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