src/tex2lyx/text.C

   1 /**
   2  * \file tex2lyx/text.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  * \author Jean-Marc Lasgouttes
   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 #include "context.h"
  18 #include "FloatList.h"
  19 #include "support/lstrings.h"
  20 #include "support/tostr.h"
  21 #include "support/filetools.h"
  22
  23 #include <iostream>
  24 #include <map>
  25 #include <sstream>
  26 #include <vector>
  27
  28 using std::cerr;
  29 using std::endl;
  30
  31 using std::map;
  32 using std::ostream;
  33 using std::ostringstream;
  34 using std::istringstream;
  35 using std::string;
  36 using std::vector;
  37
  38 using lyx::support::rtrim;
  39 using lyx::support::suffixIs;
  40
  41
  42 // thin wrapper around parse_text using a string
  43 string parse_text(Parser & p, unsigned flags, const bool outer,
  44                   Context & context)
  45 {
  46         ostringstream os;
  47         parse_text(p, os, flags, outer, context);
  48         return os.str();
  49 }
  50
  51 // parses a subdocument, usually useful in insets (whence the name)
  52 void parse_text_in_inset(Parser & p, ostream & os, unsigned flags, bool outer,
  53                 Context & context)
  54 {
  55         Context newcontext(true, context.textclass);
  56         parse_text(p, os, flags, outer, newcontext);
  57         newcontext.check_end_layout(os);
  58 }
  59
  60
  61 // parses a paragraph snippet, useful for example for \emph{...}
  62 void parse_text_snippet(Parser & p, ostream & os, unsigned flags, bool outer,
  63                 Context & context)
  64 {
  65         Context newcontext(false, context.textclass);
  66         parse_text(p, os, flags, outer, newcontext);
  67         // should not be needed
  68         newcontext.check_end_layout(os);
  69 }
  70
  71
  72 namespace {
  73
  74 char const * known_latex_commands[] = { "ref", "cite", "label", "index",
  75 "printindex", "pageref", "url", "vref", "vpageref", "prettyref", "eqref", 0 };
  76
  77 // LaTeX names for quotes
  78 char const * known_quotes[] = { "glqq", "grqq", "quotedblbase",
  79 "textquotedblleft", "quotesinglbase", "guilsinglleft", "guilsinglright", 0};
  80
  81 // the same as known_quotes with .lyx names
  82 char const * known_coded_quotes[] = { "gld", "grd", "gld",
  83 "grd", "gls", "fls", "frd", 0};
  84
  85 char const * known_sizes[] = { "tiny", "scriptsize", "footnotesize",
  86 "small", "normalsize", "large", "Large", "LARGE", "huge", "Huge", 0};
  87
  88 char const * known_coded_sizes[] = { "tiny", "scriptsize", "footnotesize",
  89 "small", "normal", "large", "larger", "largest",  "huge", "giant", 0};
  90
  91 // splits "x=z, y=b" into a map
  92 map<string, string> split_map(string const & s)
  93 {
  94         map<string, string> res;
  95         vector<string> v;
  96         split(s, v);
  97         for (size_t i = 0; i < v.size(); ++i) {
  98                 size_t const pos   = v[i].find('=');
  99                 string const index = v[i].substr(0, pos);
 100                 string const value = v[i].substr(pos + 1, string::npos);
 101                 res[trim(index)] = trim(value);
 102         }
 103         return res;
 104 }
 105
 106 // A simple function to translate a latex length to something lyx can
 107 // understand. Not perfect, but rather best-effort.
 108 string translate_len(string const & len)
 109 {
 110         const string::size_type i = len.find_first_not_of(" -0123456789.,");
 111         //'4,5' is a valid LaTeX length number. Change it to '4.5'
 112         string const length = lyx::support::subst(len, ',', '.');
 113         // a normal length
 114         if (i == string::npos || len[i]  != '\\')
 115                 return length;
 116         double val;
 117         if (i == 0) {
 118                 // We had something like \textwidth without a factor
 119                 val = 100;
 120         } else {
 121                 istringstream iss(string(length, 0, i));
 122                 iss >> val;
 123                 val = val * 100;
 124         }
 125         ostringstream oss;
 126         oss << val;
 127         string const valstring = oss.str();
 128         const string::size_type i2 = length.find(" ", i);
 129         string const unit = string(len, i, i2 - i);
 130         string const endlen = (i2 == string::npos) ? string() : string(len, i2);
 131         if (unit == "\\textwidth")
 132                 return valstring + "text%" + endlen;
 133         else if (unit == "\\columnwidth")
 134                 return valstring + "col%" + endlen;
 135         else if (unit == "\\paperwidth")
 136                 return valstring + "page%" + endlen;
 137         else if (unit == "\\linewidth")
 138                 return valstring + "line%" + endlen;
 139         else if (unit == "\\paperheight")
 140                 return valstring + "pheight%" + endlen;
 141         else if (unit == "\\textheight")
 142                 return valstring + "theight%" + endlen;
 143         else
 144                 return length;
 145 }
 146
 147
 148 void begin_inset(ostream & os, string const & name)
 149 {
 150         os << "\n\\begin_inset " << name;
 151 }
 152
 153
 154 void end_inset(ostream & os)
 155 {
 156         os << "\n\\end_inset \n\n";
 157 }
 158
 159
 160 void skip_braces(Parser & p)
 161 {
 162         if (p.next_token().cat() != catBegin)
 163                 return;
 164         p.get_token();
 165         if (p.next_token().cat() == catEnd) {
 166                 p.get_token();
 167                 return;
 168         }
 169         p.putback();
 170 }
 171
 172
 173 void handle_ert(ostream & os, string const & s, Context & context, bool check_layout = true)
 174 {
 175         if (check_layout) {
 176                 // We must have a valid layout before outputting the ERT inset.
 177                 context.check_layout(os);
 178         }
 179         Context newcontext(true, context.textclass);
 180         begin_inset(os, "ERT");
 181         os << "\nstatus Collapsed\n";
 182         newcontext.check_layout(os);
 183         for (string::const_iterator it = s.begin(), et = s.end(); it != et; ++it) {
 184                 if (*it == '\\')
 185                         os << "\n\\backslash \n";
 186                 else
 187                         os << *it;
 188         }
 189         newcontext.check_end_layout(os);
 190         end_inset(os);
 191 }
 192
 193
 194 void handle_comment(ostream & os, string const & s, Context & context)
 195 {
 196         // TODO: Handle this better
 197         Context newcontext(true, context.textclass);
 198         begin_inset(os, "ERT");
 199         os << "\nstatus Collapsed\n";
 200         newcontext.check_layout(os);
 201         for (string::const_iterator it = s.begin(), et = s.end(); it != et; ++it) {
 202                 if (*it == '\\')
 203                         os << "\n\\backslash \n";
 204                 else
 205                         os << *it;
 206         }
 207         // make sure that our comment is the last thing on the line
 208         os << "\n\\newline";
 209         newcontext.check_end_layout(os);
 210         end_inset(os);
 211 }
 212
 213
 214 struct isLayout {
 215         isLayout(string const name) : name_(name) {}
 216         bool operator()(LyXLayout_ptr const & ptr) {
 217                 return ptr.get() && ptr->latexname() == name_;
 218         }
 219 private:
 220         string const name_;
 221 };
 222
 223
 224 LyXLayout_ptr findLayout(LyXTextClass const & textclass,
 225                          string const & name)
 226 {
 227         LyXTextClass::const_iterator it  = textclass.begin();
 228         LyXTextClass::const_iterator end = textclass.end();
 229         it = std::find_if(it, end, isLayout(name));
 230         return (it == end) ? LyXLayout_ptr() : *it;
 231 }
 232
 233
 234 void output_command_layout(ostream & os, Parser & p, bool outer,
 235                            Context & parent_context,
 236                            LyXLayout_ptr newlayout)
 237 {
 238         parent_context.check_end_layout(os);
 239         Context context(true, parent_context.textclass, newlayout,
 240                         parent_context.layout);
 241         context.check_deeper(os);
 242         context.check_layout(os);
 243         if (context.layout->optionalargs > 0) {
 244                 p.skip_spaces();
 245                 if (p.next_token().character() == '[') {
 246                         p.get_token(); // eat '['
 247                         begin_inset(os, "OptArg\n");
 248                         os << "collapsed true\n\n";
 249                         parse_text_in_inset(p, os, FLAG_BRACK_LAST, outer, context);
 250                         end_inset(os);
 251                 }
 252         }
 253         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
 254         context.check_end_layout(os);
 255         context.check_end_deeper(os);
 256         // We don't need really a new paragraph, but
 257         // we must make sure that the next item gets a \begin_layout.
 258         parent_context.new_paragraph(os);
 259 }
 260
 261
 262 /*!
 263  * Output a space if necessary.
 264  * This function gets called for every whitespace token.
 265  *
 266  * We have three cases here:
 267  * 1. A space must be suppressed. Example: The lyxcode case below
 268  * 2. A space may be suppressed. Example: Spaces before "\par"
 269  * 3. A space must not be suppressed. Example: A space between two words
 270  *
 271  * We currently handle only 1. and 3 and from 2. only the case of
 272  * spaces before newlines as a side effect.
 273  *
 274  * 2. could be used to suppress as many spaces as possible. This has two effects:
 275  * - Reimporting LyX generated LaTeX files changes almost no whitespace
 276  * - Superflous whitespace from non LyX generated LaTeX files is removed.
 277  * The drawback is that the logic inside the function becomes
 278  * complicated, and that is the reason why it is not implemented.
 279  */
 280 void check_space(Parser const & p, ostream & os, Context & context)
 281 {
 282         Token const next = p.next_token();
 283         Token const curr = p.curr_token();
 284         // A space before a single newline and vice versa must be ignored
 285         // LyX emits a newline before \end{lyxcode}.
 286         // This newline must be ignored,
 287         // otherwise LyX will add an additional protected space.
 288         if (next.cat() == catSpace ||
 289             next.cat() == catNewline ||
 290             (next.cs() == "end" && context.layout->free_spacing && curr.cat() == catNewline)) {
 291                 return;
 292         }
 293         context.check_layout(os);
 294         os << ' ';
 295 }
 296
 297
 298 /*!
 299  * Check wether \param command is a known command. If yes,
 300  * handle the command with all arguments.
 301  * \return true if the command was parsed, false otherwise.
 302  */
 303 bool parse_command(string const & command, Parser & p, ostream & os,
 304                    bool outer, Context & context)
 305 {
 306         if (known_commands.find(command) != known_commands.end()) {
 307                 vector<ArgumentType> const & template_arguments = known_commands[command];
 308                 string ert = command;
 309                 size_t no_arguments = template_arguments.size();
 310                 for (size_t i = 0; i < no_arguments; ++i) {
 311                         switch (template_arguments[i]) {
 312                         case required:
 313                                 // This argument contains regular LaTeX
 314                                 handle_ert(os, ert + '{', context);
 315                                 parse_text(p, os, FLAG_ITEM, outer, context);
 316                                 ert = "}";
 317                                 break;
 318                         case verbatim:
 319                                 // This argument may contain special characters
 320                                 ert += '{' + p.verbatim_item() + '}';
 321                                 break;
 322                         case optional:
 323                                 ert += p.getOpt();
 324                                 break;
 325                         }
 326                 }
 327                 handle_ert(os, ert, context);
 328                 return true;
 329         }
 330         return false;
 331 }
 332
 333
 334 void parse_environment(Parser & p, ostream & os, bool outer,
 335                        Context & parent_context)
 336 {
 337         LyXLayout_ptr newlayout;
 338         string const name = p.getArg('{', '}');
 339         const bool is_starred = suffixIs(name, '*');
 340         string const unstarred_name = rtrim(name, "*");
 341         active_environments.push_back(name);
 342         p.skip_spaces();
 343
 344         if (is_math_env(name)) {
 345                 parent_context.check_layout(os);
 346                 begin_inset(os, "Formula ");
 347                 os << "\\begin{" << name << "}";
 348                 parse_math(p, os, FLAG_END, MATH_MODE);
 349                 os << "\\end{" << name << "}";
 350                 end_inset(os);
 351         }
 352
 353         else if (name == "tabular") {
 354                 parent_context.check_layout(os);
 355                 begin_inset(os, "Tabular ");
 356                 handle_tabular(p, os, parent_context);
 357                 end_inset(os);
 358         }
 359
 360         else if (parent_context.textclass.floats().typeExist(unstarred_name)) {
 361                 parent_context.check_layout(os);
 362                 begin_inset(os, "Float " + unstarred_name + "\n");
 363                 if (p.next_token().asInput() == "[") {
 364                         os << "placement " << p.getArg('[', ']') << '\n';
 365                 }
 366                 os << "wide " << tostr(is_starred)
 367                    << "\ncollapsed false\n\n";
 368                 parse_text_in_inset(p, os, FLAG_END, outer, parent_context);
 369                 end_inset(os);
 370                 // We don't need really a new paragraph, but
 371                 // we must make sure that the next item gets a \begin_layout.
 372                 parent_context.new_paragraph(os);
 373         }
 374
 375         else if (name == "minipage") {
 376                 string position = "1";
 377                 string inner_pos = "0";
 378                 string height = "0pt";
 379                 string latex_position;
 380                 string latex_inner_pos;
 381                 string latex_height;
 382                 if (p.next_token().asInput() == "[") {
 383                         latex_position = p.getArg('[', ']');
 384                         switch(latex_position[0]) {
 385                         case 't': position = "0"; break;
 386                         case 'c': position = "1"; break;
 387                         case 'b': position = "2"; break;
 388                         default:
 389                                 cerr << "invalid position for minipage"
 390                                      << endl;
 391                                 break;
 392                         }
 393                         if (p.next_token().asInput() == "[") {
 394                                 latex_height = p.getArg('[', ']');
 395                                 height = translate_len(latex_height);
 396
 397                                 if (p.next_token().asInput() == "[") {
 398                                         latex_inner_pos = p.getArg('[', ']');
 399                                         switch(latex_inner_pos[0]) {
 400                                         case 'c': inner_pos = "0"; break;
 401                                         case 't': inner_pos = "1"; break;
 402                                         case 'b': inner_pos = "2"; break;
 403                                         case 's': inner_pos = "3"; break;
 404                                         default:
 405                                                 cerr << "invalid inner_pos for minipage"
 406                                                      << endl;
 407                                                 break;
 408                                         }
 409                                 }
 410                         }
 411                 }
 412                 string width = translate_len(p.verbatim_item());
 413                 if (width[0] == '\\') {
 414                         // lyx can't handle length variables
 415                         ostringstream ss;
 416                         ss << "\\begin{minipage}";
 417                         if (!latex_position.empty())
 418                                 ss << '[' << latex_position << ']';
 419                         if (!latex_height.empty())
 420                                 ss << '[' << latex_height << ']';
 421                         if (!latex_inner_pos.empty())
 422                                 ss << '[' << latex_inner_pos << ']';
 423                         ss << "{" << width << "}";
 424                         handle_ert(os, ss.str(), parent_context);
 425                         parent_context.new_paragraph(os);
 426                         parse_text_in_inset(p, os, FLAG_END, outer, parent_context);
 427                         handle_ert(os, "\\end{minipage}", parent_context);
 428                 } else {
 429                         parent_context.check_layout(os);
 430                         begin_inset(os, "Minipage\n");
 431                         os << "position " << position << '\n';
 432                         os << "inner_position " << inner_pos << '\n';
 433                         os << "height \"" << height << "\"\n";
 434                         os << "width \"" << width << "\"\n";
 435                         os << "collapsed false\n\n";
 436                         parse_text_in_inset(p, os, FLAG_END, outer, parent_context);
 437                         end_inset(os);
 438                 }
 439         }
 440
 441         // Alignment settings
 442         else if (name == "center" || name == "flushleft" || name == "flushright" ||
 443                  name == "centering" || name == "raggedright" || name == "raggedleft") {
 444                 // We must begin a new paragraph if not already done
 445                 if (! parent_context.atParagraphStart()) {
 446                         parent_context.check_end_layout(os);
 447                         parent_context.new_paragraph(os);
 448                 }
 449                 if (name == "flushleft" || name == "raggedright")
 450                         parent_context.extra_stuff += "\\align left ";
 451                 else if (name == "flushright" || name == "raggedleft")
 452                         parent_context.extra_stuff += "\\align right ";
 453                 else
 454                         parent_context.extra_stuff += "\\align center ";
 455                 parse_text(p, os, FLAG_END, outer, parent_context);
 456                 // Just in case the environment is empty ..
 457                 parent_context.extra_stuff.erase();
 458                 // We must begin a new paragraph to reset the alignment
 459                 parent_context.new_paragraph(os);
 460         }
 461
 462         // The single '=' is meant here.
 463         else if ((newlayout = findLayout(parent_context.textclass, name)).get() &&
 464                    newlayout->isEnvironment()) {
 465                 Context context(true, parent_context.textclass, newlayout,
 466                                 parent_context.layout);
 467                 parent_context.check_end_layout(os);
 468                 switch (context.layout->latextype) {
 469                 case  LATEX_LIST_ENVIRONMENT:
 470                         context.extra_stuff = "\\labelwidthstring "
 471                                 + p.verbatim_item() + '\n';
 472                         p.skip_spaces();
 473                         break;
 474                 case  LATEX_BIB_ENVIRONMENT:
 475                         p.verbatim_item(); // swallow next arg
 476                         p.skip_spaces();
 477                         break;
 478                 default:
 479                         break;
 480                 }
 481                 context.check_deeper(os);
 482                 parse_text(p, os, FLAG_END, outer, context);
 483                 context.check_end_layout(os);
 484                 context.check_end_deeper(os);
 485                 parent_context.new_paragraph(os);
 486         }
 487
 488         else if (name == "appendix") {
 489                 // This is no good latex style, but it works and is used in some documents...
 490                 parent_context.check_end_layout(os);
 491                 Context context(true, parent_context.textclass, parent_context.layout,
 492                                 parent_context.layout);
 493                 context.check_layout(os);
 494                 os << "\\start_of_appendix\n";
 495                 parse_text(p, os, FLAG_END, outer, context);
 496                 context.check_end_layout(os);
 497         }
 498
 499         else if (name == "comment") {
 500                 parent_context.check_layout(os);
 501                 begin_inset(os, "Comment\n");
 502                 os << "collapsed false\n";
 503                 parse_text_in_inset(p, os, FLAG_END, outer, parent_context);
 504                 end_inset(os);
 505         }
 506
 507         else if (name == "tabbing") {
 508                 // We need to remember that we have to handle '\=' specially
 509                 handle_ert(os, "\\begin{" + name + "}", parent_context);
 510                 parse_text_snippet(p, os, FLAG_END | FLAG_TABBING, outer, parent_context);
 511                 handle_ert(os, "\\end{" + name + "}", parent_context);
 512         }
 513
 514         else {
 515                 handle_ert(os, "\\begin{" + name + "}", parent_context);
 516                 parse_text_snippet(p, os, FLAG_END, outer, parent_context);
 517                 handle_ert(os, "\\end{" + name + "}", parent_context);
 518         }
 519
 520         active_environments.pop_back();
 521         if (name != "math")
 522                 p.skip_spaces();
 523 }
 524
 525 } // anonymous namespace
 526
 527
 528
 529
 530 void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
 531                 Context & context)
 532 {
 533         LyXLayout_ptr newlayout;
 534         // Store the latest bibliographystyle (needed for bibtex inset)
 535         string bibliographystyle;
 536         while (p.good()) {
 537                 Token const & t = p.get_token();
 538
 539 #ifdef FILEDEBUG
 540                 cerr << "t: " << t << " flags: " << flags << "\n";
 541 #endif
 542
 543                 if (flags & FLAG_ITEM) {
 544                         if (t.cat() == catSpace)
 545                                 continue;
 546
 547                         flags &= ~FLAG_ITEM;
 548                         if (t.cat() == catBegin) {
 549                                 // skip the brace and collect everything to the next matching
 550                                 // closing brace
 551                                 flags |= FLAG_BRACE_LAST;
 552                                 continue;
 553                         }
 554
 555                         // handle only this single token, leave the loop if done
 556                         flags |= FLAG_LEAVE;
 557                 }
 558
 559                 if (t.character() == ']' && (flags & FLAG_BRACK_LAST))
 560                         return;
 561
 562                 //
 563                 // cat codes
 564                 //
 565                 if (t.cat() == catMath) {
 566                         // we are inside some text mode thingy, so opening new math is allowed
 567                         context.check_layout(os);
 568                         begin_inset(os, "Formula ");
 569                         Token const & n = p.get_token();
 570                         if (n.cat() == catMath && outer) {
 571                                 // TeX's $$...$$ syntax for displayed math
 572                                 os << "\\[";
 573                                 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
 574                                 os << "\\]";
 575                                 p.get_token(); // skip the second '$' token
 576                         } else {
 577                                 // simple $...$  stuff
 578                                 p.putback();
 579                                 os << '$';
 580                                 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
 581                                 os << '$';
 582                         }
 583                         end_inset(os);
 584                 }
 585
 586                 else if (t.cat() == catSuper || t.cat() == catSub)
 587                         cerr << "catcode " << t << " illegal in text mode\n";
 588
 589                 // Basic support for english quotes. This should be
 590                 // extended to other quotes, but is not so easy (a
 591                 // left english quote is the same as a right german
 592                 // quote...)
 593                 else if (t.asInput() == "`"
 594                          && p.next_token().asInput() == "`") {
 595                         context.check_layout(os);
 596                         begin_inset(os, "Quotes ");
 597                         os << "eld";
 598                         end_inset(os);
 599                         p.get_token();
 600                         skip_braces(p);
 601                 }
 602                 else if (t.asInput() == "'"
 603                          && p.next_token().asInput() == "'") {
 604                         context.check_layout(os);
 605                         begin_inset(os, "Quotes ");
 606                         os << "erd";
 607                         end_inset(os);
 608                         p.get_token();
 609                         skip_braces(p);
 610                 }
 611
 612                 else if (t.cat() == catSpace || (t.cat() == catNewline && t.cs().size() == 1))
 613                         check_space(p, os, context);
 614
 615                 else if (t.cat() == catLetter ||
 616                                t.cat() == catOther ||
 617                                t.cat() == catAlign ||
 618                                t.cat() == catParameter) {
 619                         context.check_layout(os);
 620                         os << t.character();
 621                 }
 622
 623                 else if (t.cat() == catNewline || (t.cat() == catEscape && t.cs() == "par")) {
 624                         p.skip_spaces();
 625                         context.new_paragraph(os);
 626                 }
 627
 628                 else if (t.cat() == catActive) {
 629                         context.check_layout(os);
 630                         if (t.character() == '~') {
 631                                 if (context.layout->free_spacing)
 632                                         os << ' ';
 633                                 else
 634                                         os << "\\InsetSpace ~\n";
 635                         } else
 636                                 os << t.character();
 637                 }
 638
 639                 else if (t.cat() == catBegin) {
 640                         // special handling of size changes
 641                         context.check_layout(os);
 642                         bool const is_size = is_known(p.next_token().cs(), known_sizes);
 643                         Token const prev = p.prev_token();
 644                         string const s = parse_text(p, FLAG_BRACE_LAST, outer, context);
 645                         if (s.empty() && (p.next_token().character() == '`' ||
 646                                           (prev.character() == '-' && p.next_token().character())))
 647                                 ; // ignore it in {}`` or -{}-
 648                         else if (is_size || s == "[" || s == "]" || s == "*")
 649                                 os << s;
 650                         else {
 651                                 handle_ert(os, "{", context, false);
 652                                 // s will end the current layout and begin a new one if necessary
 653                                 os << s;
 654                                 handle_ert(os, "}", context);
 655                         }
 656                 }
 657
 658                 else if (t.cat() == catEnd) {
 659                         if (flags & FLAG_BRACE_LAST) {
 660                                 return;
 661                         }
 662                         cerr << "stray '}' in text\n";
 663                         handle_ert(os, "}", context);
 664                 }
 665
 666                 else if (t.cat() == catComment) {
 667                         context.check_layout(os);
 668                         if (!t.cs().empty()) {
 669                                 handle_comment(os, '%' + t.cs(), context);
 670                                 if (p.next_token().cat() == catNewline) {
 671                                         // A newline after a comment line starts a new paragraph
 672                                         context.new_paragraph(os);
 673                                         p.skip_spaces();
 674                                 }
 675                         } else {
 676                                 // "%\n" combination
 677                                 p.skip_spaces();
 678                         }
 679                 }
 680
 681                 //
 682                 // control sequences
 683                 //
 684
 685                 else if (t.cs() == "(") {
 686                         context.check_layout(os);
 687                         begin_inset(os, "Formula");
 688                         os << " \\(";
 689                         parse_math(p, os, FLAG_SIMPLE2, MATH_MODE);
 690                         os << "\\)";
 691                         end_inset(os);
 692                 }
 693
 694                 else if (t.cs() == "[") {
 695                         context.check_layout(os);
 696                         begin_inset(os, "Formula");
 697                         os << " \\[";
 698                         parse_math(p, os, FLAG_EQUATION, MATH_MODE);
 699                         os << "\\]";
 700                         end_inset(os);
 701                 }
 702
 703                 else if (t.cs() == "begin")
 704                         parse_environment(p, os, outer, context);
 705
 706                 else if (t.cs() == "end") {
 707                         if (flags & FLAG_END) {
 708                                 // eat environment name
 709                                 string const name = p.getArg('{', '}');
 710                                 if (name != active_environment())
 711                                         cerr << "\\end{" + name + "} does not match \\begin{"
 712                                                 + active_environment() + "}\n";
 713                                 return;
 714                         }
 715                         p.error("found 'end' unexpectedly");
 716                 }
 717
 718                 else if (t.cs() == "item") {
 719                         p.skip_spaces();
 720                         string s;
 721                         bool optarg = false;
 722                         if (p.next_token().character() == '[') {
 723                                 p.get_token(); // eat '['
 724                                 Context newcontext(false, context.textclass);
 725                                 s = parse_text(p, FLAG_BRACK_LAST, outer, newcontext);
 726                                 optarg = true;
 727                         }
 728                         context.set_item();
 729                         context.check_layout(os);
 730                         if (optarg) {
 731                                 if (context.layout->labeltype != LABEL_MANUAL) {
 732                                         // lyx does not support \item[\mybullet] in itemize environments
 733                                         handle_ert(os, "[", context);
 734                                         os << s;
 735                                         handle_ert(os, "]", context);
 736                                 } else if (!s.empty()) {
 737                                         // The space is needed to separate the item from the rest of the sentence.
 738                                         os << s << ' ';
 739                                         p.skip_spaces();
 740                                 }
 741                         }
 742                 }
 743
 744                 else if (t.cs() == "bibitem") {
 745                         context.set_item();
 746                         context.check_layout(os);
 747                         os << "\\bibitem ";
 748                         os << p.getOpt();
 749                         os << '{' << p.verbatim_item() << '}' << "\n";
 750                 }
 751
 752                 else if (t.cs() == "def") {
 753                         p.skip_spaces();
 754                         context.check_layout(os);
 755                         string name = p.get_token().cs();
 756                         while (p.next_token().cat() != catBegin)
 757                                 name += p.get_token().asString();
 758                         handle_ert(os, "\\def\\" + name + '{' + p.verbatim_item() + '}', context);
 759                 }
 760
 761                 else if (t.cs() == "noindent") {
 762                         p.skip_spaces();
 763                         context.extra_stuff += "\\noindent ";
 764                 }
 765
 766                 else if (t.cs() == "appendix") {
 767                         p.skip_spaces();
 768                         context.extra_stuff += "\\start_of_appendix ";
 769                 }
 770
 771                 // Must attempt to parse "Section*" before "Section".
 772                 else if ((p.next_token().asInput() == "*") &&
 773                          // The single '=' is meant here.
 774                          (newlayout = findLayout(context.textclass,
 775                                                  t.cs() + '*')).get() &&
 776                          newlayout->isCommand()) {
 777                         p.get_token();
 778                         output_command_layout(os, p, outer, context, newlayout);
 779                         p.skip_spaces();
 780                 }
 781
 782                 // The single '=' is meant here.
 783                 else if ((newlayout = findLayout(context.textclass, t.cs())).get() &&
 784                          newlayout->isCommand()) {
 785                         output_command_layout(os, p, outer, context, newlayout);
 786                         p.skip_spaces();
 787                 }
 788
 789                 else if (t.cs() == "includegraphics") {
 790                         map<string, string> opts = split_map(p.getArg('[', ']'));
 791                         string name = p.verbatim_item();
 792
 793                         context.check_layout(os);
 794                         begin_inset(os, "Graphics ");
 795                         os << "\n\tfilename " << name << '\n';
 796                         if (opts.find("width") != opts.end())
 797                                 os << "\twidth "
 798                                    << translate_len(opts["width"]) << '\n';
 799                         if (opts.find("height") != opts.end())
 800                                 os << "\theight "
 801                                    << translate_len(opts["height"]) << '\n';
 802                         if (opts.find("scale") != opts.end()) {
 803                                 istringstream iss(opts["scale"]);
 804                                 double val;
 805                                 iss >> val;
 806                                 val = val*100;
 807                                 os << "\tscale " << val << '\n';
 808                         }
 809                         if (opts.find("angle") != opts.end())
 810                                 os << "\trotateAngle "
 811                                    << opts["angle"] << '\n';
 812                         if (opts.find("origin") != opts.end()) {
 813                                 ostringstream ss;
 814                                 string const opt = opts["origin"];
 815                                 if (opt.find('l') != string::npos) ss << "left";
 816                                 if (opt.find('r') != string::npos) ss << "right";
 817                                 if (opt.find('c') != string::npos) ss << "center";
 818                                 if (opt.find('t') != string::npos) ss << "Top";
 819                                 if (opt.find('b') != string::npos) ss << "Bottom";
 820                                 if (opt.find('B') != string::npos) ss << "Baseline";
 821                                 if (!ss.str().empty())
 822                                         os << "\trotateOrigin " << ss.str() << '\n';
 823                                 else
 824                                         cerr << "Warning: Ignoring unknown includegraphics origin argument '" << opt << "'\n";
 825                         }
 826                         if (opts.find("keepaspectratio") != opts.end())
 827                                 os << "\tkeepAspectRatio\n";
 828                         if (opts.find("clip") != opts.end())
 829                                 os << "\tclip\n";
 830                         if (opts.find("draft") != opts.end())
 831                                 os << "\tdraft\n";
 832                         if (opts.find("bb") != opts.end())
 833                                 os << "\tBoundingBox "
 834                                    << opts["bb"] << '\n';
 835                         int numberOfbbOptions = 0;
 836                         if (opts.find("bbllx") != opts.end())
 837                                 numberOfbbOptions++;
 838                         if (opts.find("bblly") != opts.end())
 839                                 numberOfbbOptions++;
 840                         if (opts.find("bburx") != opts.end())
 841                                 numberOfbbOptions++;
 842                         if (opts.find("bbury") != opts.end())
 843                                 numberOfbbOptions++;
 844                         if (numberOfbbOptions == 4)
 845                                 os << "\tBoundingBox "
 846                                    << opts["bbllx"] << opts["bblly"]
 847                                    << opts["bburx"] << opts["bbury"] << '\n';
 848                         else if (numberOfbbOptions > 0)
 849                                 cerr << "Warning: Ignoring incomplete includegraphics boundingbox arguments.\n";
 850                         numberOfbbOptions = 0;
 851                         if (opts.find("natwidth") != opts.end())
 852                                 numberOfbbOptions++;
 853                         if (opts.find("natheight") != opts.end())
 854                                 numberOfbbOptions++;
 855                         if (numberOfbbOptions == 2)
 856                                 os << "\tBoundingBox 0bp 0bp "
 857                                    << opts["natwidth"] << opts["natheight"] << '\n';
 858                         else if (numberOfbbOptions > 0)
 859                                 cerr << "Warning: Ignoring incomplete includegraphics boundingbox arguments.\n";
 860                         ostringstream special;
 861                         if (opts.find("hiresbb") != opts.end())
 862                                 special << "hiresbb,";
 863                         if (opts.find("trim") != opts.end())
 864                                 special << "trim,";
 865                         if (opts.find("viewport") != opts.end())
 866                                 special << "viewport=" << opts["viewport"] << ',';
 867                         if (opts.find("totalheight") != opts.end())
 868                                 special << "totalheight=" << opts["totalheight"] << ',';
 869                         if (opts.find("type") != opts.end())
 870                                 special << "type=" << opts["type"] << ',';
 871                         if (opts.find("ext") != opts.end())
 872                                 special << "ext=" << opts["ext"] << ',';
 873                         if (opts.find("read") != opts.end())
 874                                 special << "read=" << opts["read"] << ',';
 875                         if (opts.find("command") != opts.end())
 876                                 special << "command=" << opts["command"] << ',';
 877                         string s_special = special.str();
 878                         if (!s_special.empty()) {
 879                                 // We had special arguments. Remove the trailing ','.
 880                                 os << "\tspecial " << s_special.substr(0, s_special.size() - 1) << '\n';
 881                         }
 882                         // TODO: Handle the unknown settings better.
 883                         // Warn about invalid options.
 884                         // Check wether some option was given twice.
 885                         end_inset(os);
 886                 }
 887
 888                 else if (t.cs() == "footnote") {
 889                         p.skip_spaces();
 890                         context.check_layout(os);
 891                         begin_inset(os, "Foot\n");
 892                         os << "collapsed true\n\n";
 893                         parse_text_in_inset(p, os, FLAG_ITEM, false, context);
 894                         end_inset(os);
 895                 }
 896
 897                 else if (t.cs() == "marginpar") {
 898                         p.skip_spaces();
 899                         context.check_layout(os);
 900                         begin_inset(os, "Marginal\n");
 901                         os << "collapsed true\n\n";
 902                         parse_text_in_inset(p, os, FLAG_ITEM, false, context);
 903                         end_inset(os);
 904                 }
 905
 906                 else if (t.cs() == "ensuremath") {
 907                         p.skip_spaces();
 908                         context.check_layout(os);
 909                         Context newcontext(false, context.textclass);
 910                         string s = parse_text(p, FLAG_ITEM, false, newcontext);
 911                         if (s == "±" || s == "³" || s == "²" || s == "µ")
 912                                 os << s;
 913                         else
 914                                 handle_ert(os, "\\ensuremath{" + s + "}",
 915                                            context);
 916                 }
 917
 918                 else if (t.cs() == "hfill") {
 919                         context.check_layout(os);
 920                         os << "\n\\hfill\n";
 921                         skip_braces(p);
 922                         p.skip_spaces();
 923                 }
 924
 925                 else if (t.cs() == "makeindex" || t.cs() == "maketitle") {
 926                         p.skip_spaces();
 927                         skip_braces(p); // swallow this
 928                 }
 929
 930                 else if (t.cs() == "tableofcontents") {
 931                         p.skip_spaces();
 932                         context.check_layout(os);
 933                         begin_inset(os, "LatexCommand \\tableofcontents\n");
 934                         end_inset(os);
 935                         skip_braces(p); // swallow this
 936                 }
 937
 938                 else if (t.cs() == "listoffigures") {
 939                         p.skip_spaces();
 940                         context.check_layout(os);
 941                         begin_inset(os, "FloatList figure\n");
 942                         end_inset(os);
 943                         skip_braces(p); // swallow this
 944                 }
 945
 946                 else if (t.cs() == "listoftables") {
 947                         p.skip_spaces();
 948                         context.check_layout(os);
 949                         begin_inset(os, "FloatList table\n");
 950                         end_inset(os);
 951                         skip_braces(p); // swallow this
 952                 }
 953
 954                 else if (t.cs() == "listof") {
 955                         p.skip_spaces(true);
 956                         string const name = p.get_token().asString();
 957                         if (context.textclass.floats().typeExist(name)) {
 958                                 context.check_layout(os);
 959                                 begin_inset(os, "FloatList ");
 960                                 os << name << "\n";
 961                                 end_inset(os);
 962                                 p.get_token(); // swallow second arg
 963                         } else
 964                                 handle_ert(os, "\\listof{" + name + "}", context);
 965                 }
 966
 967                 else if (t.cs() == "textrm") {
 968                         context.check_layout(os);
 969                         os << "\n\\family roman \n";
 970                         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
 971                         os << "\n\\family default \n";
 972                 }
 973
 974                 else if (t.cs() == "textsf") {
 975                         context.check_layout(os);
 976                         os << "\n\\family sans \n";
 977                         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
 978                         os << "\n\\family default \n";
 979                 }
 980
 981                 else if (t.cs() == "textsl") {
 982                         context.check_layout(os);
 983                         os << "\n\\shape slanted \n";
 984                         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
 985                         os << "\n\\shape default \n";
 986                 }
 987
 988                 else if (t.cs() == "texttt") {
 989                         context.check_layout(os);
 990                         os << "\n\\family typewriter \n";
 991                         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
 992                         os << "\n\\family default \n";
 993                 }
 994
 995                 else if (t.cs() == "textit") {
 996                         context.check_layout(os);
 997                         os << "\n\\shape italic \n";
 998                         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
 999                         os << "\n\\shape default \n";
1000                 }
1001
1002                 else if (t.cs() == "textsc") {
1003                         context.check_layout(os);
1004                         os << "\n\\noun on \n";
1005                         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
1006                         os << "\n\\noun default \n";
1007                 }
1008
1009                 else if (t.cs() == "textbf") {
1010                         context.check_layout(os);
1011                         os << "\n\\series bold \n";
1012                         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
1013                         os << "\n\\series default \n";
1014                 }
1015
1016                 else if (t.cs() == "underbar") {
1017                         context.check_layout(os);
1018                         os << "\n\\bar under \n";
1019                         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
1020                         os << "\n\\bar default \n";
1021                 }
1022
1023                 else if (t.cs() == "emph" || t.cs() == "noun") {
1024                         context.check_layout(os);
1025                         os << "\n\\" << t.cs() << " on \n";
1026                         parse_text_snippet(p, os, FLAG_ITEM, outer, context);
1027                         os << "\n\\" << t.cs() << " default \n";
1028                 }
1029
1030                 else if (is_known(t.cs(), known_latex_commands)) {
1031                         context.check_layout(os);
1032                         begin_inset(os, "LatexCommand ");
1033                         os << '\\' << t.cs();
1034                         os << p.getOpt();
1035                         os << p.getOpt();
1036                         os << '{' << p.verbatim_item() << "}\n";
1037                         end_inset(os);
1038                 }
1039
1040                 else if (is_known(t.cs(), known_quotes)) {
1041                         char const ** where = is_known(t.cs(), known_quotes);
1042                         context.check_layout(os);
1043                         begin_inset(os, "Quotes ");
1044                         os << known_coded_quotes[where - known_quotes];
1045                         end_inset(os);
1046                         skip_braces(p);
1047                 }
1048
1049                 else if (is_known(t.cs(), known_sizes)) {
1050                         char const ** where = is_known(t.cs(), known_sizes);
1051                         context.check_layout(os);
1052                         os << "\n\\size " << known_coded_sizes[where - known_sizes] << "\n";
1053                         p.skip_spaces();
1054                 }
1055
1056                 else if (t.cs() == "LyX" || t.cs() == "TeX"
1057                          || t.cs() == "LaTeX") {
1058                         context.check_layout(os);
1059                         os << t.cs();
1060                         skip_braces(p); // eat {}
1061                 }
1062
1063                 else if (t.cs() == "LaTeXe") {
1064                         context.check_layout(os);
1065                         os << "LaTeX2e";
1066                         skip_braces(p); // eat {}
1067                 }
1068
1069                 else if (t.cs() == "ldots") {
1070                         context.check_layout(os);
1071                         skip_braces(p);
1072                         os << "\\SpecialChar \\ldots{}\n";
1073                 }
1074
1075                 else if (t.cs() == "lyxarrow") {
1076                         context.check_layout(os);
1077                         os << "\\SpecialChar \\menuseparator\n";
1078                         skip_braces(p);
1079                 }
1080
1081                 else if (t.cs() == "textcompwordmark") {
1082                         context.check_layout(os);
1083                         os << "\\SpecialChar \\textcompwordmark{}\n";
1084                         skip_braces(p);
1085                 }
1086
1087                 else if (t.cs() == "@" && p.next_token().asInput() == ".") {
1088                         context.check_layout(os);
1089                         os << "\\SpecialChar \\@.\n";
1090                         p.get_token();
1091                 }
1092
1093                 else if (t.cs() == "-") {
1094                         context.check_layout(os);
1095                         os << "\\SpecialChar \\-\n";
1096                 }
1097
1098                 else if (t.cs() == "textasciitilde") {
1099                         context.check_layout(os);
1100                         os << '~';
1101                         skip_braces(p);
1102                 }
1103
1104                 else if (t.cs() == "textasciicircum") {
1105                         context.check_layout(os);
1106                         os << '^';
1107                         skip_braces(p);
1108                 }
1109
1110                 else if (t.cs() == "textbackslash") {
1111                         context.check_layout(os);
1112                         os << "\n\\backslash \n";
1113                         skip_braces(p);
1114                 }
1115
1116                 else if (t.cs() == "_" || t.cs() == "&" || t.cs() == "#"
1117                             || t.cs() == "$" || t.cs() == "{" || t.cs() == "}"
1118                             || t.cs() == "%") {
1119                         context.check_layout(os);
1120                         os << t.cs();
1121                 }
1122
1123                 else if (t.cs() == "char") {
1124                         context.check_layout(os);
1125                         if (p.next_token().character() == '`') {
1126                                 p.get_token();
1127                                 if (p.next_token().cs() == "\"") {
1128                                         p.get_token();
1129                                         os << '"';
1130                                         skip_braces(p);
1131                                 } else {
1132                                         handle_ert(os, "\\char`", context);
1133                                 }
1134                         } else {
1135                                 handle_ert(os, "\\char", context);
1136                         }
1137                 }
1138
1139                 else if (t.cs() == "\"") {
1140                         context.check_layout(os);
1141                         string const name = p.verbatim_item();
1142                              if (name == "a") os << 'ä';
1143                         else if (name == "o") os << 'ö';
1144                         else if (name == "u") os << 'ü';
1145                         else if (name == "A") os << 'Ä';
1146                         else if (name == "O") os << 'Ö';
1147                         else if (name == "U") os << 'Ü';
1148                         else handle_ert(os, "\"{" + name + "}", context);
1149                 }
1150
1151                 // Problem: \= creates a tabstop inside the tabbing environment
1152                 // and else an accent. In the latter case we really would want
1153                 // \={o} instead of \= o.
1154                 else if (t.cs() == "H" || t.cs() == "c" || t.cs() == "^" || t.cs() == "'"
1155                       || t.cs() == "~" || t.cs() == "." || (t.cs() == "=" && ! (flags & FLAG_TABBING))) {
1156                         // we need the trim as the LyX parser chokes on such spaces
1157                         context.check_layout(os);
1158                         os << "\n\\i \\" << t.cs() << "{"
1159                            << trim(parse_text(p, FLAG_ITEM, outer, context), " ") << "}\n";
1160                 }
1161
1162                 else if (t.cs() == "ss") {
1163                         context.check_layout(os);
1164                         os << "ß";
1165                 }
1166
1167                 else if (t.cs() == "i" || t.cs() == "j") {
1168                         context.check_layout(os);
1169                         os << "\\" << t.cs() << ' ';
1170                 }
1171
1172                 else if (t.cs() == "\\") {
1173                         context.check_layout(os);
1174                         string const next = p.next_token().asInput();
1175                         if (next == "[")
1176                                 handle_ert(os, "\\\\" + p.getOpt(), context);
1177                         else if (next == "*") {
1178                                 p.get_token();
1179                                 handle_ert(os, "\\\\*" + p.getOpt(), context);
1180                         }
1181                         else {
1182                                 os << "\n\\newline \n";
1183                         }
1184                 }
1185
1186                 else if (t.cs() == "input" || t.cs() == "include"
1187                          || t.cs() == "verbatiminput") {
1188                         string name = '\\' + t.cs();
1189                         if (t.cs() == "verbatiminput"
1190                             && p.next_token().asInput() == "*")
1191                                 name += p.get_token().asInput();
1192                         context.check_layout(os);
1193                         begin_inset(os, "Include ");
1194                         string filename(p.getArg('{', '}'));
1195                         string lyxname(lyx::support::ChangeExtension(filename, ".lyx"));
1196                         if (tex2lyx(filename, lyxname)) {
1197                                 os << name << '{' << lyxname << "}\n";
1198                         } else {
1199                                 os << name << '{' << filename << "}\n";
1200                         }
1201                         os << "preview false\n";
1202                         end_inset(os);
1203                 }
1204
1205                 else if (t.cs() == "fancyhead") {
1206                         context.check_layout(os);
1207                         ostringstream ss;
1208                         ss << "\\fancyhead";
1209                         ss << p.getOpt();
1210                         ss << '{' << p.verbatim_item() << "}\n";
1211                         handle_ert(os, ss.str(), context);
1212                 }
1213
1214                 else if (t.cs() == "bibliographystyle") {
1215                         // store new bibliographystyle
1216                         bibliographystyle = p.verbatim_item();
1217                         // output new bibliographystyle.
1218                         // This is only necessary if used in some other macro than \bibliography.
1219                         handle_ert(os, "\\bibliographystyle{" + bibliographystyle + "}", context);
1220                 }
1221
1222                 else if (t.cs() == "bibliography") {
1223                         context.check_layout(os);
1224                         begin_inset(os, "LatexCommand ");
1225                         os << "\\bibtex";
1226                         // Do we have a bibliographystyle set?
1227                         if (!bibliographystyle.empty()) {
1228                                 os << '[' << bibliographystyle << ']';
1229                         }
1230                         os << '{' << p.verbatim_item() << "}\n";
1231                         end_inset(os);
1232                 }
1233
1234                 else if ( t.cs() == "smallskip" ||
1235                           t.cs() == "medskip" ||
1236                           t.cs() == "bigskip" ||
1237                           t.cs() == "vfill" ||
1238                          (t.cs() == "vspace" && p.next_token().asInput() != "*")) {
1239                         string arg;
1240                         if (t.cs() == "vspace")
1241                                 arg = p.getArg('{', '}');
1242                         else
1243                                 arg = t.cs();
1244                         // We may only add the vspace to the current context if the
1245                         // current paragraph is not empty.
1246                         if (context.atParagraphStart()
1247                             && (p.next_token().cat() != catNewline || p.next_token().cs().size() == 1)
1248                             && (! (p.next_token().cat() == catEscape && p.next_token().cs() == "end"))
1249                             && (! (p.next_token().cat() == catEscape && p.next_token().cs() == "par"))) {
1250                                 context.extra_stuff += "\\added_space_top " + arg + " ";
1251                                 p.skip_spaces();
1252                         } else {
1253                                 if (t.cs() == "vspace")
1254                                         handle_ert(os, t.asInput() + '{' + arg + '}', context);
1255                                 else
1256                                         handle_ert(os, t.asInput(), context);
1257                         }
1258                         // Would be nice to recognize added_space_bottom too...
1259                         // At the moment this is parsed as added_space_top of the
1260                         // next paragraph.
1261                 }
1262
1263                 else {
1264                         //cerr << "#: " << t << " mode: " << mode << endl;
1265                         // heuristic: read up to next non-nested space
1266                         /*
1267                         string s = t.asInput();
1268                         string z = p.verbatim_item();
1269                         while (p.good() && z != " " && z.size()) {
1270                                 //cerr << "read: " << z << endl;
1271                                 s += z;
1272                                 z = p.verbatim_item();
1273                         }
1274                         cerr << "found ERT: " << s << endl;
1275                         handle_ert(os, s + ' ', context);
1276                         */
1277                         string name = t.asInput();
1278                         if (p.next_token().asInput() == "*") {
1279                                 // Starred commands like \vspace*{}
1280                                 p.get_token();                          // Eat '*'
1281                                 name += '*';
1282                         }
1283                         if (! parse_command(t.asInput(), p, os, outer, context))
1284                                 handle_ert(os, name, context);
1285                 }
1286
1287                 if (flags & FLAG_LEAVE) {
1288                         flags &= ~FLAG_LEAVE;
1289                         break;
1290                 }
1291         }
1292 }
1293
1294
1295 // }])