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_language_package        = "default";
 172 string h_font_roman              = "default";
 173 string h_font_sans               = "default";
 174 string h_font_typewriter         = "default";
 175 string h_font_default_family     = "default";
 176 string h_font_sc                 = "false";
 177 string h_font_osf                = "false";
 178 string h_font_sf_scale           = "100";
 179 string h_font_tt_scale           = "100";
 180 string h_graphics                = "default";
 181 string h_float_placement;
 182 string h_paperfontsize           = "default";
 183 string h_spacing                 = "single";
 184 string h_use_hyperref            = "0";
 185 string h_pdf_title;
 186 string h_pdf_author;
 187 string h_pdf_subject;
 188 string h_pdf_keywords;
 189 string h_pdf_bookmarks           = "1";
 190 string h_pdf_bookmarksnumbered   = "0";
 191 string h_pdf_bookmarksopen       = "0";
 192 string h_pdf_bookmarksopenlevel  = "1";
 193 string h_pdf_breaklinks          = "0";
 194 string h_pdf_pdfborder           = "0";
 195 string h_pdf_colorlinks          = "0";
 196 string h_pdf_backref             = "section";
 197 string h_pdf_pdfusetitle         = "1";
 198 string h_pdf_pagemode;
 199 string h_pdf_quoted_options;
 200 string h_papersize               = "default";
 201 string h_use_geometry            = "false";
 202 string h_use_amsmath             = "1";
 203 string h_use_esint               = "1";
 204 string h_use_mhchem              = "0";
 205 string h_use_mathdots            = "0";
 206 string h_cite_engine             = "basic";
 207 string h_use_bibtopic            = "false";
 208 string h_paperorientation        = "portrait";
 209 string h_suppress_date           = "false";
 210 string h_use_refstyle            = "0";
 211 string h_notefontcolor;
 212 string h_secnumdepth             = "3";
 213 string h_tocdepth                = "3";
 214 string h_defskip                 = "medskip";
 215 string h_paragraph_indentation   = "default";
 216 string h_quotes_language         = "english";
 217 string h_papercolumns            = "1";
 218 string h_papersides;
 219 string h_paperpagestyle          = "default";
 220 string h_listings_params;
 221 string h_tracking_changes        = "false";
 222 string h_output_changes          = "false";
 223 string h_html_math_output        = "0";
 224 string h_html_css_as_file        = "0";
 225 string h_html_be_strict          = "false";
 226 string h_margins;
 227
 228
 229 // returns true if at least one of the options in what has been found
 230 bool handle_opt(vector<string> & opts, char const * const * what, string & target)
 231 {
 232         if (opts.empty())
 233                 return false;
 234
 235         bool found = false;
 236         // the last language option is the document language (for babel and LyX)
 237         // the last size option is the document font size
 238         vector<string>::iterator it;
 239         vector<string>::iterator position = opts.begin();
 240         for (; *what; ++what) {
 241                 it = find(opts.begin(), opts.end(), *what);
 242                 if (it != opts.end()) {
 243                         if (it >= position) {
 244                                 found = true;
 245                                 target = *what;
 246                                 position = it;
 247                         }
 248                 }
 249         }
 250         return found;
 251 }
 252
 253
 254 void delete_opt(vector<string> & opts, char const * const * what)
 255 {
 256         if (opts.empty())
 257                 return;
 258
 259         // remove found options from the list
 260         // do this after handle_opt to avoid potential memory leaks
 261         vector<string>::iterator it;
 262         for (; *what; ++what) {
 263                 it = find(opts.begin(), opts.end(), *what);
 264                 if (it != opts.end())
 265                         opts.erase(it);
 266         }
 267 }
 268
 269
 270 /*!
 271  * Split a package options string (keyval format) into a vector.
 272  * Example input:
 273  *   authorformat=smallcaps,
 274  *   commabeforerest,
 275  *   titleformat=colonsep,
 276  *   bibformat={tabular,ibidem,numbered}
 277  */
 278 vector<string> split_options(string const & input)
 279 {
 280         vector<string> options;
 281         string option;
 282         Parser p(input);
 283         while (p.good()) {
 284                 Token const & t = p.get_token();
 285                 if (t.asInput() == ",") {
 286                         options.push_back(trimSpaceAndEol(option));
 287                         option.erase();
 288                 } else if (t.asInput() == "=") {
 289                         option += '=';
 290                         p.skip_spaces(true);
 291                         if (p.next_token().asInput() == "{")
 292                                 option += '{' + p.getArg('{', '}') + '}';
 293                 } else if (t.cat() != catSpace)
 294                         option += t.asInput();
 295         }
 296
 297         if (!option.empty())
 298                 options.push_back(trimSpaceAndEol(option));
 299
 300         return options;
 301 }
 302
 303
 304 /*!
 305  * Retrieve a keyval option "name={value with=sign}" named \p name from
 306  * \p options and return the value.
 307  * The found option is also removed from \p options.
 308  */
 309 string process_keyval_opt(vector<string> & options, string name)
 310 {
 311         for (size_t i = 0; i < options.size(); ++i) {
 312                 vector<string> option;
 313                 split(options[i], option, '=');
 314                 if (option.size() < 2)
 315                         continue;
 316                 if (option[0] == name) {
 317                         options.erase(options.begin() + i);
 318                         option.erase(option.begin());
 319                         return join(option, "=");
 320                 }
 321         }
 322         return "";
 323 }
 324
 325
 326 /*!
 327  * Add package \p name with options \p options to used_packages.
 328  * Remove options from \p options that we don't want to output.
 329  */
 330 void add_package(string const & name, vector<string> & options)
 331 {
 332         // every package inherits the global options
 333         if (used_packages.find(name) == used_packages.end())
 334                 used_packages[name] = split_options(h_options);
 335
 336         vector<string> & v = used_packages[name];
 337         v.insert(v.end(), options.begin(), options.end());
 338         if (name == "jurabib") {
 339                 // Don't output the order argument (see the cite command
 340                 // handling code in text.cpp).
 341                 vector<string>::iterator end =
 342                         remove(options.begin(), options.end(), "natbiborder");
 343                 end = remove(options.begin(), end, "jurabiborder");
 344                 options.erase(end, options.end());
 345         }
 346 }
 347
 348
 349 // Given is a string like "scaled=0.9", return 0.9 * 100
 350 string const scale_as_percentage(string const & scale)
 351 {
 352         string::size_type pos = scale.find('=');
 353         if (pos != string::npos) {
 354                 string value = scale.substr(pos + 1);
 355                 if (isStrDbl(value))
 356                         return convert<string>(100 * convert<double>(value));
 357         }
 358         // If the input string didn't match our expectations.
 359         // return the default value "100"
 360         return "100";
 361 }
 362
 363
 364 string remove_braces(string const & value)
 365 {
 366         if (value.empty())
 367                 return value;
 368         if (value[0] == '{' && value[value.length()-1] == '}')
 369                 return value.substr(1, value.length()-2);
 370         return value;
 371 }
 372
 373
 374 void handle_hyperref(vector<string> & options)
 375 {
 376         // FIXME swallow inputencoding changes that might surround the
 377         //       hyperref setup if it was written by LyX
 378         h_use_hyperref = "1";
 379         // swallow "unicode=true", since LyX does always write that
 380         vector<string>::iterator it =
 381                 find(options.begin(), options.end(), "unicode=true");
 382         if (it != options.end())
 383                 options.erase(it);
 384         it = find(options.begin(), options.end(), "pdfusetitle");
 385         if (it != options.end()) {
 386                 h_pdf_pdfusetitle = "1";
 387                 options.erase(it);
 388         }
 389         string bookmarks = process_keyval_opt(options, "bookmarks");
 390         if (bookmarks == "true")
 391                 h_pdf_bookmarks = "1";
 392         else if (bookmarks == "false")
 393                 h_pdf_bookmarks = "0";
 394         if (h_pdf_bookmarks == "1") {
 395                 string bookmarksnumbered =
 396                         process_keyval_opt(options, "bookmarksnumbered");
 397                 if (bookmarksnumbered == "true")
 398                         h_pdf_bookmarksnumbered = "1";
 399                 else if (bookmarksnumbered == "false")
 400                         h_pdf_bookmarksnumbered = "0";
 401                 string bookmarksopen =
 402                         process_keyval_opt(options, "bookmarksopen");
 403                 if (bookmarksopen == "true")
 404                         h_pdf_bookmarksopen = "1";
 405                 else if (bookmarksopen == "false")
 406                         h_pdf_bookmarksopen = "0";
 407                 if (h_pdf_bookmarksopen == "1") {
 408                         string bookmarksopenlevel =
 409                                 process_keyval_opt(options, "bookmarksopenlevel");
 410                         if (!bookmarksopenlevel.empty())
 411                                 h_pdf_bookmarksopenlevel = bookmarksopenlevel;
 412                 }
 413         }
 414         string breaklinks = process_keyval_opt(options, "breaklinks");
 415         if (breaklinks == "true")
 416                 h_pdf_breaklinks = "1";
 417         else if (breaklinks == "false")
 418                 h_pdf_breaklinks = "0";
 419         string pdfborder = process_keyval_opt(options, "pdfborder");
 420         if (pdfborder == "{0 0 0}")
 421                 h_pdf_pdfborder = "1";
 422         else if (pdfborder == "{0 0 1}")
 423                 h_pdf_pdfborder = "0";
 424         string backref = process_keyval_opt(options, "backref");
 425         if (!backref.empty())
 426                 h_pdf_backref = backref;
 427         string colorlinks = process_keyval_opt(options, "colorlinks");
 428         if (colorlinks == "true")
 429                 h_pdf_colorlinks = "1";
 430         else if (colorlinks == "false")
 431                 h_pdf_colorlinks = "0";
 432         string pdfpagemode = process_keyval_opt(options, "pdfpagemode");
 433         if (!pdfpagemode.empty())
 434                 h_pdf_pagemode = pdfpagemode;
 435         string pdftitle = process_keyval_opt(options, "pdftitle");
 436         if (!pdftitle.empty()) {
 437                 h_pdf_title = remove_braces(pdftitle);
 438         }
 439         string pdfauthor = process_keyval_opt(options, "pdfauthor");
 440         if (!pdfauthor.empty()) {
 441                 h_pdf_author = remove_braces(pdfauthor);
 442         }
 443         string pdfsubject = process_keyval_opt(options, "pdfsubject");
 444         if (!pdfsubject.empty())
 445                 h_pdf_subject = remove_braces(pdfsubject);
 446         string pdfkeywords = process_keyval_opt(options, "pdfkeywords");
 447         if (!pdfkeywords.empty())
 448                 h_pdf_keywords = remove_braces(pdfkeywords);
 449         if (!options.empty()) {
 450                 if (!h_pdf_quoted_options.empty())
 451                         h_pdf_quoted_options += ',';
 452                 h_pdf_quoted_options += join(options, ",");
 453                 options.clear();
 454         }
 455 }
 456
 457
 458 void handle_package(Parser &p, string const & name, string const & opts,
 459                     bool in_lyx_preamble)
 460 {
 461         vector<string> options = split_options(opts);
 462         add_package(name, options);
 463         string scale;
 464
 465         // roman fonts
 466         if (is_known(name, known_roman_fonts)) {
 467                 h_font_roman = name;
 468                 p.skip_spaces();
 469         }
 470
 471         if (name == "fourier") {
 472                 h_font_roman = "utopia";
 473                 // when font uses real small capitals
 474                 if (opts == "expert")
 475                         h_font_sc = "true";
 476         }
 477
 478         if (name == "mathpazo")
 479                 h_font_roman = "palatino";
 480
 481         if (name == "mathptmx")
 482                 h_font_roman = "times";
 483
 484         // sansserif fonts
 485         if (is_known(name, known_sans_fonts)) {
 486                 h_font_sans = name;
 487                 if (!opts.empty()) {
 488                         scale = opts;
 489                         h_font_sf_scale = scale_as_percentage(scale);
 490                 }
 491         }
 492
 493         // typewriter fonts
 494         if (is_known(name, known_typewriter_fonts)) {
 495                 h_font_typewriter = name;
 496                 if (!opts.empty()) {
 497                         scale = opts;
 498                         h_font_tt_scale = scale_as_percentage(scale);
 499                 }
 500         }
 501
 502         // font uses old-style figure
 503         if (name == "eco")
 504                 h_font_osf = "true";
 505
 506         else if (name == "amsmath" || name == "amssymb")
 507                 h_use_amsmath = "2";
 508
 509         else if (name == "esint")
 510                 h_use_esint = "2";
 511
 512         else if (name == "mhchem")
 513                 h_use_mhchem = "2";
 514
 515         else if (name == "mathdots")
 516                 h_use_mathdots = "2";
 517
 518         else if (name == "babel" && !opts.empty()) {
 519                 // check if more than one option was used - used later for inputenc
 520                 // in case inputenc is parsed before babel, set the encoding to auto
 521                 if (options.begin() != options.end() - 1) {
 522                         one_language = false;
 523                         h_inputencoding = "auto";
 524                 }
 525                 // babel takes the last language of the option of its \usepackage
 526                 // call as document language. If there is no such language option, the
 527                 // last language in the documentclass options is used.
 528                 handle_opt(options, known_languages, h_language);
 529                 delete_opt(options, known_languages);
 530         }
 531
 532         else if (name == "fontenc")
 533                  ;// ignore this
 534
 535         else if (name == "inputenc" || name == "luainputenc") {
 536                 // h_inputencoding is only set when there is not more than one
 537                 // inputenc option because otherwise h_inputencoding must be
 538                 // set to "auto" (the default encoding of the document language)
 539                 // Therefore check for the "," character.
 540                 // It is also only set when there is not more then one babel
 541                 // language option but this is handled in the routine for babel.
 542                 if (opts.find(",") == string::npos && one_language == true)
 543                         h_inputencoding = opts;
 544                 if (!options.empty())
 545                         p.setEncoding(options.back());
 546                 options.clear();
 547         }
 548
 549         else if (name == "makeidx")
 550                 ; // ignore this
 551
 552         else if (name == "prettyref")
 553                 ; // ignore this
 554
 555         else if (name == "varioref")
 556                 ; // ignore this
 557
 558         else if (name == "verbatim")
 559                 ; // ignore this
 560
 561         else if (name == "nomencl")
 562                 ; // ignore this
 563
 564         else if (name == "textcomp")
 565                 ; // ignore this
 566
 567         else if (name == "url")
 568                 ; // ignore this
 569
 570         else if (name == "subscript")
 571                 ; // ignore this
 572
 573         else if (name == "color") {
 574                 // with the following command this package is only loaded when needed for
 575                 // undefined colors, since we only support the predefined colors
 576                 h_preamble << "\\@ifundefined{definecolor}\n {\\usepackage{color}}{}\n";
 577         }
 578
 579         else if (name == "graphicx")
 580                 ; // ignore this
 581
 582         else if (name == "setspace")
 583                 ; // ignore this
 584
 585         else if (name == "geometry")
 586                 ; // Ignore this, the geometry settings are made by the \geometry
 587                   // command. This command is handled below.
 588
 589         else if (is_known(name, known_languages))
 590                 h_language = name;
 591
 592         else if (name == "natbib") {
 593                 h_cite_engine = "natbib_authoryear";
 594                 vector<string>::iterator it =
 595                         find(options.begin(), options.end(), "authoryear");
 596                 if (it != options.end())
 597                         options.erase(it);
 598                 else {
 599                         it = find(options.begin(), options.end(), "numbers");
 600                         if (it != options.end()) {
 601                                 h_cite_engine = "natbib_numerical";
 602                                 options.erase(it);
 603                         }
 604                 }
 605         }
 606
 607         else if (name == "jurabib")
 608                 h_cite_engine = "jurabib";
 609
 610         else if (name == "hyperref")
 611                 handle_hyperref(options);
 612
 613         else if (!in_lyx_preamble) {
 614                 if (options.empty())
 615                         h_preamble << "\\usepackage{" << name << "}";
 616                 else {
 617                         h_preamble << "\\usepackage[" << opts << "]{"
 618                                    << name << "}";
 619                         options.clear();
 620                 }
 621         }
 622
 623         // We need to do something with the options...
 624         if (!options.empty())
 625                 cerr << "Ignoring options '" << join(options, ",")
 626                      << "' of package " << name << '.' << endl;
 627
 628         // remove the whitespace
 629         p.skip_spaces();
 630 }
 631
 632
 633 void handle_if(Parser & p, bool in_lyx_preamble)
 634 {
 635         while (p.good()) {
 636                 Token t = p.get_token();
 637                 if (t.cat() == catEscape &&
 638                     is_known(t.cs(), known_if_commands))
 639                         handle_if(p, in_lyx_preamble);
 640                 else {
 641                         if (!in_lyx_preamble)
 642                                 h_preamble << t.asInput();
 643                         if (t.cat() == catEscape && t.cs() == "fi")
 644                                 return;
 645                 }
 646         }
 647 }
 648
 649
 650 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 651 {
 652         // translate from babel to LyX names
 653         h_language = babel2lyx(h_language);
 654
 655         // set the quote language
 656         // LyX only knows the following quotes languages:
 657         // english, swedish, german, polish, french and danish
 658         // (quotes for "japanese" and "chinese-traditional" are missing because
 659         //  they wouldn't be useful: http://www.lyx.org/trac/ticket/6383)
 660         // conversion list taken from
 661         // http://en.wikipedia.org/wiki/Quotation_mark,_non-English_usage
 662         // (quotes for kazakh and interlingua are unknown)
 663         // danish
 664         if (h_language == "danish")
 665                 h_quotes_language = "danish";
 666         // french
 667         else if (is_known(h_language, known_french_quotes_languages))
 668                 h_quotes_language = "french";
 669         // german
 670         else if (is_known(h_language, known_german_quotes_languages))
 671                 h_quotes_language = "german";
 672         // polish
 673         else if (is_known(h_language, known_polish_quotes_languages))
 674                 h_quotes_language = "polish";
 675         // swedish
 676         else if (is_known(h_language, known_swedish_quotes_languages))
 677                 h_quotes_language = "swedish";
 678         //english
 679         else if (is_known(h_language, known_english_quotes_languages))
 680                 h_quotes_language = "english";
 681
 682         // output the LyX file settings
 683         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 684            << "\\lyxformat " << LYX_FORMAT << '\n'
 685            << "\\begin_document\n"
 686            << "\\begin_header\n"
 687            << "\\textclass " << h_textclass << "\n";
 688         if (!h_preamble.str().empty())
 689                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 690         if (!h_options.empty())
 691                 os << "\\options " << h_options << "\n";
 692         os << "\\use_default_options " << h_use_default_options << "\n"
 693            << modules_placeholder
 694            << "\\language " << h_language << "\n"
 695            << "\\language_package " << h_language_package << "\n"
 696            << "\\inputencoding " << h_inputencoding << "\n"
 697            << "\\font_roman " << h_font_roman << "\n"
 698            << "\\font_sans " << h_font_sans << "\n"
 699            << "\\font_typewriter " << h_font_typewriter << "\n"
 700            << "\\font_default_family " << h_font_default_family << "\n"
 701            << "\\font_sc " << h_font_sc << "\n"
 702            << "\\font_osf " << h_font_osf << "\n"
 703            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 704            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 705            << "\\graphics " << h_graphics << "\n";
 706         if (!h_float_placement.empty())
 707                 os << "\\float_placement " << h_float_placement << "\n";
 708         os << "\\paperfontsize " << h_paperfontsize << "\n"
 709            << "\\spacing " << h_spacing << "\n"
 710            << "\\use_hyperref " << h_use_hyperref << '\n';
 711         if (h_use_hyperref == "1") {
 712                 if (!h_pdf_title.empty())
 713                         os << "\\pdf_title \"" << h_pdf_title << "\"\n";
 714                 if (!h_pdf_author.empty())
 715                         os << "\\pdf_author \"" << h_pdf_author << "\"\n";
 716                 if (!h_pdf_subject.empty())
 717                         os << "\\pdf_subject \"" << h_pdf_subject << "\"\n";
 718                 if (!h_pdf_keywords.empty())
 719                         os << "\\pdf_keywords \"" << h_pdf_keywords << "\"\n";
 720                 os << "\\pdf_bookmarks " << h_pdf_bookmarks << "\n"
 721                       "\\pdf_bookmarksnumbered " << h_pdf_bookmarksnumbered << "\n"
 722                       "\\pdf_bookmarksopen " << h_pdf_bookmarksopen << "\n"
 723                       "\\pdf_bookmarksopenlevel " << h_pdf_bookmarksopenlevel << "\n"
 724                       "\\pdf_breaklinks " << h_pdf_breaklinks << "\n"
 725                       "\\pdf_pdfborder " << h_pdf_pdfborder << "\n"
 726                       "\\pdf_colorlinks " << h_pdf_colorlinks << "\n"
 727                       "\\pdf_backref " << h_pdf_backref << "\n"
 728                       "\\pdf_pdfusetitle " << h_pdf_pdfusetitle << '\n';
 729                 if (!h_pdf_pagemode.empty())
 730                         os << "\\pdf_pagemode " << h_pdf_pagemode << '\n';
 731                 if (!h_pdf_quoted_options.empty())
 732                         os << "\\pdf_quoted_options \"" << h_pdf_quoted_options << "\"\n";
 733         }
 734         os << "\\papersize " << h_papersize << "\n"
 735            << "\\use_geometry " << h_use_geometry << "\n"
 736            << "\\use_amsmath " << h_use_amsmath << "\n"
 737            << "\\use_esint " << h_use_esint << "\n"
 738            << "\\use_mhchem " << h_use_mhchem << "\n"
 739            << "\\use_mathdots " << h_use_mathdots << "\n"
 740            << "\\cite_engine " << h_cite_engine << "\n"
 741            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 742            << "\\paperorientation " << h_paperorientation << '\n'
 743            << "\\suppress_date " << h_suppress_date << '\n'
 744            << "\\use_refstyle " << h_use_refstyle << '\n';
 745         if (!h_notefontcolor.empty())
 746                 os << "\\notefontcolor " << h_notefontcolor << '\n';
 747         os << h_margins
 748            << "\\secnumdepth " << h_secnumdepth << "\n"
 749            << "\\tocdepth " << h_tocdepth << "\n"
 750            << "\\paragraph_separation " << h_paragraph_separation << "\n";
 751         if (h_paragraph_separation == "skip")
 752                 os << "\\defskip " << h_defskip << "\n";
 753         else
 754                 os << "\\paragraph_indentation " << h_paragraph_indentation << "\n";
 755         os << "\\quotes_language " << h_quotes_language << "\n"
 756            << "\\papercolumns " << h_papercolumns << "\n"
 757            << "\\papersides " << h_papersides << "\n"
 758            << "\\paperpagestyle " << h_paperpagestyle << "\n";
 759         if (!h_listings_params.empty())
 760                 os << "\\listings_params " << h_listings_params << "\n";
 761         os << "\\tracking_changes " << h_tracking_changes << "\n"
 762            << "\\output_changes " << h_output_changes << "\n"
 763            << "\\html_math_output " << h_html_math_output << "\n"
 764            << "\\html_css_as_file " << h_html_css_as_file << "\n"
 765            << "\\html_be_strict " << h_html_be_strict << "\n"
 766            << "\\end_header\n\n"
 767            << "\\begin_body\n";
 768         // clear preamble for subdocuments
 769         h_preamble.str("");
 770 }
 771
 772 } // anonymous namespace
 773
 774
 775 void parse_preamble(Parser & p, ostream & os,
 776         string const & forceclass, TeX2LyXDocClass & tc)
 777 {
 778         // initialize fixed types
 779         special_columns['D'] = 3;
 780         bool is_full_document = false;
 781         bool is_lyx_file = false;
 782         bool in_lyx_preamble = false;
 783
 784         // determine whether this is a full document or a fragment for inclusion
 785         while (p.good()) {
 786                 Token const & t = p.get_token();
 787
 788                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 789                         is_full_document = true;
 790                         break;
 791                 }
 792         }
 793         p.reset();
 794
 795         while (is_full_document && p.good()) {
 796                 Token const & t = p.get_token();
 797
 798 #ifdef FILEDEBUG
 799                 cerr << "t: " << t << "\n";
 800 #endif
 801
 802                 //
 803                 // cat codes
 804                 //
 805                 if (!in_lyx_preamble &&
 806                     (t.cat() == catLetter ||
 807                      t.cat() == catSuper ||
 808                      t.cat() == catSub ||
 809                      t.cat() == catOther ||
 810                      t.cat() == catMath ||
 811                      t.cat() == catActive ||
 812                      t.cat() == catBegin ||
 813                      t.cat() == catEnd ||
 814                      t.cat() == catAlign ||
 815                      t.cat() == catParameter))
 816                         h_preamble << t.cs();
 817
 818                 else if (!in_lyx_preamble &&
 819                          (t.cat() == catSpace || t.cat() == catNewline))
 820                         h_preamble << t.asInput();
 821
 822                 else if (t.cat() == catComment) {
 823                         static regex const islyxfile("%% LyX .* created this file");
 824                         static regex const usercommands("User specified LaTeX commands");
 825
 826                         string const comment = t.asInput();
 827
 828                         // magically switch encoding default if it looks like XeLaTeX
 829                         static string const magicXeLaTeX =
 830                                 "% This document must be compiled with XeLaTeX ";
 831                         if (comment.size() > magicXeLaTeX.size()
 832                                   && comment.substr(0, magicXeLaTeX.size()) == magicXeLaTeX
 833                                   && h_inputencoding == "auto") {
 834                                 cerr << "XeLaTeX comment found, switching to UTF8\n";
 835                                 h_inputencoding = "utf8";
 836                         }
 837                         smatch sub;
 838                         if (regex_search(comment, sub, islyxfile)) {
 839                                 is_lyx_file = true;
 840                                 in_lyx_preamble = true;
 841                         } else if (is_lyx_file
 842                                    && regex_search(comment, sub, usercommands))
 843                                 in_lyx_preamble = false;
 844                         else if (!in_lyx_preamble)
 845                                 h_preamble << t.asInput();
 846                 }
 847
 848                 else if (t.cs() == "pagestyle")
 849                         h_paperpagestyle = p.verbatim_item();
 850
 851                 else if (t.cs() == "makeatletter") {
 852                         // LyX takes care of this
 853                         p.setCatCode('@', catLetter);
 854                 }
 855
 856                 else if (t.cs() == "makeatother") {
 857                         // LyX takes care of this
 858                         p.setCatCode('@', catOther);
 859                 }
 860
 861                 else if (t.cs() == "newcommand" || t.cs() == "newcommandx"
 862                       || t.cs() == "renewcommand" || t.cs() == "renewcommandx"
 863                       || t.cs() == "providecommand" || t.cs() == "providecommandx"
 864                                 || t.cs() == "DeclareRobustCommand"
 865                       || t.cs() == "DeclareRobustCommandx"
 866                                 || t.cs() == "ProvideTextCommandDefault"
 867                                 || t.cs() == "DeclareMathAccent") {
 868                         bool star = false;
 869                         if (p.next_token().character() == '*') {
 870                                 p.get_token();
 871                                 star = true;
 872                         }
 873                         string const name = p.verbatim_item();
 874                         string const opt1 = p.getFullOpt();
 875                         string const opt2 = p.getFullOpt();
 876                         string const body = p.verbatim_item();
 877                         // font settings
 878                         if (name == "\\rmdefault")
 879                                 if (is_known(body, known_roman_fonts))
 880                                         h_font_roman = body;
 881                         if (name == "\\sfdefault")
 882                                 if (is_known(body, known_sans_fonts))
 883                                         h_font_sans = body;
 884                         if (name == "\\ttdefault")
 885                                 if (is_known(body, known_typewriter_fonts))
 886                                         h_font_typewriter = body;
 887                         if (name == "\\familydefault") {
 888                                 string family = body;
 889                                 // remove leading "\"
 890                                 h_font_default_family = family.erase(0,1);
 891                         }
 892
 893                         // Add the command to the known commands
 894                         add_known_command(name, opt1, !opt2.empty(), from_utf8(body));
 895
 896                         // only non-lyxspecific stuff
 897                         if (!in_lyx_preamble) {
 898                                 ostringstream ss;
 899                                 ss << '\\' << t.cs();
 900                                 if (star)
 901                                         ss << '*';
 902                                 ss << '{' << name << '}' << opt1 << opt2
 903                                    << '{' << body << "}";
 904                                 h_preamble << ss.str();
 905 /*
 906                                 ostream & out = in_preamble ? h_preamble : os;
 907                                 out << "\\" << t.cs() << "{" << name << "}"
 908                                     << opts << "{" << body << "}";
 909 */
 910                         }
 911                 }
 912
 913                 else if (t.cs() == "documentclass") {
 914                         vector<string>::iterator it;
 915                         vector<string> opts = split_options(p.getArg('[', ']'));
 916                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 917                         delete_opt(opts, known_fontsizes);
 918                         // delete "pt" at the end
 919                         string::size_type i = h_paperfontsize.find("pt");
 920                         if (i != string::npos)
 921                                 h_paperfontsize.erase(i);
 922                         // The documentclass options are always parsed before the options
 923                         // of the babel call so that a language cannot overwrite the babel
 924                         // options.
 925                         handle_opt(opts, known_languages, h_language);
 926                         delete_opt(opts, known_languages);
 927
 928                         // paper orientation
 929                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
 930                                 h_paperorientation = "landscape";
 931                                 opts.erase(it);
 932                         }
 933                         // paper sides
 934                         if ((it = find(opts.begin(), opts.end(), "oneside"))
 935                                  != opts.end()) {
 936                                 h_papersides = "1";
 937                                 opts.erase(it);
 938                         }
 939                         if ((it = find(opts.begin(), opts.end(), "twoside"))
 940                                  != opts.end()) {
 941                                 h_papersides = "2";
 942                                 opts.erase(it);
 943                         }
 944                         // paper columns
 945                         if ((it = find(opts.begin(), opts.end(), "onecolumn"))
 946                                  != opts.end()) {
 947                                 h_papercolumns = "1";
 948                                 opts.erase(it);
 949                         }
 950                         if ((it = find(opts.begin(), opts.end(), "twocolumn"))
 951                                  != opts.end()) {
 952                                 h_papercolumns = "2";
 953                                 opts.erase(it);
 954                         }
 955                         // paper sizes
 956                         // some size options are know to any document classes, other sizes
 957                         // are handled by the \geometry command of the geometry package
 958                         handle_opt(opts, known_class_paper_sizes, h_papersize);
 959                         delete_opt(opts, known_class_paper_sizes);
 960                         // the remaining options
 961                         h_options = join(opts, ",");
 962                         // FIXME This does not work for classes that have a
 963                         //       different name in LyX than in LaTeX
 964                         h_textclass = p.getArg('{', '}');
 965                 }
 966
 967                 else if (t.cs() == "usepackage") {
 968                         string const options = p.getArg('[', ']');
 969                         string const name = p.getArg('{', '}');
 970                         vector<string> vecnames;
 971                         split(name, vecnames, ',');
 972                         vector<string>::const_iterator it  = vecnames.begin();
 973                         vector<string>::const_iterator end = vecnames.end();
 974                         for (; it != end; ++it)
 975                                 handle_package(p, trimSpaceAndEol(*it), options,
 976                                                in_lyx_preamble);
 977                 }
 978
 979                 else if (t.cs() == "inputencoding") {
 980                         string const encoding = p.getArg('{','}');
 981                         h_inputencoding = encoding;
 982                         p.setEncoding(encoding);
 983                 }
 984
 985                 else if (t.cs() == "newenvironment") {
 986                         string const name = p.getArg('{', '}');
 987                         string const opt1 = p.getFullOpt();
 988                         string const opt2 = p.getFullOpt();
 989                         string const beg = p.verbatim_item();
 990                         string const end = p.verbatim_item();
 991                         if (!in_lyx_preamble) {
 992                                 h_preamble << "\\newenvironment{" << name
 993                                            << '}' << opt1 << opt2 << '{'
 994                                            << beg << "}{" << end << '}';
 995                         }
 996                         add_known_environment(name, opt1, !opt2.empty(),
 997                                               from_utf8(beg), from_utf8(end));
 998
 999                 }
1000
1001                 else if (t.cs() == "def") {
1002                         string name = p.get_token().cs();
1003                         while (p.next_token().cat() != catBegin)
1004                                 name += p.get_token().cs();
1005                         if (!in_lyx_preamble)
1006                                 h_preamble << "\\def\\" << name << '{'
1007                                            << p.verbatim_item() << "}";
1008                 }
1009
1010                 else if (t.cs() == "newcolumntype") {
1011                         string const name = p.getArg('{', '}');
1012                         trimSpaceAndEol(name);
1013                         int nargs = 0;
1014                         string opts = p.getOpt();
1015                         if (!opts.empty()) {
1016                                 istringstream is(string(opts, 1));
1017                                 is >> nargs;
1018                         }
1019                         special_columns[name[0]] = nargs;
1020                         h_preamble << "\\newcolumntype{" << name << "}";
1021                         if (nargs)
1022                                 h_preamble << "[" << nargs << "]";
1023                         h_preamble << "{" << p.verbatim_item() << "}";
1024                 }
1025
1026                 else if (t.cs() == "setcounter") {
1027                         string const name = p.getArg('{', '}');
1028                         string const content = p.getArg('{', '}');
1029                         if (name == "secnumdepth")
1030                                 h_secnumdepth = content;
1031                         else if (name == "tocdepth")
1032                                 h_tocdepth = content;
1033                         else
1034                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
1035                 }
1036
1037                 else if (t.cs() == "setlength") {
1038                         string const name = p.verbatim_item();
1039                         string const content = p.verbatim_item();
1040                         // the paragraphs are only not indented when \parindent is set to zero
1041                         if (name == "\\parindent" && content != "") {
1042                                 if (content[0] == '0')
1043                                         h_paragraph_separation = "skip";
1044                                 else
1045                                         h_paragraph_indentation = translate_len(content);
1046                         } else if (name == "\\parskip") {
1047                                 if (content == "\\smallskipamount")
1048                                         h_defskip = "smallskip";
1049                                 else if (content == "\\medskipamount")
1050                                         h_defskip = "medskip";
1051                                 else if (content == "\\bigskipamount")
1052                                         h_defskip = "bigskip";
1053                                 else
1054                                         h_defskip = content;
1055                         } else
1056                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
1057                 }
1058
1059                 else if (t.cs() == "onehalfspacing")
1060                         h_spacing = "onehalf";
1061
1062                 else if (t.cs() == "doublespacing")
1063                         h_spacing = "double";
1064
1065                 else if (t.cs() == "setstretch")
1066                         h_spacing = "other " + p.verbatim_item();
1067
1068                 else if (t.cs() == "begin") {
1069                         string const name = p.getArg('{', '}');
1070                         if (name == "document")
1071                                 break;
1072                         h_preamble << "\\begin{" << name << "}";
1073                 }
1074
1075                 else if (t.cs() == "geometry") {
1076                         h_use_geometry = "true";
1077                         vector<string> opts = split_options(p.getArg('{', '}'));
1078                         vector<string>::iterator it;
1079                         // paper orientation
1080                         if ((it = find(opts.begin(), opts.end(), "landscape")) != opts.end()) {
1081                                 h_paperorientation = "landscape";
1082                                 opts.erase(it);
1083                         }
1084                         // paper size
1085                         handle_opt(opts, known_paper_sizes, h_papersize);
1086                         delete_opt(opts, known_paper_sizes);
1087                         // page margins
1088                         char const * const * margin = known_paper_margins;
1089                         int k = -1;
1090                         for (; *margin; ++margin) {
1091                                 k += 1;
1092                                 // search for the "=" in e.g. "lmargin=2cm" to get the value
1093                                 for(size_t i = 0; i != opts.size(); i++) {
1094                                         if (opts.at(i).find(*margin) != string::npos) {
1095                                                 string::size_type pos = opts.at(i).find("=");
1096                                                 string value = opts.at(i).substr(pos + 1);
1097                                                 string name = known_coded_paper_margins[k];
1098                                                 h_margins += "\\" + name + " " + value + "\n";
1099                                         }
1100                                 }
1101                         }
1102                 }
1103
1104                 else if (t.cs() == "definecolor") {
1105                         string const color = p.getArg('{', '}');
1106                         string const space = p.getArg('{', '}');
1107                         string const value = p.getArg('{', '}');
1108                         if (color == "note_fontcolor" && space == "rgb") {
1109                                 RGBColor c(RGBColorFromLaTeX(value));
1110                                 h_notefontcolor = X11hexname(c);
1111                         } else {
1112                                 h_preamble << "\\definecolor{" << color
1113                                            << "}{" << space << "}{" << value
1114                                            << '}';
1115                         }
1116                 }
1117
1118                 else if (t.cs() == "jurabibsetup") {
1119                         // FIXME p.getArg('{', '}') is most probably wrong (it
1120                         //       does not handle nested braces).
1121                         //       Use p.verbatim_item() instead.
1122                         vector<string> jurabibsetup =
1123                                 split_options(p.getArg('{', '}'));
1124                         // add jurabibsetup to the jurabib package options
1125                         add_package("jurabib", jurabibsetup);
1126                         if (!jurabibsetup.empty()) {
1127                                 h_preamble << "\\jurabibsetup{"
1128                                            << join(jurabibsetup, ",") << '}';
1129                         }
1130                 }
1131
1132                 else if (t.cs() == "hypersetup") {
1133                         vector<string> hypersetup =
1134                                 split_options(p.verbatim_item());
1135                         // add hypersetup to the hyperref package options
1136                         handle_hyperref(hypersetup);
1137                         if (!hypersetup.empty()) {
1138                                 h_preamble << "\\hypersetup{"
1139                                            << join(hypersetup, ",") << '}';
1140                         }
1141                 }
1142
1143                 else if (is_known(t.cs(), known_if_3arg_commands)) {
1144                         // prevent misparsing of \usepackage if it is used
1145                         // as an argument (see e.g. our own output of
1146                         // \@ifundefined above)
1147                         string const arg1 = p.verbatim_item();
1148                         string const arg2 = p.verbatim_item();
1149                         string const arg3 = p.verbatim_item();
1150                         if (!in_lyx_preamble) {
1151                                 h_preamble << t.asInput()
1152                                            << '{' << arg1 << '}'
1153                                            << '{' << arg2 << '}'
1154                                            << '{' << arg3 << '}';
1155                         }
1156                 }
1157
1158                 else if (is_known(t.cs(), known_if_commands)) {
1159                         // must not parse anything in conditional code, since
1160                         // LyX would output the parsed contents unconditionally
1161                         if (!in_lyx_preamble)
1162                                 h_preamble << t.asInput();
1163                         handle_if(p, in_lyx_preamble);
1164                 }
1165
1166                 else if (!t.cs().empty() && !in_lyx_preamble)
1167                         h_preamble << '\\' << t.cs();
1168         }
1169
1170         // remove the whitespace
1171         p.skip_spaces();
1172
1173         // Force textclass if the user wanted it
1174         if (!forceclass.empty())
1175                 h_textclass = forceclass;
1176         if (noweb_mode && !prefixIs(h_textclass, "literate-"))
1177                 h_textclass.insert(0, "literate-");
1178         tc.setName(h_textclass);
1179         if (!tc.load()) {
1180                 cerr << "Error: Could not read layout file for textclass \"" << h_textclass << "\"." << endl;
1181                 exit(EXIT_FAILURE);
1182         }
1183         if (h_papersides.empty()) {
1184                 ostringstream ss;
1185                 ss << tc.sides();
1186                 h_papersides = ss.str();
1187         }
1188         end_preamble(os, tc);
1189 }
1190
1191
1192 /// translates a babel language name to a LyX language name
1193 string babel2lyx(string const & language)
1194 {
1195         char const * const * where = is_known(language, known_languages);
1196         if (where)
1197                 return known_coded_languages[where - known_languages];
1198         return language;
1199 }
1200
1201 // }])
1202
1203
1204 } // namespace lyx