src/mathed/math_parser.C

   1 /** The math parser
   2     \author André Pönitz (2001)
   3  */
   4
   5 /*
   6
   7 If someone desperately needs partial "structures" (such as a few
   8 cells of an array inset or similar) (s)he could uses the
   9 following hack as starting point to write some macros:
  10
  11   \newif\ifcomment
  12   \commentfalse
  13   \ifcomment
  14           \def\makeamptab{\catcode`\&=4\relax}
  15           \def\makeampletter{\catcode`\&=11\relax}
  16     \def\b{\makeampletter\expandafter\makeamptab\bi}
  17     \long\def\bi#1\e{}
  18   \else
  19     \def\b{}\def\e{}
  20   \fi
  21
  22   ...
  23
  24   \[\begin{array}{ccc}
  25 1
  26 &
  27
  28   \end{array}\]
  29
  30 */
  31
  32
  33 #include <config.h>
  34
  35
  36 #include "math_parser.h"
  37 #include "math_inset.h"
  38 #include "math_arrayinset.h"
  39 #include "math_braceinset.h"
  40 #include "math_boxinset.h"
  41 #include "math_charinset.h"
  42 #include "math_commentinset.h"
  43 #include "math_deliminset.h"
  44 #include "math_envinset.h"
  45 #include "math_extern.h"
  46 #include "math_factory.h"
  47 #include "math_kerninset.h"
  48 #include "math_macro.h"
  49 #include "math_macrotemplate.h"
  50 #include "math_hullinset.h"
  51 #include "math_parboxinset.h"
  52 #include "math_parinset.h"
  53 #include "math_rootinset.h"
  54 #include "math_scriptinset.h"
  55 #include "math_sizeinset.h"
  56 #include "math_sqrtinset.h"
  57 #include "math_stringinset.h"
  58 #include "math_support.h"
  59 #include "math_tabularinset.h"
  60 #include "math_xyarrowinset.h"
  61
  62 //#include "insets/insetref.h"
  63 #include "ref_inset.h"
  64
  65 #include "lyxlex.h"
  66 #include "debug.h"
  67 #include "support/LAssert.h"
  68 #include "support/lstrings.h"
  69
  70 #include <cctype>
  71 #include <algorithm>
  72
  73 using std::istream;
  74 using std::ostream;
  75 using std::ios;
  76 using std::endl;
  77 using std::fill;
  78 using std::vector;
  79 using std::atoi;
  80
  81
  82 //#define FILEDEBUG
  83
  84
  85 namespace {
  86
  87 MathInset::mode_type asMode(MathInset::mode_type oldmode, string const & str)
  88 {
  89         if (str == "mathmode")
  90                 return MathInset::MATH_MODE;
  91         if (str == "textmode" || str == "forcetext")
  92                 return MathInset::TEXT_MODE;
  93         return oldmode;
  94 }
  95
  96
  97 bool stared(string const & s)
  98 {
  99         string::size_type const n = s.size();
 100         return n && s[n - 1] == '*';
 101 }
 102
 103
 104 // These are TeX's catcodes
 105 enum CatCode {
 106         catEscape,     // 0    backslash
 107         catBegin,      // 1    {
 108         catEnd,        // 2    }
 109         catMath,       // 3    $
 110         catAlign,      // 4    &
 111         catNewline,    // 5    ^^M
 112         catParameter,  // 6    #
 113         catSuper,      // 7    ^
 114         catSub,        // 8    _
 115         catIgnore,     // 9
 116         catSpace,      // 10   space
 117         catLetter,     // 11   a-zA-Z
 118         catOther,      // 12   none of the above
 119         catActive,     // 13   ~
 120         catComment,    // 14   %
 121         catInvalid     // 15   <delete>
 122 };
 123
 124 CatCode theCatcode[256];
 125
 126
 127 inline CatCode catcode(unsigned char c)
 128 {
 129         return theCatcode[c];
 130 }
 131
 132
 133 enum {
 134         FLAG_BRACE_LAST = 1 << 1,  //  last closing brace ends the parsing
 135         FLAG_RIGHT      = 1 << 2,  //  next \\right ends the parsing process
 136         FLAG_END        = 1 << 3,  //  next \\end ends the parsing process
 137         FLAG_BRACK_LAST = 1 << 4,  //  next closing bracket ends the parsing
 138         FLAG_TEXTMODE   = 1 << 5,  //  we are in a box
 139         FLAG_ITEM       = 1 << 6,  //  read a (possibly braced token)
 140         FLAG_LEAVE      = 1 << 7,  //  leave the loop at the end
 141         FLAG_SIMPLE     = 1 << 8,  //  next $ leaves the loop
 142         FLAG_EQUATION   = 1 << 9,  //  next \] leaves the loop
 143         FLAG_SIMPLE2    = 1 << 10, //  next \) leaves the loop
 144         FLAG_OPTION     = 1 << 11, //  read [...] style option
 145         FLAG_BRACED     = 1 << 12  //  read {...} style argument
 146 };
 147
 148
 149 void catInit()
 150 {
 151         fill(theCatcode, theCatcode + 256, catOther);
 152         fill(theCatcode + 'a', theCatcode + 'z' + 1, catLetter);
 153         fill(theCatcode + 'A', theCatcode + 'Z' + 1, catLetter);
 154
 155         theCatcode['\\'] = catEscape;
 156         theCatcode['{']  = catBegin;
 157         theCatcode['}']  = catEnd;
 158         theCatcode['$']  = catMath;
 159         theCatcode['&']  = catAlign;
 160         theCatcode['\n'] = catNewline;
 161         theCatcode['#']  = catParameter;
 162         theCatcode['^']  = catSuper;
 163         theCatcode['_']  = catSub;
 164         theCatcode['\7f'] = catIgnore;
 165         theCatcode[' ']  = catSpace;
 166         theCatcode['\t'] = catSpace;
 167         theCatcode['\r'] = catNewline;
 168         theCatcode['~']  = catActive;
 169         theCatcode['%']  = catComment;
 170 }
 171
 172
 173
 174 //
 175 // Helper class for parsing
 176 //
 177
 178 class Token {
 179 public:
 180         ///
 181         Token() : cs_(), char_(0), cat_(catIgnore) {}
 182         ///
 183         Token(char c, CatCode cat) : cs_(), char_(c), cat_(cat) {}
 184         ///
 185         Token(string const & cs) : cs_(cs), char_(0), cat_(catIgnore) {}
 186
 187         ///
 188         string const & cs() const { return cs_; }
 189         ///
 190         CatCode cat() const { return cat_; }
 191         ///
 192         char character() const { return char_; }
 193         ///
 194         string asString() const { return cs_.size() ? cs_ : string(1, char_); }
 195
 196 private:
 197         ///
 198         string cs_;
 199         ///
 200         char char_;
 201         ///
 202         CatCode cat_;
 203 };
 204
 205 ostream & operator<<(ostream & os, Token const & t)
 206 {
 207         if (t.cs().size())
 208                 os << '\\' << t.cs();
 209         else
 210                 os << '[' << t.character() << ',' << t.cat() << ']';
 211         return os;
 212 }
 213
 214
 215 class Parser {
 216
 217 public:
 218         ///
 219         typedef  MathInset::mode_type mode_type;
 220
 221         ///
 222         Parser(LyXLex & lex);
 223         ///
 224         Parser(istream & is);
 225
 226         ///
 227         bool parse(MathAtom & at);
 228         ///
 229         void parse(MathArray & array, unsigned flags, mode_type mode);
 230         ///
 231         void parse1(MathGridInset & grid, unsigned flags, mode_type mode,
 232                 bool numbered);
 233         ///
 234         MathArray parse(unsigned flags, mode_type mode);
 235         ///
 236         int lineno() const { return lineno_; }
 237         ///
 238         void putback();
 239
 240 private:
 241         ///
 242         void parse2(MathAtom & at, unsigned flags, mode_type mode, bool numbered);
 243         /// get arg delimited by 'left' and 'right'
 244         string getArg(char left, char right);
 245         ///
 246         char getChar();
 247         ///
 248         void error(string const & msg);
 249         /// dump contents to screen
 250         void dump() const;
 251         ///
 252         void tokenize(istream & is);
 253         ///
 254         void tokenize(string const & s);
 255         ///
 256         void skipSpaceTokens(istream & is, char c);
 257         ///
 258         void push_back(Token const & t);
 259         ///
 260         void pop_back();
 261         ///
 262         Token const & prevToken() const;
 263         ///
 264         Token const & nextToken() const;
 265         ///
 266         Token const & getToken();
 267         /// skips spaces if any
 268         void skipSpaces();
 269         ///
 270         void lex(string const & s);
 271         ///
 272         bool good() const;
 273         ///
 274         string parse_verbatim_item();
 275         ///
 276         string parse_verbatim_option();
 277
 278         ///
 279         int lineno_;
 280         ///
 281         vector<Token> tokens_;
 282         ///
 283         unsigned pos_;
 284 };
 285
 286
 287 Parser::Parser(LyXLex & lexer)
 288         : lineno_(lexer.getLineNo()), pos_(0)
 289 {
 290         tokenize(lexer.getStream());
 291         lexer.eatLine();
 292 }
 293
 294
 295 Parser::Parser(istream & is)
 296         : lineno_(0), pos_(0)
 297 {
 298         tokenize(is);
 299 }
 300
 301
 302 void Parser::push_back(Token const & t)
 303 {
 304         tokens_.push_back(t);
 305 }
 306
 307
 308 void Parser::pop_back()
 309 {
 310         tokens_.pop_back();
 311 }
 312
 313
 314 Token const & Parser::prevToken() const
 315 {
 316         static const Token dummy;
 317         return pos_ > 0 ? tokens_[pos_ - 1] : dummy;
 318 }
 319
 320
 321 Token const & Parser::nextToken() const
 322 {
 323         static const Token dummy;
 324         return good() ? tokens_[pos_] : dummy;
 325 }
 326
 327
 328 Token const & Parser::getToken()
 329 {
 330         static const Token dummy;
 331         //lyxerr << "looking at token " << tokens_[pos_] << " pos: " << pos_ << '\n';
 332         return good() ? tokens_[pos_++] : dummy;
 333 }
 334
 335
 336 void Parser::skipSpaces()
 337 {
 338         while (nextToken().cat() == catSpace || nextToken().cat() == catNewline)
 339                 getToken();
 340 }
 341
 342
 343 void Parser::putback()
 344 {
 345         --pos_;
 346 }
 347
 348
 349 bool Parser::good() const
 350 {
 351         return pos_ < tokens_.size();
 352 }
 353
 354
 355 char Parser::getChar()
 356 {
 357         if (!good())
 358                 error("The input stream is not well...");
 359         return tokens_[pos_++].character();
 360 }
 361
 362
 363 string Parser::getArg(char left, char right)
 364 {
 365         skipSpaces();
 366
 367         string result;
 368         char c = getChar();
 369
 370         if (c != left)
 371                 putback();
 372         else
 373                 while ((c = getChar()) != right && good())
 374                         result += c;
 375
 376         return result;
 377 }
 378
 379
 380 void Parser::skipSpaceTokens(istream & is, char c)
 381 {
 382         // skip trailing spaces
 383         while (catcode(c) == catSpace || catcode(c) == catNewline)
 384                 if (!is.get(c))
 385                         break;
 386         //lyxerr << "putting back: " << c << "\n";
 387         is.putback(c);
 388 }
 389
 390
 391 void Parser::tokenize(istream & is)
 392 {
 393         // eat everything up to the next \end_inset or end of stream
 394         // and store it in s for further tokenization
 395         string s;
 396         char c;
 397         while (is.get(c)) {
 398                 s += c;
 399                 if (s.size() >= 10 && s.substr(s.size() - 10) == "\\end_inset") {
 400                         s = s.substr(0, s.size() - 10);
 401                         break;
 402                 }
 403         }
 404         // Remove the space after \end_inset
 405         if (is.get(c) && c != ' ')
 406                 is.unget();
 407
 408         // tokenize buffer
 409         tokenize(s);
 410 }
 411
 412
 413 void Parser::tokenize(string const & buffer)
 414 {
 415         static bool init_done = false;
 416
 417         if (!init_done) {
 418                 catInit();
 419                 init_done = true;
 420         }
 421
 422         istringstream is(buffer.c_str(), ios::in | ios::binary);
 423
 424         char c;
 425         while (is.get(c)) {
 426                 //lyxerr << "reading c: " << c << "\n";
 427
 428                 switch (catcode(c)) {
 429                         case catNewline: {
 430                                 ++lineno_;
 431                                 is.get(c);
 432                                 if (catcode(c) == catNewline)
 433                                         ; //push_back(Token("par"));
 434                                 else {
 435                                         push_back(Token('\n', catNewline));
 436                                         is.putback(c);
 437                                 }
 438                                 break;
 439                         }
 440
 441 /*
 442                         case catComment: {
 443                                 while (is.get(c) && catcode(c) != catNewline)
 444                                         ;
 445                                 ++lineno_;
 446                                 break;
 447                         }
 448 */
 449
 450                         case catEscape: {
 451                                 is.get(c);
 452                                 if (!is) {
 453                                         error("unexpected end of input");
 454                                 } else {
 455                                         string s(1, c);
 456                                         if (catcode(c) == catLetter) {
 457                                                 // collect letters
 458                                                 while (is.get(c) && catcode(c) == catLetter)
 459                                                         s += c;
 460                                                 skipSpaceTokens(is, c);
 461                                         }
 462                                         push_back(Token(s));
 463                                 }
 464                                 break;
 465                         }
 466
 467                         case catSuper:
 468                         case catSub: {
 469                                 push_back(Token(c, catcode(c)));
 470                                 is.get(c);
 471                                 skipSpaceTokens(is, c);
 472                                 break;
 473                         }
 474
 475                         case catIgnore: {
 476                                 lyxerr << "ignoring a char: " << int(c) << "\n";
 477                                 break;
 478                         }
 479
 480                         default:
 481                                 push_back(Token(c, catcode(c)));
 482                 }
 483         }
 484
 485 #ifdef FILEDEBUG
 486         dump();
 487 #endif
 488 }
 489
 490
 491 void Parser::dump() const
 492 {
 493         lyxerr << "\nTokens: ";
 494         for (unsigned i = 0; i < tokens_.size(); ++i) {
 495                 if (i == pos_)
 496                         lyxerr << " <#> ";
 497                 lyxerr << tokens_[i];
 498         }
 499         lyxerr << " pos: " << pos_ << "\n";
 500 }
 501
 502
 503 void Parser::error(string const & msg)
 504 {
 505         lyxerr << "Line ~" << lineno_ << ": Math parse error: " << msg << endl;
 506         dump();
 507         //exit(1);
 508 }
 509
 510
 511 bool Parser::parse(MathAtom & at)
 512 {
 513         skipSpaces();
 514         MathArray ar;
 515         parse(ar, false, MathInset::UNDECIDED_MODE);
 516         if (ar.size() != 1 || ar.front()->getType() == "none") {
 517                 lyxerr << "unusual contents found: " << ar << endl;
 518                 at = MathAtom(new MathParInset(ar));
 519                 //if (at->nargs() > 0)
 520                 //      at.nucleus()->cell(0) = ar;
 521                 //else
 522                 //      lyxerr << "unusual contents found: " << ar << endl;
 523                 return true;
 524         }
 525         at = ar[0];
 526         return true;
 527 }
 528
 529
 530 string Parser::parse_verbatim_option()
 531 {
 532         string res;
 533         if (nextToken().character() == '[') {
 534                 Token t = getToken();
 535                 for (Token t = getToken(); t.character() != ']' && good(); t = getToken()) {
 536                         if (t.cat() == catBegin) {
 537                                 putback();
 538                                 res += '{' + parse_verbatim_item() + '}';
 539                         } else
 540                                 res += t.asString();
 541                 }
 542         }
 543         return res;
 544 }
 545
 546
 547 string Parser::parse_verbatim_item()
 548 {
 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, unsigned flags, mode_type mode,
 582         bool numbered)
 583 {
 584         parse1(*(at.nucleus()->asGridInset()), flags, mode, numbered);
 585 }
 586
 587
 588 void Parser::parse1(MathGridInset & grid, unsigned flags,
 589         mode_type mode, 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 << "grid: " << grid << endl;
 601
 602         while (good()) {
 603                 Token const & t = getToken();
 604
 605 #ifdef FILEDEBUG
 606                 lyxerr << "t: " << t << " flags: " << flags << "\n";
 607                 cell->dump();
 608                 lyxerr << "\n";
 609 #endif
 610
 611                 if (flags & FLAG_ITEM) {
 612                         if (t.cat() == catSpace)
 613                                 continue;
 614
 615                         flags &= ~FLAG_ITEM;
 616                         if (t.cat() == catBegin) {
 617                                 // skip the brace and collect everything to the next matching
 618                                 // closing brace
 619                                 flags |= FLAG_BRACE_LAST;
 620                                 continue;
 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\n");
 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                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 696
 697                 else if (t.cat() == catParameter) {
 698                         Token const & n = getToken();
 699                         cell->push_back(MathAtom(new MathMacroArgument(n.character()-'0')));
 700                 }
 701
 702                 else if (t.cat() == catActive)
 703                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 704
 705                 else if (t.cat() == catBegin) {
 706                         MathArray ar;
 707                         parse(ar, FLAG_BRACE_LAST, mode);
 708                         // do not create a BraceInset if they were written by LyX
 709                         // this helps to keep the annoyance of  "a choose b"  to a minimum
 710                         if (ar.size() == 1 && ar[0]->extraBraces())
 711                                 cell->append(ar);
 712                         else
 713                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
 714                 }
 715
 716                 else if (t.cat() == catEnd) {
 717                         if (flags & FLAG_BRACE_LAST)
 718                                 return;
 719                         error("found '}' unexpectedly");
 720                         //lyx::Assert(0);
 721                         //add(cell, '}', LM_TC_TEX);
 722                 }
 723
 724                 else if (t.cat() == catAlign) {
 725                         ++cellcol;
 726                         //lyxerr << " column now " << cellcol << " max: " << grid.ncols() << "\n";
 727                         if (cellcol == grid.ncols()) {
 728                                 //lyxerr << "adding column " << cellcol << "\n";
 729                                 grid.addCol(cellcol - 1);
 730                         }
 731                         cell = &grid.cell(grid.index(cellrow, cellcol));
 732                 }
 733
 734                 else if (t.cat() == catSuper || t.cat() == catSub) {
 735                         bool up = (t.cat() == catSuper);
 736                         // we need no new script inset if the last thing was a scriptinset,
 737                         // which has that script already not the same script already
 738                         if (!cell->size())
 739                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 740                         else if (cell->back()->asScriptInset() &&
 741                                         !cell->back()->asScriptInset()->has(up))
 742                                 cell->back().nucleus()->asScriptInset()->ensure(up);
 743                         else if (cell->back()->asScriptInset())
 744                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 745                         else
 746                                 cell->back() = MathAtom(new MathScriptInset(cell->back(), up));
 747                         MathScriptInset * p = cell->back().nucleus()->asScriptInset();
 748                         // special handling of {}-bases
 749                         // is this always correct?
 750                         // It appears that this is wrong (Dekel)
 751                         //if (p->nuc().size() == 1 && p->nuc().back()->asNestInset() &&
 752                         //    p->nuc().back()->extraBraces())
 753                         //      p->nuc() = p->nuc().back()->asNestInset()->cell(0);
 754                         parse(p->cell(up), FLAG_ITEM, mode);
 755                         if (limits) {
 756                                 p->limits(limits);
 757                                 limits = 0;
 758                         }
 759                 }
 760
 761                 else if (t.character() == ']' && (flags & FLAG_BRACK_LAST)) {
 762                         //lyxerr << "finished reading option\n";
 763                         return;
 764                 }
 765
 766                 else if (t.cat() == catOther)
 767                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 768
 769                 else if (t.cat() == catComment) {
 770                         string s;
 771                         while (good()) {
 772                                 Token const & t = getToken();
 773                                 if (t.cat() == catNewline)
 774                                         break;
 775                                 s += t.asString();
 776                         }
 777                         cell->push_back(MathAtom(new MathCommentInset(s)));
 778                         skipSpaces();
 779                 }
 780
 781                 //
 782                 // control sequences
 783                 //
 784
 785                 else if (t.cs() == "lyxlock") {
 786                         if (cell->size())
 787                                 cell->back().nucleus()->lock(true);
 788                 }
 789
 790                 else if (t.cs() == "def" || t.cs() == "newcommand") {
 791                         string name;
 792                         int nargs = 0;
 793                         if (t.cs() == "def") {
 794                                 // get name
 795                                 name = getToken().cs();
 796
 797                                 // read parameter
 798                                 string pars;
 799                                 while (good() && nextToken().cat() != catBegin) {
 800                                         pars += getToken().cs();
 801                                         ++nargs;
 802                                 }
 803                                 nargs /= 2;
 804                                 //lyxerr << "read \\def parameter list '" << pars << "'\n";
 805
 806                         } else { // t.cs() == "newcommand"
 807
 808                                 if (getToken().cat() != catBegin) {
 809                                         error("'{' in \\newcommand expected (1) \n");
 810                                         return;
 811                                 }
 812
 813                                 name = getToken().cs();
 814
 815                                 if (getToken().cat() != catEnd) {
 816                                         error("'}' in \\newcommand expected\n");
 817                                         return;
 818                                 }
 819
 820                                 string arg  = getArg('[', ']');
 821                                 if (!arg.empty())
 822                                         nargs = atoi(arg.c_str());
 823
 824                         }
 825
 826                         MathArray ar1;
 827                         parse(ar1, FLAG_ITEM, MathInset::UNDECIDED_MODE);
 828
 829                         // we cannot handle recursive stuff at all
 830                         //MathArray test;
 831                         //test.push_back(createMathInset(name));
 832                         //if (ar1.contains(test)) {
 833                         //      error("we cannot handle recursive macros at all.\n");
 834                         //      return;
 835                         //}
 836
 837                         // is a version for display attached?
 838                         skipSpaces();
 839                         MathArray ar2;
 840                         if (nextToken().cat() == catBegin) {
 841                                 parse(ar2, FLAG_ITEM, MathInset::MATH_MODE);
 842                         }
 843
 844                         cell->push_back(MathAtom(new MathMacroTemplate(name, nargs, ar1, ar2)));
 845                 }
 846
 847                 else if (t.cs() == "(") {
 848                         cell->push_back(MathAtom(new MathHullInset("simple")));
 849                         parse2(cell->back(), FLAG_SIMPLE2, MathInset::MATH_MODE, false);
 850                 }
 851
 852                 else if (t.cs() == "[") {
 853                         cell->push_back(MathAtom(new MathHullInset("equation")));
 854                         parse2(cell->back(), FLAG_EQUATION, MathInset::MATH_MODE, false);
 855                 }
 856
 857                 else if (t.cs() == "protect")
 858                         // ignore \\protect, will hopefully be re-added during output
 859                         ;
 860
 861                 else if (t.cs() == "end") {
 862                         if (flags & FLAG_END) {
 863                                 // eat environment name
 864                                 //string const name =
 865                                 getArg('{', '}');
 866                                 // FIXME: check that we ended the correct environment
 867                                 return;
 868                         }
 869                         error("found 'end' unexpectedly");
 870                 }
 871
 872                 else if (t.cs() == ")") {
 873                         if (flags & FLAG_SIMPLE2)
 874                                 return;
 875                         error("found '\\)' unexpectedly");
 876                 }
 877
 878                 else if (t.cs() == "]") {
 879                         if (flags & FLAG_EQUATION)
 880                                 return;
 881                         error("found '\\]' unexpectedly");
 882                 }
 883
 884                 else if (t.cs() == "\\") {
 885                         grid.vcrskip(LyXLength(getArg('[', ']')), cellrow);
 886                         ++cellrow;
 887                         cellcol = 0;
 888                         if (cellrow == grid.nrows())
 889                                 grid.addRow(cellrow - 1);
 890                         if (grid.asHullInset())
 891                                 grid.asHullInset()->numbered(cellrow, numbered);
 892                         cell = &grid.cell(grid.index(cellrow, cellcol));
 893                 }
 894
 895 #if 0
 896                 else if (t.cs() == "multicolumn") {
 897                         // extract column count and insert dummy cells
 898                         MathArray count;
 899                         parse(count, FLAG_ITEM, mode);
 900                         int cols = 1;
 901                         if (!extractNumber(count, cols)) {
 902                                 lyxerr << " can't extract number of cells from " << count << "\n";
 903                         }
 904                         // resize the table if necessary
 905                         for (int i = 0; i < cols; ++i) {
 906                                 ++cellcol;
 907                                 if (cellcol == grid.ncols()) {
 908                                         //lyxerr << "adding column " << cellcol << "\n";
 909                                         grid.addCol(cellcol - 1);
 910                                 }
 911                                 cell = &grid.cell(grid.index(cellrow, cellcol));
 912                                 // mark this as dummy
 913                                 grid.cellinfo(grid.index(cellrow, cellcol)).dummy_ = true;
 914                         }
 915                         // the last cell is the real thng, not a dummy
 916                         grid.cellinfo(grid.index(cellrow, cellcol)).dummy_ = false;
 917
 918                         // read special alignment
 919                         MathArray align;
 920                         parse(align, FLAG_ITEM, mode);
 921                         //grid.cellinfo(grid.index(cellrow, cellcol)).align_ = extractString(align);
 922
 923                         // parse the remaining contents into the "real" cell
 924                         parse(*cell, FLAG_ITEM, mode);
 925                 }
 926 #endif
 927
 928                 else if (t.cs() == "limits")
 929                         limits = 1;
 930
 931                 else if (t.cs() == "nolimits")
 932                         limits = -1;
 933
 934                 else if (t.cs() == "nonumber") {
 935                         if (grid.asHullInset())
 936                                 grid.asHullInset()->numbered(cellrow, false);
 937                 }
 938
 939                 else if (t.cs() == "number") {
 940                         if (grid.asHullInset())
 941                                 grid.asHullInset()->numbered(cellrow, true);
 942                 }
 943
 944                 else if (t.cs() == "hline") {
 945                         grid.rowinfo(cellrow).lines_ ++;
 946                 }
 947
 948                 else if (t.cs() == "sqrt") {
 949                         MathArray ar;
 950                         parse(ar, FLAG_OPTION, mode);
 951                         if (ar.size()) {
 952                                 cell->push_back(MathAtom(new MathRootInset));
 953                                 cell->back().nucleus()->cell(0) = ar;
 954                                 parse(cell->back().nucleus()->cell(1), FLAG_ITEM, mode);
 955                         } else {
 956                                 cell->push_back(MathAtom(new MathSqrtInset));
 957                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
 958                         }
 959                 }
 960
 961                 else if (t.cs() == "xrightarrow" || t.cs() == "xleftarrow") {
 962                         cell->push_back(createMathInset(t.cs()));
 963                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
 964                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
 965                 }
 966
 967                 else if (t.cs() == "ref" || t.cs() == "prettyref" ||
 968                                 t.cs() == "pageref" || t.cs() == "vpageref" || t.cs() == "vref") {
 969                         cell->push_back(MathAtom(new RefInset(t.cs())));
 970                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
 971                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
 972                 }
 973
 974                 else if (t.cs() == "left") {
 975                         string l = getToken().asString();
 976                         MathArray ar;
 977                         parse(ar, FLAG_RIGHT, mode);
 978                         string r = getToken().asString();
 979                         cell->push_back(MathAtom(new MathDelimInset(l, r, ar)));
 980                 }
 981
 982                 else if (t.cs() == "right") {
 983                         if (flags & FLAG_RIGHT)
 984                                 return;
 985                         //lyxerr << "got so far: '" << cell << "'\n";
 986                         error("Unmatched right delimiter");
 987                         return;
 988                 }
 989
 990                 else if (t.cs() == "begin") {
 991                         string const name = getArg('{', '}');
 992                         skipSpaces();
 993
 994                         if (name == "array" || name == "subarray") {
 995                                 string const valign = parse_verbatim_option() + 'c';
 996                                 string const halign = parse_verbatim_item();
 997                                 cell->push_back(MathAtom(new MathArrayInset(name, valign[0], halign)));
 998                                 parse2(cell->back(), FLAG_END, mode, false);
 999                         }
1000
1001                         if (name == "tabular") {
1002                                 string const valign = parse_verbatim_option() + 'c';
1003                                 string const halign = parse_verbatim_item();
1004                                 cell->push_back(MathAtom(new MathTabularInset(name, valign[0], halign)));
1005                                 parse2(cell->back(), FLAG_END, MathInset::TEXT_MODE, false);
1006                         }
1007
1008                         else if (name == "split" || name == "cases" ||
1009                                          name == "gathered" || name == "aligned") {
1010                                 cell->push_back(createMathInset(name));
1011                                 parse2(cell->back(), FLAG_END, mode, false);
1012                         }
1013
1014                         else if (name == "math") {
1015                                 cell->push_back(MathAtom(new MathHullInset("simple")));
1016                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, true);
1017                         }
1018
1019                         else if (name == "equation" || name == "equation*"
1020                                         || name == "displaymath") {
1021                                 cell->push_back(MathAtom(new MathHullInset("equation")));
1022                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, (name == "equation"));
1023                         }
1024
1025                         else if (name == "eqnarray" || name == "eqnarray*") {
1026                                 cell->push_back(MathAtom(new MathHullInset("eqnarray")));
1027                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1028                         }
1029
1030                         else if (name == "align" || name == "align*") {
1031                                 cell->push_back(MathAtom(new MathHullInset("align")));
1032                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1033                         }
1034
1035                         else if (name == "flalign" || name == "flalign*") {
1036                                 cell->push_back(MathAtom(new MathHullInset("flalign")));
1037                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1038                         }
1039
1040                         else if (name == "alignat" || name == "alignat*") {
1041                                 // ignore this for a while
1042                                 getArg('{', '}');
1043                                 cell->push_back(MathAtom(new MathHullInset("alignat")));
1044                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1045                         }
1046
1047                         else if (name == "xalignat" || name == "xalignat*") {
1048                                 // ignore this for a while
1049                                 getArg('{', '}');
1050                                 cell->push_back(MathAtom(new MathHullInset("xalignat")));
1051                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1052                         }
1053
1054                         else if (name == "xxalignat") {
1055                                 // ignore this for a while
1056                                 getArg('{', '}');
1057                                 cell->push_back(MathAtom(new MathHullInset("xxalignat")));
1058                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1059                         }
1060
1061                         else if (name == "multline" || name == "multline*") {
1062                                 cell->push_back(MathAtom(new MathHullInset("multline")));
1063                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1064                         }
1065
1066                         else if (name == "gather" || name == "gather*") {
1067                                 cell->push_back(MathAtom(new MathHullInset("gather")));
1068                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1069                         }
1070
1071                         else if (latexkeys const * l = in_word_set(name)) {
1072                                 if (l->inset == "matrix") {
1073                                         cell->push_back(createMathInset(name));
1074                                         parse2(cell->back(), FLAG_END, mode, false);
1075                                 }
1076                         }
1077
1078                         else {
1079                                 // lyxerr << "unknow math inset begin '" << name << "'\n";
1080                                 // create generic environment inset
1081                                 cell->push_back(MathAtom(new MathEnvInset(name)));
1082                                 parse(cell->back().nucleus()->cell(0), FLAG_END, mode);
1083                         }
1084                 }
1085
1086                 else if (t.cs() == "kern") {
1087 #ifdef WITH_WARNINGS
1088 #warning A hack...
1089 #endif
1090                         string s;
1091                         while (true) {
1092                                 Token const & t = getToken();
1093                                 if (!good()) {
1094                                         putback();
1095                                         break;
1096                                 }
1097                                 s += t.character();
1098                                 if (isValidLength(s))
1099                                         break;
1100                         }
1101                         cell->push_back(MathAtom(new MathKernInset(s)));
1102                 }
1103
1104                 else if (t.cs() == "label") {
1105                         string label = parse_verbatim_item();
1106                         if (grid.asHullInset()) {
1107                                 grid.asHullInset()->label(cellrow, label);
1108                         } else {
1109                                 cell->push_back(createMathInset(t.cs()));
1110                                 cell->push_back(MathAtom(new MathBraceInset(asArray(label))));
1111                         }
1112                 }
1113
1114                 else if (t.cs() == "choose" || t.cs() == "over" || t.cs() == "atop") {
1115                         MathAtom at = createMathInset(t.cs());
1116                         at.nucleus()->cell(0) = *cell;
1117                         cell->clear();
1118                         parse(at.nucleus()->cell(1), flags, mode);
1119                         cell->push_back(at);
1120                         return;
1121                 }
1122
1123                 else if (t.cs() == "substack") {
1124                         cell->push_back(createMathInset(t.cs()));
1125                         parse2(cell->back(), FLAG_ITEM, mode, false);
1126                 }
1127
1128                 else if (t.cs() == "xymatrix") {
1129                         cell->push_back(createMathInset(t.cs()));
1130                         parse2(cell->back(), FLAG_ITEM, mode, false);
1131                 }
1132
1133                 else if (t.cs() == "framebox") {
1134                         cell->push_back(createMathInset(t.cs()));
1135                         parse(cell->back().nucleus()->cell(0), FLAG_OPTION, MathInset::TEXT_MODE);
1136                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, MathInset::TEXT_MODE);
1137                         parse(cell->back().nucleus()->cell(2), FLAG_ITEM, MathInset::TEXT_MODE);
1138                 }
1139
1140 #if 0
1141                 else if (t.cs() == "infer") {
1142                         MathArray ar;
1143                         parse(ar, FLAG_OPTION, mode);
1144                         cell->push_back(createMathInset(t.cs()));
1145                         parse2(cell->back(), FLAG_ITEM, mode, false);
1146                 }
1147
1148                 // Disabled
1149                 else if (1 && t.cs() == "ar") {
1150                         MathXYArrowInset * p = new MathXYArrowInset;
1151                         // try to read target
1152                         parse(p->cell(0), FLAG_OTPTION, mode);
1153                         // try to read label
1154                         if (nextToken().cat() == catSuper || nextToken().cat() == catSub) {
1155                                 p->up_ = nextToken().cat() == catSuper;
1156                                 getToken();
1157                                 parse(p->cell(1), FLAG_ITEM, mode);
1158                                 //lyxerr << "read label: " << p->cell(1) << "\n";
1159                         }
1160
1161                         cell->push_back(MathAtom(p));
1162                         //lyxerr << "read cell: " << cell << "\n";
1163                 }
1164 #endif
1165
1166                 else if (t.cs().size()) {
1167                         latexkeys const * l = in_word_set(t.cs());
1168                         if (l) {
1169                                 if (l->inset == "font") {
1170                                         cell->push_back(createMathInset(t.cs()));
1171                                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, asMode(mode, l->extra));
1172                                 }
1173
1174                                 else if (l->inset == "oldfont") {
1175                                         cell->push_back(createMathInset(t.cs()));
1176                                         parse(cell->back().nucleus()->cell(0), flags, asMode(mode, l->extra));
1177                                         return;
1178                                 }
1179
1180                                 else if (l->inset == "style") {
1181                                         cell->push_back(createMathInset(t.cs()));
1182                                         parse(cell->back().nucleus()->cell(0), flags, mode);
1183                                         return;
1184                                 }
1185
1186                                 else if (l->inset == "parbox") {
1187                                         // read optional positioning and width
1188                                         string pos   = parse_verbatim_option();
1189                                         string width = parse_verbatim_item();
1190                                         cell->push_back(createMathInset(t.cs()));
1191                                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, MathInset::TEXT_MODE);
1192                                         cell->back().nucleus()->asParboxInset()->setPosition(pos);
1193                                         cell->back().nucleus()->asParboxInset()->setWidth(width);
1194                                 }
1195
1196                                 else {
1197                                         MathAtom at = createMathInset(t.cs());
1198                                         for (MathInset::idx_type i = 0; i < at->nargs(); ++i)
1199                                                 parse(at.nucleus()->cell(i), FLAG_ITEM, asMode(mode, l->extra));
1200                                         cell->push_back(at);
1201                                 }
1202                         }
1203
1204                         else {
1205                                 MathAtom at = createMathInset(t.cs());
1206                                 MathInset::mode_type m = mode;
1207                                 if (m == MathInset::UNDECIDED_MODE)
1208                                         m = at->currentMode();
1209                                 MathInset::idx_type start = 0;
1210                                 // this fails on \bigg[...\bigg]
1211                                 //MathArray opt;
1212                                 //parse(opt, FLAG_OPTION, MathInset::VERBATIM_MODE);
1213                                 //if (opt.size()) {
1214                                 //      start = 1;
1215                                 //      at.nucleus()->cell(0) = opt;
1216                                 //}
1217                                 for (MathInset::idx_type i = start; i < at->nargs(); ++i)
1218                                         parse(at.nucleus()->cell(i), FLAG_ITEM, m);
1219                                 cell->push_back(at);
1220                         }
1221                 }
1222
1223
1224                 if (flags & FLAG_LEAVE) {
1225                         flags &= ~FLAG_LEAVE;
1226                         break;
1227                 }
1228         }
1229 }
1230
1231
1232
1233 } // anonymous namespace
1234
1235
1236 void mathed_parse_cell(MathArray & ar, string const & str)
1237 {
1238         istringstream is(str.c_str());
1239         mathed_parse_cell(ar, is);
1240 }
1241
1242
1243 void mathed_parse_cell(MathArray & ar, istream & is)
1244 {
1245         Parser(is).parse(ar, 0, MathInset::MATH_MODE);
1246 }
1247
1248
1249 bool mathed_parse_normal(MathAtom & t, string const & str)
1250 {
1251         istringstream is(str.c_str());
1252         return Parser(is).parse(t);
1253 }
1254
1255
1256 bool mathed_parse_normal(MathAtom & t, istream & is)
1257 {
1258         return Parser(is).parse(t);
1259 }
1260
1261
1262 bool mathed_parse_normal(MathAtom & t, LyXLex & lex)
1263 {
1264         return Parser(lex).parse(t);
1265 }
1266
1267
1268 void mathed_parse_normal(MathGridInset & grid, string const & str)
1269 {
1270         istringstream is(str.c_str());
1271         Parser(is).parse1(grid, 0, MathInset::MATH_MODE, false);
1272 }