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 namespace lyx::support;
  13
  14 using std::sort;
  15 using std::ostream;
  16 using std::ios;
  17 using std::istream;
  18 using std::endl;
  19 using std::lower_bound;
  20 using std::vector;
  21 using std::getline;
  22
  23 // namespace {
  24 struct compare_tags {
  25         // used by lower_bound, sort and sorted
  26         inline
  27         int operator()(keyword_item const & a, keyword_item const & b) const {
  28                 // we use the ascii version, because in turkish, 'i'
  29                 // is not the lowercase version of 'I', and thus
  30                 // turkish locale breaks parsing of tags.
  31                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  32         }
  33 };
  34 // } // end of anon namespace
  35
  36
  37 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  38         : is(&fb__), table(tab), no_items(num),
  39           status(0), lineno(0), commentChar('#')
  40 {
  41         verifyTable();
  42 }
  43
  44
  45 string const LyXLex::Pimpl::getString() const
  46 {
  47         return string(buff.begin(), buff.end());
  48 }
  49
  50
  51 void LyXLex::Pimpl::printError(string const & message) const
  52 {
  53         string const tmpmsg = subst(message, "$$Token", getString());
  54         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  55                << " of file " << MakeDisplayPath(name) << ']' << endl;
  56 }
  57
  58
  59 void LyXLex::Pimpl::printTable(ostream & os)
  60 {
  61         os << "\nNumber of tags: " << no_items << '\n';
  62         for (int i= 0; i < no_items; ++i)
  63                 os << "table[" << i
  64                    << "]:  tag: `" << table[i].tag
  65                    << "'  code:" << table[i].code << '\n';
  66         os.flush();
  67 }
  68
  69
  70 void LyXLex::Pimpl::verifyTable()
  71 {
  72         // Check if the table is sorted and if not, sort it.
  73         if (table
  74             && !lyx::sorted(table, table + no_items, compare_tags())) {
  75                 lyxerr << "The table passed to LyXLex is not sorted!\n"
  76                        << "Tell the developers to fix it!" << endl;
  77                 // We sort it anyway to avoid problems.
  78                 lyxerr << "\nUnsorted:\n";
  79                 printTable(lyxerr);
  80
  81                 sort(table, table + no_items, compare_tags());
  82                 lyxerr << "\nSorted:\n";
  83                 printTable(lyxerr);
  84         }
  85 }
  86
  87
  88 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
  89 {
  90         pushed_table tmppu(table, no_items);
  91         pushed.push(tmppu);
  92
  93         table = tab;
  94         no_items = num;
  95
  96         verifyTable();
  97 }
  98
  99
 100 void LyXLex::Pimpl::popTable()
 101 {
 102         if (pushed.empty()) {
 103                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 104                 return;
 105         }
 106
 107         pushed_table tmp = pushed.top();
 108         pushed.pop();
 109         table = tmp.table_elem;
 110         no_items = tmp.table_siz;
 111 }
 112
 113
 114 bool LyXLex::Pimpl::setFile(string const & filename)
 115 {
 116
 117         // Check the format of the file.
 118         string const format = getExtFromContents(filename);
 119
 120         if (format == "gzip" || format == "zip" || format == "compress") {
 121                 lyxerr << "lyxlex: compressed" << endl;
 122
 123                 // The check only outputs a debug message, because it triggers
 124                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 125                 // a fresh new filebuf.  (JMarc)
 126                 if (gz__.is_open() || is.tellg() > 0)
 127                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 128                                 "file or stream already set." << endl;
 129                 gz__.open(filename.c_str(), ios::in);
 130                 is.rdbuf(&gz__);
 131                 name = filename;
 132                 lineno = 0;
 133                 return gz__.is_open() && is.good();
 134         } else {
 135                 lyxerr << "lyxlex: UNcompressed" << endl;
 136
 137                 // The check only outputs a debug message, because it triggers
 138                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 139                 // a fresh new filebuf.  (JMarc)
 140                 if (fb__.is_open() || is.tellg() > 0)
 141                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 142                                 "file or stream already set." << endl;
 143                 fb__.open(filename.c_str(), ios::in);
 144                 is.rdbuf(&fb__);
 145                 name = filename;
 146                 lineno = 0;
 147                 return fb__.is_open() && is.good();
 148         }
 149 }
 150
 151
 152 void LyXLex::Pimpl::setStream(istream & i)
 153 {
 154         if (fb__.is_open() || is.tellg() > 0)
 155                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 156                         "file or stream already set." << endl;
 157         is.rdbuf(i.rdbuf());
 158         lineno = 0;
 159 }
 160
 161
 162 void LyXLex::Pimpl::setCommentChar(char c)
 163 {
 164         commentChar = c;
 165 }
 166
 167
 168 bool LyXLex::Pimpl::next(bool esc /* = false */)
 169 {
 170         if (!pushTok.empty()) {
 171                 // There can have been a whole line pushed so
 172                 // we extract the first word and leaves the rest
 173                 // in pushTok. (Lgb)
 174                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 175                         string tmp;
 176                         pushTok = split(pushTok, tmp, ' ');
 177                         buff.assign(tmp.begin(), tmp.end());
 178                         return true;
 179                 } else {
 180                         buff.assign(pushTok.begin(), pushTok.end());
 181                         pushTok.erase();
 182                         return true;
 183                 }
 184         }
 185         if (!esc) {
 186                 unsigned char c = 0; // getc() returns an int
 187                 char cc = 0;
 188                 status = 0;
 189                 while (is && !status) {
 190                         is.get(cc);
 191                         c = cc;
 192                         if (c == commentChar) {
 193                                 // Read rest of line (fast :-)
 194 #if 1
 195                                 // That is not fast... (Lgb)
 196                                 string dummy;
 197                                 getline(is, dummy);
 198
 199                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 200                                                       << dummy << '\'' << endl;
 201 #else
 202                                 // unfortunately ignore is buggy (Lgb)
 203                                 is.ignore(100, '\n');
 204 #endif
 205                                 ++lineno;
 206                                 continue;
 207                         }
 208
 209                         if (c == '\"') {
 210                                 buff.clear();
 211
 212                                 do {
 213                                         is.get(cc);
 214                                         c = cc;
 215                                         if (c != '\r')
 216                                                 buff.push_back(c);
 217                                 } while (c != '\"' && c != '\n' && is);
 218
 219                                 if (c != '\"') {
 220                                         printError("Missing quote");
 221                                         if (c == '\n')
 222                                                 ++lineno;
 223                                 }
 224
 225                                 buff.pop_back();
 226                                 status = LEX_DATA;
 227                                 break;
 228                         }
 229
 230                         if (c == ',')
 231                                 continue;              /* Skip ','s */
 232
 233                                 // using relational operators with chars other
 234                                 // than == and != is not safe. And if it is done
 235                                 // the type _have_ to be unsigned. It usually a
 236                                 // lot better to use the functions from cctype
 237                         if (c > ' ' && is)  {
 238                                 buff.clear();
 239
 240                                 do {
 241                                         buff.push_back(c);
 242                                         is.get(cc);
 243                                         c = cc;
 244                                 } while (c > ' ' && c != ',' && is);
 245
 246                                 status = LEX_TOKEN;
 247                         }
 248
 249                         if (c == '\r' && is) {
 250                                 // The Windows support has lead to the
 251                                 // possibility of "\r\n" at the end of
 252                                 // a line.  This will stop LyX choking
 253                                 // when it expected to find a '\n'
 254                                 is.get(cc);
 255                                 c = cc;
 256                         }
 257
 258                         if (c == '\n')
 259                                 ++lineno;
 260
 261                 }
 262                 if (status) return true;
 263
 264                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 265                 buff.clear();
 266                 return false;
 267         } else {
 268                 unsigned char c = 0; // getc() returns an int
 269                 char cc = 0;
 270
 271                 status = 0;
 272                 while (is && !status) {
 273                         is.get(cc);
 274                         c = cc;
 275
 276                         // skip ','s
 277                         if (c == ',') continue;
 278
 279                         if (c == '\\') {
 280                                 // escape
 281                                 buff.clear();
 282
 283                                 do {
 284                                         if (c == '\\') {
 285                                                 // escape the next char
 286                                                 is.get(cc);
 287                                                 c = cc;
 288                                         }
 289                                         buff.push_back(c);
 290                                         is.get(cc);
 291                                         c = cc;
 292                                 } while (c > ' ' && c != ',' && is);
 293
 294                                 status = LEX_TOKEN;
 295                                 continue;
 296                         }
 297
 298                         if (c == commentChar) {
 299                                 // Read rest of line (fast :-)
 300 #if 1
 301                                 // That is still not fast... (Lgb)
 302                                 string dummy;
 303                                 getline(is, dummy);
 304
 305                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 306                                                       << dummy << '\'' << endl;
 307 #else
 308                                 // but ignore is also still buggy (Lgb)
 309                                 // This is fast (Lgb)
 310                                 is.ignore(100, '\n');
 311 #endif
 312                                 ++lineno;
 313                                 continue;
 314                         }
 315
 316                         // string
 317                         if (c == '\"') {
 318                                 buff.clear();
 319
 320                                 bool escaped = false;
 321                                 do {
 322                                         escaped = false;
 323                                         is.get(cc);
 324                                         c = cc;
 325                                         if (c == '\r') continue;
 326                                         if (c == '\\') {
 327                                                 // escape the next char
 328                                                 is.get(cc);
 329                                                 c = cc;
 330                                                 if (c == '\"' || c == '\\')
 331                                                         escaped = true;
 332                                                 else
 333                                                         buff.push_back('\\');
 334                                         }
 335                                         buff.push_back(c);
 336
 337                                         if (!escaped && c == '\"') break;
 338                                 } while (c != '\n' && is);
 339
 340                                 if (c != '\"') {
 341                                         printError("Missing quote");
 342                                         if (c == '\n')
 343                                                 ++lineno;
 344                                 }
 345
 346                                 buff.pop_back();
 347                                 status = LEX_DATA;
 348                                 break;
 349                         }
 350
 351                         if (c > ' ' && is) {
 352                                 buff.clear();
 353
 354                                 do {
 355                                         if (c == '\\') {
 356                                                 // escape the next char
 357                                                 is.get(cc);
 358                                                 c = cc;
 359                                                 //escaped = true;
 360                                         }
 361                                         buff.push_back(c);
 362                                         is.get(cc);
 363                                         c = cc;
 364                                 } while (c > ' ' && c != ',' && is);
 365
 366                                 status = LEX_TOKEN;
 367                         }
 368                         // new line
 369                         if (c == '\n')
 370                                 ++lineno;
 371                 }
 372
 373                 if (status) return true;
 374
 375                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 376                 buff.clear();
 377                 return false;
 378         }
 379 }
 380
 381
 382 int LyXLex::Pimpl::search_kw(char const * const tag) const
 383 {
 384         keyword_item search_tag = { tag, 0 };
 385         keyword_item * res =
 386                 lower_bound(table, table + no_items,
 387                             search_tag, compare_tags());
 388         // use the compare_ascii_no_case instead of compare_no_case,
 389         // because in turkish, 'i' is not the lowercase version of 'I',
 390         // and thus turkish locale breaks parsing of tags.
 391         if (res != table + no_items
 392             && !compare_ascii_no_case(res->tag, tag))
 393                 return res->code;
 394         return LEX_UNDEF;
 395 }
 396
 397
 398 int LyXLex::Pimpl::lex()
 399 {
 400         //NOTE: possible bug.
 401         if (next() && status == LEX_TOKEN) {
 402                 return search_kw(getString().c_str());
 403         } else
 404                 return status;
 405 }
 406
 407
 408 bool LyXLex::Pimpl::eatLine()
 409 {
 410         buff.clear();
 411
 412         unsigned char c = '\0';
 413         char cc = 0;
 414         while (is && c != '\n') {
 415                 is.get(cc);
 416                 c = cc;
 417                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 418                 //                    << c << '\'' << endl;
 419                 if (c != '\r')
 420                         buff.push_back(c);
 421         }
 422
 423         if (c == '\n') {
 424                 ++lineno;
 425                 buff.pop_back();
 426                 status = LEX_DATA;
 427                 return true;
 428         } else {
 429                 return false;
 430         }
 431 }
 432
 433
 434 bool LyXLex::Pimpl::nextToken()
 435 {
 436         if (!pushTok.empty()) {
 437                 // There can have been a whole line pushed so
 438                 // we extract the first word and leaves the rest
 439                 // in pushTok. (Lgb)
 440                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 441                         string tmp;
 442                         pushTok = split(pushTok, tmp, ' ');
 443                         buff.assign(tmp.begin(), tmp.end());
 444                         return true;
 445                 } else {
 446                         buff.assign(pushTok.begin(), pushTok.end());
 447                         pushTok.erase();
 448                         return true;
 449                 }
 450         }
 451
 452         status = 0;
 453         while (is && !status) {
 454                 unsigned char c = 0;
 455                 char cc = 0;
 456                 is.get(cc);
 457                 c = cc;
 458                 if (c >= ' ' && is) {
 459                         buff.clear();
 460
 461                         if (c == '\\') { // first char == '\\'
 462                                 do {
 463                                         buff.push_back(c);
 464                                         is.get(cc);
 465                                         c = cc;
 466                                 } while (c > ' ' && c != '\\' && is);
 467                         } else {
 468                                 do {
 469                                         buff.push_back(c);
 470                                         is.get(cc);
 471                                         c = cc;
 472                                 } while (c >= ' ' && c != '\\' && is);
 473                         }
 474
 475                         if (c == '\\') is.putback(c); // put it back
 476                         status = LEX_TOKEN;
 477                 }
 478
 479                 if (c == '\n')
 480                         ++lineno;
 481
 482         }
 483         if (status)  return true;
 484
 485         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 486         buff.clear();
 487         return false;
 488 }
 489
 490
 491 void LyXLex::Pimpl::pushToken(string const & pt)
 492 {
 493         pushTok = pt;
 494 }