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 string h_inputencoding = "auto";
  52 string h_paragraph_separation    = "indent";
  53
  54 namespace {
  55
  56 //add this to known_languages when updating to lyxformat 266:
  57 // "armenian" (needs special handling since not supported by standard babel)
  58 //add these to known_languages when updating to lyxformat 268:
  59 //"chinese-simplified", "chinese-traditional", "japanese", "korean"
  60 // Both changes require first that support for non-babel languages (CJK,
  61 // armtex) is added.
  62 // add turkmen for lyxformat 383
  63 /**
  64  * known babel language names (including synonyms)
  65  * not in standard babel: arabic, arabtex, armenian, belarusian, serbian-latin, thai
  66  * not yet supported by LyX: kurmanji
  67  * please keep this in sync with known_coded_languages line by line!
  68  */
  69 const char * const known_languages[] = {"acadian", "afrikaans", "albanian",
  70 "american", "arabic", "arabtex", "austrian", "bahasa", "bahasai", "bahasam",
  71 "basque", "belarusian", "brazil", "brazilian", "breton", "british", "bulgarian",
  72 "canadian", "canadien", "catalan", "croatian", "czech", "danish", "dutch",
  73 "english", "esperanto", "estonian", "farsi", "finnish", "francais", "french",
  74 "frenchb", "frenchle", "frenchpro", "galician", "german", "germanb", "greek",
  75 "hebrew", "hungarian", "icelandic", "indon", "indonesian", "interlingua",
  76 "irish", "italian", "kazakh", "latin", "latvian", "lithuanian", "lowersorbian",
  77 "lsorbian", "magyar", "malay", "meyalu", "mongolian", "naustrian", "newzealand",
  78 "ngerman", "ngermanb", "norsk", "nynorsk", "polutonikogreek", "polish",
  79 "portuges", "portuguese", "romanian", "russian", "russianb", "samin",
  80 "scottish", "serbian", "serbian-latin", "slovak", "slovene", "spanish",
  81 "swedish", "thai", "turkish", "turkmen", "ukraineb", "ukrainian",
  82 "uppersorbian", "UKenglish", "USenglish", "usorbian", "vietnam", "welsh",
  83 0};
  84
  85 /**
  86  * the same as known_languages with .lyx names
  87  * please keep this in sync with known_languages line by line!
  88  */
  89 const char * const known_coded_languages[] = {"french", "afrikaans", "albanian",
  90 "american", "arabic_arabi", "arabic_arabtex", "austrian", "bahasa", "bahasa", "bahasam",
  91 "basque", "belarusian", "brazilian", "brazilian", "breton", "british", "bulgarian",
  92 "canadian", "canadien", "catalan", "croatian", "czech", "danish", "dutch",
  93 "english", "esperanto", "estonian", "farsi", "finnish", "french", "french",
  94 "french", "french", "french", "galician", "german", "german", "greek",
  95 "hebrew", "magyar", "icelandic", "bahasa", "bahasa", "interlingua",
  96 "irish", "italian", "kazakh", "latin", "latvian", "lithuanian", "lowersorbian",
  97 "lowersorbian", "magyar", "bahasam", "bahasam", "mongolian", "naustrian", "english",
  98 "ngerman", "ngerman", "norsk", "nynorsk", "polutonikogreek", "polish",
  99 "portuguese", "portuguese", "romanian", "russian", "russian", "samin",
 100 "scottish", "serbian", "serbian-latin", "slovak", "slovene", "spanish",
 101 "swedish", "thai", "turkish", "turkmen", "ukrainian", "ukrainian",
 102 "uppersorbian", "uppersorbian", "english", "english", "vietnamese", "welsh",
 103 0};
 104
 105 /// languages with english quotes (.lyx names)
 106 const char * const known_english_quotes_languages[] = {"american", "bahasa",
 107 "bahasam", "brazilian", "canadian", "chinese-simplified", "english",
 108 "esperanto", "hebrew", "irish", "korean", "portuguese", "scottish", "thai", 0};
 109
 110 /// languages with french quotes (.lyx names)
 111 const char * const known_french_quotes_languages[] = {"albanian",
 112 "arabic_arabi", "arabic_arabtex", "basque", "canadien", "catalan", "french",
 113 "galician", "greek", "italian", "norsk", "nynorsk", "polutonikogreek",
 114 "russian", "spanish", "spanish-mexico", "turkish", "turkmen", "ukrainian",
 115 "vietnamese", 0};
 116
 117 /// languages with german quotes (.lyx names)
 118 const char * const known_german_quotes_languages[] = {"austrian", "bulgarian",
 119 "czech", "german", "icelandic", "lithuanian", "lowersorbian", "naustrian",
 120 "ngerman", "serbian", "serbian-latin", "slovak", "slovene", "uppersorbian", 0};
 121
 122 /// languages with polish quotes (.lyx names)
 123 const char * const known_polish_quotes_languages[] = {"afrikaans", "croatian",
 124 "dutch", "estonian", "magyar", "polish", "romanian", 0};
 125
 126 /// languages with swedish quotes (.lyx names)
 127 const char * const known_swedish_quotes_languages[] = {"finnish",
 128 "swedish", 0};
 129
 130 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
 131
 132 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
 133 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
 134
 135 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
 136 "helvet", "lmss", 0};
 137
 138 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
 139 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
 140 "newcent", 0};
 141
 142 const char * const known_paper_sizes[] = { "a3paper", "b3paper", "a4paper",
 143 "b4paper", "a5paper", "b5paper", "executivepaper", "legalpaper",
 144 "letterpaper", 0};
 145
 146 const char * const known_class_paper_sizes[] = { "a4paper", "a5paper",
 147 "executivepaper", "legalpaper", "letterpaper", 0};
 148
 149 const char * const known_paper_margins[] = { "lmargin", "tmargin", "rmargin",
 150 "bmargin", "headheight", "headsep", "footskip", "columnsep", 0};
 151
 152 const char * const known_coded_paper_margins[] = { "leftmargin", "topmargin",
 153 "rightmargin", "bottommargin", "headheight", "headsep", "footskip",
 154 "columnsep", 0};
 155
 156 /// commands that can start an \if...\else...\endif sequence
 157 const char * const known_if_commands[] = {"if", "ifarydshln", "ifbraket",
 158 "ifcancel", "ifcolortbl", "ifeurosym", "ifmarginnote", "ifmmode", "ifpdf",
 159 "ifsidecap", "ifupgreek", 0};
 160
 161 /// conditional commands with three arguments like \@ifundefined{}{}{}
 162 const char * const known_if_3arg_commands[] = {"@ifundefined", "IfFileExists",
 163 0};
 164
 165 // default settings
 166 ostringstream h_preamble;
 167 string h_textclass               = "article";
 168 string h_use_default_options     = "false";
 169 string h_options;
 170 string h_language                = "english";
 171 string h_font_roman              = "default";
 172 string h_font_sans               = "default";
 173 string h_font_typewriter         = "default";
 174 string h_font_default_family     = "default";
 175 string h_font_sc                 = "false";
 176 string h_font_osf                = "false";
 177 string h_font_sf_scale           = "100";
 178 string h_font_tt_scale           = "100";
 179 string h_graphics                = "default";
 180 string h_float_placement;
 181 string h_paperfontsize           = "default";
 182 string h_spacing                 = "single";
 183 string h_use_hyperref            = "0";
 184 string h_pdf_title;
 185 string h_pdf_author;
 186 string h_pdf_subject;
 187 string h_pdf_keywords;
 188 string h_pdf_bookmarks           = "1";
 189 string h_pdf_bookmarksnumbered   = "0";
 190 string h_pdf_bookmarksopen       = "0";
 191 string h_pdf_bookmarksopenlevel  = "1";
 192 string h_pdf_breaklinks          = "0";
 193 string h_pdf_pdfborder           = "0";
 194 string h_pdf_colorlinks          = "0";
 195 string h_pdf_backref             = "section";
 196 string h_pdf_pdfusetitle         = "1";
 197 string h_pdf_pagemode;
 198 string h_pdf_quoted_options;
 199 string h_papersize               = "default";
 200 string h_use_geometry            = "false";
 201 string h_use_amsmath             = "1";
 202 string h_use_esint               = "1";
 203 string h_use_mhchem              = "0";
 204 string h_use_mathdots            = "0";
 205 string h_cite_engine             = "basic";
 206 string h_use_bibtopic            = "false";
 207 string h_paperorientation        = "portrait";
 208 string h_suppress_date           = "false";
 209 string h_use_refstyle            = "0";
 210 string h_notefontcolor;
 211 string h_secnumdepth             = "3";
 212 string h_tocdepth                = "3";
 213 string h_defskip                 = "medskip";
 214 string h_paragraph_indentation   = "default";
 215 string h_quotes_language         = "english";
 216 string h_papercolumns            = "1";
 217 string h_papersides;
 218 string h_paperpagestyle          = "default";
 219 string h_listings_params;
 220 string h_tracking_changes        = "false";
 221 string h_output_changes          = "false";
 222 string h_html_math_output        = "0";
 223 string h_html_be_strict          = "false";
 224 string h_margins;
 225
 226
 227 // returns true if at least one of the options in what has been found
 228 bool handle_opt(vector<string> & opts, char const * const * what, string & target)
 229 {
 230         if (opts.empty())
 231                 return false;
 232
 233         bool found = false;
 234         // the last language option is the document language (for babel and LyX)
 235         // the last size option is the document font size
 236         vector<string>::iterator it;
 237         vector<string>::iterator position = opts.begin();
 238         for (; *what; ++what) {
 239                 it = find(opts.begin(), opts.end(), *what);
 240                 if (it != opts.end()) {
 241                         if (it >= position) {
 242                                 found = true;
 243                                 target = *what;
 244                                 position = it;
 245                         }
 246                 }
 247         }
 248         return found;
 249 }
 250
 251
 252 void delete_opt(vector<string> & opts, char const * const * what)
 253 {
 254         if (opts.empty())
 255                 return;
 256
 257         // remove found options from the list
 258         // do this after handle_opt to avoid potential memory leaks
 259         vector<string>::iterator it;
 260         for (; *what; ++what) {
 261                 it = find(opts.begin(), opts.end(), *what);
 262                 if (it != opts.end())
 263                         opts.erase(it);
 264         }
 265 }
 266
 267
 268 /*!
 269  * Split a package options string (keyval format) into a vector.
 270  * Example input:
 271  *   authorformat=smallcaps,
 272  *   commabeforerest,
 273  *   titleformat=colonsep,
 274  *   bibformat={tabular,ibidem,numbered}
 275  */
 276 vector<string> split_options(string const & input)
 277 {
 278         vector<string> options;
 279         string option;
 280         Parser p(input);
 281         while (p.good()) {
 282                 Token const & t = p.get_token();
 283                 if (t.asInput() == ",") {
 284                         options.push_back(trimSpaceAndEol(option));
 285                         option.erase();
 286                 } else if (t.asInput() == "=") {
 287                         option += '=';
 288                         p.skip_spaces(true);
 289                         if (p.next_token().asInput() == "{")
 290                                 option += '{' + p.getArg('{', '}') + '}';
 291                 } else if (t.cat() != catSpace)
 292                         option += t.asInput();
 293         }
 294
 295         if (!option.empty())
 296                 options.push_back(trimSpaceAndEol(option));
 297
 298         return options;
 299 }
 300
 301
 302 /*!
 303  * Retrieve a keyval option "name={value with=sign}" named \p name from
 304  * \p options and return the value.
 305  * The found option is also removed from \p options.
 306  */
 307 string process_keyval_opt(vector<string> & options, string name)
 308 {
 309         for (size_t i = 0; i < options.size(); ++i) {
 310                 vector<string> option;
 311                 split(options[i], option, '=');
 312                 if (option.size() < 2)
 313                         continue;
 314                 if (option[0] == name) {
 315                         options.erase(options.begin() + i);
 316                         option.erase(option.begin());
 317                         return join(option, "=");
 318                 }
 319         }
 320         return "";
 321 }
 322
 323
 324 /*!
 325  * Add package \p name with options \p options to used_packages.
 326  * Remove options from \p options that we don't want to output.
 327  */
 328 void add_package(string const & name, vector<string> & options)
 329 {
 330         // every package inherits the global options
 331         if (used_packages.find(name) == used_packages.end())
 332                 used_packages[name] = split_options(h_options);
 333
 334         vector<string> & v = used_packages[name];
 335         v.insert(v.end(), options.begin(), options.end());
 336         if (name == "jurabib") {
 337                 // Don't output the order argument (see the cite command
 338                 // handling code in text.cpp).
 339                 vector<string>::iterator end =
 340                         remove(options.begin(), options.end(), "natbiborder");
 341                 end = remove(options.begin(), end, "jurabiborder");
 342                 options.erase(end, options.end());
 343         }
 344 }
 345
 346
 347 // Given is a string like "scaled=0.9", return 0.9 * 100
 348 string const scale_as_percentage(string const & scale)
 349 {
 350         string::size_type pos = scale.find('=');
 351         if (pos != string::npos) {
 352                 string value = scale.substr(pos + 1);
 353                 if (isStrDbl(value))
 354                         return convert<string>(100 * convert<double>(value));
 355         }
 356         // If the input string didn't match our expectations.
 357         // return the default value "100"
 358         return "100";
 359 }
 360
 361
 362 string remove_braces(string const & value)
 363 {
 364         if (value.empty())
 365                 return value;
 366         if (value[0] == '{' && value[value.length()-1] == '}')
 367                 return value.substr(1, value.length()-2);
 368         return value;
 369 }
 370
 371
 372 void handle_hyperref(vector<string> & options)
 373 {
 374         // FIXME swallow inputencoding changes that might surround the
 375         //       hyperref setup if it was written by LyX
 376         h_use_hyperref = "1";
 377         // swallow "unicode=true", since LyX does always write that
 378         vector<string>::iterator it =
 379                 find(options.begin(), options.end(), "unicode=true");
 380         if (it != options.end())
 381                 options.erase(it);
 382         it = find(options.begin(), options.end(), "pdfusetitle");
 383         if (it != options.end()) {
 384                 h_pdf_pdfusetitle = "1";
 385                 options.erase(it);
 386         }
 387         string bookmarks = process_keyval_opt(options, "bookmarks");
 388         if (bookmarks == "true")
 389                 h_pdf_bookmarks = "1";
 390         else if (bookmarks == "false")
 391                 h_pdf_bookmarks = "0";
 392         if (h_pdf_bookmarks == "1") {
 393                 string bookmarksnumbered =
 394                         process_keyval_opt(options, "bookmarksnumbered");
 395                 if (bookmarksnumbered == "true")
 396                         h_pdf_bookmarksnumbered = "1";
 397                 else if (bookmarksnumbered == "false")
 398                         h_pdf_bookmarksnumbered = "0";
 399                 string bookmarksopen =
 400                         process_keyval_opt(options, "bookmarksopen");
 401                 if (bookmarksopen == "true")
 402                         h_pdf_bookmarksopen = "1";
 403                 else if (bookmarksopen == "false")
 404                         h_pdf_bookmarksopen = "0";
 405                 if (h_pdf_bookmarksopen == "1") {
 406                         string bookmarksopenlevel =
 407                                 process_keyval_opt(options, "bookmarksopenlevel");
 408                         if (!bookmarksopenlevel.empty())
 409                                 h_pdf_bookmarksopenlevel = bookmarksopenlevel;
 410                 }
 411         }
 412         string breaklinks = process_keyval_opt(options, "breaklinks");
 413         if (breaklinks == "true")
 414                 h_pdf_breaklinks = "1";
 415         else if (breaklinks == "false")
 416                 h_pdf_breaklinks = "0";
 417         string pdfborder = process_keyval_opt(options, "pdfborder");
 418         if (pdfborder == "{0 0 0}")
 419                 h_pdf_pdfborder = "1";
 420         else if (pdfborder == "{0 0 1}")
 421                 h_pdf_pdfborder = "0";
 422         string backref = process_keyval_opt(options, "backref");
 423         if (!backref.empty())
 424                 h_pdf_backref = backref;
 425         string colorlinks = process_keyval_opt(options, "colorlinks");
 426         if (colorlinks == "true")
 427                 h_pdf_colorlinks = "1";
 428         else if (colorlinks == "false")
 429                 h_pdf_colorlinks = "0";
 430         string pdfpagemode = process_keyval_opt(options, "pdfpagemode");
 431         if (!pdfpagemode.empty())
 432                 h_pdf_pagemode = pdfpagemode;
 433         string pdftitle = process_keyval_opt(options, "pdftitle");
 434         if (!pdftitle.empty()) {
 435                 h_pdf_title = remove_braces(pdftitle);
 436         }
 437         string pdfauthor = process_keyval_opt(options, "pdfauthor");
 438         if (!pdfauthor.empty()) {
 439                 h_pdf_author = remove_braces(pdfauthor);
 440         }
 441         string pdfsubject = process_keyval_opt(options, "pdfsubject");
 442         if (!pdfsubject.empty())
 443                 h_pdf_subject = remove_braces(pdfsubject);
 444         string pdfkeywords = process_keyval_opt(options, "pdfkeywords");
 445         if (!pdfkeywords.empty())
 446                 h_pdf_keywords = remove_braces(pdfkeywords);
 447         if (!options.empty()) {
 448                 if (!h_pdf_quoted_options.empty())
 449                         h_pdf_quoted_options += ',';
 450                 h_pdf_quoted_options += join(options, ",");
 451                 options.clear();
 452         }
 453 }
 454
 455
 456 void handle_package(Parser &p, string const & name, string const & opts,
 457                     bool in_lyx_preamble)
 458 {
 459         vector<string> options = split_options(opts);
 460         add_package(name, options);
 461         string scale;
 462
 463         // roman fonts
 464         if (is_known(name, known_roman_fonts)) {
 465                 h_font_roman = name;
 466                 p.skip_spaces();
 467         }
 468
 469         if (name == "fourier") {
 470                 h_font_roman = "utopia";
 471                 // when font uses real small capitals
 472                 if (opts == "expert")
 473                         h_font_sc = "true";
 474         }
 475
 476         if (name == "mathpazo")
 477                 h_font_roman = "palatino";
 478
 479         if (name == "mathptmx")
 480                 h_font_roman = "times";
 481
 482         // sansserif fonts
 483         if (is_known(name, known_sans_fonts)) {
 484                 h_font_sans = name;
 485                 if (!opts.empty()) {
 486                         scale = opts;
 487                         h_font_sf_scale = scale_as_percentage(scale);
 488                 }
 489         }
 490
 491         // typewriter fonts
 492         if (is_known(name, known_typewriter_fonts)) {
 493                 h_font_typewriter = name;
 494                 if (!opts.empty()) {
 495                         scale = opts;
 496                         h_font_tt_scale = scale_as_percentage(scale);
 497                 }
 498         }
 499
 500         // font uses old-style figure
 501         if (name == "eco")
 502                 h_font_osf = "true";
 503
 504         else if (name == "amsmath" || name == "amssymb")
 505                 h_use_amsmath = "2";
 506
 507         else if (name == "esint")
 508                 h_use_esint = "2";
 509
 510         else if (name == "mhchem")
 511                 h_use_mhchem = "2";
 512
 513         else if (name == "mathdots")
 514                 h_use_mathdots = "2";
 515
 516         else if (name == "babel" && !opts.empty()) {
 517                 // check if more than one option was used - used later for inputenc
 518                 // in case inputenc is parsed before babel, set the encoding to auto
 519                 if (options.begin() != options.end() - 1) {
 520                         one_language = false;
 521                         h_inputencoding = "auto";
 522                 }
 523                 // babel takes the last language of the option of its \usepackage
 524                 // call as document language. If there is no such language option, the
 525                 // last language in the documentclass options is used.
 526                 handle_opt(options, known_languages, h_language);
 527                 delete_opt(options, known_languages);
 528         }
 529
 530         else if (name == "fontenc")
 531                  ;// ignore this
 532
 533         else if (name == "inputenc" || name == "luainputenc") {
 534                 // h_inputencoding is only set when there is not more than one
 535                 // inputenc option because otherwise h_inputencoding must be
 536                 // set to "auto" (the default encoding of the document language)
 537                 // Therefore check for the "," character.
 538                 // It is also only set when there is not more then one babel
 539                 // language option but this is handled in the routine for babel.
 540                 if (opts.find(",") == string::npos && one_language == true)
 541                         h_inputencoding = opts;
 542                 if (!options.empty())
 543                         p.setEncoding(options.back());
 544                 options.clear();
 545         }
 546
 547         else if (name == "makeidx")
 548                 ; // ignore this
 549
 550         else if (name == "prettyref")
 551                 ; // ignore this
 552
 553         else if (name == "varioref")
 554                 ; // ignore this
 555
 556         else if (name == "verbatim")
 557                 ; // ignore this
 558
 559         else if (name == "nomencl")
 560                 ; // ignore this
 561
 562         else if (name == "textcomp")
 563                 ; // ignore this
 564
 565         else if (name == "url")
 566                 ; // ignore this
 567
 568         else if (name == "subscript")
 569                 ; // ignore this
 570
 571         else if (name == "color") {
 572                 // with the following command this package is only loaded when needed for
 573                 // undefined colors, since we only support the predefined colors
 574                 h_preamble << "\\@ifundefined{definecolor}\n {\\usepackage{color}}{}\n";
 575         }
 576
 577         else if (name == "graphicx")
 578                 ; // ignore this
 579
 580         else if (name == "setspace")
 581                 ; // ignore this
 582
 583         else if (name == "geometry")
 584                 ; // Ignore this, the geometry settings are made by the \geometry
 585                   // command. This command is handled below.
 586
 587         else if (is_known(name, known_languages))
 588                 h_language = name;
 589
 590         else if (name == "natbib") {
 591                 h_cite_engine = "natbib_authoryear";
 592                 vector<string>::iterator it =
 593                         find(options.begin(), options.end(), "authoryear");
 594                 if (it != options.end())
 595                         options.erase(it);
 596                 else {
 597                         it = find(options.begin(), options.end(), "numbers");
 598                         if (it != options.end()) {
 599                                 h_cite_engine = "natbib_numerical";
 600                                 options.erase(it);
 601                         }
 602                 }
 603         }
 604
 605         else if (name == "jurabib")
 606                 h_cite_engine = "jurabib";
 607
 608         else if (name == "hyperref")
 609                 handle_hyperref(options);
 610
 611         else if (!in_lyx_preamble) {
 612                 if (options.empty())
 613                         h_preamble << "\\usepackage{" << name << "}";
 614                 else {
 615                         h_preamble << "\\usepackage[" << opts << "]{"
 616                                    << name << "}";
 617                         options.clear();
 618                 }
 619         }
 620
 621         // We need to do something with the options...
 622         if (!options.empty())
 623                 cerr << "Ignoring options '" << join(options, ",")
 624                      << "' of package " << name << '.' << endl;
 625
 626         // remove the whitespace
 627         p.skip_spaces();
 628 }
 629
 630
 631 void handle_if(Parser & p, bool in_lyx_preamble)
 632 {
 633         while (p.good()) {
 634                 Token t = p.get_token();
 635                 if (t.cat() == catEscape &&
 636                     is_known(t.cs(), known_if_commands))
 637                         handle_if(p, in_lyx_preamble);
 638                 else {
 639                         if (!in_lyx_preamble)
 640                                 h_preamble << t.asInput();
 641                         if (t.cat() == catEscape && t.cs() == "fi")
 642                                 return;
 643                 }
 644         }
 645 }
 646
 647
 648 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 649 {
 650         // translate from babel to LyX names
 651         h_language = babel2lyx(h_language);
 652
 653         // set the quote language
 654         // LyX only knows the following quotes languages:
 655         // english, swedish, german, polish, french and danish
 656         // (quotes for "japanese" and "chinese-traditional" are missing because
 657         //  they wouldn't be useful: http://www.lyx.org/trac/ticket/6383)
 658         // conversion list taken from
 659         // http://en.wikipedia.org/wiki/Quotation_mark,_non-English_usage
 660         // (quotes for kazakh and interlingua are unknown)
 661         // danish
 662         if (h_language == "danish")
 663                 h_quotes_language = "danish";
 664         // french
 665         else if (is_known(h_language, known_french_quotes_languages))
 666                 h_quotes_language = "french";
 667         // german
 668         else if (is_known(h_language, known_german_quotes_languages))
 669                 h_quotes_language = "german";
 670         // polish
 671         else if (is_known(h_language, known_polish_quotes_languages))
 672                 h_quotes_language = "polish";
 673         // swedish
 674         else if (is_known(h_language, known_swedish_quotes_languages))
 675                 h_quotes_language = "swedish";
 676         //english
 677         else if (is_known(h_language, known_english_quotes_languages))
 678                 h_quotes_language = "english";
 679
 680         // output the LyX file settings
 681         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 682            << "\\lyxformat " << LYX_FORMAT << '\n'
 683            << "\\begin_document\n"
 684            << "\\begin_header\n"
 685            << "\\textclass " << h_textclass << "\n";
 686         if (!h_preamble.str().empty())
 687                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 688         if (!h_options.empty())
 689                 os << "\\options " << h_options << "\n";
 690         os << "\\use_default_options " << h_use_default_options << "\n"
 691            << modules_placeholder
 692            << "\\language " << h_language << "\n"
 693            << "\\inputencoding " << h_inputencoding << "\n"
 694            << "\\font_roman " << h_font_roman << "\n"
 695            << "\\font_sans " << h_font_sans << "\n"
 696            << "\\font_typewriter " << h_font_typewriter << "\n"
 697            << "\\font_default_family " << h_font_default_family << "\n"
 698            << "\\font_sc " << h_font_sc << "\n"
 699            << "\\font_osf " << h_font_osf << "\n"
 700            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 701            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 702            << "\\graphics " << h_graphics << "\n";
 703         if (!h_float_placement.empty())
 704                 os << "\\float_placement " << h_float_placement << "\n";
 705         os << "\\paperfontsize " << h_paperfontsize << "\n"
 706            << "\\spacing " << h_spacing << "\n"
 707            << "\\use_hyperref " << h_use_hyperref << '\n';
 708         if (h_use_hyperref == "1") {
 709                 if (!h_pdf_title.empty())
 710                         os << "\\pdf_title \"" << h_pdf_title << "\"\n";
 711                 if (!h_pdf_author.empty())
 712                         os << "\\pdf_author \"" << h_pdf_author << "\"\n";
 713                 if (!h_pdf_subject.empty())
 714                         os << "\\pdf_subject \"" << h_pdf_subject << "\"\n";
 715                 if (!h_pdf_keywords.empty())
 716                         os << "\\pdf_keywords \"" << h_pdf_keywords << "\"\n";
 717                 os << "\\pdf_bookmarks " << h_pdf_bookmarks << "\n"
 718                       "\\pdf_bookmarksnumbered " << h_pdf_bookmarksnumbered << "\n"
 719                       "\\pdf_bookmarksopen " << h_pdf_bookmarksopen << "\n"
 720                       "\\pdf_bookmarksopenlevel " << h_pdf_bookmarksopenlevel << "\n"
 721                       "\\pdf_breaklinks " << h_pdf_breaklinks << "\n"
 722                       "\\pdf_pdfborder " << h_pdf_pdfborder << "\n"
 723                       "\\pdf_colorlinks " << h_pdf_colorlinks << "\n"
 724                       "\\pdf_backref " << h_pdf_backref << "\n"
 725                       "\\pdf_pdfusetitle " << h_pdf_pdfusetitle << '\n';
 726                 if (!h_pdf_pagemode.empty())
 727                         os << "\\pdf_pagemode " << h_pdf_pagemode << '\n';
 728                 if (!h_pdf_quoted_options.empty())
 729                         os << "\\pdf_quoted_options \"" << h_pdf_quoted_options << "\"\n";
 730         }
 731         os << "\\papersize " << h_papersize << "\n"
 732            << "\\use_geometry " << h_use_geometry << "\n"
 733            << "\\use_amsmath " << h_use_amsmath << "\n"
 734            << "\\use_esint " << h_use_esint << "\n"
 735            << "\\use_mhchem " << h_use_mhchem << "\n"
 736            << "\\use_mathdots " << h_use_mathdots << "\n"
 737            << "\\cite_engine " << h_cite_engine << "\n"
 738            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 739            << "\\paperorientation " << h_paperorientation << '\n'
 740            << "\\suppress_date " << h_suppress_date << '\n'
 741            << "\\use_refstyle " << h_use_refstyle << '\n';
 742         if (!h_notefontcolor.empty())
 743                 os << "\\notefontcolor " << h_notefontcolor << '\n';
 744         os << h_margins
 745            << "\\secnumdepth " << h_secnumdepth << "\n"
 746            << "\\tocdepth " << h_tocdepth << "\n"
 747            << "\\paragraph_separation " << h_paragraph_separation << "\n";
 748         if (h_paragraph_separation == "skip")
 749                 os << "\\defskip " << h_defskip << "\n";
 750         else
 751                 os << "\\paragraph_indentation " << h_paragraph_indentation << "\n";
 752         os << "\\quotes_language " << h_quotes_language << "\n"
 753            << "\\papercolumns " << h_papercolumns << "\n"
 754            << "\\papersides " << h_papersides << "\n"
 755            << "\\paperpagestyle " << h_paperpagestyle << "\n";
 756         if (!h_listings_params.empty())
 757                 os << "\\listings_params " << h_listings_params << "\n";
 758         os << "\\tracking_changes " << h_tracking_changes << "\n"
 759            << "\\output_changes " << h_output_changes << "\n"
 760            << "\\html_math_output " << h_html_math_output << "\n"
 761            << "\\html_be_strict " << h_html_be_strict << "\n"
 762            << "\\end_header\n\n"
 763            << "\\begin_body\n";
 764         // clear preamble for subdocuments
 765         h_preamble.str("");
 766 }
 767
 768 } // anonymous namespace
 769
 770
 771 void parse_preamble(Parser & p, ostream & os,
 772         string const & forceclass, TeX2LyXDocClass & tc)
 773 {
 774         // initialize fixed types
 775         special_columns['D'] = 3;
 776         bool is_full_document = false;
 777         bool is_lyx_file = false;
 778         bool in_lyx_preamble = false;
 779
 780         // determine whether this is a full document or a fragment for inclusion
 781         while (p.good()) {
 782                 Token const & t = p.get_token();
 783
 784                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 785                         is_full_document = true;
 786                         break;
 787                 }
 788         }
 789         p.reset();
 790
 791         while (is_full_document && p.good()) {
 792                 Token const & t = p.get_token();
 793
 794 #ifdef FILEDEBUG
 795                 cerr << "t: " << t << "\n";
 796 #endif
 797
 798                 //
 799                 // cat codes
 800                 //
 801                 if (!in_lyx_preamble &&
 802                     (t.cat() == catLetter ||
 803                      t.cat() == catSuper ||
 804                      t.cat() == catSub ||
 805                      t.cat() == catOther ||
 806                      t.cat() == catMath ||
 807                      t.cat() == catActive ||
 808                      t.cat() == catBegin ||
 809                      t.cat() == catEnd ||
 810                      t.cat() == catAlign ||
 811                      t.cat() == catParameter))
 812                         h_preamble << t.cs();
 813
 814                 else if (!in_lyx_preamble &&
 815                          (t.cat() == catSpace || t.cat() == catNewline))
 816                         h_preamble << t.asInput();
 817
 818                 else if (t.cat() == catComment) {
 819                         static regex const islyxfile("%% LyX .* created this file");
 820                         static regex const usercommands("User specified LaTeX commands");
 821
 822                         string const comment = t.asInput();
 823
 824                         // magically switch encoding default if it looks like XeLaTeX
 825                         static string const magicXeLaTeX =
 826                                 "% This document must be compiled with XeLaTeX ";
 827                         if (comment.size() > magicXeLaTeX.size()
 828                                   && comment.substr(0, magicXeLaTeX.size()) == magicXeLaTeX
 829                                   && h_inputencoding == "auto") {
 830                                 cerr << "XeLaTeX comment found, switching to UTF8\n";
 831                                 h_inputencoding = "utf8";
 832                         }
 833                         smatch sub;
 834                         if (regex_search(comment, sub, islyxfile)) {
 835                                 is_lyx_file = true;
 836                                 in_lyx_preamble = true;
 837                         } else if (is_lyx_file
 838                                    && regex_search(comment, sub, usercommands))
 839                                 in_lyx_preamble = false;
 840                         else if (!in_lyx_preamble)
 841                                 h_preamble << t.asInput();
 842                 }
 843
 844                 else if (t.cs() == "pagestyle")
 845                         h_paperpagestyle = p.verbatim_item();
 846
 847                 else if (t.cs() == "makeatletter") {
 848                         // LyX takes care of this
 849                         p.setCatCode('@', catLetter);
 850                 }
 851
 852                 else if (t.cs() == "makeatother") {
 853                         // LyX takes care of this
 854                         p.setCatCode('@', catOther);
 855                 }
 856
 857                 else if (t.cs() == "newcommand" || t.cs() == "newcommandx"
 858                       || t.cs() == "renewcommand" || t.cs() == "renewcommandx"
 859                       || t.cs() == "providecommand" || t.cs() == "providecommandx"
 860                                 || t.cs() == "DeclareRobustCommand"
 861                       || t.cs() == "DeclareRobustCommandx"
 862                                 || t.cs() == "ProvideTextCommandDefault"
 863                                 || t.cs() == "DeclareMathAccent") {
 864                         bool star = false;
 865                         if (p.next_token().character() == '*') {
 866                                 p.get_token();
 867                                 star = true;
 868                         }
 869                         string const name = p.verbatim_item();
 870                         string const opt1 = p.getFullOpt();
 871                         string const opt2 = p.getFullOpt();
 872                         string const body = p.verbatim_item();
 873                         // font settings
 874                         if (name == "\\rmdefault")
 875                                 if (is_known(body, known_roman_fonts))
 876                                         h_font_roman = body;
 877                         if (name == "\\sfdefault")
 878                                 if (is_known(body, known_sans_fonts))
 879                                         h_font_sans = body;
 880                         if (name == "\\ttdefault")
 881                                 if (is_known(body, known_typewriter_fonts))
 882                                         h_font_typewriter = body;
 883                         if (name == "\\familydefault") {
 884                                 string family = body;
 885                                 // remove leading "\"
 886                                 h_font_default_family = family.erase(0,1);
 887                         }
 888
 889                         // Add the command to the known commands
 890                         add_known_command(name, opt1, !opt2.empty(), from_utf8(body));
 891
 892                         // only non-lyxspecific stuff
 893                         if (!in_lyx_preamble) {
 894                                 ostringstream ss;
 895                                 ss << '\\' << t.cs();
 896                                 if (star)
 897                                         ss << '*';
 898                                 ss << '{' << name << '}' << opt1 << opt2
 899                                    << '{' << body << "}";
 900                                 h_preamble << ss.str();
 901 /*
 902                                 ostream & out = in_preamble ? h_preamble : os;
 903                                 out << "\\" << t.cs() << "{" << name << "}"
 904                                     << opts << "{" << body << "}";
 905 */
 906                         }
 907                 }
 908
 909                 else if (t.cs() == "documentclass") {
 910                         vector<string>::iterator it;
 911                         vector<string> opts = split_options(p.getArg('[', ']'));
 912                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 913                         delete_opt(opts, known_fontsizes);
 914                         // delete "pt" at the end
 915                         string::size_type i = h_paperfontsize.find("pt");
 916                         if (i != string::npos)
 917                                 h_paperfontsize.erase(i);
 918                         // The documentclass options are always parsed before the options
 919                         // of the babel call so that a language cannot overwrite the babel
 920                         // options.
 921                         handle_opt(opts, known_languages, h_language);
 922                         delete_opt(opts, known_languages);
 923
 924                         // paper orientation
 925                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 926                                 h_paperorientation = "landscape";
 927                                 opts.erase(it);
 928                         }
 929                         // paper sides
 930                         if ((it = find(opts.begin(), opts.end(), "oneside"))
 931                                  != opts.end()) {
 932                                 h_papersides = "1";
 933                                 opts.erase(it);
 934                         }
 935                         if ((it = find(opts.begin(), opts.end(), "twoside"))
 936                                  != opts.end()) {
 937                                 h_papersides = "2";
 938                                 opts.erase(it);
 939                         }
 940                         // paper columns
 941                         if ((it = find(opts.begin(), opts.end(), "onecolumn"))
 942                                  != opts.end()) {
 943                                 h_papercolumns = "1";
 944                                 opts.erase(it);
 945                         }
 946                         if ((it = find(opts.begin(), opts.end(), "twocolumn"))
 947                                  != opts.end()) {
 948                                 h_papercolumns = "2";
 949                                 opts.erase(it);
 950                         }
 951                         // paper sizes
 952                         // some size options are know to any document classes, other sizes
 953                         // are handled by the \geometry command of the geometry package
 954                         handle_opt(opts, known_class_paper_sizes, h_papersize);
 955                         delete_opt(opts, known_class_paper_sizes);
 956                         // the remaining options
 957                         h_options = join(opts, ",");
 958                         // FIXME This does not work for classes that have a
 959                         //       different name in LyX than in LaTeX
 960                         h_textclass = p.getArg('{', '}');
 961                 }
 962
 963                 else if (t.cs() == "usepackage") {
 964                         string const options = p.getArg('[', ']');
 965                         string const name = p.getArg('{', '}');
 966                         vector<string> vecnames;
 967                         split(name, vecnames, ',');
 968                         vector<string>::const_iterator it  = vecnames.begin();
 969                         vector<string>::const_iterator end = vecnames.end();
 970                         for (; it != end; ++it)
 971                                 handle_package(p, trimSpaceAndEol(*it), options,
 972                                                in_lyx_preamble);
 973                 }
 974
 975                 else if (t.cs() == "inputencoding") {
 976                         string const encoding = p.getArg('{','}');
 977                         h_inputencoding = encoding;
 978                         p.setEncoding(encoding);
 979                 }
 980
 981                 else if (t.cs() == "newenvironment") {
 982                         string const name = p.getArg('{', '}');
 983                         string const opt1 = p.getFullOpt();
 984                         string const opt2 = p.getFullOpt();
 985                         string const beg = p.verbatim_item();
 986                         string const end = p.verbatim_item();
 987                         if (!in_lyx_preamble) {
 988                                 h_preamble << "\\newenvironment{" << name
 989                                            << '}' << opt1 << opt2 << '{'
 990                                            << beg << "}{" << end << '}';
 991                         }
 992                         add_known_environment(name, opt1, !opt2.empty(),
 993                                               from_utf8(beg), from_utf8(end));
 994
 995                 }
 996
 997                 else if (t.cs() == "def") {
 998                         string name = p.get_token().cs();
 999                         while (p.next_token().cat() != catBegin)
1000                                 name += p.get_token().cs();
1001                         if (!in_lyx_preamble)
1002                                 h_preamble << "\\def\\" << name << '{'
1003                                            << p.verbatim_item() << "}";
1004                 }
1005
1006                 else if (t.cs() == "newcolumntype") {
1007                         string const name = p.getArg('{', '}');
1008                         trimSpaceAndEol(name);
1009                         int nargs = 0;
1010                         string opts = p.getOpt();
1011                         if (!opts.empty()) {
1012                                 istringstream is(string(opts, 1));
1013                                 is >> nargs;
1014                         }
1015                         special_columns[name[0]] = nargs;
1016                         h_preamble << "\\newcolumntype{" << name << "}";
1017                         if (nargs)
1018                                 h_preamble << "[" << nargs << "]";
1019                         h_preamble << "{" << p.verbatim_item() << "}";
1020                 }
1021
1022                 else if (t.cs() == "setcounter") {
1023                         string const name = p.getArg('{', '}');
1024                         string const content = p.getArg('{', '}');
1025                         if (name == "secnumdepth")
1026                                 h_secnumdepth = content;
1027                         else if (name == "tocdepth")
1028                                 h_tocdepth = content;
1029                         else
1030                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
1031                 }
1032
1033                 else if (t.cs() == "setlength") {
1034                         string const name = p.verbatim_item();
1035                         string const content = p.verbatim_item();
1036                         // the paragraphs are only not indented when \parindent is set to zero
1037                         if (name == "\\parindent" && content != "") {
1038                                 if (content[0] == '0')
1039                                         h_paragraph_separation = "skip";
1040                                 else
1041                                         h_paragraph_indentation = translate_len(content);
1042                         } else if (name == "\\parskip") {
1043                                 if (content == "\\smallskipamount")
1044                                         h_defskip = "smallskip";
1045                                 else if (content == "\\medskipamount")
1046                                         h_defskip = "medskip";
1047                                 else if (content == "\\bigskipamount")
1048                                         h_defskip = "bigskip";
1049                                 else
1050                                         h_defskip = content;
1051                         } else
1052                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
1053                 }
1054
1055                 else if (t.cs() == "onehalfspacing")
1056                         h_spacing = "onehalf";
1057
1058                 else if (t.cs() == "doublespacing")
1059                         h_spacing = "double";
1060
1061                 else if (t.cs() == "setstretch")
1062                         h_spacing = "other " + p.verbatim_item();
1063
1064                 else if (t.cs() == "begin") {
1065                         string const name = p.getArg('{', '}');
1066                         if (name == "document")
1067                                 break;
1068                         h_preamble << "\\begin{" << name << "}";
1069                 }
1070
1071                 else if (t.cs() == "geometry") {
1072                         h_use_geometry = "true";
1073                         vector<string> opts = split_options(p.getArg('{', '}'));
1074                         vector<string>::iterator it;
1075                         // paper orientation
1076                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
1077                                 h_paperorientation = "landscape";
1078                                 opts.erase(it);
1079                         }
1080                         // paper size
1081                         handle_opt(opts, known_paper_sizes, h_papersize);
1082                         delete_opt(opts, known_paper_sizes);
1083                         // page margins
1084                         char const * const * margin = known_paper_margins;
1085                         int k = -1;
1086                         for (; *margin; ++margin) {
1087                                 k += 1;
1088                                 // search for the "=" in e.g. "lmargin=2cm" to get the value
1089                                 for(size_t i = 0; i != opts.size(); i++) {
1090                                         if (opts.at(i).find(*margin) != string::npos) {
1091                                                 string::size_type pos = opts.at(i).find("=");
1092                                                 string value = opts.at(i).substr(pos + 1);
1093                                                 string name = known_coded_paper_margins[k];
1094                                                 h_margins += "\\" + name + " " + value + "\n";
1095                                         }
1096                                 }
1097                         }
1098                 }
1099
1100                 else if (t.cs() == "definecolor") {
1101                         string const color = p.getArg('{', '}');
1102                         string const space = p.getArg('{', '}');
1103                         string const value = p.getArg('{', '}');
1104                         if (color == "note_fontcolor" && space == "rgb") {
1105                                 RGBColor c(RGBColorFromLaTeX(value));
1106                                 h_notefontcolor = X11hexname(c);
1107                         } else {
1108                                 h_preamble << "\\definecolor{" << color
1109                                            << "}{" << space << "}{" << value
1110                                            << '}';
1111                         }
1112                 }
1113
1114                 else if (t.cs() == "jurabibsetup") {
1115                         // FIXME p.getArg('{', '}') is most probably wrong (it
1116                         //       does not handle nested braces).
1117                         //       Use p.verbatim_item() instead.
1118                         vector<string> jurabibsetup =
1119                                 split_options(p.getArg('{', '}'));
1120                         // add jurabibsetup to the jurabib package options
1121                         add_package("jurabib", jurabibsetup);
1122                         if (!jurabibsetup.empty()) {
1123                                 h_preamble << "\\jurabibsetup{"
1124                                            << join(jurabibsetup, ",") << '}';
1125                         }
1126                 }
1127
1128                 else if (t.cs() == "hypersetup") {
1129                         vector<string> hypersetup =
1130                                 split_options(p.verbatim_item());
1131                         // add hypersetup to the hyperref package options
1132                         handle_hyperref(hypersetup);
1133                         if (!hypersetup.empty()) {
1134                                 h_preamble << "\\hypersetup{"
1135                                            << join(hypersetup, ",") << '}';
1136                         }
1137                 }
1138
1139                 else if (is_known(t.cs(), known_if_3arg_commands)) {
1140                         // prevent misparsing of \usepackage if it is used
1141                         // as an argument (see e.g. our own output of
1142                         // \@ifundefined above)
1143                         string const arg1 = p.verbatim_item();
1144                         string const arg2 = p.verbatim_item();
1145                         string const arg3 = p.verbatim_item();
1146                         if (!in_lyx_preamble) {
1147                                 h_preamble << t.asInput()
1148                                            << '{' << arg1 << '}'
1149                                            << '{' << arg2 << '}'
1150                                            << '{' << arg3 << '}';
1151                         }
1152                 }
1153
1154                 else if (is_known(t.cs(), known_if_commands)) {
1155                         // must not parse anything in conditional code, since
1156                         // LyX would output the parsed contents unconditionally
1157                         if (!in_lyx_preamble)
1158                                 h_preamble << t.asInput();
1159                         handle_if(p, in_lyx_preamble);
1160                 }
1161
1162                 else if (!t.cs().empty() && !in_lyx_preamble)
1163                         h_preamble << '\\' << t.cs();
1164         }
1165
1166         // remove the whitespace
1167         p.skip_spaces();
1168
1169         // Force textclass if the user wanted it
1170         if (!forceclass.empty())
1171                 h_textclass = forceclass;
1172         if (noweb_mode && !prefixIs(h_textclass, "literate-"))
1173                 h_textclass.insert(0, "literate-");
1174         tc.setName(h_textclass);
1175         if (!tc.load()) {
1176                 cerr << "Error: Could not read layout file for textclass \"" << h_textclass << "\"." << endl;
1177                 exit(EXIT_FAILURE);
1178         }
1179         if (h_papersides.empty()) {
1180                 ostringstream ss;
1181                 ss << tc.sides();
1182                 h_papersides = ss.str();
1183         }
1184         end_preamble(os, tc);
1185 }
1186
1187
1188 /// translates a babel language name to a LyX language name
1189 string babel2lyx(string const & language)
1190 {
1191         char const * const * where = is_known(language, known_languages);
1192         if (where)
1193                 return known_coded_languages[where - known_languages];
1194         return language;
1195 }
1196
1197 // }])
1198
1199
1200 } // namespace lyx