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_colorinset.h"
  46 #include "math_commentinset.h"
  47 #include "math_deliminset.h"
  48 #include "math_envinset.h"
  49 #include "math_factory.h"
  50 #include "math_kerninset.h"
  51 #include "math_macro.h"
  52 #include "math_macroarg.h"
  53 #include "math_macrotemplate.h"
  54 #include "math_parinset.h"
  55 #include "math_rootinset.h"
  56 #include "math_scriptinset.h"
  57 #include "math_sqrtinset.h"
  58 #include "math_support.h"
  59 #include "math_tabularinset.h"
  60
  61 //#include "insets/insetref.h"
  62 #include "ref_inset.h"
  63
  64 #include "lyxlex.h"
  65 #include "debug.h"
  66
  67 #include "support/convert.h"
  68
  69 #include <sstream>
  70
  71 using std::endl;
  72 using std::fill;
  73
  74 using std::string;
  75 using std::ios;
  76 using std::istream;
  77 using std::istringstream;
  78 using std::ostream;
  79 using std::vector;
  80
  81
  82 //#define FILEDEBUG
  83
  84
  85 namespace {
  86
  87 MathInset::mode_type asMode(MathInset::mode_type oldmode, string const & str)
  88 {
  89         //lyxerr << "handling mode: '" << str << "'" << endl;
  90         if (str == "mathmode")
  91                 return MathInset::MATH_MODE;
  92         if (str == "textmode" || str == "forcetext")
  93                 return MathInset::TEXT_MODE;
  94         return oldmode;
  95 }
  96
  97
  98 bool stared(string const & s)
  99 {
 100         string::size_type const n = s.size();
 101         return n && s[n - 1] == '*';
 102 }
 103
 104
 105 /*!
 106  * Add the row \p cellrow to \p grid.
 107  * \returns wether the row could be added. Adding a row can fail for
 108  * environments like "equation" that have a fixed number of rows.
 109  */
 110 bool addRow(MathGridInset & grid, MathGridInset::row_type & cellrow,
 111             string const & vskip)
 112 {
 113         ++cellrow;
 114         if (cellrow == grid.nrows()) {
 115                 //lyxerr << "adding row " << cellrow << endl;
 116                 grid.addRow(cellrow - 1);
 117                 if (cellrow == grid.nrows()) {
 118                         // We can't add a row to this grid, so let's
 119                         // append the content of this cell to the previous
 120                         // one.
 121                         // This does not happen in well formed .lyx files,
 122                         // but LyX versions 1.3.x and older could create
 123                         // such files and tex2lyx can still do that.
 124                         --cellrow;
 125                         lyxerr << "ignoring extra row";
 126                         if (!vskip.empty())
 127                                 lyxerr << " with extra space " << vskip;
 128                         lyxerr << '.' << endl;
 129                         return false;
 130                 }
 131         }
 132         grid.vcrskip(LyXLength(vskip), cellrow - 1);
 133         return true;
 134 }
 135
 136
 137 /*!
 138  * Add the column \p cellcol to \p grid.
 139  * \returns wether the column could be added. Adding a column can fail for
 140  * environments like "eqnarray" that have a fixed number of columns.
 141  */
 142 bool addCol(MathGridInset & grid, MathGridInset::col_type & cellcol)
 143 {
 144         ++cellcol;
 145         if (cellcol == grid.ncols()) {
 146                 //lyxerr << "adding column " << cellcol << endl;
 147                 grid.addCol(cellcol - 1);
 148                 if (cellcol == grid.ncols()) {
 149                         // We can't add a column to this grid, so let's
 150                         // append the content of this cell to the previous
 151                         // one.
 152                         // This does not happen in well formed .lyx files,
 153                         // but LyX versions 1.3.x and older could create
 154                         // such files and tex2lyx can still do that.
 155                         --cellcol;
 156                         lyxerr << "ignoring extra column." << endl;
 157                         return false;
 158                 }
 159         }
 160         return true;
 161 }
 162
 163
 164 /*!
 165  * Check wether the last row is empty and remove it if yes.
 166  * Otherwise the following code
 167  * \verbatim
 168 \begin{array}{|c|c|}
 169 \hline
 170 1 & 2 \\ \hline
 171 3 & 4 \\ \hline
 172 \end{array}
 173  * \endverbatim
 174  * will result in a grid with 3 rows (+ the dummy row that is always present),
 175  * because the last '\\' opens a new row.
 176  */
 177 void delEmptyLastRow(MathGridInset & grid)
 178 {
 179         MathGridInset::row_type const row = grid.nrows() - 1;
 180         for (MathGridInset::col_type col = 0; col < grid.ncols(); ++col) {
 181                 if (!grid.cell(grid.index(row, col)).empty())
 182                         return;
 183         }
 184         // Remove the dummy row, so that the previous last row (that would
 185         // contain the last hline in the example above) becomes the dummy row.
 186         grid.delRow(row + 1);
 187 }
 188
 189
 190 // These are TeX's catcodes
 191 enum CatCode {
 192         catEscape,     // 0    backslash
 193         catBegin,      // 1    {
 194         catEnd,        // 2    }
 195         catMath,       // 3    $
 196         catAlign,      // 4    &
 197         catNewline,    // 5    ^^M
 198         catParameter,  // 6    #
 199         catSuper,      // 7    ^
 200         catSub,        // 8    _
 201         catIgnore,     // 9
 202         catSpace,      // 10   space
 203         catLetter,     // 11   a-zA-Z
 204         catOther,      // 12   none of the above
 205         catActive,     // 13   ~
 206         catComment,    // 14   %
 207         catInvalid     // 15   <delete>
 208 };
 209
 210 CatCode theCatcode[256];
 211
 212
 213 inline CatCode catcode(unsigned char c)
 214 {
 215         return theCatcode[c];
 216 }
 217
 218
 219 enum {
 220         FLAG_ALIGN      = 1 << 0,  //  next & or \\ ends the parsing process
 221         FLAG_BRACE_LAST = 1 << 1,  //  next closing brace ends the parsing
 222         FLAG_RIGHT      = 1 << 2,  //  next \\right ends the parsing process
 223         FLAG_END        = 1 << 3,  //  next \\end ends the parsing process
 224         FLAG_BRACK_LAST = 1 << 4,  //  next closing bracket ends the parsing
 225         FLAG_TEXTMODE   = 1 << 5,  //  we are in a box
 226         FLAG_ITEM       = 1 << 6,  //  read a (possibly braced token)
 227         FLAG_LEAVE      = 1 << 7,  //  leave the loop at the end
 228         FLAG_SIMPLE     = 1 << 8,  //  next $ leaves the loop
 229         FLAG_EQUATION   = 1 << 9,  //  next \] leaves the loop
 230         FLAG_SIMPLE2    = 1 << 10, //  next \) leaves the loop
 231         FLAG_OPTION     = 1 << 11, //  read [...] style option
 232         FLAG_BRACED     = 1 << 12  //  read {...} style argument
 233 };
 234
 235
 236 //
 237 // Helper class for parsing
 238 //
 239
 240 class Token {
 241 public:
 242         ///
 243         Token() : cs_(), char_(0), cat_(catIgnore) {}
 244         ///
 245         Token(char c, CatCode cat) : cs_(), char_(c), cat_(cat) {}
 246         ///
 247         Token(string const & cs) : cs_(cs), char_(0), cat_(catIgnore) {}
 248
 249         ///
 250         string const & cs() const { return cs_; }
 251         ///
 252         CatCode cat() const { return cat_; }
 253         ///
 254         char character() const { return char_; }
 255         ///
 256         string asString() const { return cs_.size() ? cs_ : string(1, char_); }
 257
 258 private:
 259         ///
 260         string cs_;
 261         ///
 262         char char_;
 263         ///
 264         CatCode cat_;
 265 };
 266
 267 ostream & operator<<(ostream & os, Token const & t)
 268 {
 269         if (t.cs().size())
 270                 os << '\\' << t.cs();
 271         else if (t.cat() == catLetter)
 272                 os << t.character();
 273         else
 274                 os << '[' << t.character() << ',' << t.cat() << ']';
 275         return os;
 276 }
 277
 278
 279 class Parser {
 280 public:
 281         ///
 282         typedef  MathInset::mode_type mode_type;
 283
 284         ///
 285         Parser(LyXLex & lex);
 286         ///
 287         Parser(istream & is);
 288
 289         ///
 290         bool parse(MathAtom & at);
 291         ///
 292         void parse(MathArray & array, unsigned flags, mode_type mode);
 293         ///
 294         void parse1(MathGridInset & grid, unsigned flags, mode_type mode,
 295                 bool numbered);
 296         ///
 297         MathArray parse(unsigned flags, mode_type mode);
 298         ///
 299         int lineno() const { return lineno_; }
 300         ///
 301         void putback();
 302
 303 private:
 304         ///
 305         void parse2(MathAtom & at, unsigned flags, mode_type mode, bool numbered);
 306         /// get arg delimited by 'left' and 'right'
 307         string getArg(char left, char right);
 308         ///
 309         char getChar();
 310         ///
 311         void error(string const & msg);
 312         /// dump contents to screen
 313         void dump() const;
 314         ///
 315         void tokenize(istream & is);
 316         ///
 317         void tokenize(string const & s);
 318         ///
 319         void skipSpaceTokens(istream & is, char c);
 320         ///
 321         void push_back(Token const & t);
 322         ///
 323         void pop_back();
 324         ///
 325         Token const & prevToken() const;
 326         ///
 327         Token const & nextToken() const;
 328         ///
 329         Token const & getToken();
 330         /// skips spaces if any
 331         void skipSpaces();
 332         ///
 333         void lex(string const & s);
 334         ///
 335         bool good() const;
 336         ///
 337         string parse_verbatim_item();
 338         ///
 339         string parse_verbatim_option();
 340
 341         ///
 342         int lineno_;
 343         ///
 344         vector<Token> tokens_;
 345         ///
 346         unsigned pos_;
 347         /// Stack of active environments
 348         vector<string> environments_;
 349 };
 350
 351
 352 Parser::Parser(LyXLex & lexer)
 353         : lineno_(lexer.getLineNo()), pos_(0)
 354 {
 355         tokenize(lexer.getStream());
 356         lexer.eatLine();
 357 }
 358
 359
 360 Parser::Parser(istream & is)
 361         : lineno_(0), pos_(0)
 362 {
 363         tokenize(is);
 364 }
 365
 366
 367 void Parser::push_back(Token const & t)
 368 {
 369         tokens_.push_back(t);
 370 }
 371
 372
 373 void Parser::pop_back()
 374 {
 375         tokens_.pop_back();
 376 }
 377
 378
 379 Token const & Parser::prevToken() const
 380 {
 381         static const Token dummy;
 382         return pos_ > 0 ? tokens_[pos_ - 1] : dummy;
 383 }
 384
 385
 386 Token const & Parser::nextToken() const
 387 {
 388         static const Token dummy;
 389         return good() ? tokens_[pos_] : dummy;
 390 }
 391
 392
 393 Token const & Parser::getToken()
 394 {
 395         static const Token dummy;
 396         //lyxerr << "looking at token " << tokens_[pos_] << " pos: " << pos_ << endl;
 397         return good() ? tokens_[pos_++] : dummy;
 398 }
 399
 400
 401 void Parser::skipSpaces()
 402 {
 403         while (nextToken().cat() == catSpace || nextToken().cat() == catNewline)
 404                 getToken();
 405 }
 406
 407
 408 void Parser::putback()
 409 {
 410         --pos_;
 411 }
 412
 413
 414 bool Parser::good() const
 415 {
 416         return pos_ < tokens_.size();
 417 }
 418
 419
 420 char Parser::getChar()
 421 {
 422         if (!good())
 423                 error("The input stream is not well...");
 424         return tokens_[pos_++].character();
 425 }
 426
 427
 428 string Parser::getArg(char left, char right)
 429 {
 430         skipSpaces();
 431
 432         string result;
 433         char c = getChar();
 434
 435         if (c != left)
 436                 putback();
 437         else
 438                 while ((c = getChar()) != right && good())
 439                         result += c;
 440
 441         return result;
 442 }
 443
 444
 445 void Parser::skipSpaceTokens(istream & is, char c)
 446 {
 447         // skip trailing spaces
 448         while (catcode(c) == catSpace || catcode(c) == catNewline)
 449                 if (!is.get(c))
 450                         break;
 451         //lyxerr << "putting back: " << c << endl;
 452         is.putback(c);
 453 }
 454
 455
 456 void Parser::tokenize(istream & is)
 457 {
 458         // eat everything up to the next \end_inset or end of stream
 459         // and store it in s for further tokenization
 460         string s;
 461         char c;
 462         while (is.get(c)) {
 463                 s += c;
 464                 if (s.size() >= 10 && s.substr(s.size() - 10) == "\\end_inset") {
 465                         s = s.substr(0, s.size() - 10);
 466                         break;
 467                 }
 468         }
 469         // Remove the space after \end_inset
 470         if (is.get(c) && c != ' ')
 471                 is.unget();
 472
 473         // tokenize buffer
 474         tokenize(s);
 475 }
 476
 477
 478 void Parser::tokenize(string const & buffer)
 479 {
 480         istringstream is(buffer, ios::in | ios::binary);
 481
 482         char c;
 483         while (is.get(c)) {
 484                 //lyxerr << "reading c: " << c << endl;
 485
 486                 switch (catcode(c)) {
 487                         case catNewline: {
 488                                 ++lineno_;
 489                                 is.get(c);
 490                                 if (catcode(c) == catNewline)
 491                                         ; //push_back(Token("par"));
 492                                 else {
 493                                         push_back(Token('\n', catNewline));
 494                                         is.putback(c);
 495                                 }
 496                                 break;
 497                         }
 498
 499 /*
 500                         case catComment: {
 501                                 while (is.get(c) && catcode(c) != catNewline)
 502                                         ;
 503                                 ++lineno_;
 504                                 break;
 505                         }
 506 */
 507
 508                         case catEscape: {
 509                                 is.get(c);
 510                                 if (!is) {
 511                                         error("unexpected end of input");
 512                                 } else {
 513                                         string s(1, c);
 514                                         if (catcode(c) == catLetter) {
 515                                                 // collect letters
 516                                                 while (is.get(c) && catcode(c) == catLetter)
 517                                                         s += c;
 518                                                 skipSpaceTokens(is, c);
 519                                         }
 520                                         push_back(Token(s));
 521                                 }
 522                                 break;
 523                         }
 524
 525                         case catSuper:
 526                         case catSub: {
 527                                 push_back(Token(c, catcode(c)));
 528                                 is.get(c);
 529                                 skipSpaceTokens(is, c);
 530                                 break;
 531                         }
 532
 533                         case catIgnore: {
 534                                 lyxerr << "ignoring a char: " << int(c) << endl;
 535                                 break;
 536                         }
 537
 538                         default:
 539                                 push_back(Token(c, catcode(c)));
 540                 }
 541         }
 542
 543 #ifdef FILEDEBUG
 544         dump();
 545 #endif
 546 }
 547
 548
 549 void Parser::dump() const
 550 {
 551         lyxerr << "\nTokens: ";
 552         for (unsigned i = 0; i < tokens_.size(); ++i) {
 553                 if (i == pos_)
 554                         lyxerr << " <#> ";
 555                 lyxerr << tokens_[i];
 556         }
 557         lyxerr << " pos: " << pos_ << endl;
 558 }
 559
 560
 561 void Parser::error(string const & msg)
 562 {
 563         lyxerr << "Line ~" << lineno_ << ": Math parse error: " << msg << endl;
 564         dump();
 565         //exit(1);
 566 }
 567
 568
 569 bool Parser::parse(MathAtom & at)
 570 {
 571         skipSpaces();
 572         MathArray ar;
 573         parse(ar, false, MathInset::UNDECIDED_MODE);
 574         if (ar.size() != 1 || ar.front()->getType() == "none") {
 575                 lyxerr << "unusual contents found: " << ar << endl;
 576                 at = MathAtom(new MathParInset(ar));
 577                 //if (at->nargs() > 0)
 578                 //      at.nucleus()->cell(0) = ar;
 579                 //else
 580                 //      lyxerr << "unusual contents found: " << ar << endl;
 581                 return true;
 582         }
 583         at = ar[0];
 584         return true;
 585 }
 586
 587
 588 string Parser::parse_verbatim_option()
 589 {
 590         skipSpaces();
 591         string res;
 592         if (nextToken().character() == '[') {
 593                 Token t = getToken();
 594                 for (Token t = getToken(); t.character() != ']' && good(); t = getToken()) {
 595                         if (t.cat() == catBegin) {
 596                                 putback();
 597                                 res += '{' + parse_verbatim_item() + '}';
 598                         } else
 599                                 res += t.asString();
 600                 }
 601         }
 602         return res;
 603 }
 604
 605
 606 string Parser::parse_verbatim_item()
 607 {
 608         skipSpaces();
 609         string res;
 610         if (nextToken().cat() == catBegin) {
 611                 Token t = getToken();
 612                 for (Token t = getToken(); t.cat() != catEnd && good(); t = getToken()) {
 613                         if (t.cat() == catBegin) {
 614                                 putback();
 615                                 res += '{' + parse_verbatim_item() + '}';
 616                         }
 617                         else
 618                                 res += t.asString();
 619                 }
 620         }
 621         return res;
 622 }
 623
 624
 625 MathArray Parser::parse(unsigned flags, mode_type mode)
 626 {
 627         MathArray ar;
 628         parse(ar, flags, mode);
 629         return ar;
 630 }
 631
 632
 633 void Parser::parse(MathArray & array, unsigned flags, mode_type mode)
 634 {
 635         MathGridInset grid(1, 1);
 636         parse1(grid, flags, mode, false);
 637         array = grid.cell(0);
 638 }
 639
 640
 641 void Parser::parse2(MathAtom & at, const unsigned flags, const mode_type mode,
 642         const bool numbered)
 643 {
 644         parse1(*(at.nucleus()->asGridInset()), flags, mode, numbered);
 645 }
 646
 647
 648 void Parser::parse1(MathGridInset & grid, unsigned flags,
 649         const mode_type mode, const bool numbered)
 650 {
 651         int limits = 0;
 652         MathGridInset::row_type cellrow = 0;
 653         MathGridInset::col_type cellcol = 0;
 654         MathArray * cell = &grid.cell(grid.index(cellrow, cellcol));
 655
 656         if (grid.asHullInset())
 657                 grid.asHullInset()->numbered(cellrow, numbered);
 658
 659         //dump();
 660         //lyxerr << " flags: " << flags << endl;
 661         //lyxerr << " mode: " << mode  << endl;
 662         //lyxerr << "grid: " << grid << endl;
 663
 664         while (good()) {
 665                 Token const & t = getToken();
 666
 667 #ifdef FILEDEBUG
 668                 lyxerr << "t: " << t << " flags: " << flags << endl;
 669                 lyxerr << "mode: " << mode  << endl;
 670                 cell->dump();
 671                 lyxerr << endl;
 672 #endif
 673
 674                 if (flags & FLAG_ITEM) {
 675
 676                         if (t.cat() == catBegin) {
 677                                 // skip the brace and collect everything to the next matching
 678                                 // closing brace
 679                                 parse1(grid, FLAG_BRACE_LAST, mode, numbered);
 680                                 return;
 681                         }
 682
 683                         // handle only this single token, leave the loop if done
 684                         flags = FLAG_LEAVE;
 685                 }
 686
 687
 688                 if (flags & FLAG_BRACED) {
 689                         if (t.cat() == catSpace)
 690                                 continue;
 691
 692                         if (t.cat() != catBegin) {
 693                                 error("opening brace expected");
 694                                 return;
 695                         }
 696
 697                         // skip the brace and collect everything to the next matching
 698                         // closing brace
 699                         flags = FLAG_BRACE_LAST;
 700                 }
 701
 702
 703                 if (flags & FLAG_OPTION) {
 704                         if (t.cat() == catOther && t.character() == '[') {
 705                                 MathArray ar;
 706                                 parse(ar, FLAG_BRACK_LAST, mode);
 707                                 cell->append(ar);
 708                         } else {
 709                                 // no option found, put back token and we are done
 710                                 putback();
 711                         }
 712                         return;
 713                 }
 714
 715                 //
 716                 // cat codes
 717                 //
 718                 if (t.cat() == catMath) {
 719                         if (mode != MathInset::MATH_MODE) {
 720                                 // we are inside some text mode thingy, so opening new math is allowed
 721                                 Token const & n = getToken();
 722                                 if (n.cat() == catMath) {
 723                                         // TeX's $$...$$ syntax for displayed math
 724                                         cell->push_back(MathAtom(new MathHullInset("equation")));
 725                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 726                                         getToken(); // skip the second '$' token
 727                                 } else {
 728                                         // simple $...$  stuff
 729                                         putback();
 730                                         cell->push_back(MathAtom(new MathHullInset("simple")));
 731                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 732                                 }
 733                         }
 734
 735                         else if (flags & FLAG_SIMPLE) {
 736                                 // this is the end of the formula
 737                                 return;
 738                         }
 739
 740                         else {
 741                                 error("something strange in the parser");
 742                                 break;
 743                         }
 744                 }
 745
 746                 else if (t.cat() == catLetter)
 747                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 748
 749                 else if (t.cat() == catSpace && mode != MathInset::MATH_MODE) {
 750                         if (cell->empty() || cell->back()->getChar() != ' ')
 751                                 cell->push_back(MathAtom(new MathCharInset(t.character())));
 752                 }
 753
 754                 else if (t.cat() == catNewline && mode != MathInset::MATH_MODE) {
 755                         if (cell->empty() || cell->back()->getChar() != ' ')
 756                                 cell->push_back(MathAtom(new MathCharInset(' ')));
 757                 }
 758
 759                 else if (t.cat() == catParameter) {
 760                         Token const & n = getToken();
 761                         cell->push_back(MathAtom(new MathMacroArgument(n.character()-'0')));
 762                 }
 763
 764                 else if (t.cat() == catActive)
 765                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 766
 767                 else if (t.cat() == catBegin) {
 768                         MathArray ar;
 769                         parse(ar, FLAG_BRACE_LAST, mode);
 770                         // do not create a BraceInset if they were written by LyX
 771                         // this helps to keep the annoyance of  "a choose b"  to a minimum
 772                         if (ar.size() == 1 && ar[0]->extraBraces())
 773                                 cell->append(ar);
 774                         else
 775                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
 776                 }
 777
 778                 else if (t.cat() == catEnd) {
 779                         if (flags & FLAG_BRACE_LAST)
 780                                 return;
 781                         error("found '}' unexpectedly");
 782                         //BOOST_ASSERT(false);
 783                         //add(cell, '}', LM_TC_TEX);
 784                 }
 785
 786                 else if (t.cat() == catAlign) {
 787                         //lyxerr << " column now " << (cellcol + 1)
 788                         //       << " max: " << grid.ncols() << endl;
 789                         if (flags & FLAG_ALIGN)
 790                                 return;
 791                         if (addCol(grid, cellcol))
 792                                 cell = &grid.cell(grid.index(cellrow, cellcol));
 793                 }
 794
 795                 else if (t.cat() == catSuper || t.cat() == catSub) {
 796                         bool up = (t.cat() == catSuper);
 797                         // we need no new script inset if the last thing was a scriptinset,
 798                         // which has that script already not the same script already
 799                         if (!cell->size())
 800                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 801                         else if (cell->back()->asScriptInset() &&
 802                                         !cell->back()->asScriptInset()->has(up))
 803                                 cell->back().nucleus()->asScriptInset()->ensure(up);
 804                         else if (cell->back()->asScriptInset())
 805                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 806                         else
 807                                 cell->back() = MathAtom(new MathScriptInset(cell->back(), up));
 808                         MathScriptInset * p = cell->back().nucleus()->asScriptInset();
 809                         // special handling of {}-bases
 810                         // is this always correct?
 811                         // It appears that this is wrong (Dekel)
 812                         //if (p->nuc().size() == 1 && p->nuc().back()->asNestInset() &&
 813                         //    p->nuc().back()->extraBraces())
 814                         //      p->nuc() = p->nuc().back()->asNestInset()->cell(0);
 815                         parse(p->cell(p->idxOfScript(up)), FLAG_ITEM, mode);
 816                         if (limits) {
 817                                 p->limits(limits);
 818                                 limits = 0;
 819                         }
 820                 }
 821
 822                 else if (t.character() == ']' && (flags & FLAG_BRACK_LAST)) {
 823                         //lyxerr << "finished reading option" << endl;
 824                         return;
 825                 }
 826
 827                 else if (t.cat() == catOther)
 828                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 829
 830                 else if (t.cat() == catComment) {
 831                         string s;
 832                         while (good()) {
 833                                 Token const & t = getToken();
 834                                 if (t.cat() == catNewline)
 835                                         break;
 836                                 s += t.asString();
 837                         }
 838                         cell->push_back(MathAtom(new MathCommentInset(s)));
 839                         skipSpaces();
 840                 }
 841
 842                 //
 843                 // control sequences
 844                 //
 845
 846                 else if (t.cs() == "lyxlock") {
 847                         if (cell->size())
 848                                 cell->back().nucleus()->lock(true);
 849                 }
 850
 851                 else if (t.cs() == "def" ||
 852                         t.cs() == "newcommand" ||
 853                         t.cs() == "renewcommand")
 854                 {
 855                         string const type = t.cs();
 856                         string name;
 857                         int nargs = 0;
 858                         if (t.cs() == "def") {
 859                                 // get name
 860                                 name = getToken().cs();
 861
 862                                 // read parameter
 863                                 string pars;
 864                                 while (good() && nextToken().cat() != catBegin) {
 865                                         pars += getToken().cs();
 866                                         ++nargs;
 867                                 }
 868                                 nargs /= 2;
 869                                 //lyxerr << "read \\def parameter list '" << pars << "'" << endl;
 870
 871                         } else { // t.cs() == "newcommand" || t.cs() == "renewcommand"
 872
 873                                 if (getToken().cat() != catBegin) {
 874                                         error("'{' in \\newcommand expected (1) ");
 875                                         return;
 876                                 }
 877
 878                                 name = getToken().cs();
 879
 880                                 if (getToken().cat() != catEnd) {
 881                                         error("'}' in \\newcommand expected");
 882                                         return;
 883                                 }
 884
 885                                 string const arg  = getArg('[', ']');
 886                                 if (!arg.empty())
 887                                         nargs = convert<int>(arg);
 888
 889                         }
 890
 891                         MathArray ar1;
 892                         parse(ar1, FLAG_ITEM, MathInset::UNDECIDED_MODE);
 893
 894                         // we cannot handle recursive stuff at all
 895                         //MathArray test;
 896                         //test.push_back(createMathInset(name));
 897                         //if (ar1.contains(test)) {
 898                         //      error("we cannot handle recursive macros at all.");
 899                         //      return;
 900                         //}
 901
 902                         // is a version for display attached?
 903                         skipSpaces();
 904                         MathArray ar2;
 905                         if (nextToken().cat() == catBegin)
 906                                 parse(ar2, FLAG_ITEM, MathInset::MATH_MODE);
 907
 908                         cell->push_back(MathAtom(new MathMacroTemplate(name, nargs, type,
 909                                 ar1, ar2)));
 910                 }
 911
 912                 else if (t.cs() == "(") {
 913                         cell->push_back(MathAtom(new MathHullInset("simple")));
 914                         parse2(cell->back(), FLAG_SIMPLE2, MathInset::MATH_MODE, false);
 915                 }
 916
 917                 else if (t.cs() == "[") {
 918                         cell->push_back(MathAtom(new MathHullInset("equation")));
 919                         parse2(cell->back(), FLAG_EQUATION, MathInset::MATH_MODE, false);
 920                 }
 921
 922                 else if (t.cs() == "protect")
 923                         // ignore \\protect, will hopefully be re-added during output
 924                         ;
 925
 926                 else if (t.cs() == "end") {
 927                         if (flags & FLAG_END) {
 928                                 // eat environment name
 929                                 string const name = getArg('{', '}');
 930                                 if (environments_.empty())
 931                                         error("'found \\end{" + name +
 932                                               "}' without matching '\\begin{" +
 933                                               name + "}'");
 934                                 else if (name != environments_.back())
 935                                         error("'\\end{" + name +
 936                                               "}' does not match '\\begin{" +
 937                                               environments_.back() + "}'");
 938                                 else {
 939                                         environments_.pop_back();
 940                                         // Delete empty last row in matrix
 941                                         // like insets.
 942                                         // If you abuse MathGridInset for
 943                                         // non-matrix like structures you
 944                                         // probably need to refine this test.
 945                                         // Right now we only have to test for
 946                                         // single line hull insets.
 947                                         if (grid.nrows() > 1)
 948                                                 delEmptyLastRow(grid);
 949                                         return;
 950                                 }
 951                         } else
 952                                 error("found 'end' unexpectedly");
 953                 }
 954
 955                 else if (t.cs() == ")") {
 956                         if (flags & FLAG_SIMPLE2)
 957                                 return;
 958                         error("found '\\)' unexpectedly");
 959                 }
 960
 961                 else if (t.cs() == "]") {
 962                         if (flags & FLAG_EQUATION)
 963                                 return;
 964                         error("found '\\]' unexpectedly");
 965                 }
 966
 967                 else if (t.cs() == "\\") {
 968                         if (flags & FLAG_ALIGN)
 969                                 return;
 970                         if (addRow(grid, cellrow, getArg('[', ']'))) {
 971                                 cellcol = 0;
 972                                 if (grid.asHullInset())
 973                                         grid.asHullInset()->numbered(
 974                                                         cellrow, numbered);
 975                                 cell = &grid.cell(grid.index(cellrow,
 976                                                              cellcol));
 977                         }
 978                 }
 979
 980 #if 0
 981                 else if (t.cs() == "multicolumn") {
 982                         // extract column count and insert dummy cells
 983                         MathArray count;
 984                         parse(count, FLAG_ITEM, mode);
 985                         int cols = 1;
 986                         if (!extractNumber(count, cols)) {
 987                                 lyxerr << " can't extract number of cells from " << count << endl;
 988                         }
 989                         // resize the table if necessary
 990                         for (int i = 0; i < cols; ++i) {
 991                                 if (addCol(grid, cellcol)) {
 992                                         cell = &grid.cell(grid.index(
 993                                                         cellrow, cellcol));
 994                                         // mark this as dummy
 995                                         grid.cellinfo(grid.index(
 996                                                 cellrow, cellcol)).dummy_ = true;
 997                                 }
 998                         }
 999                         // the last cell is the real thing, not a dummy
1000                         grid.cellinfo(grid.index(cellrow, cellcol)).dummy_ = false;
1001
1002                         // read special alignment
1003                         MathArray align;
1004                         parse(align, FLAG_ITEM, mode);
1005                         //grid.cellinfo(grid.index(cellrow, cellcol)).align_ = extractString(align);
1006
1007                         // parse the remaining contents into the "real" cell
1008                         parse(*cell, FLAG_ITEM, mode);
1009                 }
1010 #endif
1011
1012                 else if (t.cs() == "limits")
1013                         limits = 1;
1014
1015                 else if (t.cs() == "nolimits")
1016                         limits = -1;
1017
1018                 else if (t.cs() == "nonumber") {
1019                         if (grid.asHullInset())
1020                                 grid.asHullInset()->numbered(cellrow, false);
1021                 }
1022
1023                 else if (t.cs() == "number") {
1024                         if (grid.asHullInset())
1025                                 grid.asHullInset()->numbered(cellrow, true);
1026                 }
1027
1028                 else if (t.cs() == "hline") {
1029                         grid.rowinfo(cellrow).lines_ ++;
1030                 }
1031
1032                 else if (t.cs() == "sqrt") {
1033                         MathArray ar;
1034                         parse(ar, FLAG_OPTION, mode);
1035                         if (ar.size()) {
1036                                 cell->push_back(MathAtom(new MathRootInset));
1037                                 cell->back().nucleus()->cell(0) = ar;
1038                                 parse(cell->back().nucleus()->cell(1), FLAG_ITEM, mode);
1039                         } else {
1040                                 cell->push_back(MathAtom(new MathSqrtInset));
1041                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1042                         }
1043                 }
1044
1045                 else if (t.cs() == "xrightarrow" || t.cs() == "xleftarrow") {
1046                         cell->push_back(createMathInset(t.cs()));
1047                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
1048                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1049                 }
1050
1051                 else if (t.cs() == "ref" || t.cs() == "prettyref" ||
1052                                 t.cs() == "pageref" || t.cs() == "vpageref" || t.cs() == "vref") {
1053                         cell->push_back(MathAtom(new RefInset(t.cs())));
1054                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
1055                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1056                 }
1057
1058                 else if (t.cs() == "left") {
1059                         skipSpaces();
1060                         Token const & tl = getToken();
1061                         // \| and \Vert are equivalent, and MathDelimInset
1062                         // can't handle \|
1063                         // FIXME: fix this in MathDelimInset itself!
1064                         string const l = tl.cs() == "|" ? "Vert" : tl.asString();
1065                         MathArray ar;
1066                         parse(ar, FLAG_RIGHT, mode);
1067                         skipSpaces();
1068                         Token const & tr = getToken();
1069                         string const r = tr.cs() == "|" ? "Vert" : tr.asString();
1070                         cell->push_back(MathAtom(new MathDelimInset(l, r, ar)));
1071                 }
1072
1073                 else if (t.cs() == "right") {
1074                         if (flags & FLAG_RIGHT)
1075                                 return;
1076                         //lyxerr << "got so far: '" << cell << "'" << endl;
1077                         error("Unmatched right delimiter");
1078                         return;
1079                 }
1080
1081                 else if (t.cs() == "begin") {
1082                         string const name = getArg('{', '}');
1083                         environments_.push_back(name);
1084
1085                         if (name == "array" || name == "subarray") {
1086                                 string const valign = parse_verbatim_option() + 'c';
1087                                 string const halign = parse_verbatim_item();
1088                                 cell->push_back(MathAtom(new MathArrayInset(name, valign[0], halign)));
1089                                 parse2(cell->back(), FLAG_END, mode, false);
1090                         }
1091
1092                         else if (name == "tabular") {
1093                                 string const valign = parse_verbatim_option() + 'c';
1094                                 string const halign = parse_verbatim_item();
1095                                 cell->push_back(MathAtom(new MathTabularInset(name, valign[0], halign)));
1096                                 parse2(cell->back(), FLAG_END, MathInset::TEXT_MODE, false);
1097                         }
1098
1099                         else if (name == "split" || name == "cases" ||
1100                                  name == "gathered" || name == "aligned") {
1101                                 cell->push_back(createMathInset(name));
1102                                 parse2(cell->back(), FLAG_END, mode, false);
1103                         }
1104
1105                         else if (name == "alignedat") {
1106                                 // ignore this for a while
1107                                 getArg('{', '}');
1108                                 cell->push_back(createMathInset(name));
1109                                 parse2(cell->back(), FLAG_END, mode, false);
1110                         }
1111
1112                         else if (name == "math") {
1113                                 cell->push_back(MathAtom(new MathHullInset("simple")));
1114                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, true);
1115                         }
1116
1117                         else if (name == "equation" || name == "equation*"
1118                                         || name == "displaymath") {
1119                                 cell->push_back(MathAtom(new MathHullInset("equation")));
1120                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, (name == "equation"));
1121                         }
1122
1123                         else if (name == "eqnarray" || name == "eqnarray*") {
1124                                 cell->push_back(MathAtom(new MathHullInset("eqnarray")));
1125                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1126                         }
1127
1128                         else if (name == "align" || name == "align*") {
1129                                 cell->push_back(MathAtom(new MathHullInset("align")));
1130                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1131                         }
1132
1133                         else if (name == "flalign" || name == "flalign*") {
1134                                 cell->push_back(MathAtom(new MathHullInset("flalign")));
1135                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1136                         }
1137
1138                         else if (name == "alignat" || name == "alignat*") {
1139                                 // ignore this for a while
1140                                 getArg('{', '}');
1141                                 cell->push_back(MathAtom(new MathHullInset("alignat")));
1142                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1143                         }
1144
1145                         else if (name == "xalignat" || name == "xalignat*") {
1146                                 // ignore this for a while
1147                                 getArg('{', '}');
1148                                 cell->push_back(MathAtom(new MathHullInset("xalignat")));
1149                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1150                         }
1151
1152                         else if (name == "xxalignat") {
1153                                 // ignore this for a while
1154                                 getArg('{', '}');
1155                                 cell->push_back(MathAtom(new MathHullInset("xxalignat")));
1156                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1157                         }
1158
1159                         else if (name == "multline" || name == "multline*") {
1160                                 cell->push_back(MathAtom(new MathHullInset("multline")));
1161                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1162                         }
1163
1164                         else if (name == "gather" || name == "gather*") {
1165                                 cell->push_back(MathAtom(new MathHullInset("gather")));
1166                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1167                         }
1168
1169                         else if (latexkeys const * l = in_word_set(name)) {
1170                                 if (l->inset == "matrix") {
1171                                         cell->push_back(createMathInset(name));
1172                                         parse2(cell->back(), FLAG_END, mode, false);
1173                                 }
1174                         }
1175
1176                         else {
1177                                 dump();
1178                                 lyxerr << "found unknown math environment '" << name << "'" << endl;
1179                                 // create generic environment inset
1180                                 cell->push_back(MathAtom(new MathEnvInset(name)));
1181                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1182                         }
1183                 }
1184
1185                 else if (t.cs() == "kern") {
1186 #ifdef WITH_WARNINGS
1187 #warning A hack...
1188 #endif
1189                         string s;
1190                         while (true) {
1191                                 Token const & t = getToken();
1192                                 if (!good()) {
1193                                         putback();
1194                                         break;
1195                                 }
1196                                 s += t.character();
1197                                 if (isValidLength(s))
1198                                         break;
1199                         }
1200                         cell->push_back(MathAtom(new MathKernInset(s)));
1201                 }
1202
1203                 else if (t.cs() == "label") {
1204                         // FIXME: This is swallowed in inline formulas
1205                         string label = parse_verbatim_item();
1206                         MathArray ar;
1207                         asArray(label, ar);
1208                         if (grid.asHullInset()) {
1209                                 grid.asHullInset()->label(cellrow, label);
1210                         } else {
1211                                 cell->push_back(createMathInset(t.cs()));
1212                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
1213                         }
1214                 }
1215
1216                 else if (t.cs() == "choose" || t.cs() == "over" || t.cs() == "atop") {
1217                         MathAtom at = createMathInset(t.cs());
1218                         at.nucleus()->cell(0) = *cell;
1219                         cell->clear();
1220                         parse(at.nucleus()->cell(1), flags, mode);
1221                         cell->push_back(at);
1222                         return;
1223                 }
1224
1225                 else if (t.cs() == "color") {
1226                         string const color = parse_verbatim_item();
1227                         cell->push_back(MathAtom(new MathColorInset(true, color)));
1228                         parse(cell->back().nucleus()->cell(0), flags, mode);
1229                         return;
1230                 }
1231
1232                 else if (t.cs() == "textcolor") {
1233                         string const color = parse_verbatim_item();
1234                         cell->push_back(MathAtom(new MathColorInset(false, color)));
1235                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, MathInset::TEXT_MODE);
1236                 }
1237
1238                 else if (t.cs() == "normalcolor") {
1239                         cell->push_back(createMathInset(t.cs()));
1240                         parse(cell->back().nucleus()->cell(0), flags, mode);
1241                         return;
1242                 }
1243
1244                 else if (t.cs() == "substack") {
1245                         cell->push_back(createMathInset(t.cs()));
1246                         parse2(cell->back(), FLAG_ITEM, mode, false);
1247                 }
1248
1249                 else if (t.cs() == "framebox" || t.cs() == "makebox") {
1250                         cell->push_back(createMathInset(t.cs()));
1251                         parse(cell->back().nucleus()->cell(0), FLAG_OPTION, MathInset::TEXT_MODE);
1252                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, MathInset::TEXT_MODE);
1253                         parse(cell->back().nucleus()->cell(2), FLAG_ITEM, MathInset::TEXT_MODE);
1254                 }
1255
1256 #if 0
1257                 else if (t.cs() == "infer") {
1258                         MathArray ar;
1259                         parse(ar, FLAG_OPTION, mode);
1260                         cell->push_back(createMathInset(t.cs()));
1261                         parse2(cell->back(), FLAG_ITEM, mode, false);
1262                 }
1263
1264                 // Disabled
1265                 else if (1 && t.cs() == "ar") {
1266                         auto_ptr<MathXYArrowInset> p(new MathXYArrowInset);
1267                         // try to read target
1268                         parse(p->cell(0), FLAG_OTPTION, mode);
1269                         // try to read label
1270                         if (nextToken().cat() == catSuper || nextToken().cat() == catSub) {
1271                                 p->up_ = nextToken().cat() == catSuper;
1272                                 getToken();
1273                                 parse(p->cell(1), FLAG_ITEM, mode);
1274                                 //lyxerr << "read label: " << p->cell(1) << endl;
1275                         }
1276
1277                         cell->push_back(MathAtom(p.release()));
1278                         //lyxerr << "read cell: " << cell << endl;
1279                 }
1280 #endif
1281
1282                 else if (t.cs().size()) {
1283                         latexkeys const * l = in_word_set(t.cs());
1284                         if (l) {
1285                                 if (l->inset == "font") {
1286                                         cell->push_back(createMathInset(t.cs()));
1287                                         parse(cell->back().nucleus()->cell(0),
1288                                                 FLAG_ITEM, asMode(mode, l->extra));
1289                                 }
1290
1291                                 else if (l->inset == "oldfont") {
1292                                         cell->push_back(createMathInset(t.cs()));
1293                                         parse(cell->back().nucleus()->cell(0),
1294                                                 flags | FLAG_ALIGN, asMode(mode, l->extra));
1295                                         if (prevToken().cat() != catAlign &&
1296                                             prevToken().cs() != "\\")
1297                                                 return;
1298                                         putback();
1299                                 }
1300
1301                                 else if (l->inset == "style") {
1302                                         cell->push_back(createMathInset(t.cs()));
1303                                         parse(cell->back().nucleus()->cell(0),
1304                                                 flags | FLAG_ALIGN, mode);
1305                                         if (prevToken().cat() != catAlign &&
1306                                             prevToken().cs() != "\\")
1307                                                 return;
1308                                         putback();
1309                                 }
1310
1311                                 else {
1312                                         MathAtom at = createMathInset(t.cs());
1313                                         for (MathInset::idx_type i = 0; i < at->nargs(); ++i)
1314                                                 parse(at.nucleus()->cell(i),
1315                                                         FLAG_ITEM, asMode(mode, l->extra));
1316                                         cell->push_back(at);
1317                                 }
1318                         }
1319
1320                         else {
1321                                 MathAtom at = createMathInset(t.cs());
1322                                 MathInset::mode_type m = mode;
1323                                 //if (m == MathInset::UNDECIDED_MODE)
1324                                 //lyxerr << "default creation: m1: " << m << endl;
1325                                 if (at->currentMode() != MathInset::UNDECIDED_MODE)
1326                                         m = at->currentMode();
1327                                 //lyxerr << "default creation: m2: " << m << endl;
1328                                 MathInset::idx_type start = 0;
1329                                 // this fails on \bigg[...\bigg]
1330                                 //MathArray opt;
1331                                 //parse(opt, FLAG_OPTION, MathInset::VERBATIM_MODE);
1332                                 //if (opt.size()) {
1333                                 //      start = 1;
1334                                 //      at.nucleus()->cell(0) = opt;
1335                                 //}
1336                                 for (MathInset::idx_type i = start; i < at->nargs(); ++i) {
1337                                         parse(at.nucleus()->cell(i), FLAG_ITEM, m);
1338                                         skipSpaces();
1339                                 }
1340                                 cell->push_back(at);
1341                         }
1342                 }
1343
1344
1345                 if (flags & FLAG_LEAVE) {
1346                         flags &= ~FLAG_LEAVE;
1347                         break;
1348                 }
1349         }
1350 }
1351
1352
1353
1354 } // anonymous namespace
1355
1356
1357 void mathed_parse_cell(MathArray & ar, string const & str)
1358 {
1359         istringstream is(str);
1360         mathed_parse_cell(ar, is);
1361 }
1362
1363
1364 void mathed_parse_cell(MathArray & ar, istream & is)
1365 {
1366         Parser(is).parse(ar, 0, MathInset::MATH_MODE);
1367 }
1368
1369
1370 bool mathed_parse_normal(MathAtom & t, string const & str)
1371 {
1372         istringstream is(str);
1373         return Parser(is).parse(t);
1374 }
1375
1376
1377 bool mathed_parse_normal(MathAtom & t, istream & is)
1378 {
1379         return Parser(is).parse(t);
1380 }
1381
1382
1383 bool mathed_parse_normal(MathAtom & t, LyXLex & lex)
1384 {
1385         return Parser(lex).parse(t);
1386 }
1387
1388
1389 void mathed_parse_normal(MathGridInset & grid, string const & str)
1390 {
1391         istringstream is(str);
1392         Parser(is).parse1(grid, 0, MathInset::MATH_MODE, false);
1393 }
1394
1395
1396 void initParser()
1397 {
1398         fill(theCatcode, theCatcode + 256, catOther);
1399         fill(theCatcode + 'a', theCatcode + 'z' + 1, catLetter);
1400         fill(theCatcode + 'A', theCatcode + 'Z' + 1, catLetter);
1401
1402         theCatcode[int('\\')] = catEscape;
1403         theCatcode[int('{')]  = catBegin;
1404         theCatcode[int('}')]  = catEnd;
1405         theCatcode[int('$')]  = catMath;
1406         theCatcode[int('&')]  = catAlign;
1407         theCatcode[int('\n')] = catNewline;
1408         theCatcode[int('#')]  = catParameter;
1409         theCatcode[int('^')]  = catSuper;
1410         theCatcode[int('_')]  = catSub;
1411         theCatcode[int(0x7f)] = catIgnore;
1412         theCatcode[int(' ')]  = catSpace;
1413         theCatcode[int('\t')] = catSpace;
1414         theCatcode[int('\r')] = catNewline;
1415         theCatcode[int('~')]  = catActive;
1416         theCatcode[int('%')]  = catComment;
1417 }