src/tex2lyx/Parser.cpp

   1 /**
   2  * \file Parser.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  *
   8  * Full author contact details are available in file CREDITS.
   9  */
  10
  11 #include <config.h>
  12
  13 #include "Encoding.h"
  14 #include "Parser.h"
  15
  16 #include <iostream>
  17
  18 using namespace std;
  19
  20 namespace lyx {
  21
  22 namespace {
  23
  24 CatCode theCatcode[256];
  25
  26 void catInit()
  27 {
  28         static bool init_done = false;
  29         if (init_done)
  30                 return;
  31         init_done = true;
  32
  33         fill(theCatcode, theCatcode + 256, catOther);
  34         fill(theCatcode + 'a', theCatcode + 'z' + 1, catLetter);
  35         fill(theCatcode + 'A', theCatcode + 'Z' + 1, catLetter);
  36
  37         theCatcode[int('\\')] = catEscape;
  38         theCatcode[int('{')]  = catBegin;
  39         theCatcode[int('}')]  = catEnd;
  40         theCatcode[int('$')]  = catMath;
  41         theCatcode[int('&')]  = catAlign;
  42         theCatcode[int('\n')] = catNewline;
  43         theCatcode[int('#')]  = catParameter;
  44         theCatcode[int('^')]  = catSuper;
  45         theCatcode[int('_')]  = catSub;
  46         theCatcode[0x7f]      = catIgnore;
  47         theCatcode[int(' ')]  = catSpace;
  48         theCatcode[int('\t')] = catSpace;
  49         theCatcode[int('\r')] = catNewline;
  50         theCatcode[int('~')]  = catActive;
  51         theCatcode[int('%')]  = catComment;
  52
  53         // This is wrong!
  54         theCatcode[int('@')]  = catLetter;
  55 }
  56
  57 /*!
  58  * Translate a line ending to '\n'.
  59  * \p c must have catcode catNewline, and it must be the last character read
  60  * from \p is.
  61  */
  62 char_type getNewline(idocstream & is, char_type c)
  63 {
  64         // we have to handle 3 different line endings:
  65         // - UNIX (\n)
  66         // - MAC  (\r)
  67         // - DOS  (\r\n)
  68         if (c == '\r') {
  69                 // MAC or DOS
  70                 char_type wc;
  71                 if (is.get(wc) && wc != '\n') {
  72                         // MAC
  73                         is.putback(wc);
  74                 }
  75                 return '\n';
  76         }
  77         // UNIX
  78         return c;
  79 }
  80
  81 CatCode catcode(char_type c)
  82 {
  83         if (c < 256)
  84                 return theCatcode[(unsigned char)c];
  85         return catOther;
  86 }
  87
  88 }
  89
  90
  91 //
  92 // Token
  93 //
  94
  95 ostream & operator<<(ostream & os, Token const & t)
  96 {
  97         if (t.cat() == catComment)
  98                 os << '%' << t.cs() << '\n';
  99         else if (t.cat() == catSpace)
 100                 os << t.cs();
 101         else if (t.cat() == catEscape)
 102                 os << '\\' << t.cs() << ' ';
 103         else if (t.cat() == catLetter)
 104                 os << t.cs();
 105         else if (t.cat() == catNewline)
 106                 os << "[" << t.cs().size() << "\\n," << t.cat() << "]\n";
 107         else
 108                 os << '[' << t.cs() << ',' << t.cat() << ']';
 109         return os;
 110 }
 111
 112
 113 string Token::asString() const
 114 {
 115         return cs_;
 116 }
 117
 118
 119 string Token::asInput() const
 120 {
 121         if (cat_ == catComment)
 122                 return '%' + cs_ + '\n';
 123         if (cat_ == catEscape)
 124                 return '\\' + cs_;
 125         return cs_;
 126 }
 127
 128
 129 //
 130 // Parser
 131 //
 132
 133
 134 Parser::Parser(idocstream & is)
 135         : lineno_(0), pos_(0), iss_(0), is_(is), encoding_latex_("utf8")
 136 {
 137 }
 138
 139
 140 Parser::Parser(string const & s)
 141         : lineno_(0), pos_(0),
 142           iss_(new idocstringstream(from_utf8(s))), is_(*iss_),
 143           encoding_latex_("utf8")
 144 {
 145 }
 146
 147
 148 Parser::~Parser()
 149 {
 150         delete iss_;
 151 }
 152
 153
 154 void Parser::setEncoding(std::string const & e)
 155 {
 156         Encoding const * enc = encodings.fromLaTeXName(e);
 157         if (!enc) {
 158                 cerr << "Unknown encoding " << e << ". Ignoring." << std::endl;
 159                 return;
 160         }
 161         //cerr << "setting encoding to " << enc->iconvName() << std::endl;
 162         is_ << lyx::setEncoding(enc->iconvName());
 163         encoding_latex_ = e;
 164 }
 165
 166
 167 void Parser::push_back(Token const & t)
 168 {
 169         tokens_.push_back(t);
 170 }
 171
 172
 173 // We return a copy here because the tokens_ vector may get reallocated
 174 Token const Parser::prev_token() const
 175 {
 176         static const Token dummy;
 177         return pos_ > 1 ? tokens_[pos_ - 2] : dummy;
 178 }
 179
 180
 181 // We return a copy here because the tokens_ vector may get reallocated
 182 Token const Parser::curr_token() const
 183 {
 184         static const Token dummy;
 185         return pos_ > 0 ? tokens_[pos_ - 1] : dummy;
 186 }
 187
 188
 189 // We return a copy here because the tokens_ vector may get reallocated
 190 Token const Parser::next_token()
 191 {
 192         static const Token dummy;
 193         return good() ? tokens_[pos_] : dummy;
 194 }
 195
 196
 197 // We return a copy here because the tokens_ vector may get reallocated
 198 Token const Parser::get_token()
 199 {
 200         static const Token dummy;
 201         //cerr << "looking at token " << tokens_[pos_] << " pos: " << pos_ << '\n';
 202         return good() ? tokens_[pos_++] : dummy;
 203 }
 204
 205
 206 bool Parser::isParagraph()
 207 {
 208         // A new paragraph in TeX ist started
 209         // - either by a newline, following any amount of whitespace
 210         //   characters (including zero), and another newline
 211         // - or the token \par
 212         if (curr_token().cat() == catNewline &&
 213             (curr_token().cs().size() > 1 ||
 214              (next_token().cat() == catSpace &&
 215               pos_ < tokens_.size() - 1 &&
 216               tokens_[pos_ + 1].cat() == catNewline)))
 217                 return true;
 218         if (curr_token().cat() == catEscape && curr_token().cs() == "par")
 219                 return true;
 220         return false;
 221 }
 222
 223
 224 void Parser::skip_spaces(bool skip_comments)
 225 {
 226         // We just silently return if we have no more tokens.
 227         // skip_spaces() should be callable at any time,
 228         // the caller must check p::good() anyway.
 229         while (good()) {
 230                 get_token();
 231                 if (isParagraph()) {
 232                         putback();
 233                         break;
 234                 }
 235                 if ( curr_token().cat() == catSpace ||
 236                      curr_token().cat() == catNewline ||
 237                     (curr_token().cat() == catComment && curr_token().cs().empty()))
 238                         continue;
 239                 if (skip_comments && curr_token().cat() == catComment)
 240                         cerr << "  Ignoring comment: " << curr_token().asInput();
 241                 else {
 242                         putback();
 243                         break;
 244                 }
 245         }
 246 }
 247
 248
 249 void Parser::unskip_spaces(bool skip_comments)
 250 {
 251         while (pos_ > 0) {
 252                 if ( curr_token().cat() == catSpace ||
 253                     (curr_token().cat() == catNewline && curr_token().cs().size() == 1))
 254                         putback();
 255                 else if (skip_comments && curr_token().cat() == catComment) {
 256                         // TODO: Get rid of this
 257                         cerr << "Unignoring comment: " << curr_token().asInput();
 258                         putback();
 259                 }
 260                 else
 261                         break;
 262         }
 263 }
 264
 265
 266 void Parser::putback()
 267 {
 268         --pos_;
 269 }
 270
 271
 272 bool Parser::good()
 273 {
 274         if (pos_ < tokens_.size())
 275                 return true;
 276         tokenize_one();
 277         return pos_ < tokens_.size();
 278 }
 279
 280
 281 char Parser::getChar()
 282 {
 283         if (!good())
 284                 error("The input stream is not well...");
 285         return get_token().character();
 286 }
 287
 288
 289 Parser::Arg Parser::getFullArg(char left, char right)
 290 {
 291         skip_spaces(true);
 292
 293         // This is needed if a partial file ends with a command without arguments,
 294         // e. g. \medskip
 295         if (! good())
 296                 return make_pair(false, string());
 297
 298         string result;
 299         char c = getChar();
 300
 301         if (c != left) {
 302                 putback();
 303                 return make_pair(false, string());
 304         } else
 305                 while ((c = getChar()) != right && good()) {
 306                         // Ignore comments
 307                         if (curr_token().cat() == catComment) {
 308                                 if (!curr_token().cs().empty())
 309                                         cerr << "Ignoring comment: " << curr_token().asInput();
 310                         }
 311                         else
 312                                 result += curr_token().asInput();
 313                 }
 314
 315         return make_pair(true, result);
 316 }
 317
 318
 319 string Parser::getArg(char left, char right)
 320 {
 321         return getFullArg(left, right).second;
 322 }
 323
 324
 325 string Parser::getFullOpt()
 326 {
 327         Arg arg = getFullArg('[', ']');
 328         if (arg.first)
 329                 return '[' + arg.second + ']';
 330         return string();
 331 }
 332
 333
 334 string Parser::getOpt()
 335 {
 336         string const res = getArg('[', ']');
 337         return res.empty() ? string() : '[' + res + ']';
 338 }
 339
 340
 341 string Parser::getOptContent()
 342 // the same as getOpt but without the brackets
 343 {
 344         string const res = getArg('[', ']');
 345         return res.empty() ? string() : res;
 346 }
 347
 348
 349 string Parser::getFullParentheseArg()
 350 {
 351         Arg arg = getFullArg('(', ')');
 352         if (arg.first)
 353                 return '(' + arg.second + ')';
 354         return string();
 355 }
 356
 357
 358 string const Parser::verbatimEnvironment(string const & name)
 359 {
 360         if (!good())
 361                 return string();
 362
 363         ostringstream os;
 364         for (Token t = get_token(); good(); t = get_token()) {
 365                 if (t.cat() == catBegin) {
 366                         putback();
 367                         os << '{' << verbatim_item() << '}';
 368                 } else if (t.asInput() == "\\begin") {
 369                         string const env = getArg('{', '}');
 370                         os << "\\begin{" << env << '}'
 371                            << verbatimEnvironment(env)
 372                            << "\\end{" << env << '}';
 373                 } else if (t.asInput() == "\\end") {
 374                         string const end = getArg('{', '}');
 375                         if (end != name)
 376                                 cerr << "\\end{" << end
 377                                      << "} does not match \\begin{" << name
 378                                      << "}." << endl;
 379                         return os.str();
 380                 } else
 381                         os << t.asInput();
 382         }
 383         cerr << "unexpected end of input" << endl;
 384         return os.str();
 385 }
 386
 387
 388 void Parser::tokenize_one()
 389 {
 390         catInit();
 391         char_type c;
 392         if (!is_.get(c))
 393                 return;
 394
 395         switch (catcode(c)) {
 396         case catSpace: {
 397                 docstring s(1, c);
 398                 while (is_.get(c) && catcode(c) == catSpace)
 399                         s += c;
 400                 if (catcode(c) != catSpace)
 401                         is_.putback(c);
 402                 push_back(Token(s, catSpace));
 403                 break;
 404         }
 405
 406         case catNewline: {
 407                 ++lineno_;
 408                 docstring s(1, getNewline(is_, c));
 409                 while (is_.get(c) && catcode(c) == catNewline) {
 410                         ++lineno_;
 411                         s += getNewline(is_, c);
 412                 }
 413                 if (catcode(c) != catNewline)
 414                         is_.putback(c);
 415                 push_back(Token(s, catNewline));
 416                 break;
 417         }
 418
 419         case catComment: {
 420                 // We don't treat "%\n" combinations here specially because
 421                 // we want to preserve them in the preamble
 422                 docstring s;
 423                 while (is_.get(c) && catcode(c) != catNewline)
 424                         s += c;
 425                 // handle possible DOS line ending
 426                 if (catcode(c) == catNewline)
 427                         c = getNewline(is_, c);
 428                 // Note: The '%' at the beginning and the '\n' at the end
 429                 // of the comment are not stored.
 430                 ++lineno_;
 431                 push_back(Token(s, catComment));
 432                 break;
 433         }
 434
 435         case catEscape: {
 436                 is_.get(c);
 437                 if (!is_) {
 438                         error("unexpected end of input");
 439                 } else {
 440                         docstring s(1, c);
 441                         if (catcode(c) == catLetter) {
 442                                 // collect letters
 443                                 while (is_.get(c) && catcode(c) == catLetter)
 444                                         s += c;
 445                                 if (catcode(c) != catLetter)
 446                                         is_.putback(c);
 447                         }
 448                         push_back(Token(s, catEscape));
 449                 }
 450                 break;
 451         }
 452
 453         case catIgnore: {
 454                 cerr << "ignoring a char: " << c << "\n";
 455                 break;
 456         }
 457
 458         default:
 459                 push_back(Token(docstring(1, c), catcode(c)));
 460         }
 461         //cerr << tokens_.back();
 462 }
 463
 464
 465 void Parser::dump() const
 466 {
 467         cerr << "\nTokens: ";
 468         for (unsigned i = 0; i < tokens_.size(); ++i) {
 469                 if (i == pos_)
 470                         cerr << " <#> ";
 471                 cerr << tokens_[i];
 472         }
 473         cerr << " pos: " << pos_ << "\n";
 474 }
 475
 476
 477 void Parser::error(string const & msg)
 478 {
 479         cerr << "Line ~" << lineno_ << ":  parse error: " << msg << endl;
 480         dump();
 481         //exit(1);
 482 }
 483
 484
 485 string Parser::verbatimOption()
 486 {
 487         string res;
 488         if (next_token().character() == '[') {
 489                 Token t = get_token();
 490                 for (t = get_token(); t.character() != ']' && good(); t = get_token()) {
 491                         if (t.cat() == catBegin) {
 492                                 putback();
 493                                 res += '{' + verbatim_item() + '}';
 494                         } else
 495                                 res += t.asString();
 496                 }
 497         }
 498         return res;
 499 }
 500
 501
 502 string Parser::verbatim_item()
 503 {
 504         if (!good())
 505                 error("stream bad");
 506         skip_spaces();
 507         if (next_token().cat() == catBegin) {
 508                 Token t = get_token(); // skip brace
 509                 string res;
 510                 for (Token t = get_token(); t.cat() != catEnd && good(); t = get_token()) {
 511                         if (t.cat() == catBegin) {
 512                                 putback();
 513                                 res += '{' + verbatim_item() + '}';
 514                         }
 515                         else
 516                                 res += t.asInput();
 517                 }
 518                 return res;
 519         }
 520         return get_token().asInput();
 521 }
 522
 523
 524 void Parser::reset()
 525 {
 526         pos_ = 0;
 527 }
 528
 529
 530 void Parser::setCatCode(char c, CatCode cat)
 531 {
 532         theCatcode[(unsigned char)c] = cat;
 533 }
 534
 535
 536 CatCode Parser::getCatCode(char c) const
 537 {
 538         return theCatcode[(unsigned char)c];
 539 }
 540
 541
 542 } // namespace lyx