math environment '" << name << "'" << endl;
1088                                 // create generic environment inset
1089                                 cell->push_back(MathAtom(new MathEnvInset(name)));
1090                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1091                         }
1092                 }
1093
1094                 else if (t.cs() == "kern") {
1095 #ifdef WITH_WARNINGS
1096 #warning A hack...
1097 #endif
1098                         string s;
1099                         while (true) {
1100                                 Token const & t = getToken();
1101                                 if (!good()) {
1102                                         putback();
1103                                         break;
1104                                 }
1105                                 s += t.character();
1106                                 if (isValidLength(s))
1107                                         break;
1108                         }
1109                         cell->push_back(MathAtom(new MathKernInset(s)));
1110                 }
1111
1112                 else if (t.cs() == "label") {
1113                         string label = parse_verbatim_item();
1114                         MathArray ar;
1115                         asArray(label, ar);
1116                         if (grid.asHullInset()) {
1117                                 grid.asHullInset()->label(cellrow, label);
1118                         } else {
1119                                 cell->push_back(createMathInset(t.cs()));
1120                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
1121                         }
1122                 }
1123
1124                 else if (t.cs() == "choose" || t.cs() == "over" || t.cs() == "atop") {
1125                         MathAtom at = createMathInset(t.cs());
1126                         at.nucleus()->cell(0) = *cell;
1127                         cell->clear();
1128                         parse(at.nucleus()->cell(1), flags, mode);
1129                         cell->push_back(at);
1130                         return;
1131                 }
1132
1133                 else if (t.cs() == "substack") {
1134                         cell->push_back(createMathInset(t.cs()));
1135                         parse2(cell->back(), FLAG_ITEM, mode, false);
1136                 }
1137
1138                 else if (t.cs() == "framebox" || t.cs() == "makebox") {
1139                         cell->push_back(createMathInset(t.cs()));
1140                         parse(cell->back().nucleus()->cell(0), FLAG_OPTION, MathInset::TEXT_MODE);
1141                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, MathInset::TEXT_MODE);
1142                         parse(cell->back().nucleus()->cell(2), FLAG_ITEM, MathInset::TEXT_MODE);
1143                 }
1144
1145 #if 0
1146                 else if (t.cs() == "infer") {
1147                         MathArray ar;
1148                         parse(ar, FLAG_OPTION, mode);
1149                         cell->push_back(createMathInset(t.cs()));
1150                         parse2(cell->back(), FLAG_ITEM, mode, false);
1151                 }
1152
1153                 // Disabled
1154                 else if (1 && t.cs() == "ar") {
1155                         MathXYArrowInset * p = new MathXYArrowInset;
1156                         // try to read target
1157                         parse(p->cell(0), FLAG_OTPTION, mode);
1158                         // try to read label
1159                         if (nextToken().cat() == catSuper || nextToken().cat() == catSub) {
1160                                 p->up_ = nextToken().cat() == catSuper;
1161                                 getToken();
1162                                 parse(p->cell(1), FLAG_ITEM, mode);
1163                                 //lyxerr << "read label: " << p->cell(1) << endl;
1164                         }
1165
1166                         cell->push_back(MathAtom(p));
1167                         //lyxerr << "read cell: " << cell << endl;
1168                 }
1169 #endif
1170
1171                 else if (t.cs().size()) {
1172                         latexkeys const * l = in_word_set(t.cs());
1173                         if (l) {
1174                                 if (l->inset == "font") {
1175                                         cell->push_back(createMathInset(t.cs()));
1176                                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, asMode(mode, l->extra));
1177                                 }
1178
1179                                 else if (l->inset == "oldfont") {
1180                                         cell->push_back(createMathInset(t.cs()));
1181                                         parse(cell->back().nucleus()->cell(0), flags, asMode(mode, l->extra));
1182                                         return;
1183                                 }
1184
1185                                 else if (l->inset == "style") {
1186                                         cell->push_back(createMathInset(t.cs()));
1187                                         parse(cell->back().nucleus()->cell(0), flags, mode);
1188                                         return;
1189                                 }
1190
1191                                 else if (l->inset == "parbox") {
1192                                         // read optional positioning and width
1193                                         string pos   = parse_verbatim_option();
1194                                         string width = parse_verbatim_item();
1195                                         cell->push_back(createMathInset(t.cs()));
1196                                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, MathInset::TEXT_MODE);
1197                                         cell->back().nucleus()->asParboxInset()->setPosition(pos);
1198                                         cell->back().nucleus()->asParboxInset()->setWidth(width);
1199                                 }
1200
1201                                 else {
1202                                         MathAtom at = createMathInset(t.cs());
1203                                         for (MathInset::idx_type i = 0; i < at->nargs(); ++i)
1204                                                 parse(at.nucleus()->cell(i), FLAG_ITEM, asMode(mode, l->extra));
1205                                         cell->push_back(at);
1206                                 }
1207                         }
1208
1209                         else {
1210                                 MathAtom at = createMathInset(t.cs());
1211                                 MathInset::mode_type m = mode;
1212                                 //if (m == MathInset::UNDECIDED_MODE)
1213                                 //lyxerr << "default creation: m1: " << m << endl;
1214                                 if (at->currentMode() != MathInset::UNDECIDED_MODE)
1215                                         m = at->currentMode();
1216                                 //lyxerr << "default creation: m2: " << m << endl;
1217                                 MathInset::idx_type start = 0;
1218                                 // this fails on \bigg[...\bigg]
1219                                 //MathArray opt;
1220                                 //parse(opt, FLAG_OPTION, MathInset::VERBATIM_MODE);
1221                                 //if (opt.size()) {
1222                                 //      start = 1;
1223                                 //      at.nucleus()->cell(0) = opt;
1224                                 //}
1225                                 for (MathInset::idx_type i = start; i < at->nargs(); ++i)
1226                                         parse(at.nucleus()->cell(i), FLAG_ITEM, m);
1227                                 cell->push_back(at);
1228                         }
1229                 }
1230
1231
1232                 if (flags & FLAG_LEAVE) {
1233                         flags &= ~FLAG_LEAVE;
1234                         break;
1235                 }
1236         }
1237 }
1238
1239
1240
1241 } // anonymous namespace
1242
1243
1244 void mathed_parse_cell(MathArray & ar, string const & str)
1245 {
1246         istringstream is(str.c_str());
1247         mathed_parse_cell(ar, is);
1248 }
1249
1250
1251 void mathed_parse_cell(MathArray & ar, istream & is)
1252 {
1253         Parser(is).parse(ar, 0, MathInset::MATH_MODE);
1254 }
1255
1256
1257 bool mathed_parse_normal(MathAtom & t, string const & str)
1258 {
1259         istringstream is(str.c_str());
1260         return Parser(is).parse(t);
1261 }
1262
1263
1264 bool mathed_parse_normal(MathAtom & t, istream & is)
1265 {
1266         return Parser(is).parse(t);
1267 }
1268
1269
1270 bool mathed_parse_normal(MathAtom & t, LyXLex & lex)
1271 {
1272         return Parser(lex).parse(t);
1273 }
1274
1275
1276 void mathed_parse_normal(MathGridInset & grid, string const & str)
1277 {
1278         istringstream is(str.c_str());
1279         Parser(is).parse1(grid, 0, MathInset::MATH_MODE, false);
1280 }