src/lyxlex_pimpl.C

   1 /**
   2  * \file lyxlex_pimpl.C
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author Lars Gullik Bjønnes
   7  * \author Jean-Marc Lasgouttes
   8  * \author Jürgen Vigna
   9  *
  10  * Full author contact details are available in file CREDITS.
  11  */
  12
  13 #include <config.h>
  14
  15 #include "lyxlex_pimpl.h"
  16
  17 #include "debug.h"
  18
  19 #include "support/filetools.h"
  20 #include "support/lyxalgo.h"
  21 #include "support/lstrings.h"
  22
  23 using lyx::support::compare_ascii_no_case;
  24 using lyx::support::getExtFromContents;
  25 using lyx::support::MakeDisplayPath;
  26 using lyx::support::split;
  27 using lyx::support::subst;
  28
  29 using std::endl;
  30 using std::getline;
  31 using std::lower_bound;
  32 using std::sort;
  33 using std::string;
  34 using std::ios;
  35 using std::istream;
  36 using std::ostream;
  37
  38 namespace {
  39
  40 struct compare_tags : public std::binary_function<keyword_item, keyword_item, int> {
  41         // used by lower_bound, sort and sorted
  42         int operator()(keyword_item const & a, keyword_item const & b) const {
  43                 // we use the ascii version, because in turkish, 'i'
  44                 // is not the lowercase version of 'I', and thus
  45                 // turkish locale breaks parsing of tags.
  46                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  47         }
  48 };
  49
  50 } // end of anon namespace
  51
  52
  53 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  54         : is(&fb_), table(tab), no_items(num),
  55           status(0), lineno(0), commentChar('#')
  56 {
  57         verifyTable();
  58 }
  59
  60
  61 string const LyXLex::Pimpl::getString() const
  62 {
  63         return string(buff.begin(), buff.end());
  64 }
  65
  66
  67 void LyXLex::Pimpl::printError(string const & message) const
  68 {
  69         string const tmpmsg = subst(message, "$$Token", getString());
  70         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  71                << " of file " << MakeDisplayPath(name) << ']' << endl;
  72 }
  73
  74
  75 void LyXLex::Pimpl::printTable(ostream & os)
  76 {
  77         os << "\nNumber of tags: " << no_items << endl;
  78         for (int i= 0; i < no_items; ++i)
  79                 os << "table[" << i
  80                    << "]:  tag: `" << table[i].tag
  81                    << "'  code:" << table[i].code << '\n';
  82         os.flush();
  83 }
  84
  85
  86 void LyXLex::Pimpl::verifyTable()
  87 {
  88         // Check if the table is sorted and if not, sort it.
  89         if (table
  90             && !lyx::sorted(table, table + no_items, compare_tags())) {
  91                 lyxerr << "The table passed to LyXLex is not sorted!\n"
  92                        << "Tell the developers to fix it!" << endl;
  93                 // We sort it anyway to avoid problems.
  94                 lyxerr << "\nUnsorted:" << endl;
  95                 printTable(lyxerr);
  96
  97                 sort(table, table + no_items, compare_tags());
  98                 lyxerr << "\nSorted:" << endl;
  99                 printTable(lyxerr);
 100         }
 101 }
 102
 103
 104 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
 105 {
 106         pushed_table tmppu(table, no_items);
 107         pushed.push(tmppu);
 108
 109         table = tab;
 110         no_items = num;
 111
 112         verifyTable();
 113 }
 114
 115
 116 void LyXLex::Pimpl::popTable()
 117 {
 118         if (pushed.empty()) {
 119                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 120                 return;
 121         }
 122
 123         pushed_table tmp = pushed.top();
 124         pushed.pop();
 125         table = tmp.table_elem;
 126         no_items = tmp.table_siz;
 127 }
 128
 129
 130 bool LyXLex::Pimpl::setFile(string const & filename)
 131 {
 132
 133         // Check the format of the file.
 134         string const format = getExtFromContents(filename);
 135
 136         if (format == "gzip" || format == "zip" || format == "compress") {
 137                 lyxerr[Debug::LYXLEX] << "lyxlex: compressed" << endl;
 138
 139                 // The check only outputs a debug message, because it triggers
 140                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 141                 // a fresh new filebuf.  (JMarc)
 142                 if (gz_.is_open() || istream::off_type(is.tellg()) > -1)
 143                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 144                                 "file or stream already set." << endl;
 145                 gz_.open(filename.c_str(), ios::in);
 146                 is.rdbuf(&gz_);
 147                 name = filename;
 148                 lineno = 0;
 149                 return gz_.is_open() && is.good();
 150         } else {
 151                 lyxerr[Debug::LYXLEX] << "lyxlex: UNcompressed" << endl;
 152
 153                 // The check only outputs a debug message, because it triggers
 154                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 155                 // a fresh new filebuf.  (JMarc)
 156                 if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 157                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 158                                 "file or stream already set." << endl;
 159                 fb_.open(filename.c_str(), ios::in);
 160                 is.rdbuf(&fb_);
 161                 name = filename;
 162                 lineno = 0;
 163                 return fb_.is_open() && is.good();
 164         }
 165 }
 166
 167
 168 void LyXLex::Pimpl::setStream(istream & i)
 169 {
 170         if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 171                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 172                         "file or stream already set." << endl;
 173         is.rdbuf(i.rdbuf());
 174         lineno = 0;
 175 }
 176
 177
 178 void LyXLex::Pimpl::setCommentChar(char c)
 179 {
 180         commentChar = c;
 181 }
 182
 183
 184 bool LyXLex::Pimpl::next(bool esc /* = false */)
 185 {
 186         if (!pushTok.empty()) {
 187                 // There can have been a whole line pushed so
 188                 // we extract the first word and leaves the rest
 189                 // in pushTok. (Lgb)
 190                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 191                         string tmp;
 192                         pushTok = split(pushTok, tmp, ' ');
 193                         buff.assign(tmp.begin(), tmp.end());
 194                         return true;
 195                 } else {
 196                         buff.assign(pushTok.begin(), pushTok.end());
 197                         pushTok.erase();
 198                         return true;
 199                 }
 200         }
 201         if (!esc) {
 202                 unsigned char c = 0; // getc() returns an int
 203                 char cc = 0;
 204                 status = 0;
 205                 while (is && !status) {
 206                         is.get(cc);
 207                         c = cc;
 208                         if (c == commentChar) {
 209                                 // Read rest of line (fast :-)
 210 #if 1
 211                                 // That is not fast... (Lgb)
 212                                 string dummy;
 213                                 getline(is, dummy);
 214
 215                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 216                                                       << dummy << '\'' << endl;
 217 #else
 218                                 // unfortunately ignore is buggy (Lgb)
 219                                 is.ignore(100, '\n');
 220 #endif
 221                                 ++lineno;
 222                                 continue;
 223                         }
 224
 225                         if (c == '\"') {
 226                                 buff.clear();
 227
 228                                 do {
 229                                         is.get(cc);
 230                                         c = cc;
 231                                         if (c != '\r')
 232                                                 buff.push_back(c);
 233                                 } while (c != '\"' && c != '\n' && is);
 234
 235                                 if (c != '\"') {
 236                                         printError("Missing quote");
 237                                         if (c == '\n')
 238                                                 ++lineno;
 239                                 }
 240
 241                                 buff.pop_back();
 242                                 status = LEX_DATA;
 243                                 break;
 244                         }
 245
 246                         if (c == ',')
 247                                 continue;              /* Skip ','s */
 248
 249                                 // using relational operators with chars other
 250                                 // than == and != is not safe. And if it is done
 251                                 // the type _have_ to be unsigned. It usually a
 252                                 // lot better to use the functions from cctype
 253                         if (c > ' ' && is)  {
 254                                 buff.clear();
 255
 256                                 do {
 257                                         buff.push_back(c);
 258                                         is.get(cc);
 259                                         c = cc;
 260                                 } while (c > ' ' && c != ',' && is);
 261
 262                                 status = LEX_TOKEN;
 263                         }
 264
 265                         if (c == '\r' && is) {
 266                                 // The Windows support has lead to the
 267                                 // possibility of "\r\n" at the end of
 268                                 // a line.  This will stop LyX choking
 269                                 // when it expected to find a '\n'
 270                                 is.get(cc);
 271                                 c = cc;
 272                         }
 273
 274                         if (c == '\n')
 275                                 ++lineno;
 276
 277                 }
 278                 if (status) return true;
 279
 280                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 281                 buff.clear();
 282                 return false;
 283         } else {
 284                 unsigned char c = 0; // getc() returns an int
 285                 char cc = 0;
 286
 287                 status = 0;
 288                 while (is && !status) {
 289                         is.get(cc);
 290                         c = cc;
 291
 292                         // skip ','s
 293                         if (c == ',') continue;
 294
 295                         if (c == '\\') {
 296                                 // escape
 297                                 buff.clear();
 298
 299                                 do {
 300                                         if (c == '\\') {
 301                                                 // escape the next char
 302                                                 is.get(cc);
 303                                                 c = cc;
 304                                         }
 305                                         buff.push_back(c);
 306                                         is.get(cc);
 307                                         c = cc;
 308                                 } while (c > ' ' && c != ',' && is);
 309
 310                                 status = LEX_TOKEN;
 311                                 continue;
 312                         }
 313
 314                         if (c == commentChar) {
 315                                 // Read rest of line (fast :-)
 316 #if 1
 317                                 // That is still not fast... (Lgb)
 318                                 string dummy;
 319                                 getline(is, dummy);
 320
 321                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 322                                                       << dummy << '\'' << endl;
 323 #else
 324                                 // but ignore is also still buggy (Lgb)
 325                                 // This is fast (Lgb)
 326                                 is.ignore(100, '\n');
 327 #endif
 328                                 ++lineno;
 329                                 continue;
 330                         }
 331
 332                         // string
 333                         if (c == '\"') {
 334                                 buff.clear();
 335
 336                                 bool escaped = false;
 337                                 do {
 338                                         escaped = false;
 339                                         is.get(cc);
 340                                         c = cc;
 341                                         if (c == '\r') continue;
 342                                         if (c == '\\') {
 343                                                 // escape the next char
 344                                                 is.get(cc);
 345                                                 c = cc;
 346                                                 if (c == '\"' || c == '\\')
 347                                                         escaped = true;
 348                                                 else
 349                                                         buff.push_back('\\');
 350                                         }
 351                                         buff.push_back(c);
 352
 353                                         if (!escaped && c == '\"') break;
 354                                 } while (c != '\n' && is);
 355
 356                                 if (c != '\"') {
 357                                         printError("Missing quote");
 358                                         if (c == '\n')
 359                                                 ++lineno;
 360                                 }
 361
 362                                 buff.pop_back();
 363                                 status = LEX_DATA;
 364                                 break;
 365                         }
 366
 367                         if (c > ' ' && is) {
 368                                 buff.clear();
 369
 370                                 do {
 371                                         if (c == '\\') {
 372                                                 // escape the next char
 373                                                 is.get(cc);
 374                                                 c = cc;
 375                                                 //escaped = true;
 376                                         }
 377                                         buff.push_back(c);
 378                                         is.get(cc);
 379                                         c = cc;
 380                                 } while (c > ' ' && c != ',' && is);
 381
 382                                 status = LEX_TOKEN;
 383                         }
 384                         // new line
 385                         if (c == '\n')
 386                                 ++lineno;
 387                 }
 388
 389                 if (status) return true;
 390
 391                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 392                 buff.clear();
 393                 return false;
 394         }
 395 }
 396
 397
 398 int LyXLex::Pimpl::search_kw(char const * const tag) const
 399 {
 400         keyword_item search_tag = { tag, 0 };
 401         keyword_item * res =
 402                 lower_bound(table, table + no_items,
 403                             search_tag, compare_tags());
 404         // use the compare_ascii_no_case instead of compare_no_case,
 405         // because in turkish, 'i' is not the lowercase version of 'I',
 406         // and thus turkish locale breaks parsing of tags.
 407         if (res != table + no_items
 408             && !compare_ascii_no_case(res->tag, tag))
 409                 return res->code;
 410         return LEX_UNDEF;
 411 }
 412
 413
 414 int LyXLex::Pimpl::lex()
 415 {
 416         //NOTE: possible bug.
 417         if (next() && status == LEX_TOKEN) {
 418                 return search_kw(getString().c_str());
 419         } else
 420                 return status;
 421 }
 422
 423
 424 bool LyXLex::Pimpl::eatLine()
 425 {
 426         buff.clear();
 427
 428         unsigned char c = '\0';
 429         char cc = 0;
 430         while (is && c != '\n') {
 431                 is.get(cc);
 432                 c = cc;
 433                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 434                 //                    << c << '\'' << endl;
 435                 if (c != '\r')
 436                         buff.push_back(c);
 437         }
 438
 439         if (c == '\n') {
 440                 ++lineno;
 441                 buff.pop_back();
 442                 status = LEX_DATA;
 443                 return true;
 444         } else {
 445                 return false;
 446         }
 447 }
 448
 449
 450 bool LyXLex::Pimpl::nextToken()
 451 {
 452         if (!pushTok.empty()) {
 453                 // There can have been a whole line pushed so
 454                 // we extract the first word and leaves the rest
 455                 // in pushTok. (Lgb)
 456                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 457                         string tmp;
 458                         pushTok = split(pushTok, tmp, ' ');
 459                         buff.assign(tmp.begin(), tmp.end());
 460                         return true;
 461                 } else {
 462                         buff.assign(pushTok.begin(), pushTok.end());
 463                         pushTok.erase();
 464                         return true;
 465                 }
 466         }
 467
 468         status = 0;
 469         while (is && !status) {
 470                 unsigned char c = 0;
 471                 char cc = 0;
 472                 is.get(cc);
 473                 c = cc;
 474                 if (c >= ' ' && is) {
 475                         buff.clear();
 476
 477                         if (c == '\\') { // first char == '\\'
 478                                 do {
 479                                         buff.push_back(c);
 480                                         is.get(cc);
 481                                         c = cc;
 482                                 } while (c > ' ' && c != '\\' && is);
 483                         } else {
 484                                 do {
 485                                         buff.push_back(c);
 486                                         is.get(cc);
 487                                         c = cc;
 488                                 } while (c >= ' ' && c != '\\' && is);
 489                         }
 490
 491                         if (c == '\\') is.putback(c); // put it back
 492                         status = LEX_TOKEN;
 493                 }
 494
 495                 if (c == '\n')
 496                         ++lineno;
 497
 498         }
 499         if (status)  return true;
 500
 501         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 502         buff.clear();
 503         return false;
 504 }
 505
 506
 507 void LyXLex::Pimpl::pushToken(string const & pt)
 508 {
 509         pushTok = pt;
 510 }