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 /**
  61  * known babel language names (including synonyms)
  62  * not in standard babel: arabic, arabtex, belarusian, serbian-latin, thai
  63  * not yet supported by LyX: kurmanji
  64  * please keep this in sync with known_coded_languages line by line!
  65  */
  66 const char * const known_languages[] = {"acadian", "afrikaans", "albanian",
  67 "american", "arabic", "arabtex", "austrian", "bahasa", "bahasai", "bahasam",
  68 "basque", "belarusian", "brazil", "brazilian", "breton", "british", "bulgarian",
  69 "canadian", "canadien", "catalan", "croatian", "czech", "danish", "dutch",
  70 "english", "esperanto", "estonian", "farsi", "finnish", "francais", "french",
  71 "frenchb", "frenchle", "frenchpro", "galician", "german", "germanb", "greek",
  72 "hebrew", "hungarian", "icelandic", "indon", "indonesian", "interlingua",
  73 "irish", "italian", "kazakh", "latin", "latvian", "lithuanian", "lowersorbian",
  74 "lsorbian", "magyar", "malay", "meyalu", "mongolian", "naustrian", "newzealand",
  75 "ngerman", "ngermanb", "norsk", "nynorsk", "polutonikogreek", "polish",
  76 "portuges", "portuguese", "romanian", "russian", "russianb", "samin",
  77 "scottish", "serbian", "serbian-latin", "slovak", "slovene", "spanish",
  78 "swedish", "thai", "turkish", "ukraineb", "ukrainian", "uppersorbian",
  79 "UKenglish", "USenglish", "usorbian", "vietnam", "welsh", 0};
  80
  81 /**
  82  * the same as known_languages with .lyx names
  83  * please keep this in sync with known_languages line by line!
  84  */
  85 const char * const known_coded_languages[] = {"french", "afrikaans", "albanian",
  86 "american", "arabic_arabi", "arabic_arabtex", "austrian", "bahasa", "bahasa", "bahasam",
  87 "basque", "belarusian", "brazilian", "brazilian", "breton", "british", "bulgarian",
  88 "canadian", "canadien", "catalan", "croatian", "czech", "danish", "dutch",
  89 "english", "esperanto", "estonian", "farsi", "finnish", "french", "french",
  90 "french", "french", "french", "galician", "german", "german", "greek",
  91 "hebrew", "magyar", "icelandic", "bahasa", "bahasa", "interlingua",
  92 "irish", "italian", "kazakh", "latin", "latvian", "lithuanian", "lowersorbian",
  93 "lowersorbian", "magyar", "bahasam", "bahasam", "mongolian", "naustrian", "english",
  94 "ngerman", "ngerman", "norsk", "nynorsk", "polutonikogreek", "polish",
  95 "portuguese", "portuguese", "romanian", "russian", "russian", "samin",
  96 "scottish", "serbian", "serbian-latin", "slovak", "slovene", "spanish",
  97 "swedish", "thai", "turkish", "ukrainian", "ukrainian", "uppersorbian",
  98 "uppersorbian", "english", "english", "vietnamese", "welsh", 0};
  99
 100 /// languages with english quotes (.lyx names)
 101 const char * const known_english_quotes_languages[] = {"american", "bahasa",
 102 "bahasam", "brazilian", "canadian", "chinese-simplified", "english",
 103 "esperanto", "hebrew", "irish", "korean", "portuguese", "scottish", "thai", 0};
 104
 105 //add this to known_french_quotes_languages when updating to
 106 //lyxformat 383: "turkmen"
 107 /// languages with french quotes (.lyx names)
 108 const char * const known_french_quotes_languages[] = {"albanian",
 109 "arabic_arabi", "arabic_arabtex", "basque", "canadien", "catalan", "french",
 110 "galician", "greek", "italian", "norsk", "nynorsk", "polutonikogreek",
 111 "russian", "spanish", "spanish-mexico", "turkish", "ukrainian", "vietnamese", 0};
 112
 113 /// languages with german quotes (.lyx names)
 114 const char * const known_german_quotes_languages[] = {"austrian", "bulgarian",
 115 "czech", "german", "icelandic", "lithuanian", "lowersorbian", "naustrian",
 116 "ngerman", "serbian", "serbian-latin", "slovak", "slovene", "uppersorbian", 0};
 117
 118 /// languages with polish quotes (.lyx names)
 119 const char * const known_polish_quotes_languages[] = {"afrikaans", "croatian",
 120 "dutch", "estonian", "magyar", "polish", "romanian", 0};
 121
 122 /// languages with swedish quotes (.lyx names)
 123 const char * const known_swedish_quotes_languages[] = {"finnish",
 124 "swedish", 0};
 125
 126 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
 127
 128 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
 129 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
 130
 131 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
 132 "helvet", "lmss", 0};
 133
 134 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
 135 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
 136 "newcent", 0};
 137
 138 const char * const known_paper_sizes[] = { "a3paper", "b3paper", "a4paper",
 139 "b4paper", "a5paper", "b5paper", "executivepaper", "legalpaper",
 140 "letterpaper", 0};
 141
 142 const char * const known_class_paper_sizes[] = { "a4paper", "a5paper",
 143 "executivepaper", "legalpaper", "letterpaper", 0};
 144
 145 const char * const known_paper_margins[] = { "lmargin", "tmargin", "rmargin",
 146 "bmargin", "headheight", "headsep", "footskip", "columnsep", 0};
 147
 148 const char * const known_coded_paper_margins[] = { "leftmargin", "topmargin",
 149 "rightmargin", "bottommargin", "headheight", "headsep", "footskip",
 150 "columnsep", 0};
 151
 152 /// commands that can start an \if...\else...\endif sequence
 153 const char * const known_if_commands[] = {"if", "ifarydshln", "ifbraket",
 154 "ifcancel", "ifcolortbl", "ifeurosym", "ifmarginnote", "ifmmode", "ifpdf",
 155 "ifsidecap", "ifupgreek", 0};
 156
 157 /// conditional commands with three arguments like \@ifundefined{}{}{}
 158 const char * const known_if_3arg_commands[] = {"@ifundefined", "IfFileExists",
 159 0};
 160
 161 // default settings
 162 ostringstream h_preamble;
 163 string h_textclass               = "article";
 164 string h_use_default_options     = "false";
 165 string h_options;
 166 string h_language                = "english";
 167 string h_inputencoding           = "auto";
 168 string h_font_roman              = "default";
 169 string h_font_sans               = "default";
 170 string h_font_typewriter         = "default";
 171 string h_font_default_family     = "default";
 172 string h_font_sc                 = "false";
 173 string h_font_osf                = "false";
 174 string h_font_sf_scale           = "100";
 175 string h_font_tt_scale           = "100";
 176 string h_graphics                = "default";
 177 string h_float_placement;
 178 string h_paperfontsize           = "default";
 179 string h_spacing                 = "single";
 180 string h_use_hyperref            = "0";
 181 string h_pdf_title;
 182 string h_pdf_author;
 183 string h_pdf_subject;
 184 string h_pdf_keywords;
 185 string h_pdf_bookmarks           = "1";
 186 string h_pdf_bookmarksnumbered   = "0";
 187 string h_pdf_bookmarksopen       = "0";
 188 string h_pdf_bookmarksopenlevel  = "1";
 189 string h_pdf_breaklinks          = "0";
 190 string h_pdf_pdfborder           = "0";
 191 string h_pdf_colorlinks          = "0";
 192 string h_pdf_backref             = "section";
 193 string h_pdf_pdfusetitle         = "1";
 194 string h_pdf_pagemode;
 195 string h_pdf_quoted_options;
 196 string h_papersize               = "default";
 197 string h_use_geometry            = "false";
 198 string h_use_amsmath             = "1";
 199 string h_use_esint               = "1";
 200 string h_cite_engine             = "basic";
 201 string h_use_bibtopic            = "false";
 202 string h_paperorientation        = "portrait";
 203 string h_secnumdepth             = "3";
 204 string h_tocdepth                = "3";
 205 string h_paragraph_separation    = "indent";
 206 string h_defskip                 = "medskip";
 207 string h_paragraph_indentation   = "default";
 208 string h_quotes_language         = "english";
 209 string h_papercolumns            = "1";
 210 string h_papersides;
 211 string h_paperpagestyle          = "default";
 212 string h_listings_params;
 213 string h_tracking_changes        = "false";
 214 string h_output_changes          = "false";
 215 string h_margins;
 216
 217
 218 // returns true if at least one of the options in what has been found
 219 bool handle_opt(vector<string> & opts, char const * const * what, string & target)
 220 {
 221         if (opts.empty())
 222                 return false;
 223
 224         bool found = false;
 225         // the last language option is the document language (for babel and LyX)
 226         // the last size option is the document font size
 227         vector<string>::iterator it;
 228         vector<string>::iterator position = opts.begin();
 229         for (; *what; ++what) {
 230                 it = find(opts.begin(), opts.end(), *what);
 231                 if (it != opts.end()) {
 232                         if (it >= position) {
 233                                 found = true;
 234                                 target = *what;
 235                                 position = it;
 236                         }
 237                 }
 238         }
 239         return found;
 240 }
 241
 242
 243 void delete_opt(vector<string> & opts, char const * const * what)
 244 {
 245         if (opts.empty())
 246                 return;
 247
 248         // remove found options from the list
 249         // do this after handle_opt to avoid potential memory leaks
 250         vector<string>::iterator it;
 251         for (; *what; ++what) {
 252                 it = find(opts.begin(), opts.end(), *what);
 253                 if (it != opts.end())
 254                         opts.erase(it);
 255         }
 256 }
 257
 258
 259 /*!
 260  * Split a package options string (keyval format) into a vector.
 261  * Example input:
 262  *   authorformat=smallcaps,
 263  *   commabeforerest,
 264  *   titleformat=colonsep,
 265  *   bibformat={tabular,ibidem,numbered}
 266  */
 267 vector<string> split_options(string const & input)
 268 {
 269         vector<string> options;
 270         string option;
 271         Parser p(input);
 272         while (p.good()) {
 273                 Token const & t = p.get_token();
 274                 if (t.asInput() == ",") {
 275                         options.push_back(trim(option));
 276                         option.erase();
 277                 } else if (t.asInput() == "=") {
 278                         option += '=';
 279                         p.skip_spaces(true);
 280                         if (p.next_token().asInput() == "{")
 281                                 option += '{' + p.getArg('{', '}') + '}';
 282                 } else if (t.cat() != catSpace)
 283                         option += t.asInput();
 284         }
 285
 286         if (!option.empty())
 287                 options.push_back(trim(option));
 288
 289         return options;
 290 }
 291
 292
 293 /*!
 294  * Retrieve a keyval option "name={value with=sign}" named \p name from
 295  * \p options and return the value.
 296  * The found option is also removed from \p options.
 297  */
 298 string process_keyval_opt(vector<string> & options, string name)
 299 {
 300         for (size_t i = 0; i < options.size(); ++i) {
 301                 vector<string> option;
 302                 split(options[i], option, '=');
 303                 if (option.size() < 2)
 304                         continue;
 305                 if (option[0] == name) {
 306                         options.erase(options.begin() + i);
 307                         option.erase(option.begin());
 308                         return join(option, "=");
 309                 }
 310         }
 311         return "";
 312 }
 313
 314
 315 /*!
 316  * Add package \p name with options \p options to used_packages.
 317  * Remove options from \p options that we don't want to output.
 318  */
 319 void add_package(string const & name, vector<string> & options)
 320 {
 321         // every package inherits the global options
 322         if (used_packages.find(name) == used_packages.end())
 323                 used_packages[name] = split_options(h_options);
 324
 325         vector<string> & v = used_packages[name];
 326         v.insert(v.end(), options.begin(), options.end());
 327         if (name == "jurabib") {
 328                 // Don't output the order argument (see the cite command
 329                 // handling code in text.cpp).
 330                 vector<string>::iterator end =
 331                         remove(options.begin(), options.end(), "natbiborder");
 332                 end = remove(options.begin(), end, "jurabiborder");
 333                 options.erase(end, options.end());
 334         }
 335 }
 336
 337
 338 // Given is a string like "scaled=0.9", return 0.9 * 100
 339 string const scale_as_percentage(string const & scale)
 340 {
 341         string::size_type pos = scale.find('=');
 342         if (pos != string::npos) {
 343                 string value = scale.substr(pos + 1);
 344                 if (isStrDbl(value))
 345                         return convert<string>(100 * convert<double>(value));
 346         }
 347         // If the input string didn't match our expectations.
 348         // return the default value "100"
 349         return "100";
 350 }
 351
 352
 353 string remove_braces(string const & value)
 354 {
 355         if (value.empty())
 356                 return value;
 357         if (value[0] == '{' && value[value.length()-1] == '}')
 358                 return value.substr(1, value.length()-2);
 359         return value;
 360 }
 361
 362
 363 void handle_hyperref(vector<string> & options)
 364 {
 365         // FIXME swallow inputencoding changes that might surround the
 366         //       hyperref setup if it was written by LyX
 367         h_use_hyperref = "1";
 368         // swallow "unicode=true", since LyX does always write that
 369         vector<string>::iterator it =
 370                 find(options.begin(), options.end(), "unicode=true");
 371         if (it != options.end())
 372                 options.erase(it);
 373         it = find(options.begin(), options.end(), "pdfusetitle");
 374         if (it != options.end()) {
 375                 h_pdf_pdfusetitle = "1";
 376                 options.erase(it);
 377         }
 378         string bookmarks = process_keyval_opt(options, "bookmarks");
 379         if (bookmarks == "true")
 380                 h_pdf_bookmarks = "1";
 381         else if (bookmarks == "false")
 382                 h_pdf_bookmarks = "0";
 383         if (h_pdf_bookmarks == "1") {
 384                 string bookmarksnumbered =
 385                         process_keyval_opt(options, "bookmarksnumbered");
 386                 if (bookmarksnumbered == "true")
 387                         h_pdf_bookmarksnumbered = "1";
 388                 else if (bookmarksnumbered == "false")
 389                         h_pdf_bookmarksnumbered = "0";
 390                 string bookmarksopen =
 391                         process_keyval_opt(options, "bookmarksopen");
 392                 if (bookmarksopen == "true")
 393                         h_pdf_bookmarksopen = "1";
 394                 else if (bookmarksopen == "false")
 395                         h_pdf_bookmarksopen = "0";
 396                 if (h_pdf_bookmarksopen == "1") {
 397                         string bookmarksopenlevel =
 398                                 process_keyval_opt(options, "bookmarksopenlevel");
 399                         if (!bookmarksopenlevel.empty())
 400                                 h_pdf_bookmarksopenlevel = bookmarksopenlevel;
 401                 }
 402         }
 403         string breaklinks = process_keyval_opt(options, "breaklinks");
 404         if (breaklinks == "true")
 405                 h_pdf_breaklinks = "1";
 406         else if (breaklinks == "false")
 407                 h_pdf_breaklinks = "0";
 408         string pdfborder = process_keyval_opt(options, "pdfborder");
 409         if (pdfborder == "{0 0 0}")
 410                 h_pdf_pdfborder = "1";
 411         else if (pdfborder == "{0 0 1}")
 412                 h_pdf_pdfborder = "0";
 413         string backref = process_keyval_opt(options, "backref");
 414         if (!backref.empty())
 415                 h_pdf_backref = backref;
 416         string colorlinks = process_keyval_opt(options, "colorlinks");
 417         if (colorlinks == "true")
 418                 h_pdf_colorlinks = "1";
 419         else if (colorlinks == "false")
 420                 h_pdf_colorlinks = "0";
 421         string pdfpagemode = process_keyval_opt(options, "pdfpagemode");
 422         if (!pdfpagemode.empty())
 423                 h_pdf_pagemode = pdfpagemode;
 424         string pdftitle = process_keyval_opt(options, "pdftitle");
 425         if (!pdftitle.empty()) {
 426                 h_pdf_title = remove_braces(pdftitle);
 427         }
 428         string pdfauthor = process_keyval_opt(options, "pdfauthor");
 429         if (!pdfauthor.empty()) {
 430                 h_pdf_author = remove_braces(pdfauthor);
 431         }
 432         string pdfsubject = process_keyval_opt(options, "pdfsubject");
 433         if (!pdfsubject.empty())
 434                 h_pdf_subject = remove_braces(pdfsubject);
 435         string pdfkeywords = process_keyval_opt(options, "pdfkeywords");
 436         if (!pdfkeywords.empty())
 437                 h_pdf_keywords = remove_braces(pdfkeywords);
 438         if (!options.empty()) {
 439                 if (!h_pdf_quoted_options.empty())
 440                         h_pdf_quoted_options += ',';
 441                 h_pdf_quoted_options += join(options, ",");
 442                 options.clear();
 443         }
 444 }
 445
 446
 447 void handle_package(Parser &p, string const & name, string const & opts,
 448                     bool in_lyx_preamble)
 449 {
 450         vector<string> options = split_options(opts);
 451         add_package(name, options);
 452         string scale;
 453
 454         // roman fonts
 455         if (is_known(name, known_roman_fonts)) {
 456                 h_font_roman = name;
 457                 p.skip_spaces();
 458         }
 459
 460         if (name == "fourier") {
 461                 h_font_roman = "utopia";
 462                 // when font uses real small capitals
 463                 if (opts == "expert")
 464                         h_font_sc = "true";
 465         }
 466
 467         if (name == "mathpazo")
 468                 h_font_roman = "palatino";
 469
 470         if (name == "mathptmx")
 471                 h_font_roman = "times";
 472
 473         // sansserif fonts
 474         if (is_known(name, known_sans_fonts)) {
 475                 h_font_sans = name;
 476                 if (!opts.empty()) {
 477                         scale = opts;
 478                         h_font_sf_scale = scale_as_percentage(scale);
 479                 }
 480         }
 481
 482         // typewriter fonts
 483         if (is_known(name, known_typewriter_fonts)) {
 484                 h_font_typewriter = name;
 485                 if (!opts.empty()) {
 486                         scale = opts;
 487                         h_font_tt_scale = scale_as_percentage(scale);
 488                 }
 489         }
 490
 491         // font uses old-style figure
 492         if (name == "eco")
 493                 h_font_osf = "true";
 494
 495         else if (name == "amsmath" || name == "amssymb")
 496                 h_use_amsmath = "2";
 497
 498         else if (name == "esint")
 499                 h_use_esint = "2";
 500
 501         else if (name == "babel" && !opts.empty()) {
 502                 // check if more than one option was used - used later for inputenc
 503                 // in case inputenc is parsed before babel, set the encoding to auto
 504                 if (options.begin() != options.end() - 1) {
 505                         one_language = false;
 506                         h_inputencoding = "auto";
 507                 }
 508                 // babel takes the last language of the option of its \usepackage
 509                 // call as document language. If there is no such language option, the
 510                 // last language in the documentclass options is used.
 511                 handle_opt(options, known_languages, h_language);
 512                 delete_opt(options, known_languages);
 513         }
 514
 515         else if (name == "fontenc")
 516                  ;// ignore this
 517
 518         else if (name == "inputenc") {
 519                 // h_inputencoding is only set when there is not more than one
 520                 // inputenc option because otherwise h_inputencoding must be
 521                 // set to "auto" (the default encoding of the document language)
 522                 // Therefore check for the "," character.
 523                 // It is also only set when there is not more then one babel
 524                 // language option but this is handled in the routine for babel.
 525                 if (opts.find(",") == string::npos && one_language == true)
 526                         h_inputencoding = opts;
 527                 if (!options.empty())
 528                         p.setEncoding(options.back());
 529                 options.clear();
 530         }
 531
 532         else if (name == "makeidx")
 533                 ; // ignore this
 534
 535         else if (name == "prettyref")
 536                 ; // ignore this
 537
 538         else if (name == "varioref")
 539                 ; // ignore this
 540
 541         else if (name == "verbatim")
 542                 ; // ignore this
 543
 544         else if (name == "nomencl")
 545                 ; // ignore this
 546
 547         else if (name == "textcomp")
 548                 ; // ignore this
 549
 550         else if (name == "url")
 551                 ; // ignore this
 552
 553         else if (LYX_FORMAT >= 408 && name == "subscript")
 554                 ; // ignore this
 555
 556         else if (name == "color") {
 557                 // with the following command this package is only loaded when needed for
 558                 // undefined colors, since we only support the predefined colors
 559                 h_preamble << "\\@ifundefined{definecolor}\n {\\usepackage{color}}{}\n";
 560         }
 561
 562         else if (name == "graphicx")
 563                 ; // ignore this
 564
 565         else if (name == "setspace")
 566                 ; // ignore this
 567
 568         else if (name == "geometry")
 569                 ; // Ignore this, the geometry settings are made by the \geometry
 570                   // command. This command is handled below.
 571
 572         else if (is_known(name, known_languages))
 573                 h_language = name;
 574
 575         else if (name == "natbib") {
 576                 h_cite_engine = "natbib_authoryear";
 577                 vector<string>::iterator it =
 578                         find(options.begin(), options.end(), "authoryear");
 579                 if (it != options.end())
 580                         options.erase(it);
 581                 else {
 582                         it = find(options.begin(), options.end(), "numbers");
 583                         if (it != options.end()) {
 584                                 h_cite_engine = "natbib_numerical";
 585                                 options.erase(it);
 586                         }
 587                 }
 588         }
 589
 590         else if (name == "jurabib")
 591                 h_cite_engine = "jurabib";
 592
 593         else if (name == "hyperref")
 594                 handle_hyperref(options);
 595
 596         else if (!in_lyx_preamble) {
 597                 if (options.empty())
 598                         h_preamble << "\\usepackage{" << name << "}";
 599                 else {
 600                         h_preamble << "\\usepackage[" << opts << "]{"
 601                                    << name << "}";
 602                         options.clear();
 603                 }
 604         }
 605
 606         // We need to do something with the options...
 607         if (!options.empty())
 608                 cerr << "Ignoring options '" << join(options, ",")
 609                      << "' of package " << name << '.' << endl;
 610
 611         // remove the whitespace
 612         p.skip_spaces();
 613 }
 614
 615
 616 void handle_if(Parser & p, bool in_lyx_preamble)
 617 {
 618         while (p.good()) {
 619                 Token t = p.get_token();
 620                 if (t.cat() == catEscape &&
 621                     is_known(t.cs(), known_if_commands))
 622                         handle_if(p, in_lyx_preamble);
 623                 else {
 624                         if (!in_lyx_preamble)
 625                                 h_preamble << t.asInput();
 626                         if (t.cat() == catEscape && t.cs() == "fi")
 627                                 return;
 628                 }
 629         }
 630 }
 631
 632
 633 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 634 {
 635         // translate from babel to LyX names
 636         h_language = babel2lyx(h_language);
 637
 638         // set the quote language
 639         // LyX only knows the following quotes languages:
 640         // english, swedish, german, polish, french and danish
 641         // (quotes for "japanese" and "chinese-traditional" are missing because
 642         //  they wouldn't be useful: http://www.lyx.org/trac/ticket/6383)
 643         // conversion list taken from
 644         // http://en.wikipedia.org/wiki/Quotation_mark,_non-English_usage
 645         // (quotes for kazakh and interlingua are unknown)
 646         // danish
 647         if (h_language == "danish")
 648                 h_quotes_language = "danish";
 649         // french
 650         else if (is_known(h_language, known_french_quotes_languages))
 651                 h_quotes_language = "french";
 652         // german
 653         else if (is_known(h_language, known_german_quotes_languages))
 654                 h_quotes_language = "german";
 655         // polish
 656         else if (is_known(h_language, known_polish_quotes_languages))
 657                 h_quotes_language = "polish";
 658         // swedish
 659         else if (is_known(h_language, known_swedish_quotes_languages))
 660                 h_quotes_language = "swedish";
 661         //english
 662         else if (is_known(h_language, known_english_quotes_languages))
 663                 h_quotes_language = "english";
 664
 665         // output the LyX file settings
 666         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 667            << "\\lyxformat " << LYX_FORMAT << '\n'
 668            << "\\begin_document\n"
 669            << "\\begin_header\n"
 670            << "\\textclass " << h_textclass << "\n";
 671         if (!h_preamble.str().empty())
 672                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 673         if (!h_options.empty())
 674                 os << "\\options " << h_options << "\n";
 675         os << "\\use_default_options " << h_use_default_options << "\n"
 676            << "\\language " << h_language << "\n"
 677            << "\\inputencoding " << h_inputencoding << "\n"
 678            << "\\font_roman " << h_font_roman << "\n"
 679            << "\\font_sans " << h_font_sans << "\n"
 680            << "\\font_typewriter " << h_font_typewriter << "\n"
 681            << "\\font_default_family " << h_font_default_family << "\n"
 682            << "\\font_sc " << h_font_sc << "\n"
 683            << "\\font_osf " << h_font_osf << "\n"
 684            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 685            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 686            << "\\graphics " << h_graphics << "\n";
 687         if (!h_float_placement.empty())
 688                 os << "\\float_placement " << h_float_placement << "\n";
 689         os << "\\paperfontsize " << h_paperfontsize << "\n"
 690            << "\\spacing " << h_spacing << "\n"
 691            << "\\use_hyperref " << h_use_hyperref << '\n';
 692         if (h_use_hyperref == "1") {
 693                 if (!h_pdf_title.empty())
 694                         os << "\\pdf_title \"" << h_pdf_title << "\"\n";
 695                 if (!h_pdf_author.empty())
 696                         os << "\\pdf_author \"" << h_pdf_author << "\"\n";
 697                 if (!h_pdf_subject.empty())
 698                         os << "\\pdf_subject \"" << h_pdf_subject << "\"\n";
 699                 if (!h_pdf_keywords.empty())
 700                         os << "\\pdf_keywords \"" << h_pdf_keywords << "\"\n";
 701                 os << "\\pdf_bookmarks " << h_pdf_bookmarks << "\n"
 702                       "\\pdf_bookmarksnumbered " << h_pdf_bookmarksnumbered << "\n"
 703                       "\\pdf_bookmarksopen " << h_pdf_bookmarksopen << "\n"
 704                       "\\pdf_bookmarksopenlevel " << h_pdf_bookmarksopenlevel << "\n"
 705                       "\\pdf_breaklinks " << h_pdf_breaklinks << "\n"
 706                       "\\pdf_pdfborder " << h_pdf_pdfborder << "\n"
 707                       "\\pdf_colorlinks " << h_pdf_colorlinks << "\n"
 708                       "\\pdf_backref " << h_pdf_backref << "\n"
 709                       "\\pdf_pdfusetitle " << h_pdf_pdfusetitle << '\n';
 710                 if (!h_pdf_pagemode.empty())
 711                         os << "\\pdf_pagemode " << h_pdf_pagemode << '\n';
 712                 if (!h_pdf_quoted_options.empty())
 713                         os << "\\pdf_quoted_options \"" << h_pdf_quoted_options << "\"\n";
 714         }
 715         os << "\\papersize " << h_papersize << "\n"
 716            << "\\use_geometry " << h_use_geometry << "\n"
 717            << "\\use_amsmath " << h_use_amsmath << "\n"
 718            << "\\use_esint " << h_use_esint << "\n"
 719            << "\\cite_engine " << h_cite_engine << "\n"
 720            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 721            << "\\paperorientation " << h_paperorientation << "\n"
 722            << h_margins
 723            << "\\secnumdepth " << h_secnumdepth << "\n"
 724            << "\\tocdepth " << h_tocdepth << "\n"
 725            << "\\paragraph_separation " << h_paragraph_separation << "\n";
 726         if (LYX_FORMAT < 365 || h_paragraph_separation == "skip")
 727                 os << "\\defskip " << h_defskip << "\n";
 728         else
 729                 os << "\\paragraph_indentation " << h_paragraph_indentation << "\n";
 730         os << "\\quotes_language " << h_quotes_language << "\n"
 731            << "\\papercolumns " << h_papercolumns << "\n"
 732            << "\\papersides " << h_papersides << "\n"
 733            << "\\paperpagestyle " << h_paperpagestyle << "\n";
 734         if (!h_listings_params.empty())
 735                 os << "\\listings_params " << h_listings_params << "\n";
 736         os << "\\tracking_changes " << h_tracking_changes << "\n"
 737            << "\\output_changes " << h_output_changes << "\n"
 738            << "\\end_header\n\n"
 739            << "\\begin_body\n";
 740         // clear preamble for subdocuments
 741         h_preamble.str("");
 742 }
 743
 744 } // anonymous namespace
 745
 746
 747 void parse_preamble(Parser & p, ostream & os,
 748         string const & forceclass, TeX2LyXDocClass & tc)
 749 {
 750         // initialize fixed types
 751         special_columns['D'] = 3;
 752         bool is_full_document = false;
 753         bool is_lyx_file = false;
 754         bool in_lyx_preamble = false;
 755
 756         // determine whether this is a full document or a fragment for inclusion
 757         while (p.good()) {
 758                 Token const & t = p.get_token();
 759
 760                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 761                         is_full_document = true;
 762                         break;
 763                 }
 764         }
 765         p.reset();
 766
 767         while (is_full_document && p.good()) {
 768                 Token const & t = p.get_token();
 769
 770 #ifdef FILEDEBUG
 771                 cerr << "t: " << t << "\n";
 772 #endif
 773
 774                 //
 775                 // cat codes
 776                 //
 777                 if (!in_lyx_preamble &&
 778                     (t.cat() == catLetter ||
 779                      t.cat() == catSuper ||
 780                      t.cat() == catSub ||
 781                      t.cat() == catOther ||
 782                      t.cat() == catMath ||
 783                      t.cat() == catActive ||
 784                      t.cat() == catBegin ||
 785                      t.cat() == catEnd ||
 786                      t.cat() == catAlign ||
 787                      t.cat() == catParameter))
 788                         h_preamble << t.cs();
 789
 790                 else if (!in_lyx_preamble &&
 791                          (t.cat() == catSpace || t.cat() == catNewline))
 792                         h_preamble << t.asInput();
 793
 794                 else if (t.cat() == catComment) {
 795                         // regex to parse comments (currently not used)
 796                         static regex const islyxfile("%% LyX .* created this file");
 797                         static regex const usercommands("User specified LaTeX commands");
 798
 799                         string const comment = t.asInput();
 800
 801                         // magically switch encoding default if it looks like XeLaTeX
 802                         static string const magicXeLaTeX =
 803                                 "% This document must be compiled with XeLaTeX ";
 804                         if (comment.size() > magicXeLaTeX.size()
 805                                   && comment.substr(0, magicXeLaTeX.size()) == magicXeLaTeX
 806                                   && h_inputencoding == "auto") {
 807                                 cerr << "XeLaTeX comment found, switching to UTF8\n";
 808                                 h_inputencoding = "utf8";
 809                         }
 810                         smatch sub;
 811                         if (regex_search(comment, sub, islyxfile)) {
 812                                 is_lyx_file = true;
 813                                 in_lyx_preamble = true;
 814                         } else if (is_lyx_file
 815                                    && regex_search(comment, sub, usercommands))
 816                                 in_lyx_preamble = false;
 817                         else if (!in_lyx_preamble)
 818                                 h_preamble << t.asInput();
 819                 }
 820
 821                 else if (t.cs() == "pagestyle")
 822                         h_paperpagestyle = p.verbatim_item();
 823
 824                 else if (t.cs() == "makeatletter") {
 825                         // LyX takes care of this
 826                         p.setCatCode('@', catLetter);
 827                 }
 828
 829                 else if (t.cs() == "makeatother") {
 830                         // LyX takes care of this
 831                         p.setCatCode('@', catOther);
 832                 }
 833
 834                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 835                             || t.cs() == "providecommand"
 836                                 || t.cs() == "DeclareRobustCommand"
 837                                 || t.cs() == "ProvideTextCommandDefault"
 838                                 || t.cs() == "DeclareMathAccent") {
 839                         bool star = false;
 840                         if (p.next_token().character() == '*') {
 841                                 p.get_token();
 842                                 star = true;
 843                         }
 844                         string const name = p.verbatim_item();
 845                         string const opt1 = p.getOpt();
 846                         string const opt2 = p.getFullOpt();
 847                         string const body = p.verbatim_item();
 848                         // font settings
 849                         if (name == "\\rmdefault")
 850                                 if (is_known(body, known_roman_fonts))
 851                                         h_font_roman = body;
 852                         if (name == "\\sfdefault")
 853                                 if (is_known(body, known_sans_fonts))
 854                                         h_font_sans = body;
 855                         if (name == "\\ttdefault")
 856                                 if (is_known(body, known_typewriter_fonts))
 857                                         h_font_typewriter = body;
 858                         if (name == "\\familydefault") {
 859                                 string family = body;
 860                                 // remove leading "\"
 861                                 h_font_default_family = family.erase(0,1);
 862                         }
 863                         // only non-lyxspecific stuff
 864                         if (!in_lyx_preamble) {
 865                                 ostringstream ss;
 866                                 ss << '\\' << t.cs();
 867                                 if (star)
 868                                         ss << '*';
 869                                 ss << '{' << name << '}' << opt1 << opt2
 870                                    << '{' << body << "}";
 871                                 h_preamble << ss.str();
 872
 873                                 // Add the command to the known commands
 874                                 add_known_command(name, opt1, !opt2.empty());
 875 /*
 876                                 ostream & out = in_preamble ? h_preamble : os;
 877                                 out << "\\" << t.cs() << "{" << name << "}"
 878                                     << opts << "{" << body << "}";
 879 */
 880                         }
 881                 }
 882
 883                 else if (t.cs() == "documentclass") {
 884                         vector<string>::iterator it;
 885                         vector<string> opts = split_options(p.getArg('[', ']'));
 886                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 887                         delete_opt(opts, known_fontsizes);
 888                         // delete "pt" at the end
 889                         string::size_type i = h_paperfontsize.find("pt");
 890                         if (i != string::npos)
 891                                 h_paperfontsize.erase(i);
 892                         // The documentclass options are always parsed before the options
 893                         // of the babel call so that a language cannot overwrite the babel
 894                         // options.
 895                         handle_opt(opts, known_languages, h_language);
 896                         delete_opt(opts, known_languages);
 897
 898                         // paper orientation
 899                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 900                                 h_paperorientation = "landscape";
 901                                 opts.erase(it);
 902                         }
 903                         // paper sides
 904                         if ((it = find(opts.begin(), opts.end(), "oneside"))
 905                                  != opts.end()) {
 906                                 h_papersides = "1";
 907                                 opts.erase(it);
 908                         }
 909                         if ((it = find(opts.begin(), opts.end(), "twoside"))
 910                                  != opts.end()) {
 911                                 h_papersides = "2";
 912                                 opts.erase(it);
 913                         }
 914                         // paper columns
 915                         if ((it = find(opts.begin(), opts.end(), "onecolumn"))
 916                                  != opts.end()) {
 917                                 h_papercolumns = "1";
 918                                 opts.erase(it);
 919                         }
 920                         if ((it = find(opts.begin(), opts.end(), "twocolumn"))
 921                                  != opts.end()) {
 922                                 h_papercolumns = "2";
 923                                 opts.erase(it);
 924                         }
 925                         // paper sizes
 926                         // some size options are know to any document classes, other sizes
 927                         // are handled by the \geometry command of the geometry package
 928                         handle_opt(opts, known_class_paper_sizes, h_papersize);
 929                         delete_opt(opts, known_class_paper_sizes);
 930                         // the remaining options
 931                         h_options = join(opts, ",");
 932                         h_textclass = p.getArg('{', '}');
 933                 }
 934
 935                 else if (t.cs() == "usepackage") {
 936                         string const options = p.getArg('[', ']');
 937                         string const name = p.getArg('{', '}');
 938                         vector<string> vecnames;
 939                         split(name, vecnames, ',');
 940                         vector<string>::const_iterator it  = vecnames.begin();
 941                         vector<string>::const_iterator end = vecnames.end();
 942                         for (; it != end; ++it)
 943                                 handle_package(p, trim(*it), options,
 944                                                in_lyx_preamble);
 945                 }
 946
 947                 else if (t.cs() == "inputencoding") {
 948                         string const encoding = p.getArg('{','}');
 949                         h_inputencoding = encoding;
 950                         p.setEncoding(encoding);
 951                 }
 952
 953                 else if (t.cs() == "newenvironment") {
 954                         string const name = p.getArg('{', '}');
 955                         ostringstream ss;
 956                         ss << "\\newenvironment{" << name << "}";
 957                         ss << p.getOpt();
 958                         ss << p.getOpt();
 959                         ss << '{' << p.verbatim_item() << '}';
 960                         ss << '{' << p.verbatim_item() << '}';
 961                         if (!in_lyx_preamble)
 962                                 h_preamble << ss.str();
 963                 }
 964
 965                 else if (t.cs() == "def") {
 966                         string name = p.get_token().cs();
 967                         while (p.next_token().cat() != catBegin)
 968                                 name += p.get_token().cs();
 969                         if (!in_lyx_preamble)
 970                                 h_preamble << "\\def\\" << name << '{'
 971                                            << p.verbatim_item() << "}";
 972                 }
 973
 974                 else if (t.cs() == "newcolumntype") {
 975                         string const name = p.getArg('{', '}');
 976                         trim(name);
 977                         int nargs = 0;
 978                         string opts = p.getOpt();
 979                         if (!opts.empty()) {
 980                                 istringstream is(string(opts, 1));
 981                                 is >> nargs;
 982                         }
 983                         special_columns[name[0]] = nargs;
 984                         h_preamble << "\\newcolumntype{" << name << "}";
 985                         if (nargs)
 986                                 h_preamble << "[" << nargs << "]";
 987                         h_preamble << "{" << p.verbatim_item() << "}";
 988                 }
 989
 990                 else if (t.cs() == "setcounter") {
 991                         string const name = p.getArg('{', '}');
 992                         string const content = p.getArg('{', '}');
 993                         if (name == "secnumdepth")
 994                                 h_secnumdepth = content;
 995                         else if (name == "tocdepth")
 996                                 h_tocdepth = content;
 997                         else
 998                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
 999                 }
1000
1001                 else if (t.cs() == "setlength") {
1002                         string const name = p.verbatim_item();
1003                         string const content = p.verbatim_item();
1004                         // the paragraphs are only not indented when \parindent is set to zero
1005                         if (name == "\\parindent" && content != "") {
1006                                 if (content[0] == '0')
1007                                         h_paragraph_separation = "skip";
1008                                 else if (LYX_FORMAT >= 365)
1009                                         h_paragraph_indentation = translate_len(content);
1010                                 else
1011                                         h_preamble << "\\setlength{" << name
1012                                                    << "}{" << content << "}";
1013                         } else if (name == "\\parskip") {
1014                                 if (content == "\\smallskipamount")
1015                                         h_defskip = "smallskip";
1016                                 else if (content == "\\medskipamount")
1017                                         h_defskip = "medskip";
1018                                 else if (content == "\\bigskipamount")
1019                                         h_defskip = "bigskip";
1020                                 else
1021                                         h_defskip = content;
1022                         } else
1023                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
1024                 }
1025
1026                 else if (t.cs() == "onehalfspacing")
1027                         h_spacing = "onehalf";
1028
1029                 else if (t.cs() == "doublespacing")
1030                         h_spacing = "double";
1031
1032                 else if (t.cs() == "setstretch")
1033                         h_spacing = "other " + p.verbatim_item();
1034
1035                 else if (t.cs() == "begin") {
1036                         string const name = p.getArg('{', '}');
1037                         if (name == "document")
1038                                 break;
1039                         h_preamble << "\\begin{" << name << "}";
1040                 }
1041
1042                 else if (t.cs() == "geometry") {
1043                         h_use_geometry = "true";
1044                         vector<string> opts = split_options(p.getArg('{', '}'));
1045                         vector<string>::iterator it;
1046                         // paper orientation
1047                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
1048                                 h_paperorientation = "landscape";
1049                                 opts.erase(it);
1050                         }
1051                         // paper size
1052                         handle_opt(opts, known_paper_sizes, h_papersize);
1053                         delete_opt(opts, known_paper_sizes);
1054                         // page margins
1055                         char const * const * margin = known_paper_margins;
1056                         int k = -1;
1057                         for (; *margin; ++margin) {
1058                                 k += 1;
1059                                 // search for the "=" in e.g. "lmargin=2cm" to get the value
1060                                 for(size_t i = 0; i != opts.size(); i++) {
1061                                         if (opts.at(i).find(*margin) != string::npos) {
1062                                                 string::size_type pos = opts.at(i).find("=");
1063                                                 string value = opts.at(i).substr(pos + 1);
1064                                                 string name = known_coded_paper_margins[k];
1065                                                 h_margins += "\\" + name + " " + value + "\n";
1066                                         }
1067                                 }
1068                         }
1069                 }
1070
1071                 else if (t.cs() == "jurabibsetup") {
1072                         // FIXME p.getArg('{', '}') is most probably wrong (it
1073                         //       does not handle nested braces).
1074                         //       Use p.verbatim_item() instead.
1075                         vector<string> jurabibsetup =
1076                                 split_options(p.getArg('{', '}'));
1077                         // add jurabibsetup to the jurabib package options
1078                         add_package("jurabib", jurabibsetup);
1079                         if (!jurabibsetup.empty()) {
1080                                 h_preamble << "\\jurabibsetup{"
1081                                            << join(jurabibsetup, ",") << '}';
1082                         }
1083                 }
1084
1085                 else if (t.cs() == "hypersetup") {
1086                         vector<string> hypersetup =
1087                                 split_options(p.verbatim_item());
1088                         // add hypersetup to the hyperref package options
1089                         handle_hyperref(hypersetup);
1090                         if (!hypersetup.empty()) {
1091                                 h_preamble << "\\hypersetup{"
1092                                            << join(hypersetup, ",") << '}';
1093                         }
1094                 }
1095
1096                 else if (is_known(t.cs(), known_if_3arg_commands)) {
1097                         // prevent misparsing of \usepackage if it is used
1098                         // as an argument (see e.g. our own output of
1099                         // \@ifundefined above)
1100                         string const arg1 = p.verbatim_item();
1101                         string const arg2 = p.verbatim_item();
1102                         string const arg3 = p.verbatim_item();
1103                         if (!in_lyx_preamble) {
1104                                 h_preamble << t.asInput()
1105                                            << '{' << arg1 << '}'
1106                                            << '{' << arg2 << '}'
1107                                            << '{' << arg3 << '}';
1108                         }
1109                 }
1110
1111                 else if (is_known(t.cs(), known_if_commands)) {
1112                         // must not parse anything in conditional code, since
1113                         // LyX would output the parsed contents unconditionally
1114                         if (!in_lyx_preamble)
1115                                 h_preamble << t.asInput();
1116                         handle_if(p, in_lyx_preamble);
1117                 }
1118
1119                 else if (!t.cs().empty() && !in_lyx_preamble)
1120                         h_preamble << '\\' << t.cs();
1121         }
1122
1123         // remove the whitespace
1124         p.skip_spaces();
1125
1126         // Force textclass if the user wanted it
1127         if (!forceclass.empty())
1128                 h_textclass = forceclass;
1129         if (noweb_mode && !prefixIs(h_textclass, "literate-"))
1130                 h_textclass.insert(0, "literate-");
1131         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
1132         if (layoutfilename.empty()) {
1133                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
1134                 exit(1);
1135         }
1136         tc.read(layoutfilename);
1137         if (h_papersides.empty()) {
1138                 ostringstream ss;
1139                 ss << tc.sides();
1140                 h_papersides = ss.str();
1141         }
1142         end_preamble(os, tc);
1143 }
1144
1145
1146 /// translates a babel language name to a LyX language name
1147 string babel2lyx(string const & language)
1148 {
1149         char const * const * where = is_known(language, known_languages);
1150         if (where)
1151                 return known_coded_languages[where - known_languages];
1152         return language;
1153 }
1154
1155 // }])
1156
1157
1158 } // namespace lyx