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