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