src/mathed/math_parser.C

   1 /**
   2  * \file math_parser.C
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author André Pönitz
   7  *
   8  * Full author contact details are available in file CREDITS.
   9  */
  10
  11 /*
  12
  13 If someone desperately needs partial "structures" (such as a few
  14 cells of an array inset or similar) (s)he could uses the
  15 following hack as starting point to write some macros:
  16
  17   \newif\ifcomment
  18   \commentfalse
  19   \ifcomment
  20           \def\makeamptab{\catcode`\&=4\relax}
  21           \def\makeampletter{\catcode`\&=11\relax}
  22     \def\b{\makeampletter\expandafter\makeamptab\bi}
  23     \long\def\bi#1\e{}
  24   \else
  25     \def\b{}\def\e{}
  26   \fi
  27
  28   ...
  29
  30   \[\begin{array}{ccc}
  31 1
  32 &
  33
  34   \end{array}\]
  35
  36 */
  37
  38
  39 #include <config.h>
  40
  41 #include "math_parser.h"
  42 #include "math_arrayinset.h"
  43 #include "math_braceinset.h"
  44 #include "math_charinset.h"
  45 #include "math_commentinset.h"
  46 #include "math_deliminset.h"
  47 #include "math_envinset.h"
  48 #include "math_factory.h"
  49 #include "math_kerninset.h"
  50 #include "math_macro.h"
  51 #include "math_macroarg.h"
  52 #include "math_macrotemplate.h"
  53 #include "math_parinset.h"
  54 #include "math_rootinset.h"
  55 #include "math_scriptinset.h"
  56 #include "math_sqrtinset.h"
  57 #include "math_support.h"
  58 #include "math_tabularinset.h"
  59
  60 //#include "insets/insetref.h"
  61 #include "ref_inset.h"
  62
  63 #include "lyxlex.h"
  64 #include "debug.h"
  65
  66 #include "support/convert.h"
  67
  68 #include <sstream>
  69
  70 using std::endl;
  71 using std::fill;
  72
  73 using std::string;
  74 using std::ios;
  75 using std::istream;
  76 using std::istringstream;
  77 using std::ostream;
  78 using std::vector;
  79
  80
  81 //#define FILEDEBUG
  82
  83
  84 namespace {
  85
  86 MathInset::mode_type asMode(MathInset::mode_type oldmode, string const & str)
  87 {
  88         //lyxerr << "handling mode: '" << str << "'" << endl;
  89         if (str == "mathmode")
  90                 return MathInset::MATH_MODE;
  91         if (str == "textmode" || str == "forcetext")
  92                 return MathInset::TEXT_MODE;
  93         return oldmode;
  94 }
  95
  96
  97 bool stared(string const & s)
  98 {
  99         string::size_type const n = s.size();
 100         return n && s[n - 1] == '*';
 101 }
 102
 103
 104 /*!
 105  * Add the row \p cellrow to \p grid.
 106  * \returns wether the row could be added. Adding a row can fail for
 107  * environments like "equation" that have a fixed number of rows.
 108  */
 109 bool addRow(MathGridInset & grid, MathGridInset::row_type & cellrow,
 110             string const & vskip)
 111 {
 112         ++cellrow;
 113         if (cellrow == grid.nrows()) {
 114                 //lyxerr << "adding row " << cellrow << endl;
 115                 grid.addRow(cellrow - 1);
 116                 if (cellrow == grid.nrows()) {
 117                         // We can't add a row to this grid, so let's
 118                         // append the content of this cell to the previous
 119                         // one.
 120                         // This does not happen in well formed .lyx files,
 121                         // but LyX versions 1.3.x and older could create
 122                         // such files and tex2lyx can still do that.
 123                         --cellrow;
 124                         lyxerr << "ignoring extra row";
 125                         if (!vskip.empty())
 126                                 lyxerr << " with extra space " << vskip;
 127                         lyxerr << '.' << endl;
 128                         return false;
 129                 }
 130         }
 131         grid.vcrskip(LyXLength(vskip), cellrow - 1);
 132         return true;
 133 }
 134
 135
 136 /*!
 137  * Add the column \p cellcol to \p grid.
 138  * \returns wether the column could be added. Adding a column can fail for
 139  * environments like "eqnarray" that have a fixed number of columns.
 140  */
 141 bool addCol(MathGridInset & grid, MathGridInset::col_type & cellcol)
 142 {
 143         ++cellcol;
 144         if (cellcol == grid.ncols()) {
 145                 //lyxerr << "adding column " << cellcol << endl;
 146                 grid.addCol(cellcol - 1);
 147                 if (cellcol == grid.ncols()) {
 148                         // We can't add a column to this grid, so let's
 149                         // append the content of this cell to the previous
 150                         // one.
 151                         // This does not happen in well formed .lyx files,
 152                         // but LyX versions 1.3.x and older could create
 153                         // such files and tex2lyx can still do that.
 154                         --cellcol;
 155                         lyxerr << "ignoring extra column." << endl;
 156                         return false;
 157                 }
 158         }
 159         return true;
 160 }
 161
 162
 163 /*!
 164  * Check wether the last row is empty and remove it if yes.
 165  * Otherwise the following code
 166  * \verbatim
 167 \begin{array}{|c|c|}
 168 \hline
 169 1 & 2 \\ \hline
 170 3 & 4 \\ \hline
 171 \end{array}
 172  * \endverbatim
 173  * will result in a grid with 3 rows (+ the dummy row that is always present),
 174  * because the last '\\' opens a new row.
 175  */
 176 void delEmptyLastRow(MathGridInset & grid)
 177 {
 178         MathGridInset::row_type const row = grid.nrows() - 1;
 179         for (MathGridInset::col_type col = 0; col < grid.ncols(); ++col) {
 180                 if (!grid.cell(grid.index(row, col)).empty())
 181                         return;
 182         }
 183         grid.delRow(row + 1);
 184 }
 185
 186
 187 // These are TeX's catcodes
 188 enum CatCode {
 189         catEscape,     // 0    backslash
 190         catBegin,      // 1    {
 191         catEnd,        // 2    }
 192         catMath,       // 3    $
 193         catAlign,      // 4    &
 194         catNewline,    // 5    ^^M
 195         catParameter,  // 6    #
 196         catSuper,      // 7    ^
 197         catSub,        // 8    _
 198         catIgnore,     // 9
 199         catSpace,      // 10   space
 200         catLetter,     // 11   a-zA-Z
 201         catOther,      // 12   none of the above
 202         catActive,     // 13   ~
 203         catComment,    // 14   %
 204         catInvalid     // 15   <delete>
 205 };
 206
 207 CatCode theCatcode[256];
 208
 209
 210 inline CatCode catcode(unsigned char c)
 211 {
 212         return theCatcode[c];
 213 }
 214
 215
 216 enum {
 217         FLAG_ALIGN      = 1 << 0,  //  next & or \\ ends the parsing process
 218         FLAG_BRACE_LAST = 1 << 1,  //  next closing brace ends the parsing
 219         FLAG_RIGHT      = 1 << 2,  //  next \\right ends the parsing process
 220         FLAG_END        = 1 << 3,  //  next \\end ends the parsing process
 221         FLAG_BRACK_LAST = 1 << 4,  //  next closing bracket ends the parsing
 222         FLAG_TEXTMODE   = 1 << 5,  //  we are in a box
 223         FLAG_ITEM       = 1 << 6,  //  read a (possibly braced token)
 224         FLAG_LEAVE      = 1 << 7,  //  leave the loop at the end
 225         FLAG_SIMPLE     = 1 << 8,  //  next $ leaves the loop
 226         FLAG_EQUATION   = 1 << 9,  //  next \] leaves the loop
 227         FLAG_SIMPLE2    = 1 << 10, //  next \) leaves the loop
 228         FLAG_OPTION     = 1 << 11, //  read [...] style option
 229         FLAG_BRACED     = 1 << 12  //  read {...} style argument
 230 };
 231
 232
 233 //
 234 // Helper class for parsing
 235 //
 236
 237 class Token {
 238 public:
 239         ///
 240         Token() : cs_(), char_(0), cat_(catIgnore) {}
 241         ///
 242         Token(char c, CatCode cat) : cs_(), char_(c), cat_(cat) {}
 243         ///
 244         Token(string const & cs) : cs_(cs), char_(0), cat_(catIgnore) {}
 245
 246         ///
 247         string const & cs() const { return cs_; }
 248         ///
 249         CatCode cat() const { return cat_; }
 250         ///
 251         char character() const { return char_; }
 252         ///
 253         string asString() const { return cs_.size() ? cs_ : string(1, char_); }
 254
 255 private:
 256         ///
 257         string cs_;
 258         ///
 259         char char_;
 260         ///
 261         CatCode cat_;
 262 };
 263
 264 ostream & operator<<(ostream & os, Token const & t)
 265 {
 266         if (t.cs().size())
 267                 os << '\\' << t.cs();
 268         else if (t.cat() == catLetter)
 269                 os << t.character();
 270         else
 271                 os << '[' << t.character() << ',' << t.cat() << ']';
 272         return os;
 273 }
 274
 275
 276 class Parser {
 277 public:
 278         ///
 279         typedef  MathInset::mode_type mode_type;
 280
 281         ///
 282         Parser(LyXLex & lex);
 283         ///
 284         Parser(istream & is);
 285
 286         ///
 287         bool parse(MathAtom & at);
 288         ///
 289         void parse(MathArray & array, unsigned flags, mode_type mode);
 290         ///
 291         void parse1(MathGridInset & grid, unsigned flags, mode_type mode,
 292                 bool numbered);
 293         ///
 294         MathArray parse(unsigned flags, mode_type mode);
 295         ///
 296         int lineno() const { return lineno_; }
 297         ///
 298         void putback();
 299
 300 private:
 301         ///
 302         void parse2(MathAtom & at, unsigned flags, mode_type mode, bool numbered);
 303         /// get arg delimited by 'left' and 'right'
 304         string getArg(char left, char right);
 305         ///
 306         char getChar();
 307         ///
 308         void error(string const & msg);
 309         /// dump contents to screen
 310         void dump() const;
 311         ///
 312         void tokenize(istream & is);
 313         ///
 314         void tokenize(string const & s);
 315         ///
 316         void skipSpaceTokens(istream & is, char c);
 317         ///
 318         void push_back(Token const & t);
 319         ///
 320         void pop_back();
 321         ///
 322         Token const & prevToken() const;
 323         ///
 324         Token const & nextToken() const;
 325         ///
 326         Token const & getToken();
 327         /// skips spaces if any
 328         void skipSpaces();
 329         ///
 330         void lex(string const & s);
 331         ///
 332         bool good() const;
 333         ///
 334         string parse_verbatim_item();
 335         ///
 336         string parse_verbatim_option();
 337
 338         ///
 339         int lineno_;
 340         ///
 341         vector<Token> tokens_;
 342         ///
 343         unsigned pos_;
 344         /// Stack of active environments
 345         vector<string> environments_;
 346 };
 347
 348
 349 Parser::Parser(LyXLex & lexer)
 350         : lineno_(lexer.getLineNo()), pos_(0)
 351 {
 352         tokenize(lexer.getStream());
 353         lexer.eatLine();
 354 }
 355
 356
 357 Parser::Parser(istream & is)
 358         : lineno_(0), pos_(0)
 359 {
 360         tokenize(is);
 361 }
 362
 363
 364 void Parser::push_back(Token const & t)
 365 {
 366         tokens_.push_back(t);
 367 }
 368
 369
 370 void Parser::pop_back()
 371 {
 372         tokens_.pop_back();
 373 }
 374
 375
 376 Token const & Parser::prevToken() const
 377 {
 378         static const Token dummy;
 379         return pos_ > 0 ? tokens_[pos_ - 1] : dummy;
 380 }
 381
 382
 383 Token const & Parser::nextToken() const
 384 {
 385         static const Token dummy;
 386         return good() ? tokens_[pos_] : dummy;
 387 }
 388
 389
 390 Token const & Parser::getToken()
 391 {
 392         static const Token dummy;
 393         //lyxerr << "looking at token " << tokens_[pos_] << " pos: " << pos_ << endl;
 394         return good() ? tokens_[pos_++] : dummy;
 395 }
 396
 397
 398 void Parser::skipSpaces()
 399 {
 400         while (nextToken().cat() == catSpace || nextToken().cat() == catNewline)
 401                 getToken();
 402 }
 403
 404
 405 void Parser::putback()
 406 {
 407         --pos_;
 408 }
 409
 410
 411 bool Parser::good() const
 412 {
 413         return pos_ < tokens_.size();
 414 }
 415
 416
 417 char Parser::getChar()
 418 {
 419         if (!good())
 420                 error("The input stream is not well...");
 421         return tokens_[pos_++].character();
 422 }
 423
 424
 425 string Parser::getArg(char left, char right)
 426 {
 427         skipSpaces();
 428
 429         string result;
 430         char c = getChar();
 431
 432         if (c != left)
 433                 putback();
 434         else
 435                 while ((c = getChar()) != right && good())
 436                         result += c;
 437
 438         return result;
 439 }
 440
 441
 442 void Parser::skipSpaceTokens(istream & is, char c)
 443 {
 444         // skip trailing spaces
 445         while (catcode(c) == catSpace || catcode(c) == catNewline)
 446                 if (!is.get(c))
 447                         break;
 448         //lyxerr << "putting back: " << c << endl;
 449         is.putback(c);
 450 }
 451
 452
 453 void Parser::tokenize(istream & is)
 454 {
 455         // eat everything up to the next \end_inset or end of stream
 456         // and store it in s for further tokenization
 457         string s;
 458         char c;
 459         while (is.get(c)) {
 460                 s += c;
 461                 if (s.size() >= 10 && s.substr(s.size() - 10) == "\\end_inset") {
 462                         s = s.substr(0, s.size() - 10);
 463                         break;
 464                 }
 465         }
 466         // Remove the space after \end_inset
 467         if (is.get(c) && c != ' ')
 468                 is.unget();
 469
 470         // tokenize buffer
 471         tokenize(s);
 472 }
 473
 474
 475 void Parser::tokenize(string const & buffer)
 476 {
 477         istringstream is(buffer, ios::in | ios::binary);
 478
 479         char c;
 480         while (is.get(c)) {
 481                 //lyxerr << "reading c: " << c << endl;
 482
 483                 switch (catcode(c)) {
 484                         case catNewline: {
 485                                 ++lineno_;
 486                                 is.get(c);
 487                                 if (catcode(c) == catNewline)
 488                                         ; //push_back(Token("par"));
 489                                 else {
 490                                         push_back(Token('\n', catNewline));
 491                                         is.putback(c);
 492                                 }
 493                                 break;
 494                         }
 495
 496 /*
 497                         case catComment: {
 498                                 while (is.get(c) && catcode(c) != catNewline)
 499                                         ;
 500                                 ++lineno_;
 501                                 break;
 502                         }
 503 */
 504
 505                         case catEscape: {
 506                                 is.get(c);
 507                                 if (!is) {
 508                                         error("unexpected end of input");
 509                                 } else {
 510                                         string s(1, c);
 511                                         if (catcode(c) == catLetter) {
 512                                                 // collect letters
 513                                                 while (is.get(c) && catcode(c) == catLetter)
 514                                                         s += c;
 515                                                 skipSpaceTokens(is, c);
 516                                         }
 517                                         push_back(Token(s));
 518                                 }
 519                                 break;
 520                         }
 521
 522                         case catSuper:
 523                         case catSub: {
 524                                 push_back(Token(c, catcode(c)));
 525                                 is.get(c);
 526                                 skipSpaceTokens(is, c);
 527                                 break;
 528                         }
 529
 530                         case catIgnore: {
 531                                 lyxerr << "ignoring a char: " << int(c) << endl;
 532                                 break;
 533                         }
 534
 535                         default:
 536                                 push_back(Token(c, catcode(c)));
 537                 }
 538         }
 539
 540 #ifdef FILEDEBUG
 541         dump();
 542 #endif
 543 }
 544
 545
 546 void Parser::dump() const
 547 {
 548         lyxerr << "\nTokens: ";
 549         for (unsigned i = 0; i < tokens_.size(); ++i) {
 550                 if (i == pos_)
 551                         lyxerr << " <#> ";
 552                 lyxerr << tokens_[i];
 553         }
 554         lyxerr << " pos: " << pos_ << endl;
 555 }
 556
 557
 558 void Parser::error(string const & msg)
 559 {
 560         lyxerr << "Line ~" << lineno_ << ": Math parse error: " << msg << endl;
 561         dump();
 562         //exit(1);
 563 }
 564
 565
 566 bool Parser::parse(MathAtom & at)
 567 {
 568         skipSpaces();
 569         MathArray ar;
 570         parse(ar, false, MathInset::UNDECIDED_MODE);
 571         if (ar.size() != 1 || ar.front()->getType() == "none") {
 572                 lyxerr << "unusual contents found: " << ar << endl;
 573                 at = MathAtom(new MathParInset(ar));
 574                 //if (at->nargs() > 0)
 575                 //      at.nucleus()->cell(0) = ar;
 576                 //else
 577                 //      lyxerr << "unusual contents found: " << ar << endl;
 578                 return true;
 579         }
 580         at = ar[0];
 581         return true;
 582 }
 583
 584
 585 string Parser::parse_verbatim_option()
 586 {
 587         skipSpaces();
 588         string res;
 589         if (nextToken().character() == '[') {
 590                 Token t = getToken();
 591                 for (Token t = getToken(); t.character() != ']' && good(); t = getToken()) {
 592                         if (t.cat() == catBegin) {
 593                                 putback();
 594                                 res += '{' + parse_verbatim_item() + '}';
 595                         } else
 596                                 res += t.asString();
 597                 }
 598         }
 599         return res;
 600 }
 601
 602
 603 string Parser::parse_verbatim_item()
 604 {
 605         skipSpaces();
 606         string res;
 607         if (nextToken().cat() == catBegin) {
 608                 Token t = getToken();
 609                 for (Token t = getToken(); t.cat() != catEnd && good(); t = getToken()) {
 610                         if (t.cat() == catBegin) {
 611                                 putback();
 612                                 res += '{' + parse_verbatim_item() + '}';
 613                         }
 614                         else
 615                                 res += t.asString();
 616                 }
 617         }
 618         return res;
 619 }
 620
 621
 622 MathArray Parser::parse(unsigned flags, mode_type mode)
 623 {
 624         MathArray ar;
 625         parse(ar, flags, mode);
 626         return ar;
 627 }
 628
 629
 630 void Parser::parse(MathArray & array, unsigned flags, mode_type mode)
 631 {
 632         MathGridInset grid(1, 1);
 633         parse1(grid, flags, mode, false);
 634         array = grid.cell(0);
 635 }
 636
 637
 638 void Parser::parse2(MathAtom & at, const unsigned flags, const mode_type mode,
 639         const bool numbered)
 640 {
 641         parse1(*(at.nucleus()->asGridInset()), flags, mode, numbered);
 642 }
 643
 644
 645 void Parser::parse1(MathGridInset & grid, unsigned flags,
 646         const mode_type mode, const bool numbered)
 647 {
 648         int limits = 0;
 649         MathGridInset::row_type cellrow = 0;
 650         MathGridInset::col_type cellcol = 0;
 651         MathArray * cell = &grid.cell(grid.index(cellrow, cellcol));
 652
 653         if (grid.asHullInset())
 654                 grid.asHullInset()->numbered(cellrow, numbered);
 655
 656         //dump();
 657         //lyxerr << " flags: " << flags << endl;
 658         //lyxerr << " mode: " << mode  << endl;
 659         //lyxerr << "grid: " << grid << endl;
 660
 661         while (good()) {
 662                 Token const & t = getToken();
 663
 664 #ifdef FILEDEBUG
 665                 lyxerr << "t: " << t << " flags: " << flags << endl;
 666                 lyxerr << "mode: " << mode  << endl;
 667                 cell->dump();
 668                 lyxerr << endl;
 669 #endif
 670
 671                 if (flags & FLAG_ITEM) {
 672
 673                         if (t.cat() == catBegin) {
 674                                 // skip the brace and collect everything to the next matching
 675                                 // closing brace
 676                                 parse1(grid, FLAG_BRACE_LAST, mode, numbered);
 677                                 return;
 678                         }
 679
 680                         // handle only this single token, leave the loop if done
 681                         flags = FLAG_LEAVE;
 682                 }
 683
 684
 685                 if (flags & FLAG_BRACED) {
 686                         if (t.cat() == catSpace)
 687                                 continue;
 688
 689                         if (t.cat() != catBegin) {
 690                                 error("opening brace expected");
 691                                 return;
 692                         }
 693
 694                         // skip the brace and collect everything to the next matching
 695                         // closing brace
 696                         flags = FLAG_BRACE_LAST;
 697                 }
 698
 699
 700                 if (flags & FLAG_OPTION) {
 701                         if (t.cat() == catOther && t.character() == '[') {
 702                                 MathArray ar;
 703                                 parse(ar, FLAG_BRACK_LAST, mode);
 704                                 cell->append(ar);
 705                         } else {
 706                                 // no option found, put back token and we are done
 707                                 putback();
 708                         }
 709                         return;
 710                 }
 711
 712                 //
 713                 // cat codes
 714                 //
 715                 if (t.cat() == catMath) {
 716                         if (mode != MathInset::MATH_MODE) {
 717                                 // we are inside some text mode thingy, so opening new math is allowed
 718                                 Token const & n = getToken();
 719                                 if (n.cat() == catMath) {
 720                                         // TeX's $$...$$ syntax for displayed math
 721                                         cell->push_back(MathAtom(new MathHullInset("equation")));
 722                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 723                                         getToken(); // skip the second '$' token
 724                                 } else {
 725                                         // simple $...$  stuff
 726                                         putback();
 727                                         cell->push_back(MathAtom(new MathHullInset("simple")));
 728                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 729                                 }
 730                         }
 731
 732                         else if (flags & FLAG_SIMPLE) {
 733                                 // this is the end of the formula
 734                                 return;
 735                         }
 736
 737                         else {
 738                                 error("something strange in the parser");
 739                                 break;
 740                         }
 741                 }
 742
 743                 else if (t.cat() == catLetter)
 744                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 745
 746                 else if (t.cat() == catSpace && mode != MathInset::MATH_MODE) {
 747                         if (cell->empty() || cell->back()->getChar() != ' ')
 748                                 cell->push_back(MathAtom(new MathCharInset(t.character())));
 749                 }
 750
 751                 else if (t.cat() == catNewline && mode != MathInset::MATH_MODE) {
 752                         if (cell->empty() || cell->back()->getChar() != ' ')
 753                                 cell->push_back(MathAtom(new MathCharInset(' ')));
 754                 }
 755
 756                 else if (t.cat() == catParameter) {
 757                         Token const & n = getToken();
 758                         cell->push_back(MathAtom(new MathMacroArgument(n.character()-'0')));
 759                 }
 760
 761                 else if (t.cat() == catActive)
 762                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 763
 764                 else if (t.cat() == catBegin) {
 765                         MathArray ar;
 766                         parse(ar, FLAG_BRACE_LAST, mode);
 767                         // do not create a BraceInset if they were written by LyX
 768                         // this helps to keep the annoyance of  "a choose b"  to a minimum
 769                         if (ar.size() == 1 && ar[0]->extraBraces())
 770                                 cell->append(ar);
 771                         else
 772                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
 773                 }
 774
 775                 else if (t.cat() == catEnd) {
 776                         if (flags & FLAG_BRACE_LAST)
 777                                 return;
 778                         error("found '}' unexpectedly");
 779                         //BOOST_ASSERT(false);
 780                         //add(cell, '}', LM_TC_TEX);
 781                 }
 782
 783                 else if (t.cat() == catAlign) {
 784                         //lyxerr << " column now " << (cellcol + 1)
 785                         //       << " max: " << grid.ncols() << endl;
 786                         if (flags & FLAG_ALIGN)
 787                                 return;
 788                         if (addCol(grid, cellcol))
 789                                 cell = &grid.cell(grid.index(cellrow, cellcol));
 790                 }
 791
 792                 else if (t.cat() == catSuper || t.cat() == catSub) {
 793                         bool up = (t.cat() == catSuper);
 794                         // we need no new script inset if the last thing was a scriptinset,
 795                         // which has that script already not the same script already
 796                         if (!cell->size())
 797                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 798                         else if (cell->back()->asScriptInset() &&
 799                                         !cell->back()->asScriptInset()->has(up))
 800                                 cell->back().nucleus()->asScriptInset()->ensure(up);
 801                         else if (cell->back()->asScriptInset())
 802                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 803                         else
 804                                 cell->back() = MathAtom(new MathScriptInset(cell->back(), up));
 805                         MathScriptInset * p = cell->back().nucleus()->asScriptInset();
 806                         // special handling of {}-bases
 807                         // is this always correct?
 808                         // It appears that this is wrong (Dekel)
 809                         //if (p->nuc().size() == 1 && p->nuc().back()->asNestInset() &&
 810                         //    p->nuc().back()->extraBraces())
 811                         //      p->nuc() = p->nuc().back()->asNestInset()->cell(0);
 812                         parse(p->cell(p->idxOfScript(up)), FLAG_ITEM, mode);
 813                         if (limits) {
 814                                 p->limits(limits);
 815                                 limits = 0;
 816                         }
 817                 }
 818
 819                 else if (t.character() == ']' && (flags & FLAG_BRACK_LAST)) {
 820                         //lyxerr << "finished reading option" << endl;
 821                         return;
 822                 }
 823
 824                 else if (t.cat() == catOther)
 825                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 826
 827                 else if (t.cat() == catComment) {
 828                         string s;
 829                         while (good()) {
 830                                 Token const & t = getToken();
 831                                 if (t.cat() == catNewline)
 832                                         break;
 833                                 s += t.asString();
 834                         }
 835                         cell->push_back(MathAtom(new MathCommentInset(s)));
 836                         skipSpaces();
 837                 }
 838
 839                 //
 840                 // control sequences
 841                 //
 842
 843                 else if (t.cs() == "lyxlock") {
 844                         if (cell->size())
 845                                 cell->back().nucleus()->lock(true);
 846                 }
 847
 848                 else if (t.cs() == "def" ||
 849                         t.cs() == "newcommand" ||
 850                         t.cs() == "renewcommand")
 851                 {
 852                         string const type = t.cs();
 853                         string name;
 854                         int nargs = 0;
 855                         if (t.cs() == "def") {
 856                                 // get name
 857                                 name = getToken().cs();
 858
 859                                 // read parameter
 860                                 string pars;
 861                                 while (good() && nextToken().cat() != catBegin) {
 862                                         pars += getToken().cs();
 863                                         ++nargs;
 864                                 }
 865                                 nargs /= 2;
 866                                 //lyxerr << "read \\def parameter list '" << pars << "'" << endl;
 867
 868                         } else { // t.cs() == "newcommand" || t.cs() == "renewcommand"
 869
 870                                 if (getToken().cat() != catBegin) {
 871                                         error("'{' in \\newcommand expected (1) ");
 872                                         return;
 873                                 }
 874
 875                                 name = getToken().cs();
 876
 877                                 if (getToken().cat() != catEnd) {
 878                                         error("'}' in \\newcommand expected");
 879                                         return;
 880                                 }
 881
 882                                 string const arg  = getArg('[', ']');
 883                                 if (!arg.empty())
 884                                         nargs = convert<int>(arg);
 885
 886                         }
 887
 888                         MathArray ar1;
 889                         parse(ar1, FLAG_ITEM, MathInset::UNDECIDED_MODE);
 890
 891                         // we cannot handle recursive stuff at all
 892                         //MathArray test;
 893                         //test.push_back(createMathInset(name));
 894                         //if (ar1.contains(test)) {
 895                         //      error("we cannot handle recursive macros at all.");
 896                         //      return;
 897                         //}
 898
 899                         // is a version for display attached?
 900                         skipSpaces();
 901                         MathArray ar2;
 902                         if (nextToken().cat() == catBegin)
 903                                 parse(ar2, FLAG_ITEM, MathInset::MATH_MODE);
 904
 905                         cell->push_back(MathAtom(new MathMacroTemplate(name, nargs, type,
 906                                 ar1, ar2)));
 907                 }
 908
 909                 else if (t.cs() == "(") {
 910                         cell->push_back(MathAtom(new MathHullInset("simple")));
 911                         parse2(cell->back(), FLAG_SIMPLE2, MathInset::MATH_MODE, false);
 912                 }
 913
 914                 else if (t.cs() == "[") {
 915                         cell->push_back(MathAtom(new MathHullInset("equation")));
 916                         parse2(cell->back(), FLAG_EQUATION, MathInset::MATH_MODE, false);
 917                 }
 918
 919                 else if (t.cs() == "protect")
 920                         // ignore \\protect, will hopefully be re-added during output
 921                         ;
 922
 923                 else if (t.cs() == "end") {
 924                         if (flags & FLAG_END) {
 925                                 // eat environment name
 926                                 string const name = getArg('{', '}');
 927                                 if (environments_.empty())
 928                                         error("'found \\end{" + name +
 929                                               "}' without matching '\\begin{" +
 930                                               name + "}'");
 931                                 else if (name != environments_.back())
 932                                         error("'\\end{" + name +
 933                                               "}' does not match '\\begin{" +
 934                                               environments_.back() + "}'");
 935                                 else {
 936                                         environments_.pop_back();
 937                                         if (name == "array" ||
 938                                             name == "subarray")
 939                                                 delEmptyLastRow(grid);
 940                                         return;
 941                                 }
 942                         } else
 943                                 error("found 'end' unexpectedly");
 944                 }
 945
 946                 else if (t.cs() == ")") {
 947                         if (flags & FLAG_SIMPLE2)
 948                                 return;
 949                         error("found '\\)' unexpectedly");
 950                 }
 951
 952                 else if (t.cs() == "]") {
 953                         if (flags & FLAG_EQUATION)
 954                                 return;
 955                         error("found '\\]' unexpectedly");
 956                 }
 957
 958                 else if (t.cs() == "\\") {
 959                         if (flags & FLAG_ALIGN)
 960                                 return;
 961                         if (addRow(grid, cellrow, getArg('[', ']'))) {
 962                                 cellcol = 0;
 963                                 if (grid.asHullInset())
 964                                         grid.asHullInset()->numbered(
 965                                                         cellrow, numbered);
 966                                 cell = &grid.cell(grid.index(cellrow,
 967                                                              cellcol));
 968                         }
 969                 }
 970
 971 #if 0
 972                 else if (t.cs() == "multicolumn") {
 973                         // extract column count and insert dummy cells
 974                         MathArray count;
 975                         parse(count, FLAG_ITEM, mode);
 976                         int cols = 1;
 977                         if (!extractNumber(count, cols)) {
 978                                 lyxerr << " can't extract number of cells from " << count << endl;
 979                         }
 980                         // resize the table if necessary
 981                         for (int i = 0; i < cols; ++i) {
 982                                 if (addCol(grid, cellcol)) {
 983                                         cell = &grid.cell(grid.index(
 984                                                         cellrow, cellcol));
 985                                         // mark this as dummy
 986                                         grid.cellinfo(grid.index(
 987                                                 cellrow, cellcol)).dummy_ = true;
 988                                 }
 989                         }
 990                         // the last cell is the real thing, not a dummy
 991                         grid.cellinfo(grid.index(cellrow, cellcol)).dummy_ = false;
 992
 993                         // read special alignment
 994                         MathArray align;
 995                         parse(align, FLAG_ITEM, mode);
 996                         //grid.cellinfo(grid.index(cellrow, cellcol)).align_ = extractString(align);
 997
 998                         // parse the remaining contents into the "real" cell
 999                         parse(*cell, FLAG_ITEM, mode);
1000                 }
1001 #endif
1002
1003                 else if (t.cs() == "limits")
1004                         limits = 1;
1005
1006                 else if (t.cs() == "nolimits")
1007                         limits = -1;
1008
1009                 else if (t.cs() == "nonumber") {
1010                         if (grid.asHullInset())
1011                                 grid.asHullInset()->numbered(cellrow, false);
1012                 }
1013
1014                 else if (t.cs() == "number") {
1015                         if (grid.asHullInset())
1016                                 grid.asHullInset()->numbered(cellrow, true);
1017                 }
1018
1019                 else if (t.cs() == "hline") {
1020                         grid.rowinfo(cellrow).lines_ ++;
1021                 }
1022
1023                 else if (t.cs() == "sqrt") {
1024                         MathArray ar;
1025                         parse(ar, FLAG_OPTION, mode);
1026                         if (ar.size()) {
1027                                 cell->push_back(MathAtom(new MathRootInset));
1028                                 cell->back().nucleus()->cell(0) = ar;
1029                                 parse(cell->back().nucleus()->cell(1), FLAG_ITEM, mode);
1030                         } else {
1031                                 cell->push_back(MathAtom(new MathSqrtInset));
1032                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1033                         }
1034                 }
1035
1036                 else if (t.cs() == "xrightarrow" || t.cs() == "xleftarrow") {
1037                         cell->push_back(createMathInset(t.cs()));
1038                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
1039                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1040                 }
1041
1042                 else if (t.cs() == "ref" || t.cs() == "prettyref" ||
1043                                 t.cs() == "pageref" || t.cs() == "vpageref" || t.cs() == "vref") {
1044                         cell->push_back(MathAtom(new RefInset(t.cs())));
1045                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
1046                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1047                 }
1048
1049                 else if (t.cs() == "left") {
1050                         skipSpaces();
1051                         Token const & tl = getToken();
1052                         // \| and \Vert are equivalent, and MathDelimInset
1053                         // can't handle \|
1054                         // FIXME: fix this in MathDelimInset itself!
1055                         string const l = tl.cs() == "|" ? "Vert" : tl.asString();
1056                         MathArray ar;
1057                         parse(ar, FLAG_RIGHT, mode);
1058                         skipSpaces();
1059                         Token const & tr = getToken();
1060                         string const r = tr.cs() == "|" ? "Vert" : tr.asString();
1061                         cell->push_back(MathAtom(new MathDelimInset(l, r, ar)));
1062                 }
1063
1064                 else if (t.cs() == "right") {
1065                         if (flags & FLAG_RIGHT)
1066                                 return;
1067                         //lyxerr << "got so far: '" << cell << "'" << endl;
1068                         error("Unmatched right delimiter");
1069                         return;
1070                 }
1071
1072                 else if (t.cs() == "begin") {
1073                         string const name = getArg('{', '}');
1074                         environments_.push_back(name);
1075
1076                         if (name == "array" || name == "subarray") {
1077                                 string const valign = parse_verbatim_option() + 'c';
1078                                 string const halign = parse_verbatim_item();
1079                                 cell->push_back(MathAtom(new MathArrayInset(name, valign[0], halign)));
1080                                 parse2(cell->back(), FLAG_END, mode, false);
1081                         }
1082
1083                         else if (name == "tabular") {
1084                                 string const valign = parse_verbatim_option() + 'c';
1085                                 string const halign = parse_verbatim_item();
1086                                 cell->push_back(MathAtom(new MathTabularInset(name, valign[0], halign)));
1087                                 parse2(cell->back(), FLAG_END, MathInset::TEXT_MODE, false);
1088                         }
1089
1090                         else if (name == "split" || name == "cases" ||
1091                                  name == "gathered" || name == "aligned" ||
1092                                  name == "alignedat") {
1093                                 cell->push_back(createMathInset(name));
1094                                 parse2(cell->back(), FLAG_END, mode, false);
1095                         }
1096
1097                         else if (name == "math") {
1098                                 cell->push_back(MathAtom(new MathHullInset("simple")));
1099                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, true);
1100                         }
1101
1102                         else if (name == "equation" || name == "equation*"
1103                                         || name == "displaymath") {
1104                                 cell->push_back(MathAtom(new MathHullInset("equation")));
1105                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, (name == "equation"));
1106                         }
1107
1108                         else if (name == "eqnarray" || name == "eqnarray*") {
1109                                 cell->push_back(MathAtom(new MathHullInset("eqnarray")));
1110                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1111                         }
1112
1113                         else if (name == "align" || name == "align*") {
1114                                 cell->push_back(MathAtom(new MathHullInset("align")));
1115                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1116                         }
1117
1118                         else if (name == "flalign" || name == "flalign*") {
1119                                 cell->push_back(MathAtom(new MathHullInset("flalign")));
1120                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1121                         }
1122
1123                         else if (name == "alignat" || name == "alignat*") {
1124                                 // ignore this for a while
1125                                 getArg('{', '}');
1126                                 cell->push_back(MathAtom(new MathHullInset("alignat")));
1127                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1128                         }
1129
1130                         else if (name == "xalignat" || name == "xalignat*") {
1131                                 // ignore this for a while
1132                                 getArg('{', '}');
1133                                 cell->push_back(MathAtom(new MathHullInset("xalignat")));
1134                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1135                         }
1136
1137                         else if (name == "xxalignat") {
1138                                 // ignore this for a while
1139                                 getArg('{', '}');
1140                                 cell->push_back(MathAtom(new MathHullInset("xxalignat")));
1141                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1142                         }
1143
1144                         else if (name == "multline" || name == "multline*") {
1145                                 cell->push_back(MathAtom(new MathHullInset("multline")));
1146                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1147                         }
1148
1149                         else if (name == "gather" || name == "gather*") {
1150                                 cell->push_back(MathAtom(new MathHullInset("gather")));
1151                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1152                         }
1153
1154                         else if (latexkeys const * l = in_word_set(name)) {
1155                                 if (l->inset == "matrix") {
1156                                         cell->push_back(createMathInset(name));
1157                                         parse2(cell->back(), FLAG_END, mode, false);
1158                                 }
1159                         }
1160
1161                         else {
1162                                 dump();
1163                                 lyxerr << "found unknown math environment '" << name << "'" << endl;
1164                                 // create generic environment inset
1165                                 cell->push_back(MathAtom(new MathEnvInset(name)));
1166                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1167                         }
1168                 }
1169
1170                 else if (t.cs() == "kern") {
1171 #ifdef WITH_WARNINGS
1172 #warning A hack...
1173 #endif
1174                         string s;
1175                         while (true) {
1176                                 Token const & t = getToken();
1177                                 if (!good()) {
1178                                         putback();
1179                                         break;
1180                                 }
1181                                 s += t.character();
1182                                 if (isValidLength(s))
1183                                         break;
1184                         }
1185                         cell->push_back(MathAtom(new MathKernInset(s)));
1186                 }
1187
1188                 else if (t.cs() == "label") {
1189                         string label = parse_verbatim_item();
1190                         MathArray ar;
1191                         asArray(label, ar);
1192                         if (grid.asHullInset()) {
1193                                 grid.asHullInset()->label(cellrow, label);
1194                         } else {
1195                                 cell->push_back(createMathInset(t.cs()));
1196                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
1197                         }
1198                 }
1199
1200                 else if (t.cs() == "choose" || t.cs() == "over" || t.cs() == "atop") {
1201                         MathAtom at = createMathInset(t.cs());
1202                         at.nucleus()->cell(0) = *cell;
1203                         cell->clear();
1204                         parse(at.nucleus()->cell(1), flags, mode);
1205                         cell->push_back(at);
1206                         return;
1207                 }
1208
1209                 else if (t.cs() == "color") {
1210                         MathAtom at = createMathInset(t.cs());
1211                         parse(at.nucleus()->cell(0), FLAG_ITEM, MathInset::TEXT_MODE);
1212                         parse(at.nucleus()->cell(1), flags, mode);
1213                         cell->push_back(at);
1214                         return;
1215                 }
1216
1217                 else if (t.cs() == "substack") {
1218                         cell->push_back(createMathInset(t.cs()));
1219                         parse2(cell->back(), FLAG_ITEM, mode, false);
1220                 }
1221
1222                 else if (t.cs() == "framebox" || t.cs() == "makebox") {
1223                         cell->push_back(createMathInset(t.cs()));
1224                         parse(cell->back().nucleus()->cell(0), FLAG_OPTION, MathInset::TEXT_MODE);
1225                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, MathInset::TEXT_MODE);
1226                         parse(cell->back().nucleus()->cell(2), FLAG_ITEM, MathInset::TEXT_MODE);
1227                 }
1228
1229 #if 0
1230                 else if (t.cs() == "infer") {
1231                         MathArray ar;
1232                         parse(ar, FLAG_OPTION, mode);
1233                         cell->push_back(createMathInset(t.cs()));
1234                         parse2(cell->back(), FLAG_ITEM, mode, false);
1235                 }
1236
1237                 // Disabled
1238                 else if (1 && t.cs() == "ar") {
1239                         auto_ptr<MathXYArrowInset> p(new MathXYArrowInset);
1240                         // try to read target
1241                         parse(p->cell(0), FLAG_OTPTION, mode);
1242                         // try to read label
1243                         if (nextToken().cat() == catSuper || nextToken().cat() == catSub) {
1244                                 p->up_ = nextToken().cat() == catSuper;
1245                                 getToken();
1246                                 parse(p->cell(1), FLAG_ITEM, mode);
1247                                 //lyxerr << "read label: " << p->cell(1) << endl;
1248                         }
1249
1250                         cell->push_back(MathAtom(p.release()));
1251                         //lyxerr << "read cell: " << cell << endl;
1252                 }
1253 #endif
1254
1255                 else if (t.cs().size()) {
1256                         latexkeys const * l = in_word_set(t.cs());
1257                         if (l) {
1258                                 if (l->inset == "font") {
1259                                         cell->push_back(createMathInset(t.cs()));
1260                                         parse(cell->back().nucleus()->cell(0),
1261                                                 FLAG_ITEM, asMode(mode, l->extra));
1262                                 }
1263
1264                                 else if (l->inset == "oldfont") {
1265                                         cell->push_back(createMathInset(t.cs()));
1266                                         parse(cell->back().nucleus()->cell(0),
1267                                                 flags | FLAG_ALIGN, asMode(mode, l->extra));
1268                                         if (prevToken().cat() != catAlign &&
1269                                             prevToken().cs() != "\\")
1270                                                 return;
1271                                         putback();
1272                                 }
1273
1274                                 else if (l->inset == "style") {
1275                                         cell->push_back(createMathInset(t.cs()));
1276                                         parse(cell->back().nucleus()->cell(0),
1277                                                 flags | FLAG_ALIGN, mode);
1278                                         if (prevToken().cat() != catAlign &&
1279                                             prevToken().cs() != "\\")
1280                                                 return;
1281                                         putback();
1282                                 }
1283
1284                                 else {
1285                                         MathAtom at = createMathInset(t.cs());
1286                                         for (MathInset::idx_type i = 0; i < at->nargs(); ++i)
1287                                                 parse(at.nucleus()->cell(i),
1288                                                         FLAG_ITEM, asMode(mode, l->extra));
1289                                         cell->push_back(at);
1290                                 }
1291                         }
1292
1293                         else {
1294                                 MathAtom at = createMathInset(t.cs());
1295                                 MathInset::mode_type m = mode;
1296                                 //if (m == MathInset::UNDECIDED_MODE)
1297                                 //lyxerr << "default creation: m1: " << m << endl;
1298                                 if (at->currentMode() != MathInset::UNDECIDED_MODE)
1299                                         m = at->currentMode();
1300                                 //lyxerr << "default creation: m2: " << m << endl;
1301                                 MathInset::idx_type start = 0;
1302                                 // this fails on \bigg[...\bigg]
1303                                 //MathArray opt;
1304                                 //parse(opt, FLAG_OPTION, MathInset::VERBATIM_MODE);
1305                                 //if (opt.size()) {
1306                                 //      start = 1;
1307                                 //      at.nucleus()->cell(0) = opt;
1308                                 //}
1309                                 for (MathInset::idx_type i = start; i < at->nargs(); ++i) {
1310                                         parse(at.nucleus()->cell(i), FLAG_ITEM, m);
1311                                         skipSpaces();
1312                                 }
1313                                 cell->push_back(at);
1314                         }
1315                 }
1316
1317
1318                 if (flags & FLAG_LEAVE) {
1319                         flags &= ~FLAG_LEAVE;
1320                         break;
1321                 }
1322         }
1323 }
1324
1325
1326
1327 } // anonymous namespace
1328
1329
1330 void mathed_parse_cell(MathArray & ar, string const & str)
1331 {
1332         istringstream is(str);
1333         mathed_parse_cell(ar, is);
1334 }
1335
1336
1337 void mathed_parse_cell(MathArray & ar, istream & is)
1338 {
1339         Parser(is).parse(ar, 0, MathInset::MATH_MODE);
1340 }
1341
1342
1343 bool mathed_parse_normal(MathAtom & t, string const & str)
1344 {
1345         istringstream is(str);
1346         return Parser(is).parse(t);
1347 }
1348
1349
1350 bool mathed_parse_normal(MathAtom & t, istream & is)
1351 {
1352         return Parser(is).parse(t);
1353 }
1354
1355
1356 bool mathed_parse_normal(MathAtom & t, LyXLex & lex)
1357 {
1358         return Parser(lex).parse(t);
1359 }
1360
1361
1362 void mathed_parse_normal(MathGridInset & grid, string const & str)
1363 {
1364         istringstream is(str);
1365         Parser(is).parse1(grid, 0, MathInset::MATH_MODE, false);
1366 }
1367
1368
1369 void initParser()
1370 {
1371         fill(theCatcode, theCatcode + 256, catOther);
1372         fill(theCatcode + 'a', theCatcode + 'z' + 1, catLetter);
1373         fill(theCatcode + 'A', theCatcode + 'Z' + 1, catLetter);
1374
1375         theCatcode[int('\\')] = catEscape;
1376         theCatcode[int('{')]  = catBegin;
1377         theCatcode[int('}')]  = catEnd;
1378         theCatcode[int('$')]  = catMath;
1379         theCatcode[int('&')]  = catAlign;
1380         theCatcode[int('\n')] = catNewline;
1381         theCatcode[int('#')]  = catParameter;
1382         theCatcode[int('^')]  = catSuper;
1383         theCatcode[int('_')]  = catSub;
1384         theCatcode[int(0x7f)] = catIgnore;
1385         theCatcode[int(' ')]  = catSpace;
1386         theCatcode[int('\t')] = catSpace;
1387         theCatcode[int('\r')] = catNewline;
1388         theCatcode[int('~')]  = catActive;
1389         theCatcode[int('%')]  = catComment;
1390 }