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 const char * const modules_placeholder = "\001modules\001";
  48
  49 // needed to handle encodings with babel
  50 bool one_language = true;
  51
  52 namespace {
  53
  54 //add this to known_languages when updating to lyxformat 266:
  55 // "armenian"
  56 //add these to known_languages when updating to lyxformat 268:
  57 //"chinese-simplified", "chinese-traditional", "japanese", "korean"
  58 // Both changes require first that support for non-babel languages (CJK,
  59 // armtex) is added.
  60 // add turkmen for lyxformat 383
  61 /**
  62  * known babel language names (including synonyms)
  63  * not in standard babel: arabic, arabtex, belarusian, serbian-latin, thai
  64  * not yet supported by LyX: kurmanji
  65  * please keep this in sync with known_coded_languages line by line!
  66  */
  67 const char * const known_languages[] = {"acadian", "afrikaans", "albanian",
  68 "american", "arabic", "arabtex", "austrian", "bahasa", "bahasai", "bahasam",
  69 "basque", "belarusian", "brazil", "brazilian", "breton", "british", "bulgarian",
  70 "canadian", "canadien", "catalan", "croatian", "czech", "danish", "dutch",
  71 "english", "esperanto", "estonian", "farsi", "finnish", "francais", "french",
  72 "frenchb", "frenchle", "frenchpro", "galician", "german", "germanb", "greek",
  73 "hebrew", "hungarian", "icelandic", "indon", "indonesian", "interlingua",
  74 "irish", "italian", "kazakh", "latin", "latvian", "lithuanian", "lowersorbian",
  75 "lsorbian", "magyar", "malay", "meyalu", "mongolian", "naustrian", "newzealand",
  76 "ngerman", "ngermanb", "norsk", "nynorsk", "polutonikogreek", "polish",
  77 "portuges", "portuguese", "romanian", "russian", "russianb", "samin",
  78 "scottish", "serbian", "serbian-latin", "slovak", "slovene", "spanish",
  79 "swedish", "thai", "turkish", "ukraineb", "ukrainian", "uppersorbian",
  80 "UKenglish", "USenglish", "usorbian", "vietnam", "welsh", 0};
  81
  82 /**
  83  * the same as known_languages with .lyx names
  84  * please keep this in sync with known_languages line by line!
  85  */
  86 const char * const known_coded_languages[] = {"french", "afrikaans", "albanian",
  87 "american", "arabic_arabi", "arabic_arabtex", "austrian", "bahasa", "bahasa", "bahasam",
  88 "basque", "belarusian", "brazilian", "brazilian", "breton", "british", "bulgarian",
  89 "canadian", "canadien", "catalan", "croatian", "czech", "danish", "dutch",
  90 "english", "esperanto", "estonian", "farsi", "finnish", "french", "french",
  91 "french", "french", "french", "galician", "german", "german", "greek",
  92 "hebrew", "magyar", "icelandic", "bahasa", "bahasa", "interlingua",
  93 "irish", "italian", "kazakh", "latin", "latvian", "lithuanian", "lowersorbian",
  94 "lowersorbian", "magyar", "bahasam", "bahasam", "mongolian", "naustrian", "english",
  95 "ngerman", "ngerman", "norsk", "nynorsk", "polutonikogreek", "polish",
  96 "portuguese", "portuguese", "romanian", "russian", "russian", "samin",
  97 "scottish", "serbian", "serbian-latin", "slovak", "slovene", "spanish",
  98 "swedish", "thai", "turkish", "ukrainian", "ukrainian", "uppersorbian",
  99 "uppersorbian", "english", "english", "vietnamese", "welsh", 0};
 100
 101 /// languages with english quotes (.lyx names)
 102 const char * const known_english_quotes_languages[] = {"american", "bahasa",
 103 "bahasam", "brazilian", "canadian", "chinese-simplified", "english",
 104 "esperanto", "hebrew", "irish", "korean", "portuguese", "scottish", "thai", 0};
 105
 106 //add this to known_french_quotes_languages when updating to
 107 //lyxformat 383: "turkmen"
 108 /// languages with french quotes (.lyx names)
 109 const char * const known_french_quotes_languages[] = {"albanian",
 110 "arabic_arabi", "arabic_arabtex", "basque", "canadien", "catalan", "french",
 111 "galician", "greek", "italian", "norsk", "nynorsk", "polutonikogreek",
 112 "russian", "spanish", "spanish-mexico", "turkish", "ukrainian", "vietnamese", 0};
 113
 114 /// languages with german quotes (.lyx names)
 115 const char * const known_german_quotes_languages[] = {"austrian", "bulgarian",
 116 "czech", "german", "icelandic", "lithuanian", "lowersorbian", "naustrian",
 117 "ngerman", "serbian", "serbian-latin", "slovak", "slovene", "uppersorbian", 0};
 118
 119 /// languages with polish quotes (.lyx names)
 120 const char * const known_polish_quotes_languages[] = {"afrikaans", "croatian",
 121 "dutch", "estonian", "magyar", "polish", "romanian", 0};
 122
 123 /// languages with swedish quotes (.lyx names)
 124 const char * const known_swedish_quotes_languages[] = {"finnish",
 125 "swedish", 0};
 126
 127 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
 128
 129 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
 130 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
 131
 132 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
 133 "helvet", "lmss", 0};
 134
 135 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
 136 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
 137 "newcent", 0};
 138
 139 const char * const known_paper_sizes[] = { "a3paper", "b3paper", "a4paper",
 140 "b4paper", "a5paper", "b5paper", "executivepaper", "legalpaper",
 141 "letterpaper", 0};
 142
 143 const char * const known_class_paper_sizes[] = { "a4paper", "a5paper",
 144 "executivepaper", "legalpaper", "letterpaper", 0};
 145
 146 const char * const known_paper_margins[] = { "lmargin", "tmargin", "rmargin",
 147 "bmargin", "headheight", "headsep", "footskip", "columnsep", 0};
 148
 149 const char * const known_coded_paper_margins[] = { "leftmargin", "topmargin",
 150 "rightmargin", "bottommargin", "headheight", "headsep", "footskip",
 151 "columnsep", 0};
 152
 153 /// commands that can start an \if...\else...\endif sequence
 154 const char * const known_if_commands[] = {"if", "ifarydshln", "ifbraket",
 155 "ifcancel", "ifcolortbl", "ifeurosym", "ifmarginnote", "ifmmode", "ifpdf",
 156 "ifsidecap", "ifupgreek", 0};
 157
 158 /// conditional commands with three arguments like \@ifundefined{}{}{}
 159 const char * const known_if_3arg_commands[] = {"@ifundefined", "IfFileExists",
 160 0};
 161
 162 // default settings
 163 ostringstream h_preamble;
 164 string h_textclass               = "article";
 165 string h_use_default_options     = "false";
 166 string h_options;
 167 string h_language                = "english";
 168 string h_inputencoding           = "auto";
 169 string h_font_roman              = "default";
 170 string h_font_sans               = "default";
 171 string h_font_typewriter         = "default";
 172 string h_font_default_family     = "default";
 173 string h_font_sc                 = "false";
 174 string h_font_osf                = "false";
 175 string h_font_sf_scale           = "100";
 176 string h_font_tt_scale           = "100";
 177 string h_graphics                = "default";
 178 string h_float_placement;
 179 string h_paperfontsize           = "default";
 180 string h_spacing                 = "single";
 181 string h_use_hyperref            = "0";
 182 string h_pdf_title;
 183 string h_pdf_author;
 184 string h_pdf_subject;
 185 string h_pdf_keywords;
 186 string h_pdf_bookmarks           = "1";
 187 string h_pdf_bookmarksnumbered   = "0";
 188 string h_pdf_bookmarksopen       = "0";
 189 string h_pdf_bookmarksopenlevel  = "1";
 190 string h_pdf_breaklinks          = "0";
 191 string h_pdf_pdfborder           = "0";
 192 string h_pdf_colorlinks          = "0";
 193 string h_pdf_backref             = "section";
 194 string h_pdf_pdfusetitle         = "1";
 195 string h_pdf_pagemode;
 196 string h_pdf_quoted_options;
 197 string h_papersize               = "default";
 198 string h_use_geometry            = "false";
 199 string h_use_amsmath             = "1";
 200 string h_use_esint               = "1";
 201 string h_cite_engine             = "basic";
 202 string h_use_bibtopic            = "false";
 203 string h_paperorientation        = "portrait";
 204 string h_notefontcolor;
 205 string h_secnumdepth             = "3";
 206 string h_tocdepth                = "3";
 207 string h_paragraph_separation    = "indent";
 208 string h_defskip                 = "medskip";
 209 string h_paragraph_indentation   = "default";
 210 string h_quotes_language         = "english";
 211 string h_papercolumns            = "1";
 212 string h_papersides;
 213 string h_paperpagestyle          = "default";
 214 string h_listings_params;
 215 string h_tracking_changes        = "false";
 216 string h_output_changes          = "false";
 217 string h_margins;
 218
 219
 220 // returns true if at least one of the options in what has been found
 221 bool handle_opt(vector<string> & opts, char const * const * what, string & target)
 222 {
 223         if (opts.empty())
 224                 return false;
 225
 226         bool found = false;
 227         // the last language option is the document language (for babel and LyX)
 228         // the last size option is the document font size
 229         vector<string>::iterator it;
 230         vector<string>::iterator position = opts.begin();
 231         for (; *what; ++what) {
 232                 it = find(opts.begin(), opts.end(), *what);
 233                 if (it != opts.end()) {
 234                         if (it >= position) {
 235                                 found = true;
 236                                 target = *what;
 237                                 position = it;
 238                         }
 239                 }
 240         }
 241         return found;
 242 }
 243
 244
 245 void delete_opt(vector<string> & opts, char const * const * what)
 246 {
 247         if (opts.empty())
 248                 return;
 249
 250         // remove found options from the list
 251         // do this after handle_opt to avoid potential memory leaks
 252         vector<string>::iterator it;
 253         for (; *what; ++what) {
 254                 it = find(opts.begin(), opts.end(), *what);
 255                 if (it != opts.end())
 256                         opts.erase(it);
 257         }
 258 }
 259
 260
 261 /*!
 262  * Split a package options string (keyval format) into a vector.
 263  * Example input:
 264  *   authorformat=smallcaps,
 265  *   commabeforerest,
 266  *   titleformat=colonsep,
 267  *   bibformat={tabular,ibidem,numbered}
 268  */
 269 vector<string> split_options(string const & input)
 270 {
 271         vector<string> options;
 272         string option;
 273         Parser p(input);
 274         while (p.good()) {
 275                 Token const & t = p.get_token();
 276                 if (t.asInput() == ",") {
 277                         options.push_back(trim(option));
 278                         option.erase();
 279                 } else if (t.asInput() == "=") {
 280                         option += '=';
 281                         p.skip_spaces(true);
 282                         if (p.next_token().asInput() == "{")
 283                                 option += '{' + p.getArg('{', '}') + '}';
 284                 } else if (t.cat() != catSpace)
 285                         option += t.asInput();
 286         }
 287
 288         if (!option.empty())
 289                 options.push_back(trim(option));
 290
 291         return options;
 292 }
 293
 294
 295 /*!
 296  * Retrieve a keyval option "name={value with=sign}" named \p name from
 297  * \p options and return the value.
 298  * The found option is also removed from \p options.
 299  */
 300 string process_keyval_opt(vector<string> & options, string name)
 301 {
 302         for (size_t i = 0; i < options.size(); ++i) {
 303                 vector<string> option;
 304                 split(options[i], option, '=');
 305                 if (option.size() < 2)
 306                         continue;
 307                 if (option[0] == name) {
 308                         options.erase(options.begin() + i);
 309                         option.erase(option.begin());
 310                         return join(option, "=");
 311                 }
 312         }
 313         return "";
 314 }
 315
 316
 317 /*!
 318  * Add package \p name with options \p options to used_packages.
 319  * Remove options from \p options that we don't want to output.
 320  */
 321 void add_package(string const & name, vector<string> & options)
 322 {
 323         // every package inherits the global options
 324         if (used_packages.find(name) == used_packages.end())
 325                 used_packages[name] = split_options(h_options);
 326
 327         vector<string> & v = used_packages[name];
 328         v.insert(v.end(), options.begin(), options.end());
 329         if (name == "jurabib") {
 330                 // Don't output the order argument (see the cite command
 331                 // handling code in text.cpp).
 332                 vector<string>::iterator end =
 333                         remove(options.begin(), options.end(), "natbiborder");
 334                 end = remove(options.begin(), end, "jurabiborder");
 335                 options.erase(end, options.end());
 336         }
 337 }
 338
 339
 340 // Given is a string like "scaled=0.9", return 0.9 * 100
 341 string const scale_as_percentage(string const & scale)
 342 {
 343         string::size_type pos = scale.find('=');
 344         if (pos != string::npos) {
 345                 string value = scale.substr(pos + 1);
 346                 if (isStrDbl(value))
 347                         return convert<string>(100 * convert<double>(value));
 348         }
 349         // If the input string didn't match our expectations.
 350         // return the default value "100"
 351         return "100";
 352 }
 353
 354
 355 string remove_braces(string const & value)
 356 {
 357         if (value.empty())
 358                 return value;
 359         if (value[0] == '{' && value[value.length()-1] == '}')
 360                 return value.substr(1, value.length()-2);
 361         return value;
 362 }
 363
 364
 365 void handle_hyperref(vector<string> & options)
 366 {
 367         // FIXME swallow inputencoding changes that might surround the
 368         //       hyperref setup if it was written by LyX
 369         h_use_hyperref = "1";
 370         // swallow "unicode=true", since LyX does always write that
 371         vector<string>::iterator it =
 372                 find(options.begin(), options.end(), "unicode=true");
 373         if (it != options.end())
 374                 options.erase(it);
 375         it = find(options.begin(), options.end(), "pdfusetitle");
 376         if (it != options.end()) {
 377                 h_pdf_pdfusetitle = "1";
 378                 options.erase(it);
 379         }
 380         string bookmarks = process_keyval_opt(options, "bookmarks");
 381         if (bookmarks == "true")
 382                 h_pdf_bookmarks = "1";
 383         else if (bookmarks == "false")
 384                 h_pdf_bookmarks = "0";
 385         if (h_pdf_bookmarks == "1") {
 386                 string bookmarksnumbered =
 387                         process_keyval_opt(options, "bookmarksnumbered");
 388                 if (bookmarksnumbered == "true")
 389                         h_pdf_bookmarksnumbered = "1";
 390                 else if (bookmarksnumbered == "false")
 391                         h_pdf_bookmarksnumbered = "0";
 392                 string bookmarksopen =
 393                         process_keyval_opt(options, "bookmarksopen");
 394                 if (bookmarksopen == "true")
 395                         h_pdf_bookmarksopen = "1";
 396                 else if (bookmarksopen == "false")
 397                         h_pdf_bookmarksopen = "0";
 398                 if (h_pdf_bookmarksopen == "1") {
 399                         string bookmarksopenlevel =
 400                                 process_keyval_opt(options, "bookmarksopenlevel");
 401                         if (!bookmarksopenlevel.empty())
 402                                 h_pdf_bookmarksopenlevel = bookmarksopenlevel;
 403                 }
 404         }
 405         string breaklinks = process_keyval_opt(options, "breaklinks");
 406         if (breaklinks == "true")
 407                 h_pdf_breaklinks = "1";
 408         else if (breaklinks == "false")
 409                 h_pdf_breaklinks = "0";
 410         string pdfborder = process_keyval_opt(options, "pdfborder");
 411         if (pdfborder == "{0 0 0}")
 412                 h_pdf_pdfborder = "1";
 413         else if (pdfborder == "{0 0 1}")
 414                 h_pdf_pdfborder = "0";
 415         string backref = process_keyval_opt(options, "backref");
 416         if (!backref.empty())
 417                 h_pdf_backref = backref;
 418         string colorlinks = process_keyval_opt(options, "colorlinks");
 419         if (colorlinks == "true")
 420                 h_pdf_colorlinks = "1";
 421         else if (colorlinks == "false")
 422                 h_pdf_colorlinks = "0";
 423         string pdfpagemode = process_keyval_opt(options, "pdfpagemode");
 424         if (!pdfpagemode.empty())
 425                 h_pdf_pagemode = pdfpagemode;
 426         string pdftitle = process_keyval_opt(options, "pdftitle");
 427         if (!pdftitle.empty()) {
 428                 h_pdf_title = remove_braces(pdftitle);
 429         }
 430         string pdfauthor = process_keyval_opt(options, "pdfauthor");
 431         if (!pdfauthor.empty()) {
 432                 h_pdf_author = remove_braces(pdfauthor);
 433         }
 434         string pdfsubject = process_keyval_opt(options, "pdfsubject");
 435         if (!pdfsubject.empty())
 436                 h_pdf_subject = remove_braces(pdfsubject);
 437         string pdfkeywords = process_keyval_opt(options, "pdfkeywords");
 438         if (!pdfkeywords.empty())
 439                 h_pdf_keywords = remove_braces(pdfkeywords);
 440         if (!options.empty()) {
 441                 if (!h_pdf_quoted_options.empty())
 442                         h_pdf_quoted_options += ',';
 443                 h_pdf_quoted_options += join(options, ",");
 444                 options.clear();
 445         }
 446 }
 447
 448
 449 void handle_package(Parser &p, string const & name, string const & opts,
 450                     bool in_lyx_preamble)
 451 {
 452         vector<string> options = split_options(opts);
 453         add_package(name, options);
 454         string scale;
 455
 456         // roman fonts
 457         if (is_known(name, known_roman_fonts)) {
 458                 h_font_roman = name;
 459                 p.skip_spaces();
 460         }
 461
 462         if (name == "fourier") {
 463                 h_font_roman = "utopia";
 464                 // when font uses real small capitals
 465                 if (opts == "expert")
 466                         h_font_sc = "true";
 467         }
 468
 469         if (name == "mathpazo")
 470                 h_font_roman = "palatino";
 471
 472         if (name == "mathptmx")
 473                 h_font_roman = "times";
 474
 475         // sansserif fonts
 476         if (is_known(name, known_sans_fonts)) {
 477                 h_font_sans = name;
 478                 if (!opts.empty()) {
 479                         scale = opts;
 480                         h_font_sf_scale = scale_as_percentage(scale);
 481                 }
 482         }
 483
 484         // typewriter fonts
 485         if (is_known(name, known_typewriter_fonts)) {
 486                 h_font_typewriter = name;
 487                 if (!opts.empty()) {
 488                         scale = opts;
 489                         h_font_tt_scale = scale_as_percentage(scale);
 490                 }
 491         }
 492
 493         // font uses old-style figure
 494         if (name == "eco")
 495                 h_font_osf = "true";
 496
 497         else if (name == "amsmath" || name == "amssymb")
 498                 h_use_amsmath = "2";
 499
 500         else if (name == "esint")
 501                 h_use_esint = "2";
 502
 503         else if (name == "babel" && !opts.empty()) {
 504                 // check if more than one option was used - used later for inputenc
 505                 // in case inputenc is parsed before babel, set the encoding to auto
 506                 if (options.begin() != options.end() - 1) {
 507                         one_language = false;
 508                         h_inputencoding = "auto";
 509                 }
 510                 // babel takes the last language of the option of its \usepackage
 511                 // call as document language. If there is no such language option, the
 512                 // last language in the documentclass options is used.
 513                 handle_opt(options, known_languages, h_language);
 514                 delete_opt(options, known_languages);
 515         }
 516
 517         else if (name == "fontenc")
 518                  ;// ignore this
 519
 520         else if (name == "inputenc") {
 521                 // h_inputencoding is only set when there is not more than one
 522                 // inputenc option because otherwise h_inputencoding must be
 523                 // set to "auto" (the default encoding of the document language)
 524                 // Therefore check for the "," character.
 525                 // It is also only set when there is not more then one babel
 526                 // language option but this is handled in the routine for babel.
 527                 if (opts.find(",") == string::npos && one_language == true)
 528                         h_inputencoding = opts;
 529                 if (!options.empty())
 530                         p.setEncoding(options.back());
 531                 options.clear();
 532         }
 533
 534         else if (name == "makeidx")
 535                 ; // ignore this
 536
 537         else if (name == "prettyref")
 538                 ; // ignore this
 539
 540         else if (name == "varioref")
 541                 ; // ignore this
 542
 543         else if (name == "verbatim")
 544                 ; // ignore this
 545
 546         else if (name == "nomencl")
 547                 ; // ignore this
 548
 549         else if (name == "textcomp")
 550                 ; // ignore this
 551
 552         else if (name == "url")
 553                 ; // ignore this
 554
 555         else if (LYX_FORMAT >= 408 && name == "subscript")
 556                 ; // ignore this
 557
 558         else if (name == "color") {
 559                 // with the following command this package is only loaded when needed for
 560                 // undefined colors, since we only support the predefined colors
 561                 h_preamble << "\\@ifundefined{definecolor}\n {\\usepackage{color}}{}\n";
 562         }
 563
 564         else if (name == "graphicx")
 565                 ; // ignore this
 566
 567         else if (name == "setspace")
 568                 ; // ignore this
 569
 570         else if (name == "geometry")
 571                 ; // Ignore this, the geometry settings are made by the \geometry
 572                   // command. This command is handled below.
 573
 574         else if (is_known(name, known_languages))
 575                 h_language = name;
 576
 577         else if (name == "natbib") {
 578                 h_cite_engine = "natbib_authoryear";
 579                 vector<string>::iterator it =
 580                         find(options.begin(), options.end(), "authoryear");
 581                 if (it != options.end())
 582                         options.erase(it);
 583                 else {
 584                         it = find(options.begin(), options.end(), "numbers");
 585                         if (it != options.end()) {
 586                                 h_cite_engine = "natbib_numerical";
 587                                 options.erase(it);
 588                         }
 589                 }
 590         }
 591
 592         else if (name == "jurabib")
 593                 h_cite_engine = "jurabib";
 594
 595         else if (name == "hyperref")
 596                 handle_hyperref(options);
 597
 598         else if (!in_lyx_preamble) {
 599                 if (options.empty())
 600                         h_preamble << "\\usepackage{" << name << "}";
 601                 else {
 602                         h_preamble << "\\usepackage[" << opts << "]{"
 603                                    << name << "}";
 604                         options.clear();
 605                 }
 606         }
 607
 608         // We need to do something with the options...
 609         if (!options.empty())
 610                 cerr << "Ignoring options '" << join(options, ",")
 611                      << "' of package " << name << '.' << endl;
 612
 613         // remove the whitespace
 614         p.skip_spaces();
 615 }
 616
 617
 618 void handle_if(Parser & p, bool in_lyx_preamble)
 619 {
 620         while (p.good()) {
 621                 Token t = p.get_token();
 622                 if (t.cat() == catEscape &&
 623                     is_known(t.cs(), known_if_commands))
 624                         handle_if(p, in_lyx_preamble);
 625                 else {
 626                         if (!in_lyx_preamble)
 627                                 h_preamble << t.asInput();
 628                         if (t.cat() == catEscape && t.cs() == "fi")
 629                                 return;
 630                 }
 631         }
 632 }
 633
 634
 635 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 636 {
 637         // translate from babel to LyX names
 638         h_language = babel2lyx(h_language);
 639
 640         // set the quote language
 641         // LyX only knows the following quotes languages:
 642         // english, swedish, german, polish, french and danish
 643         // (quotes for "japanese" and "chinese-traditional" are missing because
 644         //  they wouldn't be useful: http://www.lyx.org/trac/ticket/6383)
 645         // conversion list taken from
 646         // http://en.wikipedia.org/wiki/Quotation_mark,_non-English_usage
 647         // (quotes for kazakh and interlingua are unknown)
 648         // danish
 649         if (h_language == "danish")
 650                 h_quotes_language = "danish";
 651         // french
 652         else if (is_known(h_language, known_french_quotes_languages))
 653                 h_quotes_language = "french";
 654         // german
 655         else if (is_known(h_language, known_german_quotes_languages))
 656                 h_quotes_language = "german";
 657         // polish
 658         else if (is_known(h_language, known_polish_quotes_languages))
 659                 h_quotes_language = "polish";
 660         // swedish
 661         else if (is_known(h_language, known_swedish_quotes_languages))
 662                 h_quotes_language = "swedish";
 663         //english
 664         else if (is_known(h_language, known_english_quotes_languages))
 665                 h_quotes_language = "english";
 666
 667         // output the LyX file settings
 668         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 669            << "\\lyxformat " << LYX_FORMAT << '\n'
 670            << "\\begin_document\n"
 671            << "\\begin_header\n"
 672            << "\\textclass " << h_textclass << "\n";
 673         if (!h_preamble.str().empty())
 674                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 675         if (!h_options.empty())
 676                 os << "\\options " << h_options << "\n";
 677         os << "\\use_default_options " << h_use_default_options << "\n"
 678            << modules_placeholder
 679            << "\\language " << h_language << "\n"
 680            << "\\inputencoding " << h_inputencoding << "\n"
 681            << "\\font_roman " << h_font_roman << "\n"
 682            << "\\font_sans " << h_font_sans << "\n"
 683            << "\\font_typewriter " << h_font_typewriter << "\n"
 684            << "\\font_default_family " << h_font_default_family << "\n"
 685            << "\\font_sc " << h_font_sc << "\n"
 686            << "\\font_osf " << h_font_osf << "\n"
 687            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 688            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 689            << "\\graphics " << h_graphics << "\n";
 690         if (!h_float_placement.empty())
 691                 os << "\\float_placement " << h_float_placement << "\n";
 692         os << "\\paperfontsize " << h_paperfontsize << "\n"
 693            << "\\spacing " << h_spacing << "\n"
 694            << "\\use_hyperref " << h_use_hyperref << '\n';
 695         if (h_use_hyperref == "1") {
 696                 if (!h_pdf_title.empty())
 697                         os << "\\pdf_title \"" << h_pdf_title << "\"\n";
 698                 if (!h_pdf_author.empty())
 699                         os << "\\pdf_author \"" << h_pdf_author << "\"\n";
 700                 if (!h_pdf_subject.empty())
 701                         os << "\\pdf_subject \"" << h_pdf_subject << "\"\n";
 702                 if (!h_pdf_keywords.empty())
 703                         os << "\\pdf_keywords \"" << h_pdf_keywords << "\"\n";
 704                 os << "\\pdf_bookmarks " << h_pdf_bookmarks << "\n"
 705                       "\\pdf_bookmarksnumbered " << h_pdf_bookmarksnumbered << "\n"
 706                       "\\pdf_bookmarksopen " << h_pdf_bookmarksopen << "\n"
 707                       "\\pdf_bookmarksopenlevel " << h_pdf_bookmarksopenlevel << "\n"
 708                       "\\pdf_breaklinks " << h_pdf_breaklinks << "\n"
 709                       "\\pdf_pdfborder " << h_pdf_pdfborder << "\n"
 710                       "\\pdf_colorlinks " << h_pdf_colorlinks << "\n"
 711                       "\\pdf_backref " << h_pdf_backref << "\n"
 712                       "\\pdf_pdfusetitle " << h_pdf_pdfusetitle << '\n';
 713                 if (!h_pdf_pagemode.empty())
 714                         os << "\\pdf_pagemode " << h_pdf_pagemode << '\n';
 715                 if (!h_pdf_quoted_options.empty())
 716                         os << "\\pdf_quoted_options \"" << h_pdf_quoted_options << "\"\n";
 717         }
 718         os << "\\papersize " << h_papersize << "\n"
 719            << "\\use_geometry " << h_use_geometry << "\n"
 720            << "\\use_amsmath " << h_use_amsmath << "\n"
 721            << "\\use_esint " << h_use_esint << "\n"
 722            << "\\cite_engine " << h_cite_engine << "\n"
 723            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 724            << "\\paperorientation " << h_paperorientation << '\n';
 725         if (LYX_FORMAT >= 382 && !h_notefontcolor.empty())
 726                 os << "\\notefontcolor " << h_notefontcolor << '\n';
 727         os << h_margins
 728            << "\\secnumdepth " << h_secnumdepth << "\n"
 729            << "\\tocdepth " << h_tocdepth << "\n"
 730            << "\\paragraph_separation " << h_paragraph_separation << "\n";
 731         if (LYX_FORMAT < 365 || h_paragraph_separation == "skip")
 732                 os << "\\defskip " << h_defskip << "\n";
 733         else
 734                 os << "\\paragraph_indentation " << h_paragraph_indentation << "\n";
 735         os << "\\quotes_language " << h_quotes_language << "\n"
 736            << "\\papercolumns " << h_papercolumns << "\n"
 737            << "\\papersides " << h_papersides << "\n"
 738            << "\\paperpagestyle " << h_paperpagestyle << "\n";
 739         if (!h_listings_params.empty())
 740                 os << "\\listings_params " << h_listings_params << "\n";
 741         os << "\\tracking_changes " << h_tracking_changes << "\n"
 742            << "\\output_changes " << h_output_changes << "\n"
 743            << "\\end_header\n\n"
 744            << "\\begin_body\n";
 745         // clear preamble for subdocuments
 746         h_preamble.str("");
 747 }
 748
 749 } // anonymous namespace
 750
 751
 752 void parse_preamble(Parser & p, ostream & os,
 753         string const & forceclass, TeX2LyXDocClass & tc)
 754 {
 755         // initialize fixed types
 756         special_columns['D'] = 3;
 757         bool is_full_document = false;
 758         bool is_lyx_file = false;
 759         bool in_lyx_preamble = false;
 760
 761         // determine whether this is a full document or a fragment for inclusion
 762         while (p.good()) {
 763                 Token const & t = p.get_token();
 764
 765                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 766                         is_full_document = true;
 767                         break;
 768                 }
 769         }
 770         p.reset();
 771
 772         while (is_full_document && p.good()) {
 773                 Token const & t = p.get_token();
 774
 775 #ifdef FILEDEBUG
 776                 cerr << "t: " << t << "\n";
 777 #endif
 778
 779                 //
 780                 // cat codes
 781                 //
 782                 if (!in_lyx_preamble &&
 783                     (t.cat() == catLetter ||
 784                      t.cat() == catSuper ||
 785                      t.cat() == catSub ||
 786                      t.cat() == catOther ||
 787                      t.cat() == catMath ||
 788                      t.cat() == catActive ||
 789                      t.cat() == catBegin ||
 790                      t.cat() == catEnd ||
 791                      t.cat() == catAlign ||
 792                      t.cat() == catParameter))
 793                         h_preamble << t.cs();
 794
 795                 else if (!in_lyx_preamble &&
 796                          (t.cat() == catSpace || t.cat() == catNewline))
 797                         h_preamble << t.asInput();
 798
 799                 else if (t.cat() == catComment) {
 800                         static regex const islyxfile("%% LyX .* created this file");
 801                         static regex const usercommands("User specified LaTeX commands");
 802
 803                         string const comment = t.asInput();
 804
 805                         // magically switch encoding default if it looks like XeLaTeX
 806                         static string const magicXeLaTeX =
 807                                 "% This document must be compiled with XeLaTeX ";
 808                         if (comment.size() > magicXeLaTeX.size()
 809                                   && comment.substr(0, magicXeLaTeX.size()) == magicXeLaTeX
 810                                   && h_inputencoding == "auto") {
 811                                 cerr << "XeLaTeX comment found, switching to UTF8\n";
 812                                 h_inputencoding = "utf8";
 813                         }
 814                         smatch sub;
 815                         if (regex_search(comment, sub, islyxfile)) {
 816                                 is_lyx_file = true;
 817                                 in_lyx_preamble = true;
 818                         } else if (is_lyx_file
 819                                    && regex_search(comment, sub, usercommands))
 820                                 in_lyx_preamble = false;
 821                         else if (!in_lyx_preamble)
 822                                 h_preamble << t.asInput();
 823                 }
 824
 825                 else if (t.cs() == "pagestyle")
 826                         h_paperpagestyle = p.verbatim_item();
 827
 828                 else if (t.cs() == "makeatletter") {
 829                         // LyX takes care of this
 830                         p.setCatCode('@', catLetter);
 831                 }
 832
 833                 else if (t.cs() == "makeatother") {
 834                         // LyX takes care of this
 835                         p.setCatCode('@', catOther);
 836                 }
 837
 838                 else if (t.cs() == "newcommand" || t.cs() == "newcommandx"
 839                       || t.cs() == "renewcommand" || t.cs() == "renewcommandx"
 840                       || t.cs() == "providecommand" || t.cs() == "providecommandx"
 841                                 || t.cs() == "DeclareRobustCommand"
 842                       || t.cs() == "DeclareRobustCommandx"
 843                                 || t.cs() == "ProvideTextCommandDefault"
 844                                 || t.cs() == "DeclareMathAccent") {
 845                         bool star = false;
 846                         if (p.next_token().character() == '*') {
 847                                 p.get_token();
 848                                 star = true;
 849                         }
 850                         string const name = p.verbatim_item();
 851                         string const opt1 = p.getFullOpt();
 852                         string const opt2 = p.getFullOpt();
 853                         string const body = p.verbatim_item();
 854                         // font settings
 855                         if (name == "\\rmdefault")
 856                                 if (is_known(body, known_roman_fonts))
 857                                         h_font_roman = body;
 858                         if (name == "\\sfdefault")
 859                                 if (is_known(body, known_sans_fonts))
 860                                         h_font_sans = body;
 861                         if (name == "\\ttdefault")
 862                                 if (is_known(body, known_typewriter_fonts))
 863                                         h_font_typewriter = body;
 864                         if (name == "\\familydefault") {
 865                                 string family = body;
 866                                 // remove leading "\"
 867                                 h_font_default_family = family.erase(0,1);
 868                         }
 869
 870                         // Add the command to the known commands
 871                         add_known_command(name, opt1, !opt2.empty(), from_utf8(body));
 872
 873                         // only non-lyxspecific stuff
 874                         if (!in_lyx_preamble) {
 875                                 ostringstream ss;
 876                                 ss << '\\' << t.cs();
 877                                 if (star)
 878                                         ss << '*';
 879                                 ss << '{' << name << '}' << opt1 << opt2
 880                                    << '{' << body << "}";
 881                                 h_preamble << ss.str();
 882 /*
 883                                 ostream & out = in_preamble ? h_preamble : os;
 884                                 out << "\\" << t.cs() << "{" << name << "}"
 885                                     << opts << "{" << body << "}";
 886 */
 887                         }
 888                 }
 889
 890                 else if (t.cs() == "documentclass") {
 891                         vector<string>::iterator it;
 892                         vector<string> opts = split_options(p.getArg('[', ']'));
 893                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 894                         delete_opt(opts, known_fontsizes);
 895                         // delete "pt" at the end
 896                         string::size_type i = h_paperfontsize.find("pt");
 897                         if (i != string::npos)
 898                                 h_paperfontsize.erase(i);
 899                         // The documentclass options are always parsed before the options
 900                         // of the babel call so that a language cannot overwrite the babel
 901                         // options.
 902                         handle_opt(opts, known_languages, h_language);
 903                         delete_opt(opts, known_languages);
 904
 905                         // paper orientation
 906                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 907                                 h_paperorientation = "landscape";
 908                                 opts.erase(it);
 909                         }
 910                         // paper sides
 911                         if ((it = find(opts.begin(), opts.end(), "oneside"))
 912                                  != opts.end()) {
 913                                 h_papersides = "1";
 914                                 opts.erase(it);
 915                         }
 916                         if ((it = find(opts.begin(), opts.end(), "twoside"))
 917                                  != opts.end()) {
 918                                 h_papersides = "2";
 919                                 opts.erase(it);
 920                         }
 921                         // paper columns
 922                         if ((it = find(opts.begin(), opts.end(), "onecolumn"))
 923                                  != opts.end()) {
 924                                 h_papercolumns = "1";
 925                                 opts.erase(it);
 926                         }
 927                         if ((it = find(opts.begin(), opts.end(), "twocolumn"))
 928                                  != opts.end()) {
 929                                 h_papercolumns = "2";
 930                                 opts.erase(it);
 931                         }
 932                         // paper sizes
 933                         // some size options are know to any document classes, other sizes
 934                         // are handled by the \geometry command of the geometry package
 935                         handle_opt(opts, known_class_paper_sizes, h_papersize);
 936                         delete_opt(opts, known_class_paper_sizes);
 937                         // the remaining options
 938                         h_options = join(opts, ",");
 939                         // FIXME This does not work for classes that have a
 940                         //       different name in LyX than in LaTeX
 941                         h_textclass = p.getArg('{', '}');
 942                 }
 943
 944                 else if (t.cs() == "usepackage") {
 945                         string const options = p.getArg('[', ']');
 946                         string const name = p.getArg('{', '}');
 947                         vector<string> vecnames;
 948                         split(name, vecnames, ',');
 949                         vector<string>::const_iterator it  = vecnames.begin();
 950                         vector<string>::const_iterator end = vecnames.end();
 951                         for (; it != end; ++it)
 952                                 handle_package(p, trim(*it), options,
 953                                                in_lyx_preamble);
 954                 }
 955
 956                 else if (t.cs() == "inputencoding") {
 957                         string const encoding = p.getArg('{','}');
 958                         h_inputencoding = encoding;
 959                         p.setEncoding(encoding);
 960                 }
 961
 962                 else if (t.cs() == "newenvironment") {
 963                         string const name = p.getArg('{', '}');
 964                         string const opt1 = p.getFullOpt();
 965                         string const opt2 = p.getFullOpt();
 966                         string const beg = p.verbatim_item();
 967                         string const end = p.verbatim_item();
 968                         if (!in_lyx_preamble) {
 969                                 h_preamble << "\\newenvironment{" << name
 970                                            << '}' << opt1 << opt2 << '{'
 971                                            << beg << "}{" << end << '}';
 972                         }
 973                         add_known_environment(name, opt1, !opt2.empty(),
 974                                               from_utf8(beg), from_utf8(end));
 975
 976                 }
 977
 978                 else if (t.cs() == "def") {
 979                         string name = p.get_token().cs();
 980                         while (p.next_token().cat() != catBegin)
 981                                 name += p.get_token().cs();
 982                         if (!in_lyx_preamble)
 983                                 h_preamble << "\\def\\" << name << '{'
 984                                            << p.verbatim_item() << "}";
 985                 }
 986
 987                 else if (t.cs() == "newcolumntype") {
 988                         string const name = p.getArg('{', '}');
 989                         trim(name);
 990                         int nargs = 0;
 991                         string opts = p.getOpt();
 992                         if (!opts.empty()) {
 993                                 istringstream is(string(opts, 1));
 994                                 is >> nargs;
 995                         }
 996                         special_columns[name[0]] = nargs;
 997                         h_preamble << "\\newcolumntype{" << name << "}";
 998                         if (nargs)
 999                                 h_preamble << "[" << nargs << "]";
1000                         h_preamble << "{" << p.verbatim_item() << "}";
1001                 }
1002
1003                 else if (t.cs() == "setcounter") {
1004                         string const name = p.getArg('{', '}');
1005                         string const content = p.getArg('{', '}');
1006                         if (name == "secnumdepth")
1007                                 h_secnumdepth = content;
1008                         else if (name == "tocdepth")
1009                                 h_tocdepth = content;
1010                         else
1011                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
1012                 }
1013
1014                 else if (t.cs() == "setlength") {
1015                         string const name = p.verbatim_item();
1016                         string const content = p.verbatim_item();
1017                         // the paragraphs are only not indented when \parindent is set to zero
1018                         if (name == "\\parindent" && content != "") {
1019                                 if (content[0] == '0')
1020                                         h_paragraph_separation = "skip";
1021                                 else if (LYX_FORMAT >= 365)
1022                                         h_paragraph_indentation = translate_len(content);
1023                                 else
1024                                         h_preamble << "\\setlength{" << name
1025                                                    << "}{" << content << "}";
1026                         } else if (name == "\\parskip") {
1027                                 if (content == "\\smallskipamount")
1028                                         h_defskip = "smallskip";
1029                                 else if (content == "\\medskipamount")
1030                                         h_defskip = "medskip";
1031                                 else if (content == "\\bigskipamount")
1032                                         h_defskip = "bigskip";
1033                                 else
1034                                         h_defskip = content;
1035                         } else
1036                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
1037                 }
1038
1039                 else if (t.cs() == "onehalfspacing")
1040                         h_spacing = "onehalf";
1041
1042                 else if (t.cs() == "doublespacing")
1043                         h_spacing = "double";
1044
1045                 else if (t.cs() == "setstretch")
1046                         h_spacing = "other " + p.verbatim_item();
1047
1048                 else if (t.cs() == "begin") {
1049                         string const name = p.getArg('{', '}');
1050                         if (name == "document")
1051                                 break;
1052                         h_preamble << "\\begin{" << name << "}";
1053                 }
1054
1055                 else if (t.cs() == "geometry") {
1056                         h_use_geometry = "true";
1057                         vector<string> opts = split_options(p.getArg('{', '}'));
1058                         vector<string>::iterator it;
1059                         // paper orientation
1060                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
1061                                 h_paperorientation = "landscape";
1062                                 opts.erase(it);
1063                         }
1064                         // paper size
1065                         handle_opt(opts, known_paper_sizes, h_papersize);
1066                         delete_opt(opts, known_paper_sizes);
1067                         // page margins
1068                         char const * const * margin = known_paper_margins;
1069                         int k = -1;
1070                         for (; *margin; ++margin) {
1071                                 k += 1;
1072                                 // search for the "=" in e.g. "lmargin=2cm" to get the value
1073                                 for(size_t i = 0; i != opts.size(); i++) {
1074                                         if (opts.at(i).find(*margin) != string::npos) {
1075                                                 string::size_type pos = opts.at(i).find("=");
1076                                                 string value = opts.at(i).substr(pos + 1);
1077                                                 string name = known_coded_paper_margins[k];
1078                                                 h_margins += "\\" + name + " " + value + "\n";
1079                                         }
1080                                 }
1081                         }
1082                 }
1083
1084                 else if (t.cs() == "definecolor") {
1085                         string const color = p.getArg('{', '}');
1086                         string const space = p.getArg('{', '}');
1087                         string const value = p.getArg('{', '}');
1088                         if (LYX_FORMAT >= 382 &&
1089                             color == "note_fontcolor" && space == "rgb") {
1090                                 RGBColor c(RGBColorFromLaTeX(value));
1091                                 h_notefontcolor = X11hexname(c);
1092                         } else {
1093                                 h_preamble << "\\definecolor{" << color
1094                                            << "}{" << space << "}{" << value
1095                                            << '}';
1096                         }
1097                 }
1098
1099                 else if (t.cs() == "jurabibsetup") {
1100                         // FIXME p.getArg('{', '}') is most probably wrong (it
1101                         //       does not handle nested braces).
1102                         //       Use p.verbatim_item() instead.
1103                         vector<string> jurabibsetup =
1104                                 split_options(p.getArg('{', '}'));
1105                         // add jurabibsetup to the jurabib package options
1106                         add_package("jurabib", jurabibsetup);
1107                         if (!jurabibsetup.empty()) {
1108                                 h_preamble << "\\jurabibsetup{"
1109                                            << join(jurabibsetup, ",") << '}';
1110                         }
1111                 }
1112
1113                 else if (t.cs() == "hypersetup") {
1114                         vector<string> hypersetup =
1115                                 split_options(p.verbatim_item());
1116                         // add hypersetup to the hyperref package options
1117                         handle_hyperref(hypersetup);
1118                         if (!hypersetup.empty()) {
1119                                 h_preamble << "\\hypersetup{"
1120                                            << join(hypersetup, ",") << '}';
1121                         }
1122                 }
1123
1124                 else if (is_known(t.cs(), known_if_3arg_commands)) {
1125                         // prevent misparsing of \usepackage if it is used
1126                         // as an argument (see e.g. our own output of
1127                         // \@ifundefined above)
1128                         string const arg1 = p.verbatim_item();
1129                         string const arg2 = p.verbatim_item();
1130                         string const arg3 = p.verbatim_item();
1131                         if (!in_lyx_preamble) {
1132                                 h_preamble << t.asInput()
1133                                            << '{' << arg1 << '}'
1134                                            << '{' << arg2 << '}'
1135                                            << '{' << arg3 << '}';
1136                         }
1137                 }
1138
1139                 else if (is_known(t.cs(), known_if_commands)) {
1140                         // must not parse anything in conditional code, since
1141                         // LyX would output the parsed contents unconditionally
1142                         if (!in_lyx_preamble)
1143                                 h_preamble << t.asInput();
1144                         handle_if(p, in_lyx_preamble);
1145                 }
1146
1147                 else if (!t.cs().empty() && !in_lyx_preamble)
1148                         h_preamble << '\\' << t.cs();
1149         }
1150
1151         // remove the whitespace
1152         p.skip_spaces();
1153
1154         // Force textclass if the user wanted it
1155         if (!forceclass.empty())
1156                 h_textclass = forceclass;
1157         if (noweb_mode && !prefixIs(h_textclass, "literate-"))
1158                 h_textclass.insert(0, "literate-");
1159         tc.setName(h_textclass);
1160         if (!tc.load()) {
1161                 cerr << "Error: Could not read layout file for textclass \"" << h_textclass << "\"." << endl;
1162                 exit(EXIT_FAILURE);
1163         }
1164         if (h_papersides.empty()) {
1165                 ostringstream ss;
1166                 ss << tc.sides();
1167                 h_papersides = ss.str();
1168         }
1169         end_preamble(os, tc);
1170 }
1171
1172
1173 /// translates a babel language name to a LyX language name
1174 string babel2lyx(string const & language)
1175 {
1176         char const * const * where = is_known(language, known_languages);
1177         if (where)
1178                 return known_coded_languages[where - known_languages];
1179         return language;
1180 }
1181
1182 // }])
1183
1184
1185 } // namespace lyx