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         // Copy the row information of the empty row (which would contain the
 185         // last hline in the example above) to the dummy row and delete the
 186         // empty row.
 187         grid.rowinfo(row + 1) = grid.rowinfo(row);
 188         grid.delRow(row);
 189 }
 190
 191
 192 // These are TeX's catcodes
 193 enum CatCode {
 194         catEscape,     // 0    backslash
 195         catBegin,      // 1    {
 196         catEnd,        // 2    }
 197         catMath,       // 3    $
 198         catAlign,      // 4    &
 199         catNewline,    // 5    ^^M
 200         catParameter,  // 6    #
 201         catSuper,      // 7    ^
 202         catSub,        // 8    _
 203         catIgnore,     // 9
 204         catSpace,      // 10   space
 205         catLetter,     // 11   a-zA-Z
 206         catOther,      // 12   none of the above
 207         catActive,     // 13   ~
 208         catComment,    // 14   %
 209         catInvalid     // 15   <delete>
 210 };
 211
 212 CatCode theCatcode[256];
 213
 214
 215 inline CatCode catcode(unsigned char c)
 216 {
 217         return theCatcode[c];
 218 }
 219
 220
 221 enum {
 222         FLAG_ALIGN      = 1 << 0,  //  next & or \\ ends the parsing process
 223         FLAG_BRACE_LAST = 1 << 1,  //  next closing brace ends the parsing
 224         FLAG_RIGHT      = 1 << 2,  //  next \\right ends the parsing process
 225         FLAG_END        = 1 << 3,  //  next \\end ends the parsing process
 226         FLAG_BRACK_LAST = 1 << 4,  //  next closing bracket ends the parsing
 227         FLAG_TEXTMODE   = 1 << 5,  //  we are in a box
 228         FLAG_ITEM       = 1 << 6,  //  read a (possibly braced token)
 229         FLAG_LEAVE      = 1 << 7,  //  leave the loop at the end
 230         FLAG_SIMPLE     = 1 << 8,  //  next $ leaves the loop
 231         FLAG_EQUATION   = 1 << 9,  //  next \] leaves the loop
 232         FLAG_SIMPLE2    = 1 << 10, //  next \) leaves the loop
 233         FLAG_OPTION     = 1 << 11, //  read [...] style option
 234         FLAG_BRACED     = 1 << 12  //  read {...} style argument
 235 };
 236
 237
 238 //
 239 // Helper class for parsing
 240 //
 241
 242 class Token {
 243 public:
 244         ///
 245         Token() : cs_(), char_(0), cat_(catIgnore) {}
 246         ///
 247         Token(char c, CatCode cat) : cs_(), char_(c), cat_(cat) {}
 248         ///
 249         Token(string const & cs) : cs_(cs), char_(0), cat_(catIgnore) {}
 250
 251         ///
 252         string const & cs() const { return cs_; }
 253         ///
 254         CatCode cat() const { return cat_; }
 255         ///
 256         char character() const { return char_; }
 257         ///
 258         string asString() const { return cs_.size() ? cs_ : string(1, char_); }
 259
 260 private:
 261         ///
 262         string cs_;
 263         ///
 264         char char_;
 265         ///
 266         CatCode cat_;
 267 };
 268
 269 ostream & operator<<(ostream & os, Token const & t)
 270 {
 271         if (t.cs().size())
 272                 os << '\\' << t.cs();
 273         else if (t.cat() == catLetter)
 274                 os << t.character();
 275         else
 276                 os << '[' << t.character() << ',' << t.cat() << ']';
 277         return os;
 278 }
 279
 280
 281 class Parser {
 282 public:
 283         ///
 284         typedef  MathInset::mode_type mode_type;
 285
 286         ///
 287         Parser(LyXLex & lex);
 288         ///
 289         Parser(istream & is);
 290
 291         ///
 292         bool parse(MathAtom & at);
 293         ///
 294         void parse(MathArray & array, unsigned flags, mode_type mode);
 295         ///
 296         void parse1(MathGridInset & grid, unsigned flags, mode_type mode,
 297                 bool numbered);
 298         ///
 299         MathArray parse(unsigned flags, mode_type mode);
 300         ///
 301         int lineno() const { return lineno_; }
 302         ///
 303         void putback();
 304
 305 private:
 306         ///
 307         void parse2(MathAtom & at, unsigned flags, mode_type mode, bool numbered);
 308         /// get arg delimited by 'left' and 'right'
 309         string getArg(char left, char right);
 310         ///
 311         char getChar();
 312         ///
 313         void error(string const & msg);
 314         /// dump contents to screen
 315         void dump() const;
 316         ///
 317         void tokenize(istream & is);
 318         ///
 319         void tokenize(string const & s);
 320         ///
 321         void skipSpaceTokens(istream & is, char c);
 322         ///
 323         void push_back(Token const & t);
 324         ///
 325         void pop_back();
 326         ///
 327         Token const & prevToken() const;
 328         ///
 329         Token const & nextToken() const;
 330         ///
 331         Token const & getToken();
 332         /// skips spaces if any
 333         void skipSpaces();
 334         ///
 335         void lex(string const & s);
 336         ///
 337         bool good() const;
 338         ///
 339         string parse_verbatim_item();
 340         ///
 341         string parse_verbatim_option();
 342
 343         ///
 344         int lineno_;
 345         ///
 346         vector<Token> tokens_;
 347         ///
 348         unsigned pos_;
 349         /// Stack of active environments
 350         vector<string> environments_;
 351 };
 352
 353
 354 Parser::Parser(LyXLex & lexer)
 355         : lineno_(lexer.getLineNo()), pos_(0)
 356 {
 357         tokenize(lexer.getStream());
 358         lexer.eatLine();
 359 }
 360
 361
 362 Parser::Parser(istream & is)
 363         : lineno_(0), pos_(0)
 364 {
 365         tokenize(is);
 366 }
 367
 368
 369 void Parser::push_back(Token const & t)
 370 {
 371         tokens_.push_back(t);
 372 }
 373
 374
 375 void Parser::pop_back()
 376 {
 377         tokens_.pop_back();
 378 }
 379
 380
 381 Token const & Parser::prevToken() const
 382 {
 383         static const Token dummy;
 384         return pos_ > 0 ? tokens_[pos_ - 1] : dummy;
 385 }
 386
 387
 388 Token const & Parser::nextToken() const
 389 {
 390         static const Token dummy;
 391         return good() ? tokens_[pos_] : dummy;
 392 }
 393
 394
 395 Token const & Parser::getToken()
 396 {
 397         static const Token dummy;
 398         //lyxerr << "looking at token " << tokens_[pos_] << " pos: " << pos_ << endl;
 399         return good() ? tokens_[pos_++] : dummy;
 400 }
 401
 402
 403 void Parser::skipSpaces()
 404 {
 405         while (nextToken().cat() == catSpace || nextToken().cat() == catNewline)
 406                 getToken();
 407 }
 408
 409
 410 void Parser::putback()
 411 {
 412         --pos_;
 413 }
 414
 415
 416 bool Parser::good() const
 417 {
 418         return pos_ < tokens_.size();
 419 }
 420
 421
 422 char Parser::getChar()
 423 {
 424         if (!good())
 425                 error("The input stream is not well...");
 426         return tokens_[pos_++].character();
 427 }
 428
 429
 430 string Parser::getArg(char left, char right)
 431 {
 432         skipSpaces();
 433
 434         string result;
 435         char c = getChar();
 436
 437         if (c != left)
 438                 putback();
 439         else
 440                 while ((c = getChar()) != right && good())
 441                         result += c;
 442
 443         return result;
 444 }
 445
 446
 447 void Parser::skipSpaceTokens(istream & is, char c)
 448 {
 449         // skip trailing spaces
 450         while (catcode(c) == catSpace || catcode(c) == catNewline)
 451                 if (!is.get(c))
 452                         break;
 453         //lyxerr << "putting back: " << c << endl;
 454         is.putback(c);
 455 }
 456
 457
 458 void Parser::tokenize(istream & is)
 459 {
 460         // eat everything up to the next \end_inset or end of stream
 461         // and store it in s for further tokenization
 462         string s;
 463         char c;
 464         while (is.get(c)) {
 465                 s += c;
 466                 if (s.size() >= 10 && s.substr(s.size() - 10) == "\\end_inset") {
 467                         s = s.substr(0, s.size() - 10);
 468                         break;
 469                 }
 470         }
 471         // Remove the space after \end_inset
 472         if (is.get(c) && c != ' ')
 473                 is.unget();
 474
 475         // tokenize buffer
 476         tokenize(s);
 477 }
 478
 479
 480 void Parser::tokenize(string const & buffer)
 481 {
 482         istringstream is(buffer, ios::in | ios::binary);
 483
 484         char c;
 485         while (is.get(c)) {
 486                 //lyxerr << "reading c: " << c << endl;
 487
 488                 switch (catcode(c)) {
 489                         case catNewline: {
 490                                 ++lineno_;
 491                                 is.get(c);
 492                                 if (catcode(c) == catNewline)
 493                                         ; //push_back(Token("par"));
 494                                 else {
 495                                         push_back(Token('\n', catNewline));
 496                                         is.putback(c);
 497                                 }
 498                                 break;
 499                         }
 500
 501 /*
 502                         case catComment: {
 503                                 while (is.get(c) && catcode(c) != catNewline)
 504                                         ;
 505                                 ++lineno_;
 506                                 break;
 507                         }
 508 */
 509
 510                         case catEscape: {
 511                                 is.get(c);
 512                                 if (!is) {
 513                                         error("unexpected end of input");
 514                                 } else {
 515                                         string s(1, c);
 516                                         if (catcode(c) == catLetter) {
 517                                                 // collect letters
 518                                                 while (is.get(c) && catcode(c) == catLetter)
 519                                                         s += c;
 520                                                 skipSpaceTokens(is, c);
 521                                         }
 522                                         push_back(Token(s));
 523                                 }
 524                                 break;
 525                         }
 526
 527                         case catSuper:
 528                         case catSub: {
 529                                 push_back(Token(c, catcode(c)));
 530                                 is.get(c);
 531                                 skipSpaceTokens(is, c);
 532                                 break;
 533                         }
 534
 535                         case catIgnore: {
 536                                 lyxerr << "ignoring a char: " << int(c) << endl;
 537                                 break;
 538                         }
 539
 540                         default:
 541                                 push_back(Token(c, catcode(c)));
 542                 }
 543         }
 544
 545 #ifdef FILEDEBUG
 546         dump();
 547 #endif
 548 }
 549
 550
 551 void Parser::dump() const
 552 {
 553         lyxerr << "\nTokens: ";
 554         for (unsigned i = 0; i < tokens_.size(); ++i) {
 555                 if (i == pos_)
 556                         lyxerr << " <#> ";
 557                 lyxerr << tokens_[i];
 558         }
 559         lyxerr << " pos: " << pos_ << endl;
 560 }
 561
 562
 563 void Parser::error(string const & msg)
 564 {
 565         lyxerr << "Line ~" << lineno_ << ": Math parse error: " << msg << endl;
 566         dump();
 567         //exit(1);
 568 }
 569
 570
 571 bool Parser::parse(MathAtom & at)
 572 {
 573         skipSpaces();
 574         MathArray ar;
 575         parse(ar, false, MathInset::UNDECIDED_MODE);
 576         if (ar.size() != 1 || ar.front()->getType() == "none") {
 577                 lyxerr << "unusual contents found: " << ar << endl;
 578                 at = MathAtom(new MathParInset(ar));
 579                 //if (at->nargs() > 0)
 580                 //      at.nucleus()->cell(0) = ar;
 581                 //else
 582                 //      lyxerr << "unusual contents found: " << ar << endl;
 583                 return true;
 584         }
 585         at = ar[0];
 586         return true;
 587 }
 588
 589
 590 string Parser::parse_verbatim_option()
 591 {
 592         skipSpaces();
 593         string res;
 594         if (nextToken().character() == '[') {
 595                 Token t = getToken();
 596                 for (Token t = getToken(); t.character() != ']' && good(); t = getToken()) {
 597                         if (t.cat() == catBegin) {
 598                                 putback();
 599                                 res += '{' + parse_verbatim_item() + '}';
 600                         } else
 601                                 res += t.asString();
 602                 }
 603         }
 604         return res;
 605 }
 606
 607
 608 string Parser::parse_verbatim_item()
 609 {
 610         skipSpaces();
 611         string res;
 612         if (nextToken().cat() == catBegin) {
 613                 Token t = getToken();
 614                 for (Token t = getToken(); t.cat() != catEnd && good(); t = getToken()) {
 615                         if (t.cat() == catBegin) {
 616                                 putback();
 617                                 res += '{' + parse_verbatim_item() + '}';
 618                         }
 619                         else
 620                                 res += t.asString();
 621                 }
 622         }
 623         return res;
 624 }
 625
 626
 627 MathArray Parser::parse(unsigned flags, mode_type mode)
 628 {
 629         MathArray ar;
 630         parse(ar, flags, mode);
 631         return ar;
 632 }
 633
 634
 635 void Parser::parse(MathArray & array, unsigned flags, mode_type mode)
 636 {
 637         MathGridInset grid(1, 1);
 638         parse1(grid, flags, mode, false);
 639         array = grid.cell(0);
 640 }
 641
 642
 643 void Parser::parse2(MathAtom & at, const unsigned flags, const mode_type mode,
 644         const bool numbered)
 645 {
 646         parse1(*(at.nucleus()->asGridInset()), flags, mode, numbered);
 647 }
 648
 649
 650 void Parser::parse1(MathGridInset & grid, unsigned flags,
 651         const mode_type mode, const bool numbered)
 652 {
 653         int limits = 0;
 654         MathGridInset::row_type cellrow = 0;
 655         MathGridInset::col_type cellcol = 0;
 656         MathArray * cell = &grid.cell(grid.index(cellrow, cellcol));
 657
 658         if (grid.asHullInset())
 659                 grid.asHullInset()->numbered(cellrow, numbered);
 660
 661         //dump();
 662         //lyxerr << " flags: " << flags << endl;
 663         //lyxerr << " mode: " << mode  << endl;
 664         //lyxerr << "grid: " << grid << endl;
 665
 666         while (good()) {
 667                 Token const & t = getToken();
 668
 669 #ifdef FILEDEBUG
 670                 lyxerr << "t: " << t << " flags: " << flags << endl;
 671                 lyxerr << "mode: " << mode  << endl;
 672                 cell->dump();
 673                 lyxerr << endl;
 674 #endif
 675
 676                 if (flags & FLAG_ITEM) {
 677
 678                         if (t.cat() == catBegin) {
 679                                 // skip the brace and collect everything to the next matching
 680                                 // closing brace
 681                                 parse1(grid, FLAG_BRACE_LAST, mode, numbered);
 682                                 return;
 683                         }
 684
 685                         // handle only this single token, leave the loop if done
 686                         flags = FLAG_LEAVE;
 687                 }
 688
 689
 690                 if (flags & FLAG_BRACED) {
 691                         if (t.cat() == catSpace)
 692                                 continue;
 693
 694                         if (t.cat() != catBegin) {
 695                                 error("opening brace expected");
 696                                 return;
 697                         }
 698
 699                         // skip the brace and collect everything to the next matching
 700                         // closing brace
 701                         flags = FLAG_BRACE_LAST;
 702                 }
 703
 704
 705                 if (flags & FLAG_OPTION) {
 706                         if (t.cat() == catOther && t.character() == '[') {
 707                                 MathArray ar;
 708                                 parse(ar, FLAG_BRACK_LAST, mode);
 709                                 cell->append(ar);
 710                         } else {
 711                                 // no option found, put back token and we are done
 712                                 putback();
 713                         }
 714                         return;
 715                 }
 716
 717                 //
 718                 // cat codes
 719                 //
 720                 if (t.cat() == catMath) {
 721                         if (mode != MathInset::MATH_MODE) {
 722                                 // we are inside some text mode thingy, so opening new math is allowed
 723                                 Token const & n = getToken();
 724                                 if (n.cat() == catMath) {
 725                                         // TeX's $$...$$ syntax for displayed math
 726                                         cell->push_back(MathAtom(new MathHullInset("equation")));
 727                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 728                                         getToken(); // skip the second '$' token
 729                                 } else {
 730                                         // simple $...$  stuff
 731                                         putback();
 732                                         cell->push_back(MathAtom(new MathHullInset("simple")));
 733                                         parse2(cell->back(), FLAG_SIMPLE, MathInset::MATH_MODE, false);
 734                                 }
 735                         }
 736
 737                         else if (flags & FLAG_SIMPLE) {
 738                                 // this is the end of the formula
 739                                 return;
 740                         }
 741
 742                         else {
 743                                 error("something strange in the parser");
 744                                 break;
 745                         }
 746                 }
 747
 748                 else if (t.cat() == catLetter)
 749                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 750
 751                 else if (t.cat() == catSpace && mode != MathInset::MATH_MODE) {
 752                         if (cell->empty() || cell->back()->getChar() != ' ')
 753                                 cell->push_back(MathAtom(new MathCharInset(t.character())));
 754                 }
 755
 756                 else if (t.cat() == catNewline && mode != MathInset::MATH_MODE) {
 757                         if (cell->empty() || cell->back()->getChar() != ' ')
 758                                 cell->push_back(MathAtom(new MathCharInset(' ')));
 759                 }
 760
 761                 else if (t.cat() == catParameter) {
 762                         Token const & n = getToken();
 763                         cell->push_back(MathAtom(new MathMacroArgument(n.character()-'0')));
 764                 }
 765
 766                 else if (t.cat() == catActive)
 767                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 768
 769                 else if (t.cat() == catBegin) {
 770                         MathArray ar;
 771                         parse(ar, FLAG_BRACE_LAST, mode);
 772                         // do not create a BraceInset if they were written by LyX
 773                         // this helps to keep the annoyance of  "a choose b"  to a minimum
 774                         if (ar.size() == 1 && ar[0]->extraBraces())
 775                                 cell->append(ar);
 776                         else
 777                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
 778                 }
 779
 780                 else if (t.cat() == catEnd) {
 781                         if (flags & FLAG_BRACE_LAST)
 782                                 return;
 783                         error("found '}' unexpectedly");
 784                         //BOOST_ASSERT(false);
 785                         //add(cell, '}', LM_TC_TEX);
 786                 }
 787
 788                 else if (t.cat() == catAlign) {
 789                         //lyxerr << " column now " << (cellcol + 1)
 790                         //       << " max: " << grid.ncols() << endl;
 791                         if (flags & FLAG_ALIGN)
 792                                 return;
 793                         if (addCol(grid, cellcol))
 794                                 cell = &grid.cell(grid.index(cellrow, cellcol));
 795                 }
 796
 797                 else if (t.cat() == catSuper || t.cat() == catSub) {
 798                         bool up = (t.cat() == catSuper);
 799                         // we need no new script inset if the last thing was a scriptinset,
 800                         // which has that script already not the same script already
 801                         if (!cell->size())
 802                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 803                         else if (cell->back()->asScriptInset() &&
 804                                         !cell->back()->asScriptInset()->has(up))
 805                                 cell->back().nucleus()->asScriptInset()->ensure(up);
 806                         else if (cell->back()->asScriptInset())
 807                                 cell->push_back(MathAtom(new MathScriptInset(up)));
 808                         else
 809                                 cell->back() = MathAtom(new MathScriptInset(cell->back(), up));
 810                         MathScriptInset * p = cell->back().nucleus()->asScriptInset();
 811                         // special handling of {}-bases
 812                         // is this always correct?
 813                         // It appears that this is wrong (Dekel)
 814                         //if (p->nuc().size() == 1 && p->nuc().back()->asNestInset() &&
 815                         //    p->nuc().back()->extraBraces())
 816                         //      p->nuc() = p->nuc().back()->asNestInset()->cell(0);
 817                         parse(p->cell(p->idxOfScript(up)), FLAG_ITEM, mode);
 818                         if (limits) {
 819                                 p->limits(limits);
 820                                 limits = 0;
 821                         }
 822                 }
 823
 824                 else if (t.character() == ']' && (flags & FLAG_BRACK_LAST)) {
 825                         //lyxerr << "finished reading option" << endl;
 826                         return;
 827                 }
 828
 829                 else if (t.cat() == catOther)
 830                         cell->push_back(MathAtom(new MathCharInset(t.character())));
 831
 832                 else if (t.cat() == catComment) {
 833                         string s;
 834                         while (good()) {
 835                                 Token const & t = getToken();
 836                                 if (t.cat() == catNewline)
 837                                         break;
 838                                 s += t.asString();
 839                         }
 840                         cell->push_back(MathAtom(new MathCommentInset(s)));
 841                         skipSpaces();
 842                 }
 843
 844                 //
 845                 // control sequences
 846                 //
 847
 848                 else if (t.cs() == "lyxlock") {
 849                         if (cell->size())
 850                                 cell->back().nucleus()->lock(true);
 851                 }
 852
 853                 else if (t.cs() == "def" ||
 854                         t.cs() == "newcommand" ||
 855                         t.cs() == "renewcommand")
 856                 {
 857                         string const type = t.cs();
 858                         string name;
 859                         int nargs = 0;
 860                         if (t.cs() == "def") {
 861                                 // get name
 862                                 name = getToken().cs();
 863
 864                                 // read parameter
 865                                 string pars;
 866                                 while (good() && nextToken().cat() != catBegin) {
 867                                         pars += getToken().cs();
 868                                         ++nargs;
 869                                 }
 870                                 nargs /= 2;
 871                                 //lyxerr << "read \\def parameter list '" << pars << "'" << endl;
 872
 873                         } else { // t.cs() == "newcommand" || t.cs() == "renewcommand"
 874
 875                                 if (getToken().cat() != catBegin) {
 876                                         error("'{' in \\newcommand expected (1) ");
 877                                         return;
 878                                 }
 879
 880                                 name = getToken().cs();
 881
 882                                 if (getToken().cat() != catEnd) {
 883                                         error("'}' in \\newcommand expected");
 884                                         return;
 885                                 }
 886
 887                                 string const arg  = getArg('[', ']');
 888                                 if (!arg.empty())
 889                                         nargs = convert<int>(arg);
 890
 891                         }
 892
 893                         MathArray ar1;
 894                         parse(ar1, FLAG_ITEM, MathInset::UNDECIDED_MODE);
 895
 896                         // we cannot handle recursive stuff at all
 897                         //MathArray test;
 898                         //test.push_back(createMathInset(name));
 899                         //if (ar1.contains(test)) {
 900                         //      error("we cannot handle recursive macros at all.");
 901                         //      return;
 902                         //}
 903
 904                         // is a version for display attached?
 905                         skipSpaces();
 906                         MathArray ar2;
 907                         if (nextToken().cat() == catBegin)
 908                                 parse(ar2, FLAG_ITEM, MathInset::MATH_MODE);
 909
 910                         cell->push_back(MathAtom(new MathMacroTemplate(name, nargs, type,
 911                                 ar1, ar2)));
 912                 }
 913
 914                 else if (t.cs() == "(") {
 915                         cell->push_back(MathAtom(new MathHullInset("simple")));
 916                         parse2(cell->back(), FLAG_SIMPLE2, MathInset::MATH_MODE, false);
 917                 }
 918
 919                 else if (t.cs() == "[") {
 920                         cell->push_back(MathAtom(new MathHullInset("equation")));
 921                         parse2(cell->back(), FLAG_EQUATION, MathInset::MATH_MODE, false);
 922                 }
 923
 924                 else if (t.cs() == "protect")
 925                         // ignore \\protect, will hopefully be re-added during output
 926                         ;
 927
 928                 else if (t.cs() == "end") {
 929                         if (flags & FLAG_END) {
 930                                 // eat environment name
 931                                 string const name = getArg('{', '}');
 932                                 if (environments_.empty())
 933                                         error("'found \\end{" + name +
 934                                               "}' without matching '\\begin{" +
 935                                               name + "}'");
 936                                 else if (name != environments_.back())
 937                                         error("'\\end{" + name +
 938                                               "}' does not match '\\begin{" +
 939                                               environments_.back() + "}'");
 940                                 else {
 941                                         environments_.pop_back();
 942                                         // Delete empty last row in matrix
 943                                         // like insets.
 944                                         // If you abuse MathGridInset for
 945                                         // non-matrix like structures you
 946                                         // probably need to refine this test.
 947                                         // Right now we only have to test for
 948                                         // single line hull insets.
 949                                         if (grid.nrows() > 1)
 950                                                 delEmptyLastRow(grid);
 951                                         return;
 952                                 }
 953                         } else
 954                                 error("found 'end' unexpectedly");
 955                 }
 956
 957                 else if (t.cs() == ")") {
 958                         if (flags & FLAG_SIMPLE2)
 959                                 return;
 960                         error("found '\\)' unexpectedly");
 961                 }
 962
 963                 else if (t.cs() == "]") {
 964                         if (flags & FLAG_EQUATION)
 965                                 return;
 966                         error("found '\\]' unexpectedly");
 967                 }
 968
 969                 else if (t.cs() == "\\") {
 970                         if (flags & FLAG_ALIGN)
 971                                 return;
 972                         if (addRow(grid, cellrow, getArg('[', ']'))) {
 973                                 cellcol = 0;
 974                                 if (grid.asHullInset())
 975                                         grid.asHullInset()->numbered(
 976                                                         cellrow, numbered);
 977                                 cell = &grid.cell(grid.index(cellrow,
 978                                                              cellcol));
 979                         }
 980                 }
 981
 982 #if 0
 983                 else if (t.cs() == "multicolumn") {
 984                         // extract column count and insert dummy cells
 985                         MathArray count;
 986                         parse(count, FLAG_ITEM, mode);
 987                         int cols = 1;
 988                         if (!extractNumber(count, cols)) {
 989                                 lyxerr << " can't extract number of cells from " << count << endl;
 990                         }
 991                         // resize the table if necessary
 992                         for (int i = 0; i < cols; ++i) {
 993                                 if (addCol(grid, cellcol)) {
 994                                         cell = &grid.cell(grid.index(
 995                                                         cellrow, cellcol));
 996                                         // mark this as dummy
 997                                         grid.cellinfo(grid.index(
 998                                                 cellrow, cellcol)).dummy_ = true;
 999                                 }
1000                         }
1001                         // the last cell is the real thing, not a dummy
1002                         grid.cellinfo(grid.index(cellrow, cellcol)).dummy_ = false;
1003
1004                         // read special alignment
1005                         MathArray align;
1006                         parse(align, FLAG_ITEM, mode);
1007                         //grid.cellinfo(grid.index(cellrow, cellcol)).align_ = extractString(align);
1008
1009                         // parse the remaining contents into the "real" cell
1010                         parse(*cell, FLAG_ITEM, mode);
1011                 }
1012 #endif
1013
1014                 else if (t.cs() == "limits")
1015                         limits = 1;
1016
1017                 else if (t.cs() == "nolimits")
1018                         limits = -1;
1019
1020                 else if (t.cs() == "nonumber") {
1021                         if (grid.asHullInset())
1022                                 grid.asHullInset()->numbered(cellrow, false);
1023                 }
1024
1025                 else if (t.cs() == "number") {
1026                         if (grid.asHullInset())
1027                                 grid.asHullInset()->numbered(cellrow, true);
1028                 }
1029
1030                 else if (t.cs() == "hline") {
1031                         grid.rowinfo(cellrow).lines_ ++;
1032                 }
1033
1034                 else if (t.cs() == "sqrt") {
1035                         MathArray ar;
1036                         parse(ar, FLAG_OPTION, mode);
1037                         if (ar.size()) {
1038                                 cell->push_back(MathAtom(new MathRootInset));
1039                                 cell->back().nucleus()->cell(0) = ar;
1040                                 parse(cell->back().nucleus()->cell(1), FLAG_ITEM, mode);
1041                         } else {
1042                                 cell->push_back(MathAtom(new MathSqrtInset));
1043                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1044                         }
1045                 }
1046
1047                 else if (t.cs() == "xrightarrow" || t.cs() == "xleftarrow") {
1048                         cell->push_back(createMathInset(t.cs()));
1049                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
1050                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1051                 }
1052
1053                 else if (t.cs() == "ref" || t.cs() == "prettyref" ||
1054                                 t.cs() == "pageref" || t.cs() == "vpageref" || t.cs() == "vref") {
1055                         cell->push_back(MathAtom(new RefInset(t.cs())));
1056                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, mode);
1057                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1058                 }
1059
1060                 else if (t.cs() == "left") {
1061                         skipSpaces();
1062                         Token const & tl = getToken();
1063                         // \| and \Vert are equivalent, and MathDelimInset
1064                         // can't handle \|
1065                         // FIXME: fix this in MathDelimInset itself!
1066                         string const l = tl.cs() == "|" ? "Vert" : tl.asString();
1067                         MathArray ar;
1068                         parse(ar, FLAG_RIGHT, mode);
1069                         skipSpaces();
1070                         Token const & tr = getToken();
1071                         string const r = tr.cs() == "|" ? "Vert" : tr.asString();
1072                         cell->push_back(MathAtom(new MathDelimInset(l, r, ar)));
1073                 }
1074
1075                 else if (t.cs() == "right") {
1076                         if (flags & FLAG_RIGHT)
1077                                 return;
1078                         //lyxerr << "got so far: '" << cell << "'" << endl;
1079                         error("Unmatched right delimiter");
1080                         return;
1081                 }
1082
1083                 else if (t.cs() == "begin") {
1084                         string const name = getArg('{', '}');
1085                         environments_.push_back(name);
1086
1087                         if (name == "array" || name == "subarray") {
1088                                 string const valign = parse_verbatim_option() + 'c';
1089                                 string const halign = parse_verbatim_item();
1090                                 cell->push_back(MathAtom(new MathArrayInset(name, valign[0], halign)));
1091                                 parse2(cell->back(), FLAG_END, mode, false);
1092                         }
1093
1094                         else if (name == "tabular") {
1095                                 string const valign = parse_verbatim_option() + 'c';
1096                                 string const halign = parse_verbatim_item();
1097                                 cell->push_back(MathAtom(new MathTabularInset(name, valign[0], halign)));
1098                                 parse2(cell->back(), FLAG_END, MathInset::TEXT_MODE, false);
1099                         }
1100
1101                         else if (name == "split" || name == "cases" ||
1102                                  name == "gathered" || name == "aligned") {
1103                                 cell->push_back(createMathInset(name));
1104                                 parse2(cell->back(), FLAG_END, mode, false);
1105                         }
1106
1107                         else if (name == "alignedat") {
1108                                 // ignore this for a while
1109                                 getArg('{', '}');
1110                                 cell->push_back(createMathInset(name));
1111                                 parse2(cell->back(), FLAG_END, mode, false);
1112                         }
1113
1114                         else if (name == "math") {
1115                                 cell->push_back(MathAtom(new MathHullInset("simple")));
1116                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, true);
1117                         }
1118
1119                         else if (name == "equation" || name == "equation*"
1120                                         || name == "displaymath") {
1121                                 cell->push_back(MathAtom(new MathHullInset("equation")));
1122                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, (name == "equation"));
1123                         }
1124
1125                         else if (name == "eqnarray" || name == "eqnarray*") {
1126                                 cell->push_back(MathAtom(new MathHullInset("eqnarray")));
1127                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1128                         }
1129
1130                         else if (name == "align" || name == "align*") {
1131                                 cell->push_back(MathAtom(new MathHullInset("align")));
1132                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1133                         }
1134
1135                         else if (name == "flalign" || name == "flalign*") {
1136                                 cell->push_back(MathAtom(new MathHullInset("flalign")));
1137                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1138                         }
1139
1140                         else if (name == "alignat" || name == "alignat*") {
1141                                 // ignore this for a while
1142                                 getArg('{', '}');
1143                                 cell->push_back(MathAtom(new MathHullInset("alignat")));
1144                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1145                         }
1146
1147                         else if (name == "xalignat" || name == "xalignat*") {
1148                                 // ignore this for a while
1149                                 getArg('{', '}');
1150                                 cell->push_back(MathAtom(new MathHullInset("xalignat")));
1151                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1152                         }
1153
1154                         else if (name == "xxalignat") {
1155                                 // ignore this for a while
1156                                 getArg('{', '}');
1157                                 cell->push_back(MathAtom(new MathHullInset("xxalignat")));
1158                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1159                         }
1160
1161                         else if (name == "multline" || name == "multline*") {
1162                                 cell->push_back(MathAtom(new MathHullInset("multline")));
1163                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1164                         }
1165
1166                         else if (name == "gather" || name == "gather*") {
1167                                 cell->push_back(MathAtom(new MathHullInset("gather")));
1168                                 parse2(cell->back(), FLAG_END, MathInset::MATH_MODE, !stared(name));
1169                         }
1170
1171                         else if (latexkeys const * l = in_word_set(name)) {
1172                                 if (l->inset == "matrix") {
1173                                         cell->push_back(createMathInset(name));
1174                                         parse2(cell->back(), FLAG_END, mode, false);
1175                                 }
1176                         }
1177
1178                         else {
1179                                 dump();
1180                                 lyxerr << "found unknown math environment '" << name << "'" << endl;
1181                                 // create generic environment inset
1182                                 cell->push_back(MathAtom(new MathEnvInset(name)));
1183                                 parse(cell->back().nucleus()->cell(0), FLAG_ITEM, mode);
1184                         }
1185                 }
1186
1187                 else if (t.cs() == "kern") {
1188 #ifdef WITH_WARNINGS
1189 #warning A hack...
1190 #endif
1191                         string s;
1192                         while (true) {
1193                                 Token const & t = getToken();
1194                                 if (!good()) {
1195                                         putback();
1196                                         break;
1197                                 }
1198                                 s += t.character();
1199                                 if (isValidLength(s))
1200                                         break;
1201                         }
1202                         cell->push_back(MathAtom(new MathKernInset(s)));
1203                 }
1204
1205                 else if (t.cs() == "label") {
1206                         // FIXME: This is swallowed in inline formulas
1207                         string label = parse_verbatim_item();
1208                         MathArray ar;
1209                         asArray(label, ar);
1210                         if (grid.asHullInset()) {
1211                                 grid.asHullInset()->label(cellrow, label);
1212                         } else {
1213                                 cell->push_back(createMathInset(t.cs()));
1214                                 cell->push_back(MathAtom(new MathBraceInset(ar)));
1215                         }
1216                 }
1217
1218                 else if (t.cs() == "choose" || t.cs() == "over" || t.cs() == "atop") {
1219                         MathAtom at = createMathInset(t.cs());
1220                         at.nucleus()->cell(0) = *cell;
1221                         cell->clear();
1222                         parse(at.nucleus()->cell(1), flags, mode);
1223                         cell->push_back(at);
1224                         return;
1225                 }
1226
1227                 else if (t.cs() == "color") {
1228                         string const color = parse_verbatim_item();
1229                         cell->push_back(MathAtom(new MathColorInset(true, color)));
1230                         parse(cell->back().nucleus()->cell(0), flags, mode);
1231                         return;
1232                 }
1233
1234                 else if (t.cs() == "textcolor") {
1235                         string const color = parse_verbatim_item();
1236                         cell->push_back(MathAtom(new MathColorInset(false, color)));
1237                         parse(cell->back().nucleus()->cell(0), FLAG_ITEM, MathInset::TEXT_MODE);
1238                 }
1239
1240                 else if (t.cs() == "normalcolor") {
1241                         cell->push_back(createMathInset(t.cs()));
1242                         parse(cell->back().nucleus()->cell(0), flags, mode);
1243                         return;
1244                 }
1245
1246                 else if (t.cs() == "substack") {
1247                         cell->push_back(createMathInset(t.cs()));
1248                         parse2(cell->back(), FLAG_ITEM, mode, false);
1249                 }
1250
1251                 else if (t.cs() == "framebox" || t.cs() == "makebox") {
1252                         cell->push_back(createMathInset(t.cs()));
1253                         parse(cell->back().nucleus()->cell(0), FLAG_OPTION, MathInset::TEXT_MODE);
1254                         parse(cell->back().nucleus()->cell(1), FLAG_OPTION, MathInset::TEXT_MODE);
1255                         parse(cell->back().nucleus()->cell(2), FLAG_ITEM, MathInset::TEXT_MODE);
1256                 }
1257
1258 #if 0
1259                 else if (t.cs() == "infer") {
1260                         MathArray ar;
1261                         parse(ar, FLAG_OPTION, mode);
1262                         cell->push_back(createMathInset(t.cs()));
1263                         parse2(cell->back(), FLAG_ITEM, mode, false);
1264                 }
1265
1266                 // Disabled
1267                 else if (1 && t.cs() == "ar") {
1268                         auto_ptr<MathXYArrowInset> p(new MathXYArrowInset);
1269                         // try to read target
1270                         parse(p->cell(0), FLAG_OTPTION, mode);
1271                         // try to read label
1272                         if (nextToken().cat() == catSuper || nextToken().cat() == catSub) {
1273                                 p->up_ = nextToken().cat() == catSuper;
1274                                 getToken();
1275                                 parse(p->cell(1), FLAG_ITEM, mode);
1276                                 //lyxerr << "read label: " << p->cell(1) << endl;
1277                         }
1278
1279                         cell->push_back(MathAtom(p.release()));
1280                         //lyxerr << "read cell: " << cell << endl;
1281                 }
1282 #endif
1283
1284                 else if (t.cs().size()) {
1285                         latexkeys const * l = in_word_set(t.cs());
1286                         if (l) {
1287                                 if (l->inset == "font") {
1288                                         cell->push_back(createMathInset(t.cs()));
1289                                         parse(cell->back().nucleus()->cell(0),
1290                                                 FLAG_ITEM, asMode(mode, l->extra));
1291                                 }
1292
1293                                 else if (l->inset == "oldfont") {
1294                                         cell->push_back(createMathInset(t.cs()));
1295                                         parse(cell->back().nucleus()->cell(0),
1296                                                 flags | FLAG_ALIGN, asMode(mode, l->extra));
1297                                         if (prevToken().cat() != catAlign &&
1298                                             prevToken().cs() != "\\")
1299                                                 return;
1300                                         putback();
1301                                 }
1302
1303                                 else if (l->inset == "style") {
1304                                         cell->push_back(createMathInset(t.cs()));
1305                                         parse(cell->back().nucleus()->cell(0),
1306                                                 flags | FLAG_ALIGN, mode);
1307                                         if (prevToken().cat() != catAlign &&
1308                                             prevToken().cs() != "\\")
1309                                                 return;
1310                                         putback();
1311                                 }
1312
1313                                 else {
1314                                         MathAtom at = createMathInset(t.cs());
1315                                         for (MathInset::idx_type i = 0; i < at->nargs(); ++i)
1316                                                 parse(at.nucleus()->cell(i),
1317                                                         FLAG_ITEM, asMode(mode, l->extra));
1318                                         cell->push_back(at);
1319                                 }
1320                         }
1321
1322                         else {
1323                                 MathAtom at = createMathInset(t.cs());
1324                                 MathInset::mode_type m = mode;
1325                                 //if (m == MathInset::UNDECIDED_MODE)
1326                                 //lyxerr << "default creation: m1: " << m << endl;
1327                                 if (at->currentMode() != MathInset::UNDECIDED_MODE)
1328                                         m = at->currentMode();
1329                                 //lyxerr << "default creation: m2: " << m << endl;
1330                                 MathInset::idx_type start = 0;
1331                                 // this fails on \bigg[...\bigg]
1332                                 //MathArray opt;
1333                                 //parse(opt, FLAG_OPTION, MathInset::VERBATIM_MODE);
1334                                 //if (opt.size()) {
1335                                 //      start = 1;
1336                                 //      at.nucleus()->cell(0) = opt;
1337                                 //}
1338                                 for (MathInset::idx_type i = start; i < at->nargs(); ++i) {
1339                                         parse(at.nucleus()->cell(i), FLAG_ITEM, m);
1340                                         skipSpaces();
1341                                 }
1342                                 cell->push_back(at);
1343                         }
1344                 }
1345
1346
1347                 if (flags & FLAG_LEAVE) {
1348                         flags &= ~FLAG_LEAVE;
1349                         break;
1350                 }
1351         }
1352 }
1353
1354
1355
1356 } // anonymous namespace
1357
1358
1359 void mathed_parse_cell(MathArray & ar, string const & str)
1360 {
1361         istringstream is(str);
1362         mathed_parse_cell(ar, is);
1363 }
1364
1365
1366 void mathed_parse_cell(MathArray & ar, istream & is)
1367 {
1368         Parser(is).parse(ar, 0, MathInset::MATH_MODE);
1369 }
1370
1371
1372 bool mathed_parse_normal(MathAtom & t, string const & str)
1373 {
1374         istringstream is(str);
1375         return Parser(is).parse(t);
1376 }
1377
1378
1379 bool mathed_parse_normal(MathAtom & t, istream & is)
1380 {
1381         return Parser(is).parse(t);
1382 }
1383
1384
1385 bool mathed_parse_normal(MathAtom & t, LyXLex & lex)
1386 {
1387         return Parser(lex).parse(t);
1388 }
1389
1390
1391 void mathed_parse_normal(MathGridInset & grid, string const & str)
1392 {
1393         istringstream is(str);
1394         Parser(is).parse1(grid, 0, MathInset::MATH_MODE, false);
1395 }
1396
1397
1398 void initParser()
1399 {
1400         fill(theCatcode, theCatcode + 256, catOther);
1401         fill(theCatcode + 'a', theCatcode + 'z' + 1, catLetter);
1402         fill(theCatcode + 'A', theCatcode + 'Z' + 1, catLetter);
1403
1404         theCatcode[int('\\')] = catEscape;
1405         theCatcode[int('{')]  = catBegin;
1406         theCatcode[int('}')]  = catEnd;
1407         theCatcode[int('$')]  = catMath;
1408         theCatcode[int('&')]  = catAlign;
1409         theCatcode[int('\n')] = catNewline;
1410         theCatcode[int('#')]  = catParameter;
1411         theCatcode[int('^')]  = catSuper;
1412         theCatcode[int('_')]  = catSub;
1413         theCatcode[int(0x7f)] = catIgnore;
1414         theCatcode[int(' ')]  = catSpace;
1415         theCatcode[int('\t')] = catSpace;
1416         theCatcode[int('\r')] = catNewline;
1417         theCatcode[int('~')]  = catActive;
1418         theCatcode[int('%')]  = catComment;
1419 }