src/tex2lyx/text.C

   1 /** The .tex to .lyx converter
   2     \author André Pönitz (2003)
   3  */
   4
   5 // {[(
   6
   7 #include <config.h>
   8
   9 #include "Lsstream.h"
  10 #include "tex2lyx.h"
  11
  12 #include <iostream>
  13 #include <map>
  14 #include <vector>
  15
  16 using std::cerr;
  17 using std::endl;
  18 using std::map;
  19 using std::ostream;
  20 using std::ostringstream;
  21 using std::vector;
  22
  23
  24 namespace {
  25
  26 char const * known_headings[] = { "caption", "title", "author", "date",
  27 "paragraph", "chapter", "section", "subsection", "subsubsection", 0 };
  28
  29 char const * known_latex_commands[] = { "ref", "cite", "label", "index",
  30 "printindex", "pageref", "url", 0 };
  31
  32 // LaTeX names for quotes
  33 char const * known_quotes[] = { "glqq", "grqq", "quotedblbase",
  34 "textquotedblleft", "quotesinglbase", "guilsinglleft", "guilsinglright", 0};
  35
  36 // the same as known_quotes with .lyx names
  37 char const * known_coded_quotes[] = { "gld", "grd", "gld",
  38 "grd", "gls", "fls", "frd", 0};
  39
  40 char const * known_sizes[] = { "tiny", "scriptsize", "footnotesize",
  41 "small", "normalsize", "large", "Large", "LARGE", "huge", "Huge", 0};
  42
  43 char const * known_coded_sizes[] = { "tiny", "scriptsize", "footnotesize",
  44 "small", "normal", "large", "larger", "largest",  "huge", "giant", 0};
  45
  46
  47 string cap(string s)
  48 {
  49         if (s.size())
  50                 s[0] = toupper(s[0]);
  51         return s;
  52 }
  53
  54
  55 // splits "x=z, y=b" into a map
  56 map<string, string> split_map(string const & s)
  57 {
  58         map<string, string> res;
  59         vector<string> v;
  60         split(s, v);
  61         for (size_t i = 0; i < v.size(); ++i) {
  62                 size_t const pos   = v[i].find('=');
  63                 string const index = v[i].substr(0, pos);
  64                 string const value = v[i].substr(pos + 1, string::npos);
  65                 res[trim(index)] = trim(value);
  66         }
  67         return res;
  68 }
  69
  70
  71 void begin_inset(ostream & os, string const & name)
  72 {
  73         os << "\n\\begin_inset " << name;
  74 }
  75
  76
  77 void end_inset(ostream & os)
  78 {
  79         os << "\n\\end_inset\n\n";
  80 }
  81
  82
  83 void skip_braces(Parser & p)
  84 {
  85         if (p.next_token().cat() != catBegin)
  86                 return;
  87         p.get_token();
  88         if (p.next_token().cat() == catEnd) {
  89                 p.get_token();
  90                 return;
  91         }
  92         p.putback();
  93 }
  94
  95
  96 void handle_ert(ostream & os, string const & s)
  97 {
  98         begin_inset(os, "ERT");
  99         os << "\nstatus Collapsed\n\n\\layout Standard\n\n";
 100         for (string::const_iterator it = s.begin(), et = s.end(); it != et; ++it) {
 101                 if (*it == '\\')
 102                         os << "\n\\backslash\n";
 103                 else
 104                         os << *it;
 105         }
 106         end_inset(os);
 107 }
 108
 109
 110 void handle_par(ostream & os)
 111 {
 112         if (active_environments.empty())
 113                 return;
 114         os << "\n\\layout ";
 115         string s = active_environment();
 116         if (s == "document" || s == "table")
 117                 os << "Standard\n\n";
 118         else if (s == "center")
 119                 os << "Standard\n\n\\align center\n";
 120         else if (s == "lyxcode")
 121                 os << "LyX-Code\n\n";
 122         else if (s == "lyxlist")
 123                 os << "List\n\n";
 124         else if (s == "thebibliography")
 125                 os << "Bibliography\n\n";
 126         else
 127                 os << cap(s) << "\n\n";
 128 }
 129
 130
 131 } // anonymous namespace
 132
 133
 134 void parse_text(Parser & p, ostream & os, unsigned flags, bool outer)
 135 {
 136         while (p.good()) {
 137                 Token const & t = p.get_token();
 138
 139 #ifdef FILEDEBUG
 140                 cerr << "t: " << t << " flags: " << flags << "\n";
 141 #endif
 142
 143                 if (flags & FLAG_ITEM) {
 144                         if (t.cat() == catSpace)
 145                                 continue;
 146
 147                         flags &= ~FLAG_ITEM;
 148                         if (t.cat() == catBegin) {
 149                                 // skip the brace and collect everything to the next matching
 150                                 // closing brace
 151                                 flags |= FLAG_BRACE_LAST;
 152                                 continue;
 153                         }
 154
 155                         // handle only this single token, leave the loop if done
 156                         flags |= FLAG_LEAVE;
 157                 }
 158
 159                 if (t.character() == ']' && (flags & FLAG_BRACK_LAST))
 160                         return;
 161
 162                 //
 163                 // cat codes
 164                 //
 165                 if (t.cat() == catMath) {
 166                         // we are inside some text mode thingy, so opening new math is allowed
 167                         begin_inset(os, "Formula ");
 168                         Token const & n = p.get_token();
 169                         if (n.cat() == catMath && outer) {
 170                                 // TeX's $$...$$ syntax for displayed math
 171                                 os << "\\[";
 172                                 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
 173                                 os << "\\]";
 174                                 p.get_token(); // skip the second '$' token
 175                         } else {
 176                                 // simple $...$  stuff
 177                                 p.putback();
 178                                 os << '$';
 179                                 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
 180                                 os << '$';
 181                         }
 182                         end_inset(os);
 183                 }
 184
 185                 else if (t.cat() == catSuper || t.cat() == catSub)
 186                         cerr << "catcode " << t << " illegal in text mode\n";
 187
 188                 else if (t.cat() == catLetter ||
 189                                t.cat() == catSpace ||
 190                                t.cat() == catOther ||
 191                                t.cat() == catAlign ||
 192                                t.cat() == catParameter)
 193                         os << t.character();
 194
 195                 else if (t.cat() == catNewline) {
 196                         if (p.next_token().cat() == catNewline) {
 197                                 p.get_token();
 198                                 handle_par(os);
 199                         } else {
 200                                 os << " "; // note the space
 201                         }
 202                 }
 203
 204                 else if (t.cat() == catActive) {
 205                         if (t.character() == '~') {
 206                                 if (active_environment() == "lyxcode")
 207                                         os << ' ';
 208                                 else
 209                                         os << "\\SpecialChar ~\n";
 210                         } else
 211                                 os << t.character();
 212                 }
 213
 214                 else if (t.cat() == catBegin) {
 215                         // special handling of size changes
 216                         bool const is_size = is_known(p.next_token().cs(), known_sizes);
 217                         string const s = parse_text(p, FLAG_BRACE_LAST, outer);
 218                         if (s.empty() && p.next_token().character() == '`')
 219                                 ; // ignore it in  {}``
 220                         else if (is_size || s == "[" || s == "]" || s == "*")
 221                                 os << s;
 222                         else {
 223                                 handle_ert(os, "{");
 224                                 os << s;
 225                                 handle_ert(os, "}");
 226                         }
 227                 }
 228
 229                 else if (t.cat() == catEnd) {
 230                         if (flags & FLAG_BRACE_LAST)
 231                                 return;
 232                         cerr << "stray '}' in text\n";
 233                         handle_ert(os, "}");
 234                 }
 235
 236                 else if (t.cat() == catOther)
 237                         os << string(1, t.character());
 238
 239                 else if (t.cat() == catComment)
 240                         handle_comment(p);
 241
 242                 //
 243                 // control sequences
 244                 //
 245
 246                 else if (t.cs() == "ldots") {
 247                         skip_braces(p);
 248                         os << "\n\\SpecialChar \\ldots{}\n";
 249                 }
 250
 251                 else if (t.cs() == "(") {
 252                         begin_inset(os, "Formula");
 253                         os << " \\(";
 254                         parse_math(p, os, FLAG_SIMPLE2, MATH_MODE);
 255                         os << "\\)";
 256                         end_inset(os);
 257                 }
 258
 259                 else if (t.cs() == "[") {
 260                         begin_inset(os, "Formula");
 261                         os << " \\[";
 262                         parse_math(p, os, FLAG_EQUATION, MATH_MODE);
 263                         os << "\\]";
 264                         end_inset(os);
 265                 }
 266
 267                 else if (t.cs() == "begin") {
 268                         string const name = p.getArg('{', '}');
 269                         active_environments.push_back(name);
 270                         if (is_math_env(name)) {
 271                                 begin_inset(os, "Formula ");
 272                                 os << "\\begin{" << name << "}";
 273                                 parse_math(p, os, FLAG_END, MATH_MODE);
 274                                 os << "\\end{" << name << "}";
 275                                 end_inset(os);
 276                         } else if (name == "tabular") {
 277                                 begin_inset(os, "Tabular ");
 278                                 handle_tabular(p, os);
 279                                 end_inset(os);
 280                         } else if (name == "table" || name == "figure") {
 281                                 string opts = p.getOpt();
 282                                 begin_inset(os, "Float " + name + "\n");
 283                                 if (opts.size())
 284                                         os << "placement " << opts << '\n';
 285                                 os << "wide false\ncollapsed false\n\n"
 286                                          << "\\layout Standard\n";
 287                                 parse_text(p, os, FLAG_END, outer);
 288                                 end_inset(os);
 289                         } else if (name == "center") {
 290                                 handle_par(os);
 291                                 parse_text(p, os, FLAG_END, outer);
 292                         } else if (name == "enumerate" || name == "itemize"
 293                                         || name == "lyxlist") {
 294                                 size_t const n = active_environments.size();
 295                                 string const s = active_environments[n - 2];
 296                                 bool const deeper = s == "enumerate" || s == "itemize"
 297                                         || s == "lyxlist";
 298                                 if (deeper)
 299                                         os << "\n\\begin_deeper";
 300                                 os << "\n\\layout " << cap(name) << "\n\n";
 301                                 if (name == "lyxlist")
 302                                         p.verbatim_item(); // swallow next arg
 303                                 parse_text(p, os, FLAG_END, outer);
 304                                 if (deeper)
 305                                         os << "\n\\end_deeper\n";
 306                                 handle_par(os);
 307                         } else if (name == "thebibliography") {
 308                                 p.verbatim_item(); // swallow next arg
 309                                 parse_text(p, os, FLAG_END, outer);
 310                                 os << "\n\\layout Bibliography\n\n";
 311                         } else {
 312                                 handle_par(os);
 313                                 parse_text(p, os, FLAG_END, outer);
 314                         }
 315                 }
 316
 317                 else if (t.cs() == "end") {
 318                         if (flags & FLAG_END) {
 319                                 // eat environment name
 320                                 string const name = p.getArg('{', '}');
 321                                 if (name != active_environment())
 322                                         cerr << "\\end{" + name + "} does not match \\begin{"
 323                                                 + active_environment() + "}\n";
 324                                 active_environments.pop_back();
 325                                 handle_par(os);
 326                                 return;
 327                         }
 328                         p.error("found 'end' unexpectedly");
 329                 }
 330
 331                 else if (t.cs() == "item") {
 332                         p.skip_spaces();
 333                         string s;
 334                         if (p.next_token().character() == '[') {
 335                                 p.get_token(); // eat '['
 336                                 s = parse_text(p, FLAG_BRACK_LAST, outer);
 337                         }
 338                         handle_par(os);
 339                         os << s << ' ';
 340                 }
 341
 342                 else if (t.cs() == "def") {
 343                         string name = p.get_token().cs();
 344                         while (p.next_token().cat() != catBegin)
 345                                 name += p.get_token().asString();
 346                         handle_ert(os, "\\def\\" + name + '{' + p.verbatim_item() + '}');
 347                 }
 348
 349                 else if (t.cs() == "par") {
 350                         p.skip_spaces();
 351                         if (p.next_token().cs() != "\\begin")
 352                                 handle_par(os);
 353                         //cerr << "next token: '" << p.next_token().cs() << "'\n";
 354                 }
 355
 356                 else if (is_known(t.cs(), known_headings)) {
 357                         string name = t.cs();
 358                         if (p.next_token().asInput() == "*") {
 359                                 p.get_token();
 360                                 name += "*";
 361                         }
 362                         os << "\n\n\\layout " << cap(name) << "\n\n";
 363                         string opt = p.getOpt();
 364                         if (opt.size()) {
 365                                 begin_inset(os, "OptArg\n");
 366                                 os << "collapsed true\n\n\\layout Standard\n\n" << opt;
 367                                 end_inset(os);
 368                         }
 369                         parse_text(p, os, FLAG_ITEM, outer);
 370                         os << "\n\n\\layout Standard\n\n";
 371                 }
 372
 373                 else if (t.cs() == "includegraphics") {
 374                         map<string, string> opts = split_map(p.getArg('[', ']'));
 375                         string name = p.verbatim_item();
 376                         begin_inset(os, "Graphics ");
 377                         os << "\n\tfilename " << name << '\n';
 378                         if (opts.find("width") != opts.end())
 379                                 os << "\twidth " << opts["width"] << '\n';
 380                         if (opts.find("height") != opts.end())
 381                                 os << "\theight " << opts["height"] << '\n';
 382                         end_inset(os);
 383                 }
 384
 385                 else if (t.cs() == "footnote") {
 386                         begin_inset(os, "Foot\n");
 387                         os << "collapsed true\n\n\\layout Standard\n\n";
 388                         parse_text(p, os, FLAG_ITEM, false);
 389                         end_inset(os);
 390                 }
 391
 392                 else if (t.cs() == "ensuremath") {
 393                         string s = parse_text(p, FLAG_ITEM, false);
 394                         if (s == "±" || s == "³" || s == "²" || s == "µ")
 395                                 os << s;
 396                         else
 397                                 handle_ert(os, "\\ensuremath{" + s + "}");
 398                 }
 399
 400                 else if (t.cs() == "marginpar") {
 401                         begin_inset(os, "Marginal\n");
 402                         os << "collapsed true\n\n\\layout Standard\n\n";
 403                         parse_text(p, os, FLAG_ITEM, false);
 404                         end_inset(os);
 405                 }
 406
 407                 else if (t.cs() == "hfill") {
 408                         os << "\n\\hfill\n";
 409                         skip_braces(p);
 410                 }
 411
 412                 else if (t.cs() == "makeindex" || t.cs() == "maketitle")
 413                         skip_braces(p); // swallow this
 414
 415                 else if (t.cs() == "tableofcontents")
 416                         skip_braces(p); // swallow this
 417
 418                 else if (t.cs() == "textrm") {
 419                         os << "\n\\family roman \n";
 420                         parse_text(p, os, FLAG_ITEM, outer);
 421                         os << "\n\\family default \n";
 422                 }
 423
 424                 else if (t.cs() == "textsf") {
 425                         os << "\n\\family sans \n";
 426                         parse_text(p, os, FLAG_ITEM, outer);
 427                         os << "\n\\family default \n";
 428                 }
 429
 430                 else if (t.cs() == "texttt") {
 431                         os << "\n\\family typewriter \n";
 432                         parse_text(p, os, FLAG_ITEM, outer);
 433                         os << "\n\\family default \n";
 434                 }
 435
 436                 else if (t.cs() == "textit") {
 437                         os << "\n\\shape italic \n";
 438                         parse_text(p, os, FLAG_ITEM, outer);
 439                         os << "\n\\shape default \n";
 440                 }
 441
 442                 else if (t.cs() == "textsc") {
 443                         os << "\n\\noun on \n";
 444                         parse_text(p, os, FLAG_ITEM, outer);
 445                         os << "\n\\noun default \n";
 446                 }
 447
 448                 else if (t.cs() == "textbf") {
 449                         os << "\n\\series bold \n";
 450                         parse_text(p, os, FLAG_ITEM, outer);
 451                         os << "\n\\series default \n";
 452                 }
 453
 454                 else if (t.cs() == "underbar") {
 455                         os << "\n\\bar under \n";
 456                         parse_text(p, os, FLAG_ITEM, outer);
 457                         os << "\n\\bar default \n";
 458                 }
 459
 460                 else if (t.cs() == "emph" || t.cs() == "noun") {
 461                         os << "\n\\" << t.cs() << " on \n";
 462                         parse_text(p, os, FLAG_ITEM, outer);
 463                         os << "\n\\" << t.cs() << " default \n";
 464                 }
 465
 466                 else if (t.cs() == "bibitem") {
 467                         os << "\n\\layout Bibliography\n\\bibitem ";
 468                         os << p.getOpt();
 469                         os << '{' << p.verbatim_item() << '}' << "\n\n";
 470                 }
 471
 472                 else if (is_known(t.cs(), known_latex_commands)) {
 473                         begin_inset(os, "LatexCommand ");
 474                         os << '\\' << t.cs();
 475                         os << p.getOpt();
 476                         os << p.getOpt();
 477                         os << '{' << p.verbatim_item() << '}';
 478                         end_inset(os);
 479                 }
 480
 481                 else if (is_known(t.cs(), known_quotes)) {
 482                   char const ** where = is_known(t.cs(), known_quotes);
 483                         begin_inset(os, "Quotes ");
 484                         os << known_coded_quotes[where - known_quotes];
 485                         end_inset(os);
 486                         skip_braces(p);
 487                 }
 488
 489                 else if (is_known(t.cs(), known_sizes)) {
 490                   char const ** where = is_known(t.cs(), known_sizes);
 491                         os << "\n\\size " << known_coded_sizes[where - known_sizes] << "\n";
 492                 }
 493
 494                 else if (t.cs() == "LyX" || t.cs() == "TeX"
 495                       || t.cs() == "LaTeX" || t.cs() == "LaTeXe") {
 496                         os << t.cs();
 497                         skip_braces(p); // eat {}
 498                 }
 499
 500                 else if (t.cs() == "lyxarrow") {
 501                         os << "\\SpecialChar \\menuseparator\n";
 502                         skip_braces(p);
 503                 }
 504
 505                 else if (t.cs() == "ldots") {
 506                         os << "\\SpecialChar \\ldots{}\n";
 507                         skip_braces(p);
 508                 }
 509
 510                 else if (t.cs() == "@") {
 511                         os << "\\SpecialChar \\@";
 512                         skip_braces(p);
 513                 }
 514
 515                 else if (t.cs() == "textasciitilde") {
 516                         os << '~';
 517                         skip_braces(p);
 518                 }
 519
 520                 else if (t.cs() == "textasciicircum") {
 521                         os << '^';
 522                         skip_braces(p);
 523                 }
 524
 525                 else if (t.cs() == "textbackslash") {
 526                         os << "\n\\backslash\n";
 527                         skip_braces(p);
 528                 }
 529
 530                 else if (t.cs() == "_" || t.cs() == "&" || t.cs() == "#" || t.cs() == "$"
 531                             || t.cs() == "{" || t.cs() == "}" || t.cs() == "%")
 532                         os << t.cs();
 533
 534                 else if (t.cs() == "char") {
 535                         if (p.next_token().character() == '`') {
 536                                 p.get_token();
 537                                 if (p.next_token().cs() == "\"") {
 538                                         p.get_token();
 539                                         os << '"';
 540                                         skip_braces(p);
 541                                 } else {
 542                                         handle_ert(os, "\\char`");
 543                                 }
 544                         } else {
 545                                 handle_ert(os, "\\char");
 546                         }
 547                 }
 548
 549                 else if (t.cs() == "\"") {
 550                         string const name = p.verbatim_item();
 551                              if (name == "a") os << 'ä';
 552                         else if (name == "o") os << 'ö';
 553                         else if (name == "u") os << 'ü';
 554                         else if (name == "A") os << 'Ä';
 555                         else if (name == "O") os << 'Ö';
 556                         else if (name == "U") os << 'Ü';
 557                         else handle_ert(os, "\"{" + name + "}");
 558                 }
 559
 560                 else if (t.cs() == "=" || t.cs() == "H" || t.cs() == "c"
 561                       || t.cs() == "^" || t.cs() == "'" || t.cs() == "~") {
 562                         // we need the trim as the LyX parser chokes on such spaces
 563                         os << "\n\\i \\" << t.cs() << "{"
 564                            << trim(parse_text(p, FLAG_ITEM, outer), " ") << "}\n";
 565                 }
 566
 567                 else if (t.cs() == "ss")
 568                         os << "ß";
 569
 570                 else if (t.cs() == "i" || t.cs() == "j")
 571                         os << "\\" << t.cs() << ' ';
 572
 573                 else if (t.cs() == "-")
 574                         os << "\\SpecialChar \\-\n";
 575
 576                 else if (t.cs() == "\\")
 577                         os << "\n\\newline\n";
 578
 579                 else if (t.cs() == "lyxrightaddress") {
 580                         os << "\n\\layout Right Address\n";
 581                         parse_text(p, os, FLAG_ITEM, outer);
 582                         os << "\n\\layout Standard\n";
 583                 }
 584
 585                 else if (t.cs() == "lyxaddress") {
 586                         os << "\n\\layout Address\n";
 587                         parse_text(p, os, FLAG_ITEM, outer);
 588                         os << "\n\\layout Standard\n";
 589                 }
 590
 591                 else if (t.cs() == "input")
 592                         handle_ert(os, "\\input{" + p.verbatim_item() + "}\n");
 593
 594                 else if (t.cs() == "fancyhead") {
 595                         ostringstream ss;
 596                         ss << "\\fancyhead";
 597                         ss << p.getOpt();
 598                         ss << '{' << p.verbatim_item() << "}\n";
 599                         handle_ert(os, STRCONV(ss.str()));
 600                 }
 601
 602                 else {
 603                         //cerr << "#: " << t << " mode: " << mode << endl;
 604                         // heuristic: read up to next non-nested space
 605                         /*
 606                         string s = t.asInput();
 607                         string z = p.verbatim_item();
 608                         while (p.good() && z != " " && z.size()) {
 609                                 //cerr << "read: " << z << endl;
 610                                 s += z;
 611                                 z = p.verbatim_item();
 612                         }
 613                         cerr << "found ERT: " << s << endl;
 614                         handle_ert(os, s + ' ');
 615                         */
 616                         handle_ert(os, t.asInput() + ' ');
 617                 }
 618
 619                 if (flags & FLAG_LEAVE) {
 620                         flags &= ~FLAG_LEAVE;
 621                         break;
 622                 }
 623         }
 624 }
 625
 626
 627 string parse_text(Parser & p, unsigned flags, const bool outer)
 628 {
 629         ostringstream os;
 630         parse_text(p, os, flags, outer);
 631         return STRCONV(os.str());
 632 }
 633
 634
 635 // }])