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