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; true; ++col) {
 592                         //lyxerr << "reading cell " << row << " " << col << " "
 593                         // << p->ncols() << "\n";
 594                         //lyxerr << "ncols: " << p->ncols() << "\n";
 595
 596                         if (col >= p->ncols()) {
 597                                 //lyxerr << "adding col " << col << "\n";
 598                                 p->addCol(p->ncols());
 599                         }
 600
 601                         MathArray & ar = p->cell(col + row * p->ncols());
 602                         parse_into(ar, FLAG_BLOCK);
 603                         // remove 'unnecessary' braces:
 604                         if (ar.size() == 1 && ar.back()->asBraceInset())
 605                                 ar = ar.back()->asBraceInset()->cell(0);
 606                         //lyxerr << "ar: " << ar << "\n";
 607
 608                         // break if cell is not followed by an ampersand
 609                         if (nextToken().cat() != catAlign) {
 610                                 //lyxerr << "less cells read than normal in row/col: "
 611                                 //      << row << " " << col << "\n";
 612                                 break;
 613                         }
 614
 615                         // skip the ampersand
 616                         getToken();
 617                 }
 618
 619                 if (outmost) {
 620                         MathHullInset * m = t->asHullInset();
 621                         if (!m) {
 622                                 lyxerr << "error in Parser::parse_lines() 2\n";
 623                                 return false;
 624                         }
 625                         m->numbered(row, curr_num_);
 626                         m->label(row, curr_label_);
 627                         if (curr_skip_.size()) {
 628                                 m->vcrskip(LyXLength(curr_skip_), row);
 629                                 curr_skip_.erase();
 630                         }
 631                 }
 632
 633                 // is a \\ coming?
 634                 if (nextToken().isCR()) {
 635                         // skip the cr-token
 636                         getToken();
 637
 638                         // try to read a length
 639                         //get
 640
 641                         // read hlines for next row
 642                         p->rowinfo(row + 1).lines_ = readHLines();
 643                 }
 644
 645                 // we are finished if the next token is an 'end'
 646                 if (nextToken().cs() == "end") {
 647                         // skip the end-token
 648                         getToken();
 649                         getArg('{','}');
 650
 651                         // leave the 'read a line'-loop
 652                         break;
 653                 }
 654
 655                 // otherwise, we have to start a new row
 656                 p->appendRow();
 657         }
 658
 659         // restore "global" variables
 660         curr_num_   = saved_num;
 661         curr_label_ = saved_label;
 662
 663         return true;
 664 }
 665
 666
 667 bool Parser::parse_lines2(MathAtom & t, bool braced)
 668 {
 669         MathGridInset * p = t->asGridInset();
 670         if (!p) {
 671                 lyxerr << "error in Parser::parse_lines() 1\n";
 672                 return false;
 673         }
 674
 675         for (int row = 0; true; ++row) {
 676                 // reading a row
 677                 for (MathInset::col_type col = 0; true; ++col) {
 678                         //lyxerr << "reading cell " << row << " " << col << " " << p->ncols() << "\n";
 679
 680                         if (col >= p->ncols()) {
 681                                 //lyxerr << "adding col " << col << "\n";
 682                                 p->addCol(p->ncols());
 683                         }
 684
 685                         parse_into(p->cell(col + row * p->ncols()), FLAG_BLOCK2);
 686                         //lyxerr << "read cell: " << p->cell(col + row * p->ncols()) << "\n";
 687
 688                         // break if cell is not followed by an ampersand
 689                         if (nextToken().cat() != catAlign) {
 690                                 //lyxerr << "less cells read than normal in row/col: " << row << " " << col << "\n";
 691                                 break;
 692                         }
 693
 694                         // skip the ampersand
 695                         getToken();
 696                 }
 697
 698                 // is a \\ coming?
 699                 if (nextToken().isCR()) {
 700                         // skip the cr-token
 701                         getToken();
 702                 }
 703
 704                 // we are finished if the next token is the one we expected
 705                 // skip the end-token
 706                 // leave the 'read a line'-loop
 707                 if (braced) {
 708                         if (nextToken().cat() == catEnd) {
 709                                 getToken();
 710                                 break;
 711                         }
 712                 } else {
 713                         if (nextToken().cs() == "end") {
 714                                 getToken();
 715                                 getArg('{','}');
 716                                 break;
 717                         }
 718                 }
 719
 720                 // otherwise, we have to start a new row
 721                 p->appendRow();
 722         }
 723
 724         return true;
 725 }
 726
 727
 728
 729 bool Parser::parse_macro(string & name)
 730 {
 731         name = "{error}";
 732         skipSpaces();
 733
 734         if (getToken().cs() != "newcommand") {
 735                 lyxerr << "\\newcommand expected\n";
 736                 return false;
 737         }
 738
 739         if (getToken().cat() != catBegin) {
 740                 lyxerr << "'{' in \\newcommand expected (1)\n";
 741                 return false;
 742         }
 743
 744         name = getToken().cs();
 745
 746         if (getToken().cat() != catEnd) {
 747                 lyxerr << "'}' expected\n";
 748                 return false;
 749         }
 750
 751         string    arg  = getArg('[', ']');
 752         int       narg = arg.empty() ? 0 : atoi(arg.c_str());
 753
 754         if (getToken().cat() != catBegin) {
 755                 lyxerr << "'{' in \\newcommand expected (2)\n";
 756                 return false;
 757         }
 758
 759         MathArray ar;
 760         parse_into(ar, FLAG_BRACE_LAST);
 761
 762         // we cannot handle recursive stuff at all
 763         MathArray test;
 764         test.push_back(createMathInset(name));
 765         if (ar.contains(test)) {
 766                 lyxerr << "we cannot handle recursive macros at all.\n";
 767                 return false;
 768         }
 769
 770         MathMacroTable::create(name, narg, ar);
 771         return true;
 772 }
 773
 774
 775 bool Parser::parse_normal(MathAtom & matrix)
 776 {
 777         skipSpaces();
 778         Token const & t = getToken();
 779
 780         if (t.cs() == "(") {
 781                 matrix = MathAtom(new MathHullInset(LM_OT_SIMPLE));
 782                 parse_into(matrix->cell(0), 0);
 783                 return true;
 784         }
 785
 786         if (t.cat() == catMath) {
 787                 Token const & n = getToken();
 788                 if (n.cat() == catMath) {
 789                         // TeX's $$...$$ syntax for displayed math
 790                         matrix = MathAtom(new MathHullInset(LM_OT_EQUATION));
 791                         MathHullInset * p = matrix->asHullInset();
 792                         parse_into(p->cell(0), 0);
 793                         p->numbered(0, curr_num_);
 794                         p->label(0, curr_label_);
 795                 } else {
 796                         // simple $...$  stuff
 797                         putback();
 798                         matrix = MathAtom(new MathHullInset(LM_OT_SIMPLE));
 799                         parse_into(matrix->cell(0), 0);
 800                 }
 801                 return true;
 802         }
 803
 804         if (!t.cs().size()) {
 805                 lyxerr << "start of math expected, got '" << t << "'\n";
 806                 return false;
 807         }
 808
 809         string const & cs = t.cs();
 810
 811         if (cs == "[") {
 812                 curr_num_ = 0;
 813                 curr_label_.erase();
 814                 matrix = MathAtom(new MathHullInset(LM_OT_EQUATION));
 815                 MathHullInset * p = matrix->asHullInset();
 816                 parse_into(p->cell(0), 0);
 817                 p->numbered(0, curr_num_);
 818                 p->label(0, curr_label_);
 819                 return true;
 820         }
 821
 822         if (cs != "begin") {
 823                 lyxerr << "'begin' of un-simple math expected, got '" << cs << "'\n";
 824                 return false;
 825         }
 826
 827         string const name = getArg('{', '}');
 828
 829         if (name == "math") {
 830                 matrix = MathAtom(new MathHullInset(LM_OT_SIMPLE));
 831                 parse_into(matrix->cell(0), 0);
 832                 return true;
 833         }
 834
 835         if (name == "equation" || name == "equation*" || name == "displaymath") {
 836                 curr_num_ = (name == "equation");
 837                 curr_label_.erase();
 838                 matrix = MathAtom(new MathHullInset(LM_OT_EQUATION));
 839                 MathHullInset * p = matrix->asHullInset();
 840                 parse_into(p->cell(0), FLAG_END);
 841                 p->numbered(0, curr_num_);
 842                 p->label(0, curr_label_);
 843                 return true;
 844         }
 845
 846         if (name == "eqnarray" || name == "eqnarray*") {
 847                 matrix = MathAtom(new MathHullInset(LM_OT_EQNARRAY));
 848                 return parse_lines(matrix, !stared(name), true);
 849         }
 850
 851         if (name == "align" || name == "align*") {
 852                 matrix = MathAtom(new MathHullInset(LM_OT_ALIGN));
 853                 return parse_lines(matrix, !stared(name), true);
 854         }
 855
 856         if (name == "alignat" || name == "alignat*") {
 857                 int nc = 2 * atoi(getArg('{', '}').c_str());
 858                 matrix = MathAtom(new MathHullInset(LM_OT_ALIGNAT, nc));
 859                 return parse_lines(matrix, !stared(name), true);
 860         }
 861
 862         if (name == "xalignat" || name == "xalignat*") {
 863                 int nc = 2 * atoi(getArg('{', '}').c_str());
 864                 matrix = MathAtom(new MathHullInset(LM_OT_XALIGNAT, nc));
 865                 return parse_lines(matrix, !stared(name), true);
 866         }
 867
 868         if (name == "xxalignat") {
 869                 int nc = 2 * atoi(getArg('{', '}').c_str());
 870                 matrix = MathAtom(new MathHullInset(LM_OT_XXALIGNAT, nc));
 871                 return parse_lines(matrix, !stared(name), true);
 872         }
 873
 874         if (name == "multline" || name == "multline*") {
 875                 matrix = MathAtom(new MathHullInset(LM_OT_MULTLINE));
 876                 return parse_lines(matrix, !stared(name), true);
 877         }
 878
 879         if (name == "gather" || name == "gather*") {
 880                 matrix = MathAtom(new MathHullInset(LM_OT_GATHER));
 881                 return parse_lines(matrix, !stared(name), true);
 882         }
 883
 884         lyxerr[Debug::MATHED] << "1: unknown math environment: " << name << "\n";
 885         lyxerr << "1: unknown math environment: " << name << "\n";
 886         return false;
 887 }
 888
 889
 890 void Parser::parse_into(MathArray & array, unsigned flags, MathTextCodes code)
 891 {
 892         parse_into1(array, flags, code);
 893         // remove 'unnecessary' braces:
 894         if (array.size() == 1 && array.back()->asBraceInset()) {
 895                 lyxerr << "extra braces removed\n";
 896                 array = array.back()->asBraceInset()->cell(0);
 897         }
 898 }
 899
 900
 901 void Parser::parse_into1(MathArray & array, unsigned flags, MathTextCodes code)
 902 {
 903         bool panic  = false;
 904         int  limits = 0;
 905
 906         while (good()) {
 907                 Token const & t = getToken();
 908
 909 #ifdef FILEDEBUG
 910                 lyxerr << "t: " << t << " flags: " << flags << "\n";
 911                 //array.dump();
 912                 lyxerr << "\n";
 913 #endif
 914
 915                 if (flags & FLAG_ITEM) {
 916                         flags &= ~FLAG_ITEM;
 917                         if (t.cat() == catBegin) {
 918                                 // skip the brace and collect everything to the next matching
 919                                 // closing brace
 920                                 flags |= FLAG_BRACE_LAST;
 921                                 continue;
 922                         } else {
 923                                 // handle only this single token, leave the loop if done
 924                                 flags |= FLAG_LEAVE;
 925                         }
 926                 }
 927
 928                 if (flags & FLAG_BLOCK) {
 929                         if (t.cat() == catAlign || t.isCR() || t.cs() == "end") {
 930                                 putback();
 931                                 return;
 932                         }
 933                 }
 934
 935                 if (flags & FLAG_BLOCK2) {
 936                         if (t.cat() == catAlign || t.isCR() || t.cs() == "end"
 937                                         || t.cat() == catEnd) {
 938                                 putback();
 939                                 return;
 940                         }
 941                 }
 942
 943                 //
 944                 // cat codes
 945                 //
 946                 if (t.cat() == catMath) {
 947                         if (flags & FLAG_BOX) {
 948                                 // we are inside an mbox, so opening new math is allowed
 949                                 array.push_back(MathAtom(new MathHullInset(LM_OT_SIMPLE)));
 950                                 parse_into(array.back()->cell(0), 0);
 951                         } else {
 952                                 // otherwise this is the end of the formula
 953                                 break;
 954                         }
 955                 }
 956
 957                 else if (t.cat() == catLetter)
 958                         add(array, t.character(), code);
 959
 960                 else if (t.cat() == catSpace && code == LM_TC_TEXTRM)
 961                         add(array, t.character(), code);
 962
 963                 else if (t.cat() == catParameter) {
 964                         Token const & n = getToken();
 965                         array.push_back(MathAtom(new MathMacroArgument(n.character()-'0', code)));
 966                 }
 967
 968                 else if (t.cat() == catBegin) {
 969                         MathArray ar;
 970                         parse_into(ar, FLAG_BRACE_LAST);
 971 #ifndef WITH_WARNINGS
 972 #warning this might be wrong in general!
 973 #endif
 974                         // ignore braces around simple items
 975                         if ((ar.size() == 1 && !ar.front()->needsBraces()
 976        || (ar.size() == 2 && !ar.front()->needsBraces()
 977                                             && ar.back()->asScriptInset()))
 978        || (ar.size() == 0 && array.size() == 0))
 979                         {
 980                                 array.push_back(ar);
 981                         } else {
 982                                 array.push_back(MathAtom(new MathBraceInset));
 983                                 array.back()->cell(0).swap(ar);
 984                         }
 985                 }
 986
 987                 else if (t.cat() == catEnd) {
 988                         if (flags & FLAG_BRACE_LAST)
 989                                 return;
 990                         lyxerr << "found '}' unexpectedly, array: '" << array << "'\n";
 991                         //lyxerr << "found '}' unexpectedly\n";
 992                         lyx::Assert(0);
 993                         add(array, '}', LM_TC_TEX);
 994                 }
 995
 996                 else if (t.cat() == catAlign) {
 997                         lyxerr << "found tab unexpectedly, array: '" << array << "'\n";
 998                         //lyxerr << "found tab unexpectedly\n";
 999                         add(array, '&', LM_TC_TEX);
1000                 }
1001
1002                 else if (t.cat() == catSuper || t.cat() == catSub) {
1003                         bool up = (t.cat() == catSuper);
1004                         MathScriptInset * p = 0;
1005                         if (array.size())
1006                                 p = array.back()->asScriptInset();
1007                         if (!p || p->has(up)) {
1008                                 array.push_back(MathAtom(new MathScriptInset(up)));
1009                                 p = array.back()->asScriptInset();
1010                         }
1011                         p->ensure(up);
1012                         parse_into(p->cell(up), FLAG_ITEM);
1013                         p->limits(limits);
1014                         limits = 0;
1015                 }
1016
1017                 else if (t.character() == ']' && (flags & FLAG_BRACK_END))
1018                         return;
1019
1020                 else if (t.cat() == catOther)
1021                         add(array, t.character(), code);
1022
1023                 //
1024                 // control sequences
1025                 //
1026                 else if (t.cs() == "protect")
1027                         // ignore \\protect, will be re-added during output
1028                         ;
1029
1030                 else if (t.cs() == "end")
1031                         break;
1032
1033                 else if (t.cs() == ")")
1034                         break;
1035
1036                 else if (t.cs() == "]")
1037                         break;
1038
1039                 else if (t.cs() == "\\") {
1040                         curr_skip_ = getArg('[', ']');
1041                         //lyxerr << "found newline unexpectedly, array: '" << array << "'\n";
1042                         lyxerr << "found newline unexpectedly\n";
1043                         array.push_back(createMathInset("\\"));
1044                 }
1045
1046                 else if (t.cs() == "limits")
1047                         limits = 1;
1048
1049                 else if (t.cs() == "nolimits")
1050                         limits = -1;
1051
1052                 else if (t.cs() == "nonumber")
1053                         curr_num_ = false;
1054
1055                 else if (t.cs() == "number")
1056                         curr_num_ = true;
1057
1058                 else if (t.cs() == "sqrt") {
1059                         char c = getChar();
1060                         if (c == '[') {
1061                                 array.push_back(MathAtom(new MathRootInset));
1062                                 parse_into(array.back()->cell(0), FLAG_BRACK_END);
1063                                 parse_into(array.back()->cell(1), FLAG_ITEM);
1064                         } else {
1065                                 putback();
1066                                 array.push_back(MathAtom(new MathSqrtInset));
1067                                 parse_into(array.back()->cell(0), FLAG_ITEM);
1068                         }
1069                 }
1070
1071                 else if (t.cs() == "left") {
1072                         string l = getToken().asString();
1073                         MathArray ar;
1074                         parse_into(ar, FLAG_RIGHT);
1075                         string r = getToken().asString();
1076                         MathAtom dl(new MathDelimInset(l, r));
1077                         dl->cell(0) = ar;
1078                         array.push_back(dl);
1079                 }
1080
1081                 else if (t.cs() == "right") {
1082                         if (!(flags & FLAG_RIGHT)) {
1083                                 //lyxerr << "got so far: '" << array << "'\n";
1084                                 error("Unmatched right delimiter");
1085                         }
1086                         return;
1087                 }
1088
1089                 else if (t.cs() == "begin") {
1090                         string const name = getArg('{', '}');
1091                         if (name == "array" || name == "subarray") {
1092                                 string const valign = getArg('[', ']') + 'c';
1093                                 string const halign = getArg('{', '}');
1094                                 array.push_back(MathAtom(new MathArrayInset(name, valign[0], halign)));
1095                                 parse_lines(array.back(), false, false);
1096                         } else if (name == "split" || name == "cases" ||
1097                                          name == "gathered" || name == "aligned") {
1098                                 array.push_back(createMathInset(name));
1099                                 parse_lines(array.back(), false, false);
1100                         } else if (name == "matrix"  || name == "pmatrix" || name == "bmatrix" ||
1101                                          name == "vmatrix" || name == "Vmatrix") {
1102                                 array.push_back(createMathInset(name));
1103                                 parse_lines2(array.back(), false);
1104                         } else
1105                                 lyxerr << "unknow math inset begin '" << name << "'\n";
1106                 }
1107
1108                 else if (t.cs() == "kern") {
1109 #ifdef WITH_WARNINGS
1110 #warning A hack...
1111 #endif
1112                         string s;
1113                         while (1) {
1114                                 Token const & t = getToken();
1115                                 if (!good()) {
1116                                         putback();
1117                                         break;
1118                                 }
1119                                 s += t.character();
1120                                 if (isValidLength(s))
1121                                         break;
1122                         }
1123                         array.push_back(MathAtom(new MathKernInset(s)));
1124                 }
1125
1126                 else if (t.cs() == "label") {
1127                         curr_label_ = getArg('{', '}');
1128                 }
1129
1130                 else if (t.cs() == "choose" || t.cs() == "over" || t.cs() == "atop") {
1131                         MathAtom p = createMathInset(t.cs());
1132                         array.swap(p->cell(0));
1133                         parse_into(p->cell(1), flags, code);
1134                         array.push_back(p);
1135                         return;
1136                 }
1137
1138                 else if (t.cs() == "substack") {
1139                         array.push_back(createMathInset(t.cs()));
1140                         skipBegin();
1141                         parse_lines2(array.back(), true);
1142                 }
1143
1144                 else if (t.cs() == "xymatrix") {
1145                         array.push_back(createMathInset(t.cs()));
1146                         skipBegin();
1147                         parse_lines2(array.back(), true);
1148                 }
1149
1150 #if 0
1151                 // Disabled
1152                 else if (1 && t.cs() == "ar") {
1153                         MathXYArrowInset * p = new MathXYArrowInset;
1154
1155                         // try to read target
1156                         char c = getChar();
1157                         if (c == '[') {
1158                                 parse_into(p->cell(0), FLAG_BRACK_END);
1159                                 //lyxerr << "read target: " << p->cell(0) << "\n";
1160                         } else {
1161                                 putback();
1162                         }
1163
1164                         // try to read label
1165                         if (nextToken().cat() == catSuper || nextToken().cat() == catSub) {
1166                                 p->up_ = nextToken().cat() == catSuper;
1167                                 getToken();
1168                                 parse_into(p->cell(1), FLAG_ITEM);
1169                                 //lyxerr << "read label: " << p->cell(1) << "\n";
1170                         }
1171
1172                         array.push_back(MathAtom(p));
1173                         //lyxerr << "read array: " << array << "\n";
1174                 }
1175 #endif
1176
1177 #if 0
1178                 else if (t.cs() == "mbox" || t.cs() == "text") {
1179                         //array.push_back(createMathInset(t.cs()));
1180                         array.push_back(MathAtom(new MathBoxInset(t.cs())));
1181                         // slurp in the argument of mbox
1182
1183                         MathBoxInset * p = array.back()->asBoxInset();
1184                         //lyx::assert(p);
1185                 }
1186 #endif
1187
1188
1189                 else if (t.cs().size()) {
1190                         latexkeys const * l = in_word_set(t.cs());
1191                         if (l) {
1192                                 if (l->token == LM_TK_FONT) {
1193                                         //lyxerr << "starting font\n";
1194                                         //CatCode catSpaceSave = theCatcode[' '];
1195                                         //if (l->id == LM_TC_TEXTRM) {
1196                                         //      // temporarily change catcode
1197                                         //      theCatcode[' '] = catLetter;
1198                                         //}
1199
1200                                         MathArray ar;
1201                                         parse_into(ar, FLAG_ITEM, static_cast<MathTextCodes>(l->id));
1202                                         array.push_back(ar);
1203
1204                                         // undo catcode changes
1205                                         ////theCatcode[' '] = catSpaceSave;
1206                                         //lyxerr << "ending font\n";
1207                                 }
1208
1209                                 else if (l->token == LM_TK_OLDFONT) {
1210                                         code = static_cast<MathTextCodes>(l->id);
1211                                 }
1212
1213                                 else if (l->token == LM_TK_BOX) {
1214                                         MathAtom p = createMathInset(t.cs());
1215                                         parse_into(p->cell(0), FLAG_ITEM | FLAG_BOX, LM_TC_BOX);
1216                                         array.push_back(p);
1217                                 }
1218
1219                                 else if (l->token == LM_TK_STY) {
1220                                         MathAtom p = createMathInset(t.cs());
1221                                         parse_into(p->cell(0), flags, code);
1222                                         array.push_back(p);
1223                                         return;
1224                                 }
1225
1226                                 else {
1227                                         MathAtom p = createMathInset(t.cs());
1228                                         for (MathInset::idx_type i = 0; i < p->nargs(); ++i)
1229                                                 parse_into(p->cell(i), FLAG_ITEM);
1230                                         array.push_back(p);
1231                                 }
1232                         }
1233
1234                         else {
1235                                 MathAtom p = createMathInset(t.cs());
1236                                 for (MathInset::idx_type i = 0; i < p->nargs(); ++i)
1237                                         parse_into(p->cell(i), FLAG_ITEM);
1238                                 array.push_back(p);
1239                         }
1240                 }
1241
1242
1243                 if (flags & FLAG_LEAVE) {
1244                         flags &= ~FLAG_LEAVE;
1245                         break;
1246                 }
1247         }
1248
1249         if (panic) {
1250                 lyxerr << " Math Panic, expect problems!\n";
1251                 //   Search for the end command.
1252                 Token t;
1253                 do {
1254                         t = getToken();
1255                 } while (good() && t.cs() != "end");
1256         }
1257 }
1258
1259
1260
1261 } // anonymous namespace
1262
1263
1264 void mathed_parse_cell(MathArray & ar, string const & str)
1265 {
1266         istringstream is(str.c_str());
1267         mathed_parse_cell(ar, is);
1268 }
1269
1270
1271 void mathed_parse_cell(MathArray & ar, istream & is)
1272 {
1273         Parser(is).parse_into(ar, 0);
1274 }
1275
1276
1277
1278 bool mathed_parse_macro(string & name, string const & str)
1279 {
1280         istringstream is(str.c_str());
1281         Parser parser(is);
1282         return parser.parse_macro(name);
1283 }
1284
1285 bool mathed_parse_macro(string & name, istream & is)
1286 {
1287         Parser parser(is);
1288         return parser.parse_macro(name);
1289 }
1290
1291 bool mathed_parse_macro(string & name, LyXLex & lex)
1292 {
1293         Parser parser(lex);
1294         return parser.parse_macro(name);
1295 }
1296
1297
1298
1299 bool mathed_parse_normal(MathAtom & t, string const & str)
1300 {
1301         istringstream is(str.c_str());
1302         Parser parser(is);
1303         return parser.parse_normal(t);
1304 }
1305
1306 bool mathed_parse_normal(MathAtom & t, istream & is)
1307 {
1308         Parser parser(is);
1309         return parser.parse_normal(t);
1310 }
1311
1312 bool mathed_parse_normal(MathAtom & t, LyXLex & lex)
1313 {
1314         Parser parser(lex);
1315         return parser.parse_normal(t);
1316 }