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