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