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