src/lyxlex_pimpl.C

   1 #include <config.h>
   2
   3 #include "lyxlex_pimpl.h"
   4 #include "debug.h"
   5
   6 #include "support/lyxalgo.h"
   7 #include "support/filetools.h"
   8 #include "support/lstrings.h"
   9
  10 #include <algorithm>
  11
  12 using std::sort;
  13 using std::ostream;
  14 using std::ios;
  15 using std::istream;
  16 using std::endl;
  17 using std::lower_bound;
  18 using std::vector;
  19 using std::getline;
  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.begin(), buff.end());
  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
 138 void LyXLex::Pimpl::setCommentChar(char c)
 139 {
 140         commentChar = c;
 141 }
 142
 143
 144 bool LyXLex::Pimpl::next(bool esc /* = false */)
 145 {
 146         if (!pushTok.empty()) {
 147                 // There can have been a whole line pushed so
 148                 // we extract the first word and leaves the rest
 149                 // in pushTok. (Lgb)
 150                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 151                         string tmp;
 152                         pushTok = split(pushTok, tmp, ' ');
 153                         buff.assign(tmp.begin(), tmp.end());
 154                         return true;
 155                 } else {
 156                         buff.assign(pushTok.begin(), pushTok.end());
 157                         pushTok.erase();
 158                         return true;
 159                 }
 160         }
 161         if (!esc) {
 162                 unsigned char c = 0; // getc() returns an int
 163                 char cc = 0;
 164                 status = 0;
 165                 while (is && !status) {
 166                         is.get(cc);
 167                         c = cc;
 168                         if (c == commentChar) {
 169                                 // Read rest of line (fast :-)
 170 #if 1
 171                                 // That is not fast... (Lgb)
 172                                 string dummy;
 173                                 getline(is, dummy);
 174
 175                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 176                                                       << dummy << '\'' << endl;
 177 #else
 178                                 // unfortunately ignore is buggy (Lgb)
 179                                 is.ignore(100, '\n');
 180 #endif
 181                                 ++lineno;
 182                                 continue;
 183                         }
 184
 185                         if (c == '\"') {
 186                                 buff.clear();
 187
 188                                 do {
 189                                         is.get(cc);
 190                                         c = cc;
 191                                         if (c != '\r')
 192                                                 buff.push_back(c);
 193                                 } while (c != '\"' && c != '\n' && is);
 194
 195                                 if (c != '\"') {
 196                                         printError("Missing quote");
 197                                         if (c == '\n')
 198                                                 ++lineno;
 199                                 }
 200
 201                                 buff.pop_back();
 202                                 status = LEX_DATA;
 203                                 break;
 204                         }
 205
 206                         if (c == ',')
 207                                 continue;              /* Skip ','s */
 208
 209                                 // using relational operators with chars other
 210                                 // than == and != is not safe. And if it is done
 211                                 // the type _have_ to be unsigned. It usually a
 212                                 // lot better to use the functions from cctype
 213                         if (c > ' ' && is)  {
 214                                 buff.clear();
 215
 216                                 do {
 217                                         buff.push_back(c);
 218                                         is.get(cc);
 219                                         c = cc;
 220                                 } while (c > ' ' && c != ',' && is);
 221
 222                                 status = LEX_TOKEN;
 223                         }
 224
 225                         if (c == '\r' && is) {
 226                                 // The Windows support has lead to the
 227                                 // possibility of "\r\n" at the end of
 228                                 // a line.  This will stop LyX choking
 229                                 // when it expected to find a '\n'
 230                                 is.get(cc);
 231                                 c = cc;
 232                         }
 233
 234                         if (c == '\n')
 235                                 ++lineno;
 236
 237                 }
 238                 if (status) return true;
 239
 240                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 241                 buff.clear();
 242                 return false;
 243         } else {
 244                 unsigned char c = 0; // getc() returns an int
 245                 char cc = 0;
 246
 247                 status = 0;
 248                 while (is && !status) {
 249                         is.get(cc);
 250                         c = cc;
 251
 252                         // skip ','s
 253                         if (c == ',') continue;
 254
 255                         if (c == '\\') {
 256                                 // escape
 257                                 buff.clear();
 258
 259                                 do {
 260                                         if (c == '\\') {
 261                                                 // escape the next char
 262                                                 is.get(cc);
 263                                                 c = cc;
 264                                         }
 265                                         buff.push_back(c);
 266                                         is.get(cc);
 267                                         c = cc;
 268                                 } while (c > ' ' && c != ',' && is);
 269
 270                                 status = LEX_TOKEN;
 271                                 continue;
 272                         }
 273
 274                         if (c == commentChar) {
 275                                 // Read rest of line (fast :-)
 276 #if 1
 277                                 // That is still not fast... (Lgb)
 278                                 string dummy;
 279                                 getline(is, dummy);
 280
 281                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 282                                                       << dummy << '\'' << endl;
 283 #else
 284                                 // but ignore is also still buggy (Lgb)
 285                                 // This is fast (Lgb)
 286                                 is.ignore(100, '\n');
 287 #endif
 288                                 ++lineno;
 289                                 continue;
 290                         }
 291
 292                         // string
 293                         if (c == '\"') {
 294                                 buff.clear();
 295
 296                                 bool escaped = false;
 297                                 do {
 298                                         escaped = false;
 299                                         is.get(cc);
 300                                         c = cc;
 301                                         if (c == '\r') continue;
 302                                         if (c == '\\') {
 303                                                 // escape the next char
 304                                                 is.get(cc);
 305                                                 c = cc;
 306                                                 if (c == '\"' || c == '\\')
 307                                                         escaped = true;
 308                                                 else
 309                                                         buff.push_back('\\');
 310                                         }
 311                                         buff.push_back(c);
 312
 313                                         if (!escaped && c == '\"') break;
 314                                 } while (c != '\n' && is);
 315
 316                                 if (c != '\"') {
 317                                         printError("Missing quote");
 318                                         if (c == '\n')
 319                                                 ++lineno;
 320                                 }
 321
 322                                 buff.pop_back();
 323                                 status = LEX_DATA;
 324                                 break;
 325                         }
 326
 327                         if (c > ' ' && is) {
 328                                 buff.clear();
 329
 330                                 do {
 331                                         if (c == '\\') {
 332                                                 // escape the next char
 333                                                 is.get(cc);
 334                                                 c = cc;
 335                                                 //escaped = true;
 336                                         }
 337                                         buff.push_back(c);
 338                                         is.get(cc);
 339                                         c = cc;
 340                                 } while (c > ' ' && c != ',' && is);
 341
 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.clear();
 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         // use the compare_ascii_no_case instead of compare_no_case,
 365         // because in turkish, 'i' is not the lowercase version of 'I',
 366         // and thus turkish locale breaks parsing of tags.
 367         if (res != table + no_items
 368             && !compare_ascii_no_case(res->tag, tag))
 369                 return res->code;
 370         return LEX_UNDEF;
 371 }
 372
 373
 374 int LyXLex::Pimpl::lex()
 375 {
 376         //NOTE: possible bug.
 377         if (next() && status == LEX_TOKEN) {
 378                 return search_kw(getString().c_str());
 379         } else
 380                 return status;
 381 }
 382
 383
 384 bool LyXLex::Pimpl::eatLine()
 385 {
 386         buff.clear();
 387
 388         unsigned char c = '\0';
 389         char cc = 0;
 390         while (is && c != '\n') {
 391                 is.get(cc);
 392                 c = cc;
 393                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 394                 //                    << c << '\'' << endl;
 395                 if (c != '\r')
 396                         buff.push_back(c);
 397         }
 398
 399         if (c == '\n') {
 400                 ++lineno;
 401                 buff.pop_back();
 402                 status = LEX_DATA;
 403                 return true;
 404         } else {
 405                 return false;
 406         }
 407 }
 408
 409
 410 bool LyXLex::Pimpl::nextToken()
 411 {
 412         if (!pushTok.empty()) {
 413                 // There can have been a whole line pushed so
 414                 // we extract the first word and leaves the rest
 415                 // in pushTok. (Lgb)
 416                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 417                         string tmp;
 418                         pushTok = split(pushTok, tmp, ' ');
 419                         buff.assign(tmp.begin(), tmp.end());
 420                         return true;
 421                 } else {
 422                         buff.assign(pushTok.begin(), pushTok.end());
 423                         pushTok.erase();
 424                         return true;
 425                 }
 426         }
 427
 428         status = 0;
 429         while (is && !status) {
 430                 unsigned char c = 0;
 431                 char cc = 0;
 432                 is.get(cc);
 433                 c = cc;
 434                 if (c >= ' ' && is) {
 435                         buff.clear();
 436
 437                         if (c == '\\') { // first char == '\\'
 438                                 do {
 439                                         buff.push_back(c);
 440                                         is.get(cc);
 441                                         c = cc;
 442                                 } while (c > ' ' && c != '\\' && is);
 443                         } else {
 444                                 do {
 445                                         buff.push_back(c);
 446                                         is.get(cc);
 447                                         c = cc;
 448                                 } while (c >= ' ' && c != '\\' && is);
 449                         }
 450
 451                         if (c == '\\') is.putback(c); // put it back
 452                         status = LEX_TOKEN;
 453                 }
 454
 455                 if (c == '\n')
 456                         ++lineno;
 457
 458         }
 459         if (status)  return true;
 460
 461         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 462         buff.clear();
 463         return false;
 464 }
 465
 466
 467 void LyXLex::Pimpl::pushToken(string const & pt)
 468 {
 469         pushTok = pt;
 470 }