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