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), commentChar('#')
  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 void LyXLex::Pimpl::setCommentChar(char c)
 134 {
 135         commentChar = c;
 136 }
 137
 138
 139 bool LyXLex::Pimpl::next(bool esc /* = false */)
 140 {
 141         if (!pushTok.empty()) {
 142                 pushTok.copy(buff, string::npos);
 143                 buff[pushTok.length()] = '\0';
 144                 pushTok.erase();
 145                 return true;
 146         }
 147         if (!esc) {
 148                 unsigned char c = 0; // getc() returns an int
 149                 char cc = 0;
 150                 status = 0;
 151                 while (is && !status) {
 152                         is.get(cc);
 153                         c = cc;
 154                         if (c == commentChar) {
 155                                 // Read rest of line (fast :-)
 156                                 // That is not fast... (Lgb)
 157 #if 1
 158                                 is.getline(buff, sizeof(buff));
 159                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 160                                                       << buff << "'" << endl;
 161 #else
 162                                 // unfortunately is ignore buggy (Lgb)
 163                                 is.ignore(100, '\n');
 164 #endif
 165                                 ++lineno;
 166                                 continue;
 167                         }
 168
 169                         if (c == '\"') {
 170                                 int i = -1;
 171                                 do {
 172                                         is.get(cc);
 173                                         c = cc;
 174                                         if (c != '\r')
 175                                                 buff[++i] = c;
 176                                 } while (c != '\"' && c != '\n' && is &&
 177                                          i != (LEX_MAX_BUFF - 2));
 178
 179                                 if (i == (LEX_MAX_BUFF - 2)) {
 180                                         printError("Line too long");
 181                                         c = '\"'; // Pretend we got a "
 182                                         ++i;
 183                                 }
 184
 185                                 if (c != '\"') {
 186                                         printError("Missing quote");
 187                                         if (c == '\n')
 188                                                 ++lineno;
 189                                 }
 190
 191                                 buff[i] = '\0';
 192                                 status = LEX_DATA;
 193                                 break;
 194                         }
 195
 196                         if (c == ',')
 197                                 continue;              /* Skip ','s */
 198
 199                                 // using relational operators with chars other
 200                                 // than == and != is not safe. And if it is done
 201                                 // the type _have_ to be unsigned. It usually a
 202                                 // lot better to use the functions from cctype
 203                         if (c > ' ' && is)  {
 204                                 int i = 0;
 205                                 do {
 206                                         buff[i++] = c;
 207                                         is.get(cc);
 208                                         c = cc;
 209                                 } while (c > ' ' && c != ',' && is
 210                                          && (i != LEX_MAX_BUFF - 1) );
 211                                 if (i == LEX_MAX_BUFF - 1) {
 212                                         printError("Line too long");
 213                                 }
 214                                 buff[i] = '\0';
 215                                 status = LEX_TOKEN;
 216                         }
 217
 218                         if (c == '\r' && is) {
 219                                 // The Windows support has lead to the
 220                                 // possibility of "\r\n" at the end of
 221                                 // a line.  This will stop LyX choking
 222                                 // when it expected to find a '\n'
 223                                 is.get(cc);
 224                                 c = cc;
 225                         }
 226
 227                         if (c == '\n')
 228                                 ++lineno;
 229
 230                 }
 231                 if (status) return true;
 232
 233                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 234                 buff[0] = '\0';
 235                 return false;
 236         } else {
 237                 unsigned char c = 0; // getc() returns an int
 238                 char cc = 0;
 239
 240                 status = 0;
 241                 while (is && !status) {
 242                         is.get(cc);
 243                         c = cc;
 244
 245                         // skip ','s
 246                         if (c == ',') continue;
 247
 248                         if (c == '\\') {
 249                                 // escape
 250                                 int i = 0;
 251                                 do {
 252                                         if (c == '\\') {
 253                                                 // escape the next char
 254                                                 is.get(cc);
 255                                                 c = cc;
 256                                         }
 257                                         buff[i++] = c;
 258                                         is.get(cc);
 259                                         c = cc;
 260                                 } while (c > ' ' && c != ',' && is
 261                                          && (i != LEX_MAX_BUFF - 1) );
 262                                 if (i == LEX_MAX_BUFF - 1) {
 263                                         printError("Line too long");
 264                                 }
 265                                 buff[i] = '\0';
 266                                 status = LEX_TOKEN;
 267                                 continue;
 268                         }
 269
 270                         if (c == commentChar) {
 271                                 // Read rest of line (fast :-)
 272                                 // That is still not fast... (Lgb)
 273 #if 1
 274                                 is.getline(buff, sizeof(buff));
 275                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 276                                                       << buff << "'" << endl;
 277 #else
 278                                 // but ignore is also still buggy (Lgb)
 279                                 // This is fast (Lgb)
 280                                 is.ignore(100, '\n');
 281 #endif
 282                                 ++lineno;
 283                                 continue;
 284                         }
 285
 286                         // string
 287                         if (c == '\"') {
 288                                 int i = -1;
 289                                 bool escaped = false;
 290                                 do {
 291                                         escaped = false;
 292                                         is.get(cc);
 293                                         c = cc;
 294                                         if (c == '\r') continue;
 295                                         if (c == '\\') {
 296                                                 // escape the next char
 297                                                 is.get(cc);
 298                                                 c = cc;
 299                                                 escaped = true;
 300                                         }
 301                                         buff[++i] = c;
 302
 303                                         if (!escaped && c == '\"') break;
 304                                 } while (c != '\n' && is &&
 305                                          i != (LEX_MAX_BUFF - 2));
 306
 307                                 if (i == (LEX_MAX_BUFF - 2)) {
 308                                         printError("Line too long");
 309                                         c = '\"'; // Pretend we got a "
 310                                         ++i;
 311                                 }
 312
 313                                 if (c != '\"') {
 314                                         printError("Missing quote");
 315                                         if (c == '\n')
 316                                                 ++lineno;
 317                                 }
 318
 319                                 buff[i] = '\0';
 320                                 status = LEX_DATA;
 321                                 break;
 322                         }
 323
 324                         if (c > ' ' && is) {
 325                                 int i = 0;
 326                                 do {
 327                                         if (c == '\\') {
 328                                                 // escape the next char
 329                                                 is.get(cc);
 330                                                 c = cc;
 331                                                 //escaped = true;
 332                                         }
 333                                         buff[i++] = c;
 334                                         is.get(cc);
 335                                         c = cc;
 336                                 } while (c > ' ' && c != ',' && is
 337                                          && (i != LEX_MAX_BUFF-1) );
 338                                 if (i == LEX_MAX_BUFF-1) {
 339                                         printError("Line too long");
 340                                 }
 341                                 buff[i] = '\0';
 342                                 status = LEX_TOKEN;
 343                         }
 344                         // new line
 345                         if (c == '\n')
 346                                 ++lineno;
 347                 }
 348
 349                 if (status) return true;
 350
 351                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 352                 buff[0] = '\0';
 353                 return false;
 354         }
 355 }
 356
 357
 358 int LyXLex::Pimpl::search_kw(char const * const tag) const
 359 {
 360         keyword_item search_tag = { tag, 0 };
 361         keyword_item * res =
 362                 lower_bound(table, table + no_items,
 363                             search_tag, compare_tags());
 364         if (res != table + no_items
 365             && !compare_no_case(res->tag, tag))
 366                 return res->code;
 367         return LEX_UNDEF;
 368 }
 369
 370
 371 int LyXLex::Pimpl::lex()
 372 {
 373         //NOTE: possible bug.
 374         if (next() && status == LEX_TOKEN)
 375                 return search_kw(buff);
 376         else
 377                 return status;
 378 }
 379
 380
 381 bool LyXLex::Pimpl::EatLine()
 382 {
 383         int i = 0;
 384         unsigned char c = '\0';
 385         char cc = 0;
 386         while(is && c != '\n' && i != (LEX_MAX_BUFF - 1)) {
 387                 is.get(cc);
 388                 c = cc;
 389                 lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 390                                       << c << "'" << endl;
 391                 if (c != '\r')
 392                         buff[i++] = c;
 393         }
 394         if (i == (LEX_MAX_BUFF - 1) && c != '\n') {
 395                 printError("Line too long");
 396                 c = '\n'; // Pretend we had an end of line
 397                 --lineno; // but don't increase line counter (netto effect)
 398                 ++i; // and preserve last character read.
 399         }
 400         if (c == '\n') {
 401                 ++lineno;
 402                 buff[--i] = '\0'; // i can never be 0 here, so no danger
 403                 status = LEX_DATA;
 404                 return true;
 405         } else {
 406                 buff[i] = '\0';
 407                 return false;
 408         }
 409 }
 410
 411
 412 bool LyXLex::Pimpl::nextToken()
 413 {
 414         if (!pushTok.empty()) {
 415                 pushTok.copy(buff, string::npos);
 416                 buff[pushTok.length()] = '\0';
 417                 pushTok.erase();
 418                 return true;
 419         }
 420
 421         status = 0;
 422         while (is && !status) {
 423                 unsigned char c = 0;
 424                 char cc = 0;
 425                 is.get(cc);
 426                 c = cc;
 427                 if (c >= ' ' && is) {
 428                         int i = 0;
 429                         if (c == '\\') { // first char == '\\'
 430                                 do {
 431                                         buff[i++] = c;
 432                                         is.get(cc);
 433                                         c = cc;
 434                                 } while (c > ' ' && c != '\\' && is
 435                                          && i != (LEX_MAX_BUFF-1));
 436                         } else {
 437                                 do {
 438                                         buff[i++] = c;
 439                                         is.get(cc);
 440                                         c = cc;
 441                                 } while (c >= ' ' && c != '\\' && is
 442                                          && i != (LEX_MAX_BUFF-1));
 443                         }
 444
 445                         if (i == (LEX_MAX_BUFF - 1)) {
 446                                 printError("Line too long");
 447                         }
 448
 449                         if (c == '\\') is.putback(c); // put it back
 450                         buff[i] = '\0';
 451                         status = LEX_TOKEN;
 452                 }
 453
 454                 if (c == '\n')
 455                         ++lineno;
 456
 457         }
 458         if (status)  return true;
 459
 460         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 461         buff[0] = '\0';
 462         return false;
 463 }
 464
 465
 466 void LyXLex::Pimpl::pushToken(string const & pt)
 467 {
 468         pushTok = pt;
 469 }