src/tex2lyx/preamble.cpp

   1 /**
   2  * \file preamble.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author André Pönitz
   7  * \author Uwe Stöhr
   8  *
   9  * Full author contact details are available in file CREDITS.
  10  */
  11
  12 // {[(
  13
  14 #include <config.h>
  15
  16 #include "tex2lyx.h"
  17
  18 #include "LayoutFile.h"
  19 #include "Layout.h"
  20 #include "Lexer.h"
  21 #include "TextClass.h"
  22
  23 #include "support/convert.h"
  24 #include "support/FileName.h"
  25 #include "support/filetools.h"
  26 #include "support/lstrings.h"
  27
  28 #include "support/regex.h"
  29
  30 #include <algorithm>
  31 #include <iostream>
  32 #include <sstream>
  33 #include <string>
  34 #include <vector>
  35 #include <map>
  36
  37 using namespace std;
  38 using namespace lyx::support;
  39
  40
  41 namespace lyx {
  42
  43 // special columntypes
  44 extern map<char, int> special_columns;
  45
  46 map<string, vector<string> > used_packages;
  47
  48 // needed to handle encodings with babel
  49 bool one_language = true;
  50
  51 namespace {
  52
  53 //add this to known_languages when updating to lyxformat 266:
  54 // "armenian"
  55 //add these to known_languages when updating to lyxformat 268:
  56 //"chinese-simplified", "chinese-traditional", "japanese", "korean"
  57 // Both changes require first that support for non-babel languages (CJK,
  58 // armtex) is added.
  59 // add turkmen for lyxformat 383
  60 const char * const known_languages[] = { "afrikaans", "albanian", "american",
  61 "arabic", "arabtex", "austrian", "bahasa", "bahasai", "bahasam", "basque",
  62 "belarusian", "brazil", "brazilian", "breton", "british", "bulgarian",
  63 "canadian", "canadien", "catalan", "croatian", "czech", "danish", "dutch",
  64 "english", "esperanto", "estonian", "finnish", "francais", "french",
  65 "frenchb", "frenchle", "frenchpro", "galician", "german", "germanb", "greek",
  66 "hebrew", "icelandic", "indon", "indonesian", "interlingua", "irish",
  67 "italian", "kazakh", "latin", "latvian", "lithuanian", "lsorbian", "magyar",
  68 "malay", "meyalu", "mongolian", "naustrian", "ngerman", "ngermanb", "norsk",
  69 "nynorsk", "polutonikogreek", "polish", "portuges", "portuguese", "romanian",
  70 "russian", "russianb", "samin", "scottish", "serbian", "serbian-latin",
  71 "slovak", "slovene", "spanish", "swedish", "thai", "turkish", "ukraineb",
  72 "ukrainian", "usorbian", "vietnam", "welsh", 0};
  73
  74 const char * const known_bahasa_languages[] = {"bahasa", "bahasai",
  75                                                 "indon", "indonesian", 0};
  76 const char * const known_bahasam_languages[] = {"bahasam", "malay",
  77                                                 "meyalu", 0};
  78 const char * const known_brazilian_languages[] = {"brazil", "brazilian", 0};
  79 const char * const known_french_languages[] = {"french", "frenchb", "francais",
  80                                                 "frenchle", "frenchpro", 0};
  81 const char * const known_german_languages[] = {"german", "germanb", 0};
  82 const char * const known_ngerman_languages[] = {"ngerman", "ngermanb", 0};
  83 const char * const known_portuguese_languages[] = {"portuges", "portuguese", 0};
  84 const char * const known_russian_languages[] = {"russian", "russianb", 0};
  85 const char * const known_ukrainian_languages[] = {"ukrainian", "ukraineb", 0};
  86
  87 //add these to known_english_quotes_languages when updating to lyxformat 268:
  88 //"chinese-simplified", "korean"
  89 // This requires first that support for non-babel languages (CJK) is added.
  90 const char * const known_english_quotes_languages[] = {"american", "canadian",
  91 "english", "esperanto", "hebrew", "irish", "scottish", "thai", 0};
  92
  93 //add this to known_french_quotes_languages when updating to
  94 //lyxformat 383: "turkmen"
  95 const char * const known_french_quotes_languages[] = {"albanian", "arabic",
  96 "basque", "canadien", "catalan", "galician", "greek", "italian", "norsk",
  97 "nynorsk", "polutonikogreek", "spanish", "spanish-mexico", "turkish",
  98 "vietnam", 0};
  99
 100 const char * const known_german_quotes_languages[] = {"austrian", "bulgarian",
 101 "czech", "icelandic", "lithuanian", "lsorbian", "naustrian", "serbian",
 102 "serbian-latin", "slovak", "slovene", "usorbian",  0};
 103
 104 const char * const known_polish_quotes_languages[] = {"afrikaans", "croatian",
 105 "dutch", "estonian", "magyar", "polish", "romanian", 0};
 106
 107 const char * const known_swedish_quotes_languages[] = {"finnish",
 108 "swedish", 0};
 109
 110 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
 111
 112 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
 113 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
 114
 115 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
 116 "helvet", "lmss", 0};
 117
 118 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
 119 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
 120 "newcent", 0};
 121
 122 const char * const known_paper_sizes[] = { "a3paper", "b3paper", "a4paper",
 123 "b4paper", "a5paper", "b5paper", "executivepaper", "legalpaper",
 124 "letterpaper", 0};
 125
 126 const char * const known_class_paper_sizes[] = { "a4paper", "a5paper",
 127 "executivepaper", "legalpaper", "letterpaper", 0};
 128
 129 const char * const known_paper_margins[] = { "lmargin", "tmargin", "rmargin",
 130 "bmargin", "headheight", "headsep", "footskip", "columnsep", 0};
 131
 132 const char * const known_coded_paper_margins[] = { "leftmargin", "topmargin",
 133 "rightmargin", "bottommargin", "headheight", "headsep", "footskip",
 134 "columnsep", 0};
 135
 136 // default settings
 137 ostringstream h_preamble;
 138 string h_textclass               = "article";
 139 string h_use_default_options     = "false";
 140 string h_options                 = string();
 141 string h_language                = "english";
 142 string h_inputencoding           = "auto";
 143 string h_font_roman              = "default";
 144 string h_font_sans               = "default";
 145 string h_font_typewriter         = "default";
 146 string h_font_default_family     = "default";
 147 string h_font_sc                 = "false";
 148 string h_font_osf                = "false";
 149 string h_font_sf_scale           = "100";
 150 string h_font_tt_scale           = "100";
 151 string h_graphics                = "default";
 152 string h_float_placement;
 153 string h_paperfontsize           = "default";
 154 string h_spacing                 = "single";
 155 string h_use_hyperref            = "0";
 156 string h_pdf_title;
 157 string h_pdf_author;
 158 string h_pdf_subject;
 159 string h_pdf_keywords;
 160 string h_pdf_bookmarks           = "1";
 161 string h_pdf_bookmarksnumbered   = "0";
 162 string h_pdf_bookmarksopen       = "0";
 163 string h_pdf_bookmarksopenlevel  = "1";
 164 string h_pdf_breaklinks          = "0";
 165 string h_pdf_pdfborder           = "0";
 166 string h_pdf_colorlinks          = "0";
 167 string h_pdf_backref             = "section";
 168 string h_pdf_pdfusetitle         = "1";
 169 string h_pdf_pagemode;
 170 string h_pdf_quoted_options;
 171 string h_papersize               = "default";
 172 string h_use_geometry            = "false";
 173 string h_use_amsmath             = "1";
 174 string h_use_esint               = "1";
 175 string h_cite_engine             = "basic";
 176 string h_use_bibtopic            = "false";
 177 string h_paperorientation        = "portrait";
 178 string h_secnumdepth             = "3";
 179 string h_tocdepth                = "3";
 180 string h_paragraph_separation    = "indent";
 181 string h_defskip                 = "medskip";
 182 string h_paragraph_indentation   = "default";
 183 string h_quotes_language         = "english";
 184 string h_papercolumns            = "1";
 185 string h_papersides              = string();
 186 string h_paperpagestyle          = "default";
 187 string h_listings_params;
 188 string h_tracking_changes        = "false";
 189 string h_output_changes          = "false";
 190 string h_margins                 = "";
 191
 192
 193 /// translates a babel language name to a LyX language name
 194 string babel2lyx(string language)
 195 {
 196         if (language == "arabtex")
 197                 return "arabic_arabtex";
 198         if (language == "arabic")
 199                 return "arabic_arabi";
 200         if (language == "lsorbian")
 201                 return "lowersorbian";
 202         if (language == "usorbian")
 203                 return "uppersorbian";
 204         return language;
 205 }
 206
 207
 208 // returns true if at least one of the options in what has been found
 209 bool handle_opt(vector<string> & opts, char const * const * what, string & target)
 210 {
 211         if (opts.empty())
 212                 return false;
 213
 214         bool found = false;
 215         // the last language option is the document language (for babel and LyX)
 216         // the last size option is the document font size
 217         vector<string>::iterator it;
 218         vector<string>::iterator position = opts.begin();
 219         for (; *what; ++what) {
 220                 it = find(opts.begin(), opts.end(), *what);
 221                 if (it != opts.end()) {
 222                         if (it >= position) {
 223                                 found = true;
 224                                 target = *what;
 225                                 position = it;
 226                         }
 227                 }
 228         }
 229         return found;
 230 }
 231
 232
 233 void delete_opt(vector<string> & opts, char const * const * what)
 234 {
 235         if (opts.empty())
 236                 return;
 237
 238         // remove found options from the list
 239         // do this after handle_opt to avoid potential memory leaks
 240         vector<string>::iterator it;
 241         for (; *what; ++what) {
 242                 it = find(opts.begin(), opts.end(), *what);
 243                 if (it != opts.end())
 244                         opts.erase(it);
 245         }
 246 }
 247
 248
 249 /*!
 250  * Split a package options string (keyval format) into a vector.
 251  * Example input:
 252  *   authorformat=smallcaps,
 253  *   commabeforerest,
 254  *   titleformat=colonsep,
 255  *   bibformat={tabular,ibidem,numbered}
 256  */
 257 vector<string> split_options(string const & input)
 258 {
 259         vector<string> options;
 260         string option;
 261         Parser p(input);
 262         while (p.good()) {
 263                 Token const & t = p.get_token();
 264                 if (t.asInput() == ",") {
 265                         options.push_back(trim(option));
 266                         option.erase();
 267                 } else if (t.asInput() == "=") {
 268                         option += '=';
 269                         p.skip_spaces(true);
 270                         if (p.next_token().asInput() == "{")
 271                                 option += '{' + p.getArg('{', '}') + '}';
 272                 } else if (t.cat() != catSpace)
 273                         option += t.asInput();
 274         }
 275
 276         if (!option.empty())
 277                 options.push_back(trim(option));
 278
 279         return options;
 280 }
 281
 282
 283 /*!
 284  * Retrieve a keyval option "name={value with=sign}" named \p name from
 285  * \p options and return the value.
 286  * The found option is also removed from \p options.
 287  */
 288 string process_keyval_opt(vector<string> & options, string name)
 289 {
 290         for (size_t i = 0; i < options.size(); ++i) {
 291                 vector<string> option;
 292                 split(options[i], option, '=');
 293                 if (option.size() < 2)
 294                         continue;
 295                 if (option[0] == name) {
 296                         options.erase(options.begin() + i);
 297                         option.erase(option.begin());
 298                         return join(option, "=");
 299                 }
 300         }
 301         return "";
 302 }
 303
 304
 305 /*!
 306  * Add package \p name with options \p options to used_packages.
 307  * Remove options from \p options that we don't want to output.
 308  */
 309 void add_package(string const & name, vector<string> & options)
 310 {
 311         // every package inherits the global options
 312         if (used_packages.find(name) == used_packages.end())
 313                 used_packages[name] = split_options(h_options);
 314
 315         vector<string> & v = used_packages[name];
 316         v.insert(v.end(), options.begin(), options.end());
 317         if (name == "jurabib") {
 318                 // Don't output the order argument (see the cite command
 319                 // handling code in text.cpp).
 320                 vector<string>::iterator end =
 321                         remove(options.begin(), options.end(), "natbiborder");
 322                 end = remove(options.begin(), end, "jurabiborder");
 323                 options.erase(end, options.end());
 324         }
 325 }
 326
 327
 328 // Given is a string like "scaled=0.9", return 0.9 * 100
 329 string const scale_as_percentage(string const & scale)
 330 {
 331         string::size_type pos = scale.find('=');
 332         if (pos != string::npos) {
 333                 string value = scale.substr(pos + 1);
 334                 if (isStrDbl(value))
 335                         return convert<string>(100 * convert<double>(value));
 336         }
 337         // If the input string didn't match our expectations.
 338         // return the default value "100"
 339         return "100";
 340 }
 341
 342
 343 string remove_braces(string const & value)
 344 {
 345         if (value.empty())
 346                 return value;
 347         if (value[0] == '{' && value[value.length()-1] == '}')
 348                 return value.substr(1, value.length()-2);
 349         return value;
 350 }
 351
 352
 353 void handle_hyperref(vector<string> & options)
 354 {
 355         // FIXME swallow inputencoding changes that might surround the
 356         //       hyperref setup if it was written by LyX
 357         h_use_hyperref = "1";
 358         // swallow "unicode=true", since LyX does always write that
 359         vector<string>::iterator it =
 360                 find(options.begin(), options.end(), "unicode=true");
 361         if (it != options.end())
 362                 options.erase(it);
 363         it = find(options.begin(), options.end(), "pdfusetitle");
 364         if (it != options.end()) {
 365                 h_pdf_pdfusetitle = "1";
 366                 options.erase(it);
 367         }
 368         string bookmarks = process_keyval_opt(options, "bookmarks");
 369         if (bookmarks == "true")
 370                 h_pdf_bookmarks = "1";
 371         else if (bookmarks == "false")
 372                 h_pdf_bookmarks = "0";
 373         if (h_pdf_bookmarks == "1") {
 374                 string bookmarksnumbered =
 375                         process_keyval_opt(options, "bookmarksnumbered");
 376                 if (bookmarksnumbered == "true")
 377                         h_pdf_bookmarksnumbered = "1";
 378                 else if (bookmarksnumbered == "false")
 379                         h_pdf_bookmarksnumbered = "0";
 380                 string bookmarksopen =
 381                         process_keyval_opt(options, "bookmarksopen");
 382                 if (bookmarksopen == "true")
 383                         h_pdf_bookmarksopen = "1";
 384                 else if (bookmarksopen == "false")
 385                         h_pdf_bookmarksopen = "0";
 386                 if (h_pdf_bookmarksopen == "1") {
 387                         string bookmarksopenlevel =
 388                                 process_keyval_opt(options, "bookmarksopenlevel");
 389                         if (!bookmarksopenlevel.empty())
 390                                 h_pdf_bookmarksopenlevel = bookmarksopenlevel;
 391                 }
 392         }
 393         string breaklinks = process_keyval_opt(options, "breaklinks");
 394         if (breaklinks == "true")
 395                 h_pdf_breaklinks = "1";
 396         else if (breaklinks == "false")
 397                 h_pdf_breaklinks = "0";
 398         string pdfborder = process_keyval_opt(options, "pdfborder");
 399         if (pdfborder == "{0 0 0}")
 400                 h_pdf_pdfborder = "1";
 401         else if (pdfborder == "{0 0 1}")
 402                 h_pdf_pdfborder = "0";
 403         string backref = process_keyval_opt(options, "backref");
 404         if (!backref.empty())
 405                 h_pdf_backref = backref;
 406         string colorlinks = process_keyval_opt(options, "colorlinks");
 407         if (colorlinks == "true")
 408                 h_pdf_colorlinks = "1";
 409         else if (colorlinks == "false")
 410                 h_pdf_colorlinks = "0";
 411         string pdfpagemode = process_keyval_opt(options, "pdfpagemode");
 412         if (!pdfpagemode.empty())
 413                 h_pdf_pagemode = pdfpagemode;
 414         string pdftitle = process_keyval_opt(options, "pdftitle");
 415         if (!pdftitle.empty()) {
 416                 h_pdf_title = remove_braces(pdftitle);
 417         }
 418         string pdfauthor = process_keyval_opt(options, "pdfauthor");
 419         if (!pdfauthor.empty()) {
 420                 h_pdf_author = remove_braces(pdfauthor);
 421         }
 422         string pdfsubject = process_keyval_opt(options, "pdfsubject");
 423         if (!pdfsubject.empty())
 424                 h_pdf_subject = remove_braces(pdfsubject);
 425         string pdfkeywords = process_keyval_opt(options, "pdfkeywords");
 426         if (!pdfkeywords.empty())
 427                 h_pdf_keywords = remove_braces(pdfkeywords);
 428         if (!options.empty()) {
 429                 if (!h_pdf_quoted_options.empty())
 430                         h_pdf_quoted_options += ',';
 431                 h_pdf_quoted_options += join(options, ",");
 432                 options.clear();
 433         }
 434 }
 435
 436
 437 void handle_package(Parser &p, string const & name, string const & opts,
 438                     bool in_lyx_preamble)
 439 {
 440         vector<string> options = split_options(opts);
 441         add_package(name, options);
 442         string scale;
 443
 444         // roman fonts
 445         if (is_known(name, known_roman_fonts)) {
 446                 h_font_roman = name;
 447                 p.skip_spaces();
 448         }
 449
 450         if (name == "fourier") {
 451                 h_font_roman = "utopia";
 452                 // when font uses real small capitals
 453                 if (opts == "expert")
 454                         h_font_sc = "true";
 455         }
 456
 457         if (name == "mathpazo")
 458                 h_font_roman = "palatino";
 459
 460         if (name == "mathptmx")
 461                 h_font_roman = "times";
 462
 463         // sansserif fonts
 464         if (is_known(name, known_sans_fonts)) {
 465                 h_font_sans = name;
 466                 if (!opts.empty()) {
 467                         scale = opts;
 468                         h_font_sf_scale = scale_as_percentage(scale);
 469                 }
 470         }
 471
 472         // typewriter fonts
 473         if (is_known(name, known_typewriter_fonts)) {
 474                 h_font_typewriter = name;
 475                 if (!opts.empty()) {
 476                         scale = opts;
 477                         h_font_tt_scale = scale_as_percentage(scale);
 478                 }
 479         }
 480
 481         // font uses old-style figure
 482         if (name == "eco")
 483                 h_font_osf = "true";
 484
 485         else if (name == "amsmath" || name == "amssymb")
 486                 h_use_amsmath = "2";
 487
 488         else if (name == "esint")
 489                 h_use_esint = "2";
 490
 491         else if (name == "babel" && !opts.empty()) {
 492                 // check if more than one option was used - used later for inputenc
 493                 // in case inputenc is parsed before babel, set the encoding to auto
 494                 if (options.begin() != options.end() - 1) {
 495                         one_language = false;
 496                         h_inputencoding = "auto";
 497                 }
 498                 // babel takes the last language of the option of its \usepackage
 499                 // call as document language. If there is no such language option, the
 500                 // last language in the documentclass options is used.
 501                 handle_opt(options, known_languages, h_language);
 502                 delete_opt(options, known_languages);
 503         }
 504
 505         else if (name == "fontenc")
 506                  ;// ignore this
 507
 508         else if (name == "inputenc") {
 509                 // h_inputencoding is only set when there is not more than one
 510                 // inputenc option because otherwise h_inputencoding must be
 511                 // set to "auto" (the default encoding of the document language)
 512                 // Therefore check for the "," character.
 513                 // It is also only set when there is not more then one babel
 514                 // language option but this is handled in the routine for babel.
 515                 if (opts.find(",") == string::npos && one_language == true)
 516                         h_inputencoding = opts;
 517                 if (!options.empty())
 518                         p.setEncoding(options.back());
 519                 options.clear();
 520         }
 521
 522         else if (name == "makeidx")
 523                 ; // ignore this
 524
 525         else if (name == "prettyref")
 526                 ; // ignore this
 527
 528         else if (name == "varioref")
 529                 ; // ignore this
 530
 531         else if (name == "verbatim")
 532                 ; // ignore this
 533
 534         else if (name == "nomencl")
 535                 ; // ignore this
 536
 537         else if (name == "textcomp")
 538                 ; // ignore this
 539
 540         else if (name == "url")
 541                 ; // ignore this
 542
 543         else if (LYX_FORMAT >= 408 && name == "subscript")
 544                 ; // ignore this
 545
 546         else if (name == "color") {
 547                 // with the following command this package is only loaded when needed for
 548                 // undefined colors, since we only support the predefined colors
 549                 h_preamble << "\\@ifundefined{definecolor}\n {\\usepackage{color}}{}\n";
 550         }
 551
 552         else if (name == "graphicx")
 553                 ; // ignore this
 554
 555         else if (name == "setspace")
 556                 ; // ignore this
 557
 558         else if (name == "geometry")
 559                 ; // Ignore this, the geometry settings are made by the \geometry
 560                   // command. This command is handled below.
 561
 562         else if (is_known(name, known_languages))
 563                 h_language = name;
 564
 565         else if (name == "natbib") {
 566                 h_cite_engine = "natbib_authoryear";
 567                 vector<string>::iterator it =
 568                         find(options.begin(), options.end(), "authoryear");
 569                 if (it != options.end())
 570                         options.erase(it);
 571                 else {
 572                         it = find(options.begin(), options.end(), "numbers");
 573                         if (it != options.end()) {
 574                                 h_cite_engine = "natbib_numerical";
 575                                 options.erase(it);
 576                         }
 577                 }
 578         }
 579
 580         else if (name == "jurabib")
 581                 h_cite_engine = "jurabib";
 582
 583         else if (name == "hyperref")
 584                 handle_hyperref(options);
 585
 586         else if (!in_lyx_preamble) {
 587                 if (options.empty())
 588                         h_preamble << "\\usepackage{" << name << "}";
 589                 else {
 590                         h_preamble << "\\usepackage[" << opts << "]{"
 591                                    << name << "}";
 592                         options.clear();
 593                 }
 594         }
 595
 596         // We need to do something with the options...
 597         if (!options.empty())
 598                 cerr << "Ignoring options '" << join(options, ",")
 599                      << "' of package " << name << '.' << endl;
 600
 601         // remove the whitespace
 602         p.skip_spaces();
 603 }
 604
 605
 606
 607 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 608 {
 609         // merge synonym languages
 610         if (is_known(h_language, known_bahasa_languages))
 611                 h_language = "bahasa";
 612         else if (is_known(h_language, known_bahasam_languages))
 613                 h_language = "bahasam";
 614         else if (is_known(h_language, known_brazilian_languages))
 615                 h_language = "brazilian";
 616         else if (is_known(h_language, known_french_languages))
 617                 h_language = "french";
 618         else if (is_known(h_language, known_german_languages))
 619                 h_language = "german";
 620         else if (is_known(h_language, known_ngerman_languages))
 621                 h_language = "ngerman";
 622         else if (is_known(h_language, known_portuguese_languages))
 623                 h_language = "portuguese";
 624         else if (is_known(h_language, known_russian_languages))
 625                 h_language = "russian";
 626         else if (is_known(h_language, known_ukrainian_languages))
 627                 h_language = "ukrainian";
 628
 629         // set the quote language
 630         // LyX only knows the following quotes languages:
 631         // english, swedish, german, polish, french and danish
 632         // (quotes for "japanese" and "chinese-traditional" are missing because
 633         //  they wouldn't be useful: http://www.lyx.org/trac/ticket/6383)
 634         // conversion list taken from
 635         // http://en.wikipedia.org/wiki/Quotation_mark,_non-English_usage
 636         // (quotes for kazakh and interlingua are unknown)
 637         // danish
 638         if (h_language == "danish")
 639                 h_quotes_language = "danish";
 640         // french
 641         else if (is_known(h_language, known_french_quotes_languages)
 642                 || is_known(h_language, known_french_languages)
 643                 || is_known(h_language, known_russian_languages)
 644                 || is_known(h_language, known_ukrainian_languages))
 645                 h_quotes_language = "french";
 646         // german
 647         else if (is_known(h_language, known_german_quotes_languages)
 648                 || is_known(h_language, known_german_languages)
 649                 || is_known(h_language, known_ngerman_languages))
 650                 h_quotes_language = "german";
 651         // polish
 652         else if (is_known(h_language, known_polish_quotes_languages))
 653                 h_quotes_language = "polish";
 654         // swedish
 655         else if (is_known(h_language, known_swedish_quotes_languages))
 656                 h_quotes_language = "swedish";
 657         //english
 658         else if (is_known(h_language, known_english_quotes_languages)
 659                 || is_known(h_language, known_bahasa_languages)
 660                 || is_known(h_language, known_bahasam_languages)
 661                 || is_known(h_language, known_brazilian_languages)
 662                 || is_known(h_language, known_portuguese_languages))
 663                 h_quotes_language = "english";
 664
 665         h_language = babel2lyx(h_language);
 666
 667         // output the LyX file settings
 668         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 669            << "\\lyxformat " << LYX_FORMAT << '\n'
 670            << "\\begin_document\n"
 671            << "\\begin_header\n"
 672            << "\\textclass " << h_textclass << "\n";
 673         if (!h_preamble.str().empty())
 674                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 675         if (!h_options.empty())
 676                 os << "\\options " << h_options << "\n";
 677         os << "\\use_default_options " << h_use_default_options << "\n"
 678            << "\\language " << h_language << "\n"
 679            << "\\inputencoding " << h_inputencoding << "\n"
 680            << "\\font_roman " << h_font_roman << "\n"
 681            << "\\font_sans " << h_font_sans << "\n"
 682            << "\\font_typewriter " << h_font_typewriter << "\n"
 683            << "\\font_default_family " << h_font_default_family << "\n"
 684            << "\\font_sc " << h_font_sc << "\n"
 685            << "\\font_osf " << h_font_osf << "\n"
 686            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 687            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 688            << "\\graphics " << h_graphics << "\n";
 689         if (!h_float_placement.empty())
 690                 os << "\\float_placement " << h_float_placement << "\n";
 691         os << "\\paperfontsize " << h_paperfontsize << "\n"
 692            << "\\spacing " << h_spacing << "\n"
 693            << "\\use_hyperref " << h_use_hyperref << '\n';
 694         if (h_use_hyperref == "1") {
 695                 if (!h_pdf_title.empty())
 696                         os << "\\pdf_title \"" << h_pdf_title << "\"\n";
 697                 if (!h_pdf_author.empty())
 698                         os << "\\pdf_author \"" << h_pdf_author << "\"\n";
 699                 if (!h_pdf_subject.empty())
 700                         os << "\\pdf_subject \"" << h_pdf_subject << "\"\n";
 701                 if (!h_pdf_keywords.empty())
 702                         os << "\\pdf_keywords \"" << h_pdf_keywords << "\"\n";
 703                 os << "\\pdf_bookmarks " << h_pdf_bookmarks << "\n"
 704                       "\\pdf_bookmarksnumbered " << h_pdf_bookmarksnumbered << "\n"
 705                       "\\pdf_bookmarksopen " << h_pdf_bookmarksopen << "\n"
 706                       "\\pdf_bookmarksopenlevel " << h_pdf_bookmarksopenlevel << "\n"
 707                       "\\pdf_breaklinks " << h_pdf_breaklinks << "\n"
 708                       "\\pdf_pdfborder " << h_pdf_pdfborder << "\n"
 709                       "\\pdf_colorlinks " << h_pdf_colorlinks << "\n"
 710                       "\\pdf_backref " << h_pdf_backref << "\n"
 711                       "\\pdf_pdfusetitle " << h_pdf_pdfusetitle << '\n';
 712                 if (!h_pdf_pagemode.empty())
 713                         os << "\\pdf_pagemode " << h_pdf_pagemode << '\n';
 714                 if (!h_pdf_quoted_options.empty())
 715                         os << "\\pdf_quoted_options \"" << h_pdf_quoted_options << "\"\n";
 716         }
 717         os << "\\papersize " << h_papersize << "\n"
 718            << "\\use_geometry " << h_use_geometry << "\n"
 719            << "\\use_amsmath " << h_use_amsmath << "\n"
 720            << "\\use_esint " << h_use_esint << "\n"
 721            << "\\cite_engine " << h_cite_engine << "\n"
 722            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 723            << "\\paperorientation " << h_paperorientation << "\n"
 724            << h_margins
 725            << "\\secnumdepth " << h_secnumdepth << "\n"
 726            << "\\tocdepth " << h_tocdepth << "\n"
 727            << "\\paragraph_separation " << h_paragraph_separation << "\n";
 728         if (LYX_FORMAT < 365 || h_paragraph_separation == "skip")
 729                 os << "\\defskip " << h_defskip << "\n";
 730         else
 731                 os << "\\paragraph_indentation " << h_paragraph_indentation << "\n";
 732         os << "\\quotes_language " << h_quotes_language << "\n"
 733            << "\\papercolumns " << h_papercolumns << "\n"
 734            << "\\papersides " << h_papersides << "\n"
 735            << "\\paperpagestyle " << h_paperpagestyle << "\n";
 736         if (!h_listings_params.empty())
 737                 os << "\\listings_params " << h_listings_params << "\n";
 738         os << "\\tracking_changes " << h_tracking_changes << "\n"
 739            << "\\output_changes " << h_output_changes << "\n"
 740            << "\\end_header\n\n"
 741            << "\\begin_body\n";
 742         // clear preamble for subdocuments
 743         h_preamble.str("");
 744 }
 745
 746 } // anonymous namespace
 747
 748 void parse_preamble(Parser & p, ostream & os,
 749         string const & forceclass, TeX2LyXDocClass & tc)
 750 {
 751         // initialize fixed types
 752         special_columns['D'] = 3;
 753         bool is_full_document = false;
 754         bool is_lyx_file = false;
 755         bool in_lyx_preamble = false;
 756
 757         // determine whether this is a full document or a fragment for inclusion
 758         while (p.good()) {
 759                 Token const & t = p.get_token();
 760
 761                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 762                         is_full_document = true;
 763                         break;
 764                 }
 765         }
 766         p.reset();
 767
 768         while (is_full_document && p.good()) {
 769                 Token const & t = p.get_token();
 770
 771 #ifdef FILEDEBUG
 772                 cerr << "t: " << t << "\n";
 773 #endif
 774
 775                 //
 776                 // cat codes
 777                 //
 778                 if (!in_lyx_preamble &&
 779                     (t.cat() == catLetter ||
 780                      t.cat() == catSuper ||
 781                      t.cat() == catSub ||
 782                      t.cat() == catOther ||
 783                      t.cat() == catMath ||
 784                      t.cat() == catActive ||
 785                      t.cat() == catBegin ||
 786                      t.cat() == catEnd ||
 787                      t.cat() == catAlign ||
 788                      t.cat() == catParameter))
 789                         h_preamble << t.cs();
 790
 791                 else if (!in_lyx_preamble &&
 792                          (t.cat() == catSpace || t.cat() == catNewline))
 793                         h_preamble << t.asInput();
 794
 795                 else if (t.cat() == catComment) {
 796                         // regex to parse comments (currently not used)
 797                         static regex const islyxfile("%% LyX .* created this file");
 798                         static regex const usercommands("User specified LaTeX commands");
 799
 800                         string const comment = t.asInput();
 801
 802                         // magically switch encoding default if it looks like XeLaTeX
 803                         static string const magicXeLaTeX =
 804                                 "% This document must be compiled with XeLaTeX ";
 805                         if (comment.size() > magicXeLaTeX.size()
 806                                   && comment.substr(0, magicXeLaTeX.size()) == magicXeLaTeX
 807                                   && h_inputencoding == "auto") {
 808                                 cerr << "XeLaTeX comment found, switching to UTF8\n";
 809                                 h_inputencoding = "utf8";
 810                         }
 811                         smatch sub;
 812                         if (regex_search(comment, sub, islyxfile)) {
 813                                 is_lyx_file = true;
 814                                 in_lyx_preamble = true;
 815                         } else if (is_lyx_file
 816                                    && regex_search(comment, sub, usercommands))
 817                                 in_lyx_preamble = false;
 818                         else if (!in_lyx_preamble)
 819                                 h_preamble << t.asInput();
 820                 }
 821
 822                 else if (t.cs() == "pagestyle")
 823                         h_paperpagestyle = p.verbatim_item();
 824
 825                 else if (t.cs() == "makeatletter") {
 826                         // LyX takes care of this
 827                         p.setCatCode('@', catLetter);
 828                 }
 829
 830                 else if (t.cs() == "makeatother") {
 831                         // LyX takes care of this
 832                         p.setCatCode('@', catOther);
 833                 }
 834
 835                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 836                             || t.cs() == "providecommand"
 837                                 || t.cs() == "DeclareRobustCommand"
 838                                 || t.cs() == "ProvideTextCommandDefault"
 839                                 || t.cs() == "DeclareMathAccent") {
 840                         bool star = false;
 841                         if (p.next_token().character() == '*') {
 842                                 p.get_token();
 843                                 star = true;
 844                         }
 845                         string const name = p.verbatim_item();
 846                         string const opt1 = p.getOpt();
 847                         string const opt2 = p.getFullOpt();
 848                         string const body = p.verbatim_item();
 849                         // font settings
 850                         if (name == "\\rmdefault")
 851                                 if (is_known(body, known_roman_fonts))
 852                                         h_font_roman = body;
 853                         if (name == "\\sfdefault")
 854                                 if (is_known(body, known_sans_fonts))
 855                                         h_font_sans = body;
 856                         if (name == "\\ttdefault")
 857                                 if (is_known(body, known_typewriter_fonts))
 858                                         h_font_typewriter = body;
 859                         if (name == "\\familydefault") {
 860                                 string family = body;
 861                                 // remove leading "\"
 862                                 h_font_default_family = family.erase(0,1);
 863                         }
 864                         // only non-lyxspecific stuff
 865                         if (!in_lyx_preamble) {
 866                                 ostringstream ss;
 867                                 ss << '\\' << t.cs();
 868                                 if (star)
 869                                         ss << '*';
 870                                 ss << '{' << name << '}' << opt1 << opt2
 871                                    << '{' << body << "}";
 872                                 h_preamble << ss.str();
 873
 874                                 // Add the command to the known commands
 875                                 add_known_command(name, opt1, !opt2.empty());
 876 /*
 877                                 ostream & out = in_preamble ? h_preamble : os;
 878                                 out << "\\" << t.cs() << "{" << name << "}"
 879                                     << opts << "{" << body << "}";
 880 */
 881                         }
 882                 }
 883
 884                 else if (t.cs() == "documentclass") {
 885                         vector<string>::iterator it;
 886                         vector<string> opts = split_options(p.getArg('[', ']'));
 887                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 888                         delete_opt(opts, known_fontsizes);
 889                         // delete "pt" at the end
 890                         string::size_type i = h_paperfontsize.find("pt");
 891                         if (i != string::npos)
 892                                 h_paperfontsize.erase(i);
 893                         // The documentclass options are always parsed before the options
 894                         // of the babel call so that a language cannot overwrite the babel
 895                         // options.
 896                         handle_opt(opts, known_languages, h_language);
 897                         delete_opt(opts, known_languages);
 898
 899                         // paper orientation
 900                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 901                                 h_paperorientation = "landscape";
 902                                 opts.erase(it);
 903                         }
 904                         // paper sides
 905                         if ((it = find(opts.begin(), opts.end(), "oneside"))
 906                                  != opts.end()) {
 907                                 h_papersides = "1";
 908                                 opts.erase(it);
 909                         }
 910                         if ((it = find(opts.begin(), opts.end(), "twoside"))
 911                                  != opts.end()) {
 912                                 h_papersides = "2";
 913                                 opts.erase(it);
 914                         }
 915                         // paper columns
 916                         if ((it = find(opts.begin(), opts.end(), "onecolumn"))
 917                                  != opts.end()) {
 918                                 h_papercolumns = "1";
 919                                 opts.erase(it);
 920                         }
 921                         if ((it = find(opts.begin(), opts.end(), "twocolumn"))
 922                                  != opts.end()) {
 923                                 h_papercolumns = "2";
 924                                 opts.erase(it);
 925                         }
 926                         // paper sizes
 927                         // some size options are know to any document classes, other sizes
 928                         // are handled by the \geometry command of the geometry package
 929                         handle_opt(opts, known_class_paper_sizes, h_papersize);
 930                         delete_opt(opts, known_class_paper_sizes);
 931                         // the remaining options
 932                         h_options = join(opts, ",");
 933                         h_textclass = p.getArg('{', '}');
 934                 }
 935
 936                 else if (t.cs() == "usepackage") {
 937                         string const options = p.getArg('[', ']');
 938                         string const name = p.getArg('{', '}');
 939                         vector<string> vecnames;
 940                         split(name, vecnames, ',');
 941                         vector<string>::const_iterator it  = vecnames.begin();
 942                         vector<string>::const_iterator end = vecnames.end();
 943                         for (; it != end; ++it)
 944                                 handle_package(p, trim(*it), options,
 945                                                in_lyx_preamble);
 946                 }
 947
 948                 else if (t.cs() == "inputencoding") {
 949                         string const encoding = p.getArg('{','}');
 950                         h_inputencoding = encoding;
 951                         p.setEncoding(encoding);
 952                 }
 953
 954                 else if (t.cs() == "newenvironment") {
 955                         string const name = p.getArg('{', '}');
 956                         ostringstream ss;
 957                         ss << "\\newenvironment{" << name << "}";
 958                         ss << p.getOpt();
 959                         ss << p.getOpt();
 960                         ss << '{' << p.verbatim_item() << '}';
 961                         ss << '{' << p.verbatim_item() << '}';
 962                         if (!in_lyx_preamble)
 963                                 h_preamble << ss.str();
 964                 }
 965
 966                 else if (t.cs() == "def") {
 967                         string name = p.get_token().cs();
 968                         while (p.next_token().cat() != catBegin)
 969                                 name += p.get_token().cs();
 970                         if (!in_lyx_preamble)
 971                                 h_preamble << "\\def\\" << name << '{'
 972                                            << p.verbatim_item() << "}";
 973                 }
 974
 975                 else if (t.cs() == "newcolumntype") {
 976                         string const name = p.getArg('{', '}');
 977                         trim(name);
 978                         int nargs = 0;
 979                         string opts = p.getOpt();
 980                         if (!opts.empty()) {
 981                                 istringstream is(string(opts, 1));
 982                                 is >> nargs;
 983                         }
 984                         special_columns[name[0]] = nargs;
 985                         h_preamble << "\\newcolumntype{" << name << "}";
 986                         if (nargs)
 987                                 h_preamble << "[" << nargs << "]";
 988                         h_preamble << "{" << p.verbatim_item() << "}";
 989                 }
 990
 991                 else if (t.cs() == "setcounter") {
 992                         string const name = p.getArg('{', '}');
 993                         string const content = p.getArg('{', '}');
 994                         if (name == "secnumdepth")
 995                                 h_secnumdepth = content;
 996                         else if (name == "tocdepth")
 997                                 h_tocdepth = content;
 998                         else
 999                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
1000                 }
1001
1002                 else if (t.cs() == "setlength") {
1003                         string const name = p.verbatim_item();
1004                         string const content = p.verbatim_item();
1005                         // the paragraphs are only not indented when \parindent is set to zero
1006                         if (name == "\\parindent" && content != "") {
1007                                 if (content[0] == '0')
1008                                         h_paragraph_separation = "skip";
1009                                 else if (LYX_FORMAT >= 365)
1010                                         h_paragraph_indentation = translate_len(content);
1011                                 else
1012                                         h_preamble << "\\setlength{" << name
1013                                                    << "}{" << content << "}";
1014                         } else if (name == "\\parskip") {
1015                                 if (content == "\\smallskipamount")
1016                                         h_defskip = "smallskip";
1017                                 else if (content == "\\medskipamount")
1018                                         h_defskip = "medskip";
1019                                 else if (content == "\\bigskipamount")
1020                                         h_defskip = "bigskip";
1021                                 else
1022                                         h_defskip = content;
1023                         } else
1024                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
1025                 }
1026
1027                 else if (t.cs() == "onehalfspacing")
1028                         h_spacing = "onehalf";
1029
1030                 else if (t.cs() == "doublespacing")
1031                         h_spacing = "double";
1032
1033                 else if (t.cs() == "setstretch")
1034                         h_spacing = "other " + p.verbatim_item();
1035
1036                 else if (t.cs() == "begin") {
1037                         string const name = p.getArg('{', '}');
1038                         if (name == "document")
1039                                 break;
1040                         h_preamble << "\\begin{" << name << "}";
1041                 }
1042
1043                 else if (t.cs() == "geometry") {
1044                         h_use_geometry = "true";
1045                         vector<string> opts = split_options(p.getArg('{', '}'));
1046                         vector<string>::iterator it;
1047                         // paper orientation
1048                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
1049                                 h_paperorientation = "landscape";
1050                                 opts.erase(it);
1051                         }
1052                         // paper size
1053                         handle_opt(opts, known_paper_sizes, h_papersize);
1054                         delete_opt(opts, known_paper_sizes);
1055                         // page margins
1056                         char const * const * margin = known_paper_margins;
1057                         int k = -1;
1058                         for (; *margin; ++margin) {
1059                                 k += 1;
1060                                 // search for the "=" in e.g. "lmargin=2cm" to get the value
1061                                 for(size_t i = 0; i != opts.size(); i++) {
1062                                         if (opts.at(i).find(*margin) != string::npos) {
1063                                                 string::size_type pos = opts.at(i).find("=");
1064                                                 string value = opts.at(i).substr(pos + 1);
1065                                                 string name = known_coded_paper_margins[k];
1066                                                 h_margins += "\\" + name + " " + value + "\n";
1067                                         }
1068                                 }
1069                         }
1070                 }
1071
1072                 else if (t.cs() == "jurabibsetup") {
1073                         // FIXME p.getArg('{', '}') is most probably wrong (it
1074                         //       does not handle nested braces).
1075                         //       Use p.verbatim_item() instead.
1076                         vector<string> jurabibsetup =
1077                                 split_options(p.getArg('{', '}'));
1078                         // add jurabibsetup to the jurabib package options
1079                         add_package("jurabib", jurabibsetup);
1080                         if (!jurabibsetup.empty()) {
1081                                 h_preamble << "\\jurabibsetup{"
1082                                            << join(jurabibsetup, ",") << '}';
1083                         }
1084                 }
1085
1086                 else if (t.cs() == "hypersetup") {
1087                         vector<string> hypersetup =
1088                                 split_options(p.verbatim_item());
1089                         // add hypersetup to the hyperref package options
1090                         handle_hyperref(hypersetup);
1091                         if (!hypersetup.empty()) {
1092                                 h_preamble << "\\hypersetup{"
1093                                            << join(hypersetup, ",") << '}';
1094                         }
1095                 }
1096
1097                 else if (!t.cs().empty() && !in_lyx_preamble)
1098                         h_preamble << '\\' << t.cs();
1099         }
1100
1101         // remove the whitespace
1102         p.skip_spaces();
1103
1104         // Force textclass if the user wanted it
1105         if (!forceclass.empty())
1106                 h_textclass = forceclass;
1107         if (noweb_mode && !prefixIs(h_textclass, "literate-"))
1108                 h_textclass.insert(0, "literate-");
1109         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
1110         if (layoutfilename.empty()) {
1111                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
1112                 exit(1);
1113         }
1114         tc.read(layoutfilename);
1115         if (h_papersides.empty()) {
1116                 ostringstream ss;
1117                 ss << tc.sides();
1118                 h_papersides = ss.str();
1119         }
1120         end_preamble(os, tc);
1121 }
1122
1123 // }])
1124
1125
1126 } // namespace lyx