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
  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
 135         // Check the format of the file.
 136         string const format = getExtFromContents(filename);
 137
 138         if (format == "gzip" || format == "zip" || format == "compress") {
 139                 lyxerr[Debug::LYXLEX] << "lyxlex: compressed" << endl;
 140
 141                 // The check only outputs a debug message, because it triggers
 142                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 143                 // a fresh new filebuf.  (JMarc)
 144                 if (gz_.is_open() || istream::off_type(is.tellg()) > -1)
 145                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 146                                 "file or stream already set." << endl;
 147                 gz_.open(filename.c_str(), ios::in);
 148                 is.rdbuf(&gz_);
 149                 name = filename;
 150                 lineno = 0;
 151                 return gz_.is_open() && is.good();
 152         } else {
 153                 lyxerr[Debug::LYXLEX] << "lyxlex: UNcompressed" << endl;
 154
 155                 // The check only outputs a debug message, because it triggers
 156                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 157                 // a fresh new filebuf.  (JMarc)
 158                 if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 159                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 160                                 "file or stream already set." << endl;
 161                 fb_.open(filename.c_str(), ios::in);
 162                 is.rdbuf(&fb_);
 163                 name = filename;
 164                 lineno = 0;
 165                 return fb_.is_open() && is.good();
 166         }
 167 }
 168
 169
 170 void LyXLex::Pimpl::setStream(istream & i)
 171 {
 172         if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 173                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 174                         "file or stream already set." << endl;
 175         is.rdbuf(i.rdbuf());
 176         lineno = 0;
 177 }
 178
 179
 180 void LyXLex::Pimpl::setCommentChar(char c)
 181 {
 182         commentChar = c;
 183 }
 184
 185
 186 bool LyXLex::Pimpl::next(bool esc /* = false */)
 187 {
 188         if (!pushTok.empty()) {
 189                 // There can have been a whole line pushed so
 190                 // we extract the first word and leaves the rest
 191                 // in pushTok. (Lgb)
 192                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 193                         string tmp;
 194                         pushTok = split(pushTok, tmp, ' ');
 195                         buff.assign(tmp.begin(), tmp.end());
 196                         return true;
 197                 } else {
 198                         buff.assign(pushTok.begin(), pushTok.end());
 199                         pushTok.erase();
 200                         return true;
 201                 }
 202         }
 203         if (!esc) {
 204                 unsigned char c = 0; // getc() returns an int
 205                 char cc = 0;
 206                 status = 0;
 207                 while (is && !status) {
 208                         is.get(cc);
 209                         c = cc;
 210                         if (c == commentChar) {
 211                                 // Read rest of line (fast :-)
 212 #if 1
 213                                 // That is not fast... (Lgb)
 214                                 string dummy;
 215                                 getline(is, dummy);
 216
 217                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 218                                                       << dummy << '\'' << endl;
 219 #else
 220                                 // unfortunately ignore is buggy (Lgb)
 221                                 is.ignore(100, '\n');
 222 #endif
 223                                 ++lineno;
 224                                 continue;
 225                         }
 226
 227                         if (c == '\"') {
 228                                 buff.clear();
 229
 230                                 do {
 231                                         is.get(cc);
 232                                         c = cc;
 233                                         if (c != '\r')
 234                                                 buff.push_back(c);
 235                                 } while (c != '\"' && c != '\n' && is);
 236
 237                                 if (c != '\"') {
 238                                         printError("Missing quote");
 239                                         if (c == '\n')
 240                                                 ++lineno;
 241                                 }
 242
 243                                 buff.pop_back();
 244                                 status = LEX_DATA;
 245                                 break;
 246                         }
 247
 248                         if (c == ',')
 249                                 continue;              /* Skip ','s */
 250
 251                                 // using relational operators with chars other
 252                                 // than == and != is not safe. And if it is done
 253                                 // the type _have_ to be unsigned. It usually a
 254                                 // lot better to use the functions from cctype
 255                         if (c > ' ' && is)  {
 256                                 buff.clear();
 257
 258                                 do {
 259                                         buff.push_back(c);
 260                                         is.get(cc);
 261                                         c = cc;
 262                                 } while (c > ' ' && c != ',' && is);
 263
 264                                 status = LEX_TOKEN;
 265                         }
 266
 267                         if (c == '\r' && is) {
 268                                 // The Windows support has lead to the
 269                                 // possibility of "\r\n" at the end of
 270                                 // a line.  This will stop LyX choking
 271                                 // when it expected to find a '\n'
 272                                 is.get(cc);
 273                                 c = cc;
 274                         }
 275
 276                         if (c == '\n')
 277                                 ++lineno;
 278
 279                 }
 280                 if (status) return true;
 281
 282                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 283                 buff.clear();
 284                 return false;
 285         } else {
 286                 unsigned char c = 0; // getc() returns an int
 287                 char cc = 0;
 288
 289                 status = 0;
 290                 while (is && !status) {
 291                         is.get(cc);
 292                         c = cc;
 293
 294                         // skip ','s
 295                         if (c == ',') continue;
 296
 297                         if (c == '\\') {
 298                                 // escape
 299                                 buff.clear();
 300
 301                                 do {
 302                                         if (c == '\\') {
 303                                                 // escape the next char
 304                                                 is.get(cc);
 305                                                 c = cc;
 306                                         }
 307                                         buff.push_back(c);
 308                                         is.get(cc);
 309                                         c = cc;
 310                                 } while (c > ' ' && c != ',' && is);
 311
 312                                 status = LEX_TOKEN;
 313                                 continue;
 314                         }
 315
 316                         if (c == commentChar) {
 317                                 // Read rest of line (fast :-)
 318 #if 1
 319                                 // That is still not fast... (Lgb)
 320                                 string dummy;
 321                                 getline(is, dummy);
 322
 323                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 324                                                       << dummy << '\'' << endl;
 325 #else
 326                                 // but ignore is also still buggy (Lgb)
 327                                 // This is fast (Lgb)
 328                                 is.ignore(100, '\n');
 329 #endif
 330                                 ++lineno;
 331                                 continue;
 332                         }
 333
 334                         // string
 335                         if (c == '\"') {
 336                                 buff.clear();
 337
 338                                 bool escaped = false;
 339                                 do {
 340                                         escaped = false;
 341                                         is.get(cc);
 342                                         c = cc;
 343                                         if (c == '\r') continue;
 344                                         if (c == '\\') {
 345                                                 // escape the next char
 346                                                 is.get(cc);
 347                                                 c = cc;
 348                                                 if (c == '\"' || c == '\\')
 349                                                         escaped = true;
 350                                                 else
 351                                                         buff.push_back('\\');
 352                                         }
 353                                         buff.push_back(c);
 354
 355                                         if (!escaped && c == '\"') break;
 356                                 } while (c != '\n' && is);
 357
 358                                 if (c != '\"') {
 359                                         printError("Missing quote");
 360                                         if (c == '\n')
 361                                                 ++lineno;
 362                                 }
 363
 364                                 buff.pop_back();
 365                                 status = LEX_DATA;
 366                                 break;
 367                         }
 368
 369                         if (c > ' ' && is) {
 370                                 buff.clear();
 371
 372                                 do {
 373                                         if (c == '\\') {
 374                                                 // escape the next char
 375                                                 is.get(cc);
 376                                                 c = cc;
 377                                                 //escaped = true;
 378                                         }
 379                                         buff.push_back(c);
 380                                         is.get(cc);
 381                                         c = cc;
 382                                 } while (c > ' ' && c != ',' && is);
 383
 384                                 status = LEX_TOKEN;
 385                         }
 386                         // new line
 387                         if (c == '\n')
 388                                 ++lineno;
 389                 }
 390
 391                 if (status) return true;
 392
 393                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 394                 buff.clear();
 395                 return false;
 396         }
 397 }
 398
 399
 400 int LyXLex::Pimpl::search_kw(char const * const tag) const
 401 {
 402         keyword_item search_tag = { tag, 0 };
 403         keyword_item * res =
 404                 lower_bound(table, table + no_items,
 405                             search_tag, compare_tags());
 406         // use the compare_ascii_no_case instead of compare_no_case,
 407         // because in turkish, 'i' is not the lowercase version of 'I',
 408         // and thus turkish locale breaks parsing of tags.
 409         if (res != table + no_items
 410             && !compare_ascii_no_case(res->tag, tag))
 411                 return res->code;
 412         return LEX_UNDEF;
 413 }
 414
 415
 416 int LyXLex::Pimpl::lex()
 417 {
 418         //NOTE: possible bug.
 419         if (next() && status == LEX_TOKEN) {
 420                 return search_kw(getString().c_str());
 421         } else
 422                 return status;
 423 }
 424
 425
 426 bool LyXLex::Pimpl::eatLine()
 427 {
 428         buff.clear();
 429
 430         unsigned char c = '\0';
 431         char cc = 0;
 432         while (is && c != '\n') {
 433                 is.get(cc);
 434                 c = cc;
 435                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 436                 //                    << c << '\'' << endl;
 437                 if (c != '\r')
 438                         buff.push_back(c);
 439         }
 440
 441         if (c == '\n') {
 442                 ++lineno;
 443                 buff.pop_back();
 444                 status = LEX_DATA;
 445                 return true;
 446         } else {
 447                 return false;
 448         }
 449 }
 450
 451
 452 bool LyXLex::Pimpl::nextToken()
 453 {
 454         if (!pushTok.empty()) {
 455                 // There can have been a whole line pushed so
 456                 // we extract the first word and leaves the rest
 457                 // in pushTok. (Lgb)
 458                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 459                         string tmp;
 460                         pushTok = split(pushTok, tmp, ' ');
 461                         buff.assign(tmp.begin(), tmp.end());
 462                         return true;
 463                 } else {
 464                         buff.assign(pushTok.begin(), pushTok.end());
 465                         pushTok.erase();
 466                         return true;
 467                 }
 468         }
 469
 470         status = 0;
 471         while (is && !status) {
 472                 unsigned char c = 0;
 473                 char cc = 0;
 474                 is.get(cc);
 475                 c = cc;
 476                 if (c >= ' ' && is) {
 477                         buff.clear();
 478
 479                         if (c == '\\') { // first char == '\\'
 480                                 do {
 481                                         buff.push_back(c);
 482                                         is.get(cc);
 483                                         c = cc;
 484                                 } while (c > ' ' && c != '\\' && is);
 485                         } else {
 486                                 do {
 487                                         buff.push_back(c);
 488                                         is.get(cc);
 489                                         c = cc;
 490                                 } while (c >= ' ' && c != '\\' && is);
 491                         }
 492
 493                         if (c == '\\') is.putback(c); // put it back
 494                         status = LEX_TOKEN;
 495                 }
 496
 497                 if (c == '\n')
 498                         ++lineno;
 499
 500         }
 501         if (status)  return true;
 502
 503         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 504         buff.clear();
 505         return false;
 506 }
 507
 508
 509 void LyXLex::Pimpl::pushToken(string const & pt)
 510 {
 511         pushTok = pt;
 512 }