src/lyxlex.C

   1 //  Generalized simple lexical analizer.
   2 //  It can be used for simple syntax parsers, like lyxrc,
   3 //  texclass and others to come.   [asierra30/03/96]
   4 //
   5 //   (C) 1996 Lyx Team.
   6
   7 #include <config.h>
   8
   9 #include <cstdlib>
  10
  11 #ifdef __GNUG__
  12 #pragma implementation "lyxlex.h"
  13 #endif
  14
  15 #include "lyxlex.h"
  16 #include "debug.h"
  17 #include "support/filetools.h"
  18
  19 LyXLex::LyXLex(keyword_item * tab, int num)
  20         : is(&fb__), table(tab), no_items(num)
  21 {
  22
  23         status = 0;
  24         pushed = 0;
  25 }
  26
  27
  28 void LyXLex::pushTable(keyword_item * tab, int num)
  29 {
  30         pushed_table * tmppu = new pushed_table;
  31         tmppu->next = pushed;
  32         tmppu->table_elem = table;
  33         tmppu->table_siz = no_items;
  34         pushed = tmppu;
  35         table = tab;
  36         no_items = num;
  37 }
  38
  39
  40 void LyXLex::popTable()
  41 {
  42         if (pushed == 0)
  43                 lyxerr << "LyXLex error: nothing to pop!" << endl;
  44
  45         pushed_table * tmp;
  46         tmp = pushed;
  47         table = tmp->table_elem;
  48         no_items = tmp->table_siz;
  49         tmp->table_elem = 0;
  50         pushed = tmp->next;
  51         delete tmp;
  52 }
  53
  54
  55 void LyXLex::printTable()
  56 {
  57         lyxerr << "\nNumber of tags: " << no_items << endl;
  58         for(int i= 0; i < no_items; ++i)
  59                 lyxerr << "table[" << i
  60                        << "]:  tag: `" << table[i].tag
  61                        << "'  code:" << table[i].code << endl;
  62         lyxerr << endl;
  63 }
  64
  65
  66 void LyXLex::printError(string const & message) const
  67 {
  68         string tmpmsg = subst(message, "$$Token", GetString());
  69         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  70                << " of file " << MakeDisplayPath(name) << ']' << endl;
  71 }
  72
  73
  74 bool LyXLex::setFile(string const & filename)
  75 {
  76         if (fb__.is_open())
  77                 lyxerr << "Error in LyXLex::setFile: "
  78                         "file or stream already set." << endl;
  79         fb__.open(filename.c_str(), ios::in);
  80         is.rdbuf(&fb__);
  81         name = filename;
  82         lineno = 0;
  83         return fb__.is_open() && is.good();
  84 }
  85
  86
  87 void LyXLex::setStream(istream & i)
  88 {
  89         if (fb__.is_open() || is.rdbuf()->in_avail())
  90                 lyxerr << "Error in LyXLex::setStream: "
  91                         "file or stream already set." << endl;
  92         is.rdbuf(i.rdbuf());
  93         lineno = 0;
  94 }
  95
  96
  97 int LyXLex::lex()
  98 {
  99         //NOTE: possible bug.
 100         if (next() && status == LEX_TOKEN)
 101                 return search_kw(buff);
 102         else
 103                 return status;
 104 }
 105
 106
 107 int LyXLex::GetInteger() const
 108 {
 109         if (buff[0] > ' ')
 110                 return atoi(buff);
 111         else {
 112                 printError("Bad integer `$$Token'");
 113                 return -1;
 114         }
 115 }
 116
 117
 118 float LyXLex::GetFloat() const
 119 {
 120    if (buff[0] > ' ')
 121        return atof(buff);
 122    else {
 123         printError("Bad float `$$Token'");
 124         return -1;
 125    }
 126 }
 127
 128
 129 string LyXLex::GetString() const
 130 {
 131         return string(buff);
 132 }
 133
 134
 135 // I would prefer to give a tag number instead of an explicit token
 136 // here, but it is not possible because Buffer::readLyXformat2 uses
 137 // explicit tokens (JMarc)
 138 string LyXLex::getLongString(string const & endtoken)
 139 {
 140         string str, prefix;
 141         bool firstline = true;
 142
 143         while (IsOK()) {
 144                 if (!EatLine())
 145                         // blank line in the file being read
 146                         continue;
 147
 148                 string const token = frontStrip(strip(GetString()), " \t");
 149
 150                 lyxerr[Debug::PARSER] << "LongString: `"
 151                                       << GetString() << '\'' << endl;
 152
 153                 // We do a case independent comparison, like search_kw
 154                 // does.
 155                 if (compare_no_case(token, endtoken) != 0) {
 156                         string tmpstr = GetString();
 157                         if (firstline) {
 158                                 unsigned int i = 0;
 159                                 while(i < tmpstr.length()
 160                                       && tmpstr[i] == ' ') {
 161                                         ++i;
 162                                         prefix += ' ';
 163                                 }
 164                                 firstline = false;
 165                                 lyxerr[Debug::PARSER] << "Prefix = `" << prefix
 166                                                       << '\'' << endl;
 167                         }
 168
 169                         if (!prefix.empty()
 170                             && prefixIs(tmpstr, prefix.c_str())) {
 171                                 tmpstr.erase(0, prefix.length() - 1);
 172                         }
 173                         str += tmpstr + '\n';
 174                 }
 175                 else // token == endtoken
 176                         break;
 177         }
 178         if (!IsOK())
 179                 printError("Long string not ended by `" + endtoken + '\'');
 180
 181         return str;
 182 }
 183
 184
 185 bool LyXLex::GetBool() const
 186 {
 187         if (compare(buff, "true") == 0)
 188                 return true;
 189         else if (compare(buff, "false") != 0)
 190                 printError("Bad boolean `$$Token'. Use \"false\" or \"true\"");
 191         return false;
 192 }
 193
 194
 195 bool LyXLex::EatLine()
 196 {
 197         int i = 0;
 198         unsigned char c = '\0';
 199
 200         while(is && c != '\n' && i != (LEX_MAX_BUFF - 1)) {
 201                 is.get(c);
 202                 if (c != '\r')
 203                         buff[i++] = c;
 204         }
 205         if (i == (LEX_MAX_BUFF - 1) && c != '\n') {
 206                 printError("Line too long");
 207                 c = '\n'; // Pretend we had an end of line
 208                 --lineno; // but don't increase line counter (netto effect)
 209                 ++i; // and preserve last character read.
 210         }
 211         if (c == '\n') {
 212                 ++lineno;
 213                 buff[--i] = '\0'; // i can never be 0 here, so no danger
 214                 status = LEX_DATA;
 215                 return true;
 216         } else {
 217                 buff[i] = '\0';
 218                 return false;
 219         }
 220 }
 221
 222
 223 int LyXLex::search_kw(char const * const tag) const
 224 {
 225         int m, k = 0 , l = 0, r = no_items;
 226
 227         while (l < r) {
 228                 m = (l + r) / 2;
 229
 230                 if (lyxerr.debugging(Debug::PARSER)) {
 231                         lyxerr << "LyXLex::search_kw: elem " << m
 232                                << " tag " << table[m].tag
 233                                << " search tag " << tag
 234                                << endl;
 235                 }
 236
 237                 if (table[m].tag)
 238                         k = compare_no_case(table[m].tag, tag);
 239                 if (k == 0)
 240                         return table[m].code;
 241                 else
 242                         if (k < 0) l = m + 1; else r = m;
 243         }
 244         return LEX_UNDEF;
 245 }
 246
 247
 248 bool LyXLex::next(bool esc)
 249 {
 250         if (!esc) {
 251                 unsigned char c; // getc() returns an int
 252
 253                 status = 0;
 254                 while (is && !status) {
 255                         is.get(c);
 256                         if (c == '#') {
 257                                 // Read rest of line (fast :-)
 258                                 is.get(buff, sizeof(buff));
 259                                 lyxerr[Debug::LYXLEX] << "Comment read: " << c << buff << endl;
 260                                 ++lineno;
 261                                 continue;
 262                         }
 263
 264                         if (c == '\"') {
 265                                 int i = -1;
 266                                 do {
 267                                         is.get(c);
 268                                         if (c != '\r')
 269                                                 buff[++i] = c;
 270                                 } while (c != '\"' && c != '\n' && is &&
 271                                          i != (LEX_MAX_BUFF - 2));
 272
 273                                 if (i == (LEX_MAX_BUFF - 2)) {
 274                                         printError("Line too long");
 275                                         c = '\"'; // Pretend we got a "
 276                                         ++i;
 277                                 }
 278
 279                                 if (c != '\"') {
 280                                         printError("Missing quote");
 281                                         if (c == '\n')
 282                                                 ++lineno;
 283                                 }
 284
 285                                 buff[i] = '\0';
 286                                 status = LEX_DATA;
 287                                 break;
 288                         }
 289
 290                         if (c == ',')
 291                                 continue;              /* Skip ','s */
 292
 293                         // using relational operators with chars other
 294                         // than == and != is not safe. And if it is done
 295                         // the type _have_ to be unsigned. It usually a
 296                         // lot better to use the functions from cctype
 297                         if (c > ' ' && is)  {
 298 #warning Verify this! (Lgb)
 299                                 //if (isalnum(static_cast<unsigned char>(c)) && is) {
 300                                 int i = 0;
 301                                 do {
 302                                         buff[i++] = c;
 303                                         is.get(c);
 304                                 } while (c > ' ' && c != ',' && is
 305                                 //} while (isalnum(static_cast<unsigned char>(c))
 306                                          //&& c != ',' && is
 307                                          && (i != LEX_MAX_BUFF - 1) );
 308                                 if (i == LEX_MAX_BUFF - 1) {
 309                                         printError("Line too long");
 310                                 }
 311                                 buff[i] = '\0';
 312                                 status = LEX_TOKEN;
 313                         }
 314
 315                         if (c == '\r' && is) {
 316                                 // The Windows support has lead to the
 317                                 // possibility of "\r\n" at the end of
 318                                 // a line.  This will stop LyX choking
 319                                 // when it expected to find a '\n'
 320                                 is.get(c);
 321                         }
 322
 323                         if (c == '\n')
 324                                 ++lineno;
 325
 326                 }
 327                 if (status) return true;
 328
 329                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 330                 buff[0] = '\0';
 331                 return false;
 332         } else {
 333                 unsigned char c; // getc() returns an int
 334
 335                 status = 0;
 336                 while (is && !status) {
 337                         is.get(c);
 338
 339                         // skip ','s
 340                         if (c == ',') continue;
 341
 342                         if (c == '\\') {
 343                                 // escape
 344                                 int i = 0;
 345                                 do {
 346                                         if (c == '\\') {
 347                                                 // escape the next char
 348                                                 is.get(c);
 349                                         }
 350                                         buff[i++] = c;
 351                                         is.get(c);
 352                                 } while (c > ' ' && c != ',' && is
 353                                 //} while (isalnum(static_cast<unsigned char>(c))
 354                                          //&& c != ',' && is
 355                                          && (i != LEX_MAX_BUFF - 1) );
 356                                 if (i == LEX_MAX_BUFF - 1) {
 357                                         printError("Line too long");
 358                                 }
 359                                 buff[i] = '\0';
 360                                 status = LEX_TOKEN;
 361                                 continue;
 362                         }
 363
 364                         if (c == '#') {
 365                                 // Read rest of line (fast :-)
 366                                 is.get(buff, sizeof(buff));
 367                                 lyxerr[Debug::LYXLEX] << "Comment read: " << c << buff << endl;
 368                                 ++lineno;
 369                                 continue;
 370                         }
 371
 372                         // string
 373                         if (c == '\"') {
 374                                 int i = -1;
 375                                 bool escaped = false;
 376                                 do {
 377                                         escaped = false;
 378                                         is.get(c);
 379                                         if (c == '\r') continue;
 380                                         if (c == '\\') {
 381                                                 // escape the next char
 382                                                 is.get(c);
 383                                                 escaped = true;
 384                                         }
 385                                         buff[++i] = c;
 386
 387                                         if (!escaped && c == '\"') break;
 388                                 } while (c != '\n' && is &&
 389                                          i != (LEX_MAX_BUFF - 2));
 390
 391                                 if (i == (LEX_MAX_BUFF - 2)) {
 392                                         printError("Line too long");
 393                                         c = '\"'; // Pretend we got a "
 394                                         ++i;
 395                                 }
 396
 397                                 if (c != '\"') {
 398                                         printError("Missing quote");
 399                                         if (c == '\n')
 400                                                 ++lineno;
 401                                 }
 402
 403                                 buff[i] = '\0';
 404                                 status = LEX_DATA;
 405                                 break;
 406                         }
 407
 408                         if (c > ' ' && is) {
 409                                 //if (isalnum(static_cast<unsigned char>(c)) && is) {
 410                                 int i = 0;
 411                                 do {
 412                                         if (c == '\\') {
 413                                                 // escape the next char
 414                                                 is.get(c);
 415                                                 //escaped = true;
 416                                         }
 417                                         buff[i++] = c;
 418                                         is.get(c);
 419                                 } while (c > ' ' && c != ',' && is
 420                                 //} while (isalnum(static_cast<unsigned char>(c))
 421                                          //!= ',' && is
 422                                          && (i != LEX_MAX_BUFF-1) );
 423                                 if (i == LEX_MAX_BUFF-1) {
 424                                         printError("Line too long");
 425                                 }
 426                                 buff[i] = '\0';
 427                                 status = LEX_TOKEN;
 428                         }
 429                         // new line
 430                         if (c == '\n')
 431                                 ++lineno;
 432                 }
 433
 434                 if (status) return true;
 435
 436                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 437                 buff[0] = '\0';
 438                 return false;
 439         }
 440 }
 441
 442
 443 bool LyXLex::nextToken()
 444 {
 445         status = 0;
 446         while (is && !status) {
 447                 unsigned char c;
 448                 is.get(c);
 449
 450                 if (c >= ' ' && is) {
 451                         //if (isprint(static_cast<unsigned char>(c)) && is) {
 452                         int i = 0;
 453                         if (c == '\\') { // first char == '\\'
 454                                 do {
 455                                         buff[i++] = c;
 456                                         is.get(c);
 457                                 } while (c > ' ' && c != '\\' && is
 458                                 //} while (isalnum(static_cast<unsigned char>(c))
 459                                 //       && c != '\\' && is
 460                                          && i != (LEX_MAX_BUFF-1));
 461                         } else {
 462                                 do {
 463                                         buff[i++] = c;
 464                                         is.get(c);
 465                                 } while (c >= ' ' && c != '\\' && is
 466                                 //} while (isprint(static_cast<unsigned char>(c))
 467                                          // && c != '\\' && is
 468                                          && i != (LEX_MAX_BUFF-1));
 469                         }
 470
 471                         if (i == (LEX_MAX_BUFF - 1)) {
 472                                 printError("Line too long");
 473                         }
 474
 475                         if (c == '\\') is.putback(c); // put it back
 476                         buff[i] = '\0';
 477                         status = LEX_TOKEN;
 478                 }
 479
 480                 if (c == '\n')
 481                         ++lineno;
 482
 483         }
 484         if (status)  return true;
 485
 486         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 487         buff[0] = '\0';
 488         return false;
 489 }
 490
 491
 492 int LyXLex::FindToken(char const * str[])
 493 {
 494    int i = -1;
 495
 496    if (next()) {
 497       if (compare(buff, "default")) {
 498          for (i = 0; str[i][0] && compare(str[i], buff); ++i);
 499          if (!str[i][0]) {
 500             printError("Unknown argument `$$Token'");
 501             i = -1;
 502          }
 503       }
 504    } else
 505      printError("file ended while scanning string token");
 506    return i;
 507 }
 508
 509
 510 int LyXLex::CheckToken(char const * str[], int print_error)
 511 {
 512    int i = -1;
 513
 514    if (compare(buff, "default")) {
 515        for (i = 0; str[i][0] && compare(str[i], buff); ++i);
 516        if (!str[i][0]) {
 517            if (print_error)
 518                printError("Unknown argument `$$Token'");
 519            i = -1;
 520        }
 521    }
 522    return i;
 523 }