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