src/lyxlex_pimpl.C

   1 #include <config.h>
   2
   3 #ifdef __GNUG__
   4 #pragma implementation
   5 #endif
   6 #include <algorithm>
   7
   8 #include "lyxlex_pimpl.h"
   9 #include "support/lyxalgo.h"
  10 #include "support/filetools.h"
  11 #include "support/lstrings.h"
  12 #include "debug.h"
  13
  14 using std::sort;
  15 using std::ostream;
  16 using std::ios;
  17 using std::istream;
  18 using std::endl;
  19 using std::lower_bound;
  20
  21 // namespace {
  22 struct compare_tags {
  23         // used by lower_bound, sort and sorted
  24         inline
  25         int operator()(keyword_item const & a, keyword_item const & b) const {
  26                 // we use the ascii version, because in turkish, 'i'
  27                 // is not the lowercase version of 'I', and thus
  28                 // turkish locale breaks parsing of tags.
  29                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  30         }
  31 };
  32 // } // end of anon namespace
  33
  34
  35 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  36         : is(&fb__), table(tab), no_items(num),
  37           status(0), lineno(0), commentChar('#')
  38 {
  39         verifyTable();
  40 }
  41
  42
  43 string const LyXLex::Pimpl::getString() const
  44 {
  45         return string(buff);
  46 }
  47
  48
  49 void LyXLex::Pimpl::printError(string const & message) const
  50 {
  51         string const tmpmsg = subst(message, "$$Token", getString());
  52         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  53                << " of file " << MakeDisplayPath(name) << ']' << endl;
  54 }
  55
  56
  57 void LyXLex::Pimpl::printTable(ostream & os)
  58 {
  59         os << "\nNumber of tags: " << no_items << '\n';
  60         for (int i= 0; i < no_items; ++i)
  61                 os << "table[" << i
  62                    << "]:  tag: `" << table[i].tag
  63                    << "'  code:" << table[i].code << '\n';
  64         os.flush();
  65 }
  66
  67
  68 void LyXLex::Pimpl::verifyTable()
  69 {
  70         // Check if the table is sorted and if not, sort it.
  71         if (table
  72             && !lyx::sorted(table, table + no_items, compare_tags())) {
  73                 lyxerr << "The table passed to LyXLex is not sorted!\n"
  74                        << "Tell the developers to fix it!" << endl;
  75                 // We sort it anyway to avoid problems.
  76                 lyxerr << "\nUnsorted:\n";
  77                 printTable(lyxerr);
  78
  79                 sort(table, table + no_items, compare_tags());
  80                 lyxerr << "\nSorted:\n";
  81                 printTable(lyxerr);
  82         }
  83 }
  84
  85
  86 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
  87 {
  88         pushed_table tmppu(table, no_items);
  89         pushed.push(tmppu);
  90
  91         table = tab;
  92         no_items = num;
  93
  94         verifyTable();
  95 }
  96
  97
  98 void LyXLex::Pimpl::popTable()
  99 {
 100         if (pushed.empty()) {
 101                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 102                 return;
 103         }
 104
 105         pushed_table tmp = pushed.top();
 106         pushed.pop();
 107         table = tmp.table_elem;
 108         no_items = tmp.table_siz;
 109 }
 110
 111
 112 bool LyXLex::Pimpl::setFile(string const & filename)
 113 {
 114         // The check only outputs a debug message, because it triggers
 115         // a bug in compaq cxx 6.2, where is_open() returns 'true' for a
 116         // fresh new filebuf.  (JMarc)
 117         if (fb__.is_open() || is.tellg() > 0)
 118                 lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 119                         "file or stream already set." << endl;
 120         fb__.open(filename.c_str(), ios::in);
 121         is.rdbuf(&fb__);
 122         name = filename;
 123         lineno = 0;
 124         return fb__.is_open() && is.good();
 125 }
 126
 127
 128 void LyXLex::Pimpl::setStream(istream & i)
 129 {
 130         if (fb__.is_open() || is.tellg() > 0)
 131                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 132                         "file or stream already set." << endl;
 133         is.rdbuf(i.rdbuf());
 134         lineno = 0;
 135 }
 136
 137 void LyXLex::Pimpl::setCommentChar(char c)
 138 {
 139         commentChar = c;
 140 }
 141
 142
 143 bool LyXLex::Pimpl::next(bool esc /* = false */)
 144 {
 145         if (!pushTok.empty()) {
 146                 // There can have been a whole line pushed so
 147                 // we extract the first word and leaves the rest
 148                 // in pushTok. (Lgb)
 149                 if (pushTok.find(' ') != string::npos) {
 150                         string tmp;
 151                         pushTok = split(pushTok, tmp, ' ');
 152                         tmp.copy(buff, string::npos);
 153                         buff[tmp.length()] = '\0';
 154                         return true;
 155                 } else {
 156                         pushTok.copy(buff, string::npos);
 157                         buff[pushTok.length()] = '\0';
 158                         pushTok.erase();
 159                         return true;
 160                 }
 161         }
 162         if (!esc) {
 163                 unsigned char c = 0; // getc() returns an int
 164                 char cc = 0;
 165                 status = 0;
 166                 while (is && !status) {
 167                         is.get(cc);
 168                         c = cc;
 169                         if (c == commentChar) {
 170                                 // Read rest of line (fast :-)
 171                                 // That is not fast... (Lgb)
 172 #if 1
 173                                 is.getline(buff, sizeof(buff));
 174                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 175                                                       << buff << "'" << endl;
 176 #else
 177                                 // unfortunately ignore is buggy (Lgb)
 178                                 is.ignore(100, '\n');
 179 #endif
 180                                 ++lineno;
 181                                 continue;
 182                         }
 183
 184                         if (c == '\"') {
 185                                 int i = -1;
 186                                 do {
 187                                         is.get(cc);
 188                                         c = cc;
 189                                         if (c != '\r')
 190                                                 buff[++i] = c;
 191                                 } while (c != '\"' && c != '\n' && is &&
 192                                          i != (LEX_MAX_BUFF - 2));
 193
 194                                 if (i == (LEX_MAX_BUFF - 2)) {
 195                                         printError("Line too long");
 196                                         c = '\"'; // Pretend we got a "
 197                                         ++i;
 198                                 }
 199
 200                                 if (c != '\"') {
 201                                         printError("Missing quote");
 202                                         if (c == '\n')
 203                                                 ++lineno;
 204                                 }
 205
 206                                 buff[i] = '\0';
 207                                 status = LEX_DATA;
 208                                 break;
 209                         }
 210
 211                         if (c == ',')
 212                                 continue;              /* Skip ','s */
 213
 214                                 // using relational operators with chars other
 215                                 // than == and != is not safe. And if it is done
 216                                 // the type _have_ to be unsigned. It usually a
 217                                 // lot better to use the functions from cctype
 218                         if (c > ' ' && is)  {
 219                                 int i = 0;
 220                                 do {
 221                                         buff[i++] = c;
 222                                         is.get(cc);
 223                                         c = cc;
 224                                 } while (c > ' ' && c != ',' && is
 225                                          && (i != LEX_MAX_BUFF - 1));
 226                                 if (i == LEX_MAX_BUFF - 1) {
 227                                         printError("Line too long");
 228                                 }
 229                                 buff[i] = '\0';
 230                                 status = LEX_TOKEN;
 231                         }
 232
 233                         if (c == '\r' && is) {
 234                                 // The Windows support has lead to the
 235                                 // possibility of "\r\n" at the end of
 236                                 // a line.  This will stop LyX choking
 237                                 // when it expected to find a '\n'
 238                                 is.get(cc);
 239                                 c = cc;
 240                         }
 241
 242                         if (c == '\n')
 243                                 ++lineno;
 244
 245                 }
 246                 if (status) return true;
 247
 248                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 249                 buff[0] = '\0';
 250                 return false;
 251         } else {
 252                 unsigned char c = 0; // getc() returns an int
 253                 char cc = 0;
 254
 255                 status = 0;
 256                 while (is && !status) {
 257                         is.get(cc);
 258                         c = cc;
 259
 260                         // skip ','s
 261                         if (c == ',') continue;
 262
 263                         if (c == '\\') {
 264                                 // escape
 265                                 int i = 0;
 266                                 do {
 267                                         if (c == '\\') {
 268                                                 // escape the next char
 269                                                 is.get(cc);
 270                                                 c = cc;
 271                                         }
 272                                         buff[i++] = c;
 273                                         is.get(cc);
 274                                         c = cc;
 275                                 } while (c > ' ' && c != ',' && is
 276                                          && (i != LEX_MAX_BUFF - 1));
 277                                 if (i == LEX_MAX_BUFF - 1) {
 278                                         printError("Line too long");
 279                                 }
 280                                 buff[i] = '\0';
 281                                 status = LEX_TOKEN;
 282                                 continue;
 283                         }
 284
 285                         if (c == commentChar) {
 286                                 // Read rest of line (fast :-)
 287                                 // That is still not fast... (Lgb)
 288 #if 1
 289                                 is.getline(buff, sizeof(buff));
 290                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 291                                                       << buff << "'" << endl;
 292 #else
 293                                 // but ignore is also still buggy (Lgb)
 294                                 // This is fast (Lgb)
 295                                 is.ignore(100, '\n');
 296 #endif
 297                                 ++lineno;
 298                                 continue;
 299                         }
 300
 301                         // string
 302                         if (c == '\"') {
 303                                 int i = -1;
 304                                 bool escaped = false;
 305                                 do {
 306                                         escaped = false;
 307                                         is.get(cc);
 308                                         c = cc;
 309                                         if (c == '\r') continue;
 310                                         if (c == '\\') {
 311                                                 // escape the next char
 312                                                 is.get(cc);
 313                                                 c = cc;
 314                                                 if (c == '\"' || c == '\\')
 315                                                         escaped = true;
 316                                                 else
 317                                                         buff[++i] = '\\';
 318                                         }
 319                                         buff[++i] = c;
 320
 321                                         if (!escaped && c == '\"') break;
 322                                 } while (c != '\n' && is &&
 323                                          i != (LEX_MAX_BUFF - 2));
 324
 325                                 if (i == (LEX_MAX_BUFF - 2)) {
 326                                         printError("Line too long");
 327                                         c = '\"'; // Pretend we got a "
 328                                         ++i;
 329                                 }
 330
 331                                 if (c != '\"') {
 332                                         printError("Missing quote");
 333                                         if (c == '\n')
 334                                                 ++lineno;
 335                                 }
 336
 337                                 buff[i] = '\0';
 338                                 status = LEX_DATA;
 339                                 break;
 340                         }
 341
 342                         if (c > ' ' && is) {
 343                                 int i = 0;
 344                                 do {
 345                                         if (c == '\\') {
 346                                                 // escape the next char
 347                                                 is.get(cc);
 348                                                 c = cc;
 349                                                 //escaped = true;
 350                                         }
 351                                         buff[i++] = c;
 352                                         is.get(cc);
 353                                         c = cc;
 354                                 } while (c > ' ' && 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                         }
 362                         // new line
 363                         if (c == '\n')
 364                                 ++lineno;
 365                 }
 366
 367                 if (status) return true;
 368
 369                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 370                 buff[0] = '\0';
 371                 return false;
 372         }
 373 }
 374
 375
 376 int LyXLex::Pimpl::search_kw(char const * const tag) const
 377 {
 378         keyword_item search_tag = { tag, 0 };
 379         keyword_item * res =
 380                 lower_bound(table, table + no_items,
 381                             search_tag, compare_tags());
 382         if (res != table + no_items
 383             && !compare_no_case(res->tag, tag))
 384                 return res->code;
 385         return LEX_UNDEF;
 386 }
 387
 388
 389 int LyXLex::Pimpl::lex()
 390 {
 391         //NOTE: possible bug.
 392         if (next() && status == LEX_TOKEN)
 393                 return search_kw(buff);
 394         else
 395                 return status;
 396 }
 397
 398
 399 bool LyXLex::Pimpl::eatLine()
 400 {
 401         int i = 0;
 402         unsigned char c = '\0';
 403         char cc = 0;
 404         while (is && c != '\n' && i != (LEX_MAX_BUFF - 1)) {
 405                 is.get(cc);
 406                 c = cc;
 407                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 408                 //                    << c << "'" << endl;
 409                 if (c != '\r')
 410                         buff[i++] = c;
 411         }
 412         if (i == (LEX_MAX_BUFF - 1) && c != '\n') {
 413                 printError("Line too long");
 414                 c = '\n'; // Pretend we had an end of line
 415                 --lineno; // but don't increase line counter (netto effect)
 416                 ++i; // and preserve last character read.
 417         }
 418         if (c == '\n') {
 419                 ++lineno;
 420                 buff[--i] = '\0'; // i can never be 0 here, so no danger
 421                 status = LEX_DATA;
 422                 return true;
 423         } else {
 424                 buff[i] = '\0';
 425                 return false;
 426         }
 427 }
 428
 429
 430 bool LyXLex::Pimpl::nextToken()
 431 {
 432         if (!pushTok.empty()) {
 433                 // There can have been a whole line pushed so
 434                 // we extract the first word and leaves the rest
 435                 // in pushTok. (Lgb)
 436                 if (pushTok.find(' ') != string::npos) {
 437                         string tmp;
 438                         pushTok = split(pushTok, tmp, ' ');
 439                         tmp.copy(buff, string::npos);
 440                         buff[tmp.length()] = '\0';
 441                         return true;
 442                 } else {
 443                         pushTok.copy(buff, string::npos);
 444                         buff[pushTok.length()] = '\0';
 445                         pushTok.erase();
 446                         return true;
 447                 }
 448         }
 449
 450         status = 0;
 451         while (is && !status) {
 452                 unsigned char c = 0;
 453                 char cc = 0;
 454                 is.get(cc);
 455                 c = cc;
 456                 if (c >= ' ' && is) {
 457                         int i = 0;
 458                         if (c == '\\') { // first char == '\\'
 459                                 do {
 460                                         buff[i++] = c;
 461                                         is.get(cc);
 462                                         c = cc;
 463                                 } while (c > ' ' && c != '\\' && is
 464                                          && i != (LEX_MAX_BUFF-1));
 465                         } else {
 466                                 do {
 467                                         buff[i++] = c;
 468                                         is.get(cc);
 469                                         c = cc;
 470                                 } while (c >= ' ' && c != '\\' && is
 471                                          && i != (LEX_MAX_BUFF-1));
 472                         }
 473
 474                         if (i == (LEX_MAX_BUFF - 1)) {
 475                                 printError("Line too long");
 476                         }
 477
 478                         if (c == '\\') is.putback(c); // put it back
 479                         buff[i] = '\0';
 480                         status = LEX_TOKEN;
 481                 }
 482
 483                 if (c == '\n')
 484                         ++lineno;
 485
 486         }
 487         if (status)  return true;
 488
 489         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 490         buff[0] = '\0';
 491         return false;
 492 }
 493
 494
 495 void LyXLex::Pimpl::pushToken(string const & pt)
 496 {
 497         pushTok = pt;
 498 }