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