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_notefontcolor;
 204 string h_secnumdepth             = "3";
 205 string h_tocdepth                = "3";
 206 string h_paragraph_separation    = "indent";
 207 string h_defskip                 = "medskip";
 208 string h_paragraph_indentation   = "default";
 209 string h_quotes_language         = "english";
 210 string h_papercolumns            = "1";
 211 string h_papersides;
 212 string h_paperpagestyle          = "default";
 213 string h_listings_params;
 214 string h_tracking_changes        = "false";
 215 string h_output_changes          = "false";
 216 string h_margins;
 217
 218
 219 // returns true if at least one of the options in what has been found
 220 bool handle_opt(vector<string> & opts, char const * const * what, string & target)
 221 {
 222         if (opts.empty())
 223                 return false;
 224
 225         bool found = false;
 226         // the last language option is the document language (for babel and LyX)
 227         // the last size option is the document font size
 228         vector<string>::iterator it;
 229         vector<string>::iterator position = opts.begin();
 230         for (; *what; ++what) {
 231                 it = find(opts.begin(), opts.end(), *what);
 232                 if (it != opts.end()) {
 233                         if (it >= position) {
 234                                 found = true;
 235                                 target = *what;
 236                                 position = it;
 237                         }
 238                 }
 239         }
 240         return found;
 241 }
 242
 243
 244 void delete_opt(vector<string> & opts, char const * const * what)
 245 {
 246         if (opts.empty())
 247                 return;
 248
 249         // remove found options from the list
 250         // do this after handle_opt to avoid potential memory leaks
 251         vector<string>::iterator it;
 252         for (; *what; ++what) {
 253                 it = find(opts.begin(), opts.end(), *what);
 254                 if (it != opts.end())
 255                         opts.erase(it);
 256         }
 257 }
 258
 259
 260 /*!
 261  * Split a package options string (keyval format) into a vector.
 262  * Example input:
 263  *   authorformat=smallcaps,
 264  *   commabeforerest,
 265  *   titleformat=colonsep,
 266  *   bibformat={tabular,ibidem,numbered}
 267  */
 268 vector<string> split_options(string const & input)
 269 {
 270         vector<string> options;
 271         string option;
 272         Parser p(input);
 273         while (p.good()) {
 274                 Token const & t = p.get_token();
 275                 if (t.asInput() == ",") {
 276                         options.push_back(trim(option));
 277                         option.erase();
 278                 } else if (t.asInput() == "=") {
 279                         option += '=';
 280                         p.skip_spaces(true);
 281                         if (p.next_token().asInput() == "{")
 282                                 option += '{' + p.getArg('{', '}') + '}';
 283                 } else if (t.cat() != catSpace)
 284                         option += t.asInput();
 285         }
 286
 287         if (!option.empty())
 288                 options.push_back(trim(option));
 289
 290         return options;
 291 }
 292
 293
 294 /*!
 295  * Retrieve a keyval option "name={value with=sign}" named \p name from
 296  * \p options and return the value.
 297  * The found option is also removed from \p options.
 298  */
 299 string process_keyval_opt(vector<string> & options, string name)
 300 {
 301         for (size_t i = 0; i < options.size(); ++i) {
 302                 vector<string> option;
 303                 split(options[i], option, '=');
 304                 if (option.size() < 2)
 305                         continue;
 306                 if (option[0] == name) {
 307                         options.erase(options.begin() + i);
 308                         option.erase(option.begin());
 309                         return join(option, "=");
 310                 }
 311         }
 312         return "";
 313 }
 314
 315
 316 /*!
 317  * Add package \p name with options \p options to used_packages.
 318  * Remove options from \p options that we don't want to output.
 319  */
 320 void add_package(string const & name, vector<string> & options)
 321 {
 322         // every package inherits the global options
 323         if (used_packages.find(name) == used_packages.end())
 324                 used_packages[name] = split_options(h_options);
 325
 326         vector<string> & v = used_packages[name];
 327         v.insert(v.end(), options.begin(), options.end());
 328         if (name == "jurabib") {
 329                 // Don't output the order argument (see the cite command
 330                 // handling code in text.cpp).
 331                 vector<string>::iterator end =
 332                         remove(options.begin(), options.end(), "natbiborder");
 333                 end = remove(options.begin(), end, "jurabiborder");
 334                 options.erase(end, options.end());
 335         }
 336 }
 337
 338
 339 // Given is a string like "scaled=0.9", return 0.9 * 100
 340 string const scale_as_percentage(string const & scale)
 341 {
 342         string::size_type pos = scale.find('=');
 343         if (pos != string::npos) {
 344                 string value = scale.substr(pos + 1);
 345                 if (isStrDbl(value))
 346                         return convert<string>(100 * convert<double>(value));
 347         }
 348         // If the input string didn't match our expectations.
 349         // return the default value "100"
 350         return "100";
 351 }
 352
 353
 354 string remove_braces(string const & value)
 355 {
 356         if (value.empty())
 357                 return value;
 358         if (value[0] == '{' && value[value.length()-1] == '}')
 359                 return value.substr(1, value.length()-2);
 360         return value;
 361 }
 362
 363
 364 void handle_hyperref(vector<string> & options)
 365 {
 366         // FIXME swallow inputencoding changes that might surround the
 367         //       hyperref setup if it was written by LyX
 368         h_use_hyperref = "1";
 369         // swallow "unicode=true", since LyX does always write that
 370         vector<string>::iterator it =
 371                 find(options.begin(), options.end(), "unicode=true");
 372         if (it != options.end())
 373                 options.erase(it);
 374         it = find(options.begin(), options.end(), "pdfusetitle");
 375         if (it != options.end()) {
 376                 h_pdf_pdfusetitle = "1";
 377                 options.erase(it);
 378         }
 379         string bookmarks = process_keyval_opt(options, "bookmarks");
 380         if (bookmarks == "true")
 381                 h_pdf_bookmarks = "1";
 382         else if (bookmarks == "false")
 383                 h_pdf_bookmarks = "0";
 384         if (h_pdf_bookmarks == "1") {
 385                 string bookmarksnumbered =
 386                         process_keyval_opt(options, "bookmarksnumbered");
 387                 if (bookmarksnumbered == "true")
 388                         h_pdf_bookmarksnumbered = "1";
 389                 else if (bookmarksnumbered == "false")
 390                         h_pdf_bookmarksnumbered = "0";
 391                 string bookmarksopen =
 392                         process_keyval_opt(options, "bookmarksopen");
 393                 if (bookmarksopen == "true")
 394                         h_pdf_bookmarksopen = "1";
 395                 else if (bookmarksopen == "false")
 396                         h_pdf_bookmarksopen = "0";
 397                 if (h_pdf_bookmarksopen == "1") {
 398                         string bookmarksopenlevel =
 399                                 process_keyval_opt(options, "bookmarksopenlevel");
 400                         if (!bookmarksopenlevel.empty())
 401                                 h_pdf_bookmarksopenlevel = bookmarksopenlevel;
 402                 }
 403         }
 404         string breaklinks = process_keyval_opt(options, "breaklinks");
 405         if (breaklinks == "true")
 406                 h_pdf_breaklinks = "1";
 407         else if (breaklinks == "false")
 408                 h_pdf_breaklinks = "0";
 409         string pdfborder = process_keyval_opt(options, "pdfborder");
 410         if (pdfborder == "{0 0 0}")
 411                 h_pdf_pdfborder = "1";
 412         else if (pdfborder == "{0 0 1}")
 413                 h_pdf_pdfborder = "0";
 414         string backref = process_keyval_opt(options, "backref");
 415         if (!backref.empty())
 416                 h_pdf_backref = backref;
 417         string colorlinks = process_keyval_opt(options, "colorlinks");
 418         if (colorlinks == "true")
 419                 h_pdf_colorlinks = "1";
 420         else if (colorlinks == "false")
 421                 h_pdf_colorlinks = "0";
 422         string pdfpagemode = process_keyval_opt(options, "pdfpagemode");
 423         if (!pdfpagemode.empty())
 424                 h_pdf_pagemode = pdfpagemode;
 425         string pdftitle = process_keyval_opt(options, "pdftitle");
 426         if (!pdftitle.empty()) {
 427                 h_pdf_title = remove_braces(pdftitle);
 428         }
 429         string pdfauthor = process_keyval_opt(options, "pdfauthor");
 430         if (!pdfauthor.empty()) {
 431                 h_pdf_author = remove_braces(pdfauthor);
 432         }
 433         string pdfsubject = process_keyval_opt(options, "pdfsubject");
 434         if (!pdfsubject.empty())
 435                 h_pdf_subject = remove_braces(pdfsubject);
 436         string pdfkeywords = process_keyval_opt(options, "pdfkeywords");
 437         if (!pdfkeywords.empty())
 438                 h_pdf_keywords = remove_braces(pdfkeywords);
 439         if (!options.empty()) {
 440                 if (!h_pdf_quoted_options.empty())
 441                         h_pdf_quoted_options += ',';
 442                 h_pdf_quoted_options += join(options, ",");
 443                 options.clear();
 444         }
 445 }
 446
 447
 448 void handle_package(Parser &p, string const & name, string const & opts,
 449                     bool in_lyx_preamble)
 450 {
 451         vector<string> options = split_options(opts);
 452         add_package(name, options);
 453         string scale;
 454
 455         // roman fonts
 456         if (is_known(name, known_roman_fonts)) {
 457                 h_font_roman = name;
 458                 p.skip_spaces();
 459         }
 460
 461         if (name == "fourier") {
 462                 h_font_roman = "utopia";
 463                 // when font uses real small capitals
 464                 if (opts == "expert")
 465                         h_font_sc = "true";
 466         }
 467
 468         if (name == "mathpazo")
 469                 h_font_roman = "palatino";
 470
 471         if (name == "mathptmx")
 472                 h_font_roman = "times";
 473
 474         // sansserif fonts
 475         if (is_known(name, known_sans_fonts)) {
 476                 h_font_sans = name;
 477                 if (!opts.empty()) {
 478                         scale = opts;
 479                         h_font_sf_scale = scale_as_percentage(scale);
 480                 }
 481         }
 482
 483         // typewriter fonts
 484         if (is_known(name, known_typewriter_fonts)) {
 485                 h_font_typewriter = name;
 486                 if (!opts.empty()) {
 487                         scale = opts;
 488                         h_font_tt_scale = scale_as_percentage(scale);
 489                 }
 490         }
 491
 492         // font uses old-style figure
 493         if (name == "eco")
 494                 h_font_osf = "true";
 495
 496         else if (name == "amsmath" || name == "amssymb")
 497                 h_use_amsmath = "2";
 498
 499         else if (name == "esint")
 500                 h_use_esint = "2";
 501
 502         else if (name == "babel" && !opts.empty()) {
 503                 // check if more than one option was used - used later for inputenc
 504                 // in case inputenc is parsed before babel, set the encoding to auto
 505                 if (options.begin() != options.end() - 1) {
 506                         one_language = false;
 507                         h_inputencoding = "auto";
 508                 }
 509                 // babel takes the last language of the option of its \usepackage
 510                 // call as document language. If there is no such language option, the
 511                 // last language in the documentclass options is used.
 512                 handle_opt(options, known_languages, h_language);
 513                 delete_opt(options, known_languages);
 514         }
 515
 516         else if (name == "fontenc")
 517                  ;// ignore this
 518
 519         else if (name == "inputenc") {
 520                 // h_inputencoding is only set when there is not more than one
 521                 // inputenc option because otherwise h_inputencoding must be
 522                 // set to "auto" (the default encoding of the document language)
 523                 // Therefore check for the "," character.
 524                 // It is also only set when there is not more then one babel
 525                 // language option but this is handled in the routine for babel.
 526                 if (opts.find(",") == string::npos && one_language == true)
 527                         h_inputencoding = opts;
 528                 if (!options.empty())
 529                         p.setEncoding(options.back());
 530                 options.clear();
 531         }
 532
 533         else if (name == "makeidx")
 534                 ; // ignore this
 535
 536         else if (name == "prettyref")
 537                 ; // ignore this
 538
 539         else if (name == "varioref")
 540                 ; // ignore this
 541
 542         else if (name == "verbatim")
 543                 ; // ignore this
 544
 545         else if (name == "nomencl")
 546                 ; // ignore this
 547
 548         else if (name == "textcomp")
 549                 ; // ignore this
 550
 551         else if (name == "url")
 552                 ; // ignore this
 553
 554         else if (LYX_FORMAT >= 408 && name == "subscript")
 555                 ; // ignore this
 556
 557         else if (name == "color") {
 558                 // with the following command this package is only loaded when needed for
 559                 // undefined colors, since we only support the predefined colors
 560                 h_preamble << "\\@ifundefined{definecolor}\n {\\usepackage{color}}{}\n";
 561         }
 562
 563         else if (name == "graphicx")
 564                 ; // ignore this
 565
 566         else if (name == "setspace")
 567                 ; // ignore this
 568
 569         else if (name == "geometry")
 570                 ; // Ignore this, the geometry settings are made by the \geometry
 571                   // command. This command is handled below.
 572
 573         else if (is_known(name, known_languages))
 574                 h_language = name;
 575
 576         else if (name == "natbib") {
 577                 h_cite_engine = "natbib_authoryear";
 578                 vector<string>::iterator it =
 579                         find(options.begin(), options.end(), "authoryear");
 580                 if (it != options.end())
 581                         options.erase(it);
 582                 else {
 583                         it = find(options.begin(), options.end(), "numbers");
 584                         if (it != options.end()) {
 585                                 h_cite_engine = "natbib_numerical";
 586                                 options.erase(it);
 587                         }
 588                 }
 589         }
 590
 591         else if (name == "jurabib")
 592                 h_cite_engine = "jurabib";
 593
 594         else if (name == "hyperref")
 595                 handle_hyperref(options);
 596
 597         else if (!in_lyx_preamble) {
 598                 if (options.empty())
 599                         h_preamble << "\\usepackage{" << name << "}";
 600                 else {
 601                         h_preamble << "\\usepackage[" << opts << "]{"
 602                                    << name << "}";
 603                         options.clear();
 604                 }
 605         }
 606
 607         // We need to do something with the options...
 608         if (!options.empty())
 609                 cerr << "Ignoring options '" << join(options, ",")
 610                      << "' of package " << name << '.' << endl;
 611
 612         // remove the whitespace
 613         p.skip_spaces();
 614 }
 615
 616
 617 void handle_if(Parser & p, bool in_lyx_preamble)
 618 {
 619         while (p.good()) {
 620                 Token t = p.get_token();
 621                 if (t.cat() == catEscape &&
 622                     is_known(t.cs(), known_if_commands))
 623                         handle_if(p, in_lyx_preamble);
 624                 else {
 625                         if (!in_lyx_preamble)
 626                                 h_preamble << t.asInput();
 627                         if (t.cat() == catEscape && t.cs() == "fi")
 628                                 return;
 629                 }
 630         }
 631 }
 632
 633
 634 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 635 {
 636         // translate from babel to LyX names
 637         h_language = babel2lyx(h_language);
 638
 639         // set the quote language
 640         // LyX only knows the following quotes languages:
 641         // english, swedish, german, polish, french and danish
 642         // (quotes for "japanese" and "chinese-traditional" are missing because
 643         //  they wouldn't be useful: http://www.lyx.org/trac/ticket/6383)
 644         // conversion list taken from
 645         // http://en.wikipedia.org/wiki/Quotation_mark,_non-English_usage
 646         // (quotes for kazakh and interlingua are unknown)
 647         // danish
 648         if (h_language == "danish")
 649                 h_quotes_language = "danish";
 650         // french
 651         else if (is_known(h_language, known_french_quotes_languages))
 652                 h_quotes_language = "french";
 653         // german
 654         else if (is_known(h_language, known_german_quotes_languages))
 655                 h_quotes_language = "german";
 656         // polish
 657         else if (is_known(h_language, known_polish_quotes_languages))
 658                 h_quotes_language = "polish";
 659         // swedish
 660         else if (is_known(h_language, known_swedish_quotes_languages))
 661                 h_quotes_language = "swedish";
 662         //english
 663         else if (is_known(h_language, known_english_quotes_languages))
 664                 h_quotes_language = "english";
 665
 666         // output the LyX file settings
 667         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 668            << "\\lyxformat " << LYX_FORMAT << '\n'
 669            << "\\begin_document\n"
 670            << "\\begin_header\n"
 671            << "\\textclass " << h_textclass << "\n";
 672         if (!h_preamble.str().empty())
 673                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 674         if (!h_options.empty())
 675                 os << "\\options " << h_options << "\n";
 676         os << "\\use_default_options " << h_use_default_options << "\n"
 677            << "\\language " << h_language << "\n"
 678            << "\\inputencoding " << h_inputencoding << "\n"
 679            << "\\font_roman " << h_font_roman << "\n"
 680            << "\\font_sans " << h_font_sans << "\n"
 681            << "\\font_typewriter " << h_font_typewriter << "\n"
 682            << "\\font_default_family " << h_font_default_family << "\n"
 683            << "\\font_sc " << h_font_sc << "\n"
 684            << "\\font_osf " << h_font_osf << "\n"
 685            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 686            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 687            << "\\graphics " << h_graphics << "\n";
 688         if (!h_float_placement.empty())
 689                 os << "\\float_placement " << h_float_placement << "\n";
 690         os << "\\paperfontsize " << h_paperfontsize << "\n"
 691            << "\\spacing " << h_spacing << "\n"
 692            << "\\use_hyperref " << h_use_hyperref << '\n';
 693         if (h_use_hyperref == "1") {
 694                 if (!h_pdf_title.empty())
 695                         os << "\\pdf_title \"" << h_pdf_title << "\"\n";
 696                 if (!h_pdf_author.empty())
 697                         os << "\\pdf_author \"" << h_pdf_author << "\"\n";
 698                 if (!h_pdf_subject.empty())
 699                         os << "\\pdf_subject \"" << h_pdf_subject << "\"\n";
 700                 if (!h_pdf_keywords.empty())
 701                         os << "\\pdf_keywords \"" << h_pdf_keywords << "\"\n";
 702                 os << "\\pdf_bookmarks " << h_pdf_bookmarks << "\n"
 703                       "\\pdf_bookmarksnumbered " << h_pdf_bookmarksnumbered << "\n"
 704                       "\\pdf_bookmarksopen " << h_pdf_bookmarksopen << "\n"
 705                       "\\pdf_bookmarksopenlevel " << h_pdf_bookmarksopenlevel << "\n"
 706                       "\\pdf_breaklinks " << h_pdf_breaklinks << "\n"
 707                       "\\pdf_pdfborder " << h_pdf_pdfborder << "\n"
 708                       "\\pdf_colorlinks " << h_pdf_colorlinks << "\n"
 709                       "\\pdf_backref " << h_pdf_backref << "\n"
 710                       "\\pdf_pdfusetitle " << h_pdf_pdfusetitle << '\n';
 711                 if (!h_pdf_pagemode.empty())
 712                         os << "\\pdf_pagemode " << h_pdf_pagemode << '\n';
 713                 if (!h_pdf_quoted_options.empty())
 714                         os << "\\pdf_quoted_options \"" << h_pdf_quoted_options << "\"\n";
 715         }
 716         os << "\\papersize " << h_papersize << "\n"
 717            << "\\use_geometry " << h_use_geometry << "\n"
 718            << "\\use_amsmath " << h_use_amsmath << "\n"
 719            << "\\use_esint " << h_use_esint << "\n"
 720            << "\\cite_engine " << h_cite_engine << "\n"
 721            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 722            << "\\paperorientation " << h_paperorientation << '\n';
 723         if (LYX_FORMAT >= 382 && !h_notefontcolor.empty())
 724                 os << "\\notefontcolor " << h_notefontcolor << '\n';
 725         os << h_margins
 726            << "\\secnumdepth " << h_secnumdepth << "\n"
 727            << "\\tocdepth " << h_tocdepth << "\n"
 728            << "\\paragraph_separation " << h_paragraph_separation << "\n";
 729         if (LYX_FORMAT < 365 || h_paragraph_separation == "skip")
 730                 os << "\\defskip " << h_defskip << "\n";
 731         else
 732                 os << "\\paragraph_indentation " << h_paragraph_indentation << "\n";
 733         os << "\\quotes_language " << h_quotes_language << "\n"
 734            << "\\papercolumns " << h_papercolumns << "\n"
 735            << "\\papersides " << h_papersides << "\n"
 736            << "\\paperpagestyle " << h_paperpagestyle << "\n";
 737         if (!h_listings_params.empty())
 738                 os << "\\listings_params " << h_listings_params << "\n";
 739         os << "\\tracking_changes " << h_tracking_changes << "\n"
 740            << "\\output_changes " << h_output_changes << "\n"
 741            << "\\end_header\n\n"
 742            << "\\begin_body\n";
 743         // clear preamble for subdocuments
 744         h_preamble.str("");
 745 }
 746
 747 } // anonymous namespace
 748
 749
 750 void parse_preamble(Parser & p, ostream & os,
 751         string const & forceclass, TeX2LyXDocClass & tc)
 752 {
 753         // initialize fixed types
 754         special_columns['D'] = 3;
 755         bool is_full_document = false;
 756         bool is_lyx_file = false;
 757         bool in_lyx_preamble = false;
 758
 759         // determine whether this is a full document or a fragment for inclusion
 760         while (p.good()) {
 761                 Token const & t = p.get_token();
 762
 763                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 764                         is_full_document = true;
 765                         break;
 766                 }
 767         }
 768         p.reset();
 769
 770         while (is_full_document && p.good()) {
 771                 Token const & t = p.get_token();
 772
 773 #ifdef FILEDEBUG
 774                 cerr << "t: " << t << "\n";
 775 #endif
 776
 777                 //
 778                 // cat codes
 779                 //
 780                 if (!in_lyx_preamble &&
 781                     (t.cat() == catLetter ||
 782                      t.cat() == catSuper ||
 783                      t.cat() == catSub ||
 784                      t.cat() == catOther ||
 785                      t.cat() == catMath ||
 786                      t.cat() == catActive ||
 787                      t.cat() == catBegin ||
 788                      t.cat() == catEnd ||
 789                      t.cat() == catAlign ||
 790                      t.cat() == catParameter))
 791                         h_preamble << t.cs();
 792
 793                 else if (!in_lyx_preamble &&
 794                          (t.cat() == catSpace || t.cat() == catNewline))
 795                         h_preamble << t.asInput();
 796
 797                 else if (t.cat() == catComment) {
 798                         // regex to parse comments (currently not used)
 799                         static regex const islyxfile("%% LyX .* created this file");
 800                         static regex const usercommands("User specified LaTeX commands");
 801
 802                         string const comment = t.asInput();
 803
 804                         // magically switch encoding default if it looks like XeLaTeX
 805                         static string const magicXeLaTeX =
 806                                 "% This document must be compiled with XeLaTeX ";
 807                         if (comment.size() > magicXeLaTeX.size()
 808                                   && comment.substr(0, magicXeLaTeX.size()) == magicXeLaTeX
 809                                   && h_inputencoding == "auto") {
 810                                 cerr << "XeLaTeX comment found, switching to UTF8\n";
 811                                 h_inputencoding = "utf8";
 812                         }
 813                         smatch sub;
 814                         if (regex_search(comment, sub, islyxfile)) {
 815                                 is_lyx_file = true;
 816                                 in_lyx_preamble = true;
 817                         } else if (is_lyx_file
 818                                    && regex_search(comment, sub, usercommands))
 819                                 in_lyx_preamble = false;
 820                         else if (!in_lyx_preamble)
 821                                 h_preamble << t.asInput();
 822                 }
 823
 824                 else if (t.cs() == "pagestyle")
 825                         h_paperpagestyle = p.verbatim_item();
 826
 827                 else if (t.cs() == "makeatletter") {
 828                         // LyX takes care of this
 829                         p.setCatCode('@', catLetter);
 830                 }
 831
 832                 else if (t.cs() == "makeatother") {
 833                         // LyX takes care of this
 834                         p.setCatCode('@', catOther);
 835                 }
 836
 837                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 838                             || t.cs() == "providecommand"
 839                                 || t.cs() == "DeclareRobustCommand"
 840                                 || t.cs() == "ProvideTextCommandDefault"
 841                                 || t.cs() == "DeclareMathAccent") {
 842                         bool star = false;
 843                         if (p.next_token().character() == '*') {
 844                                 p.get_token();
 845                                 star = true;
 846                         }
 847                         string const name = p.verbatim_item();
 848                         string const opt1 = p.getOpt();
 849                         string const opt2 = p.getFullOpt();
 850                         string const body = p.verbatim_item();
 851                         // font settings
 852                         if (name == "\\rmdefault")
 853                                 if (is_known(body, known_roman_fonts))
 854                                         h_font_roman = body;
 855                         if (name == "\\sfdefault")
 856                                 if (is_known(body, known_sans_fonts))
 857                                         h_font_sans = body;
 858                         if (name == "\\ttdefault")
 859                                 if (is_known(body, known_typewriter_fonts))
 860                                         h_font_typewriter = body;
 861                         if (name == "\\familydefault") {
 862                                 string family = body;
 863                                 // remove leading "\"
 864                                 h_font_default_family = family.erase(0,1);
 865                         }
 866                         // only non-lyxspecific stuff
 867                         if (!in_lyx_preamble) {
 868                                 ostringstream ss;
 869                                 ss << '\\' << t.cs();
 870                                 if (star)
 871                                         ss << '*';
 872                                 ss << '{' << name << '}' << opt1 << opt2
 873                                    << '{' << body << "}";
 874                                 h_preamble << ss.str();
 875
 876                                 // Add the command to the known commands
 877                                 add_known_command(name, opt1, !opt2.empty());
 878 /*
 879                                 ostream & out = in_preamble ? h_preamble : os;
 880                                 out << "\\" << t.cs() << "{" << name << "}"
 881                                     << opts << "{" << body << "}";
 882 */
 883                         }
 884                 }
 885
 886                 else if (t.cs() == "documentclass") {
 887                         vector<string>::iterator it;
 888                         vector<string> opts = split_options(p.getArg('[', ']'));
 889                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 890                         delete_opt(opts, known_fontsizes);
 891                         // delete "pt" at the end
 892                         string::size_type i = h_paperfontsize.find("pt");
 893                         if (i != string::npos)
 894                                 h_paperfontsize.erase(i);
 895                         // The documentclass options are always parsed before the options
 896                         // of the babel call so that a language cannot overwrite the babel
 897                         // options.
 898                         handle_opt(opts, known_languages, h_language);
 899                         delete_opt(opts, known_languages);
 900
 901                         // paper orientation
 902                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 903                                 h_paperorientation = "landscape";
 904                                 opts.erase(it);
 905                         }
 906                         // paper sides
 907                         if ((it = find(opts.begin(), opts.end(), "oneside"))
 908                                  != opts.end()) {
 909                                 h_papersides = "1";
 910                                 opts.erase(it);
 911                         }
 912                         if ((it = find(opts.begin(), opts.end(), "twoside"))
 913                                  != opts.end()) {
 914                                 h_papersides = "2";
 915                                 opts.erase(it);
 916                         }
 917                         // paper columns
 918                         if ((it = find(opts.begin(), opts.end(), "onecolumn"))
 919                                  != opts.end()) {
 920                                 h_papercolumns = "1";
 921                                 opts.erase(it);
 922                         }
 923                         if ((it = find(opts.begin(), opts.end(), "twocolumn"))
 924                                  != opts.end()) {
 925                                 h_papercolumns = "2";
 926                                 opts.erase(it);
 927                         }
 928                         // paper sizes
 929                         // some size options are know to any document classes, other sizes
 930                         // are handled by the \geometry command of the geometry package
 931                         handle_opt(opts, known_class_paper_sizes, h_papersize);
 932                         delete_opt(opts, known_class_paper_sizes);
 933                         // the remaining options
 934                         h_options = join(opts, ",");
 935                         h_textclass = p.getArg('{', '}');
 936                 }
 937
 938                 else if (t.cs() == "usepackage") {
 939                         string const options = p.getArg('[', ']');
 940                         string const name = p.getArg('{', '}');
 941                         vector<string> vecnames;
 942                         split(name, vecnames, ',');
 943                         vector<string>::const_iterator it  = vecnames.begin();
 944                         vector<string>::const_iterator end = vecnames.end();
 945                         for (; it != end; ++it)
 946                                 handle_package(p, trim(*it), options,
 947                                                in_lyx_preamble);
 948                 }
 949
 950                 else if (t.cs() == "inputencoding") {
 951                         string const encoding = p.getArg('{','}');
 952                         h_inputencoding = encoding;
 953                         p.setEncoding(encoding);
 954                 }
 955
 956                 else if (t.cs() == "newenvironment") {
 957                         string const name = p.getArg('{', '}');
 958                         ostringstream ss;
 959                         ss << "\\newenvironment{" << name << "}";
 960                         ss << p.getOpt();
 961                         ss << p.getOpt();
 962                         ss << '{' << p.verbatim_item() << '}';
 963                         ss << '{' << p.verbatim_item() << '}';
 964                         if (!in_lyx_preamble)
 965                                 h_preamble << ss.str();
 966                 }
 967
 968                 else if (t.cs() == "def") {
 969                         string name = p.get_token().cs();
 970                         while (p.next_token().cat() != catBegin)
 971                                 name += p.get_token().cs();
 972                         if (!in_lyx_preamble)
 973                                 h_preamble << "\\def\\" << name << '{'
 974                                            << p.verbatim_item() << "}";
 975                 }
 976
 977                 else if (t.cs() == "newcolumntype") {
 978                         string const name = p.getArg('{', '}');
 979                         trim(name);
 980                         int nargs = 0;
 981                         string opts = p.getOpt();
 982                         if (!opts.empty()) {
 983                                 istringstream is(string(opts, 1));
 984                                 is >> nargs;
 985                         }
 986                         special_columns[name[0]] = nargs;
 987                         h_preamble << "\\newcolumntype{" << name << "}";
 988                         if (nargs)
 989                                 h_preamble << "[" << nargs << "]";
 990                         h_preamble << "{" << p.verbatim_item() << "}";
 991                 }
 992
 993                 else if (t.cs() == "setcounter") {
 994                         string const name = p.getArg('{', '}');
 995                         string const content = p.getArg('{', '}');
 996                         if (name == "secnumdepth")
 997                                 h_secnumdepth = content;
 998                         else if (name == "tocdepth")
 999                                 h_tocdepth = content;
1000                         else
1001                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
1002                 }
1003
1004                 else if (t.cs() == "setlength") {
1005                         string const name = p.verbatim_item();
1006                         string const content = p.verbatim_item();
1007                         // the paragraphs are only not indented when \parindent is set to zero
1008                         if (name == "\\parindent" && content != "") {
1009                                 if (content[0] == '0')
1010                                         h_paragraph_separation = "skip";
1011                                 else if (LYX_FORMAT >= 365)
1012                                         h_paragraph_indentation = translate_len(content);
1013                                 else
1014                                         h_preamble << "\\setlength{" << name
1015                                                    << "}{" << content << "}";
1016                         } else if (name == "\\parskip") {
1017                                 if (content == "\\smallskipamount")
1018                                         h_defskip = "smallskip";
1019                                 else if (content == "\\medskipamount")
1020                                         h_defskip = "medskip";
1021                                 else if (content == "\\bigskipamount")
1022                                         h_defskip = "bigskip";
1023                                 else
1024                                         h_defskip = content;
1025                         } else
1026                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
1027                 }
1028
1029                 else if (t.cs() == "onehalfspacing")
1030                         h_spacing = "onehalf";
1031
1032                 else if (t.cs() == "doublespacing")
1033                         h_spacing = "double";
1034
1035                 else if (t.cs() == "setstretch")
1036                         h_spacing = "other " + p.verbatim_item();
1037
1038                 else if (t.cs() == "begin") {
1039                         string const name = p.getArg('{', '}');
1040                         if (name == "document")
1041                                 break;
1042                         h_preamble << "\\begin{" << name << "}";
1043                 }
1044
1045                 else if (t.cs() == "geometry") {
1046                         h_use_geometry = "true";
1047                         vector<string> opts = split_options(p.getArg('{', '}'));
1048                         vector<string>::iterator it;
1049                         // paper orientation
1050                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
1051                                 h_paperorientation = "landscape";
1052                                 opts.erase(it);
1053                         }
1054                         // paper size
1055                         handle_opt(opts, known_paper_sizes, h_papersize);
1056                         delete_opt(opts, known_paper_sizes);
1057                         // page margins
1058                         char const * const * margin = known_paper_margins;
1059                         int k = -1;
1060                         for (; *margin; ++margin) {
1061                                 k += 1;
1062                                 // search for the "=" in e.g. "lmargin=2cm" to get the value
1063                                 for(size_t i = 0; i != opts.size(); i++) {
1064                                         if (opts.at(i).find(*margin) != string::npos) {
1065                                                 string::size_type pos = opts.at(i).find("=");
1066                                                 string value = opts.at(i).substr(pos + 1);
1067                                                 string name = known_coded_paper_margins[k];
1068                                                 h_margins += "\\" + name + " " + value + "\n";
1069                                         }
1070                                 }
1071                         }
1072                 }
1073
1074                 else if (t.cs() == "definecolor") {
1075                         string const color = p.getArg('{', '}');
1076                         string const space = p.getArg('{', '}');
1077                         string const value = p.getArg('{', '}');
1078                         if (LYX_FORMAT >= 382 &&
1079                             color == "note_fontcolor" && space == "rgb") {
1080                                 RGBColor c(RGBColorFromLaTeX(value));
1081                                 h_notefontcolor = X11hexname(c);
1082                         } else {
1083                                 h_preamble << "\\definecolor{" << color
1084                                            << "}{" << space << "}{" << value
1085                                            << '}';
1086                         }
1087                 }
1088
1089                 else if (t.cs() == "jurabibsetup") {
1090                         // FIXME p.getArg('{', '}') is most probably wrong (it
1091                         //       does not handle nested braces).
1092                         //       Use p.verbatim_item() instead.
1093                         vector<string> jurabibsetup =
1094                                 split_options(p.getArg('{', '}'));
1095                         // add jurabibsetup to the jurabib package options
1096                         add_package("jurabib", jurabibsetup);
1097                         if (!jurabibsetup.empty()) {
1098                                 h_preamble << "\\jurabibsetup{"
1099                                            << join(jurabibsetup, ",") << '}';
1100                         }
1101                 }
1102
1103                 else if (t.cs() == "hypersetup") {
1104                         vector<string> hypersetup =
1105                                 split_options(p.verbatim_item());
1106                         // add hypersetup to the hyperref package options
1107                         handle_hyperref(hypersetup);
1108                         if (!hypersetup.empty()) {
1109                                 h_preamble << "\\hypersetup{"
1110                                            << join(hypersetup, ",") << '}';
1111                         }
1112                 }
1113
1114                 else if (is_known(t.cs(), known_if_3arg_commands)) {
1115                         // prevent misparsing of \usepackage if it is used
1116                         // as an argument (see e.g. our own output of
1117                         // \@ifundefined above)
1118                         string const arg1 = p.verbatim_item();
1119                         string const arg2 = p.verbatim_item();
1120                         string const arg3 = p.verbatim_item();
1121                         if (!in_lyx_preamble) {
1122                                 h_preamble << t.asInput()
1123                                            << '{' << arg1 << '}'
1124                                            << '{' << arg2 << '}'
1125                                            << '{' << arg3 << '}';
1126                         }
1127                 }
1128
1129                 else if (is_known(t.cs(), known_if_commands)) {
1130                         // must not parse anything in conditional code, since
1131                         // LyX would output the parsed contents unconditionally
1132                         if (!in_lyx_preamble)
1133                                 h_preamble << t.asInput();
1134                         handle_if(p, in_lyx_preamble);
1135                 }
1136
1137                 else if (!t.cs().empty() && !in_lyx_preamble)
1138                         h_preamble << '\\' << t.cs();
1139         }
1140
1141         // remove the whitespace
1142         p.skip_spaces();
1143
1144         // Force textclass if the user wanted it
1145         if (!forceclass.empty())
1146                 h_textclass = forceclass;
1147         if (noweb_mode && !prefixIs(h_textclass, "literate-"))
1148                 h_textclass.insert(0, "literate-");
1149         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
1150         if (layoutfilename.empty()) {
1151                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
1152                 exit(1);
1153         }
1154         tc.read(layoutfilename);
1155         if (h_papersides.empty()) {
1156                 ostringstream ss;
1157                 ss << tc.sides();
1158                 h_papersides = ss.str();
1159         }
1160         end_preamble(os, tc);
1161 }
1162
1163
1164 /// translates a babel language name to a LyX language name
1165 string babel2lyx(string const & language)
1166 {
1167         char const * const * where = is_known(language, known_languages);
1168         if (where)
1169                 return known_coded_languages[where - known_languages];
1170         return language;
1171 }
1172
1173 // }])
1174
1175
1176 } // namespace lyx