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 "LayoutFile.h"
  19 #include "Layout.h"
  20 #include "Lexer.h"
  21 #include "TextClass.h"
  22
  23 #include "support/convert.h"
  24 #include "support/FileName.h"
  25 #include "support/filetools.h"
  26 #include "support/lstrings.h"
  27
  28 #include "support/regex.h"
  29
  30 #include <algorithm>
  31 #include <iostream>
  32 #include <sstream>
  33 #include <string>
  34 #include <vector>
  35 #include <map>
  36
  37 using namespace std;
  38 using namespace lyx::support;
  39
  40
  41 namespace lyx {
  42
  43 // special columntypes
  44 extern map<char, int> special_columns;
  45
  46 map<string, vector<string> > used_packages;
  47
  48 // needed to handle encodings with babel
  49 bool one_language = true;
  50
  51 namespace {
  52
  53 //add this to known_languages when updating to lyxformat 266:
  54 // "armenian"
  55 //add these to known_languages when updating to lyxformat 268:
  56 //"chinese-simplified", "chinese-traditional", "japanese", "korean"
  57 // Both changes require first that support for non-babel languages (CJK,
  58 // armtex) is added.
  59 // add polutonikogreek for lyxformat 338
  60 // add mongolian for lyxformat 342
  61 // add turkmen for lyxformat 383
  62 const char * const known_languages[] = { "afrikaans", "albanian", "american",
  63 "arabic", "arabtex", "austrian", "bahasa", "bahasai", "bahasam", "basque",
  64 "belarusian", "brazil", "brazilian", "breton", "british", "bulgarian",
  65 "canadian", "canadien", "catalan", "croatian", "czech", "danish", "dutch",
  66 "english", "esperanto", "estonian", "finnish", "francais", "french",
  67 "frenchb", "frenchle", "frenchpro", "galician", "german", "germanb", "greek",
  68 "hebrew", "icelandic", "indon", "indonesian", "interlingua", "irish",
  69 "italian", "kazakh", "latin", "latvian", "lithuanian", "lsorbian", "magyar",
  70 "malay", "meyalu", "naustrian", "ngerman", "ngermanb", "norsk", "nynorsk",
  71 "polish", "portuges", "portuguese", "romanian", "russian", "russianb",
  72 "samin", "scottish", "serbian", "serbian-latin", "slovak", "slovene",
  73 "spanish", "swedish", "thai", "turkish", "ukraineb", "ukrainian",
  74 "usorbian", "vietnam", "welsh", 0};
  75
  76 const char * const known_bahasa_languages[] = {"bahasa", "bahasai",
  77                                                 "indon", "indonesian", 0};
  78 const char * const known_bahasam_languages[] = {"bahasam", "malay",
  79                                                 "meyalu", 0};
  80 const char * const known_brazilian_languages[] = {"brazil", "brazilian", 0};
  81 const char * const known_french_languages[] = {"french", "frenchb", "francais",
  82                                                 "frenchle", "frenchpro", 0};
  83 const char * const known_german_languages[] = {"german", "germanb", 0};
  84 const char * const known_ngerman_languages[] = {"ngerman", "ngermanb", 0};
  85 const char * const known_portuguese_languages[] = {"portuges", "portuguese", 0};
  86 const char * const known_russian_languages[] = {"russian", "russianb", 0};
  87 const char * const known_ukrainian_languages[] = {"ukrainian", "ukraineb", 0};
  88
  89 //add these to known_english_quotes_languages when updating to lyxformat 268:
  90 //"chinese-simplified", "korean"
  91 // This requires first that support for non-babel languages (CJK) is added.
  92 const char * const known_english_quotes_languages[] = {"american", "canadian",
  93 "english", "esperanto", "hebrew", "irish", "scottish", "thai", 0};
  94
  95 //add this to known_french_quotes_languages when updating to
  96 // - lyxformat 327: "spanish-mexico"
  97 // - lyxformat 338: "polutonikogreek"
  98 // - lyxformat 383: "turkmen"
  99 const char * const known_french_quotes_languages[] = {"albanian", "arabic",
 100 "basque", "canadien", "catalan", "galician", "greek", "italian", "norsk",
 101 "nynorsk", "spanish", "turkish", "vietnam", 0};
 102
 103 const char * const known_german_quotes_languages[] = {"austrian", "bulgarian",
 104 "czech", "icelandic", "lithuanian", "lsorbian", "naustrian", "serbian",
 105 "serbian-latin", "slovak", "slovene", "usorbian",  0};
 106
 107 const char * const known_polish_quotes_languages[] = {"afrikaans", "croatian",
 108 "dutch", "estonian", "magyar", "polish", "romanian", 0};
 109
 110 const char * const known_swedish_quotes_languages[] = {"finnish",
 111 "swedish", 0};
 112
 113 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
 114
 115 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
 116 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
 117
 118 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
 119 "helvet", "lmss", 0};
 120
 121 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
 122 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
 123 "newcent", 0};
 124
 125 const char * const known_paper_sizes[] = { "a3paper", "b3paper", "a4paper",
 126 "b4paper", "a5paper", "b5paper", "executivepaper", "legalpaper",
 127 "letterpaper", 0};
 128
 129 const char * const known_class_paper_sizes[] = { "a4paper", "a5paper",
 130 "executivepaper", "legalpaper", "letterpaper", 0};
 131
 132 const char * const known_paper_margins[] = { "lmargin", "tmargin", "rmargin",
 133 "bmargin", "headheight", "headsep", "footskip", "columnsep", 0};
 134
 135 const char * const known_coded_paper_margins[] = { "leftmargin", "topmargin",
 136 "rightmargin", "bottommargin", "headheight", "headsep", "footskip",
 137 "columnsep", 0};
 138
 139 // default settings
 140 ostringstream h_preamble;
 141 string h_textclass               = "article";
 142 string h_use_default_options     = "false";
 143 string h_options                 = string();
 144 string h_language                = "english";
 145 string h_inputencoding           = "auto";
 146 string h_font_roman              = "default";
 147 string h_font_sans               = "default";
 148 string h_font_typewriter         = "default";
 149 string h_font_default_family     = "default";
 150 string h_font_sc                 = "false";
 151 string h_font_osf                = "false";
 152 string h_font_sf_scale           = "100";
 153 string h_font_tt_scale           = "100";
 154 string h_graphics                = "default";
 155 string h_float_placement;
 156 string h_paperfontsize           = "default";
 157 string h_spacing                 = "single";
 158 string h_use_hyperref            = "0";
 159 string h_pdf_title;
 160 string h_pdf_author;
 161 string h_pdf_subject;
 162 string h_pdf_keywords;
 163 string h_pdf_bookmarks           = "1";
 164 string h_pdf_bookmarksnumbered   = "0";
 165 string h_pdf_bookmarksopen       = "0";
 166 string h_pdf_bookmarksopenlevel  = "1";
 167 string h_pdf_breaklinks          = "0";
 168 string h_pdf_pdfborder           = "0";
 169 string h_pdf_colorlinks          = "0";
 170 string h_pdf_backref             = "section";
 171 string h_pdf_pdfusetitle         = "1";
 172 string h_pdf_pagemode;
 173 string h_pdf_quoted_options;
 174 string h_papersize               = "default";
 175 string h_use_geometry            = "false";
 176 string h_use_amsmath             = "1";
 177 string h_use_esint               = "1";
 178 string h_cite_engine             = "basic";
 179 string h_use_bibtopic            = "false";
 180 string h_paperorientation        = "portrait";
 181 string h_secnumdepth             = "3";
 182 string h_tocdepth                = "3";
 183 string h_paragraph_separation    = "indent";
 184 string h_defskip                 = "medskip";
 185 string h_quotes_language         = "english";
 186 string h_papercolumns            = "1";
 187 string h_papersides              = string();
 188 string h_paperpagestyle          = "default";
 189 string h_listings_params;
 190 string h_tracking_changes        = "false";
 191 string h_output_changes          = "false";
 192 string h_margins                 = "";
 193
 194
 195 /// translates a babel language name to a LyX language name
 196 string babel2lyx(string language)
 197 {
 198         if (language == "arabtex")
 199                 return "arabic_arabtex";
 200         if (language == "arabic")
 201                 return "arabic_arabi";
 202         if (language == "lsorbian")
 203                 return "lowersorbian";
 204         if (language == "usorbian")
 205                 return "uppersorbian";
 206         return language;
 207 }
 208
 209
 210 // returns true if at least one of the options in what has been found
 211 bool handle_opt(vector<string> & opts, char const * const * what, string & target)
 212 {
 213         if (opts.empty())
 214                 return false;
 215
 216         bool found = false;
 217         // the last language option is the document language (for babel and LyX)
 218         // the last size option is the document font size
 219         vector<string>::iterator it;
 220         vector<string>::iterator position = opts.begin();
 221         for (; *what; ++what) {
 222                 it = find(opts.begin(), opts.end(), *what);
 223                 if (it != opts.end()) {
 224                         if (it >= position) {
 225                                 found = true;
 226                                 target = *what;
 227                                 position = it;
 228                         }
 229                 }
 230         }
 231         return found;
 232 }
 233
 234
 235 void delete_opt(vector<string> & opts, char const * const * what)
 236 {
 237         if (opts.empty())
 238                 return;
 239
 240         // remove found options from the list
 241         // do this after handle_opt to avoid potential memory leaks
 242         vector<string>::iterator it;
 243         for (; *what; ++what) {
 244                 it = find(opts.begin(), opts.end(), *what);
 245                 if (it != opts.end())
 246                         opts.erase(it);
 247         }
 248 }
 249
 250
 251 /*!
 252  * Split a package options string (keyval format) into a vector.
 253  * Example input:
 254  *   authorformat=smallcaps,
 255  *   commabeforerest,
 256  *   titleformat=colonsep,
 257  *   bibformat={tabular,ibidem,numbered}
 258  */
 259 vector<string> split_options(string const & input)
 260 {
 261         vector<string> options;
 262         string option;
 263         Parser p(input);
 264         while (p.good()) {
 265                 Token const & t = p.get_token();
 266                 if (t.asInput() == ",") {
 267                         options.push_back(trim(option));
 268                         option.erase();
 269                 } else if (t.asInput() == "=") {
 270                         option += '=';
 271                         p.skip_spaces(true);
 272                         if (p.next_token().asInput() == "{")
 273                                 option += '{' + p.getArg('{', '}') + '}';
 274                 } else if (t.cat() != catSpace)
 275                         option += t.asInput();
 276         }
 277
 278         if (!option.empty())
 279                 options.push_back(trim(option));
 280
 281         return options;
 282 }
 283
 284
 285 /*!
 286  * Add package \p name with options \p options to used_packages.
 287  * Remove options from \p options that we don't want to output.
 288  */
 289 void add_package(string const & name, vector<string> & options)
 290 {
 291         // every package inherits the global options
 292         if (used_packages.find(name) == used_packages.end())
 293                 used_packages[name] = split_options(h_options);
 294
 295         vector<string> & v = used_packages[name];
 296         v.insert(v.end(), options.begin(), options.end());
 297         if (name == "jurabib") {
 298                 // Don't output the order argument (see the cite command
 299                 // handling code in text.cpp).
 300                 vector<string>::iterator end =
 301                         remove(options.begin(), options.end(), "natbiborder");
 302                 end = remove(options.begin(), end, "jurabiborder");
 303                 options.erase(end, options.end());
 304         }
 305 }
 306
 307
 308 // Given is a string like "scaled=0.9", return 0.9 * 100
 309 string const scale_as_percentage(string const & scale)
 310 {
 311         string::size_type pos = scale.find('=');
 312         if (pos != string::npos) {
 313                 string value = scale.substr(pos + 1);
 314                 if (isStrDbl(value))
 315                         return convert<string>(100 * convert<double>(value));
 316         }
 317         // If the input string didn't match our expectations.
 318         // return the default value "100"
 319         return "100";
 320 }
 321
 322
 323 void handle_package(Parser &p, string const & name, string const & opts,
 324                     bool in_lyx_preamble)
 325 {
 326         vector<string> options = split_options(opts);
 327         add_package(name, options);
 328         string scale;
 329
 330         // roman fonts
 331         if (is_known(name, known_roman_fonts)) {
 332                 h_font_roman = name;
 333                 p.skip_spaces();
 334         }
 335
 336         if (name == "fourier") {
 337                 h_font_roman = "utopia";
 338                 // when font uses real small capitals
 339                 if (opts == "expert")
 340                         h_font_sc = "true";
 341         }
 342
 343         if (name == "mathpazo")
 344                 h_font_roman = "palatino";
 345
 346         if (name == "mathptmx")
 347                 h_font_roman = "times";
 348
 349         // sansserif fonts
 350         if (is_known(name, known_sans_fonts)) {
 351                 h_font_sans = name;
 352                 if (!opts.empty()) {
 353                         scale = opts;
 354                         h_font_sf_scale = scale_as_percentage(scale);
 355                 }
 356         }
 357
 358         // typewriter fonts
 359         if (is_known(name, known_typewriter_fonts)) {
 360                 h_font_typewriter = name;
 361                 if (!opts.empty()) {
 362                         scale = opts;
 363                         h_font_tt_scale = scale_as_percentage(scale);
 364                 }
 365         }
 366
 367         // font uses old-style figure
 368         if (name == "eco")
 369                 h_font_osf = "true";
 370
 371         else if (name == "amsmath" || name == "amssymb")
 372                 h_use_amsmath = "2";
 373
 374         else if (name == "esint")
 375                 h_use_esint = "2";
 376
 377         else if (name == "babel" && !opts.empty()) {
 378                 // check if more than one option was used - used later for inputenc
 379                 // in case inputenc is parsed before babel, set the encoding to auto
 380                 if (options.begin() != options.end() - 1) {
 381                         one_language = false;
 382                         h_inputencoding = "auto";
 383                 }
 384                 // babel takes the last language of the option of its \usepackage
 385                 // call as document language. If there is no such language option, the
 386                 // last language in the documentclass options is used.
 387                 handle_opt(options, known_languages, h_language);
 388                 delete_opt(options, known_languages);
 389         }
 390
 391         else if (name == "fontenc")
 392                  ;// ignore this
 393
 394         else if (name == "inputenc") {
 395                 // h_inputencoding is only set when there is not more than one
 396                 // inputenc option because otherwise h_inputencoding must be
 397                 // set to "auto" (the default encoding of the document language)
 398                 // Therefore check for the "," character.
 399                 // It is also only set when there is not more then one babel
 400                 // language option but this is handled in the routine for babel.
 401                 if (opts.find(",") == string::npos && one_language == true)
 402                         h_inputencoding = opts;
 403                 if (!options.empty())
 404                         p.setEncoding(options.back());
 405                 options.clear();
 406         }
 407
 408         else if (name == "makeidx")
 409                 ; // ignore this
 410
 411         else if (name == "prettyref")
 412                 ; // ignore this
 413
 414         else if (name == "varioref")
 415                 ; // ignore this
 416
 417         else if (name == "verbatim")
 418                 ; // ignore this
 419
 420         else if (name == "nomencl")
 421                 ; // ignore this
 422
 423         else if (name == "textcomp")
 424                 ; // ignore this
 425
 426         else if (name == "url")
 427                 ; // ignore this
 428
 429         else if (LYX_FORMAT >= 408 && name == "subscript")
 430                 ; // ignore this
 431
 432         else if (name == "color") {
 433                 // with the following command this package is only loaded when needed for
 434                 // undefined colors, since we only support the predefined colors
 435                 h_preamble << "\\@ifundefined{definecolor}\n {\\usepackage{color}}{}\n";
 436         }
 437
 438         else if (name == "graphicx")
 439                 ; // ignore this
 440
 441         else if (name == "setspace")
 442                 ; // ignore this
 443
 444         else if (name == "geometry")
 445                 ; // Ignore this, the geometry settings are made by the \geometry
 446                   // command. This command is handled below.
 447
 448         else if (is_known(name, known_languages))
 449                 h_language = name;
 450
 451         else if (name == "natbib") {
 452                 h_cite_engine = "natbib_authoryear";
 453                 vector<string>::iterator it =
 454                         find(options.begin(), options.end(), "authoryear");
 455                 if (it != options.end())
 456                         options.erase(it);
 457                 else {
 458                         it = find(options.begin(), options.end(), "numbers");
 459                         if (it != options.end()) {
 460                                 h_cite_engine = "natbib_numerical";
 461                                 options.erase(it);
 462                         }
 463                 }
 464         }
 465
 466         else if (name == "jurabib")
 467                 h_cite_engine = "jurabib";
 468
 469         else if (!in_lyx_preamble) {
 470                 if (options.empty())
 471                         h_preamble << "\\usepackage{" << name << "}";
 472                 else {
 473                         h_preamble << "\\usepackage[" << opts << "]{"
 474                                    << name << "}";
 475                         options.clear();
 476                 }
 477         }
 478
 479         // We need to do something with the options...
 480         if (!options.empty())
 481                 cerr << "Ignoring options '" << join(options, ",")
 482                      << "' of package " << name << '.' << endl;
 483
 484         // remove the whitespace
 485         p.skip_spaces();
 486 }
 487
 488
 489
 490 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 491 {
 492         // merge synonym languages
 493         if (is_known(h_language, known_bahasa_languages))
 494                 h_language = "bahasa";
 495         else if (is_known(h_language, known_bahasam_languages))
 496                 h_language = "bahasam";
 497         else if (is_known(h_language, known_brazilian_languages))
 498                 h_language = "brazilian";
 499         else if (is_known(h_language, known_french_languages))
 500                 h_language = "french";
 501         else if (is_known(h_language, known_german_languages))
 502                 h_language = "german";
 503         else if (is_known(h_language, known_ngerman_languages))
 504                 h_language = "ngerman";
 505         else if (is_known(h_language, known_portuguese_languages))
 506                 h_language = "portuguese";
 507         else if (is_known(h_language, known_russian_languages))
 508                 h_language = "russian";
 509         else if (is_known(h_language, known_ukrainian_languages))
 510                 h_language = "ukrainian";
 511
 512         // set the quote language
 513         // LyX only knows the following quotes languages:
 514         // english, swedish, german, polish, french and danish
 515         // (quotes for "japanese" and "chinese-traditional" are missing because
 516         //  they wouldn't be useful: http://www.lyx.org/trac/ticket/6383)
 517         // conversion list taken from
 518         // http://en.wikipedia.org/wiki/Quotation_mark,_non-English_usage
 519         // (quotes for kazakh and interlingua are unknown)
 520         // danish
 521         if (h_language == "danish")
 522                 h_quotes_language = "danish";
 523         // french
 524         else if (is_known(h_language, known_french_quotes_languages)
 525                 || is_known(h_language, known_french_languages)
 526                 || is_known(h_language, known_russian_languages)
 527                 || is_known(h_language, known_ukrainian_languages))
 528                 h_quotes_language = "french";
 529         // german
 530         else if (is_known(h_language, known_german_quotes_languages)
 531                 || is_known(h_language, known_german_languages)
 532                 || is_known(h_language, known_ngerman_languages))
 533                 h_quotes_language = "german";
 534         // polish
 535         else if (is_known(h_language, known_polish_quotes_languages))
 536                 h_quotes_language = "polish";
 537         // swedish
 538         else if (is_known(h_language, known_swedish_quotes_languages))
 539                 h_quotes_language = "swedish";
 540         //english
 541         else if (is_known(h_language, known_english_quotes_languages)
 542                 || is_known(h_language, known_bahasa_languages)
 543                 || is_known(h_language, known_bahasam_languages)
 544                 || is_known(h_language, known_brazilian_languages)
 545                 || is_known(h_language, known_portuguese_languages))
 546                 h_quotes_language = "english";
 547
 548         h_language = babel2lyx(h_language);
 549
 550         // output the LyX file settings
 551         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 552            << "\\lyxformat " << LYX_FORMAT << '\n'
 553            << "\\begin_document\n"
 554            << "\\begin_header\n"
 555            << "\\textclass " << h_textclass << "\n";
 556         if (!h_preamble.str().empty())
 557                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 558         if (!h_options.empty())
 559                 os << "\\options " << h_options << "\n";
 560         os << "\\use_default_options " << h_use_default_options << "\n"
 561            << "\\language " << h_language << "\n"
 562            << "\\inputencoding " << h_inputencoding << "\n"
 563            << "\\font_roman " << h_font_roman << "\n"
 564            << "\\font_sans " << h_font_sans << "\n"
 565            << "\\font_typewriter " << h_font_typewriter << "\n"
 566            << "\\font_default_family " << h_font_default_family << "\n"
 567            << "\\font_sc " << h_font_sc << "\n"
 568            << "\\font_osf " << h_font_osf << "\n"
 569            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 570            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 571            << "\\graphics " << h_graphics << "\n";
 572         if (!h_float_placement.empty())
 573                 os << "\\float_placement " << h_float_placement << "\n";
 574         os << "\\paperfontsize " << h_paperfontsize << "\n"
 575            << "\\spacing " << h_spacing << "\n"
 576            << "\\use_hyperref " << h_use_hyperref << '\n';
 577         if (h_use_hyperref == "1") {
 578                 if (!h_pdf_title.empty())
 579                         os << "\\pdf_title \"" << h_pdf_title << "\"\n";
 580                 if (!h_pdf_author.empty())
 581                         os << "\\pdf_author \"" << h_pdf_author << "\"\n";
 582                 if (!h_pdf_subject.empty())
 583                         os << "\\pdf_subject \"" << h_pdf_subject << "\"\n";
 584                 if (!h_pdf_keywords.empty())
 585                         os << "\\pdf_keywords \"" << h_pdf_keywords << "\"\n";
 586                 os << "\\pdf_bookmarks " << h_pdf_bookmarks << "\n"
 587                       "\\pdf_bookmarksnumbered " << h_pdf_bookmarksnumbered << "\n"
 588                       "\\pdf_bookmarksopen " << h_pdf_bookmarksopen << "\n"
 589                       "\\pdf_bookmarksopenlevel " << h_pdf_bookmarksopenlevel << "\n"
 590                       "\\pdf_breaklinks " << h_pdf_breaklinks << "\n"
 591                       "\\pdf_pdfborder " << h_pdf_pdfborder << "\n"
 592                       "\\pdf_colorlinks " << h_pdf_colorlinks << "\n"
 593                       "\\pdf_backref " << h_pdf_backref << "\n"
 594                       "\\pdf_pdfusetitle " << h_pdf_pdfusetitle << '\n';
 595                 if (!h_pdf_pagemode.empty())
 596                         os << "\\pdf_pagemode " << h_pdf_pagemode << '\n';
 597                 if (!h_pdf_quoted_options.empty())
 598                         os << "\\pdf_quoted_options \"" << h_pdf_quoted_options << "\"\n";
 599         }
 600         os << "\\papersize " << h_papersize << "\n"
 601            << "\\use_geometry " << h_use_geometry << "\n"
 602            << "\\use_amsmath " << h_use_amsmath << "\n"
 603            << "\\use_esint " << h_use_esint << "\n"
 604            << "\\cite_engine " << h_cite_engine << "\n"
 605            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 606            << "\\paperorientation " << h_paperorientation << "\n"
 607            << h_margins
 608            << "\\secnumdepth " << h_secnumdepth << "\n"
 609            << "\\tocdepth " << h_tocdepth << "\n"
 610            << "\\paragraph_separation " << h_paragraph_separation << "\n"
 611            << "\\defskip " << h_defskip << "\n"
 612            << "\\quotes_language " << h_quotes_language << "\n"
 613            << "\\papercolumns " << h_papercolumns << "\n"
 614            << "\\papersides " << h_papersides << "\n"
 615            << "\\paperpagestyle " << h_paperpagestyle << "\n";
 616         if (!h_listings_params.empty())
 617                 os << "\\listings_params " << h_listings_params << "\n";
 618         os << "\\tracking_changes " << h_tracking_changes << "\n"
 619            << "\\output_changes " << h_output_changes << "\n"
 620            << "\\end_header\n\n"
 621            << "\\begin_body\n";
 622         // clear preamble for subdocuments
 623         h_preamble.str("");
 624 }
 625
 626 } // anonymous namespace
 627
 628 void parse_preamble(Parser & p, ostream & os,
 629         string const & forceclass, TeX2LyXDocClass & tc)
 630 {
 631         // initialize fixed types
 632         special_columns['D'] = 3;
 633         bool is_full_document = false;
 634         bool is_lyx_file = false;
 635         bool in_lyx_preamble = false;
 636
 637         // determine whether this is a full document or a fragment for inclusion
 638         while (p.good()) {
 639                 Token const & t = p.get_token();
 640
 641                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 642                         is_full_document = true;
 643                         break;
 644                 }
 645         }
 646         p.reset();
 647
 648         while (is_full_document && p.good()) {
 649                 Token const & t = p.get_token();
 650
 651 #ifdef FILEDEBUG
 652                 cerr << "t: " << t << "\n";
 653 #endif
 654
 655                 //
 656                 // cat codes
 657                 //
 658                 if (!in_lyx_preamble &&
 659                     (t.cat() == catLetter ||
 660                      t.cat() == catSuper ||
 661                      t.cat() == catSub ||
 662                      t.cat() == catOther ||
 663                      t.cat() == catMath ||
 664                      t.cat() == catActive ||
 665                      t.cat() == catBegin ||
 666                      t.cat() == catEnd ||
 667                      t.cat() == catAlign ||
 668                      t.cat() == catParameter))
 669                         h_preamble << t.cs();
 670
 671                 else if (!in_lyx_preamble &&
 672                          (t.cat() == catSpace || t.cat() == catNewline))
 673                         h_preamble << t.asInput();
 674
 675                 else if (t.cat() == catComment) {
 676                         // regex to parse comments (currently not used)
 677                         static regex const islyxfile("%% LyX .* created this file");
 678                         static regex const usercommands("User specified LaTeX commands");
 679
 680                         string const comment = t.asInput();
 681
 682                         // magically switch encoding default if it looks like XeLaTeX
 683                         static string const magicXeLaTeX =
 684                                 "% This document must be compiled with XeLaTeX ";
 685                         if (comment.size() > magicXeLaTeX.size()
 686                                   && comment.substr(0, magicXeLaTeX.size()) == magicXeLaTeX
 687                                   && h_inputencoding == "auto") {
 688                                 cerr << "XeLaTeX comment found, switching to UTF8\n";
 689                                 h_inputencoding = "utf8";
 690                         }
 691                         smatch sub;
 692                         if (regex_search(comment, sub, islyxfile)) {
 693                                 is_lyx_file = true;
 694                                 in_lyx_preamble = true;
 695                         } else if (is_lyx_file
 696                                    && regex_search(comment, sub, usercommands))
 697                                 in_lyx_preamble = false;
 698                         else if (!in_lyx_preamble)
 699                                 h_preamble << t.asInput();
 700                 }
 701
 702                 else if (t.cs() == "pagestyle")
 703                         h_paperpagestyle = p.verbatim_item();
 704
 705                 else if (t.cs() == "makeatletter") {
 706                         // LyX takes care of this
 707                         p.setCatCode('@', catLetter);
 708                 }
 709
 710                 else if (t.cs() == "makeatother") {
 711                         // LyX takes care of this
 712                         p.setCatCode('@', catOther);
 713                 }
 714
 715                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 716                             || t.cs() == "providecommand"
 717                                 || t.cs() == "DeclareRobustCommand"
 718                                 || t.cs() == "ProvideTextCommandDefault"
 719                                 || t.cs() == "DeclareMathAccent") {
 720                         bool star = false;
 721                         if (p.next_token().character() == '*') {
 722                                 p.get_token();
 723                                 star = true;
 724                         }
 725                         string const name = p.verbatim_item();
 726                         string const opt1 = p.getOpt();
 727                         string const opt2 = p.getFullOpt();
 728                         string const body = p.verbatim_item();
 729                         // font settings
 730                         if (name == "\\rmdefault")
 731                                 if (is_known(body, known_roman_fonts))
 732                                         h_font_roman = body;
 733                         if (name == "\\sfdefault")
 734                                 if (is_known(body, known_sans_fonts))
 735                                         h_font_sans = body;
 736                         if (name == "\\ttdefault")
 737                                 if (is_known(body, known_typewriter_fonts))
 738                                         h_font_typewriter = body;
 739                         if (name == "\\familydefault") {
 740                                 string family = body;
 741                                 // remove leading "\"
 742                                 h_font_default_family = family.erase(0,1);
 743                         }
 744                         // only non-lyxspecific stuff
 745                         if (!in_lyx_preamble) {
 746                                 ostringstream ss;
 747                                 ss << '\\' << t.cs();
 748                                 if (star)
 749                                         ss << '*';
 750                                 ss << '{' << name << '}' << opt1 << opt2
 751                                    << '{' << body << "}";
 752                                 h_preamble << ss.str();
 753
 754                                 // Add the command to the known commands
 755                                 add_known_command(name, opt1, !opt2.empty());
 756 /*
 757                                 ostream & out = in_preamble ? h_preamble : os;
 758                                 out << "\\" << t.cs() << "{" << name << "}"
 759                                     << opts << "{" << body << "}";
 760 */
 761                         }
 762                 }
 763
 764                 else if (t.cs() == "documentclass") {
 765                         vector<string>::iterator it;
 766                         vector<string> opts = split_options(p.getArg('[', ']'));
 767                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 768                         delete_opt(opts, known_fontsizes);
 769                         // delete "pt" at the end
 770                         string::size_type i = h_paperfontsize.find("pt");
 771                         if (i != string::npos)
 772                                 h_paperfontsize.erase(i);
 773                         // The documentclass options are always parsed before the options
 774                         // of the babel call so that a language cannot overwrite the babel
 775                         // options.
 776                         handle_opt(opts, known_languages, h_language);
 777                         delete_opt(opts, known_languages);
 778
 779                         // paper orientation
 780                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 781                                 h_paperorientation = "landscape";
 782                                 opts.erase(it);
 783                         }
 784                         // paper sides
 785                         if ((it = find(opts.begin(), opts.end(), "oneside"))
 786                                  != opts.end()) {
 787                                 h_papersides = "1";
 788                                 opts.erase(it);
 789                         }
 790                         if ((it = find(opts.begin(), opts.end(), "twoside"))
 791                                  != opts.end()) {
 792                                 h_papersides = "2";
 793                                 opts.erase(it);
 794                         }
 795                         // paper columns
 796                         if ((it = find(opts.begin(), opts.end(), "onecolumn"))
 797                                  != opts.end()) {
 798                                 h_papercolumns = "1";
 799                                 opts.erase(it);
 800                         }
 801                         if ((it = find(opts.begin(), opts.end(), "twocolumn"))
 802                                  != opts.end()) {
 803                                 h_papercolumns = "2";
 804                                 opts.erase(it);
 805                         }
 806                         // paper sizes
 807                         // some size options are know to any document classes, other sizes
 808                         // are handled by the \geometry command of the geometry package
 809                         handle_opt(opts, known_class_paper_sizes, h_papersize);
 810                         delete_opt(opts, known_class_paper_sizes);
 811                         // the remaining options
 812                         h_options = join(opts, ",");
 813                         h_textclass = p.getArg('{', '}');
 814                 }
 815
 816                 else if (t.cs() == "usepackage") {
 817                         string const options = p.getArg('[', ']');
 818                         string const name = p.getArg('{', '}');
 819                         vector<string> vecnames;
 820                         split(name, vecnames, ',');
 821                         vector<string>::const_iterator it  = vecnames.begin();
 822                         vector<string>::const_iterator end = vecnames.end();
 823                         for (; it != end; ++it)
 824                                 handle_package(p, trim(*it), options,
 825                                                in_lyx_preamble);
 826                 }
 827
 828                 else if (t.cs() == "inputencoding") {
 829                         string const encoding = p.getArg('{','}');
 830                         h_inputencoding = encoding;
 831                         p.setEncoding(encoding);
 832                 }
 833
 834                 else if (t.cs() == "newenvironment") {
 835                         string const name = p.getArg('{', '}');
 836                         ostringstream ss;
 837                         ss << "\\newenvironment{" << name << "}";
 838                         ss << p.getOpt();
 839                         ss << p.getOpt();
 840                         ss << '{' << p.verbatim_item() << '}';
 841                         ss << '{' << p.verbatim_item() << '}';
 842                         if (!in_lyx_preamble)
 843                                 h_preamble << ss.str();
 844                 }
 845
 846                 else if (t.cs() == "def") {
 847                         string name = p.get_token().cs();
 848                         while (p.next_token().cat() != catBegin)
 849                                 name += p.get_token().cs();
 850                         if (!in_lyx_preamble)
 851                                 h_preamble << "\\def\\" << name << '{'
 852                                            << p.verbatim_item() << "}";
 853                 }
 854
 855                 else if (t.cs() == "newcolumntype") {
 856                         string const name = p.getArg('{', '}');
 857                         trim(name);
 858                         int nargs = 0;
 859                         string opts = p.getOpt();
 860                         if (!opts.empty()) {
 861                                 istringstream is(string(opts, 1));
 862                                 is >> nargs;
 863                         }
 864                         special_columns[name[0]] = nargs;
 865                         h_preamble << "\\newcolumntype{" << name << "}";
 866                         if (nargs)
 867                                 h_preamble << "[" << nargs << "]";
 868                         h_preamble << "{" << p.verbatim_item() << "}";
 869                 }
 870
 871                 else if (t.cs() == "setcounter") {
 872                         string const name = p.getArg('{', '}');
 873                         string const content = p.getArg('{', '}');
 874                         if (name == "secnumdepth")
 875                                 h_secnumdepth = content;
 876                         else if (name == "tocdepth")
 877                                 h_tocdepth = content;
 878                         else
 879                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
 880                 }
 881
 882                 else if (t.cs() == "setlength") {
 883                         string const name = p.verbatim_item();
 884                         string const content = p.verbatim_item();
 885                         // the paragraphs are only not indented when \parindent is set to zero
 886                         if (name == "\\parindent" && content != "") {
 887                                 if (content[0] == '0')
 888                                         h_paragraph_separation = "skip";
 889                         } else if (name == "\\parskip") {
 890                                 if (content == "\\smallskipamount")
 891                                         h_defskip = "smallskip";
 892                                 else if (content == "\\medskipamount")
 893                                         h_defskip = "medskip";
 894                                 else if (content == "\\bigskipamount")
 895                                         h_defskip = "bigskip";
 896                                 else
 897                                         h_defskip = content;
 898                         } else
 899                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
 900                 }
 901
 902                 else if (t.cs() == "onehalfspacing")
 903                         h_spacing = "onehalf";
 904
 905                 else if (t.cs() == "doublespacing")
 906                         h_spacing = "double";
 907
 908                 else if (t.cs() == "setstretch")
 909                         h_spacing = "other " + p.verbatim_item();
 910
 911                 else if (t.cs() == "begin") {
 912                         string const name = p.getArg('{', '}');
 913                         if (name == "document")
 914                                 break;
 915                         h_preamble << "\\begin{" << name << "}";
 916                 }
 917
 918                 else if (t.cs() == "geometry") {
 919                         h_use_geometry = "true";
 920                         vector<string> opts = split_options(p.getArg('{', '}'));
 921                         vector<string>::iterator it;
 922                         // paper orientation
 923                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 924                                 h_paperorientation = "landscape";
 925                                 opts.erase(it);
 926                         }
 927                         // paper size
 928                         handle_opt(opts, known_paper_sizes, h_papersize);
 929                         delete_opt(opts, known_paper_sizes);
 930                         // page margins
 931                         char const * const * margin = known_paper_margins;
 932                         int k = -1;
 933                         for (; *margin; ++margin) {
 934                                 k += 1;
 935                                 // search for the "=" in e.g. "lmargin=2cm" to get the value
 936                                 for(size_t i = 0; i != opts.size(); i++) {
 937                                         if (opts.at(i).find(*margin) != string::npos) {
 938                                                 string::size_type pos = opts.at(i).find("=");
 939                                                 string value = opts.at(i).substr(pos + 1);
 940                                                 string name = known_coded_paper_margins[k];
 941                                                 h_margins += "\\" + name + " " + value + "\n";
 942                                         }
 943                                 }
 944                         }
 945                 }
 946
 947                 else if (t.cs() == "jurabibsetup") {
 948                         vector<string> jurabibsetup =
 949                                 split_options(p.getArg('{', '}'));
 950                         // add jurabibsetup to the jurabib package options
 951                         add_package("jurabib", jurabibsetup);
 952                         if (!jurabibsetup.empty()) {
 953                                 h_preamble << "\\jurabibsetup{"
 954                                            << join(jurabibsetup, ",") << '}';
 955                         }
 956                 }
 957
 958                 else if (!t.cs().empty() && !in_lyx_preamble)
 959                         h_preamble << '\\' << t.cs();
 960         }
 961
 962         // remove the whitespace
 963         p.skip_spaces();
 964
 965         // Force textclass if the user wanted it
 966         if (!forceclass.empty())
 967                 h_textclass = forceclass;
 968         if (noweb_mode && !prefixIs(h_textclass, "literate-"))
 969                 h_textclass.insert(0, "literate-");
 970         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
 971         if (layoutfilename.empty()) {
 972                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
 973                 exit(1);
 974         }
 975         tc.read(layoutfilename);
 976         if (h_papersides.empty()) {
 977                 ostringstream ss;
 978                 ss << tc.sides();
 979                 h_papersides = ss.str();
 980         }
 981         end_preamble(os, tc);
 982 }
 983
 984 // }])
 985
 986
 987 } // namespace lyx