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 #include <functional>
  24
  25 using lyx::support::compare_ascii_no_case;
  26 using lyx::support::getFormatFromContents;
  27 using lyx::support::makeDisplayPath;
  28 using lyx::support::split;
  29 using lyx::support::subst;
  30
  31 using std::endl;
  32 using std::getline;
  33 using std::lower_bound;
  34 using std::sort;
  35 using std::string;
  36 using std::ios;
  37 using std::istream;
  38 using std::ostream;
  39
  40 namespace {
  41
  42 class compare_tags
  43         : public std::binary_function<keyword_item, keyword_item, bool> {
  44 public:
  45         // used by lower_bound, sort and sorted
  46         bool operator()(keyword_item const & a, keyword_item const & b) const
  47         {
  48                 // we use the ascii version, because in turkish, 'i'
  49                 // is not the lowercase version of 'I', and thus
  50                 // turkish locale breaks parsing of tags.
  51                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  52         }
  53 };
  54
  55 } // end of anon namespace
  56
  57
  58 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  59         : is(&fb_), table(tab), no_items(num),
  60           status(0), lineno(0), commentChar('#')
  61 {
  62         verifyTable();
  63 }
  64
  65
  66 string const LyXLex::Pimpl::getString() const
  67 {
  68         return string(buff.begin(), buff.end());
  69 }
  70
  71
  72 void LyXLex::Pimpl::printError(string const & message) const
  73 {
  74         string const tmpmsg = subst(message, "$$Token", getString());
  75         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  76                << " of file " << makeDisplayPath(name) << ']' << endl;
  77 }
  78
  79
  80 void LyXLex::Pimpl::printTable(ostream & os)
  81 {
  82         os << "\nNumber of tags: " << no_items << endl;
  83         for (int i= 0; i < no_items; ++i)
  84                 os << "table[" << i
  85                    << "]:  tag: `" << table[i].tag
  86                    << "'  code:" << table[i].code << '\n';
  87         os.flush();
  88 }
  89
  90
  91 void LyXLex::Pimpl::verifyTable()
  92 {
  93         // Check if the table is sorted and if not, sort it.
  94         if (table
  95             && !lyx::sorted(table, table + no_items, compare_tags())) {
  96                 lyxerr << "The table passed to LyXLex is not sorted!\n"
  97                        << "Tell the developers to fix it!" << endl;
  98                 // We sort it anyway to avoid problems.
  99                 lyxerr << "\nUnsorted:" << endl;
 100                 printTable(lyxerr);
 101
 102                 sort(table, table + no_items, compare_tags());
 103                 lyxerr << "\nSorted:" << endl;
 104                 printTable(lyxerr);
 105         }
 106 }
 107
 108
 109 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
 110 {
 111         pushed_table tmppu(table, no_items);
 112         pushed.push(tmppu);
 113
 114         table = tab;
 115         no_items = num;
 116
 117         verifyTable();
 118 }
 119
 120
 121 void LyXLex::Pimpl::popTable()
 122 {
 123         if (pushed.empty()) {
 124                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 125                 return;
 126         }
 127
 128         pushed_table tmp = pushed.top();
 129         pushed.pop();
 130         table = tmp.table_elem;
 131         no_items = tmp.table_siz;
 132 }
 133
 134
 135 bool LyXLex::Pimpl::setFile(string const & filename)
 136 {
 137         // Check the format of the file.
 138         string const format = getFormatFromContents(filename);
 139
 140         if (format == "gzip" || format == "zip" || format == "compress") {
 141                 lyxerr[Debug::LYXLEX] << "lyxlex: compressed" << endl;
 142
 143 #ifdef USE_COMPRESSION
 144                 // The check only outputs a debug message, because it triggers
 145                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 146                 // a fresh new filebuf.  (JMarc)
 147                 if (!gz_.empty() || istream::off_type(is.tellg()) > -1)
 148                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 149                                 "file or stream already set." << endl;
 150                 gz_.push(io::gzip_decompressor());
 151                 gz_.push(io::file_source(filename));
 152                 is.rdbuf(&gz_);
 153                 name = filename;
 154                 lineno = 0;
 155                 return gz_.component<io::file_source>(1)->is_open() && is.good();
 156 #else
 157                 return false;
 158 #endif
 159         } else {
 160                 lyxerr[Debug::LYXLEX] << "lyxlex: UNcompressed" << endl;
 161
 162                 // The check only outputs a debug message, because it triggers
 163                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 164                 // a fresh new filebuf.  (JMarc)
 165                 if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 166                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 167                                 "file or stream already set." << endl;
 168                 fb_.open(filename.c_str(), ios::in);
 169                 is.rdbuf(&fb_);
 170                 name = filename;
 171                 lineno = 0;
 172                 return fb_.is_open() && is.good();
 173         }
 174 }
 175
 176
 177 void LyXLex::Pimpl::setStream(istream & i)
 178 {
 179         if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 180                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 181                         "file or stream already set." << endl;
 182         is.rdbuf(i.rdbuf());
 183         lineno = 0;
 184 }
 185
 186
 187 void LyXLex::Pimpl::setCommentChar(char c)
 188 {
 189         commentChar = c;
 190 }
 191
 192
 193 bool LyXLex::Pimpl::next(bool esc /* = false */)
 194 {
 195         if (!pushTok.empty()) {
 196                 // There can have been a whole line pushed so
 197                 // we extract the first word and leaves the rest
 198                 // in pushTok. (Lgb)
 199                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 200                         string tmp;
 201                         pushTok = split(pushTok, tmp, ' ');
 202                         buff.assign(tmp.begin(), tmp.end());
 203                         return true;
 204                 } else {
 205                         buff.assign(pushTok.begin(), pushTok.end());
 206                         pushTok.erase();
 207                         return true;
 208                 }
 209         }
 210         if (!esc) {
 211                 unsigned char c = 0; // getc() returns an int
 212                 char cc = 0;
 213                 status = 0;
 214                 while (is && !status) {
 215                         is.get(cc);
 216                         c = cc;
 217                         if (c == commentChar) {
 218                                 // Read rest of line (fast :-)
 219 #if 1
 220                                 // That is not fast... (Lgb)
 221                                 string dummy;
 222                                 getline(is, dummy);
 223
 224                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 225                                                       << dummy << '\'' << endl;
 226 #else
 227                                 // unfortunately ignore is buggy (Lgb)
 228                                 is.ignore(100, '\n');
 229 #endif
 230                                 ++lineno;
 231                                 continue;
 232                         }
 233
 234                         if (c == '\"') {
 235                                 buff.clear();
 236
 237                                 do {
 238                                         is.get(cc);
 239                                         c = cc;
 240                                         if (c != '\r')
 241                                                 buff.push_back(c);
 242                                 } while (c != '\"' && c != '\n' && is);
 243
 244                                 if (c != '\"') {
 245                                         printError("Missing quote");
 246                                         if (c == '\n')
 247                                                 ++lineno;
 248                                 }
 249
 250                                 buff.pop_back();
 251                                 status = LEX_DATA;
 252                                 break;
 253                         }
 254
 255                         if (c == ',')
 256                                 continue;              /* Skip ','s */
 257
 258                                 // using relational operators with chars other
 259                                 // than == and != is not safe. And if it is done
 260                                 // the type _have_ to be unsigned. It usually a
 261                                 // lot better to use the functions from cctype
 262                         if (c > ' ' && is)  {
 263                                 buff.clear();
 264
 265                                 do {
 266                                         buff.push_back(c);
 267                                         is.get(cc);
 268                                         c = cc;
 269                                 } while (c > ' ' && c != ',' && is);
 270
 271                                 status = LEX_TOKEN;
 272                         }
 273
 274                         if (c == '\r' && is) {
 275                                 // The Windows support has lead to the
 276                                 // possibility of "\r\n" at the end of
 277                                 // a line.  This will stop LyX choking
 278                                 // when it expected to find a '\n'
 279                                 is.get(cc);
 280                                 c = cc;
 281                         }
 282
 283                         if (c == '\n')
 284                                 ++lineno;
 285
 286                 }
 287                 if (status) return true;
 288
 289                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 290                 buff.clear();
 291                 return false;
 292         } else {
 293                 unsigned char c = 0; // getc() returns an int
 294                 char cc = 0;
 295
 296                 status = 0;
 297                 while (is && !status) {
 298                         is.get(cc);
 299                         c = cc;
 300
 301                         // skip ','s
 302                         if (c == ',') continue;
 303
 304                         if (c == '\\') {
 305                                 // escape
 306                                 buff.clear();
 307
 308                                 do {
 309                                         if (c == '\\') {
 310                                                 // escape the next char
 311                                                 is.get(cc);
 312                                                 c = cc;
 313                                         }
 314                                         buff.push_back(c);
 315                                         is.get(cc);
 316                                         c = cc;
 317                                 } while (c > ' ' && c != ',' && is);
 318
 319                                 status = LEX_TOKEN;
 320                                 continue;
 321                         }
 322
 323                         if (c == commentChar) {
 324                                 // Read rest of line (fast :-)
 325 #if 1
 326                                 // That is still not fast... (Lgb)
 327                                 string dummy;
 328                                 getline(is, dummy);
 329
 330                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 331                                                       << dummy << '\'' << endl;
 332 #else
 333                                 // but ignore is also still buggy (Lgb)
 334                                 // This is fast (Lgb)
 335                                 is.ignore(100, '\n');
 336 #endif
 337                                 ++lineno;
 338                                 continue;
 339                         }
 340
 341                         // string
 342                         if (c == '\"') {
 343                                 buff.clear();
 344
 345                                 bool escaped = false;
 346                                 do {
 347                                         escaped = false;
 348                                         is.get(cc);
 349                                         c = cc;
 350                                         if (c == '\r') continue;
 351                                         if (c == '\\') {
 352                                                 // escape the next char
 353                                                 is.get(cc);
 354                                                 c = cc;
 355                                                 if (c == '\"' || c == '\\')
 356                                                         escaped = true;
 357                                                 else
 358                                                         buff.push_back('\\');
 359                                         }
 360                                         buff.push_back(c);
 361
 362                                         if (!escaped && c == '\"') break;
 363                                 } while (c != '\n' && is);
 364
 365                                 if (c != '\"') {
 366                                         printError("Missing quote");
 367                                         if (c == '\n')
 368                                                 ++lineno;
 369                                 }
 370
 371                                 buff.pop_back();
 372                                 status = LEX_DATA;
 373                                 break;
 374                         }
 375
 376                         if (c > ' ' && is) {
 377                                 buff.clear();
 378
 379                                 do {
 380                                         if (c == '\\') {
 381                                                 // escape the next char
 382                                                 is.get(cc);
 383                                                 c = cc;
 384                                                 //escaped = true;
 385                                         }
 386                                         buff.push_back(c);
 387                                         is.get(cc);
 388                                         c = cc;
 389                                 } while (c > ' ' && c != ',' && is);
 390
 391                                 status = LEX_TOKEN;
 392                         }
 393                         // new line
 394                         if (c == '\n')
 395                                 ++lineno;
 396                 }
 397
 398                 if (status) return true;
 399
 400                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 401                 buff.clear();
 402                 return false;
 403         }
 404 }
 405
 406
 407 int LyXLex::Pimpl::search_kw(char const * const tag) const
 408 {
 409         keyword_item search_tag = { tag, 0 };
 410         keyword_item * res =
 411                 lower_bound(table, table + no_items,
 412                             search_tag, compare_tags());
 413         // use the compare_ascii_no_case instead of compare_no_case,
 414         // because in turkish, 'i' is not the lowercase version of 'I',
 415         // and thus turkish locale breaks parsing of tags.
 416         if (res != table + no_items
 417             && !compare_ascii_no_case(res->tag, tag))
 418                 return res->code;
 419         return LEX_UNDEF;
 420 }
 421
 422
 423 int LyXLex::Pimpl::lex()
 424 {
 425         //NOTE: possible bug.
 426         if (next() && status == LEX_TOKEN) {
 427                 return search_kw(getString().c_str());
 428         } else
 429                 return status;
 430 }
 431
 432
 433 bool LyXLex::Pimpl::eatLine()
 434 {
 435         buff.clear();
 436
 437         unsigned char c = '\0';
 438         char cc = 0;
 439         while (is && c != '\n') {
 440                 is.get(cc);
 441                 c = cc;
 442                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 443                 //                    << c << '\'' << endl;
 444                 if (c != '\r')
 445                         buff.push_back(c);
 446         }
 447
 448         if (c == '\n') {
 449                 ++lineno;
 450                 buff.pop_back();
 451                 status = LEX_DATA;
 452                 return true;
 453         } else {
 454                 return false;
 455         }
 456 }
 457
 458
 459 bool LyXLex::Pimpl::nextToken()
 460 {
 461         if (!pushTok.empty()) {
 462                 // There can have been a whole line pushed so
 463                 // we extract the first word and leaves the rest
 464                 // in pushTok. (Lgb)
 465                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 466                         string tmp;
 467                         pushTok = split(pushTok, tmp, ' ');
 468                         buff.assign(tmp.begin(), tmp.end());
 469                         return true;
 470                 } else {
 471                         buff.assign(pushTok.begin(), pushTok.end());
 472                         pushTok.erase();
 473                         return true;
 474                 }
 475         }
 476
 477         status = 0;
 478         while (is && !status) {
 479                 unsigned char c = 0;
 480                 char cc = 0;
 481                 is.get(cc);
 482                 c = cc;
 483                 if (c >= ' ' && is) {
 484                         buff.clear();
 485
 486                         if (c == '\\') { // first char == '\\'
 487                                 do {
 488                                         buff.push_back(c);
 489                                         is.get(cc);
 490                                         c = cc;
 491                                 } while (c > ' ' && c != '\\' && is);
 492                         } else {
 493                                 do {
 494                                         buff.push_back(c);
 495                                         is.get(cc);
 496                                         c = cc;
 497                                 } while (c >= ' ' && c != '\\' && is);
 498                         }
 499
 500                         if (c == '\\') is.putback(c); // put it back
 501                         status = LEX_TOKEN;
 502                 }
 503
 504                 if (c == '\n')
 505                         ++lineno;
 506
 507         }
 508         if (status)  return true;
 509
 510         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 511         buff.clear();
 512         return false;
 513 }
 514
 515
 516 void LyXLex::Pimpl::pushToken(string const & pt)
 517 {
 518         pushTok = pt;
 519 }