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::count_if;
  21 using std::cout;
  22 using std::cerr;
  23 using std::endl;
  24 using std::fill;
  25 using std::getline;
  26 using std::ios;
  27 using std::ifstream;
  28 using std::istream;
  29 using std::istringstream;
  30 using std::ostream;
  31 using std::ostringstream;
  32 using std::stack;
  33 using std::string;
  34 using std::vector;
  35
  36
  37 namespace {
  38
  39 char const OPEN = '<';
  40 char const CLOSE = '>';
  41 char const TAB = '\001';
  42 char const LINE = '\002';
  43
  44 const char * known_languages[] = { "austrian", "babel", "bahasa",
  45 "basque", "breton", "bulgarian", "catalan", "croatian", "czech", "danish",
  46 "dutch", "english", "esperanto", "estonian", "finnish", "francais",
  47 "frenchb", "galician", "germanb", "greek", "hebcal", "hebfont", "hebrew",
  48 "hebrew_newcode", "hebrew_oldcode", "hebrew_p", "hyphen", "icelandic",
  49 "irish", "italian", "latin", "lgrcmr", "lgrcmro", "lgrcmss", "lgrcmtt",
  50 "lgrenc", "lgrlcmss", "lgrlcmtt", "lheclas", "lhecmr", "lhecmss",
  51 "lhecmtt", "lhecrml", "lheenc", "lhefr", "lheredis", "lheshold",
  52 "lheshscr", "lheshstk", "lsorbian", "magyar", "naustrian", "ngermanb",
  53 "ngerman", "norsk", "polish", "portuges", "rlbabel", "romanian",
  54 "russianb", "samin", "scottish", "serbian", "slovak", "slovene", "spanish",
  55 "swedish", "turkish", "ukraineb", "usorbian", "welsh", 0};
  56
  57 const char * known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
  58
  59
  60 const char * known_math_envs[] = {"equation", "eqnarray", "eqnarray*",
  61 "align", "align*", 0};
  62
  63
  64 // some ugly stuff
  65 ostringstream h_preamble;
  66 string h_textclass               = "FIXME";
  67 string h_options                 = "FIXME";
  68 string h_language                = "FIXME";
  69 string h_inputencoding           = "FIXME";
  70 string h_fontscheme              = "FIXME";
  71 string h_graphics                = "default";
  72 string h_paperfontsize           = "FIXME";
  73 string h_spacing                 = "single";
  74 string h_papersize               = "FIXME";
  75 string h_paperpackage            = "FIXME";
  76 string h_use_geometry            = "0";
  77 string h_use_amsmath             = "0";
  78 string h_use_natbib              = "0";
  79 string h_use_numerical_citations = "0";
  80 string h_paperorientation        = "portrait";
  81 string h_secnumdepth             = "3";
  82 string h_tocdepth                = "3";
  83 string h_paragraph_separation    = "indent";
  84 string h_defskip                 = "medskip";
  85 string h_quotes_language         = "FIXME";
  86 string h_quotes_times            = "1";
  87 string h_papercolumns            = "1";
  88 string h_papersides              = "1";
  89 string h_paperpagestyle          = "default";
  90 string h_tracking_changes        = "0";
  91
  92 // indicates whether we are in the preamble
  93 bool in_preamble = true;
  94
  95 // current stack of nested environments
  96 stack<string> active_environments;
  97
  98
  99 string cap(string s)
 100 {
 101         if (s.size())
 102                 s[0] = toupper(s[0]);
 103         return s;
 104 }
 105
 106
 107 string const trim(string const & a, char const * p = " ")
 108 {
 109         // lyx::Assert(p);
 110
 111         if (a.empty() || !*p)
 112                 return a;
 113
 114         string::size_type r = a.find_last_not_of(p);
 115         string::size_type l = a.find_first_not_of(p);
 116
 117         // Is this the minimal test? (lgb)
 118         if (r == string::npos && l == string::npos)
 119                 return string();
 120
 121         return a.substr(l, r - l + 1);
 122 }
 123
 124
 125 void split(string const & s, vector<string> & result, char delim)
 126 {
 127         istringstream is(s);
 128         string t;
 129         while (getline(is, t, delim))
 130                 result.push_back(t);
 131 }
 132
 133
 134 string join(vector<string> const & input, char delim)
 135 {
 136         ostringstream os;
 137         for (size_t i = 0; i != input.size(); ++i) {
 138                 if (i)
 139                         os << delim;
 140                 os << input[i];
 141         }
 142         return os.str();
 143 }
 144
 145
 146 void handle_opt(vector<string> & opts, char const ** what, string & target)
 147 {
 148         if (opts.empty())
 149                 return;
 150
 151         for ( ; what; ++what) {
 152                 vector<string>::iterator it = find(opts.begin(), opts.end(), *what);
 153                 if (it != opts.end()) {
 154                         //cerr << "### found option '" << *what << "'\n";
 155                         target = *what;
 156                         opts.erase(it);
 157                         return;
 158                 }
 159         }
 160 }
 161
 162
 163 bool is_math_env(string const & name)
 164 {
 165         for (char const ** what = known_math_envs; *what; ++what)
 166                 if (*what == name)
 167                         return true;
 168         return false;
 169 }
 170
 171
 172 bool is_heading(string const & name)
 173 {
 174         return
 175                 name == "caption" ||
 176                 name == "title" ||
 177                 name == "author" ||
 178                 name == "paragraph" ||
 179                 name == "chapter" ||
 180                 name == "chapter*" ||
 181                 name == "section" ||
 182                 name == "section*" ||
 183                 name == "subsection" ||
 184                 name == "subsection*" ||
 185                 name == "subsubsection" ||
 186                 name == "subsubsection*";
 187 }
 188
 189
 190 bool is_latex_command(string const & name)
 191 {
 192         return
 193                 name == "ref" ||
 194                 name == "cite" ||
 195                 name == "label" ||
 196                 name == "index" ||
 197                 name == "printindex";
 198 }
 199
 200
 201 void begin_inset(ostream & os, string const & name)
 202 {
 203         os << "\n\\begin_inset " << name;
 204 }
 205
 206
 207 void end_inset(ostream & os)
 208 {
 209         os << "\n\\end_inset\n";
 210 }
 211
 212
 213 string curr_env()
 214 {
 215         return active_environments.empty() ? string() : active_environments.top();
 216 }
 217
 218
 219 void handle_ert(ostream & os, string const & s)
 220 {
 221         begin_inset(os, "ERT");
 222         os << "\nstatus Collapsed\n\n\\layout Standard\n\n";
 223         for (string::const_iterator it = s.begin(), et = s.end(); it != et; ++it) {
 224                 if (*it == '\\')
 225                         os << "\n\\backslash\n";
 226                 else
 227                         os << *it;
 228         }
 229         end_inset(os);
 230 }
 231
 232
 233 void handle_par(ostream & os)
 234 {
 235         if (active_environments.empty())
 236                 return;
 237         os << "\n\\layout ";
 238         string s = curr_env();
 239         if (s == "document" || s == "table") {
 240                 os << "Standard\n\n";
 241                 return;
 242         }
 243         if (s == "lyxcode") {
 244                 os << "LyX-Code\n\n";
 245                 return;
 246         }
 247         os << cap(s) << "\n\n";
 248 }
 249
 250
 251 void handle_package(string const & name, string const & options)
 252 {
 253         if (name == "a4wide") {
 254                 h_papersize = "a4paper";
 255                 h_paperpackage = "widemarginsa4";
 256         } else if (name == "ae")
 257                 h_fontscheme = "ae";
 258         else if (name == "aecompl")
 259                 h_fontscheme = "ae";
 260         else if (name == "amsmath")
 261                 h_use_amsmath = "1";
 262         else if (name == "amssymb")
 263                 h_use_amsmath = "1";
 264         else if (name == "babel")
 265                 ; // ignore this
 266         else if (name == "fontenc")
 267                 ; // ignore this
 268         else if (name == "inputenc")
 269                 h_inputencoding = options;
 270         else if (name == "makeidx")
 271                 ; // ignore this
 272         else if (name == "verbatim")
 273                 ; // ignore this
 274         else {
 275                 if (options.size())
 276                         h_preamble << "\\usepackage[" << options << "]{" << name << "}\n";
 277                 else
 278                         h_preamble << "\\usepackage{" << name << "}\n";
 279         }
 280 }
 281
 282
 283 void handle_table(Parser &, ostream &)
 284 {
 285         // \begin{table} has been read
 286         //parse(end
 287 }
 288
 289
 290 string wrap(string const & cmd, string const & str)
 291 {
 292         return OPEN + cmd + ' ' + str + CLOSE;
 293 }
 294
 295
 296 vector<string> extract_col_align(string const & s)
 297 {
 298         vector<string> res;
 299         for (size_t i = 0; i < s.size(); ++i) {
 300                 switch (s[i]) {
 301                         case 'c':
 302                                 res.push_back("center");
 303                                 break;
 304                         case 'l':
 305                                 res.push_back("left");
 306                                 break;
 307                         case 'r':
 308                                 res.push_back("right");
 309                                 break;
 310                         default:
 311                                 res.push_back("right");
 312                                 break;
 313                 }
 314         }
 315         return res;
 316 }
 317
 318
 319 void end_preamble(ostream & os)
 320 {
 321         in_preamble = false;
 322         os << "# tex2lyx 0.0.2 created this file\n"
 323            << "\\lyxformat 222\n"
 324            << "\\textclass " << h_textclass << "\n"
 325            << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n"
 326            << "\\options " << h_options << "\n"
 327            << "\\language " << h_language << "\n"
 328            << "\\inputencoding " << h_inputencoding << "\n"
 329            << "\\fontscheme " << h_fontscheme << "\n"
 330            << "\\graphics " << h_graphics << "\n"
 331            << "\\paperfontsize " << h_paperfontsize << "\n"
 332            << "\\spacing " << h_spacing << "\n"
 333            << "\\papersize " << h_papersize << "\n"
 334            << "\\paperpackage " << h_paperpackage << "\n"
 335            << "\\use_geometry " << h_use_geometry << "\n"
 336            << "\\use_amsmath " << h_use_amsmath << "\n"
 337            << "\\use_natbib " << h_use_natbib << "\n"
 338            << "\\use_numerical_citations " << h_use_numerical_citations << "\n"
 339            << "\\paperorientation " << h_paperorientation << "\n"
 340            << "\\secnumdepth " << h_secnumdepth << "\n"
 341            << "\\tocdepth " << h_tocdepth << "\n"
 342            << "\\paragraph_separation " << h_paragraph_separation << "\n"
 343            << "\\defskip " << h_defskip << "\n"
 344            << "\\quotes_language " << h_quotes_language << "\n"
 345            << "\\quotes_times " << h_quotes_times << "\n"
 346            << "\\papercolumns " << h_papercolumns << "\n"
 347            << "\\papersides " << h_papersides << "\n"
 348            << "\\paperpagestyle " << h_paperpagestyle << "\n"
 349            << "\\tracking_changes " << h_tracking_changes << "\n";
 350 }
 351
 352
 353 void parse(Parser & p, ostream & os, unsigned flags, mode_type mode)
 354 {
 355         while (p.good()) {
 356                 Token const & t = p.getToken();
 357
 358 #ifdef FILEDEBUG
 359                 cerr << "t: " << t << " flags: " << flags << "\n";
 360                 //cell->dump();
 361 #endif
 362
 363                 if (flags & FLAG_ITEM) {
 364                         if (t.cat() == catSpace)
 365                                 continue;
 366
 367                         flags &= ~FLAG_ITEM;
 368                         if (t.cat() == catBegin) {
 369                                 // skip the brace and collect everything to the next matching
 370                                 // closing brace
 371                                 flags |= FLAG_BRACE_LAST;
 372                                 continue;
 373                         }
 374
 375                         // handle only this single token, leave the loop if done
 376                         flags |= FLAG_LEAVE;
 377                 }
 378
 379
 380                 if (flags & FLAG_BRACED) {
 381                         if (t.cat() == catSpace)
 382                                 continue;
 383
 384                         if (t.cat() != catBegin) {
 385                                 p.error("opening brace expected");
 386                                 return;
 387                         }
 388
 389                         // skip the brace and collect everything to the next matching
 390                         // closing brace
 391                         flags = FLAG_BRACE_LAST;
 392                 }
 393
 394
 395                 if (flags & FLAG_OPTION) {
 396                         if (t.cat() == catOther && t.character() == '[') {
 397                                 parse(p, os, FLAG_BRACK_LAST, mode);
 398                         } else {
 399                                 // no option found, put back token and we are done
 400                                 p.putback();
 401                         }
 402                         return;
 403                 }
 404
 405                 //
 406                 // cat codes
 407                 //
 408                 if (t.cat() == catMath) {
 409                         if (mode == TEXT_MODE || mode == MATHTEXT_MODE) {
 410                                 // we are inside some text mode thingy, so opening new math is allowed
 411                                 if (mode == TEXT_MODE)
 412                                         begin_inset(os, "Formula ");
 413                                 Token const & n = p.getToken();
 414                                 if (n.cat() == catMath) {
 415                                         // TeX's $$...$$ syntax for displayed math
 416                                         os << "\\[";
 417                                         parse(p, os, FLAG_SIMPLE, MATH_MODE);
 418                                         os << "\\]";
 419                                         p.getToken(); // skip the second '$' token
 420                                 } else {
 421                                         // simple $...$  stuff
 422                                         p.putback();
 423                                         os << '$';
 424                                         parse(p, os, FLAG_SIMPLE, MATH_MODE);
 425                                         os << '$';
 426                                 }
 427                                 if (mode == TEXT_MODE)
 428                                         end_inset(os);
 429                         }
 430
 431                         else if (flags & FLAG_SIMPLE) {
 432                                 // this is the end of the formula
 433                                 return;
 434                         }
 435
 436                         else {
 437                                 cerr << "mode: " << mode << endl;
 438                                 p.error("something strange in the parser\n");
 439                                 break;
 440                         }
 441                 }
 442
 443                 else if (t.cat() == catLetter)
 444                         os << t.character();
 445
 446                 else if (t.cat() == catSpace)
 447                         os << t.character();
 448
 449                 else if (t.cat() == catNewline)
 450                         os << ' ';
 451
 452                 else if (t.cat() == catSuper)
 453                         os << t.character();
 454
 455                 else if (t.cat() == catSub)
 456                         os << t.character();
 457
 458                 else if (t.cat() == catParameter) {
 459                         Token const & n = p.getToken();
 460                         os << wrap("macroarg", string(1, n.character()));
 461                 }
 462
 463                 else if (t.cat() == catActive) {
 464                         if (t.character() == '~')
 465                                 os << (curr_env() == "lyxcode" ? ' ' : '~');
 466                         else
 467                                 os << t.asInput();
 468                 }
 469
 470                 else if (t.cat() == catBegin) {
 471                         if (mode == MATH_MODE)
 472                                 os << '{';
 473                         else
 474                                 handle_ert(os, "{");
 475                 }
 476
 477                 else if (t.cat() == catEnd) {
 478                         if (flags & FLAG_BRACE_LAST)
 479                                 return;
 480                         if (mode == MATH_MODE)
 481                                 os << '}';
 482                         else
 483                                 handle_ert(os, "}");
 484                 }
 485
 486                 else if (t.cat() == catAlign) {
 487                         if (mode == MATH_MODE)
 488                                 os << t.character();
 489                         else
 490                                 os << TAB;
 491                 }
 492
 493                 else if (t.cs() == "tabularnewline") {
 494                         if (mode == MATH_MODE)
 495                                 os << t.asInput();
 496                         else
 497                                 os << LINE;
 498                 }
 499
 500                 else if (t.character() == ']' && (flags & FLAG_BRACK_LAST)) {
 501                         //cerr << "finished reading option\n";
 502                         return;
 503                 }
 504
 505                 else if (t.cat() == catOther)
 506                         os << string(1, t.character());
 507
 508                 else if (t.cat() == catComment) {
 509                         string s;
 510                         while (p.good()) {
 511                                 Token const & t = p.getToken();
 512                                 if (t.cat() == catNewline)
 513                                         break;
 514                                 s += t.asString();
 515                         }
 516                         //os << wrap("comment", s);
 517                         p.skipSpaces();
 518                 }
 519
 520                 //
 521                 // control sequences
 522                 //
 523
 524                 else if (t.cs() == "lyxlock")
 525                         ; // ignored
 526
 527                 else if (t.cs() == "newcommand" || t.cs() == "providecommand") {
 528                         string const name = p.verbatimItem();
 529                         string const opts = p.getArg('[', ']');
 530                         string const body = p.verbatimItem();
 531                         // only non-lyxspecific stuff
 532                         if (name != "noun" && name != "tabularnewline") {
 533                                 ostream & out = in_preamble ? h_preamble : os;
 534                                 if (!in_preamble)
 535                                         begin_inset(os, "FormulaMacro\n");
 536                                 out << "\\" << t.cs() << "{" << name << "}";
 537                                 if (opts.size())
 538                                         out << "[" << opts << "]";
 539                                 out << "{" << body << "}";
 540                                 if (!in_preamble)
 541                                         end_inset(os);
 542                         }
 543                 }
 544
 545                 else if (t.cs() == "(") {
 546                         begin_inset(os, "Formula");
 547                         os << " \\(";
 548                         parse(p, os, FLAG_SIMPLE2, MATH_MODE);
 549                         os << "\\)";
 550                         end_inset(os);
 551                 }
 552
 553                 else if (t.cs() == "[") {
 554                         begin_inset(os, "Formula");
 555                         os << " \\[";
 556                         parse(p, os, FLAG_EQUATION, MATH_MODE);
 557                         os << "\\]";
 558                         end_inset(os);
 559                 }
 560
 561                 else if (t.cs() == "protect")
 562                         // ignore \\protect, will hopefully be re-added during output
 563                         ;
 564
 565                 else if (t.cs() == "begin") {
 566                         string const name = p.getArg('{', '}');
 567                         active_environments.push(name);
 568                         if (name == "document") {
 569                                 end_preamble(os);
 570                                 parse(p, os, FLAG_END, mode);
 571                         } else if (name == "abstract") {
 572                                 handle_par(os);
 573                                 parse(p, os, FLAG_END, mode);
 574                         } else if (is_math_env(name)) {
 575                                 begin_inset(os, "Formula ");
 576                                 os << "\\begin{" << name << "}";
 577                                 parse(p, os, FLAG_END, MATH_MODE);
 578                                 os << "\\end{" << name << "}";
 579                                 end_inset(os);
 580                         } else if (name == "tabular") {
 581                                 begin_inset(os, "Tabular \n");
 582                                 string colopts = p.verbatimItem();
 583                                 vector<string> colalign = extract_col_align(colopts);
 584                                 ostringstream ss;
 585                                 parse(p, ss, FLAG_END, mode);
 586                                 vector<string> lines;
 587                                 split(ss.str(), lines, LINE);
 588                                 const size_t cols = colalign.size();
 589                                 const size_t rows = lines.size();
 590                                 os << "<lyxtabular version=\"3\" rows=\"" << rows
 591                                    << "\" columns=\"" << cols << "\">\n"
 592                                    << "<features>\n";
 593                                 for (size_t c = 0; c < cols; ++c)
 594                                         os << "<column alignment=\"" << colalign[c] << "\""
 595                                            << " valignment=\"top\""
 596                                            << " width=\"0pt\""
 597                                            << ">\n";
 598                                 for (size_t r = 0; r < rows; ++r) {
 599                                         vector<string> cells;
 600                                         split(lines[r], cells, TAB);
 601                                         while (cells.size() < cols)
 602                                                 cells.push_back(string());
 603                                         //os << "<row bottomline=\"true\">\n";
 604                                         os << "<row>\n";
 605                                         for (size_t c = 0; c < cols; ++c) {
 606                                                 os << "<cell alignment=\"center\""
 607                                                    << " valignment=\"top\""
 608                                                    << " topline=\"true\""
 609                                                    << " leftline=\"true\""
 610                                                    << " usebox=\"none\""
 611                                                    << ">";
 612                                                 begin_inset(os, "Text");
 613                                                 os << "\n\n\\layout Standard\n\n";
 614                                                 os << cells[c];
 615                                                 end_inset(os);
 616                                                 os << "</cell>\n";
 617                                         }
 618                                         os << "</row>\n";
 619                                 }
 620                                 os << "</lyxtabular>\n";
 621                                 end_inset(os);
 622                         } else if (name == "table") {
 623                                 begin_inset(os, "Float table\n");
 624                                 os << "wide false\n"
 625                                    << "collapsed false\n"
 626                                    << "\n"
 627                                    << "\\layout Standard\n";
 628                                 parse(p, os, FLAG_END, mode);
 629                                 end_inset(os);
 630                         } else if (name == "thebibliography") {
 631                                 p.verbatimItem(); // swallow next arg
 632                                 parse(p, os, FLAG_END, mode);
 633                                 os << "\n\\layout Standard\n\n";
 634                         } else if (mode == MATH_MODE) {
 635                                 os << "\\begin{" << name << "}";
 636                                 parse(p, os, FLAG_END, mode);
 637                                 os << "\\end{" << name << "}";
 638                         } else {
 639                                 parse(p, os, FLAG_END, mode);
 640                         }
 641                 }
 642
 643                 else if (t.cs() == "end") {
 644                         if (flags & FLAG_END) {
 645                                 // eat environment name
 646                                 string const name = p.getArg('{', '}');
 647                                 if (name != curr_env())
 648                                         p.error("\\end{" + name + "} does not match \\begin{"
 649                                                 + curr_env() + "}");
 650                                 active_environments.pop();
 651                                 return;
 652                         }
 653                         p.error("found 'end' unexpectedly");
 654                 }
 655
 656                 else if (t.cs() == "item")
 657                         handle_par(os);
 658
 659                 else if (t.cs() == ")") {
 660                         if (flags & FLAG_SIMPLE2)
 661                                 return;
 662                         p.error("found '\\)' unexpectedly");
 663                 }
 664
 665                 else if (t.cs() == "]") {
 666                         if (flags & FLAG_EQUATION)
 667                                 return;
 668                         p.error("found '\\]' unexpectedly");
 669                 }
 670
 671                 else if (t.cs() == "documentclass") {
 672                         vector<string> opts;
 673                         split(p.getArg('[', ']'), opts, ',');
 674                         handle_opt(opts, known_languages, h_language);
 675                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 676                         h_quotes_language = h_language;
 677                         h_options = join(opts, ',');
 678                         h_textclass = p.getArg('{', '}');
 679                 }
 680
 681                 else if (t.cs() == "usepackage") {
 682                         string const options = p.getArg('[', ']');
 683                         string const name = p.getArg('{', '}');
 684                         if (options.empty() && name.find(',')) {
 685                                 vector<string> vecnames;
 686                                 split(name, vecnames, ',');
 687                                 vector<string>::const_iterator it  = vecnames.begin();
 688                                 vector<string>::const_iterator end = vecnames.end();
 689                                 for (; it != end; ++it)
 690                                         handle_package(trim(*it), string());
 691                         } else {
 692                                 handle_package(name, options);
 693                         }
 694                 }
 695
 696                 else if (t.cs() == "newenvironment") {
 697                         string const name = p.getArg('{', '}');
 698                         p.skipSpaces();
 699                         string const begin = p.verbatimItem();
 700                         p.skipSpaces();
 701                         string const end = p.verbatimItem();
 702                         // ignore out mess
 703                         if (name != "lyxcode")
 704                                 os << wrap("newenvironment", begin + end);
 705                 }
 706
 707                 else if (t.cs() == "def") {
 708                         string name = p.getToken().cs();
 709                         while (p.nextToken().cat() != catBegin)
 710                                 name += p.getToken().asString();
 711                         handle_ert(os, "\\def\\" + name + '{' + p.verbatimItem() + '}');
 712                 }
 713
 714                 else if (t.cs() == "setcounter") {
 715                         string const name = p.getArg('{', '}');
 716                         string const content = p.getArg('{', '}');
 717                         if (name == "secnumdepth")
 718                                 h_secnumdepth = content;
 719                         else if (name == "tocdepth")
 720                                 h_tocdepth = content;
 721                         else
 722                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}\n";
 723                 }
 724
 725                 else if (t.cs() == "setlength") {
 726                         string const name = p.getToken().cs();
 727                         string const content = p.getArg('{', '}');
 728                         if (name == "parskip")
 729                                 h_paragraph_separation = "skip";
 730                         else if (name == "parindent")
 731                                 h_paragraph_separation = "skip";
 732                         else
 733                                 h_preamble << "\\setlength{" << name << "}{" << content << "}\n";
 734                 }
 735
 736                 else if (t.cs() == "par")
 737                         handle_par(os);
 738
 739                 else if (is_heading(t.cs())) {
 740                         os << "\\layout " << cap(t.cs()) << "\n\n";
 741                         parse(p, os, FLAG_ITEM, mode);
 742                 }
 743
 744                 else if (t.cs() == "makeindex" || t.cs() == "maketitle")
 745                         ; // swallow this
 746
 747                 else if (t.cs() == "tableofcontents")
 748                         p.verbatimItem(); // swallow this
 749
 750                 else if (t.cs() == "textrm") {
 751                         os << '\\' << t.cs() << '{';
 752                         parse(p, os, FLAG_ITEM, MATHTEXT_MODE);
 753                         os << '}';
 754                 }
 755
 756                 else if ((t.cs() == "emph" || t.cs() == "noun") && mode == TEXT_MODE) {
 757                         os << "\n\\" << t.cs() << " on\n";
 758                         parse(p, os, FLAG_ITEM, mode);
 759                         os << "\n\\" << t.cs() << " default\n";
 760                 }
 761
 762                 else if (is_latex_command(t.cs()) && mode == TEXT_MODE) {
 763                         begin_inset(os, "LatexCommand ");
 764                         os << '\\' << t.cs() << '{';
 765                         parse(p, os, FLAG_ITEM, TEXT_MODE);
 766                         os << '}';
 767                         end_inset(os);
 768                 }
 769
 770                 else if (t.cs() == "bibitem") {
 771                         os << "\n\\layout Bibliography\n\\bibitem ";
 772                         string opt = p.getArg('[',']');
 773                         if (opt.size())
 774                                 os << '[' << opt << ']';
 775                         os << '{' << p.getArg('{','}') << '}' << "\n\n";
 776                 }
 777
 778                 else if (t.cs() == "textasciitilde")
 779                         os << '~';
 780
 781                 else if (t.cs() == "_" && mode == TEXT_MODE)
 782                         os << '_';
 783
 784                 else if (t.cs() == "&" && mode == TEXT_MODE)
 785                         os << '&';
 786
 787                 else {
 788                         if (mode == MATH_MODE)
 789                                 os << t.asInput();
 790                         else if (in_preamble)
 791                                 h_preamble << t.asInput();
 792                         else {
 793                                 // heuristic: read up to next non-nested space
 794                                 /*
 795                                 string s = t.asInput();
 796                                 string z = p.verbatimItem();
 797                                 while (p.good() && z != " " && z.size()) {
 798                                         //cerr << "read: " << z << endl;
 799                                         s += z;
 800                                         z = p.verbatimItem();
 801                                 }
 802                                 cerr << "found ERT: " << s << endl;
 803                                 handle_ert(os, s + ' ');
 804                                 */
 805                                 handle_ert(os, t.asInput() + ' ');
 806                         }
 807                 }
 808
 809                 if (flags & FLAG_LEAVE) {
 810                         flags &= ~FLAG_LEAVE;
 811                         break;
 812                 }
 813         }
 814 }
 815
 816
 817
 818 } // anonymous namespace
 819
 820
 821 int main(int argc, char * argv[])
 822 {
 823         if (argc <= 1) {
 824                 cerr << "Usage: " << argv[0] << " <infile.tex>" << endl;
 825                 return 2;
 826         }
 827
 828         ifstream is(argv[1]);
 829         Parser p(is);
 830         parse(p, cout, 0, TEXT_MODE);
 831         cout << "\n\\the_end";
 832
 833         return 0;
 834 }
 835
 836 // }])