src/mathed/math_parser.C

   1 /*
   2  *  File:        math_parser.C
   3  *  Purpose:     Parser for mathed
   4  *  Author:      Alejandro Aguilar Sierra <asierra@servidor.unam.mx>
   5  *  Created:     January 1996
   6  *  Description: Parse LaTeX2e math mode code.
   7  *
   8  *  Dependencies: Xlib, XForms
   9  *
  10  *  Copyright: 1996, Alejandro Aguilar Sierra
  11  *
  12  *   Version: 0.8beta.
  13  *
  14  *   You are free to use and modify this code under the terms of
  15  *   the GNU General Public Licence version 2 or later.
  16  */
  17
  18 /*
  19
  20 If someone desperately needs partial "structures" (such as a few cells of
  21 an array inset or similar) (s)he could uses the following hack as starting
  22 point to write some macros:
  23
  24   \newif\ifcomment
  25   \commentfalse
  26   \ifcomment
  27           \def\makeamptab{\catcode`\&=4\relax}
  28           \def\makeampletter{\catcode`\&=11\relax}
  29     \def\b{\makeampletter\expandafter\makeamptab\bi}
  30     \long\def\bi#1\e{}
  31   \else
  32     \def\b{}\def\e{}
  33   \fi
  34
  35   ...
  36
  37   \[\begin{array}{ccc}
  38    1 & 2\b & 3^2\\
  39    4 & 5\e & 6\\
  40    7 & 8 & 9
  41   \end{array}\]
  42
  43 */
  44
  45
  46 #include <config.h>
  47
  48 #ifdef __GNUG__
  49 #pragma implementation
  50 #endif
  51
  52 #include "math_parser.h"
  53 #include "math_inset.h"
  54 #include "math_arrayinset.h"
  55 #include "math_braceinset.h"
  56 #include "math_boxinset.h"
  57 #include "math_charinset.h"
  58 #include "math_deliminset.h"
  59 #include "math_factory.h"
  60 #include "math_funcinset.h"
  61 #include "math_kerninset.h"
  62 #include "math_macro.h"
  63 #include "math_macrotable.h"
  64 #include "math_macrotemplate.h"
  65 #include "math_hullinset.h"
  66 #include "math_rootinset.h"
  67 #include "math_sizeinset.h"
  68 #include "math_sqrtinset.h"
  69 #include "math_scriptinset.h"
  70 #include "math_specialcharinset.h"
  71 #include "math_sqrtinset.h"
  72 #include "math_support.h"
  73 #include "math_xyarrowinset.h"
  74
  75 #include "lyxlex.h"
  76 #include "debug.h"
  77 #include "support/LAssert.h"
  78 #include "support/lstrings.h"
  79
  80 #include <cctype>
  81 #include <stack>
  82 #include <algorithm>
  83
  84 using std::istream;
  85 using std::ostream;
  86 using std::ios;
  87 using std::endl;
  88 using std::stack;
  89 using std::fill;
  90
  91 //#define FILEDEBUG
  92
  93
  94 namespace {
  95
  96 bool stared(string const & s)
  97 {
  98         string::size_type const n = s.size();
  99         return n && s[n - 1] == '*';
 100 }
 101
 102
 103 void add(MathArray & ar, char c, MathTextCodes code)
 104 {
 105         ar.push_back(MathAtom(new MathCharInset(c, code)));
 106 }
 107
 108
 109 // These are TeX's catcodes
 110 enum CatCode {
 111         catEscape,     // 0    backslash
 112         catBegin,      // 1    {
 113         catEnd,        // 2    }
 114         catMath,       // 3    $
 115         catAlign,      // 4    &
 116         catNewline,    // 5    ^^M
 117         catParameter,  // 6    #
 118         catSuper,      // 7    ^
 119         catSub,        // 8    _
 120         catIgnore,     // 9
 121         catSpace,      // 10   space
 122         catLetter,     // 11   a-zA-Z
 123         catOther,      // 12   none of the above
 124         catActive,     // 13   ~
 125         catComment,    // 14   %
 126         catInvalid     // 15   <delete>
 127 };
 128
 129 CatCode theCatcode[256];
 130
 131
 132 inline CatCode catcode(unsigned char c)
 133 {
 134         return theCatcode[c];
 135 }
 136
 137
 138 enum {
 139         FLAG_BRACE_LAST = 1 << 1,  //  last closing brace ends the parsing process
 140         FLAG_RIGHT      = 1 << 2,  //  next \\right ends the parsing process
 141         FLAG_END        = 1 << 3,  //  next \\end ends the parsing process
 142         FLAG_BRACK_END  = 1 << 4,  //  next closing bracket ends the parsing process
 143         FLAG_BOX        = 1 << 5,  //  we are in a box
 144         FLAG_ITEM       = 1 << 6,  //  read a (possibly braced token)
 145         FLAG_BLOCK      = 1 << 7,  //  next block ends the parsing process
 146         FLAG_BLOCK2     = 1 << 8,  //  next block2 ends the parsing process
 147         FLAG_LEAVE      = 1 << 9   //  leave the loop at the end
 148 };
 149
 150
 151 void catInit()
 152 {
 153         fill(theCatcode, theCatcode + 256, catOther);
 154         fill(theCatcode + 'a', theCatcode + 'z' + 1, catLetter);
 155         fill(theCatcode + 'A', theCatcode + 'Z' + 1, catLetter);
 156
 157         theCatcode['\\'] = catEscape;
 158         theCatcode['{']  = catBegin;
 159         theCatcode['}']  = catEnd;
 160         theCatcode['$']  = catMath;
 161         theCatcode['&']  = catAlign;
 162         theCatcode['\n'] = catNewline;
 163         theCatcode['#']  = catParameter;
 164         theCatcode['^']  = catSuper;
 165         theCatcode['_']  = catSub;
 166         theCatcode['\7f'] = catIgnore;
 167         theCatcode[' ']  = catSpace;
 168         theCatcode['\t'] = catSpace;
 169         theCatcode['\r'] = catSpace;
 170         theCatcode['~']  = catActive;
 171         theCatcode['%']  = catComment;
 172 }
 173
 174
 175
 176 //
 177 // Helper class for parsing
 178 //
 179
 180 class Token {
 181 public:
 182         ///
 183         Token() : cs_(), char_(0), cat_(catIgnore) {}
 184         ///
 185         Token(char c, CatCode cat) : cs_(), char_(c), cat_(cat) {}
 186         ///
 187         Token(string const & cs) : cs_(cs), char_(0), cat_(catIgnore) {}
 188
 189         ///
 190         string const & cs() const { return cs_; }
 191         ///
 192         CatCode cat() const { return cat_; }
 193         ///
 194         char character() const { return char_; }
 195         ///
 196         string asString() const;
 197         ///
 198         bool isCR() const;
 199
 200 private:
 201         ///
 202         string cs_;
 203         ///
 204         char char_;
 205         ///
 206         CatCode cat_;
 207 };
 208
 209 bool Token::isCR() const
 210 {
 211         return cs_ == "\\" || cs_ == "cr" || cs_ == "crcr";
 212 }
 213
 214 string Token::asString() const
 215 {
 216         return cs_.size() ? cs_ : string(1, char_);
 217 }
 218
 219 // Angus' compiler says these are not needed
 220 //bool operator==(Token const & s, Token const & t)
 221 //{
 222 //      return s.character() == t.character()
 223 //              && s.cat() == t.cat() && s.cs() == t.cs();
 224 //}
 225 //
 226 //bool operator!=(Token const & s, Token const & t)
 227 //{
 228 //      return !(s == t);
 229 //}
 230
 231 ostream & operator<<(ostream & os, Token const & t)
 232 {
 233         if (t.cs().size())
 234                 os << "\\" << t.cs();
 235         else
 236                 os << "[" << t.character() << "," << t.cat() << "]";
 237         return os;
 238 }
 239
 240
 241 class Parser {
 242
 243 public:
 244         ///
 245         Parser(LyXLex & lex);
 246         ///
 247         Parser(istream & is);
 248
 249         ///
 250         bool parse_macro(string & name);
 251         ///
 252         bool parse_normal(MathAtom &);
 253         ///
 254         void parse_into(MathArray & array, unsigned flags, MathTextCodes = LM_TC_MIN);
 255         ///
 256         int lineno() const { return lineno_; }
 257         ///
 258         void putback();
 259
 260 private:
 261         ///
 262         void parse_into1(MathArray & array, unsigned flags, MathTextCodes);
 263         ///
 264         string getArg(char lf, char rf);
 265         ///
 266         char getChar();
 267         ///
 268         void error(string const & msg);
 269         ///
 270         bool parse_lines(MathAtom & t, bool numbered, bool outmost);
 271         /// parses {... & ... \\ ... & ... }
 272         bool parse_lines2(MathAtom & t, bool braced);
 273         /// dump contents to screen
 274         void dump() const;
 275
 276 private:
 277         ///
 278         void tokenize(istream & is);
 279         ///
 280         void tokenize(string const & s);
 281         ///
 282         void skipSpaceTokens(istream & is, char c);
 283         ///
 284         void push_back(Token const & t);
 285         ///
 286         void pop_back();
 287         ///
 288         Token const & prevToken() const;
 289         ///
 290         Token const & nextToken() const;
 291         ///
 292         Token const & getToken();
 293         /// skips spaces if any
 294         void skipSpaces();
 295         /// skips opening brace
 296         void skipBegin();
 297         /// skips closing brace
 298         void skipEnd();
 299         /// counts a sequence of hlines
 300         int readHLines();
 301         ///
 302         void lex(string const & s);
 303         ///
 304         bool good() const;
 305
 306         ///
 307         int lineno_;
 308         ///
 309         std::vector<Token> tokens_;
 310         ///
 311         unsigned pos_;
 312         ///
 313         bool   curr_num_;
 314         ///
 315         string curr_label_;
 316         ///
 317         string curr_skip_;
 318 };
 319
 320
 321 Parser::Parser(LyXLex & lexer)
 322         : lineno_(lexer.getLineNo()), pos_(0), curr_num_(false)
 323 {
 324         tokenize(lexer.getStream());
 325         lexer.eatLine();
 326 }
 327
 328
 329 Parser::Parser(istream & is)
 330         : lineno_(0), pos_(0), curr_num_(false)
 331 {
 332         tokenize(is);
 333 }
 334
 335
 336 void Parser::push_back(Token const & t)
 337 {
 338         tokens_.push_back(t);
 339 }
 340
 341
 342 void Parser::pop_back()
 343 {
 344         tokens_.pop_back();
 345 }
 346
 347
 348 Token const & Parser::prevToken() const
 349 {
 350         static const Token dummy;
 351         return pos_ > 0 ? tokens_[pos_ - 1] : dummy;
 352 }
 353
 354
 355 Token const & Parser::nextToken() const
 356 {
 357         static const Token dummy;
 358         return good() ? tokens_[pos_] : dummy;
 359 }
 360
 361
 362 Token const & Parser::getToken()
 363 {
 364         static const Token dummy;
 365         //lyxerr << "looking at token " << tokens_[pos_] << " pos: " << pos_ << '\n';
 366         return good() ? tokens_[pos_++] : dummy;
 367 }
 368
 369
 370 void Parser::skipSpaces()
 371 {
 372         while (nextToken().cat() == catSpace)
 373                 getToken();
 374 }
 375
 376
 377 void Parser::skipBegin()
 378 {
 379         if (nextToken().cat() == catBegin)
 380                 getToken();
 381         else
 382                 lyxerr << "'{' expected\n";
 383 }
 384
 385
 386 void Parser::skipEnd()
 387 {
 388         if (nextToken().cat() == catEnd)
 389                 getToken();
 390         else
 391                 lyxerr << "'}' expected\n";
 392 }
 393
 394
 395 int Parser::readHLines()
 396 {
 397         int num = 0;
 398         skipSpaces();
 399         while (nextToken().cs() == "hline") {
 400                 getToken();
 401                 ++num;
 402                 skipSpaces();
 403         }
 404         return num;
 405 }
 406
 407
 408 void Parser::putback()
 409 {
 410         --pos_;
 411 }
 412
 413
 414 bool Parser::good() const
 415 {
 416         return pos_ < tokens_.size();
 417 }
 418
 419
 420 char Parser::getChar()
 421 {
 422         if (!good())
 423                 lyxerr << "The input stream is not well..." << endl;
 424         return tokens_[pos_++].character();
 425 }
 426
 427
 428 string Parser::getArg(char lf, char rg)
 429 {
 430         skipSpaces();
 431
 432         string result;
 433         char c = getChar();
 434
 435         if (c != lf)
 436                 putback();
 437         else
 438                 while ((c = getChar()) != rg && good())
 439                         result += c;
 440
 441         return result;
 442 }
 443
 444
 445 void Parser::tokenize(istream & is)
 446 {
 447         // eat everything up to the next \end_inset or end of stream
 448         // and store it in s for further tokenization
 449         string s;
 450         char c;
 451         while (is.get(c)) {
 452                 s += c;
 453                 if (s.size() >= 10 && s.substr(s.size() - 10) == "\\end_inset") {
 454                         s = s.substr(0, s.size() - 10);
 455                         break;
 456                 }
 457         }
 458
 459         // tokenize buffer
 460         tokenize(s);
 461 }
 462
 463
 464 void Parser::skipSpaceTokens(istream & is, char c)
 465 {
 466         // skip trailing spaces
 467         while (catcode(c) == catSpace || catcode(c) == catNewline)
 468                 if (!is.get(c))
 469                         break;
 470         //lyxerr << "putting back: " << c << "\n";
 471         is.putback(c);
 472 }
 473
 474
 475 void Parser::tokenize(string const & buffer)
 476 {
 477         static bool init_done = false;
 478
 479         if (!init_done) {
 480                 catInit();
 481                 init_done = true;
 482         }
 483
 484         istringstream is(buffer.c_str(), ios::in | ios::binary);
 485
 486         char c;
 487         while (is.get(c)) {
 488                 //lyxerr << "reading c: " << c << "\n";
 489
 490                 switch (catcode(c)) {
 491                         case catNewline: {
 492                                 ++lineno_;
 493                                 is.get(c);
 494                                 if (catcode(c) == catNewline)
 495                                         ; //push_back(Token("par"));
 496                                 else {
 497                                         push_back(Token(' ', catSpace));
 498                                         is.putback(c);
 499                                 }
 500                                 break;
 501                         }
 502
 503                         case catComment: {
 504                                 while (is.get(c) && catcode(c) != catNewline)
 505                                         ;
 506                                 ++lineno_;
 507                                 break;
 508                         }
 509
 510                         case catEscape: {
 511                                 is.get(c);
 512                                 if (!is) {
 513                                         error("unexpected end of input");
 514                                 } else {
 515                                         string s(1, c);
 516                                         if (catcode(c) == catLetter) {
 517                                                 // collect letters
 518                                                 while (is.get(c) && catcode(c) == catLetter)
 519                                                         s += c;
 520                                                 skipSpaceTokens(is, c);
 521                                         }
 522                                         push_back(Token(s));
 523                                 }
 524                                 break;
 525                         }
 526
 527                         case catSuper:
 528                         case catSub: {
 529                                 push_back(Token(c, catcode(c)));
 530                                 is.get(c);
 531                                 skipSpaceTokens(is, c);
 532                                 break;
 533                         }
 534
 535                         case catIgnore: {
 536                                 lyxerr << "ignoring a char: " << int(c) << "\n";
 537                                 break;
 538                         }
 539
 540                         default:
 541                                 push_back(Token(c, catcode(c)));
 542                 }
 543         }
 544
 545 #ifdef FILEDEBUG
 546         dump();
 547 #endif
 548 }
 549
 550
 551 void Parser::dump() const
 552 {
 553         lyxerr << "\nTokens: ";
 554         for (unsigned i = 0; i < tokens_.size(); ++i)
 555                 lyxerr << tokens_[i];
 556         lyxerr << "\n";
 557 }
 558
 559
 560 void Parser::error(string const & msg)
 561 {
 562         lyxerr << "Line ~" << lineno_ << ": Math parse error: " << msg << endl;
 563         dump();
 564         //exit(1);
 565 }
 566
 567
 568
 569 bool Parser::parse_lines(MathAtom & t, bool numbered, bool outmost)
 570 {
 571         MathGridInset * p = t->asGridInset();
 572         if (!p) {
 573                 dump();
 574                 lyxerr << "error in Parser::parse_lines() 1\n";
 575                 return false;
 576         }
 577
 578         // save global variables
 579         bool   const saved_num   = curr_num_;
 580         string const saved_label = curr_label_;
 581
 582         // read initial hlines
 583         p->rowinfo(0).lines_ = readHLines();
 584
 585         for (int row = 0; true; ++row) {
 586                 // reset global variables
 587                 curr_num_   = numbered;
 588                 curr_label_.erase();
 589
 590                 // reading a row
 591                 for (MathInset::col_type col = 0; col < p->ncols(); ++col) {
 592                         //lyxerr << "reading cell " << row << " " << col << "\n";
 593                         //lyxerr << "ncols: " << p->ncols() << "\n";
 594
 595                         MathArray & ar = p->cell(col + row * p->ncols());
 596                         parse_into(ar, FLAG_BLOCK);
 597                         // remove 'unnecessary' braces:
 598                         if (ar.size() == 1 && ar.back()->asBraceInset())
 599                                 ar = ar.back()->asBraceInset()->cell(0);
 600                         //lyxerr << "ar: " << ar << "\n";
 601
 602                         // break if cell is not followed by an ampersand
 603                         if (nextToken().cat() != catAlign) {
 604                                 //lyxerr << "less cells read than normal in row/col: "
 605                                 //      << row << " " << col << "\n";
 606                                 break;
 607                         }
 608
 609                         // skip the ampersand
 610                         getToken();
 611                 }
 612
 613                 if (outmost) {
 614                         MathHullInset * m = t->asHullInset();
 615                         if (!m) {
 616                                 lyxerr << "error in Parser::parse_lines() 2\n";
 617                                 return false;
 618                         }
 619                         m->numbered(row, curr_num_);
 620                         m->label(row, curr_label_);
 621                         if (curr_skip_.size()) {
 622                                 m->vcrskip(LyXLength(curr_skip_), row);
 623                                 curr_skip_.erase();
 624                         }
 625                 }
 626
 627                 // is a \\ coming?
 628                 if (nextToken().isCR()) {
 629                         // skip the cr-token
 630                         getToken();
 631
 632                         // try to read a length
 633                         //get
 634
 635                         // read hlines for next row
 636                         p->rowinfo(row + 1).lines_ = readHLines();
 637                 }
 638
 639                 // we are finished if the next token is an 'end'
 640                 if (nextToken().cs() == "end") {
 641                         // skip the end-token
 642                         getToken();
 643                         getArg('{','}');
 644
 645                         // leave the 'read a line'-loop
 646                         break;
 647                 }
 648
 649                 // otherwise, we have to start a new row
 650                 p->appendRow();
 651         }
 652
 653         // restore "global" variables
 654         curr_num_   = saved_num;
 655         curr_label_ = saved_label;
 656
 657         return true;
 658 }
 659
 660
 661 bool Parser::parse_lines2(MathAtom & t, bool braced)
 662 {
 663         MathGridInset * p = t->asGridInset();
 664         if (!p) {
 665                 lyxerr << "error in Parser::parse_lines() 1\n";
 666                 return false;
 667         }
 668
 669         for (int row = 0; true; ++row) {
 670                 // reading a row
 671                 for (MathInset::col_type col = 0; true; ++col) {
 672                         //lyxerr << "reading cell " << row << " " << col << " " << p->ncols() << "\n";
 673
 674                         if (col >= p->ncols()) {
 675                                 //lyxerr << "adding col " << col << "\n";
 676                                 p->addCol(p->ncols());
 677                         }
 678
 679                         parse_into(p->cell(col + row * p->ncols()), FLAG_BLOCK2);
 680                         //lyxerr << "read cell: " << p->cell(col + row * p->ncols()) << "\n";
 681
 682                         // break if cell is not followed by an ampersand
 683                         if (nextToken().cat() != catAlign) {
 684                                 //lyxerr << "less cells read than normal in row/col: " << row << " " << col << "\n";
 685                                 break;
 686                         }
 687
 688                         // skip the ampersand
 689                         getToken();
 690                 }
 691
 692                 // is a \\ coming?
 693                 if (nextToken().isCR()) {
 694                         // skip the cr-token
 695                         getToken();
 696                 }
 697
 698                 // we are finished if the next token is the one we expected
 699                 // skip the end-token
 700                 // leave the 'read a line'-loop
 701                 if (braced) {
 702                         if (nextToken().cat() == catEnd) {
 703                                 getToken();
 704                                 break;
 705                         }
 706                 } else {
 707                         if (nextToken().cs() == "end") {
 708                                 getToken();
 709                                 getArg('{','}');
 710                                 break;
 711                         }
 712                 }
 713
 714                 // otherwise, we have to start a new row
 715                 p->appendRow();
 716         }
 717
 718         return true;
 719 }
 720
 721
 722
 723 bool Parser::parse_macro(string & name)
 724 {
 725         name = "{error}";
 726         skipSpaces();
 727
 728         if (getToken().cs() != "newcommand") {
 729                 lyxerr << "\\newcommand expected\n";
 730                 return false;
 731         }
 732
 733         if (getToken().cat() != catBegin) {
 734                 lyxerr << "'{' in \\newcommand expected (1)\n";
 735                 return false;
 736         }
 737
 738         name = getToken().cs();
 739
 740         if (getToken().cat() != catEnd) {
 741                 lyxerr << "'}' expected\n";
 742                 return false;
 743         }
 744
 745         string    arg  = getArg('[', ']');
 746         int       narg = arg.empty() ? 0 : atoi(arg.c_str());
 747
 748         if (getToken().cat() != catBegin) {
 749                 lyxerr << "'{' in \\newcommand expected (2)\n";
 750                 return false;
 751         }
 752
 753         MathArray ar;
 754         parse_into(ar, FLAG_BRACE_LAST);
 755
 756         // we cannot handle recursive stuff at all
 757         MathArray test;
 758         test.push_back(createMathInset(name));
 759         if (ar.contains(test)) {
 760                 lyxerr << "we cannot handle recursive macros at all.\n";
 761                 return false;
 762         }
 763
 764         MathMacroTable::create(name, narg, ar);
 765         return true;
 766 }
 767
 768
 769 bool Parser::parse_normal(MathAtom & matrix)
 770 {
 771         skipSpaces();
 772         Token const & t = getToken();
 773
 774         if (t.cs() == "(") {
 775                 matrix = MathAtom(new MathHullInset(LM_OT_SIMPLE));
 776                 parse_into(matrix->cell(0), 0);
 777                 return true;
 778         }
 779
 780         if (t.cat() == catMath) {
 781                 Token const & n = getToken();
 782                 if (n.cat() == catMath) {
 783                         // TeX's $$...$$ syntax for displayed math
 784                         matrix = MathAtom(new MathHullInset(LM_OT_EQUATION));
 785                         MathHullInset * p = matrix->asHullInset();
 786                         parse_into(p->cell(0), 0);
 787                         p->numbered(0, curr_num_);
 788                         p->label(0, curr_label_);
 789                 } else {
 790                         // simple $...$  stuff
 791                         putback();
 792                         matrix = MathAtom(new MathHullInset(LM_OT_SIMPLE));
 793                         parse_into(matrix->cell(0), 0);
 794                 }
 795                 return true;
 796         }
 797
 798         if (!t.cs().size()) {
 799                 lyxerr << "start of math expected, got '" << t << "'\n";
 800                 return false;
 801         }
 802
 803         string const & cs = t.cs();
 804
 805         if (cs == "[") {
 806                 curr_num_ = 0;
 807                 curr_label_.erase();
 808                 matrix = MathAtom(new MathHullInset(LM_OT_EQUATION));
 809                 MathHullInset * p = matrix->asHullInset();
 810                 parse_into(p->cell(0), 0);
 811                 p->numbered(0, curr_num_);
 812                 p->label(0, curr_label_);
 813                 return true;
 814         }
 815
 816         if (cs != "begin") {
 817                 lyxerr << "'begin' of un-simple math expected, got '" << cs << "'\n";
 818                 return false;
 819         }
 820
 821         string const name = getArg('{', '}');
 822
 823         if (name == "math") {
 824                 matrix = MathAtom(new MathHullInset(LM_OT_SIMPLE));
 825                 parse_into(matrix->cell(0), 0);
 826                 return true;
 827         }
 828
 829         if (name == "equation" || name == "equation*" || name == "displaymath") {
 830                 curr_num_ = (name == "equation");
 831                 curr_label_.erase();
 832                 matrix = MathAtom(new MathHullInset(LM_OT_EQUATION));
 833                 MathHullInset * p = matrix->asHullInset();
 834                 parse_into(p->cell(0), FLAG_END);
 835                 p->numbered(0, curr_num_);
 836                 p->label(0, curr_label_);
 837                 return true;
 838         }
 839
 840         if (name == "eqnarray" || name == "eqnarray*") {
 841                 matrix = MathAtom(new MathHullInset(LM_OT_EQNARRAY));
 842                 return parse_lines(matrix, !stared(name), true);
 843         }
 844
 845         if (name == "align" || name == "align*") {
 846                 matrix = MathAtom(new MathHullInset(LM_OT_ALIGN));
 847                 return parse_lines(matrix, !stared(name), true);
 848         }
 849
 850         if (name == "alignat" || name == "alignat*") {
 851                 int nc = 2 * atoi(getArg('{', '}').c_str());
 852                 matrix = MathAtom(new MathHullInset(LM_OT_ALIGNAT, nc));
 853                 return parse_lines(matrix, !stared(name), true);
 854         }
 855
 856         if (name == "xalignat" || name == "xalignat*") {
 857                 int nc = 2 * atoi(getArg('{', '}').c_str());
 858                 matrix = MathAtom(new MathHullInset(LM_OT_XALIGNAT, nc));
 859                 return parse_lines(matrix, !stared(name), true);
 860         }
 861
 862         if (name == "xxalignat") {
 863                 int nc = 2 * atoi(getArg('{', '}').c_str());
 864                 matrix = MathAtom(new MathHullInset(LM_OT_XXALIGNAT, nc));
 865                 return parse_lines(matrix, !stared(name), true);
 866         }
 867
 868         if (name == "multline" || name == "multline*") {
 869                 matrix = MathAtom(new MathHullInset(LM_OT_MULTLINE));
 870                 return parse_lines(matrix, !stared(name), true);
 871         }
 872
 873         if (name == "gather" || name == "gather*") {
 874                 matrix = MathAtom(new MathHullInset(LM_OT_GATHER));
 875                 return parse_lines(matrix, !stared(name), true);
 876         }
 877
 878         lyxerr[Debug::MATHED] << "1: unknown math environment: " << name << "\n";
 879         lyxerr << "1: unknown math environment: " << name << "\n";
 880         return false;
 881 }
 882
 883
 884 void Parser::parse_into(MathArray & array, unsigned flags, MathTextCodes code)
 885 {
 886         parse_into1(array, flags, code);
 887         // remove 'unnecessary' braces:
 888         if (array.size() == 1 && array.back()->asBraceInset()) {
 889                 lyxerr << "extra braces removed\n";
 890                 array = array.back()->asBraceInset()->cell(0);
 891         }
 892 }
 893
 894
 895 void Parser::parse_into1(MathArray & array, unsigned flags, MathTextCodes code)
 896 {
 897         bool panic  = false;
 898         int  limits = 0;
 899
 900         while (good()) {
 901                 Token const & t = getToken();
 902
 903 #ifdef FILEDEBUG
 904                 lyxerr << "t: " << t << " flags: " << flags << "\n";
 905                 //array.dump();
 906                 lyxerr << "\n";
 907 #endif
 908
 909                 if (flags & FLAG_ITEM) {
 910                         flags &= ~FLAG_ITEM;
 911                         if (t.cat() == catBegin) {
 912                                 // skip the brace and collect everything to the next matching
 913                                 // closing brace
 914                                 flags |= FLAG_BRACE_LAST;
 915                                 continue;
 916                         } else {
 917                                 // handle only this single token, leave the loop if done
 918                                 flags |= FLAG_LEAVE;
 919                         }
 920                 }
 921
 922                 if (flags & FLAG_BLOCK) {
 923                         if (t.cat() == catAlign || t.isCR() || t.cs() == "end") {
 924                                 putback();
 925                                 return;
 926                         }
 927                 }
 928
 929                 if (flags & FLAG_BLOCK2) {
 930                         if (t.cat() == catAlign || t.isCR() || t.cs() == "end"
 931                                         || t.cat() == catEnd) {
 932                                 putback();
 933                                 return;
 934                         }
 935                 }
 936
 937                 //
 938                 // cat codes
 939                 //
 940                 if (t.cat() == catMath) {
 941                         if (flags & FLAG_BOX) {
 942                                 // we are inside an mbox, so opening new math is allowed
 943                                 array.push_back(MathAtom(new MathHullInset(LM_OT_SIMPLE)));
 944                                 parse_into(array.back()->cell(0), 0);
 945                         } else {
 946                                 // otherwise this is the end of the formula
 947                                 break;
 948                         }
 949                 }
 950
 951                 else if (t.cat() == catLetter)
 952                         add(array, t.character(), code);
 953
 954                 else if (t.cat() == catSpace && code == LM_TC_TEXTRM)
 955                         add(array, t.character(), code);
 956
 957                 else if (t.cat() == catParameter) {
 958                         Token const & n = getToken();
 959                         array.push_back(MathAtom(new MathMacroArgument(n.character()-'0', code)));
 960                 }
 961
 962                 else if (t.cat() == catBegin) {
 963                         MathArray ar;
 964                         parse_into(ar, FLAG_BRACE_LAST);
 965 #ifndef WITH_WARNINGS
 966 #warning this might be wrong in general!
 967 #endif
 968                         // ignore braces around simple items
 969                         if ((ar.size() == 1 && !ar.front()->needsBraces()
 970        || (ar.size() == 2 && !ar.front()->needsBraces()
 971                                             && ar.back()->asScriptInset()))
 972        || (ar.size() == 0 && array.size() == 0))
 973                         {
 974                                 array.push_back(ar);
 975                         } else {
 976                                 array.push_back(MathAtom(new MathBraceInset));
 977                                 array.back()->cell(0).swap(ar);
 978                         }
 979                 }
 980
 981                 else if (t.cat() == catEnd) {
 982                         if (flags & FLAG_BRACE_LAST)
 983                                 return;
 984                         lyxerr << "found '}' unexpectedly, array: '" << array << "'\n";
 985                         //lyxerr << "found '}' unexpectedly\n";
 986                         lyx::Assert(0);
 987                         add(array, '}', LM_TC_TEX);
 988                 }
 989
 990                 else if (t.cat() == catAlign) {
 991                         lyxerr << "found tab unexpectedly, array: '" << array << "'\n";
 992                         //lyxerr << "found tab unexpectedly\n";
 993                         add(array, '&', LM_TC_TEX);
 994                 }
 995
 996                 else if (t.cat() == catSuper || t.cat() == catSub) {
 997                         bool up = (t.cat() == catSuper);
 998                         MathScriptInset * p = 0;
 999                         if (array.size())
1000                                 p = array.back()->asScriptInset();
1001                         if (!p || p->has(up)) {
1002                                 array.push_back(MathAtom(new MathScriptInset(up)));
1003                                 p = array.back()->asScriptInset();
1004                         }
1005                         p->ensure(up);
1006                         parse_into(p->cell(up), FLAG_ITEM);
1007                         p->limits(limits);
1008                         limits = 0;
1009                 }
1010
1011                 else if (t.character() == ']' && (flags & FLAG_BRACK_END))
1012                         return;
1013
1014                 else if (t.cat() == catOther)
1015                         add(array, t.character(), code);
1016
1017                 //
1018                 // control sequences
1019                 //
1020                 else if (t.cs() == "protect")
1021                         // ignore \\protect, will be re-added during output
1022                         ;
1023
1024                 else if (t.cs() == "end")
1025                         break;
1026
1027                 else if (t.cs() == ")")
1028                         break;
1029
1030                 else if (t.cs() == "]")
1031                         break;
1032
1033                 else if (t.cs() == "\\") {
1034                         curr_skip_ = getArg('[', ']');
1035                         //lyxerr << "found newline unexpectedly, array: '" << array << "'\n";
1036                         lyxerr << "found newline unexpectedly\n";
1037                         array.push_back(createMathInset("\\"));
1038                 }
1039
1040                 else if (t.cs() == "limits")
1041                         limits = 1;
1042
1043                 else if (t.cs() == "nolimits")
1044                         limits = -1;
1045
1046                 else if (t.cs() == "nonumber")
1047                         curr_num_ = false;
1048
1049                 else if (t.cs() == "number")
1050                         curr_num_ = true;
1051
1052                 else if (t.cs() == "sqrt") {
1053                         char c = getChar();
1054                         if (c == '[') {
1055                                 array.push_back(MathAtom(new MathRootInset));
1056                                 parse_into(array.back()->cell(0), FLAG_BRACK_END);
1057                                 parse_into(array.back()->cell(1), FLAG_ITEM);
1058                         } else {
1059                                 putback();
1060                                 array.push_back(MathAtom(new MathSqrtInset));
1061                                 parse_into(array.back()->cell(0), FLAG_ITEM);
1062                         }
1063                 }
1064
1065                 else if (t.cs() == "left") {
1066                         string l = getToken().asString();
1067                         MathArray ar;
1068                         parse_into(ar, FLAG_RIGHT);
1069                         string r = getToken().asString();
1070                         MathAtom dl(new MathDelimInset(l, r));
1071                         dl->cell(0) = ar;
1072                         array.push_back(dl);
1073                 }
1074
1075                 else if (t.cs() == "right") {
1076                         if (!(flags & FLAG_RIGHT)) {
1077                                 //lyxerr << "got so far: '" << array << "'\n";
1078                                 error("Unmatched right delimiter");
1079                         }
1080                         return;
1081                 }
1082
1083                 else if (t.cs() == "begin") {
1084                         string const name = getArg('{', '}');
1085                         if (name == "array" || name == "subarray") {
1086                                 string const valign = getArg('[', ']') + 'c';
1087                                 string const halign = getArg('{', '}');
1088                                 array.push_back(MathAtom(new MathArrayInset(name, valign[0], halign)));
1089                                 parse_lines(array.back(), false, false);
1090                         } else if (name == "split" || name == "cases") {
1091                                 array.push_back(createMathInset(name));
1092                                 parse_lines(array.back(), false, false);
1093                         } else if (name == "pmatrix" || name == "bmatrix" ||
1094                                          name == "vmatrix" || name == "Vmatrix") {
1095                                 array.push_back(createMathInset(name));
1096                                 parse_lines2(array.back(), false);
1097                         } else
1098                                 lyxerr << "unknow math inset begin '" << name << "'\n";
1099                 }
1100
1101                 else if (t.cs() == "kern") {
1102 #ifdef WITH_WARNINGS
1103 #warning A hack...
1104 #endif
1105                         string s;
1106                         while (1) {
1107                                 Token const & t = getToken();
1108                                 if (!good()) {
1109                                         putback();
1110                                         break;
1111                                 }
1112                                 s += t.character();
1113                                 if (isValidLength(s))
1114                                         break;
1115                         }
1116                         array.push_back(MathAtom(new MathKernInset(s)));
1117                 }
1118
1119                 else if (t.cs() == "label") {
1120                         curr_label_ = getArg('{', '}');
1121                 }
1122
1123                 else if (t.cs() == "choose" || t.cs() == "over" || t.cs() == "atop") {
1124                         MathAtom p = createMathInset(t.cs());
1125                         array.swap(p->cell(0));
1126                         parse_into(p->cell(1), flags, code);
1127                         array.push_back(p);
1128                         return;
1129                 }
1130
1131                 else if (t.cs() == "substack") {
1132                         array.push_back(createMathInset(t.cs()));
1133                         skipBegin();
1134                         parse_lines2(array.back(), true);
1135                 }
1136
1137                 else if (t.cs() == "xymatrix") {
1138                         array.push_back(createMathInset(t.cs()));
1139                         skipBegin();
1140                         parse_lines2(array.back(), true);
1141                 }
1142
1143 #if 0
1144                 // Disabled
1145                 else if (1 && t.cs() == "ar") {
1146                         MathXYArrowInset * p = new MathXYArrowInset;
1147
1148                         // try to read target
1149                         char c = getChar();
1150                         if (c == '[') {
1151                                 parse_into(p->cell(0), FLAG_BRACK_END);
1152                                 //lyxerr << "read target: " << p->cell(0) << "\n";
1153                         } else {
1154                                 putback();
1155                         }
1156
1157                         // try to read label
1158                         if (nextToken().cat() == catSuper || nextToken().cat() == catSub) {
1159                                 p->up_ = nextToken().cat() == catSuper;
1160                                 getToken();
1161                                 parse_into(p->cell(1), FLAG_ITEM);
1162                                 //lyxerr << "read label: " << p->cell(1) << "\n";
1163                         }
1164
1165                         array.push_back(MathAtom(p));
1166                         //lyxerr << "read array: " << array << "\n";
1167                 }
1168 #endif
1169
1170 #if 0
1171                 else if (t.cs() == "mbox" || t.cs() == "text") {
1172                         //array.push_back(createMathInset(t.cs()));
1173                         array.push_back(MathAtom(new MathBoxInset(t.cs())));
1174                         // slurp in the argument of mbox
1175
1176                         MathBoxInset * p = array.back()->asBoxInset();
1177                         //lyx::assert(p);
1178                 }
1179 #endif
1180
1181
1182                 else if (t.cs().size()) {
1183                         latexkeys const * l = in_word_set(t.cs());
1184                         if (l) {
1185                                 if (l->token == LM_TK_FONT) {
1186                                         //lyxerr << "starting font\n";
1187                                         //CatCode catSpaceSave = theCatcode[' '];
1188                                         //if (l->id == LM_TC_TEXTRM) {
1189                                         //      // temporarily change catcode
1190                                         //      theCatcode[' '] = catLetter;
1191                                         //}
1192
1193                                         MathArray ar;
1194                                         parse_into(ar, FLAG_ITEM, static_cast<MathTextCodes>(l->id));
1195                                         array.push_back(ar);
1196
1197                                         // undo catcode changes
1198                                         ////theCatcode[' '] = catSpaceSave;
1199                                         //lyxerr << "ending font\n";
1200                                 }
1201
1202                                 else if (l->token == LM_TK_OLDFONT) {
1203                                         code = static_cast<MathTextCodes>(l->id);
1204                                 }
1205
1206                                 else if (l->token == LM_TK_BOX) {
1207                                         MathAtom p = createMathInset(t.cs());
1208                                         parse_into(p->cell(0), FLAG_ITEM | FLAG_BOX, LM_TC_BOX);
1209                                         array.push_back(p);
1210                                 }
1211
1212                                 else if (l->token == LM_TK_STY) {
1213                                         MathAtom p = createMathInset(t.cs());
1214                                         parse_into(p->cell(0), flags, code);
1215                                         array.push_back(p);
1216                                         return;
1217                                 }
1218
1219                                 else {
1220                                         MathAtom p = createMathInset(t.cs());
1221                                         for (MathInset::idx_type i = 0; i < p->nargs(); ++i)
1222                                                 parse_into(p->cell(i), FLAG_ITEM);
1223                                         array.push_back(p);
1224                                 }
1225                         }
1226
1227                         else {
1228                                 MathAtom p = createMathInset(t.cs());
1229                                 for (MathInset::idx_type i = 0; i < p->nargs(); ++i)
1230                                         parse_into(p->cell(i), FLAG_ITEM);
1231                                 array.push_back(p);
1232                         }
1233                 }
1234
1235
1236                 if (flags & FLAG_LEAVE) {
1237                         flags &= ~FLAG_LEAVE;
1238                         break;
1239                 }
1240         }
1241
1242         if (panic) {
1243                 lyxerr << " Math Panic, expect problems!\n";
1244                 //   Search for the end command.
1245                 Token t;
1246                 do {
1247                         t = getToken();
1248                 } while (good() && t.cs() != "end");
1249         }
1250 }
1251
1252
1253
1254 } // anonymous namespace
1255
1256
1257 void mathed_parse_cell(MathArray & ar, string const & str)
1258 {
1259         istringstream is(str.c_str());
1260         mathed_parse_cell(ar, is);
1261 }
1262
1263
1264 void mathed_parse_cell(MathArray & ar, istream & is)
1265 {
1266         Parser(is).parse_into(ar, 0);
1267 }
1268
1269
1270
1271 bool mathed_parse_macro(string & name, string const & str)
1272 {
1273         istringstream is(str.c_str());
1274         Parser parser(is);
1275         return parser.parse_macro(name);
1276 }
1277
1278 bool mathed_parse_macro(string & name, istream & is)
1279 {
1280         Parser parser(is);
1281         return parser.parse_macro(name);
1282 }
1283
1284 bool mathed_parse_macro(string & name, LyXLex & lex)
1285 {
1286         Parser parser(lex);
1287         return parser.parse_macro(name);
1288 }
1289
1290
1291
1292 bool mathed_parse_normal(MathAtom & t, string const & str)
1293 {
1294         istringstream is(str.c_str());
1295         Parser parser(is);
1296         return parser.parse_normal(t);
1297 }
1298
1299 bool mathed_parse_normal(MathAtom & t, istream & is)
1300 {
1301         Parser parser(is);
1302         return parser.parse_normal(t);
1303 }
1304
1305 bool mathed_parse_normal(MathAtom & t, LyXLex & lex)
1306 {
1307         Parser parser(lex);
1308         return parser.parse_normal(t);
1309 }