src/tex2lyx/preamble.cpp

   1 /**
   2  * \file preamble.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author André Pönitz
   7  * \author Uwe Stöhr
   8  *
   9  * Full author contact details are available in file CREDITS.
  10  */
  11
  12 // {[(
  13
  14 #include <config.h>
  15
  16 #include "tex2lyx.h"
  17
  18 #include "Layout.h"
  19 #include "Lexer.h"
  20 #include "TextClass.h"
  21 #include "support/filetools.h"
  22 #include "support/lstrings.h"
  23
  24 #include <algorithm>
  25 #include <iostream>
  26 #include <sstream>
  27 #include <string>
  28 #include <vector>
  29 #include <map>
  30
  31
  32 namespace lyx {
  33
  34 using std::find;
  35 using std::istringstream;
  36 using std::ostream;
  37 using std::ostringstream;
  38 using std::string;
  39 using std::vector;
  40 using std::cerr;
  41 using std::endl;
  42 using std::find;
  43
  44 using support::FileName;
  45 using support::libFileSearch;
  46
  47 // special columntypes
  48 extern std::map<char, int> special_columns;
  49
  50 std::map<string, vector<string> > used_packages;
  51
  52 namespace {
  53
  54 const char * const known_languages[] = { "afrikaans", "albanian", "american",
  55 "arabic_arabtex", "arabic_arabi", "armenian", "austrian", "bahasa", "basque",
  56 "brazilian", "breton", "british", "bulgarian", "canadian", "canadien",
  57 "catalan", "chinese-simplified", "chinese-traditional", "croatian", "czech",
  58 "danish", "dutch", "english", "esperanto", "estonian", "farsi", "finnish",
  59 "francais", "french", "galician", "german", "greek", "hebrew", "icelandic",
  60 "irish", "italian", "japanese", "japanese-plain", "kazakh", "korean", "latin",
  61 "latvian", "lithuanian", "lowersorbian", "magyar", "naustrian", "ngerman",
  62 "norsk", "nynorsk ", "polish", "portuges", "romanian", "russian", "samin",
  63 "scottish", "serbian", "slovak", "slovene", "spanish", "swedish", "thai",
  64 "turkish", "ukrainian", "uppersorbian", "vietnamese", "welsh", 0};
  65
  66 const char * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
  67
  68 const char * const known_encodings[] = { "auto", "latin1", "latin2", "latin3",
  69 "latin4", "latin5", "latin9", "latin10", "iso88595", "8859-6", "iso-8859-7",
  70 "8859-8", "l7xen", "cp437", "cp437de", "cp850", "cp852", "cp855", "cp858",
  71 "cp862", "cp865", "cp866", "cp1250", "cp1251", "cp1252", "cp1255", "cp1256",
  72 "cp1257", "koi8-r", "koi8-u", "pt154", "pt254", "utf8", 0};
  73
  74 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
  75 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
  76
  77 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
  78 "helvet", "lmss", 0};
  79
  80 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
  81 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
  82 "newcent", 0};
  83
  84 // some ugly stuff
  85 ostringstream h_preamble;
  86 string h_textclass                              = "article";
  87 string h_options                                = string();
  88 string h_language                               = "english";
  89 string h_inputencoding                  = "auto";
  90 string h_font_roman                             = "default";
  91 string h_font_sans                              = "default";
  92 string h_font_typewriter                = "default";
  93 string h_font_default_family    = "default";
  94 string h_font_sc                                = "false";
  95 string h_font_osf                               = "false";
  96 string h_font_sf_scale                  = "100";
  97 string h_font_tt_scale                  = "100";
  98 string h_graphics                               = "default";
  99 string h_paperfontsize                  = "default";
 100 string h_spacing                                = "single";
 101 string h_papersize                              = "default";
 102 string h_use_geometry                   = "false";
 103 string h_use_amsmath                    = "1";
 104 string h_use_esint                              = "1";
 105 string h_cite_engine                    = "basic";
 106 string h_use_bibtopic                   = "false";
 107 string h_paperorientation               = "portrait";
 108 string h_secnumdepth                    = "3";
 109 string h_tocdepth                               = "3";
 110 string h_paragraph_separation   = "indent";
 111 string h_defskip                                = "medskip";
 112 string h_quotes_language                = "english";
 113 string h_papercolumns                   = "1";
 114 string h_papersides                             = string();
 115 string h_paperpagestyle                 = "default";
 116 string h_tracking_changes               = "false";
 117 string h_output_changes                 = "false";
 118 string h_use_hyperref                   = "false";
 119
 120
 121 void handle_opt(vector<string> & opts, char const * const * what, string & target)
 122 {
 123         if (opts.empty())
 124                 return;
 125
 126         for (; *what; ++what) {
 127                 vector<string>::iterator it = find(opts.begin(), opts.end(), *what);
 128                 // the last language option is the document language
 129                 if (it != opts.end()) {
 130                         //cerr << "### found option '" << *what << "'\n";
 131                         target = *what;
 132                         opts.erase(it);
 133                 }
 134         }
 135 }
 136
 137
 138
 139 /*!
 140  * Split a package options string (keyval format) into a vector.
 141  * Example input:
 142  *   authorformat=smallcaps,
 143  *   commabeforerest,
 144  *   titleformat=colonsep,
 145  *   bibformat={tabular,ibidem,numbered}
 146  */
 147 vector<string> split_options(string const & input)
 148 {
 149         vector<string> options;
 150         string option;
 151         Parser p(input);
 152         while (p.good()) {
 153                 Token const & t = p.get_token();
 154                 if (t.asInput() == ",") {
 155                         options.push_back(option);
 156                         option.erase();
 157                 } else if (t.asInput() == "=") {
 158                         option += '=';
 159                         p.skip_spaces(true);
 160                         if (p.next_token().asInput() == "{")
 161                                 option += '{' + p.getArg('{', '}') + '}';
 162                 } else if (t.cat() != catSpace)
 163                         option += t.asInput();
 164         }
 165
 166         if (!option.empty())
 167                 options.push_back(option);
 168
 169         return options;
 170 }
 171
 172
 173 /*!
 174  * Add package \p name with options \p options to used_packages.
 175  * Remove options from \p options that we don't want to output.
 176  */
 177 void add_package(string const & name, vector<string> & options)
 178 {
 179         // every package inherits the global options
 180         if (used_packages.find(name) == used_packages.end())
 181                 used_packages[name] = split_options(h_options);
 182
 183         vector<string> & v = used_packages[name];
 184         v.insert(v.end(), options.begin(), options.end());
 185         if (name == "jurabib") {
 186                 // Don't output the order argument (see the cite command
 187                 // handling code in text.cpp).
 188                 vector<string>::iterator end =
 189                         remove(options.begin(), options.end(), "natbiborder");
 190                 end = remove(options.begin(), end, "jurabiborder");
 191                 options.erase(end, options.end());
 192         }
 193 }
 194
 195
 196 void handle_package(string const & name, string const & opts)
 197 {
 198         vector<string> options = split_options(opts);
 199         add_package(name, options);
 200         size_t pos;
 201         string scale;
 202
 203         // cerr << "handle_package: '" << name << "'\n";
 204
 205         // roman fonts
 206         if (is_known(name, known_roman_fonts))
 207                 h_font_roman = name;
 208         if (name == "fourier")
 209                 h_font_roman = "utopia";
 210         if (name == "mathpazo")
 211                 h_font_roman = "palatino";
 212         if (name == "mathptmx")
 213                 h_font_roman = "times";
 214         // sansserif fonts
 215         if (is_known(name, known_sans_fonts)) {
 216                 h_font_sans = name;
 217                 if (!opts.empty()) {
 218                         scale = opts;
 219                         pos = scale.find(".", 0);
 220                         h_font_sf_scale = scale.erase(0, pos + 1);
 221                 }
 222         }
 223         // typewriter fonts
 224         if (is_known(name, known_typewriter_fonts)) {
 225                 h_font_typewriter = name;
 226                 if (!opts.empty()) {
 227                         scale = opts;
 228                         pos = scale.find(".", 0);
 229                         h_font_tt_scale = scale.erase(0, pos + 1);
 230                 }
 231         }
 232
 233         else if (name == "amssymb")
 234                 h_use_amsmath = "2";
 235         else if (name == "esint")
 236                 h_use_esint = "2";
 237         else if (name == "babel")
 238                 ; // ignore this
 239         else if (name == "fontenc")
 240                 ; // ignore this
 241         else if (name == "inputenc") {
 242                 // only set when there are not more than one inputenc option
 243                 // therefore check for the "," character
 244                 if ((pos = opts.find(",", 0)) == string::npos)
 245                         h_inputencoding = opts;
 246                 options.clear();
 247         } else if (name == "makeidx")
 248                 ; // ignore this
 249         else if (name == "verbatim")
 250                 ; // ignore this
 251         else if (name == "graphicx")
 252                 ; // ignore this
 253         else if (is_known(name, known_languages)) {
 254                 h_language = name;
 255                 h_quotes_language = name;
 256         } else if (name == "natbib") {
 257                 h_cite_engine = "natbib_authoryear";
 258                 vector<string>::iterator it =
 259                         find(options.begin(), options.end(), "authoryear");
 260                 if (it != options.end())
 261                         options.erase(it);
 262                 else {
 263                         it = find(options.begin(), options.end(), "numbers");
 264                         if (it != options.end()) {
 265                                 h_cite_engine = "natbib_numerical";
 266                                 options.erase(it);
 267                         }
 268                 }
 269         } else if (name == "jurabib") {
 270                 h_cite_engine = "jurabib";
 271         } else if (options.empty())
 272                 h_preamble << "\\usepackage{" << name << "}\n";
 273         else {
 274                 h_preamble << "\\usepackage[" << opts << "]{" << name << "}\n";
 275                 options.clear();
 276         }
 277
 278         // We need to do something with the options...
 279         if (!options.empty())
 280                 cerr << "Ignoring options '" << join(options, ",")
 281                      << "' of package " << name << '.' << endl;
 282 }
 283
 284
 285 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 286 {
 287         os << "#LyX file created by  tex2lyx 0.1.5\n"
 288            << "\\lyxformat 245\n"
 289            << "\\begin_document\n"
 290            << "\\begin_header\n"
 291            << "\\textclass " << h_textclass << "\n";
 292         if (!h_preamble.str().empty())
 293                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 294         if (!h_options.empty())
 295                 os << "\\options " << h_options << "\n";
 296         os << "\\language " << h_language << "\n"
 297            << "\\inputencoding " << h_inputencoding << "\n"
 298            << "\\font_roman " << h_font_roman << "\n"
 299            << "\\font_sans " << h_font_sans << "\n"
 300            << "\\font_typewriter " << h_font_typewriter << "\n"
 301            << "\\font_default_family " << h_font_default_family << "\n"
 302            << "\\font_sc " << h_font_sc << "\n"
 303            << "\\font_osf " << h_font_osf << "\n"
 304            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 305            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 306            << "\\graphics " << h_graphics << "\n"
 307            << "\\paperfontsize " << h_paperfontsize << "\n"
 308            << "\\spacing " << h_spacing << "\n"
 309            << "\\use_hyperref " << h_use_hyperref << "\n"
 310            << "\\papersize " << h_papersize << "\n"
 311            << "\\use_geometry " << h_use_geometry << "\n"
 312            << "\\use_amsmath " << h_use_amsmath << "\n"
 313            << "\\use_esint " << h_use_esint << "\n"
 314            << "\\cite_engine " << h_cite_engine << "\n"
 315            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 316            << "\\paperorientation " << h_paperorientation << "\n"
 317            << "\\secnumdepth " << h_secnumdepth << "\n"
 318            << "\\tocdepth " << h_tocdepth << "\n"
 319            << "\\paragraph_separation " << h_paragraph_separation << "\n"
 320            << "\\defskip " << h_defskip << "\n"
 321            << "\\quotes_language " << h_quotes_language << "\n"
 322            << "\\papercolumns " << h_papercolumns << "\n"
 323            << "\\papersides " << h_papersides << "\n"
 324            << "\\paperpagestyle " << h_paperpagestyle << "\n"
 325            << "\\tracking_changes " << h_tracking_changes << "\n"
 326            << "\\output_changes " << h_output_changes << "\n"
 327            << "\\end_header\n\n"
 328            << "\\begin_body\n";
 329         // clear preamble for subdocuments
 330         h_preamble.str("");
 331 }
 332
 333 } // anonymous namespace
 334
 335
 336 TextClass const parse_preamble(Parser & p, ostream & os, string const & forceclass)
 337 {
 338         // initialize fixed types
 339         special_columns['D'] = 3;
 340         bool is_full_document = false;
 341
 342         // determine whether this is a full document or a fragment for inclusion
 343         while (p.good()) {
 344                 Token const & t = p.get_token();
 345
 346                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 347                         is_full_document = true;
 348                         break;
 349                 }
 350         }
 351         p.reset();
 352
 353         while (is_full_document && p.good()) {
 354                 Token const & t = p.get_token();
 355
 356 #ifdef FILEDEBUG
 357                 cerr << "t: " << t << "\n";
 358 #endif
 359
 360                 //
 361                 // cat codes
 362                 //
 363                 if (t.cat() == catLetter ||
 364                           t.cat() == catSuper ||
 365                           t.cat() == catSub ||
 366                           t.cat() == catOther ||
 367                           t.cat() == catMath ||
 368                           t.cat() == catActive ||
 369                           t.cat() == catBegin ||
 370                           t.cat() == catEnd ||
 371                           t.cat() == catAlign ||
 372                           t.cat() == catParameter)
 373                 ;//h_preamble << t.character();
 374
 375                 else if (t.cat() == catSpace || t.cat() == catNewline)
 376                         ;//h_preamble << t.asInput();
 377
 378                 else if (t.cat() == catComment)
 379                         ;//h_preamble << t.asInput();
 380
 381                 else if (t.cs() == "pagestyle")
 382                         h_paperpagestyle = p.verbatim_item();
 383
 384                 else if (t.cs() == "makeatletter")
 385                         p.setCatCode('@', catLetter);
 386
 387                 else if (t.cs() == "makeatother")
 388                         p.setCatCode('@', catOther);
 389
 390                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 391                             || t.cs() == "providecommand") {
 392                         bool star = false;
 393                         if (p.next_token().character() == '*') {
 394                                 p.get_token();
 395                                 star = true;
 396                         }
 397                         string const name = p.verbatim_item();
 398                         string const opt1 = p.getOpt();
 399                         string const opt2 = p.getFullOpt();
 400                         string const body = p.verbatim_item();
 401                         // font settings
 402                         if (name == "\\rmdefault")
 403                                 if (is_known(body, known_roman_fonts))
 404                                         h_font_roman = body;
 405
 406                         if (name == "\\sfdefault")
 407                                 if (is_known(body, known_sans_fonts))
 408                                         h_font_sans = body;
 409
 410                         if (name == "\\ttdefault")
 411                                 if (is_known(body, known_typewriter_fonts))
 412                                         h_font_typewriter = body;
 413
 414                         if (name == "\\familydefault") {
 415                                 string family = body;
 416                                 // remove leading "\"
 417                                 h_font_default_family = family.erase(0,1);
 418                         }
 419                         // only non-lyxspecific stuff
 420                         if (   name != "\\noun"
 421                             && name != "\\tabularnewline"
 422                             && name != "\\LyX"
 423                             && name != "\\lyxline"
 424                             && name != "\\lyxaddress"
 425                             && name != "\\lyxrightaddress"
 426                             && name != "\\lyxdot"
 427                             && name != "\\boldsymbol"
 428                             && name != "\\lyxarrow"
 429                                 && name != "\\rmdefault"
 430                                 && name != "\\sfdefault"
 431                                 && name != "\\ttdefault"
 432                                 && name != "\\familydefault") {
 433                                 ostringstream ss;
 434                                 ss << '\\' << t.cs();
 435                                 if (star)
 436                                         ss << '*';
 437                                 ss << '{' << name << '}' << opt1 << opt2
 438                                    << '{' << body << "}";
 439                                 h_preamble << ss.str();
 440
 441                                 // Add the command to the known commands
 442                                 add_known_command(name, opt1, !opt2.empty());
 443 /*
 444                                 ostream & out = in_preamble ? h_preamble : os;
 445                                 out << "\\" << t.cs() << "{" << name << "}"
 446                                     << opts << "{" << body << "}";
 447 */
 448                         }
 449
 450                 }
 451
 452                 else if (t.cs() == "documentclass") {
 453                         vector<string> opts;
 454                         split(p.getArg('[', ']'), opts, ',');
 455                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 456                         handle_opt(opts, known_languages, h_language);
 457                         // delete "pt" at the end
 458                         string::size_type i = h_paperfontsize.find("pt");
 459                         if (i != string::npos)
 460                                 h_paperfontsize.erase(i);
 461                         h_quotes_language = h_language;
 462                         h_options = join(opts, ",");
 463                         h_textclass = p.getArg('{', '}');
 464                 }
 465
 466                 else if (t.cs() == "usepackage") {
 467                         string const options = p.getArg('[', ']');
 468                         string const name = p.getArg('{', '}');
 469                         if (options.empty() && name.find(',')) {
 470                                 vector<string> vecnames;
 471                                 split(name, vecnames, ',');
 472                                 vector<string>::const_iterator it  = vecnames.begin();
 473                                 vector<string>::const_iterator end = vecnames.end();
 474                                 for (; it != end; ++it)
 475                                         handle_package(trim(*it), string());
 476                         } else {
 477                                 handle_package(name, options);
 478                         }
 479                 }
 480
 481                 else if (t.cs() == "newenvironment") {
 482                         string const name = p.getArg('{', '}');
 483                         ostringstream ss;
 484                         ss << "\\newenvironment{" << name << "}";
 485                         ss << p.getOpt();
 486                         ss << p.getOpt();
 487                         ss << '{' << p.verbatim_item() << '}';
 488                         ss << '{' << p.verbatim_item() << '}';
 489                         if (name != "lyxcode" && name != "lyxlist" &&
 490                             name != "lyxrightadress" &&
 491                             name != "lyxaddress" && name != "lyxgreyedout")
 492                                 h_preamble << ss.str();
 493                 }
 494
 495                 else if (t.cs() == "def") {
 496                         string name = p.get_token().cs();
 497                         while (p.next_token().cat() != catBegin)
 498                                 name += p.get_token().asString();
 499                         h_preamble << "\\def\\" << name << '{'
 500                                    << p.verbatim_item() << "}";
 501                 }
 502
 503                 else if (t.cs() == "newcolumntype") {
 504                         string const name = p.getArg('{', '}');
 505                         trim(name);
 506                         int nargs = 0;
 507                         string opts = p.getOpt();
 508                         if (!opts.empty()) {
 509                                 istringstream is(string(opts, 1));
 510                                 //cerr << "opt: " << is.str() << "\n";
 511                                 is >> nargs;
 512                         }
 513                         special_columns[name[0]] = nargs;
 514                         h_preamble << "\\newcolumntype{" << name << "}";
 515                         if (nargs)
 516                                 h_preamble << "[" << nargs << "]";
 517                         h_preamble << "{" << p.verbatim_item() << "}";
 518                 }
 519
 520                 else if (t.cs() == "setcounter") {
 521                         string const name = p.getArg('{', '}');
 522                         string const content = p.getArg('{', '}');
 523                         if (name == "secnumdepth")
 524                                 h_secnumdepth = content;
 525                         else if (name == "tocdepth")
 526                                 h_tocdepth = content;
 527                         else
 528                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
 529                 }
 530
 531                 else if (t.cs() == "setlength") {
 532                         string const name = p.verbatim_item();
 533                         string const content = p.verbatim_item();
 534                         // Is this correct?
 535                         if (name == "parskip")
 536                                 h_paragraph_separation = "skip";
 537                         else if (name == "parindent")
 538                                 h_paragraph_separation = "skip";
 539                         else
 540                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
 541                 }
 542
 543                 else if (t.cs() == "begin") {
 544                         string const name = p.getArg('{', '}');
 545                         if (name == "document")
 546                                 break;
 547                         h_preamble << "\\begin{" << name << "}";
 548                 }
 549
 550                 else if (t.cs() == "jurabibsetup") {
 551                         vector<string> jurabibsetup =
 552                                 split_options(p.getArg('{', '}'));
 553                         // add jurabibsetup to the jurabib package options
 554                         add_package("jurabib", jurabibsetup);
 555                         if (!jurabibsetup.empty()) {
 556                                 h_preamble << "\\jurabibsetup{"
 557                                            << join(jurabibsetup, ",") << '}';
 558                         }
 559                 }
 560
 561                 else if (!t.cs().empty())
 562                         h_preamble << '\\' << t.cs();
 563         }
 564         p.skip_spaces();
 565
 566         // Force textclass if the user wanted it
 567         if (!forceclass.empty())
 568                 h_textclass = forceclass;
 569         if (noweb_mode && !lyx::support::prefixIs(h_textclass, "literate-"))
 570                 h_textclass.insert(0, "literate-");
 571         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
 572         if (layoutfilename.empty()) {
 573                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
 574                 exit(1);
 575         }
 576         TextClass textclass;
 577         textclass.read(layoutfilename);
 578         if (h_papersides.empty()) {
 579                 ostringstream ss;
 580                 ss << textclass.sides();
 581                 h_papersides = ss.str();
 582         }
 583         end_preamble(os, textclass);
 584         return textclass;
 585 }
 586
 587
 588 // }])
 589
 590
 591 } // namespace lyx