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::istringstream;
  35 using std::ostream;
  36 using std::ostringstream;
  37 using std::string;
  38 using std::vector;
  39 using std::cerr;
  40 using std::endl;
  41 using std::find;
  42
  43 using support::FileName;
  44 using support::libFileSearch;
  45
  46 // special columntypes
  47 extern std::map<char, int> special_columns;
  48
  49 std::map<string, vector<string> > used_packages;
  50
  51 namespace {
  52
  53 const char * const known_languages[] = { "afrikaans", "american", "austrian",
  54 "babel", "bahasa", "basque", "belarusian", "brazil", "breton", "british",
  55 "bulgarian", "catalan", "croatian", "czech", "danish", "dutch", "english",
  56 "esperanto", "estonian", "finnish", "francais", "french", "frenchb",
  57 "frenchle", "frenchpro", "galician", "german", "germanb", "greek", "hebcal",
  58 "hebfont", "hebrew", "hebrew_newcode", "hebrew_oldcode", "hebrew_p", "hyphen",
  59 "icelandic", "irish", "italian", "latin", "lgrcmr", "lgrcmro", "lgrcmss",
  60 "lgrcmtt", "lgrenc", "lgrlcmss", "lgrlcmtt", "lheclas", "lhecmr", "lhecmss",
  61 "lhecmtt", "lhecrml", "lheenc", "lhefr", "lheredis", "lheshold", "lheshscr",
  62 "lheshstk", "lsorbian", "magyar", "naustrian", "ngermanb", "ngerman", "norsk",
  63 "nynorsk", "polish", "portuges", "rlbabel", "romanian", "russian", "russianb",
  64 "samin", "scottish", "serbian", "slovak", "slovene", "spanish", "swedish",
  65 "thai", "turkish", "ukraineb", "ukrainian", "usorbian", "welsh", 0};
  66
  67 const char * const known_french_languages[] = {"french", "frenchb", "francais",
  68                                                "frenchle", "frenchpro", 0};
  69 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
  70
  71 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
  72 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
  73
  74 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
  75 "helvet", "lmss", 0};
  76
  77 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
  78 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
  79 "newcent", 0};
  80
  81 // some ugly stuff
  82 ostringstream h_preamble;
  83 string h_textclass               = "article";
  84 string h_options                 = string();
  85 string h_language                = "english";
  86 string h_inputencoding           = "auto";
  87 string h_font_roman              = "default";
  88 string h_font_sans               = "default";
  89 string h_font_typewriter         = "default";
  90 string h_font_default_family     = "default";
  91 string h_font_sc                 = "false";
  92 string h_font_osf                = "false";
  93 string h_font_sf_scale           = "100";
  94 string h_font_tt_scale           = "100";
  95 string h_graphics                = "default";
  96 string h_paperfontsize           = "default";
  97 string h_spacing                 = "single";
  98 string h_papersize               = "default";
  99 string h_use_geometry            = "false";
 100 string h_use_amsmath             = "0";
 101 string h_cite_engine             = "basic";
 102 string h_use_bibtopic            = "false";
 103 string h_paperorientation        = "portrait";
 104 string h_secnumdepth             = "3";
 105 string h_tocdepth                = "3";
 106 string h_paragraph_separation    = "indent";
 107 string h_defskip                 = "medskip";
 108 string h_quotes_language         = "english";
 109 string h_papercolumns            = "1";
 110 string h_papersides              = string();
 111 string h_paperpagestyle          = "default";
 112 string h_tracking_changes        = "false";
 113 string h_output_changes          = "false";
 114
 115
 116 void handle_opt(vector<string> & opts, char const * const * what, string & target)
 117 {
 118         if (opts.empty())
 119                 return;
 120
 121         // the last language option is the document language (for babel and LyX)
 122         // the last size option is the document font size
 123         vector<string>::iterator it;
 124         vector<string>::iterator position = opts.begin();
 125         for (; *what; ++what) {
 126                 it = find(opts.begin(), opts.end(), *what);
 127                 if (it != opts.end()) {
 128                         if (it >= position) {
 129                                 target = *what;
 130                                 position = it;
 131                         }
 132                         // remove found options from the list
 133                         opts.erase(it);
 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(trim(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(trim(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         // roman fonts
 204         if (is_known(name, known_roman_fonts))
 205                 h_font_roman = name;
 206         if (name == "fourier") {
 207                 h_font_roman = "utopia";
 208                 // when font uses real small capitals
 209                 if (opts == "expert")
 210                         h_font_sc = "true";
 211         }
 212         if (name == "mathpazo")
 213                 h_font_roman = "palatino";
 214         if (name == "mathptmx")
 215                 h_font_roman = "times";
 216         // sansserif fonts
 217         if (is_known(name, known_sans_fonts)) {
 218                 h_font_sans = name;
 219                 if (!opts.empty()) {
 220                         scale = opts;
 221                         pos = scale.find(".");
 222                         h_font_sf_scale = scale.erase(0, pos + 1);
 223                 }
 224         }
 225         // typewriter fonts
 226         if (is_known(name, known_typewriter_fonts)) {
 227                 h_font_typewriter = name;
 228                 if (!opts.empty()) {
 229                         scale = opts;
 230                         pos = scale.find(".");
 231                         h_font_tt_scale = scale.erase(0, pos + 1);
 232                 }
 233         }
 234         // font uses old-style figure
 235         if (name == "eco")
 236                 h_font_osf = "true";
 237
 238         else if (name == "amsmath" || name == "amssymb")
 239                 h_use_amsmath = "1";
 240         else if (name == "babel")
 241                 ; // ignore this
 242         else if (name == "fontenc")
 243                 ; // ignore this
 244         else if (name == "inputenc") {
 245                 // only set when there is not more than one inputenc option
 246                 // therefore check for the "," character
 247                 if (opts.find(",") == string::npos)
 248                         h_inputencoding = opts;
 249                 options.clear();
 250         } else if (name == "makeidx")
 251                 ; // ignore this
 252         else if (name == "verbatim")
 253                 ; // ignore this
 254         else if (name == "graphicx")
 255                 ; // ignore this
 256         else if (is_known(name, known_languages)) {
 257                 if (is_known(name, known_french_languages)) {
 258                         h_language = "french";
 259                         h_quotes_language = "french";
 260                 } else {
 261                         h_language = name;
 262                         h_quotes_language = name;
 263                 }
 264
 265         } else if (name == "natbib") {
 266                 h_cite_engine = "natbib_authoryear";
 267                 vector<string>::iterator it =
 268                         find(options.begin(), options.end(), "authoryear");
 269                 if (it != options.end())
 270                         options.erase(it);
 271                 else {
 272                         it = find(options.begin(), options.end(), "numbers");
 273                         if (it != options.end()) {
 274                                 h_cite_engine = "natbib_numerical";
 275                                 options.erase(it);
 276                         }
 277                 }
 278         } else if (name == "jurabib") {
 279                 h_cite_engine = "jurabib";
 280         } else if (options.empty())
 281                 h_preamble << "\\usepackage{" << name << "}\n";
 282         else {
 283                 h_preamble << "\\usepackage[" << opts << "]{" << name << "}\n";
 284                 options.clear();
 285         }
 286
 287         // We need to do something with the options...
 288         if (!options.empty())
 289                 cerr << "Ignoring options '" << join(options, ",")
 290                      << "' of package " << name << '.' << endl;
 291 }
 292
 293
 294
 295 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 296 {
 297         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 298            << "\\lyxformat 247\n"
 299            << "\\begin_document\n"
 300            << "\\begin_header\n"
 301            << "\\textclass " << h_textclass << "\n";
 302         if (!h_preamble.str().empty())
 303                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 304         if (!h_options.empty())
 305                 os << "\\options " << h_options << "\n";
 306         os << "\\language " << h_language << "\n"
 307            << "\\inputencoding " << h_inputencoding << "\n"
 308            << "\\font_roman " << h_font_roman << "\n"
 309            << "\\font_sans " << h_font_sans << "\n"
 310            << "\\font_typewriter " << h_font_typewriter << "\n"
 311            << "\\font_default_family " << h_font_default_family << "\n"
 312            << "\\font_sc " << h_font_sc << "\n"
 313            << "\\font_osf " << h_font_osf << "\n"
 314            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 315            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 316            << "\\graphics " << h_graphics << "\n"
 317            << "\\paperfontsize " << h_paperfontsize << "\n"
 318            << "\\spacing " << h_spacing << "\n"
 319            << "\\papersize " << h_papersize << "\n"
 320            << "\\use_geometry " << h_use_geometry << "\n"
 321            << "\\use_amsmath " << h_use_amsmath << "\n"
 322            << "\\cite_engine " << h_cite_engine << "\n"
 323            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 324            << "\\paperorientation " << h_paperorientation << "\n"
 325            << "\\secnumdepth " << h_secnumdepth << "\n"
 326            << "\\tocdepth " << h_tocdepth << "\n"
 327            << "\\paragraph_separation " << h_paragraph_separation << "\n"
 328            << "\\defskip " << h_defskip << "\n"
 329            << "\\quotes_language " << h_quotes_language << "\n"
 330            << "\\papercolumns " << h_papercolumns << "\n"
 331            << "\\papersides " << h_papersides << "\n"
 332            << "\\paperpagestyle " << h_paperpagestyle << "\n"
 333            << "\\tracking_changes " << h_tracking_changes << "\n"
 334            << "\\output_changes " << h_output_changes << "\n"
 335            << "\\end_header\n\n"
 336            << "\\begin_body\n";
 337         // clear preamble for subdocuments
 338         h_preamble.str("");
 339 }
 340
 341 } // anonymous namespace
 342
 343 TextClass const parse_preamble(Parser & p, ostream & os, string const & forceclass)
 344 {
 345         // initialize fixed types
 346         special_columns['D'] = 3;
 347         bool is_full_document = false;
 348
 349         // determine whether this is a full document or a fragment for inclusion
 350         while (p.good()) {
 351                 Token const & t = p.get_token();
 352
 353                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 354                         is_full_document = true;
 355                         break;
 356                 }
 357         }
 358         p.reset();
 359
 360         while (is_full_document && p.good()) {
 361                 Token const & t = p.get_token();
 362
 363 #ifdef FILEDEBUG
 364                 cerr << "t: " << t << "\n";
 365 #endif
 366
 367                 //
 368                 // cat codes
 369                 //
 370                 if (t.cat() == catLetter ||
 371                           t.cat() == catSuper ||
 372                           t.cat() == catSub ||
 373                           t.cat() == catOther ||
 374                           t.cat() == catMath ||
 375                           t.cat() == catActive ||
 376                           t.cat() == catBegin ||
 377                           t.cat() == catEnd ||
 378                           t.cat() == catAlign ||
 379                           t.cat() == catParameter)
 380                 h_preamble << t.character();
 381
 382                 else if (t.cat() == catSpace || t.cat() == catNewline)
 383                         h_preamble << t.asInput();
 384
 385                 else if (t.cat() == catComment)
 386                         h_preamble << t.asInput();
 387
 388                 else if (t.cs() == "pagestyle")
 389                         h_paperpagestyle = p.verbatim_item();
 390
 391                 else if (t.cs() == "makeatletter") {
 392                         p.setCatCode('@', catLetter);
 393                         h_preamble << "\\makeatletter";
 394                 }
 395
 396                 else if (t.cs() == "makeatother") {
 397                         p.setCatCode('@', catOther);
 398                         h_preamble << "\\makeatother";
 399                 }
 400
 401                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 402                             || t.cs() == "providecommand") {
 403                         bool star = false;
 404                         if (p.next_token().character() == '*') {
 405                                 p.get_token();
 406                                 star = true;
 407                         }
 408                         string const name = p.verbatim_item();
 409                         string const opt1 = p.getOpt();
 410                         string const opt2 = p.getFullOpt();
 411                         string const body = p.verbatim_item();
 412                         // font settings
 413                         if (name == "\\rmdefault")
 414                                 if (is_known(body, known_roman_fonts))
 415                                         h_font_roman = body;
 416
 417                         if (name == "\\sfdefault")
 418                                 if (is_known(body, known_sans_fonts))
 419                                         h_font_sans = body;
 420
 421                         if (name == "\\ttdefault")
 422                                 if (is_known(body, known_typewriter_fonts))
 423                                         h_font_typewriter = body;
 424
 425                         if (name == "\\familydefault") {
 426                                 string family = body;
 427                                 // remove leading "\"
 428                                 h_font_default_family = family.erase(0,1);
 429                         }
 430                         // only non-lyxspecific stuff
 431                         if (   name != "\\noun"
 432                             && name != "\\tabularnewline"
 433                             && name != "\\LyX"
 434                             && name != "\\lyxline"
 435                             && name != "\\lyxaddress"
 436                             && name != "\\lyxrightaddress"
 437                             && name != "\\lyxdot"
 438                             && name != "\\boldsymbol"
 439                             && name != "\\lyxarrow"
 440                             && name != "\\rmdefault"
 441                             && name != "\\sfdefault"
 442                             && name != "\\ttdefault"
 443                             && name != "\\familydefault") {
 444                                 ostringstream ss;
 445                                 ss << '\\' << t.cs();
 446                                 if (star)
 447                                         ss << '*';
 448                                 ss << '{' << name << '}' << opt1 << opt2
 449                                    << '{' << body << "}";
 450                                 h_preamble << ss.str();
 451
 452                                 // Add the command to the known commands
 453                                 add_known_command(name, opt1, !opt2.empty());
 454 /*
 455                                 ostream & out = in_preamble ? h_preamble : os;
 456                                 out << "\\" << t.cs() << "{" << name << "}"
 457                                     << opts << "{" << body << "}";
 458 */
 459                         }
 460                 }
 461
 462                 else if (t.cs() == "documentclass") {
 463                         vector<string> opts = split_options(p.getArg('[', ']'));
 464                         handle_opt(opts, known_languages, h_language);
 465                         if (is_known(h_language, known_french_languages))
 466                                 h_language = "french";
 467                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 468                         // delete "pt" at the end
 469                         string::size_type i = h_paperfontsize.find("pt");
 470                         if (i != string::npos)
 471                                 h_paperfontsize.erase(i);
 472                         h_quotes_language = h_language;
 473                         h_options = join(opts, ",");
 474                         h_textclass = p.getArg('{', '}');
 475                 }
 476
 477                 else if (t.cs() == "usepackage") {
 478                         string const options = p.getArg('[', ']');
 479                         string const name = p.getArg('{', '}');
 480                         if (options.empty() && name.find(',')) {
 481                                 vector<string> vecnames;
 482                                 split(name, vecnames, ',');
 483                                 vector<string>::const_iterator it  = vecnames.begin();
 484                                 vector<string>::const_iterator end = vecnames.end();
 485                                 for (; it != end; ++it)
 486                                         handle_package(trim(*it), string());
 487                         } else {
 488                                 handle_package(name, options);
 489                         }
 490                 }
 491
 492                 else if (t.cs() == "newenvironment") {
 493                         string const name = p.getArg('{', '}');
 494                         ostringstream ss;
 495                         ss << "\\newenvironment{" << name << "}";
 496                         ss << p.getOpt();
 497                         ss << p.getOpt();
 498                         ss << '{' << p.verbatim_item() << '}';
 499                         ss << '{' << p.verbatim_item() << '}';
 500                         if (name != "lyxcode" && name != "lyxlist" &&
 501                             name != "lyxrightadress" &&
 502                             name != "lyxaddress" && name != "lyxgreyedout")
 503                                 h_preamble << ss.str();
 504                 }
 505
 506                 else if (t.cs() == "def") {
 507                         string name = p.get_token().cs();
 508                         while (p.next_token().cat() != catBegin)
 509                                 name += p.get_token().asString();
 510                         h_preamble << "\\def\\" << name << '{'
 511                                    << p.verbatim_item() << "}";
 512                 }
 513
 514                 else if (t.cs() == "newcolumntype") {
 515                         string const name = p.getArg('{', '}');
 516                         trim(name);
 517                         int nargs = 0;
 518                         string opts = p.getOpt();
 519                         if (!opts.empty()) {
 520                                 istringstream is(string(opts, 1));
 521                                 //cerr << "opt: " << is.str() << "\n";
 522                                 is >> nargs;
 523                         }
 524                         special_columns[name[0]] = nargs;
 525                         h_preamble << "\\newcolumntype{" << name << "}";
 526                         if (nargs)
 527                                 h_preamble << "[" << nargs << "]";
 528                         h_preamble << "{" << p.verbatim_item() << "}";
 529                 }
 530
 531                 else if (t.cs() == "setcounter") {
 532                         string const name = p.getArg('{', '}');
 533                         string const content = p.getArg('{', '}');
 534                         if (name == "secnumdepth")
 535                                 h_secnumdepth = content;
 536                         else if (name == "tocdepth")
 537                                 h_tocdepth = content;
 538                         else
 539                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
 540                 }
 541
 542                 else if (t.cs() == "setlength") {
 543                         string const name = p.verbatim_item();
 544                         string const content = p.verbatim_item();
 545                         // Is this correct?
 546                         if (name == "parskip")
 547                                 h_paragraph_separation = "skip";
 548                         else if (name == "parindent")
 549                                 h_paragraph_separation = "skip";
 550                         else
 551                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
 552                 }
 553
 554                 else if (t.cs() == "begin") {
 555                         string const name = p.getArg('{', '}');
 556                         if (name == "document")
 557                                 break;
 558                         h_preamble << "\\begin{" << name << "}";
 559                 }
 560
 561                 else if (t.cs() == "jurabibsetup") {
 562                         vector<string> jurabibsetup =
 563                                 split_options(p.getArg('{', '}'));
 564                         // add jurabibsetup to the jurabib package options
 565                         add_package("jurabib", jurabibsetup);
 566                         if (!jurabibsetup.empty()) {
 567                                 h_preamble << "\\jurabibsetup{"
 568                                            << join(jurabibsetup, ",") << '}';
 569                         }
 570                 }
 571
 572                 else if (!t.cs().empty())
 573                         h_preamble << '\\' << t.cs();
 574         }
 575         p.skip_spaces();
 576
 577         // Force textclass if the user wanted it
 578         if (!forceclass.empty())
 579                 h_textclass = forceclass;
 580         if (noweb_mode && !lyx::support::prefixIs(h_textclass, "literate-"))
 581                 h_textclass.insert(0, "literate-");
 582         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
 583         if (layoutfilename.empty()) {
 584                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
 585                 exit(1);
 586         }
 587         TextClass textclass;
 588         textclass.read(layoutfilename);
 589         if (h_papersides.empty()) {
 590                 ostringstream ss;
 591                 ss << textclass.sides();
 592                 h_papersides = ss.str();
 593         }
 594         end_preamble(os, textclass);
 595         return textclass;
 596 }
 597
 598 // }])
 599
 600
 601 } // namespace lyx