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 // Given is a string like "scaled=0.9", return 0.9 * 100
 204 string const scale_as_percentage(string const & scale)
 205 {
 206         string::size_type pos = scale.find('=');
 207         if (pos != string::npos) {
 208                 string value = scale.substr(pos + 1);
 209                 if (isStrDbl(value))
 210                         return convert<string>(100 * convert<double>(value));
 211         }
 212         // If the input string didn't match our expectations.
 213         // return the default value "100"
 214         return "100";
 215 }
 216
 217
 218 void handle_package(string const & name, string const & opts)
 219 {
 220         vector<string> options = split_options(opts);
 221         add_package(name, options);
 222         string scale;
 223
 224         // roman fonts
 225         if (is_known(name, known_roman_fonts))
 226                 h_font_roman = name;
 227         if (name == "fourier") {
 228                 h_font_roman = "utopia";
 229                 // when font uses real small capitals
 230                 if (opts == "expert")
 231                         h_font_sc = "true";
 232         }
 233         if (name == "mathpazo")
 234                 h_font_roman = "palatino";
 235         if (name == "mathptmx")
 236                 h_font_roman = "times";
 237         // sansserif fonts
 238         if (is_known(name, known_sans_fonts)) {
 239                 h_font_sans = name;
 240                 if (!opts.empty()) {
 241                         scale = opts;
 242                         h_font_sf_scale = scale_as_percentage(scale);
 243                 }
 244         }
 245         // typewriter fonts
 246         if (is_known(name, known_typewriter_fonts)) {
 247                 h_font_typewriter = name;
 248                 if (!opts.empty()) {
 249                         scale = opts;
 250                         h_font_tt_scale = scale_as_percentage(scale);
 251                 }
 252         }
 253         // font uses old-style figure
 254         if (name == "eco")
 255                 h_font_osf = "true";
 256
 257         else if (name == "amsmath" || name == "amssymb")
 258                 h_use_amsmath = "1";
 259         else if (name == "babel")
 260                 ; // ignore this
 261         else if (name == "fontenc")
 262                 ; // ignore this
 263         else if (name == "inputenc") {
 264                 // only set when there is not more than one inputenc option
 265                 // therefore check for the "," character
 266                 if (opts.find(",") == string::npos)
 267                         h_inputencoding = opts;
 268                 options.clear();
 269         } else if (name == "makeidx")
 270                 ; // ignore this
 271         else if (name == "verbatim")
 272                 ; // ignore this
 273         else if (name == "graphicx")
 274                 ; // ignore this
 275         else if (is_known(name, known_languages)) {
 276                 if (is_known(name, known_french_languages))
 277                         h_language = "french";
 278                 else if (is_known(name, known_german_languages))
 279                         h_language = "german";
 280                 else if (is_known(name, known_ngerman_languages))
 281                         h_language = "ngerman";
 282                 else if (is_known(name, known_russian_languages))
 283                         h_language = "russian";
 284                 else if (is_known(name, known_ukrainian_languages))
 285                         h_language = "ukrainian";
 286                 else
 287                         h_language = name;
 288                 h_quotes_language = h_language;
 289
 290         } else if (name == "natbib") {
 291                 h_cite_engine = "natbib_authoryear";
 292                 vector<string>::iterator it =
 293                         find(options.begin(), options.end(), "authoryear");
 294                 if (it != options.end())
 295                         options.erase(it);
 296                 else {
 297                         it = find(options.begin(), options.end(), "numbers");
 298                         if (it != options.end()) {
 299                                 h_cite_engine = "natbib_numerical";
 300                                 options.erase(it);
 301                         }
 302                 }
 303         } else if (name == "jurabib") {
 304                 h_cite_engine = "jurabib";
 305         } else if (options.empty())
 306                 h_preamble << "\\usepackage{" << name << "}\n";
 307         else {
 308                 h_preamble << "\\usepackage[" << opts << "]{" << name << "}\n";
 309                 options.clear();
 310         }
 311
 312         // We need to do something with the options...
 313         if (!options.empty())
 314                 cerr << "Ignoring options '" << join(options, ",")
 315                      << "' of package " << name << '.' << endl;
 316 }
 317
 318
 319
 320 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 321 {
 322         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 323            << "\\lyxformat 247\n"
 324            << "\\begin_document\n"
 325            << "\\begin_header\n"
 326            << "\\textclass " << h_textclass << "\n";
 327         if (!h_preamble.str().empty())
 328                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 329         if (!h_options.empty())
 330                 os << "\\options " << h_options << "\n";
 331         os << "\\language " << h_language << "\n"
 332            << "\\inputencoding " << h_inputencoding << "\n"
 333            << "\\font_roman " << h_font_roman << "\n"
 334            << "\\font_sans " << h_font_sans << "\n"
 335            << "\\font_typewriter " << h_font_typewriter << "\n"
 336            << "\\font_default_family " << h_font_default_family << "\n"
 337            << "\\font_sc " << h_font_sc << "\n"
 338            << "\\font_osf " << h_font_osf << "\n"
 339            << "\\font_sf_scale " << h_font_sf_scale << "\n"
 340            << "\\font_tt_scale " << h_font_tt_scale << "\n"
 341            << "\\graphics " << h_graphics << "\n"
 342            << "\\paperfontsize " << h_paperfontsize << "\n"
 343            << "\\spacing " << h_spacing << "\n"
 344            << "\\papersize " << h_papersize << "\n"
 345            << "\\use_geometry " << h_use_geometry << "\n"
 346            << "\\use_amsmath " << h_use_amsmath << "\n"
 347            << "\\cite_engine " << h_cite_engine << "\n"
 348            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 349            << "\\paperorientation " << h_paperorientation << "\n"
 350            << "\\secnumdepth " << h_secnumdepth << "\n"
 351            << "\\tocdepth " << h_tocdepth << "\n"
 352            << "\\paragraph_separation " << h_paragraph_separation << "\n"
 353            << "\\defskip " << h_defskip << "\n"
 354            << "\\quotes_language " << h_quotes_language << "\n"
 355            << "\\papercolumns " << h_papercolumns << "\n"
 356            << "\\papersides " << h_papersides << "\n"
 357            << "\\paperpagestyle " << h_paperpagestyle << "\n"
 358            << "\\tracking_changes " << h_tracking_changes << "\n"
 359            << "\\output_changes " << h_output_changes << "\n"
 360            << "\\end_header\n\n"
 361            << "\\begin_body\n";
 362         // clear preamble for subdocuments
 363         h_preamble.str("");
 364 }
 365
 366 } // anonymous namespace
 367
 368 TextClass const parse_preamble(Parser & p, ostream & os, string const & forceclass)
 369 {
 370         // initialize fixed types
 371         special_columns['D'] = 3;
 372         bool is_full_document = false;
 373
 374         // determine whether this is a full document or a fragment for inclusion
 375         while (p.good()) {
 376                 Token const & t = p.get_token();
 377
 378                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 379                         is_full_document = true;
 380                         break;
 381                 }
 382         }
 383         p.reset();
 384
 385         while (is_full_document && p.good()) {
 386                 Token const & t = p.get_token();
 387
 388 #ifdef FILEDEBUG
 389                 cerr << "t: " << t << "\n";
 390 #endif
 391
 392                 //
 393                 // cat codes
 394                 //
 395                 if (t.cat() == catLetter ||
 396                           t.cat() == catSuper ||
 397                           t.cat() == catSub ||
 398                           t.cat() == catOther ||
 399                           t.cat() == catMath ||
 400                           t.cat() == catActive ||
 401                           t.cat() == catBegin ||
 402                           t.cat() == catEnd ||
 403                           t.cat() == catAlign ||
 404                           t.cat() == catParameter)
 405                 h_preamble << t.character();
 406
 407                 else if (t.cat() == catSpace || t.cat() == catNewline)
 408                         h_preamble << t.asInput();
 409
 410                 else if (t.cat() == catComment)
 411                         h_preamble << t.asInput();
 412
 413                 else if (t.cs() == "pagestyle")
 414                         h_paperpagestyle = p.verbatim_item();
 415
 416                 else if (t.cs() == "makeatletter") {
 417                         p.setCatCode('@', catLetter);
 418                 }
 419
 420                 else if (t.cs() == "makeatother") {
 421                         p.setCatCode('@', catOther);
 422                 }
 423
 424                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 425                             || t.cs() == "providecommand") {
 426                         bool star = false;
 427                         if (p.next_token().character() == '*') {
 428                                 p.get_token();
 429                                 star = true;
 430                         }
 431                         string const name = p.verbatim_item();
 432                         string const opt1 = p.getOpt();
 433                         string const opt2 = p.getFullOpt();
 434                         string const body = p.verbatim_item();
 435                         // font settings
 436                         if (name == "\\rmdefault")
 437                                 if (is_known(body, known_roman_fonts))
 438                                         h_font_roman = body;
 439
 440                         if (name == "\\sfdefault")
 441                                 if (is_known(body, known_sans_fonts))
 442                                         h_font_sans = body;
 443
 444                         if (name == "\\ttdefault")
 445                                 if (is_known(body, known_typewriter_fonts))
 446                                         h_font_typewriter = body;
 447
 448                         if (name == "\\familydefault") {
 449                                 string family = body;
 450                                 // remove leading "\"
 451                                 h_font_default_family = family.erase(0,1);
 452                         }
 453                         // only non-lyxspecific stuff
 454                         if (   name != "\\noun"
 455                             && name != "\\tabularnewline"
 456                             && name != "\\LyX"
 457                             && name != "\\lyxline"
 458                             && name != "\\lyxaddress"
 459                             && name != "\\lyxrightaddress"
 460                             && name != "\\lyxdot"
 461                             && name != "\\boldsymbol"
 462                             && name != "\\lyxarrow"
 463                             && name != "\\rmdefault"
 464                             && name != "\\sfdefault"
 465                             && name != "\\ttdefault"
 466                             && name != "\\familydefault") {
 467                                 ostringstream ss;
 468                                 ss << '\\' << t.cs();
 469                                 if (star)
 470                                         ss << '*';
 471                                 ss << '{' << name << '}' << opt1 << opt2
 472                                    << '{' << body << "}";
 473                                 h_preamble << ss.str();
 474
 475                                 // Add the command to the known commands
 476                                 add_known_command(name, opt1, !opt2.empty());
 477 /*
 478                                 ostream & out = in_preamble ? h_preamble : os;
 479                                 out << "\\" << t.cs() << "{" << name << "}"
 480                                     << opts << "{" << body << "}";
 481 */
 482                         }
 483                 }
 484
 485                 else if (t.cs() == "documentclass") {
 486                         vector<string> opts = split_options(p.getArg('[', ']'));
 487                         handle_opt(opts, known_languages, h_language);
 488                         if (is_known(h_language, known_french_languages))
 489                                 h_language = "french";
 490                         else if (is_known(h_language, known_german_languages))
 491                                 h_language = "german";
 492                         else if (is_known(h_language, known_ngerman_languages))
 493                                 h_language = "ngerman";
 494                         else if (is_known(h_language, known_russian_languages))
 495                                 h_language = "russian";
 496                         else if (is_known(h_language, known_ukrainian_languages))
 497                                 h_language = "ukrainian";
 498                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 499                         // delete "pt" at the end
 500                         string::size_type i = h_paperfontsize.find("pt");
 501                         if (i != string::npos)
 502                                 h_paperfontsize.erase(i);
 503                         h_quotes_language = h_language;
 504                         h_options = join(opts, ",");
 505                         h_textclass = p.getArg('{', '}');
 506                 }
 507
 508                 else if (t.cs() == "usepackage") {
 509                         string const options = p.getArg('[', ']');
 510                         string const name = p.getArg('{', '}');
 511                         if (options.empty() && name.find(',')) {
 512                                 vector<string> vecnames;
 513                                 split(name, vecnames, ',');
 514                                 vector<string>::const_iterator it  = vecnames.begin();
 515                                 vector<string>::const_iterator end = vecnames.end();
 516                                 for (; it != end; ++it)
 517                                         handle_package(trim(*it), string());
 518                         } else {
 519                                 handle_package(name, options);
 520                         }
 521                 }
 522
 523                 else if (t.cs() == "newenvironment") {
 524                         string const name = p.getArg('{', '}');
 525                         ostringstream ss;
 526                         ss << "\\newenvironment{" << name << "}";
 527                         ss << p.getOpt();
 528                         ss << p.getOpt();
 529                         ss << '{' << p.verbatim_item() << '}';
 530                         ss << '{' << p.verbatim_item() << '}';
 531                         if (name != "lyxcode" && name != "lyxlist" &&
 532                             name != "lyxrightadress" &&
 533                             name != "lyxaddress" && name != "lyxgreyedout")
 534                                 h_preamble << ss.str();
 535                 }
 536
 537                 else if (t.cs() == "def") {
 538                         string name = p.get_token().cs();
 539                         while (p.next_token().cat() != catBegin)
 540                                 name += p.get_token().asString();
 541                         h_preamble << "\\def\\" << name << '{'
 542                                    << p.verbatim_item() << "}";
 543                 }
 544
 545                 else if (t.cs() == "newcolumntype") {
 546                         string const name = p.getArg('{', '}');
 547                         trim(name);
 548                         int nargs = 0;
 549                         string opts = p.getOpt();
 550                         if (!opts.empty()) {
 551                                 istringstream is(string(opts, 1));
 552                                 is >> nargs;
 553                         }
 554                         special_columns[name[0]] = nargs;
 555                         h_preamble << "\\newcolumntype{" << name << "}";
 556                         if (nargs)
 557                                 h_preamble << "[" << nargs << "]";
 558                         h_preamble << "{" << p.verbatim_item() << "}";
 559                 }
 560
 561                 else if (t.cs() == "setcounter") {
 562                         string const name = p.getArg('{', '}');
 563                         string const content = p.getArg('{', '}');
 564                         if (name == "secnumdepth")
 565                                 h_secnumdepth = content;
 566                         else if (name == "tocdepth")
 567                                 h_tocdepth = content;
 568                         else
 569                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
 570                 }
 571
 572                 else if (t.cs() == "setlength") {
 573                         string const name = p.verbatim_item();
 574                         string const content = p.verbatim_item();
 575                         // Is this correct?
 576                         if (name == "parskip")
 577                                 h_paragraph_separation = "skip";
 578                         else if (name == "parindent")
 579                                 h_paragraph_separation = "skip";
 580                         else
 581                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
 582                 }
 583
 584                 else if (t.cs() == "begin") {
 585                         string const name = p.getArg('{', '}');
 586                         if (name == "document")
 587                                 break;
 588                         h_preamble << "\\begin{" << name << "}";
 589                 }
 590
 591                 else if (t.cs() == "jurabibsetup") {
 592                         vector<string> jurabibsetup =
 593                                 split_options(p.getArg('{', '}'));
 594                         // add jurabibsetup to the jurabib package options
 595                         add_package("jurabib", jurabibsetup);
 596                         if (!jurabibsetup.empty()) {
 597                                 h_preamble << "\\jurabibsetup{"
 598                                            << join(jurabibsetup, ",") << '}';
 599                         }
 600                 }
 601
 602                 else if (!t.cs().empty())
 603                         h_preamble << '\\' << t.cs();
 604         }
 605         p.skip_spaces();
 606
 607         // Force textclass if the user wanted it
 608         if (!forceclass.empty())
 609                 h_textclass = forceclass;
 610         if (noweb_mode && !lyx::support::prefixIs(h_textclass, "literate-"))
 611                 h_textclass.insert(0, "literate-");
 612         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
 613         if (layoutfilename.empty()) {
 614                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
 615                 exit(1);
 616         }
 617         TextClass textclass;
 618         textclass.read(layoutfilename);
 619         if (h_papersides.empty()) {
 620                 ostringstream ss;
 621                 ss << textclass.sides();
 622                 h_papersides = ss.str();
 623         }
 624         end_preamble(os, textclass);
 625         return textclass;
 626 }
 627
 628 // }])
 629
 630
 631 } // namespace lyx