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 turkmen for lyxformat 383
  60 const char * const known_languages[] = { "afrikaans", "albanian", "american",
  61 "arabic", "arabtex", "austrian", "bahasa", "bahasai", "bahasam", "basque",
  62 "belarusian", "brazil", "brazilian", "breton", "british", "bulgarian",
  63 "canadian", "canadien", "catalan", "croatian", "czech", "danish", "dutch",
  64 "english", "esperanto", "estonian", "finnish", "francais", "french",
  65 "frenchb", "frenchle", "frenchpro", "galician", "german", "germanb", "greek",
  66 "hebrew", "icelandic", "indon", "indonesian", "interlingua", "irish",
  67 "italian", "kazakh", "latin", "latvian", "lithuanian", "lsorbian", "magyar",
  68 "malay", "meyalu", "mongolian", "naustrian", "ngerman", "ngermanb", "norsk",
  69 "nynorsk", "polutonikogreek", "polish", "portuges", "portuguese", "romanian",
  70 "russian", "russianb", "samin", "scottish", "serbian", "serbian-latin",
  71 "slovak", "slovene", "spanish", "swedish", "thai", "turkish", "ukraineb",
  72 "ukrainian", "usorbian", "vietnam", "welsh", 0};
  73
  74 const char * const known_bahasa_languages[] = {"bahasa", "bahasai",
  75                                                 "indon", "indonesian", 0};
  76 const char * const known_bahasam_languages[] = {"bahasam", "malay",
  77                                                 "meyalu", 0};
  78 const char * const known_brazilian_languages[] = {"brazil", "brazilian", 0};
  79 const char * const known_french_languages[] = {"french", "frenchb", "francais",
  80                                                 "frenchle", "frenchpro", 0};
  81 const char * const known_german_languages[] = {"german", "germanb", 0};
  82 const char * const known_ngerman_languages[] = {"ngerman", "ngermanb", 0};
  83 const char * const known_portuguese_languages[] = {"portuges", "portuguese", 0};
  84 const char * const known_russian_languages[] = {"russian", "russianb", 0};
  85 const char * const known_ukrainian_languages[] = {"ukrainian", "ukraineb", 0};
  86
  87 //add these to known_english_quotes_languages when updating to lyxformat 268:
  88 //"chinese-simplified", "korean"
  89 // This requires first that support for non-babel languages (CJK) is added.
  90 const char * const known_english_quotes_languages[] = {"american", "canadian",
  91 "english", "esperanto", "hebrew", "irish", "scottish", "thai", 0};
  92
  93 //add this to known_french_quotes_languages when updating to
  94 //lyxformat 383: "turkmen"
  95 const char * const known_french_quotes_languages[] = {"albanian", "arabic",
  96 "basque", "canadien", "catalan", "galician", "greek", "italian", "norsk",
  97 "nynorsk", "polutonikogreek", "spanish", "spanish-mexico", "turkish",
  98 "vietnam", 0};
  99
 100 const char * const known_german_quotes_languages[] = {"austrian", "bulgarian",
 101 "czech", "icelandic", "lithuanian", "lsorbian", "naustrian", "serbian",
 102 "serbian-latin", "slovak", "slovene", "usorbian",  0};
 103
 104 const char * const known_polish_quotes_languages[] = {"afrikaans", "croatian",
 105 "dutch", "estonian", "magyar", "polish", "romanian", 0};
 106
 107 const char * const known_swedish_quotes_languages[] = {"finnish",
 108 "swedish", 0};
 109
 110 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
 111
 112 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
 113 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
 114
 115 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
 116 "helvet", "lmss", 0};
 117
 118 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
 119 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
 120 "newcent", 0};
 121
 122 const char * const known_paper_sizes[] = { "a3paper", "b3paper", "a4paper",
 123 "b4paper", "a5paper", "b5paper", "executivepaper", "legalpaper",
 124 "letterpaper", 0};
 125
 126 const char * const known_class_paper_sizes[] = { "a4paper", "a5paper",
 127 "executivepaper", "legalpaper", "letterpaper", 0};
 128
 129 const char * const known_paper_margins[] = { "lmargin", "tmargin", "rmargin",
 130 "bmargin", "headheight", "headsep", "footskip", "columnsep", 0};
 131
 132 const char * const known_coded_paper_margins[] = { "leftmargin", "topmargin",
 133 "rightmargin", "bottommargin", "headheight", "headsep", "footskip",
 134 "columnsep", 0};
 135
 136 // default settings
 137 ostringstream h_preamble;
 138 string h_textclass               = "article";
 139 string h_use_default_options     = "false";
 140 string h_options                 = string();
 141 string h_language                = "english";
 142 string h_inputencoding           = "auto";
 143 string h_font_roman              = "default";
 144 string h_font_sans               = "default";
 145 string h_font_typewriter         = "default";
 146 string h_font_default_family     = "default";
 147 string h_font_sc                 = "false";
 148 string h_font_osf                = "false";
 149 string h_font_sf_scale           = "100";
 150 string h_font_tt_scale           = "100";
 151 string h_graphics                = "default";
 152 string h_float_placement;
 153 string h_paperfontsize           = "default";
 154 string h_spacing                 = "single";
 155 string h_use_hyperref            = "0";
 156 string h_pdf_title;
 157 string h_pdf_author;
 158 string h_pdf_subject;
 159 string h_pdf_keywords;
 160 string h_pdf_bookmarks           = "1";
 161 string h_pdf_bookmarksnumbered   = "0";
 162 string h_pdf_bookmarksopen       = "0";
 163 string h_pdf_bookmarksopenlevel  = "1";
 164 string h_pdf_breaklinks          = "0";
 165 string h_pdf_pdfborder           = "0";
 166 string h_pdf_colorlinks          = "0";
 167 string h_pdf_backref             = "section";
 168 string h_pdf_pdfusetitle         = "1";
 169 string h_pdf_pagemode;
 170 string h_pdf_quoted_options;
 171 string h_papersize               = "default";
 172 string h_use_geometry            = "false";
 173 string h_use_amsmath             = "1";
 174 string h_use_esint               = "1";
 175 string h_cite_engine             = "basic";
 176 string h_use_bibtopic            = "false";
 177 string h_paperorientation        = "portrait";
 178 string h_secnumdepth             = "3";
 179 string h_tocdepth                = "3";
 180 string h_paragraph_separation    = "indent";
 181 string h_defskip                 = "medskip";
 182 string h_paragraph_indentation   = "default";
 183 string h_quotes_language         = "english";
 184 string h_papercolumns            = "1";
 185 string h_papersides              = string();
 186 string h_paperpagestyle          = "default";
 187 string h_listings_params;
 188 string h_tracking_changes        = "false";
 189 string h_output_changes          = "false";
 190 string h_margins                 = "";
 191
 192
 193 /// translates a babel language name to a LyX language name
 194 string babel2lyx(string language)
 195 {
 196         if (language == "arabtex")
 197                 return "arabic_arabtex";
 198         if (language == "arabic")
 199                 return "arabic_arabi";
 200         if (language == "lsorbian")
 201                 return "lowersorbian";
 202         if (language == "usorbian")
 203                 return "uppersorbian";
 204         return language;
 205 }
 206
 207
 208 // returns true if at least one of the options in what has been found
 209 bool handle_opt(vector<string> & opts, char const * const * what, string & target)
 210 {
 211         if (opts.empty())
 212                 return false;
 213
 214         bool found = false;
 215         // the last language option is the document language (for babel and LyX)
 216         // the last size option is the document font size
 217         vector<string>::iterator it;
 218         vector<string>::iterator position = opts.begin();
 219         for (; *what; ++what) {
 220                 it = find(opts.begin(), opts.end(), *what);
 221                 if (it != opts.end()) {
 222                         if (it >= position) {
 223                                 found = true;
 224                                 target = *what;
 225                                 position = it;
 226                         }
 227                 }
 228         }
 229         return found;
 230 }
 231
 232
 233 void delete_opt(vector<string> & opts, char const * const * what)
 234 {
 235         if (opts.empty())
 236                 return;
 237
 238         // remove found options from the list
 239         // do this after handle_opt to avoid potential memory leaks
 240         vector<string>::iterator it;
 241         for (; *what; ++what) {
 242                 it = find(opts.begin(), opts.end(), *what);
 243                 if (it != opts.end())
 244                         opts.erase(it);
 245         }
 246 }
 247
 248
 249 /*!
 250  * Split a package options string (keyval format) into a vector.
 251  * Example input:
 252  *   authorformat=smallcaps,
 253  *   commabeforerest,
 254  *   titleformat=colonsep,
 255  *   bibformat={tabular,ibidem,numbered}
 256  */
 257 vector<string> split_options(string const & input)
 258 {
 259         vector<string> options;
 260         string option;
 261         Parser p(input);
 262         while (p.good()) {
 263                 Token const & t = p.get_token();
 264                 if (t.asInput() == ",") {
 265                         options.push_back(trim(option));
 266                         option.erase();
 267                 } else if (t.asInput() == "=") {
 268                         option += '=';
 269                         p.skip_spaces(true);
 270                         if (p.next_token().asInput() == "{")
 271                                 option += '{' + p.getArg('{', '}') + '}';
 272                 } else if (t.cat() != catSpace)
 273                         option += t.asInput();
 274         }
 275
 276         if (!option.empty())
 277                 options.push_back(trim(option));
 278
 279         return options;
 280 }
 281
 282
 283 /*!
 284  * Add package \p name with options \p options to used_packages.
 285  * Remove options from \p options that we don't want to output.
 286  */
 287 void add_package(string const & name, vector<string> & options)
 288 {
 289         // every package inherits the global options
 290         if (used_packages.find(name) == used_packages.end())
 291                 used_packages[name] = split_options(h_options);
 292
 293         vector<string> & v = used_packages[name];
 294         v.insert(v.end(), options.begin(), options.end());
 295         if (name == "jurabib") {
 296                 // Don't output the order argument (see the cite command
 297                 // handling code in text.cpp).
 298                 vector<string>::iterator end =
 299                         remove(options.begin(), options.end(), "natbiborder");
 300                 end = remove(options.begin(), end, "jurabiborder");
 301                 options.erase(end, options.end());
 302         }
 303 }
 304
 305
 306 // Given is a string like "scaled=0.9", return 0.9 * 100
 307 string const scale_as_percentage(string const & scale)
 308 {
 309         string::size_type pos = scale.find('=');
 310         if (pos != string::npos) {
 311                 string value = scale.substr(pos + 1);
 312                 if (isStrDbl(value))
 313                         return convert<string>(100 * convert<double>(value));
 314         }
 315         // If the input string didn't match our expectations.
 316         // return the default value "100"
 317         return "100";
 318 }
 319
 320
 321 void handle_package(Parser &p, string const & name, string const & opts,
 322                     bool in_lyx_preamble)
 323 {
 324         vector<string> options = split_options(opts);
 325         add_package(name, options);
 326         string scale;
 327
 328         // roman fonts
 329         if (is_known(name, known_roman_fonts)) {
 330                 h_font_roman = name;
 331                 p.skip_spaces();
 332         }
 333
 334         if (name == "fourier") {
 335                 h_font_roman = "utopia";
 336                 // when font uses real small capitals
 337                 if (opts == "expert")
 338                         h_font_sc = "true";
 339         }
 340
 341         if (name == "mathpazo")
 342                 h_font_roman = "palatino";
 343
 344         if (name == "mathptmx")
 345                 h_font_roman = "times";
 346
 347         // sansserif fonts
 348         if (is_known(name, known_sans_fonts)) {
 349                 h_font_sans = name;
 350                 if (!opts.empty()) {
 351                         scale = opts;
 352                         h_font_sf_scale = scale_as_percentage(scale);
 353                 }
 354         }
 355
 356         // typewriter fonts
 357         if (is_known(name, known_typewriter_fonts)) {
 358                 h_font_typewriter = name;
 359                 if (!opts.empty()) {
 360                         scale = opts;
 361                         h_font_tt_scale = scale_as_percentage(scale);
 362                 }
 363         }
 364
 365         // font uses old-style figure
 366         if (name == "eco")
 367                 h_font_osf = "true";
 368
 369         else if (name == "amsmath" || name == "amssymb")
 370                 h_use_amsmath = "2";
 371
 372         else if (name == "esint")
 373                 h_use_esint = "2";
 374
 375         else if (name == "babel" && !opts.empty()) {
 376                 // check if more than one option was used - used later for inputenc
 377                 // in case inputenc is parsed before babel, set the encoding to auto
 378                 if (options.begin() != options.end() - 1) {
 379                         one_language = false;
 380                         h_inputencoding = "auto";
 381                 }
 382                 // babel takes the last language of the option of its \usepackage
 383                 // call as document language. If there is no such language option, the
 384                 // last language in the documentclass options is used.
 385                 handle_opt(options, known_languages, h_language);
 386                 delete_opt(options, known_languages);
 387         }
 388
 389         else if (name == "fontenc")
 390                  ;// ignore this
 391
 392         else if (name == "inputenc") {
 393                 // h_inputencoding is only set when there is not more than one
 394                 // inputenc option because otherwise h_inputencoding must be
 395                 // set to "auto" (the default encoding of the document language)
 396                 // Therefore check for the "," character.
 397                 // It is also only set when there is not more then one babel
 398                 // language option but this is handled in the routine for babel.
 399                 if (opts.find(",") == string::npos && one_language == true)
 400                         h_inputencoding = opts;
 401                 if (!options.empty())
 402                         p.setEncoding(options.back());
 403                 options.clear();
 404         }
 405
 406         else if (name == "makeidx")
 407                 ; // ignore this
 408
 409         else if (name == "prettyref")
 410                 ; // ignore this
 411
 412         else if (name == "varioref")
 413                 ; // ignore this
 414
 415         else if (name == "verbatim")
 416                 ; // ignore this
 417
 418         else if (name == "nomencl")
 419                 ; // ignore this
 420
 421         else if (name == "textcomp")
 422                 ; // ignore this
 423
 424         else if (name == "url")
 425                 ; // ignore this
 426
 427         else if (LYX_FORMAT >= 408 && name == "subscript")
 428                 ; // ignore this
 429
 430         else if (name == "color") {
 431                 // with the following command this package is only loaded when needed for
 432                 // undefined colors, since we only support the predefined colors
 433                 h_preamble << "\\@ifundefined{definecolor}\n {\\usepackage{color}}{}\n";
 434         }
 435
 436         else if (name == "graphicx")
 437                 ; // ignore this
 438
 439         else if (name == "setspace")
 440                 ; // ignore this
 441
 442         else if (name == "geometry")
 443                 ; // Ignore this, the geometry settings are made by the \geometry
 444                   // command. This command is handled below.
 445
 446         else if (is_known(name, known_languages))
 447                 h_language = name;
 448
 449         else if (name == "natbib") {
 450                 h_cite_engine = "natbib_authoryear";
 451                 vector<string>::iterator it =
 452                         find(options.begin(), options.end(), "authoryear");
 453                 if (it != options.end())
 454                         options.erase(it);
 455                 else {
 456                         it = find(options.begin(), options.end(), "numbers");
 457                         if (it != options.end()) {
 458                                 h_cite_engine = "natbib_numerical";
 459                                 options.erase(it);
 460                         }
 461                 }
 462         }
 463
 464         else if (name == "jurabib")
 465                 h_cite_engine = "jurabib";
 466
 467         else if (!in_lyx_preamble) {
 468                 if (options.empty())
 469                         h_preamble << "\\usepackage{" << name << "}";
 470                 else {
 471                         h_preamble << "\\usepackage[" << opts << "]{"
 472                                    << name << "}";
 473                         options.clear();
 474                 }
 475         }
 476
 477         // We need to do something with the options...
 478         if (!options.empty())
 479                 cerr << "Ignoring options '" << join(options, ",")
 480                      << "' of package " << name << '.' << endl;
 481
 482         // remove the whitespace
 483         p.skip_spaces();
 484 }
 485
 486
 487
 488 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 489 {
 490         // merge synonym languages
 491         if (is_known(h_language, known_bahasa_languages))
 492                 h_language = "bahasa";
 493         else if (is_known(h_language, known_bahasam_languages))
 494                 h_language = "bahasam";
 495         else if (is_known(h_language, known_brazilian_languages))
 496                 h_language = "brazilian";
 497         else if (is_known(h_language, known_french_languages))
 498                 h_language = "french";
 499         else if (is_known(h_language, known_german_languages))
 500                 h_language = "german";
 501         else if (is_known(h_language, known_ngerman_languages))
 502                 h_language = "ngerman";
 503         else if (is_known(h_language, known_portuguese_languages))
 504                 h_language = "portuguese";
 505         else if (is_known(h_language, known_russian_languages))
 506                 h_language = "russian";
 507         else if (is_known(h_language, known_ukrainian_languages))
 508                 h_language = "ukrainian";
 509
 510         // set the quote language
 511         // LyX only knows the following quotes languages:
 512         // english, swedish, german, polish, french and danish
 513         // (quotes for "japanese" and "chinese-traditional" are missing because
 514         //  they wouldn't be useful: http://www.lyx.org/trac/ticket/6383)
 515         // conversion list taken from
 516         // http://en.wikipedia.org/wiki/Quotation_mark,_non-English_usage
 517         // (quotes for kazakh and interlingua are unknown)
 518         // danish
 519         if (h_language == "danish")
 520                 h_quotes_language = "danish";
 521         // french
 522         else if (is_known(h_language, known_french_quotes_languages)
 523                 || is_known(h_language, known_french_languages)
 524                 || is_known(h_language, known_russian_languages)
 525                 || is_known(h_language, known_ukrainian_languages))
 526                 h_quotes_language = "french";
 527         // german
 528         else if (is_known(h_language, known_german_quotes_languages)
 529                 || is_known(h_language, known_german_languages)
 530                 || is_known(h_language, known_ngerman_languages))
 531                 h_quotes_language = "german";
 532         // polish
 533         else if (is_known(h_language, known_polish_quotes_languages))
 534                 h_quotes_language = "polish";
 535         // swedish
 536         else if (is_known(h_language, known_swedish_quotes_languages))
 537                 h_quotes_language = "swedish";
 538         //english
 539         else if (is_known(h_language, known_english_quotes_languages)
 540                 || is_known(h_language, known_bahasa_languages)
 541                 || is_known(h_language, known_bahasam_languages)
 542                 || is_known(h_language, known_brazilian_languages)
 543                 || is_known(h_language, known_portuguese_languages))
 544                 h_quotes_language = "english";
 545
 546         h_language = babel2lyx(h_language);
 547
 548         // output the LyX file settings
 549         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 550            << "\\lyxformat " << LYX_FORMAT << '\n'
 551            << "\\begin_document\n"
 552            << "\\begin_header\n"
 553            << "\\textclass " << h_textclass << "\n";
 554         if (!h_preamble.str().empty())
 555                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 556         if (!h_options.empty())
 557                 os << "\\options " << h_options << "\n";
 558         os << "\\use_default_options " << h_use_default_options << "\n"
 559            << "\\language " << h_language << "\n"
 560            << "\\inputencoding " << h_inputencoding << "\n"
 561            << "\\font_roman " << h_font_roman << "\n"
 562            << "\\font_sans " << h_font_sans << "\n"
 563            << "\\font_typewriter " << h_font_typewriter << "\n"
 564            << "\\font_default_family " << h_font_default_family << "\n"
 565            << "\\font_sc " << h_font_sc << "\n"
 566            << "\\font_osf " << h_font_osf << "\n"
 567            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 568            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 569            << "\\graphics " << h_graphics << "\n";
 570         if (!h_float_placement.empty())
 571                 os << "\\float_placement " << h_float_placement << "\n";
 572         os << "\\paperfontsize " << h_paperfontsize << "\n"
 573            << "\\spacing " << h_spacing << "\n"
 574            << "\\use_hyperref " << h_use_hyperref << '\n';
 575         if (h_use_hyperref == "1") {
 576                 if (!h_pdf_title.empty())
 577                         os << "\\pdf_title \"" << h_pdf_title << "\"\n";
 578                 if (!h_pdf_author.empty())
 579                         os << "\\pdf_author \"" << h_pdf_author << "\"\n";
 580                 if (!h_pdf_subject.empty())
 581                         os << "\\pdf_subject \"" << h_pdf_subject << "\"\n";
 582                 if (!h_pdf_keywords.empty())
 583                         os << "\\pdf_keywords \"" << h_pdf_keywords << "\"\n";
 584                 os << "\\pdf_bookmarks " << h_pdf_bookmarks << "\n"
 585                       "\\pdf_bookmarksnumbered " << h_pdf_bookmarksnumbered << "\n"
 586                       "\\pdf_bookmarksopen " << h_pdf_bookmarksopen << "\n"
 587                       "\\pdf_bookmarksopenlevel " << h_pdf_bookmarksopenlevel << "\n"
 588                       "\\pdf_breaklinks " << h_pdf_breaklinks << "\n"
 589                       "\\pdf_pdfborder " << h_pdf_pdfborder << "\n"
 590                       "\\pdf_colorlinks " << h_pdf_colorlinks << "\n"
 591                       "\\pdf_backref " << h_pdf_backref << "\n"
 592                       "\\pdf_pdfusetitle " << h_pdf_pdfusetitle << '\n';
 593                 if (!h_pdf_pagemode.empty())
 594                         os << "\\pdf_pagemode " << h_pdf_pagemode << '\n';
 595                 if (!h_pdf_quoted_options.empty())
 596                         os << "\\pdf_quoted_options \"" << h_pdf_quoted_options << "\"\n";
 597         }
 598         os << "\\papersize " << h_papersize << "\n"
 599            << "\\use_geometry " << h_use_geometry << "\n"
 600            << "\\use_amsmath " << h_use_amsmath << "\n"
 601            << "\\use_esint " << h_use_esint << "\n"
 602            << "\\cite_engine " << h_cite_engine << "\n"
 603            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 604            << "\\paperorientation " << h_paperorientation << "\n"
 605            << h_margins
 606            << "\\secnumdepth " << h_secnumdepth << "\n"
 607            << "\\tocdepth " << h_tocdepth << "\n"
 608            << "\\paragraph_separation " << h_paragraph_separation << "\n";
 609         if (LYX_FORMAT < 365 || h_paragraph_separation == "skip")
 610                 os << "\\defskip " << h_defskip << "\n";
 611         else
 612                 os << "\\paragraph_indentation " << h_paragraph_indentation << "\n";
 613         os << "\\quotes_language " << h_quotes_language << "\n"
 614            << "\\papercolumns " << h_papercolumns << "\n"
 615            << "\\papersides " << h_papersides << "\n"
 616            << "\\paperpagestyle " << h_paperpagestyle << "\n";
 617         if (!h_listings_params.empty())
 618                 os << "\\listings_params " << h_listings_params << "\n";
 619         os << "\\tracking_changes " << h_tracking_changes << "\n"
 620            << "\\output_changes " << h_output_changes << "\n"
 621            << "\\end_header\n\n"
 622            << "\\begin_body\n";
 623         // clear preamble for subdocuments
 624         h_preamble.str("");
 625 }
 626
 627 } // anonymous namespace
 628
 629 void parse_preamble(Parser & p, ostream & os,
 630         string const & forceclass, TeX2LyXDocClass & tc)
 631 {
 632         // initialize fixed types
 633         special_columns['D'] = 3;
 634         bool is_full_document = false;
 635         bool is_lyx_file = false;
 636         bool in_lyx_preamble = false;
 637
 638         // determine whether this is a full document or a fragment for inclusion
 639         while (p.good()) {
 640                 Token const & t = p.get_token();
 641
 642                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 643                         is_full_document = true;
 644                         break;
 645                 }
 646         }
 647         p.reset();
 648
 649         while (is_full_document && p.good()) {
 650                 Token const & t = p.get_token();
 651
 652 #ifdef FILEDEBUG
 653                 cerr << "t: " << t << "\n";
 654 #endif
 655
 656                 //
 657                 // cat codes
 658                 //
 659                 if (!in_lyx_preamble &&
 660                     (t.cat() == catLetter ||
 661                      t.cat() == catSuper ||
 662                      t.cat() == catSub ||
 663                      t.cat() == catOther ||
 664                      t.cat() == catMath ||
 665                      t.cat() == catActive ||
 666                      t.cat() == catBegin ||
 667                      t.cat() == catEnd ||
 668                      t.cat() == catAlign ||
 669                      t.cat() == catParameter))
 670                         h_preamble << t.cs();
 671
 672                 else if (!in_lyx_preamble &&
 673                          (t.cat() == catSpace || t.cat() == catNewline))
 674                         h_preamble << t.asInput();
 675
 676                 else if (t.cat() == catComment) {
 677                         // regex to parse comments (currently not used)
 678                         static regex const islyxfile("%% LyX .* created this file");
 679                         static regex const usercommands("User specified LaTeX commands");
 680
 681                         string const comment = t.asInput();
 682
 683                         // magically switch encoding default if it looks like XeLaTeX
 684                         static string const magicXeLaTeX =
 685                                 "% This document must be compiled with XeLaTeX ";
 686                         if (comment.size() > magicXeLaTeX.size()
 687                                   && comment.substr(0, magicXeLaTeX.size()) == magicXeLaTeX
 688                                   && h_inputencoding == "auto") {
 689                                 cerr << "XeLaTeX comment found, switching to UTF8\n";
 690                                 h_inputencoding = "utf8";
 691                         }
 692                         smatch sub;
 693                         if (regex_search(comment, sub, islyxfile)) {
 694                                 is_lyx_file = true;
 695                                 in_lyx_preamble = true;
 696                         } else if (is_lyx_file
 697                                    && regex_search(comment, sub, usercommands))
 698                                 in_lyx_preamble = false;
 699                         else if (!in_lyx_preamble)
 700                                 h_preamble << t.asInput();
 701                 }
 702
 703                 else if (t.cs() == "pagestyle")
 704                         h_paperpagestyle = p.verbatim_item();
 705
 706                 else if (t.cs() == "makeatletter") {
 707                         // LyX takes care of this
 708                         p.setCatCode('@', catLetter);
 709                 }
 710
 711                 else if (t.cs() == "makeatother") {
 712                         // LyX takes care of this
 713                         p.setCatCode('@', catOther);
 714                 }
 715
 716                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 717                             || t.cs() == "providecommand"
 718                                 || t.cs() == "DeclareRobustCommand"
 719                                 || t.cs() == "ProvideTextCommandDefault"
 720                                 || t.cs() == "DeclareMathAccent") {
 721                         bool star = false;
 722                         if (p.next_token().character() == '*') {
 723                                 p.get_token();
 724                                 star = true;
 725                         }
 726                         string const name = p.verbatim_item();
 727                         string const opt1 = p.getOpt();
 728                         string const opt2 = p.getFullOpt();
 729                         string const body = p.verbatim_item();
 730                         // font settings
 731                         if (name == "\\rmdefault")
 732                                 if (is_known(body, known_roman_fonts))
 733                                         h_font_roman = body;
 734                         if (name == "\\sfdefault")
 735                                 if (is_known(body, known_sans_fonts))
 736                                         h_font_sans = body;
 737                         if (name == "\\ttdefault")
 738                                 if (is_known(body, known_typewriter_fonts))
 739                                         h_font_typewriter = body;
 740                         if (name == "\\familydefault") {
 741                                 string family = body;
 742                                 // remove leading "\"
 743                                 h_font_default_family = family.erase(0,1);
 744                         }
 745                         // only non-lyxspecific stuff
 746                         if (!in_lyx_preamble) {
 747                                 ostringstream ss;
 748                                 ss << '\\' << t.cs();
 749                                 if (star)
 750                                         ss << '*';
 751                                 ss << '{' << name << '}' << opt1 << opt2
 752                                    << '{' << body << "}";
 753                                 h_preamble << ss.str();
 754
 755                                 // Add the command to the known commands
 756                                 add_known_command(name, opt1, !opt2.empty());
 757 /*
 758                                 ostream & out = in_preamble ? h_preamble : os;
 759                                 out << "\\" << t.cs() << "{" << name << "}"
 760                                     << opts << "{" << body << "}";
 761 */
 762                         }
 763                 }
 764
 765                 else if (t.cs() == "documentclass") {
 766                         vector<string>::iterator it;
 767                         vector<string> opts = split_options(p.getArg('[', ']'));
 768                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 769                         delete_opt(opts, known_fontsizes);
 770                         // delete "pt" at the end
 771                         string::size_type i = h_paperfontsize.find("pt");
 772                         if (i != string::npos)
 773                                 h_paperfontsize.erase(i);
 774                         // The documentclass options are always parsed before the options
 775                         // of the babel call so that a language cannot overwrite the babel
 776                         // options.
 777                         handle_opt(opts, known_languages, h_language);
 778                         delete_opt(opts, known_languages);
 779
 780                         // paper orientation
 781                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 782                                 h_paperorientation = "landscape";
 783                                 opts.erase(it);
 784                         }
 785                         // paper sides
 786                         if ((it = find(opts.begin(), opts.end(), "oneside"))
 787                                  != opts.end()) {
 788                                 h_papersides = "1";
 789                                 opts.erase(it);
 790                         }
 791                         if ((it = find(opts.begin(), opts.end(), "twoside"))
 792                                  != opts.end()) {
 793                                 h_papersides = "2";
 794                                 opts.erase(it);
 795                         }
 796                         // paper columns
 797                         if ((it = find(opts.begin(), opts.end(), "onecolumn"))
 798                                  != opts.end()) {
 799                                 h_papercolumns = "1";
 800                                 opts.erase(it);
 801                         }
 802                         if ((it = find(opts.begin(), opts.end(), "twocolumn"))
 803                                  != opts.end()) {
 804                                 h_papercolumns = "2";
 805                                 opts.erase(it);
 806                         }
 807                         // paper sizes
 808                         // some size options are know to any document classes, other sizes
 809                         // are handled by the \geometry command of the geometry package
 810                         handle_opt(opts, known_class_paper_sizes, h_papersize);
 811                         delete_opt(opts, known_class_paper_sizes);
 812                         // the remaining options
 813                         h_options = join(opts, ",");
 814                         h_textclass = p.getArg('{', '}');
 815                 }
 816
 817                 else if (t.cs() == "usepackage") {
 818                         string const options = p.getArg('[', ']');
 819                         string const name = p.getArg('{', '}');
 820                         vector<string> vecnames;
 821                         split(name, vecnames, ',');
 822                         vector<string>::const_iterator it  = vecnames.begin();
 823                         vector<string>::const_iterator end = vecnames.end();
 824                         for (; it != end; ++it)
 825                                 handle_package(p, trim(*it), options,
 826                                                in_lyx_preamble);
 827                 }
 828
 829                 else if (t.cs() == "inputencoding") {
 830                         string const encoding = p.getArg('{','}');
 831                         h_inputencoding = encoding;
 832                         p.setEncoding(encoding);
 833                 }
 834
 835                 else if (t.cs() == "newenvironment") {
 836                         string const name = p.getArg('{', '}');
 837                         ostringstream ss;
 838                         ss << "\\newenvironment{" << name << "}";
 839                         ss << p.getOpt();
 840                         ss << p.getOpt();
 841                         ss << '{' << p.verbatim_item() << '}';
 842                         ss << '{' << p.verbatim_item() << '}';
 843                         if (!in_lyx_preamble)
 844                                 h_preamble << ss.str();
 845                 }
 846
 847                 else if (t.cs() == "def") {
 848                         string name = p.get_token().cs();
 849                         while (p.next_token().cat() != catBegin)
 850                                 name += p.get_token().cs();
 851                         if (!in_lyx_preamble)
 852                                 h_preamble << "\\def\\" << name << '{'
 853                                            << p.verbatim_item() << "}";
 854                 }
 855
 856                 else if (t.cs() == "newcolumntype") {
 857                         string const name = p.getArg('{', '}');
 858                         trim(name);
 859                         int nargs = 0;
 860                         string opts = p.getOpt();
 861                         if (!opts.empty()) {
 862                                 istringstream is(string(opts, 1));
 863                                 is >> nargs;
 864                         }
 865                         special_columns[name[0]] = nargs;
 866                         h_preamble << "\\newcolumntype{" << name << "}";
 867                         if (nargs)
 868                                 h_preamble << "[" << nargs << "]";
 869                         h_preamble << "{" << p.verbatim_item() << "}";
 870                 }
 871
 872                 else if (t.cs() == "setcounter") {
 873                         string const name = p.getArg('{', '}');
 874                         string const content = p.getArg('{', '}');
 875                         if (name == "secnumdepth")
 876                                 h_secnumdepth = content;
 877                         else if (name == "tocdepth")
 878                                 h_tocdepth = content;
 879                         else
 880                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
 881                 }
 882
 883                 else if (t.cs() == "setlength") {
 884                         string const name = p.verbatim_item();
 885                         string const content = p.verbatim_item();
 886                         // the paragraphs are only not indented when \parindent is set to zero
 887                         if (name == "\\parindent" && content != "") {
 888                                 if (content[0] == '0')
 889                                         h_paragraph_separation = "skip";
 890                                 else if (LYX_FORMAT >= 365)
 891                                         h_paragraph_indentation = translate_len(content);
 892                                 else
 893                                         h_preamble << "\\setlength{" << name
 894                                                    << "}{" << content << "}";
 895                         } else if (name == "\\parskip") {
 896                                 if (content == "\\smallskipamount")
 897                                         h_defskip = "smallskip";
 898                                 else if (content == "\\medskipamount")
 899                                         h_defskip = "medskip";
 900                                 else if (content == "\\bigskipamount")
 901                                         h_defskip = "bigskip";
 902                                 else
 903                                         h_defskip = content;
 904                         } else
 905                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
 906                 }
 907
 908                 else if (t.cs() == "onehalfspacing")
 909                         h_spacing = "onehalf";
 910
 911                 else if (t.cs() == "doublespacing")
 912                         h_spacing = "double";
 913
 914                 else if (t.cs() == "setstretch")
 915                         h_spacing = "other " + p.verbatim_item();
 916
 917                 else if (t.cs() == "begin") {
 918                         string const name = p.getArg('{', '}');
 919                         if (name == "document")
 920                                 break;
 921                         h_preamble << "\\begin{" << name << "}";
 922                 }
 923
 924                 else if (t.cs() == "geometry") {
 925                         h_use_geometry = "true";
 926                         vector<string> opts = split_options(p.getArg('{', '}'));
 927                         vector<string>::iterator it;
 928                         // paper orientation
 929                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 930                                 h_paperorientation = "landscape";
 931                                 opts.erase(it);
 932                         }
 933                         // paper size
 934                         handle_opt(opts, known_paper_sizes, h_papersize);
 935                         delete_opt(opts, known_paper_sizes);
 936                         // page margins
 937                         char const * const * margin = known_paper_margins;
 938                         int k = -1;
 939                         for (; *margin; ++margin) {
 940                                 k += 1;
 941                                 // search for the "=" in e.g. "lmargin=2cm" to get the value
 942                                 for(size_t i = 0; i != opts.size(); i++) {
 943                                         if (opts.at(i).find(*margin) != string::npos) {
 944                                                 string::size_type pos = opts.at(i).find("=");
 945                                                 string value = opts.at(i).substr(pos + 1);
 946                                                 string name = known_coded_paper_margins[k];
 947                                                 h_margins += "\\" + name + " " + value + "\n";
 948                                         }
 949                                 }
 950                         }
 951                 }
 952
 953                 else if (t.cs() == "jurabibsetup") {
 954                         vector<string> jurabibsetup =
 955                                 split_options(p.getArg('{', '}'));
 956                         // add jurabibsetup to the jurabib package options
 957                         add_package("jurabib", jurabibsetup);
 958                         if (!jurabibsetup.empty()) {
 959                                 h_preamble << "\\jurabibsetup{"
 960                                            << join(jurabibsetup, ",") << '}';
 961                         }
 962                 }
 963
 964                 else if (!t.cs().empty() && !in_lyx_preamble)
 965                         h_preamble << '\\' << t.cs();
 966         }
 967
 968         // remove the whitespace
 969         p.skip_spaces();
 970
 971         // Force textclass if the user wanted it
 972         if (!forceclass.empty())
 973                 h_textclass = forceclass;
 974         if (noweb_mode && !prefixIs(h_textclass, "literate-"))
 975                 h_textclass.insert(0, "literate-");
 976         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
 977         if (layoutfilename.empty()) {
 978                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
 979                 exit(1);
 980         }
 981         tc.read(layoutfilename);
 982         if (h_papersides.empty()) {
 983                 ostringstream ss;
 984                 ss << tc.sides();
 985                 h_papersides = ss.str();
 986         }
 987         end_preamble(os, tc);
 988 }
 989
 990 // }])
 991
 992
 993 } // namespace lyx