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         if (fb__.is_open() || is.tellg() > 0)
 111                 lyxerr << "Error in LyXLex::setFile: "
 112                         "file or stream already set." << endl;
 113         fb__.open(filename.c_str(), ios::in);
 114         is.rdbuf(&fb__);
 115         name = filename;
 116         lineno = 0;
 117         return fb__.is_open() && is.good();
 118 }
 119
 120
 121 void LyXLex::Pimpl::setStream(istream & i)
 122 {
 123         if (fb__.is_open() || is.tellg() > 0)
 124                 lyxerr << "Error in LyXLex::setStream: "
 125                         "file or stream already set." << endl;
 126         is.rdbuf(i.rdbuf());
 127         lineno = 0;
 128 }
 129
 130
 131 bool LyXLex::Pimpl::next(bool esc /* = false */)
 132 {
 133         if (!pushTok.empty()) {
 134                 pushTok.copy(buff, string::npos);
 135                 buff[pushTok.length()] = '\0';
 136                 pushTok.erase();
 137                 return true;
 138         }
 139         if (!esc) {
 140                 unsigned char c = 0; // getc() returns an int
 141                 char cc = 0;
 142                 status = 0;
 143                 while (is && !status) {
 144                         is.get(cc);
 145                         c = cc;
 146                         if (c == '#') {
 147                                 // Read rest of line (fast :-)
 148                                 is.getline(buff, sizeof(buff));
 149                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 150                                                       << buff << "'" << endl;
 151                                 ++lineno;
 152                                 continue;
 153                         }
 154
 155                         if (c == '\"') {
 156                                 int i = -1;
 157                                 do {
 158                                         is.get(cc);
 159                                         c = cc;
 160                                         if (c != '\r')
 161                                                 buff[++i] = c;
 162                                 } while (c != '\"' && c != '\n' && is &&
 163                                          i != (LEX_MAX_BUFF - 2));
 164
 165                                 if (i == (LEX_MAX_BUFF - 2)) {
 166                                         printError("Line too long");
 167                                         c = '\"'; // Pretend we got a "
 168                                         ++i;
 169                                 }
 170
 171                                 if (c != '\"') {
 172                                         printError("Missing quote");
 173                                         if (c == '\n')
 174                                                 ++lineno;
 175                                 }
 176
 177                                 buff[i] = '\0';
 178                                 status = LEX_DATA;
 179                                 break;
 180                         }
 181
 182                         if (c == ',')
 183                                 continue;              /* Skip ','s */
 184
 185                                 // using relational operators with chars other
 186                                 // than == and != is not safe. And if it is done
 187                                 // the type _have_ to be unsigned. It usually a
 188                                 // lot better to use the functions from cctype
 189                         if (c > ' ' && is)  {
 190                                 int i = 0;
 191                                 do {
 192                                         buff[i++] = c;
 193                                         is.get(cc);
 194                                         c = cc;
 195                                 } while (c > ' ' && c != ',' && is
 196                                          && (i != LEX_MAX_BUFF - 1) );
 197                                 if (i == LEX_MAX_BUFF - 1) {
 198                                         printError("Line too long");
 199                                 }
 200                                 buff[i] = '\0';
 201                                 status = LEX_TOKEN;
 202                         }
 203
 204                         if (c == '\r' && is) {
 205                                 // The Windows support has lead to the
 206                                 // possibility of "\r\n" at the end of
 207                                 // a line.  This will stop LyX choking
 208                                 // when it expected to find a '\n'
 209                                 is.get(cc);
 210                                 c = cc;
 211                         }
 212
 213                         if (c == '\n')
 214                                 ++lineno;
 215
 216                 }
 217                 if (status) return true;
 218
 219                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 220                 buff[0] = '\0';
 221                 return false;
 222         } else {
 223                 unsigned char c = 0; // getc() returns an int
 224                 char cc = 0;
 225
 226                 status = 0;
 227                 while (is && !status) {
 228                         is.get(cc);
 229                         c = cc;
 230
 231                         // skip ','s
 232                         if (c == ',') continue;
 233
 234                         if (c == '\\') {
 235                                 // escape
 236                                 int i = 0;
 237                                 do {
 238                                         if (c == '\\') {
 239                                                 // escape the next char
 240                                                 is.get(cc);
 241                                                 c = cc;
 242                                         }
 243                                         buff[i++] = c;
 244                                         is.get(cc);
 245                                         c = cc;
 246                                 } while (c > ' ' && c != ',' && is
 247                                          && (i != LEX_MAX_BUFF - 1) );
 248                                 if (i == LEX_MAX_BUFF - 1) {
 249                                         printError("Line too long");
 250                                 }
 251                                 buff[i] = '\0';
 252                                 status = LEX_TOKEN;
 253                                 continue;
 254                         }
 255
 256                         if (c == '#') {
 257                                 // Read rest of line (fast :-)
 258                                 is.getline(buff, sizeof(buff));
 259                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 260                                                       << buff << "'" << endl;
 261                                 ++lineno;
 262                                 continue;
 263                         }
 264
 265                         // string
 266                         if (c == '\"') {
 267                                 int i = -1;
 268                                 bool escaped = false;
 269                                 do {
 270                                         escaped = false;
 271                                         is.get(cc);
 272                                         c = cc;
 273                                         if (c == '\r') continue;
 274                                         if (c == '\\') {
 275                                                 // escape the next char
 276                                                 is.get(cc);
 277                                                 c = cc;
 278                                                 escaped = true;
 279                                         }
 280                                         buff[++i] = c;
 281
 282                                         if (!escaped && c == '\"') break;
 283                                 } while (c != '\n' && is &&
 284                                          i != (LEX_MAX_BUFF - 2));
 285
 286                                 if (i == (LEX_MAX_BUFF - 2)) {
 287                                         printError("Line too long");
 288                                         c = '\"'; // Pretend we got a "
 289                                         ++i;
 290                                 }
 291
 292                                 if (c != '\"') {
 293                                         printError("Missing quote");
 294                                         if (c == '\n')
 295                                                 ++lineno;
 296                                 }
 297
 298                                 buff[i] = '\0';
 299                                 status = LEX_DATA;
 300                                 break;
 301                         }
 302
 303                         if (c > ' ' && is) {
 304                                 int i = 0;
 305                                 do {
 306                                         if (c == '\\') {
 307                                                 // escape the next char
 308                                                 is.get(cc);
 309                                                 c = cc;
 310                                                 //escaped = true;
 311                                         }
 312                                         buff[i++] = c;
 313                                         is.get(cc);
 314                                         c = cc;
 315                                 } while (c > ' ' && c != ',' && is
 316                                          && (i != LEX_MAX_BUFF-1) );
 317                                 if (i == LEX_MAX_BUFF-1) {
 318                                         printError("Line too long");
 319                                 }
 320                                 buff[i] = '\0';
 321                                 status = LEX_TOKEN;
 322                         }
 323                         // new line
 324                         if (c == '\n')
 325                                 ++lineno;
 326                 }
 327
 328                 if (status) return true;
 329
 330                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 331                 buff[0] = '\0';
 332                 return false;
 333         }
 334 }
 335
 336
 337 int LyXLex::Pimpl::search_kw(char const * const tag) const
 338 {
 339         keyword_item search_tag = { tag, 0 };
 340         keyword_item * res =
 341                 lower_bound(table, table + no_items,
 342                             search_tag, compare_tags());
 343         if (res != table + no_items
 344             && !compare_no_case(res->tag, tag))
 345                 return res->code;
 346         return LEX_UNDEF;
 347 }
 348
 349
 350 int LyXLex::Pimpl::lex()
 351 {
 352         //NOTE: possible bug.
 353         if (next() && status == LEX_TOKEN)
 354                 return search_kw(buff);
 355         else
 356                 return status;
 357 }
 358
 359
 360 bool LyXLex::Pimpl::EatLine()
 361 {
 362         int i = 0;
 363         unsigned char c = '\0';
 364         char cc = 0;
 365         while(is && c != '\n' && i != (LEX_MAX_BUFF - 1)) {
 366                 is.get(cc);
 367                 c = cc;
 368                 lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 369                                       << c << "'" << endl;
 370                 if (c != '\r')
 371                         buff[i++] = c;
 372         }
 373         if (i == (LEX_MAX_BUFF - 1) && c != '\n') {
 374                 printError("Line too long");
 375                 c = '\n'; // Pretend we had an end of line
 376                 --lineno; // but don't increase line counter (netto effect)
 377                 ++i; // and preserve last character read.
 378         }
 379         if (c == '\n') {
 380                 ++lineno;
 381                 buff[--i] = '\0'; // i can never be 0 here, so no danger
 382                 status = LEX_DATA;
 383                 return true;
 384         } else {
 385                 buff[i] = '\0';
 386                 return false;
 387         }
 388 }
 389
 390
 391 bool LyXLex::Pimpl::nextToken()
 392 {
 393         if (!pushTok.empty()) {
 394                 pushTok.copy(buff, string::npos);
 395                 buff[pushTok.length()] = '\0';
 396                 pushTok.erase();
 397                 return true;
 398         }
 399
 400         status = 0;
 401         while (is && !status) {
 402                 unsigned char c = 0;
 403                 char cc = 0;
 404                 is.get(cc);
 405                 c = cc;
 406                 if (c >= ' ' && is) {
 407                         int i = 0;
 408                         if (c == '\\') { // first char == '\\'
 409                                 do {
 410                                         buff[i++] = c;
 411                                         is.get(cc);
 412                                         c = cc;
 413                                 } while (c > ' ' && c != '\\' && is
 414                                          && i != (LEX_MAX_BUFF-1));
 415                         } else {
 416                                 do {
 417                                         buff[i++] = c;
 418                                         is.get(cc);
 419                                         c = cc;
 420                                 } while (c >= ' ' && c != '\\' && is
 421                                          && i != (LEX_MAX_BUFF-1));
 422                         }
 423
 424                         if (i == (LEX_MAX_BUFF - 1)) {
 425                                 printError("Line too long");
 426                         }
 427
 428                         if (c == '\\') is.putback(c); // put it back
 429                         buff[i] = '\0';
 430                         status = LEX_TOKEN;
 431                 }
 432
 433                 if (c == '\n')
 434                         ++lineno;
 435
 436         }
 437         if (status)  return true;
 438
 439         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 440         buff[0] = '\0';
 441         return false;
 442 }
 443
 444
 445 void LyXLex::Pimpl::pushToken(string const & pt)
 446 {
 447         pushTok = pt;
 448 }