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 "Layout.h"
  19 #include "Lexer.h"
  20 #include "TextClass.h"
  21 #include "support/convert.h"
  22 #include "support/filetools.h"
  23 #include "support/lstrings.h"
  24
  25 #include <algorithm>
  26 #include <iostream>
  27 #include <sstream>
  28 #include <string>
  29 #include <vector>
  30 #include <map>
  31
  32
  33 namespace lyx {
  34
  35 using std::istringstream;
  36 using std::ostream;
  37 using std::ostringstream;
  38 using std::string;
  39 using std::vector;
  40 using std::cerr;
  41 using std::endl;
  42 using std::find;
  43
  44 using support::FileName;
  45 using support::libFileSearch;
  46 using support::isStrDbl;
  47
  48 // special columntypes
  49 extern std::map<char, int> special_columns;
  50
  51 std::map<string, vector<string> > used_packages;
  52
  53 namespace {
  54
  55 const char * const known_languages[] = { "afrikaans", "american", "arabic",
  56 "austrian", "bahasa", "basque", "belarusian", "brazil", "breton", "british",
  57 "bulgarian", "canadian", "canadien", "catalan", "croatian", "czech", "danish",
  58 "dutch", "english", "esperanto", "estonian", "finnish", "francais", "french",
  59 "frenchb", "frenchle", "frenchpro", "galician", "german", "germanb", "greek",
  60 "hebrew", "icelandic", "irish", "italian", "lsorbian", "magyar", "naustrian",
  61 "ngerman", "ngermanb", "norsk", "nynorsk", "polish", "portuges", "romanian",
  62 "russian", "russianb", "scottish", "serbian", "slovak", "slovene", "spanish",
  63 "swedish", "thai", "turkish", "ukraineb", "ukrainian", "usorbian", "welsh", 0};
  64
  65 //note this when updating to lyxformat 305:
  66 //bahasai, indonesian, and indon = equal to bahasa
  67 //malay, and meyalu = equal to bahasam
  68
  69 const char * const known_french_languages[] = {"french", "frenchb", "francais",
  70                                                 "frenchle", "frenchpro", 0};
  71 const char * const known_german_languages[] = {"german", "germanb", 0};
  72 const char * const known_ngerman_languages[] = {"ngerman", "ngermanb", 0};
  73 const char * const known_russian_languages[] = {"russian", "russianb", 0};
  74 const char * const known_ukrainian_languages[] = {"ukrainian", "ukraineb", 0};
  75
  76 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
  77
  78 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
  79 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
  80
  81 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
  82 "helvet", "lmss", 0};
  83
  84 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
  85 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
  86 "newcent", 0};
  87
  88 // some ugly stuff
  89 ostringstream h_preamble;
  90 string h_textclass               = "article";
  91 string h_options                 = string();
  92 string h_language                = "english";
  93 string h_inputencoding           = "auto";
  94 string h_font_roman              = "default";
  95 string h_font_sans               = "default";
  96 string h_font_typewriter         = "default";
  97 string h_font_default_family     = "default";
  98 string h_font_sc                 = "false";
  99 string h_font_osf                = "false";
 100 string h_font_sf_scale           = "100";
 101 string h_font_tt_scale           = "100";
 102 string h_graphics                = "default";
 103 string h_paperfontsize           = "default";
 104 string h_spacing                 = "single";
 105 string h_papersize               = "default";
 106 string h_use_geometry            = "false";
 107 string h_use_amsmath             = "0";
 108 string h_cite_engine             = "basic";
 109 string h_use_bibtopic            = "false";
 110 string h_paperorientation        = "portrait";
 111 string h_secnumdepth             = "3";
 112 string h_tocdepth                = "3";
 113 string h_paragraph_separation    = "indent";
 114 string h_defskip                 = "medskip";
 115 string h_quotes_language         = "english";
 116 string h_papercolumns            = "1";
 117 string h_papersides              = string();
 118 string h_paperpagestyle          = "default";
 119 string h_tracking_changes        = "false";
 120 string h_output_changes          = "false";
 121
 122
 123 void handle_opt(vector<string> & opts, char const * const * what, string & target)
 124 {
 125         if (opts.empty())
 126                 return;
 127
 128         // the last language option is the document language (for babel and LyX)
 129         // the last size option is the document font size
 130         vector<string>::iterator it;
 131         vector<string>::iterator position = opts.begin();
 132         for (; *what; ++what) {
 133                 it = find(opts.begin(), opts.end(), *what);
 134                 if (it != opts.end()) {
 135                         if (it >= position) {
 136                                 target = *what;
 137                                 position = it;
 138                         }
 139                         // remove found options from the list
 140                         opts.erase(it);
 141                 }
 142         }
 143 }
 144
 145
 146 /*!
 147  * Split a package options string (keyval format) into a vector.
 148  * Example input:
 149  *   authorformat=smallcaps,
 150  *   commabeforerest,
 151  *   titleformat=colonsep,
 152  *   bibformat={tabular,ibidem,numbered}
 153  */
 154 vector<string> split_options(string const & input)
 155 {
 156         vector<string> options;
 157         string option;
 158         Parser p(input);
 159         while (p.good()) {
 160                 Token const & t = p.get_token();
 161                 if (t.asInput() == ",") {
 162                         options.push_back(trim(option));
 163                         option.erase();
 164                 } else if (t.asInput() == "=") {
 165                         option += '=';
 166                         p.skip_spaces(true);
 167                         if (p.next_token().asInput() == "{")
 168                                 option += '{' + p.getArg('{', '}') + '}';
 169                 } else if (t.cat() != catSpace)
 170                         option += t.asInput();
 171         }
 172
 173         if (!option.empty())
 174                 options.push_back(trim(option));
 175
 176         return options;
 177 }
 178
 179
 180 /*!
 181  * Add package \p name with options \p options to used_packages.
 182  * Remove options from \p options that we don't want to output.
 183  */
 184 void add_package(string const & name, vector<string> & options)
 185 {
 186         // every package inherits the global options
 187         if (used_packages.find(name) == used_packages.end())
 188                 used_packages[name] = split_options(h_options);
 189
 190         vector<string> & v = used_packages[name];
 191         v.insert(v.end(), options.begin(), options.end());
 192         if (name == "jurabib") {
 193                 // Don't output the order argument (see the cite command
 194                 // handling code in text.cpp).
 195                 vector<string>::iterator end =
 196                         remove(options.begin(), options.end(), "natbiborder");
 197                 end = remove(options.begin(), end, "jurabiborder");
 198                 options.erase(end, options.end());
 199         }
 200 }
 201
 202
 203 void handle_package(string const & name, string const & opts)
 204 {
 205         vector<string> options = split_options(opts);
 206         add_package(name, options);
 207         size_t pos;
 208         string scale;
 209
 210         // roman fonts
 211         if (is_known(name, known_roman_fonts))
 212                 h_font_roman = name;
 213         if (name == "fourier") {
 214                 h_font_roman = "utopia";
 215                 // when font uses real small capitals
 216                 if (opts == "expert")
 217                         h_font_sc = "true";
 218         }
 219         if (name == "mathpazo")
 220                 h_font_roman = "palatino";
 221         if (name == "mathptmx")
 222                 h_font_roman = "times";
 223         // sansserif fonts
 224         if (is_known(name, known_sans_fonts)) {
 225                 h_font_sans = name;
 226                 if (!opts.empty()) {
 227                         scale = opts;
 228                         // the option is in the form "scaled=0.9"
 229                         // therefore cut of before the "="
 230                         pos = scale.find("=");
 231                         if (pos != string::npos) {
 232                                 scale.erase(0, pos + 1);
 233                                 if (isStrDbl(scale)) {
 234                                         // LyX needs the scale as integer, therfore multiply by 100
 235                                         scale = convert<string>(100 * convert<double>(scale));
 236                                         h_font_sf_scale = scale;
 237                                 }
 238                         }
 239                 }
 240         }
 241         // typewriter fonts
 242         if (is_known(name, known_typewriter_fonts)) {
 243                 h_font_typewriter = name;
 244                 if (!opts.empty()) {
 245                         scale = opts;
 246                         // the option is in the form "scaled=0.9"
 247                         // therefore cut of before the "="
 248                         pos = scale.find("=");
 249                         if (pos != string::npos) {
 250                                 scale.erase(0, pos + 1);
 251                                 if (isStrDbl(scale)) {
 252                                         // LyX needs the scale as integer, therfore multiply by 100
 253                                         scale = convert<string>(100 * convert<double>(scale));
 254                                         h_font_tt_scale = scale;
 255                                 }
 256                         }
 257                 }
 258         }
 259         // font uses old-style figure
 260         if (name == "eco")
 261                 h_font_osf = "true";
 262
 263         else if (name == "amsmath" || name == "amssymb")
 264                 h_use_amsmath = "1";
 265         else if (name == "babel")
 266                 ; // ignore this
 267         else if (name == "fontenc")
 268                 ; // ignore this
 269         else if (name == "inputenc") {
 270                 // only set when there is not more than one inputenc option
 271                 // therefore check for the "," character
 272                 if (opts.find(",") == string::npos)
 273                         h_inputencoding = opts;
 274                 options.clear();
 275         } else if (name == "makeidx")
 276                 ; // ignore this
 277         else if (name == "verbatim")
 278                 ; // ignore this
 279         else if (name == "graphicx")
 280                 ; // ignore this
 281         else if (is_known(name, known_languages)) {
 282                 if (is_known(name, known_french_languages))
 283                         h_language = "french";
 284                 else if (is_known(name, known_german_languages))
 285                         h_language = "german";
 286                 else if (is_known(name, known_ngerman_languages))
 287                         h_language = "ngerman";
 288                 else if (is_known(name, known_russian_languages))
 289                         h_language = "russian";
 290                 else if (is_known(name, known_ukrainian_languages))
 291                         h_language = "ukrainian";
 292                 else
 293                         h_language = name;
 294                 h_quotes_language = h_language;
 295
 296         } else if (name == "natbib") {
 297                 h_cite_engine = "natbib_authoryear";
 298                 vector<string>::iterator it =
 299                         find(options.begin(), options.end(), "authoryear");
 300                 if (it != options.end())
 301                         options.erase(it);
 302                 else {
 303                         it = find(options.begin(), options.end(), "numbers");
 304                         if (it != options.end()) {
 305                                 h_cite_engine = "natbib_numerical";
 306                                 options.erase(it);
 307                         }
 308                 }
 309         } else if (name == "jurabib") {
 310                 h_cite_engine = "jurabib";
 311         } else if (options.empty())
 312                 h_preamble << "\\usepackage{" << name << "}\n";
 313         else {
 314                 h_preamble << "\\usepackage[" << opts << "]{" << name << "}\n";
 315                 options.clear();
 316         }
 317
 318         // We need to do something with the options...
 319         if (!options.empty())
 320                 cerr << "Ignoring options '" << join(options, ",")
 321                      << "' of package " << name << '.' << endl;
 322 }
 323
 324
 325
 326 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 327 {
 328         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 329            << "\\lyxformat 247\n"
 330            << "\\begin_document\n"
 331            << "\\begin_header\n"
 332            << "\\textclass " << h_textclass << "\n";
 333         if (!h_preamble.str().empty())
 334                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 335         if (!h_options.empty())
 336                 os << "\\options " << h_options << "\n";
 337         os << "\\language " << h_language << "\n"
 338            << "\\inputencoding " << h_inputencoding << "\n"
 339            << "\\font_roman " << h_font_roman << "\n"
 340            << "\\font_sans " << h_font_sans << "\n"
 341            << "\\font_typewriter " << h_font_typewriter << "\n"
 342            << "\\font_default_family " << h_font_default_family << "\n"
 343            << "\\font_sc " << h_font_sc << "\n"
 344            << "\\font_osf " << h_font_osf << "\n"
 345            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 346            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 347            << "\\graphics " << h_graphics << "\n"
 348            << "\\paperfontsize " << h_paperfontsize << "\n"
 349            << "\\spacing " << h_spacing << "\n"
 350            << "\\papersize " << h_papersize << "\n"
 351            << "\\use_geometry " << h_use_geometry << "\n"
 352            << "\\use_amsmath " << h_use_amsmath << "\n"
 353            << "\\cite_engine " << h_cite_engine << "\n"
 354            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 355            << "\\paperorientation " << h_paperorientation << "\n"
 356            << "\\secnumdepth " << h_secnumdepth << "\n"
 357            << "\\tocdepth " << h_tocdepth << "\n"
 358            << "\\paragraph_separation " << h_paragraph_separation << "\n"
 359            << "\\defskip " << h_defskip << "\n"
 360            << "\\quotes_language " << h_quotes_language << "\n"
 361            << "\\papercolumns " << h_papercolumns << "\n"
 362            << "\\papersides " << h_papersides << "\n"
 363            << "\\paperpagestyle " << h_paperpagestyle << "\n"
 364            << "\\tracking_changes " << h_tracking_changes << "\n"
 365            << "\\output_changes " << h_output_changes << "\n"
 366            << "\\end_header\n\n"
 367            << "\\begin_body\n";
 368         // clear preamble for subdocuments
 369         h_preamble.str("");
 370 }
 371
 372 } // anonymous namespace
 373
 374 TextClass const parse_preamble(Parser & p, ostream & os, string const & forceclass)
 375 {
 376         // initialize fixed types
 377         special_columns['D'] = 3;
 378         bool is_full_document = false;
 379
 380         // determine whether this is a full document or a fragment for inclusion
 381         while (p.good()) {
 382                 Token const & t = p.get_token();
 383
 384                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 385                         is_full_document = true;
 386                         break;
 387                 }
 388         }
 389         p.reset();
 390
 391         while (is_full_document && p.good()) {
 392                 Token const & t = p.get_token();
 393
 394 #ifdef FILEDEBUG
 395                 cerr << "t: " << t << "\n";
 396 #endif
 397
 398                 //
 399                 // cat codes
 400                 //
 401                 if (t.cat() == catLetter ||
 402                           t.cat() == catSuper ||
 403                           t.cat() == catSub ||
 404                           t.cat() == catOther ||
 405                           t.cat() == catMath ||
 406                           t.cat() == catActive ||
 407                           t.cat() == catBegin ||
 408                           t.cat() == catEnd ||
 409                           t.cat() == catAlign ||
 410                           t.cat() == catParameter)
 411                 h_preamble << t.character();
 412
 413                 else if (t.cat() == catSpace || t.cat() == catNewline)
 414                         h_preamble << t.asInput();
 415
 416                 else if (t.cat() == catComment)
 417                         h_preamble << t.asInput();
 418
 419                 else if (t.cs() == "pagestyle")
 420                         h_paperpagestyle = p.verbatim_item();
 421
 422                 else if (t.cs() == "makeatletter") {
 423                         p.setCatCode('@', catLetter);
 424                         h_preamble << "\\makeatletter";
 425                 }
 426
 427                 else if (t.cs() == "makeatother") {
 428                         p.setCatCode('@', catOther);
 429                         h_preamble << "\\makeatother";
 430                 }
 431
 432                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 433                             || t.cs() == "providecommand") {
 434                         bool star = false;
 435                         if (p.next_token().character() == '*') {
 436                                 p.get_token();
 437                                 star = true;
 438                         }
 439                         string const name = p.verbatim_item();
 440                         string const opt1 = p.getOpt();
 441                         string const opt2 = p.getFullOpt();
 442                         string const body = p.verbatim_item();
 443                         // font settings
 444                         if (name == "\\rmdefault")
 445                                 if (is_known(body, known_roman_fonts))
 446                                         h_font_roman = body;
 447
 448                         if (name == "\\sfdefault")
 449                                 if (is_known(body, known_sans_fonts))
 450                                         h_font_sans = body;
 451
 452                         if (name == "\\ttdefault")
 453                                 if (is_known(body, known_typewriter_fonts))
 454                                         h_font_typewriter = body;
 455
 456                         if (name == "\\familydefault") {
 457                                 string family = body;
 458                                 // remove leading "\"
 459                                 h_font_default_family = family.erase(0,1);
 460                         }
 461                         // only non-lyxspecific stuff
 462                         if (   name != "\\noun"
 463                             && name != "\\tabularnewline"
 464                             && name != "\\LyX"
 465                             && name != "\\lyxline"
 466                             && name != "\\lyxaddress"
 467                             && name != "\\lyxrightaddress"
 468                             && name != "\\lyxdot"
 469                             && name != "\\boldsymbol"
 470                             && name != "\\lyxarrow"
 471                             && name != "\\rmdefault"
 472                             && name != "\\sfdefault"
 473                             && name != "\\ttdefault"
 474                             && name != "\\familydefault") {
 475                                 ostringstream ss;
 476                                 ss << '\\' << t.cs();
 477                                 if (star)
 478                                         ss << '*';
 479                                 ss << '{' << name << '}' << opt1 << opt2
 480                                    << '{' << body << "}";
 481                                 h_preamble << ss.str();
 482
 483                                 // Add the command to the known commands
 484                                 add_known_command(name, opt1, !opt2.empty());
 485 /*
 486                                 ostream & out = in_preamble ? h_preamble : os;
 487                                 out << "\\" << t.cs() << "{" << name << "}"
 488                                     << opts << "{" << body << "}";
 489 */
 490                         }
 491                 }
 492
 493                 else if (t.cs() == "documentclass") {
 494                         vector<string> opts = split_options(p.getArg('[', ']'));
 495                         handle_opt(opts, known_languages, h_language);
 496                         if (is_known(h_language, known_french_languages))
 497                                 h_language = "french";
 498                         else if (is_known(h_language, known_german_languages))
 499                                 h_language = "german";
 500                         else if (is_known(h_language, known_ngerman_languages))
 501                                 h_language = "ngerman";
 502                         else if (is_known(h_language, known_russian_languages))
 503                                 h_language = "russian";
 504                         else if (is_known(h_language, known_ukrainian_languages))
 505                                 h_language = "ukrainian";
 506                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 507                         // delete "pt" at the end
 508                         string::size_type i = h_paperfontsize.find("pt");
 509                         if (i != string::npos)
 510                                 h_paperfontsize.erase(i);
 511                         h_quotes_language = h_language;
 512                         h_options = join(opts, ",");
 513                         h_textclass = p.getArg('{', '}');
 514                 }
 515
 516                 else if (t.cs() == "usepackage") {
 517                         string const options = p.getArg('[', ']');
 518                         string const name = p.getArg('{', '}');
 519                         if (options.empty() && name.find(',')) {
 520                                 vector<string> vecnames;
 521                                 split(name, vecnames, ',');
 522                                 vector<string>::const_iterator it  = vecnames.begin();
 523                                 vector<string>::const_iterator end = vecnames.end();
 524                                 for (; it != end; ++it)
 525                                         handle_package(trim(*it), string());
 526                         } else {
 527                                 handle_package(name, options);
 528                         }
 529                 }
 530
 531                 else if (t.cs() == "newenvironment") {
 532                         string const name = p.getArg('{', '}');
 533                         ostringstream ss;
 534                         ss << "\\newenvironment{" << name << "}";
 535                         ss << p.getOpt();
 536                         ss << p.getOpt();
 537                         ss << '{' << p.verbatim_item() << '}';
 538                         ss << '{' << p.verbatim_item() << '}';
 539                         if (name != "lyxcode" && name != "lyxlist" &&
 540                             name != "lyxrightadress" &&
 541                             name != "lyxaddress" && name != "lyxgreyedout")
 542                                 h_preamble << ss.str();
 543                 }
 544
 545                 else if (t.cs() == "def") {
 546                         string name = p.get_token().cs();
 547                         while (p.next_token().cat() != catBegin)
 548                                 name += p.get_token().asString();
 549                         h_preamble << "\\def\\" << name << '{'
 550                                    << p.verbatim_item() << "}";
 551                 }
 552
 553                 else if (t.cs() == "newcolumntype") {
 554                         string const name = p.getArg('{', '}');
 555                         trim(name);
 556                         int nargs = 0;
 557                         string opts = p.getOpt();
 558                         if (!opts.empty()) {
 559                                 istringstream is(string(opts, 1));
 560                                 is >> nargs;
 561                         }
 562                         special_columns[name[0]] = nargs;
 563                         h_preamble << "\\newcolumntype{" << name << "}";
 564                         if (nargs)
 565                                 h_preamble << "[" << nargs << "]";
 566                         h_preamble << "{" << p.verbatim_item() << "}";
 567                 }
 568
 569                 else if (t.cs() == "setcounter") {
 570                         string const name = p.getArg('{', '}');
 571                         string const content = p.getArg('{', '}');
 572                         if (name == "secnumdepth")
 573                                 h_secnumdepth = content;
 574                         else if (name == "tocdepth")
 575                                 h_tocdepth = content;
 576                         else
 577                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
 578                 }
 579
 580                 else if (t.cs() == "setlength") {
 581                         string const name = p.verbatim_item();
 582                         string const content = p.verbatim_item();
 583                         // Is this correct?
 584                         if (name == "parskip")
 585                                 h_paragraph_separation = "skip";
 586                         else if (name == "parindent")
 587                                 h_paragraph_separation = "skip";
 588                         else
 589                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
 590                 }
 591
 592                 else if (t.cs() == "begin") {
 593                         string const name = p.getArg('{', '}');
 594                         if (name == "document")
 595                                 break;
 596                         h_preamble << "\\begin{" << name << "}";
 597                 }
 598
 599                 else if (t.cs() == "jurabibsetup") {
 600                         vector<string> jurabibsetup =
 601                                 split_options(p.getArg('{', '}'));
 602                         // add jurabibsetup to the jurabib package options
 603                         add_package("jurabib", jurabibsetup);
 604                         if (!jurabibsetup.empty()) {
 605                                 h_preamble << "\\jurabibsetup{"
 606                                            << join(jurabibsetup, ",") << '}';
 607                         }
 608                 }
 609
 610                 else if (!t.cs().empty())
 611                         h_preamble << '\\' << t.cs();
 612         }
 613         p.skip_spaces();
 614
 615         // Force textclass if the user wanted it
 616         if (!forceclass.empty())
 617                 h_textclass = forceclass;
 618         if (noweb_mode && !lyx::support::prefixIs(h_textclass, "literate-"))
 619                 h_textclass.insert(0, "literate-");
 620         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
 621         if (layoutfilename.empty()) {
 622                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
 623                 exit(1);
 624         }
 625         TextClass textclass;
 626         textclass.read(layoutfilename);
 627         if (h_papersides.empty()) {
 628                 ostringstream ss;
 629                 ss << textclass.sides();
 630                 h_papersides = ss.str();
 631         }
 632         end_preamble(os, textclass);
 633         return textclass;
 634 }
 635
 636 // }])
 637
 638
 639 } // namespace lyx