src/tex2lyx/tex2lyx.C

   1 /** The .tex to .lyx converter
   2     \author André Pönitz (2003)
   3  */
   4
   5 // {[(
   6
   7 #include <config.h>
   8
   9 #include <algorithm>
  10 #include <cctype>
  11 #include <fstream>
  12 #include <iostream>
  13 #include <stack>
  14 #include <string>
  15 #include <vector>
  16
  17 #include "Lsstream.h"
  18 #include "texparser.h"
  19
  20 using std::cout;
  21 using std::cerr;
  22 using std::endl;
  23 using std::fill;
  24 using std::getline;
  25 using std::ios;
  26 using std::ifstream;
  27 using std::istream;
  28 using std::istringstream;
  29 using std::ostream;
  30 using std::ostringstream;
  31 using std::stack;
  32 using std::string;
  33 using std::vector;
  34
  35
  36 namespace {
  37
  38 char const OPEN = '<';
  39 char const CLOSE = '>';
  40
  41 const char * known_languages[] = { "austrian", "babel", "bahasa",
  42 "basque", "breton", "bulgarian", "catalan", "croatian", "czech", "danish",
  43 "dutch", "english", "esperanto", "estonian", "finnish", "francais",
  44 "frenchb", "galician", "germanb", "greek", "hebcal", "hebfont", "hebrew",
  45 "hebrew_newcode", "hebrew_oldcode", "hebrew_p", "hyphen", "icelandic",
  46 "irish", "italian", "latin", "lgrcmr", "lgrcmro", "lgrcmss", "lgrcmtt",
  47 "lgrenc", "lgrlcmss", "lgrlcmtt", "lheclas", "lhecmr", "lhecmss",
  48 "lhecmtt", "lhecrml", "lheenc", "lhefr", "lheredis", "lheshold",
  49 "lheshscr", "lheshstk", "lsorbian", "magyar", "naustrian", "ngermanb",
  50 "ngerman", "norsk", "polish", "portuges", "rlbabel", "romanian",
  51 "russianb", "samin", "scottish", "serbian", "slovak", "slovene", "spanish",
  52 "swedish", "turkish", "ukraineb", "usorbian", "welsh", 0};
  53
  54 const char * known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
  55
  56
  57 const char * known_math_envs[] = {"equation", "eqnarray", "eqnarray*",
  58 "align", 0};
  59
  60
  61 // some ugly stuff
  62 ostringstream h_preamble;
  63 string h_textclass               = "FIXME";
  64 string h_options                 = "FIXME";
  65 string h_language                = "FIXME";
  66 string h_inputencoding           = "FIXME";
  67 string h_fontscheme              = "FIXME";
  68 string h_graphics                = "default";
  69 string h_paperfontsize           = "FIXME";
  70 string h_spacing                 = "single";
  71 string h_papersize               = "FIXME";
  72 string h_paperpackage            = "FIXME";
  73 string h_use_geometry            = "0";
  74 string h_use_amsmath             = "0";
  75 string h_use_natbib              = "0";
  76 string h_use_numerical_citations = "0";
  77 string h_paperorientation        = "portrait";
  78 string h_secnumdepth             = "3";
  79 string h_tocdepth                = "3";
  80 string h_paragraph_separation    = "indent";
  81 string h_defskip                 = "medskip";
  82 string h_quotes_language         = "FIXME";
  83 string h_quotes_times            = "1";
  84 string h_papercolumns            = "1";
  85 string h_papersides              = "1";
  86 string h_paperpagestyle          = "default";
  87 string h_tracking_changes        = "0";
  88
  89 // indicates whether we are in the preamble
  90 bool in_preamble = true;
  91
  92 // current stack of nested environments
  93 stack<string> active_environments;
  94
  95
  96 string cap(string s)
  97 {
  98         if (s.size())
  99                 s[0] = toupper(s[0]);
 100         return s;
 101 }
 102
 103
 104 string const trim(string const & a, char const * p = " ")
 105 {
 106         // lyx::Assert(p);
 107
 108         if (a.empty() || !*p)
 109                 return a;
 110
 111         string::size_type r = a.find_last_not_of(p);
 112         string::size_type l = a.find_first_not_of(p);
 113
 114         // Is this the minimal test? (lgb)
 115         if (r == string::npos && l == string::npos)
 116                 return string();
 117
 118         return a.substr(l, r - l + 1);
 119 }
 120
 121
 122 void split(string const & s, vector<string> & result, char delim)
 123 {
 124         istringstream is(s);
 125         string t;
 126         while (getline(is, t, delim))
 127                 result.push_back(t);
 128 }
 129
 130
 131 string join(vector<string> const & input, char delim)
 132 {
 133         ostringstream os;
 134         for (size_t i = 0; i != input.size(); ++i) {
 135                 if (i)
 136                         os << delim;
 137                 os << input[i];
 138         }
 139         return os.str();
 140 }
 141
 142
 143 void handle_opt(vector<string> & opts, char const ** what, string & target)
 144 {
 145         if (opts.empty())
 146                 return;
 147
 148         for ( ; what; ++what) {
 149                 vector<string>::iterator it = find(opts.begin(), opts.end(), *what);
 150                 if (it != opts.end()) {
 151                         //cerr << "### found option '" << *what << "'\n";
 152                         target = *what;
 153                         opts.erase(it);
 154                         return;
 155                 }
 156         }
 157 }
 158
 159
 160 bool is_math_env(string const & name)
 161 {
 162         for (char const ** what = known_math_envs; *what; ++what)
 163                 if (*what == name)
 164                         return true;
 165         return false;
 166 }
 167
 168
 169 bool is_heading(string const & name)
 170 {
 171         return
 172                 name == "title" ||
 173                 name == "author" ||
 174                 name == "paragraph" ||
 175                 name == "chapter" ||
 176                 name == "chapter*" ||
 177                 name == "section" ||
 178                 name == "section*" ||
 179                 name == "subsection" ||
 180                 name == "subsection*" ||
 181                 name == "subsubsection" ||
 182                 name == "subsubsection*";
 183 }
 184
 185
 186 void begin_inset(ostream & os, string const & name)
 187 {
 188         os << "\n\\begin_inset " << name;
 189 }
 190
 191
 192 void end_inset(ostream & os)
 193 {
 194         os << "\n\\end_inset\n";
 195 }
 196
 197
 198 void handle_ert(ostream & os, string const & s)
 199 {
 200         begin_inset(os, "ERT");
 201         os << "\nstatus Collapsed\n\n\\layout Standard\n\n";
 202         for (string::const_iterator it = s.begin(), et = s.end(); it != et; ++it) {
 203                 if (*it == '\\')
 204                         os << "\n\\backslash\n";
 205                 else
 206                         os << *it;
 207         }
 208         end_inset(os);
 209 }
 210
 211
 212 void handle_par(ostream & os)
 213 {
 214         if (active_environments.empty())
 215                 return;
 216         os << "\n\\layout ";
 217         string s = active_environments.top();
 218         if (s == "document") {
 219                 os << "Standard\n\n";
 220                 return;
 221         }
 222         if (s == "lyxcode") {
 223                 os << "LyX-Code\n\n";
 224                 return;
 225         }
 226         os << cap(s) << "\n\n";
 227 }
 228
 229
 230 void handle_package(string const & name, string const & options)
 231 {
 232         if (name == "a4wide") {
 233                 h_papersize = "a4paper";
 234                 h_paperpackage = "widemarginsa4";
 235         } else if (name == "ae")
 236                 h_fontscheme = "ae";
 237         else if (name == "aecompl")
 238                 h_fontscheme = "ae";
 239         else if (name == "amsmath")
 240                 h_use_amsmath = "1";
 241         else if (name == "amssymb")
 242                 h_use_amsmath = "1";
 243         else if (name == "babel")
 244                 ; // ignore this
 245         else if (name == "fontenc")
 246                 ; // ignore this
 247         else if (name == "inputenc")
 248                 h_inputencoding = options;
 249         else if (name == "makeidx")
 250                 ; // ignore this
 251         else if (name == "verbatim")
 252                 ; // ignore this
 253         else {
 254                 if (options.size())
 255                         h_preamble << "\\usepackage[" << options << "]{" << name << "}\n";
 256                 else
 257                         h_preamble << "\\usepackage{" << name << "}\n";
 258         }
 259 }
 260
 261
 262 string wrap(string const & cmd, string const & str)
 263 {
 264         return OPEN + cmd + ' ' + str + CLOSE;
 265 }
 266
 267
 268 void end_preamble(ostream & os)
 269 {
 270         in_preamble = false;
 271         os << "# tex2lyx 0.0.2 created this file\n"
 272            << "\\lyxformat 222\n"
 273            << "\\textclass " << h_textclass << "\n"
 274            << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n"
 275            << "\\options " << h_options << "\n"
 276            << "\\language " << h_language << "\n"
 277            << "\\inputencoding " << h_inputencoding << "\n"
 278            << "\\fontscheme " << h_fontscheme << "\n"
 279            << "\\graphics " << h_graphics << "\n"
 280            << "\\paperfontsize " << h_paperfontsize << "\n"
 281            << "\\spacing " << h_spacing << "\n"
 282            << "\\papersize " << h_papersize << "\n"
 283            << "\\paperpackage " << h_paperpackage << "\n"
 284            << "\\use_geometry " << h_use_geometry << "\n"
 285            << "\\use_amsmath " << h_use_amsmath << "\n"
 286            << "\\use_natbib " << h_use_natbib << "\n"
 287            << "\\use_numerical_citations " << h_use_numerical_citations << "\n"
 288            << "\\paperorientation " << h_paperorientation << "\n"
 289            << "\\secnumdepth " << h_secnumdepth << "\n"
 290            << "\\tocdepth " << h_tocdepth << "\n"
 291            << "\\paragraph_separation " << h_paragraph_separation << "\n"
 292            << "\\defskip " << h_defskip << "\n"
 293            << "\\quotes_language " << h_quotes_language << "\n"
 294            << "\\quotes_times " << h_quotes_times << "\n"
 295            << "\\papercolumns " << h_papercolumns << "\n"
 296            << "\\papersides " << h_papersides << "\n"
 297            << "\\paperpagestyle " << h_paperpagestyle << "\n"
 298            << "\\tracking_changes " << h_tracking_changes << "\n"
 299            << h_preamble.str() << "\n";
 300 }
 301
 302
 303 void parse(Parser & p, ostream & os, unsigned flags, mode_type mode)
 304 {
 305         while (p.good()) {
 306                 Token const & t = p.getToken();
 307
 308 #ifdef FILEDEBUG
 309                 cerr << "t: " << t << " flags: " << flags << "\n";
 310                 //cell->dump();
 311 #endif
 312
 313                 if (flags & FLAG_ITEM) {
 314                         if (t.cat() == catSpace)
 315                                 continue;
 316
 317                         flags &= ~FLAG_ITEM;
 318                         if (t.cat() == catBegin) {
 319                                 // skip the brace and collect everything to the next matching
 320                                 // closing brace
 321                                 flags |= FLAG_BRACE_LAST;
 322                                 continue;
 323                         }
 324
 325                         // handle only this single token, leave the loop if done
 326                         flags |= FLAG_LEAVE;
 327                 }
 328
 329
 330                 if (flags & FLAG_BRACED) {
 331                         if (t.cat() == catSpace)
 332                                 continue;
 333
 334                         if (t.cat() != catBegin) {
 335                                 p.error("opening brace expected");
 336                                 return;
 337                         }
 338
 339                         // skip the brace and collect everything to the next matching
 340                         // closing brace
 341                         flags = FLAG_BRACE_LAST;
 342                 }
 343
 344
 345                 if (flags & FLAG_OPTION) {
 346                         if (t.cat() == catOther && t.character() == '[') {
 347                                 parse(p, os, FLAG_BRACK_LAST, mode);
 348                         } else {
 349                                 // no option found, put back token and we are done
 350                                 p.putback();
 351                         }
 352                         return;
 353                 }
 354
 355                 //
 356                 // cat codes
 357                 //
 358                 if (t.cat() == catMath) {
 359                         if (mode == TEXT_MODE || mode == MATHTEXT_MODE) {
 360                                 // we are inside some text mode thingy, so opening new math is allowed
 361                                 if (mode == TEXT_MODE)
 362                                         begin_inset(os, "Formula ");
 363                                 Token const & n = p.getToken();
 364                                 if (n.cat() == catMath) {
 365                                         // TeX's $$...$$ syntax for displayed math
 366                                         os << "\\[";
 367                                         parse(p, os, FLAG_SIMPLE, MATH_MODE);
 368                                         os << "\\]";
 369                                         p.getToken(); // skip the second '$' token
 370                                 } else {
 371                                         // simple $...$  stuff
 372                                         p.putback();
 373                                         os << '$';
 374                                         parse(p, os, FLAG_SIMPLE, MATH_MODE);
 375                                         os << '$';
 376                                 }
 377                                 if (mode == TEXT_MODE)
 378                                         end_inset(os);
 379                         }
 380
 381                         else if (flags & FLAG_SIMPLE) {
 382                                 // this is the end of the formula
 383                                 return;
 384                         }
 385
 386                         else {
 387                                 cerr << "mode: " << mode << endl;
 388                                 p.error("something strange in the parser\n");
 389                                 break;
 390                         }
 391                 }
 392
 393                 else if (t.cat() == catLetter)
 394                         os << t.character();
 395
 396                 else if (t.cat() == catSpace)
 397                         os << t.character();
 398
 399                 else if (t.cat() == catNewline)
 400                         os << ' ';
 401
 402                 else if (t.cat() == catSuper)
 403                         os << t.character();
 404
 405                 else if (t.cat() == catSub)
 406                         os << t.character();
 407
 408                 else if (t.cat() == catParameter) {
 409                         Token const & n = p.getToken();
 410                         os << wrap("macroarg", string(1, n.character()));
 411                 }
 412
 413                 else if (t.cat() == catActive)
 414                         os << wrap("active", string(1, t.character()));
 415
 416                 else if (t.cat() == catBegin) {
 417                         os << '{';
 418                         parse(p, os, FLAG_BRACE_LAST, mode);
 419                         os << '}';
 420                 }
 421
 422                 else if (t.cat() == catEnd) {
 423                         if (flags & FLAG_BRACE_LAST)
 424                                 return;
 425                         p.error("found '}' unexpectedly");
 426                         //lyx::Assert(0);
 427                         //add(cell, '}', LM_TC_TEX);
 428                 }
 429
 430                 else if (t.cat() == catAlign)
 431                         os << t.character();
 432
 433 /*
 434                         ++cellcol;
 435                         //cerr << " column now " << cellcol << " max: " << grid.ncols() << "\n";
 436                         if (cellcol == grid.ncols()) {
 437                                 //cerr << "adding column " << cellcol << "\n";
 438                                 grid.addCol(cellcol - 1);
 439                         }
 440                         cell = &grid.cell(grid.index(cellrow, cellcol));
 441                 }
 442 */
 443
 444                 else if (t.character() == ']' && (flags & FLAG_BRACK_LAST)) {
 445                         //cerr << "finished reading option\n";
 446                         return;
 447                 }
 448
 449                 else if (t.cat() == catOther)
 450                         os << string(1, t.character());
 451
 452                 else if (t.cat() == catComment) {
 453                         string s;
 454                         while (p.good()) {
 455                                 Token const & t = p.getToken();
 456                                 if (t.cat() == catNewline)
 457                                         break;
 458                                 s += t.asString();
 459                         }
 460                         //os << wrap("comment", s);
 461                         p.skipSpaces();
 462                 }
 463
 464                 //
 465                 // control sequences
 466                 //
 467
 468                 else if (t.cs() == "lyxlock") {
 469                         // ignored;
 470                 }
 471
 472                 else if (t.cs() == "newcommand" || t.cs() == "providecommand") {
 473                         string const name = p.verbatimItem();
 474                         string const opts = p.getArg('[', ']');
 475                         string const body = p.verbatimItem();
 476                         // only non-lyxspecific stuff
 477                         if (name != "noun" && name != "tabularnewline") {
 478                                 h_preamble << "\\" << t.cs() << "{" << name << "}";
 479                                 if (opts.size())
 480                                         h_preamble << "[" << opts << "]";
 481                                 h_preamble << "{" << body << "}\n";
 482                         }
 483                 }
 484
 485                 else if (t.cs() == "(") {
 486                         begin_inset(os, "Formula");
 487                         os << " \\(";
 488                         parse(p, os, FLAG_SIMPLE2, MATH_MODE);
 489                         os << "\\)";
 490                         end_inset(os);
 491                 }
 492
 493                 else if (t.cs() == "[") {
 494                         begin_inset(os, "Formula");
 495                         os << " \\[";
 496                         parse(p, os, FLAG_EQUATION, MATH_MODE);
 497                         os << "\\]";
 498                         end_inset(os);
 499                 }
 500
 501                 else if (t.cs() == "protect")
 502                         // ignore \\protect, will hopefully be re-added during output
 503                         ;
 504
 505                 else if (t.cs() == "begin") {
 506                         string const name = p.getArg('{', '}');
 507                         active_environments.push(name);
 508                         if (name == "document")
 509                                 end_preamble(os);
 510                         else if (name == "abstract")
 511                                 handle_par(os);
 512                         else if (is_math_env(name))
 513                                 begin_inset(os, "Formula ");
 514                         else
 515                                 os << "\\begin{" << name << "}";
 516                         parse(p, os, FLAG_END, mode);
 517                 }
 518
 519                 else if (t.cs() == "end") {
 520                         if (flags & FLAG_END) {
 521                                 // eat environment name
 522                                 string const name = p.getArg('{', '}');
 523                                 if (name != active_environments.top())
 524                                         p.error("\\end{" + name + "} does not match \\begin{"
 525                                                 + active_environments.top() + "}");
 526                                 active_environments.pop();
 527                                 if (name == "document" || name == "abstract")
 528                                         ;
 529                                 else if (is_math_env(name))
 530                                         end_inset(os);
 531                                 else
 532                                         os << "\\end{" << name << "}";
 533                                 return;
 534                         }
 535                         p.error("found 'end' unexpectedly");
 536                 }
 537
 538                 else if (t.cs() == ")") {
 539                         if (flags & FLAG_SIMPLE2)
 540                                 return;
 541                         p.error("found '\\)' unexpectedly");
 542                 }
 543
 544                 else if (t.cs() == "]") {
 545                         if (flags & FLAG_EQUATION)
 546                                 return;
 547                         p.error("found '\\]' unexpectedly");
 548                 }
 549
 550                 else if (t.cs() == "documentclass") {
 551                         vector<string> opts;
 552                         split(p.getArg('[', ']'), opts, ',');
 553                         handle_opt(opts, known_languages, h_language);
 554                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 555                         h_quotes_language = h_language;
 556                         h_options = join(opts, ',');
 557                         h_textclass = p.getArg('{', '}');
 558                 }
 559
 560                 else if (t.cs() == "usepackage") {
 561                         string const options = p.getArg('[', ']');
 562                         string const name = p.getArg('{', '}');
 563                         if (options.empty() && name.find(',')) {
 564                                 vector<string> vecnames;
 565                                 split(name, vecnames, ',');
 566                                 vector<string>::const_iterator it  = vecnames.begin();
 567                                 vector<string>::const_iterator end = vecnames.end();
 568                                 for (; it != end; ++it)
 569                                         handle_package(trim(*it), string());
 570                         } else {
 571                                 handle_package(name, options);
 572                         }
 573                 }
 574
 575                 else if (t.cs() == "newenvironment") {
 576                         string const name = p.getArg('{', '}');
 577                         p.skipSpaces();
 578                         string const begin = p.verbatimItem();
 579                         p.skipSpaces();
 580                         string const end = p.verbatimItem();
 581                         // ignore out mess
 582                         if (name != "lyxcode")
 583                                 os << wrap("newenvironment", begin + end);
 584                 }
 585
 586                 else if (t.cs() == "def") {
 587                         string name = p.getToken().cs();
 588                         while (p.nextToken().cat() != catBegin)
 589                                 name += p.getToken().asString();
 590                         handle_ert(os, "\\def\\" + name + '{' + p.verbatimItem() + '}');
 591                 }
 592
 593                 else if (t.cs() == "setcounter") {
 594                         string const name = p.getArg('{', '}');
 595                         string const content = p.getArg('{', '}');
 596                         if (name == "secnumdepth")
 597                                 h_secnumdepth = content;
 598                         else if (name == "tocdepth")
 599                                 h_tocdepth = content;
 600                         else
 601                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}\n";
 602                 }
 603
 604                 else if (t.cs() == "setlength") {
 605                         string const name = p.getToken().cs();
 606                         string const content = p.getArg('{', '}');
 607                         if (name == "parskip")
 608                                 h_paragraph_separation = "skip";
 609                         else if (name == "parindent")
 610                                 h_paragraph_separation = "skip";
 611                         else
 612                                 h_preamble << "\\setlength{" << name << "}{" << content << "}\n";
 613                 }
 614
 615                 else if (t.cs() == "par")
 616                         handle_par(os);
 617
 618                 else if (is_heading(t.cs())) {
 619                         os << "\\layout " << cap(t.cs()) << "\n\n";
 620                         parse(p, os, FLAG_ITEM, mode);
 621                 }
 622
 623                 else if (t.cs() == "makeindex" || t.cs() == "maketitle")
 624                         ; // swallow this
 625
 626                 else if (t.cs() == "tableofcontents")
 627                         p.verbatimItem(); // swallow this
 628
 629                 else if (t.cs() == "textrm") {
 630                         os << '\\' << t.cs() << '{';
 631                         parse(p, os, FLAG_ITEM, MATHTEXT_MODE);
 632                         os << '}';
 633                 }
 634
 635                 else if (t.cs() == "emph" && mode == TEXT_MODE) {
 636                         os << "\n\\emph on\n";
 637                         parse(p, os, FLAG_ITEM, mode);
 638                         os << "\n\\emph default\n";
 639                 }
 640
 641                 else if (t.cs() == "index") {
 642                         begin_inset(os, "LatexCommand ");
 643                         parse(p, os, FLAG_ITEM, TEXT_MODE);
 644                         end_inset(os);
 645                 }
 646
 647                 else
 648                         (in_preamble ? h_preamble : os) << '\\' << t.asInput();
 649
 650                 if (flags & FLAG_LEAVE) {
 651                         flags &= ~FLAG_LEAVE;
 652                         break;
 653                 }
 654         }
 655 }
 656
 657
 658
 659 } // anonymous namespace
 660
 661
 662 int main(int argc, char * argv[])
 663 {
 664         if (argc <= 1) {
 665                 cerr << "Usage: " << argv[0] << " <infile.tex>" << endl;
 666                 return 2;
 667         }
 668
 669         ifstream is(argv[1]);
 670         Parser p(is);
 671         parse(p, cout, 0, TEXT_MODE);
 672         cout << "\n\\the_end";
 673
 674         return 0;
 675 }
 676
 677 // }])