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/filetools.h"
  22 #include "support/lstrings.h"
  23
  24 #include <algorithm>
  25 #include <iostream>
  26 #include <sstream>
  27 #include <string>
  28 #include <vector>
  29 #include <map>
  30
  31
  32 namespace lyx {
  33
  34 using std::istringstream;
  35 using std::ostream;
  36 using std::ostringstream;
  37 using std::string;
  38 using std::vector;
  39 using std::cerr;
  40 using std::endl;
  41 using std::find;
  42
  43 using support::FileName;
  44 using support::libFileSearch;
  45
  46 // special columntypes
  47 extern std::map<char, int> special_columns;
  48
  49 std::map<string, vector<string> > used_packages;
  50
  51 namespace {
  52
  53 const char * const known_languages[] = { "austrian", "babel", "bahasa",
  54 "basque", "breton", "british", "bulgarian", "catalan", "croatian", "czech",
  55 "danish", "dutch", "english", "esperanto", "estonian", "finnish",
  56 "francais", "french", "frenchb", "frenchle", "frenchpro",
  57 "galician", "german", "germanb", "greek", "hebcal", "hebfont",
  58 "hebrew", "hebrew_newcode", "hebrew_oldcode", "hebrew_p", "hyphen",
  59 "icelandic", "irish", "italian", "latin", "lgrcmr", "lgrcmro", "lgrcmss",
  60 "lgrcmtt", "lgrenc", "lgrlcmss", "lgrlcmtt", "lheclas", "lhecmr",
  61 "lhecmss", "lhecmtt", "lhecrml", "lheenc", "lhefr", "lheredis", "lheshold",
  62 "lheshscr", "lheshstk", "lsorbian", "magyar", "naustrian", "ngermanb",
  63 "ngerman", "norsk", "polish", "portuges", "rlbabel", "romanian",
  64 "russianb", "samin", "scottish", "serbian", "slovak", "slovene", "spanish",
  65 "swedish", "turkish", "ukraineb", "usorbian", "welsh", 0};
  66
  67 const char * const known_french_languages[] = {"french", "frenchb", "francais",
  68                                                "frenchle", "frenchpro", 0};
  69 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
  70
  71 // some ugly stuff
  72 ostringstream h_preamble;
  73 string h_textclass               = "article";
  74 string h_options                 = string();
  75 string h_language                = "english";
  76 string h_inputencoding           = "auto";
  77 string h_fontscheme              = "default";
  78 string h_graphics                = "default";
  79 string h_paperfontsize           = "default";
  80 string h_spacing                 = "single";
  81 string h_papersize               = "default";
  82 string h_use_geometry            = "false";
  83 string h_use_amsmath             = "0";
  84 string h_cite_engine             = "basic";
  85 string h_use_bibtopic            = "false";
  86 string h_paperorientation        = "portrait";
  87 string h_secnumdepth             = "3";
  88 string h_tocdepth                = "3";
  89 string h_paragraph_separation    = "indent";
  90 string h_defskip                 = "medskip";
  91 string h_quotes_language         = "english";
  92 string h_papercolumns            = "1";
  93 string h_papersides              = string();
  94 string h_paperpagestyle          = "default";
  95 string h_tracking_changes        = "false";
  96 string h_output_changes          = "false";
  97
  98
  99 void handle_opt(vector<string> & opts, char const * const * what, string & target)
 100 {
 101         if (opts.empty())
 102                 return;
 103
 104         // the last language option is the document language (for babel and LyX)
 105         // the last size option is the document font size
 106         vector<string>::iterator it;
 107         vector<string>::iterator position = opts.begin();
 108         for (; *what; ++what) {
 109                 it = find(opts.begin(), opts.end(), *what);
 110                 if (it != opts.end()) {
 111                         if (it >= position) {
 112                                 target = *what;
 113                                 position = it;
 114                         }
 115                         // remove found options from the list
 116                         opts.erase(it);
 117                 }
 118         }
 119 }
 120
 121
 122 /*!
 123  * Split a package options string (keyval format) into a vector.
 124  * Example input:
 125  *   authorformat=smallcaps,
 126  *   commabeforerest,
 127  *   titleformat=colonsep,
 128  *   bibformat={tabular,ibidem,numbered}
 129  */
 130 vector<string> split_options(string const & input)
 131 {
 132         vector<string> options;
 133         string option;
 134         Parser p(input);
 135         while (p.good()) {
 136                 Token const & t = p.get_token();
 137                 if (t.asInput() == ",") {
 138                         options.push_back(trim(option));
 139                         option.erase();
 140                 } else if (t.asInput() == "=") {
 141                         option += '=';
 142                         p.skip_spaces(true);
 143                         if (p.next_token().asInput() == "{")
 144                                 option += '{' + p.getArg('{', '}') + '}';
 145                 } else if (t.cat() != catSpace)
 146                         option += t.asInput();
 147         }
 148
 149         if (!option.empty())
 150                 options.push_back(trim(option));
 151
 152         return options;
 153 }
 154
 155
 156 /*!
 157  * Add package \p name with options \p options to used_packages.
 158  * Remove options from \p options that we don't want to output.
 159  */
 160 void add_package(string const & name, vector<string> & options)
 161 {
 162         // every package inherits the global options
 163         if (used_packages.find(name) == used_packages.end())
 164                 used_packages[name] = split_options(h_options);
 165
 166         vector<string> & v = used_packages[name];
 167         v.insert(v.end(), options.begin(), options.end());
 168         if (name == "jurabib") {
 169                 // Don't output the order argument (see the cite command
 170                 // handling code in text.cpp).
 171                 vector<string>::iterator end =
 172                         remove(options.begin(), options.end(), "natbiborder");
 173                 end = remove(options.begin(), end, "jurabiborder");
 174                 options.erase(end, options.end());
 175         }
 176 }
 177
 178
 179 void handle_package(string const & name, string const & opts)
 180 {
 181         vector<string> options = split_options(opts);
 182         add_package(name, options);
 183
 184         //cerr << "handle_package: '" << name << "'\n";
 185         if (name == "ae")
 186                 h_fontscheme = "ae";
 187         else if (name == "aecompl")
 188                 h_fontscheme = "ae";
 189         else if (name == "amsmath")
 190                 h_use_amsmath = "1";
 191         else if (name == "amssymb")
 192                 h_use_amsmath = "1";
 193         else if (name == "babel")
 194                 ; // ignore this
 195         else if (name == "fontenc")
 196                 ; // ignore this
 197         else if (name == "inputenc") {
 198                 // only set when there is not more than one inputenc option
 199                 // therefore check for the "," character
 200                 if (opts.find(",", 0) == string::npos)
 201                         h_inputencoding = opts;
 202                 options.clear();
 203         } else if (name == "makeidx")
 204                 ; // ignore this
 205         else if (name == "verbatim")
 206                 ; // ignore this
 207         else if (name == "graphicx")
 208                 ; // ignore this
 209         else if (is_known(name, known_languages)) {
 210                 if (is_known(name, known_french_languages)) {
 211                         h_language = "french";
 212                         h_quotes_language = "french";
 213                 } else {
 214                         h_language = name;
 215                         h_quotes_language = name;
 216                 }
 217
 218         } else if (name == "natbib") {
 219                 h_cite_engine = "natbib_authoryear";
 220                 vector<string>::iterator it =
 221                         find(options.begin(), options.end(), "authoryear");
 222                 if (it != options.end())
 223                         options.erase(it);
 224                 else {
 225                         it = find(options.begin(), options.end(), "numbers");
 226                         if (it != options.end()) {
 227                                 h_cite_engine = "natbib_numerical";
 228                                 options.erase(it);
 229                         }
 230                 }
 231         } else if (name == "jurabib") {
 232                 h_cite_engine = "jurabib";
 233         } else if (options.empty())
 234                 h_preamble << "\\usepackage{" << name << "}\n";
 235         else {
 236                 h_preamble << "\\usepackage[" << opts << "]{" << name << "}\n";
 237                 options.clear();
 238         }
 239
 240         // We need to do something with the options...
 241         if (!options.empty())
 242                 cerr << "Ignoring options '" << join(options, ",")
 243                      << "' of package " << name << '.' << endl;
 244 }
 245
 246
 247
 248 void end_preamble(ostream & os, TextClass const & /*textclass*/)
 249 {
 250         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
 251            << "\\lyxformat 246\n"
 252            << "\\begin_document\n"
 253            << "\\begin_header\n"
 254            << "\\textclass " << h_textclass << "\n";
 255         if (!h_preamble.str().empty())
 256                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 257         if (!h_options.empty())
 258                 os << "\\options " << h_options << "\n";
 259         os << "\\language " << h_language << "\n"
 260            << "\\inputencoding " << h_inputencoding << "\n"
 261            << "\\fontscheme " << h_fontscheme << "\n"
 262            << "\\graphics " << h_graphics << "\n"
 263            << "\\paperfontsize " << h_paperfontsize << "\n"
 264            << "\\spacing " << h_spacing << "\n"
 265            << "\\papersize " << h_papersize << "\n"
 266            << "\\use_geometry " << h_use_geometry << "\n"
 267            << "\\use_amsmath " << h_use_amsmath << "\n"
 268            << "\\cite_engine " << h_cite_engine << "\n"
 269            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 270            << "\\paperorientation " << h_paperorientation << "\n"
 271            << "\\secnumdepth " << h_secnumdepth << "\n"
 272            << "\\tocdepth " << h_tocdepth << "\n"
 273            << "\\paragraph_separation " << h_paragraph_separation << "\n"
 274            << "\\defskip " << h_defskip << "\n"
 275            << "\\quotes_language " << h_quotes_language << "\n"
 276            << "\\papercolumns " << h_papercolumns << "\n"
 277            << "\\papersides " << h_papersides << "\n"
 278            << "\\paperpagestyle " << h_paperpagestyle << "\n"
 279            << "\\tracking_changes " << h_tracking_changes << "\n"
 280            << "\\output_changes " << h_output_changes << "\n"
 281            << "\\end_header\n\n"
 282            << "\\begin_body\n";
 283         // clear preamble for subdocuments
 284         h_preamble.str("");
 285 }
 286
 287 } // anonymous namespace
 288
 289 TextClass const parse_preamble(Parser & p, ostream & os, string const & forceclass)
 290 {
 291         // initialize fixed types
 292         special_columns['D'] = 3;
 293         bool is_full_document = false;
 294
 295         // determine whether this is a full document or a fragment for inclusion
 296         while (p.good()) {
 297                 Token const & t = p.get_token();
 298
 299                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 300                         is_full_document = true;
 301                         break;
 302                 }
 303         }
 304         p.reset();
 305
 306         while (is_full_document && p.good()) {
 307                 Token const & t = p.get_token();
 308
 309 #ifdef FILEDEBUG
 310                 cerr << "t: " << t << "\n";
 311 #endif
 312
 313                 //
 314                 // cat codes
 315                 //
 316                 if (t.cat() == catLetter ||
 317                           t.cat() == catSuper ||
 318                           t.cat() == catSub ||
 319                           t.cat() == catOther ||
 320                           t.cat() == catMath ||
 321                           t.cat() == catActive ||
 322                           t.cat() == catBegin ||
 323                           t.cat() == catEnd ||
 324                           t.cat() == catAlign ||
 325                           t.cat() == catParameter)
 326                 h_preamble << t.character();
 327
 328                 else if (t.cat() == catSpace || t.cat() == catNewline)
 329                         h_preamble << t.asInput();
 330
 331                 else if (t.cat() == catComment)
 332                         h_preamble << t.asInput();
 333
 334                 else if (t.cs() == "pagestyle")
 335                         h_paperpagestyle = p.verbatim_item();
 336
 337                 else if (t.cs() == "makeatletter") {
 338                         p.setCatCode('@', catLetter);
 339                         h_preamble << "\\makeatletter";
 340                 }
 341
 342                 else if (t.cs() == "makeatother") {
 343                         p.setCatCode('@', catOther);
 344                         h_preamble << "\\makeatother";
 345                 }
 346
 347                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 348                             || t.cs() == "providecommand") {
 349                         bool star = false;
 350                         if (p.next_token().character() == '*') {
 351                                 p.get_token();
 352                                 star = true;
 353                         }
 354                         string const name = p.verbatim_item();
 355                         string const opt1 = p.getOpt();
 356                         string const opt2 = p.getFullOpt();
 357                         string const body = p.verbatim_item();
 358                         // only non-lyxspecific stuff
 359                         if (   name != "\\noun"
 360                             && name != "\\tabularnewline"
 361                             && name != "\\LyX"
 362                             && name != "\\lyxline"
 363                             && name != "\\lyxaddress"
 364                             && name != "\\lyxrightaddress"
 365                             && name != "\\lyxdot"
 366                             && name != "\\boldsymbol"
 367                             && name != "\\lyxarrow") {
 368                                 ostringstream ss;
 369                                 ss << '\\' << t.cs();
 370                                 if (star)
 371                                         ss << '*';
 372                                 ss << '{' << name << '}' << opt1 << opt2
 373                                    << '{' << body << "}";
 374                                 h_preamble << ss.str();
 375
 376                                 // Add the command to the known commands
 377                                 add_known_command(name, opt1, !opt2.empty());
 378 /*
 379                                 ostream & out = in_preamble ? h_preamble : os;
 380                                 out << "\\" << t.cs() << "{" << name << "}"
 381                                     << opts << "{" << body << "}";
 382 */
 383                         }
 384                 }
 385
 386                 else if (t.cs() == "documentclass") {
 387                         vector<string> opts = split_options(p.getArg('[', ']'));
 388                         handle_opt(opts, known_languages, h_language);
 389                         if (is_known(h_language, known_french_languages))
 390                                 h_language = "french";
 391                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 392                         // delete "pt" at the end
 393                         string::size_type i = h_paperfontsize.find("pt");
 394                         if (i != string::npos)
 395                                 h_paperfontsize.erase(i);
 396                         h_quotes_language = h_language;
 397                         h_options = join(opts, ",");
 398                         h_textclass = p.getArg('{', '}');
 399                 }
 400
 401                 else if (t.cs() == "usepackage") {
 402                         string const options = p.getArg('[', ']');
 403                         string const name = p.getArg('{', '}');
 404                         if (options.empty() && name.find(',')) {
 405                                 vector<string> vecnames;
 406                                 split(name, vecnames, ',');
 407                                 vector<string>::const_iterator it  = vecnames.begin();
 408                                 vector<string>::const_iterator end = vecnames.end();
 409                                 for (; it != end; ++it)
 410                                         handle_package(trim(*it), string());
 411                         } else {
 412                                 handle_package(name, options);
 413                         }
 414                 }
 415
 416                 else if (t.cs() == "newenvironment") {
 417                         string const name = p.getArg('{', '}');
 418                         ostringstream ss;
 419                         ss << "\\newenvironment{" << name << "}";
 420                         ss << p.getOpt();
 421                         ss << p.getOpt();
 422                         ss << '{' << p.verbatim_item() << '}';
 423                         ss << '{' << p.verbatim_item() << '}';
 424                         if (name != "lyxcode" && name != "lyxlist" &&
 425                             name != "lyxrightadress" &&
 426                             name != "lyxaddress" && name != "lyxgreyedout")
 427                                 h_preamble << ss.str();
 428                 }
 429
 430                 else if (t.cs() == "def") {
 431                         string name = p.get_token().cs();
 432                         while (p.next_token().cat() != catBegin)
 433                                 name += p.get_token().asString();
 434                         h_preamble << "\\def\\" << name << '{'
 435                                    << p.verbatim_item() << "}";
 436                 }
 437
 438                 else if (t.cs() == "newcolumntype") {
 439                         string const name = p.getArg('{', '}');
 440                         trim(name);
 441                         int nargs = 0;
 442                         string opts = p.getOpt();
 443                         if (!opts.empty()) {
 444                                 istringstream is(string(opts, 1));
 445                                 //cerr << "opt: " << is.str() << "\n";
 446                                 is >> nargs;
 447                         }
 448                         special_columns[name[0]] = nargs;
 449                         h_preamble << "\\newcolumntype{" << name << "}";
 450                         if (nargs)
 451                                 h_preamble << "[" << nargs << "]";
 452                         h_preamble << "{" << p.verbatim_item() << "}";
 453                 }
 454
 455                 else if (t.cs() == "setcounter") {
 456                         string const name = p.getArg('{', '}');
 457                         string const content = p.getArg('{', '}');
 458                         if (name == "secnumdepth")
 459                                 h_secnumdepth = content;
 460                         else if (name == "tocdepth")
 461                                 h_tocdepth = content;
 462                         else
 463                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
 464                 }
 465
 466                 else if (t.cs() == "setlength") {
 467                         string const name = p.verbatim_item();
 468                         string const content = p.verbatim_item();
 469                         // Is this correct?
 470                         if (name == "parskip")
 471                                 h_paragraph_separation = "skip";
 472                         else if (name == "parindent")
 473                                 h_paragraph_separation = "skip";
 474                         else
 475                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
 476                 }
 477
 478                 else if (t.cs() == "begin") {
 479                         string const name = p.getArg('{', '}');
 480                         if (name == "document")
 481                                 break;
 482                         h_preamble << "\\begin{" << name << "}";
 483                 }
 484
 485                 else if (t.cs() == "jurabibsetup") {
 486                         vector<string> jurabibsetup =
 487                                 split_options(p.getArg('{', '}'));
 488                         // add jurabibsetup to the jurabib package options
 489                         add_package("jurabib", jurabibsetup);
 490                         if (!jurabibsetup.empty()) {
 491                                 h_preamble << "\\jurabibsetup{"
 492                                            << join(jurabibsetup, ",") << '}';
 493                         }
 494                 }
 495
 496                 else if (!t.cs().empty())
 497                         h_preamble << '\\' << t.cs();
 498         }
 499         p.skip_spaces();
 500
 501         // Force textclass if the user wanted it
 502         if (!forceclass.empty())
 503                 h_textclass = forceclass;
 504         if (noweb_mode && !lyx::support::prefixIs(h_textclass, "literate-"))
 505                 h_textclass.insert(0, "literate-");
 506         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
 507         if (layoutfilename.empty()) {
 508                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
 509                 exit(1);
 510         }
 511         TextClass textclass;
 512         textclass.read(layoutfilename);
 513         if (h_papersides.empty()) {
 514                 ostringstream ss;
 515                 ss << textclass.sides();
 516                 h_papersides = ss.str();
 517         }
 518         end_preamble(os, textclass);
 519         return textclass;
 520 }
 521
 522 // }])
 523
 524
 525 } // namespace lyx