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::getFormatFromContents;
  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
  41         : public std::binary_function<keyword_item, keyword_item, bool> {
  42         // used by lower_bound, sort and sorted
  43         bool operator()(keyword_item const & a, keyword_item const & b) const
  44         {
  45                 // we use the ascii version, because in turkish, 'i'
  46                 // is not the lowercase version of 'I', and thus
  47                 // turkish locale breaks parsing of tags.
  48                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  49         }
  50 };
  51
  52 } // end of anon namespace
  53
  54
  55 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  56         : is(&fb_), table(tab), no_items(num),
  57           status(0), lineno(0), commentChar('#')
  58 {
  59         verifyTable();
  60 }
  61
  62
  63 string const LyXLex::Pimpl::getString() const
  64 {
  65         return string(buff.begin(), buff.end());
  66 }
  67
  68
  69 void LyXLex::Pimpl::printError(string const & message) const
  70 {
  71         string const tmpmsg = subst(message, "$$Token", getString());
  72         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  73                << " of file " << MakeDisplayPath(name) << ']' << endl;
  74 }
  75
  76
  77 void LyXLex::Pimpl::printTable(ostream & os)
  78 {
  79         os << "\nNumber of tags: " << no_items << endl;
  80         for (int i= 0; i < no_items; ++i)
  81                 os << "table[" << i
  82                    << "]:  tag: `" << table[i].tag
  83                    << "'  code:" << table[i].code << '\n';
  84         os.flush();
  85 }
  86
  87
  88 void LyXLex::Pimpl::verifyTable()
  89 {
  90         // Check if the table is sorted and if not, sort it.
  91         if (table
  92             && !lyx::sorted(table, table + no_items, compare_tags())) {
  93                 lyxerr << "The table passed to LyXLex is not sorted!\n"
  94                        << "Tell the developers to fix it!" << endl;
  95                 // We sort it anyway to avoid problems.
  96                 lyxerr << "\nUnsorted:" << endl;
  97                 printTable(lyxerr);
  98
  99                 sort(table, table + no_items, compare_tags());
 100                 lyxerr << "\nSorted:" << endl;
 101                 printTable(lyxerr);
 102         }
 103 }
 104
 105
 106 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
 107 {
 108         pushed_table tmppu(table, no_items);
 109         pushed.push(tmppu);
 110
 111         table = tab;
 112         no_items = num;
 113
 114         verifyTable();
 115 }
 116
 117
 118 void LyXLex::Pimpl::popTable()
 119 {
 120         if (pushed.empty()) {
 121                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 122                 return;
 123         }
 124
 125         pushed_table tmp = pushed.top();
 126         pushed.pop();
 127         table = tmp.table_elem;
 128         no_items = tmp.table_siz;
 129 }
 130
 131
 132 bool LyXLex::Pimpl::setFile(string const & filename)
 133 {
 134         // Check the format of the file.
 135         string const format = getFormatFromContents(filename);
 136
 137         if (format == "gzip" || format == "zip" || format == "compress") {
 138                 lyxerr[Debug::LYXLEX] << "lyxlex: compressed" << endl;
 139
 140                 // The check only outputs a debug message, because it triggers
 141                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 142                 // a fresh new filebuf.  (JMarc)
 143                 if (gz_.is_open() || istream::off_type(is.tellg()) > -1)
 144                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 145                                 "file or stream already set." << endl;
 146                 gz_.open(filename.c_str(), ios::in);
 147                 is.rdbuf(&gz_);
 148                 name = filename;
 149                 lineno = 0;
 150                 return gz_.is_open() && is.good();
 151         } else {
 152                 lyxerr[Debug::LYXLEX] << "lyxlex: UNcompressed" << endl;
 153
 154                 // The check only outputs a debug message, because it triggers
 155                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 156                 // a fresh new filebuf.  (JMarc)
 157                 if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 158                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 159                                 "file or stream already set." << endl;
 160                 fb_.open(filename.c_str(), ios::in);
 161                 is.rdbuf(&fb_);
 162                 name = filename;
 163                 lineno = 0;
 164                 return fb_.is_open() && is.good();
 165         }
 166 }
 167
 168
 169 void LyXLex::Pimpl::setStream(istream & i)
 170 {
 171         if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 172                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 173                         "file or stream already set." << endl;
 174         is.rdbuf(i.rdbuf());
 175         lineno = 0;
 176 }
 177
 178
 179 void LyXLex::Pimpl::setCommentChar(char c)
 180 {
 181         commentChar = c;
 182 }
 183
 184
 185 bool LyXLex::Pimpl::next(bool esc /* = false */)
 186 {
 187         if (!pushTok.empty()) {
 188                 // There can have been a whole line pushed so
 189                 // we extract the first word and leaves the rest
 190                 // in pushTok. (Lgb)
 191                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 192                         string tmp;
 193                         pushTok = split(pushTok, tmp, ' ');
 194                         buff.assign(tmp.begin(), tmp.end());
 195                         return true;
 196                 } else {
 197                         buff.assign(pushTok.begin(), pushTok.end());
 198                         pushTok.erase();
 199                         return true;
 200                 }
 201         }
 202         if (!esc) {
 203                 unsigned char c = 0; // getc() returns an int
 204                 char cc = 0;
 205                 status = 0;
 206                 while (is && !status) {
 207                         is.get(cc);
 208                         c = cc;
 209                         if (c == commentChar) {
 210                                 // Read rest of line (fast :-)
 211 #if 1
 212                                 // That is not fast... (Lgb)
 213                                 string dummy;
 214                                 getline(is, dummy);
 215
 216                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 217                                                       << dummy << '\'' << endl;
 218 #else
 219                                 // unfortunately ignore is buggy (Lgb)
 220                                 is.ignore(100, '\n');
 221 #endif
 222                                 ++lineno;
 223                                 continue;
 224                         }
 225
 226                         if (c == '\"') {
 227                                 buff.clear();
 228
 229                                 do {
 230                                         is.get(cc);
 231                                         c = cc;
 232                                         if (c != '\r')
 233                                                 buff.push_back(c);
 234                                 } while (c != '\"' && c != '\n' && is);
 235
 236                                 if (c != '\"') {
 237                                         printError("Missing quote");
 238                                         if (c == '\n')
 239                                                 ++lineno;
 240                                 }
 241
 242                                 buff.pop_back();
 243                                 status = LEX_DATA;
 244                                 break;
 245                         }
 246
 247                         if (c == ',')
 248                                 continue;              /* Skip ','s */
 249
 250                                 // using relational operators with chars other
 251                                 // than == and != is not safe. And if it is done
 252                                 // the type _have_ to be unsigned. It usually a
 253                                 // lot better to use the functions from cctype
 254                         if (c > ' ' && is)  {
 255                                 buff.clear();
 256
 257                                 do {
 258                                         buff.push_back(c);
 259                                         is.get(cc);
 260                                         c = cc;
 261                                 } while (c > ' ' && c != ',' && is);
 262
 263                                 status = LEX_TOKEN;
 264                         }
 265
 266                         if (c == '\r' && is) {
 267                                 // The Windows support has lead to the
 268                                 // possibility of "\r\n" at the end of
 269                                 // a line.  This will stop LyX choking
 270                                 // when it expected to find a '\n'
 271                                 is.get(cc);
 272                                 c = cc;
 273                         }
 274
 275                         if (c == '\n')
 276                                 ++lineno;
 277
 278                 }
 279                 if (status) return true;
 280
 281                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 282                 buff.clear();
 283                 return false;
 284         } else {
 285                 unsigned char c = 0; // getc() returns an int
 286                 char cc = 0;
 287
 288                 status = 0;
 289                 while (is && !status) {
 290                         is.get(cc);
 291                         c = cc;
 292
 293                         // skip ','s
 294                         if (c == ',') continue;
 295
 296                         if (c == '\\') {
 297                                 // escape
 298                                 buff.clear();
 299
 300                                 do {
 301                                         if (c == '\\') {
 302                                                 // escape the next char
 303                                                 is.get(cc);
 304                                                 c = cc;
 305                                         }
 306                                         buff.push_back(c);
 307                                         is.get(cc);
 308                                         c = cc;
 309                                 } while (c > ' ' && c != ',' && is);
 310
 311                                 status = LEX_TOKEN;
 312                                 continue;
 313                         }
 314
 315                         if (c == commentChar) {
 316                                 // Read rest of line (fast :-)
 317 #if 1
 318                                 // That is still not fast... (Lgb)
 319                                 string dummy;
 320                                 getline(is, dummy);
 321
 322                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 323                                                       << dummy << '\'' << endl;
 324 #else
 325                                 // but ignore is also still buggy (Lgb)
 326                                 // This is fast (Lgb)
 327                                 is.ignore(100, '\n');
 328 #endif
 329                                 ++lineno;
 330                                 continue;
 331                         }
 332
 333                         // string
 334                         if (c == '\"') {
 335                                 buff.clear();
 336
 337                                 bool escaped = false;
 338                                 do {
 339                                         escaped = false;
 340                                         is.get(cc);
 341                                         c = cc;
 342                                         if (c == '\r') continue;
 343                                         if (c == '\\') {
 344                                                 // escape the next char
 345                                                 is.get(cc);
 346                                                 c = cc;
 347                                                 if (c == '\"' || c == '\\')
 348                                                         escaped = true;
 349                                                 else
 350                                                         buff.push_back('\\');
 351                                         }
 352                                         buff.push_back(c);
 353
 354                                         if (!escaped && c == '\"') break;
 355                                 } while (c != '\n' && is);
 356
 357                                 if (c != '\"') {
 358                                         printError("Missing quote");
 359                                         if (c == '\n')
 360                                                 ++lineno;
 361                                 }
 362
 363                                 buff.pop_back();
 364                                 status = LEX_DATA;
 365                                 break;
 366                         }
 367
 368                         if (c > ' ' && is) {
 369                                 buff.clear();
 370
 371                                 do {
 372                                         if (c == '\\') {
 373                                                 // escape the next char
 374                                                 is.get(cc);
 375                                                 c = cc;
 376                                                 //escaped = true;
 377                                         }
 378                                         buff.push_back(c);
 379                                         is.get(cc);
 380                                         c = cc;
 381                                 } while (c > ' ' && c != ',' && is);
 382
 383                                 status = LEX_TOKEN;
 384                         }
 385                         // new line
 386                         if (c == '\n')
 387                                 ++lineno;
 388                 }
 389
 390                 if (status) return true;
 391
 392                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 393                 buff.clear();
 394                 return false;
 395         }
 396 }
 397
 398
 399 int LyXLex::Pimpl::search_kw(char const * const tag) const
 400 {
 401         keyword_item search_tag = { tag, 0 };
 402         keyword_item * res =
 403                 lower_bound(table, table + no_items,
 404                             search_tag, compare_tags());
 405         // use the compare_ascii_no_case instead of compare_no_case,
 406         // because in turkish, 'i' is not the lowercase version of 'I',
 407         // and thus turkish locale breaks parsing of tags.
 408         if (res != table + no_items
 409             && !compare_ascii_no_case(res->tag, tag))
 410                 return res->code;
 411         return LEX_UNDEF;
 412 }
 413
 414
 415 int LyXLex::Pimpl::lex()
 416 {
 417         //NOTE: possible bug.
 418         if (next() && status == LEX_TOKEN) {
 419                 return search_kw(getString().c_str());
 420         } else
 421                 return status;
 422 }
 423
 424
 425 bool LyXLex::Pimpl::eatLine()
 426 {
 427         buff.clear();
 428
 429         unsigned char c = '\0';
 430         char cc = 0;
 431         while (is && c != '\n') {
 432                 is.get(cc);
 433                 c = cc;
 434                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 435                 //                    << c << '\'' << endl;
 436                 if (c != '\r')
 437                         buff.push_back(c);
 438         }
 439
 440         if (c == '\n') {
 441                 ++lineno;
 442                 buff.pop_back();
 443                 status = LEX_DATA;
 444                 return true;
 445         } else {
 446                 return false;
 447         }
 448 }
 449
 450
 451 bool LyXLex::Pimpl::nextToken()
 452 {
 453         if (!pushTok.empty()) {
 454                 // There can have been a whole line pushed so
 455                 // we extract the first word and leaves the rest
 456                 // in pushTok. (Lgb)
 457                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 458                         string tmp;
 459                         pushTok = split(pushTok, tmp, ' ');
 460                         buff.assign(tmp.begin(), tmp.end());
 461                         return true;
 462                 } else {
 463                         buff.assign(pushTok.begin(), pushTok.end());
 464                         pushTok.erase();
 465                         return true;
 466                 }
 467         }
 468
 469         status = 0;
 470         while (is && !status) {
 471                 unsigned char c = 0;
 472                 char cc = 0;
 473                 is.get(cc);
 474                 c = cc;
 475                 if (c >= ' ' && is) {
 476                         buff.clear();
 477
 478                         if (c == '\\') { // first char == '\\'
 479                                 do {
 480                                         buff.push_back(c);
 481                                         is.get(cc);
 482                                         c = cc;
 483                                 } while (c > ' ' && c != '\\' && is);
 484                         } else {
 485                                 do {
 486                                         buff.push_back(c);
 487                                         is.get(cc);
 488                                         c = cc;
 489                                 } while (c >= ' ' && c != '\\' && is);
 490                         }
 491
 492                         if (c == '\\') is.putback(c); // put it back
 493                         status = LEX_TOKEN;
 494                 }
 495
 496                 if (c == '\n')
 497                         ++lineno;
 498
 499         }
 500         if (status)  return true;
 501
 502         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 503         buff.clear();
 504         return false;
 505 }
 506
 507
 508 void LyXLex::Pimpl::pushToken(string const & pt)
 509 {
 510         pushTok = pt;
 511 }