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