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