src/tex2lyx/preamble.C

   1 /**
   2  * \file preamble.C
   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  *
   8  * Full author contact details are available in file CREDITS.
   9  */
  10
  11 // {[(
  12
  13 #include <config.h>
  14
  15 #include "tex2lyx.h"
  16
  17 #include "layout.h"
  18 #include "lyxtextclass.h"
  19 #include "lyxlex.h"
  20 #include "support/filetools.h"
  21
  22 #include <algorithm>
  23 #include <iostream>
  24 #include <sstream>
  25 #include <string>
  26 #include <vector>
  27 #include <map>
  28
  29 using std::istringstream;
  30 using std::ostream;
  31 using std::ostringstream;
  32 using std::string;
  33 using std::vector;
  34 using std::cerr;
  35 using std::endl;
  36
  37 using lyx::support::LibFileSearch;
  38
  39 // special columntypes
  40 extern std::map<char, int> special_columns;
  41
  42 std::map<string, vector<string> > used_packages;
  43
  44 namespace {
  45
  46 const char * const known_languages[] = { "austrian", "babel", "bahasa",
  47 "basque", "breton", "british", "bulgarian", "catalan", "croatian", "czech",
  48 "danish", "dutch", "english", "esperanto", "estonian", "finnish",
  49 "francais", "french", "frenchb", "frenchle", "frenchpro",
  50 "galician", "german", "germanb", "greek", "hebcal", "hebfont",
  51 "hebrew", "hebrew_newcode", "hebrew_oldcode", "hebrew_p", "hyphen",
  52 "icelandic", "irish", "italian", "latin", "lgrcmr", "lgrcmro", "lgrcmss",
  53 "lgrcmtt", "lgrenc", "lgrlcmss", "lgrlcmtt", "lheclas", "lhecmr",
  54 "lhecmss", "lhecmtt", "lhecrml", "lheenc", "lhefr", "lheredis", "lheshold",
  55 "lheshscr", "lheshstk", "lsorbian", "magyar", "naustrian", "ngermanb",
  56 "ngerman", "norsk", "polish", "portuges", "rlbabel", "romanian",
  57 "russianb", "samin", "scottish", "serbian", "slovak", "slovene", "spanish",
  58 "swedish", "turkish", "ukraineb", "usorbian", "welsh", 0};
  59
  60 const char * const known_french_languages[] = {"french", "frenchb", "francais",
  61                                                "frenchle", "frenchpro", 0};
  62 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
  63
  64 // some ugly stuff
  65 ostringstream h_preamble;
  66 string h_textclass               = "article";
  67 string h_options                 = string();
  68 string h_language                = "english";
  69 string h_inputencoding           = "latin1";
  70 string h_fontscheme              = "default";
  71 string h_graphics                = "default";
  72 string h_paperfontsize           = "default";
  73 string h_spacing                 = "single";
  74 string h_papersize               = "default";
  75 string h_use_geometry            = "false";
  76 string h_use_amsmath             = "0";
  77 string h_cite_engine             = "basic";
  78 string h_use_bibtopic            = "false";
  79 string h_paperorientation        = "portrait";
  80 string h_secnumdepth             = "3";
  81 string h_tocdepth                = "3";
  82 string h_paragraph_separation    = "indent";
  83 string h_defskip                 = "medskip";
  84 string h_quotes_language         = "english";
  85 string h_quotes_times            = "2";
  86 string h_papercolumns            = "1";
  87 string h_papersides              = string();
  88 string h_paperpagestyle          = "default";
  89 string h_tracking_changes        = "false";
  90 string h_output_changes          = "false";
  91
  92
  93 void handle_opt(vector<string> & opts, char const * const * what, string & target)
  94 {
  95         if (opts.empty())
  96                 return;
  97
  98         for ( ; *what; ++what) {
  99                 vector<string>::iterator it = find(opts.begin(), opts.end(), *what);
 100                 if (it != opts.end()) {
 101                         //cerr << "### found option '" << *what << "'\n";
 102                         target = *what;
 103                         opts.erase(it);
 104                         return;
 105                 }
 106         }
 107 }
 108
 109
 110 /*!
 111  * Split a package options string (keyval format) into a vector.
 112  * Example input:
 113  *   authorformat=smallcaps,
 114  *   commabeforerest,
 115  *   titleformat=colonsep,
 116  *   bibformat={tabular,ibidem,numbered}
 117  */
 118 vector<string> split_options(string const & input)
 119 {
 120         vector<string> options;
 121         string option;
 122         Parser p(input);
 123         while (p.good()) {
 124                 Token const & t = p.get_token();
 125                 if (t.asInput() == ",") {
 126                         options.push_back(option);
 127                         option.erase();
 128                 } else if (t.asInput() == "=") {
 129                         option += '=';
 130                         p.skip_spaces(true);
 131                         if (p.next_token().asInput() == "{")
 132                                 option += '{' + p.getArg('{', '}') + '}';
 133                 } else if (t.cat() != catSpace)
 134                         option += t.asInput();
 135         }
 136
 137         if (!option.empty())
 138                 options.push_back(option);
 139
 140         return options;
 141 }
 142
 143
 144 /*!
 145  * Add package \p name with options \p options to used_packages.
 146  * Remove options from \p options that we don't want to output.
 147  */
 148 void add_package(string const & name, vector<string> & options)
 149 {
 150         // every package inherits the global options
 151         if (used_packages.find(name) == used_packages.end())
 152                 used_packages[name] = split_options(h_options);
 153
 154         vector<string> & v = used_packages[name];
 155         v.insert(v.end(), options.begin(), options.end());
 156         if (name == "jurabib") {
 157                 // Don't output the order argument (see the cite command
 158                 // handling code in text.C).
 159                 vector<string>::iterator end =
 160                         remove(options.begin(), options.end(), "natbiborder");
 161                 end = remove(options.begin(), end, "jurabiborder");
 162                 options.erase(end, options.end());
 163         }
 164 }
 165
 166
 167 void handle_package(string const & name, string const & opts)
 168 {
 169         vector<string> options = split_options(opts);
 170         add_package(name, options);
 171
 172         //cerr << "handle_package: '" << name << "'\n";
 173         if (name == "ae")
 174                 h_fontscheme = "ae";
 175         else if (name == "aecompl")
 176                 h_fontscheme = "ae";
 177         else if (name == "amsmath")
 178                 h_use_amsmath = "1";
 179         else if (name == "amssymb")
 180                 h_use_amsmath = "1";
 181         else if (name == "babel")
 182                 ; // ignore this
 183         else if (name == "fontenc")
 184                 ; // ignore this
 185         else if (name == "inputenc") {
 186                 h_inputencoding = opts;
 187                 options.clear();
 188         } else if (name == "makeidx")
 189                 ; // ignore this
 190         else if (name == "verbatim")
 191                 ; // ignore this
 192         else if (name == "graphicx")
 193                 ; // ignore this
 194         else if (is_known(name, known_languages)) {
 195                 if (is_known(name, known_french_languages)) {
 196                         h_language = "french";
 197                         h_quotes_language = "french";
 198                 } else {
 199                         h_language = name;
 200                         h_quotes_language = name;
 201                 }
 202
 203         } else if (name == "natbib") {
 204                 h_cite_engine = "natbib_authoryear";
 205                 vector<string>::iterator it =
 206                         find(options.begin(), options.end(), "authoryear");
 207                 if (it != options.end())
 208                         options.erase(it);
 209                 else {
 210                         it = find(options.begin(), options.end(), "numbers");
 211                         if (it != options.end()) {
 212                                 h_cite_engine = "natbib_numerical";
 213                                 options.erase(it);
 214                         }
 215                 }
 216         } else if (name == "jurabib") {
 217                 h_cite_engine = "jurabib";
 218         } else if (options.empty())
 219                 h_preamble << "\\usepackage{" << name << "}\n";
 220         else {
 221                 h_preamble << "\\usepackage[" << opts << "]{" << name << "}\n";
 222                 options.clear();
 223         }
 224
 225         // We need to do something with the options...
 226         if (!options.empty())
 227                 cerr << "Ignoring options '" << join(options, ",")
 228                      << "' of package " << name << '.' << endl;
 229 }
 230
 231
 232
 233 void end_preamble(ostream & os, LyXTextClass const & /*textclass*/)
 234 {
 235         os << "#LyX file created by  tex2lyx 0.1.2\n"
 236            << "\\lyxformat 243\n"
 237            << "\\begin_document\n"
 238            << "\\begin_header\n"
 239            << "\\textclass " << h_textclass << "\n"
 240            << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
 241         if (!h_options.empty())
 242            os << "\\options " << h_options << "\n";
 243         os << "\\language " << h_language << "\n"
 244            << "\\inputencoding " << h_inputencoding << "\n"
 245            << "\\fontscheme " << h_fontscheme << "\n"
 246            << "\\graphics " << h_graphics << "\n"
 247            << "\\paperfontsize " << h_paperfontsize << "\n"
 248            << "\\spacing " << h_spacing << "\n"
 249            << "\\papersize " << h_papersize << "\n"
 250            << "\\use_geometry " << h_use_geometry << "\n"
 251            << "\\use_amsmath " << h_use_amsmath << "\n"
 252            << "\\cite_engine " << h_cite_engine << "\n"
 253            << "\\use_bibtopic " << h_use_bibtopic << "\n"
 254            << "\\paperorientation " << h_paperorientation << "\n"
 255            << "\\secnumdepth " << h_secnumdepth << "\n"
 256            << "\\tocdepth " << h_tocdepth << "\n"
 257            << "\\paragraph_separation " << h_paragraph_separation << "\n"
 258            << "\\defskip " << h_defskip << "\n"
 259            << "\\quotes_language " << h_quotes_language << "\n"
 260            << "\\quotes_times " << h_quotes_times << "\n"
 261            << "\\papercolumns " << h_papercolumns << "\n"
 262            << "\\papersides " << h_papersides << "\n"
 263            << "\\paperpagestyle " << h_paperpagestyle << "\n"
 264            << "\\tracking_changes " << h_tracking_changes << "\n"
 265            << "\\output_changes " << h_output_changes << "\n"
 266            << "\\end_header\n\n"
 267            << "\\begin_body\n";
 268         // clear preamble for subdocuments
 269         h_preamble.str("");
 270 }
 271
 272 } // anonymous namespace
 273
 274 LyXTextClass const parse_preamble(Parser & p, ostream & os, string const & forceclass)
 275 {
 276         // initialize fixed types
 277         special_columns['D'] = 3;
 278         bool is_full_document = false;
 279
 280         // determine whether this is a full document or a fragment for inclusion
 281         while (p.good()) {
 282                 Token const & t = p.get_token();
 283
 284                 if (t.cat() == catEscape && t.cs() == "documentclass") {
 285                         is_full_document = true;
 286                         break;
 287                 }
 288         }
 289         p.reset();
 290
 291         while (is_full_document && p.good()) {
 292                 Token const & t = p.get_token();
 293
 294 #ifdef FILEDEBUG
 295                 cerr << "t: " << t << "\n";
 296 #endif
 297
 298                 //
 299                 // cat codes
 300                 //
 301                 if (t.cat() == catLetter ||
 302                           t.cat() == catSuper ||
 303                           t.cat() == catSub ||
 304                           t.cat() == catOther ||
 305                           t.cat() == catMath ||
 306                           t.cat() == catActive ||
 307                           t.cat() == catBegin ||
 308                           t.cat() == catEnd ||
 309                           t.cat() == catAlign ||
 310                           t.cat() == catParameter)
 311                 h_preamble << t.character();
 312
 313                 else if (t.cat() == catSpace || t.cat() == catNewline)
 314                         h_preamble << t.asInput();
 315
 316                 else if (t.cat() == catComment)
 317                         h_preamble << t.asInput();
 318
 319                 else if (t.cs() == "pagestyle")
 320                         h_paperpagestyle = p.verbatim_item();
 321
 322                 else if (t.cs() == "makeatletter") {
 323                         p.setCatCode('@', catLetter);
 324                         h_preamble << "\\makeatletter";
 325                 }
 326
 327                 else if (t.cs() == "makeatother") {
 328                         p.setCatCode('@', catOther);
 329                         h_preamble << "\\makeatother";
 330                 }
 331
 332                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
 333                             || t.cs() == "providecommand") {
 334                         bool star = false;
 335                         if (p.next_token().character() == '*') {
 336                                 p.get_token();
 337                                 star = true;
 338                         }
 339                         string const name = p.verbatim_item();
 340                         string const opt1 = p.getOpt();
 341                         string const opt2 = p.getFullOpt();
 342                         string const body = p.verbatim_item();
 343                         // only non-lyxspecific stuff
 344                         if (   name != "\\noun"
 345                             && name != "\\tabularnewline"
 346                             && name != "\\LyX"
 347                             && name != "\\lyxline"
 348                             && name != "\\lyxaddress"
 349                             && name != "\\lyxrightaddress"
 350                             && name != "\\lyxdot"
 351                             && name != "\\boldsymbol"
 352                             && name != "\\lyxarrow") {
 353                                 ostringstream ss;
 354                                 ss << '\\' << t.cs();
 355                                 if (star)
 356                                         ss << '*';
 357                                 ss << '{' << name << '}' << opt1 << opt2
 358                                    << '{' << body << "}";
 359                                 h_preamble << ss.str();
 360
 361                                 // Add the command to the known commands
 362                                 add_known_command(name, opt1, !opt2.empty());
 363 /*
 364                                 ostream & out = in_preamble ? h_preamble : os;
 365                                 out << "\\" << t.cs() << "{" << name << "}"
 366                                     << opts << "{" << body << "}";
 367 */
 368                         }
 369                 }
 370
 371                 else if (t.cs() == "documentclass") {
 372                         vector<string> opts;
 373                         split(p.getArg('[', ']'), opts, ',');
 374                         handle_opt(opts, known_languages, h_language);
 375                         if (is_known(h_language, known_french_languages))
 376                                 h_language = "french";
 377                         handle_opt(opts, known_fontsizes, h_paperfontsize);
 378                         // delete "pt" at the end
 379                         string::size_type i = h_paperfontsize.find("pt");
 380                         if (i != string::npos)
 381                                 h_paperfontsize.erase(i);
 382                         h_quotes_language = h_language;
 383                         h_options = join(opts, ",");
 384                         h_textclass = p.getArg('{', '}');
 385                 }
 386
 387                 else if (t.cs() == "usepackage") {
 388                         string const options = p.getArg('[', ']');
 389                         string const name = p.getArg('{', '}');
 390                         if (options.empty() && name.find(',')) {
 391                                 vector<string> vecnames;
 392                                 split(name, vecnames, ',');
 393                                 vector<string>::const_iterator it  = vecnames.begin();
 394                                 vector<string>::const_iterator end = vecnames.end();
 395                                 for (; it != end; ++it)
 396                                         handle_package(trim(*it), string());
 397                         } else {
 398                                 handle_package(name, options);
 399                         }
 400                 }
 401
 402                 else if (t.cs() == "newenvironment") {
 403                         string const name = p.getArg('{', '}');
 404                         ostringstream ss;
 405                         ss << "\\newenvironment{" << name << "}";
 406                         ss << p.getOpt();
 407                         ss << p.getOpt();
 408                         ss << '{' << p.verbatim_item() << '}';
 409                         ss << '{' << p.verbatim_item() << '}';
 410                         if (name != "lyxcode" && name != "lyxlist" &&
 411                             name != "lyxrightadress" && name != "lyxaddress")
 412                                 h_preamble << ss.str();
 413                 }
 414
 415                 else if (t.cs() == "def") {
 416                         string name = p.get_token().cs();
 417                         while (p.next_token().cat() != catBegin)
 418                                 name += p.get_token().asString();
 419                         h_preamble << "\\def\\" << name << '{'
 420                                    << p.verbatim_item() << "}";
 421                 }
 422
 423                 else if (t.cs() == "newcolumntype") {
 424                         string const name = p.getArg('{', '}');
 425                         trim(name);
 426                         int nargs = 0;
 427                         string opts = p.getOpt();
 428                         if (!opts.empty()) {
 429                                 istringstream is(string(opts, 1));
 430                                 //cerr << "opt: " << is.str() << "\n";
 431                                 is >> nargs;
 432                         }
 433                         special_columns[name[0]] = nargs;
 434                         h_preamble << "\\newcolumntype{" << name << "}";
 435                         if (nargs)
 436                                 h_preamble << "[" << nargs << "]";
 437                         h_preamble << "{" << p.verbatim_item() << "}";
 438                 }
 439
 440                 else if (t.cs() == "setcounter") {
 441                         string const name = p.getArg('{', '}');
 442                         string const content = p.getArg('{', '}');
 443                         if (name == "secnumdepth")
 444                                 h_secnumdepth = content;
 445                         else if (name == "tocdepth")
 446                                 h_tocdepth = content;
 447                         else
 448                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
 449                 }
 450
 451                 else if (t.cs() == "setlength") {
 452                         string const name = p.verbatim_item();
 453                         string const content = p.verbatim_item();
 454                         // Is this correct?
 455                         if (name == "parskip")
 456                                 h_paragraph_separation = "skip";
 457                         else if (name == "parindent")
 458                                 h_paragraph_separation = "skip";
 459                         else
 460                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
 461                 }
 462
 463                 else if (t.cs() == "begin") {
 464                         string const name = p.getArg('{', '}');
 465                         if (name == "document")
 466                                 break;
 467                         h_preamble << "\\begin{" << name << "}";
 468                 }
 469
 470                 else if (t.cs() == "jurabibsetup") {
 471                         vector<string> jurabibsetup =
 472                                 split_options(p.getArg('{', '}'));
 473                         // add jurabibsetup to the jurabib package options
 474                         add_package("jurabib", jurabibsetup);
 475                         if (!jurabibsetup.empty()) {
 476                                 h_preamble << "\\jurabibsetup{"
 477                                            << join(jurabibsetup, ",") << '}';
 478                         }
 479                 }
 480
 481                 else if (!t.cs().empty())
 482                         h_preamble << '\\' << t.cs();
 483         }
 484         p.skip_spaces();
 485
 486         // Force textclass if the user wanted it
 487         if (!forceclass.empty()) {
 488                 h_textclass = forceclass;
 489         }
 490         string layoutfilename = LibFileSearch("layouts", h_textclass, "layout");
 491         if (layoutfilename.empty()) {
 492                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
 493                 exit(1);
 494         }
 495         LyXTextClass textclass;
 496         textclass.Read(layoutfilename);
 497         if (h_papersides.empty()) {
 498                 ostringstream ss;
 499                 ss << textclass.sides();
 500                 h_papersides = ss.str();
 501         }
 502         end_preamble(os, textclass);
 503         return textclass;
 504 }
 505
 506 // }])