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