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 #include "support/types.h"
  23 #include "support/unicode.h"
  24
  25 #include <functional>
  26
  27
  28 namespace lyx {
  29
  30 using support::compare_ascii_no_case;
  31 using support::FileName;
  32 using support::getFormatFromContents;
  33 using support::makeDisplayPath;
  34 using support::split;
  35 using support::subst;
  36
  37 using std::endl;
  38 using std::getline;
  39 using std::lower_bound;
  40 using std::sort;
  41 using std::string;
  42 using std::ios;
  43 using std::istream;
  44 using std::ostream;
  45
  46 namespace {
  47
  48 class compare_tags
  49         : public std::binary_function<keyword_item, keyword_item, bool> {
  50 public:
  51         // used by lower_bound, sort and sorted
  52         bool operator()(keyword_item const & a, keyword_item const & b) const
  53         {
  54                 // we use the ascii version, because in turkish, 'i'
  55                 // is not the lowercase version of 'I', and thus
  56                 // turkish locale breaks parsing of tags.
  57                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  58         }
  59 };
  60
  61 } // end of anon namespace
  62
  63
  64 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  65         : is(&fb_), table(tab), no_items(num),
  66           status(0), lineno(0), commentChar('#')
  67 {
  68         verifyTable();
  69 }
  70
  71
  72 string const LyXLex::Pimpl::getString() const
  73 {
  74         return buff;
  75 }
  76
  77
  78 docstring const LyXLex::Pimpl::getDocString() const
  79 {
  80         return from_utf8(buff);
  81 }
  82
  83
  84 void LyXLex::Pimpl::printError(string const & message) const
  85 {
  86         string const tmpmsg = subst(message, "$$Token", getString());
  87         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  88                 << " of file " << to_utf8(makeDisplayPath(name)) << ']' << endl;
  89 }
  90
  91
  92 void LyXLex::Pimpl::printTable(ostream & os)
  93 {
  94         os << "\nNumber of tags: " << no_items << endl;
  95         for (int i= 0; i < no_items; ++i)
  96                 os << "table[" << i
  97                    << "]:  tag: `" << table[i].tag
  98                    << "'  code:" << table[i].code << '\n';
  99         os.flush();
 100 }
 101
 102
 103 void LyXLex::Pimpl::verifyTable()
 104 {
 105         // Check if the table is sorted and if not, sort it.
 106         if (table
 107             && !lyx::sorted(table, table + no_items, compare_tags())) {
 108                 lyxerr << "The table passed to LyXLex is not sorted!\n"
 109                        << "Tell the developers to fix it!" << endl;
 110                 // We sort it anyway to avoid problems.
 111                 lyxerr << "\nUnsorted:" << endl;
 112                 printTable(lyxerr);
 113
 114                 sort(table, table + no_items, compare_tags());
 115                 lyxerr << "\nSorted:" << endl;
 116                 printTable(lyxerr);
 117         }
 118 }
 119
 120
 121 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
 122 {
 123         pushed_table tmppu(table, no_items);
 124         pushed.push(tmppu);
 125
 126         table = tab;
 127         no_items = num;
 128
 129         verifyTable();
 130 }
 131
 132
 133 void LyXLex::Pimpl::popTable()
 134 {
 135         if (pushed.empty()) {
 136                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 137                 return;
 138         }
 139
 140         pushed_table tmp = pushed.top();
 141         pushed.pop();
 142         table = tmp.table_elem;
 143         no_items = tmp.table_siz;
 144 }
 145
 146
 147 bool LyXLex::Pimpl::setFile(FileName const & filename)
 148 {
 149         // Check the format of the file.
 150         string const format = getFormatFromContents(filename);
 151
 152         if (format == "gzip" || format == "zip" || format == "compress") {
 153                 lyxerr[Debug::LYXLEX] << "lyxlex: compressed" << 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 (!gz_.empty() || istream::off_type(is.tellg()) > -1)
 159                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 160                                 "file or stream already set." << endl;
 161                 gz_.push(io::gzip_decompressor());
 162                 gz_.push(io::file_source(filename.toFilesystemEncoding()));
 163                 is.rdbuf(&gz_);
 164                 name = filename.absFilename();
 165                 lineno = 0;
 166                 return gz_.component<io::file_source>(1)->is_open() && is.good();
 167         } else {
 168                 lyxerr[Debug::LYXLEX] << "lyxlex: UNcompressed" << endl;
 169
 170                 // The check only outputs a debug message, because it triggers
 171                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 172                 // a fresh new filebuf.  (JMarc)
 173                 if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 174                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 175                                 "file or stream already set." << endl;
 176                 fb_.open(filename.toFilesystemEncoding().c_str(), ios::in);
 177                 is.rdbuf(&fb_);
 178                 name = filename.absFilename();
 179                 lineno = 0;
 180                 return fb_.is_open() && is.good();
 181         }
 182 }
 183
 184
 185 void LyXLex::Pimpl::setStream(istream & i)
 186 {
 187         if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 188                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 189                         "file or stream already set." << endl;
 190         is.rdbuf(i.rdbuf());
 191         lineno = 0;
 192 }
 193
 194
 195 void LyXLex::Pimpl::setCommentChar(char c)
 196 {
 197         commentChar = c;
 198 }
 199
 200
 201 bool LyXLex::Pimpl::next(bool esc /* = false */)
 202 {
 203         if (!pushTok.empty()) {
 204                 // There can have been a whole line pushed so
 205                 // we extract the first word and leaves the rest
 206                 // in pushTok. (Lgb)
 207                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 208                         buff.clear();
 209                         pushTok = split(pushTok, buff, ' ');
 210                         return true;
 211                 } else {
 212                         buff = pushTok;
 213                         pushTok.clear();
 214                         return true;
 215                 }
 216         }
 217         if (!esc) {
 218                 unsigned char c = 0; // getc() returns an int
 219                 char cc = 0;
 220                 status = 0;
 221                 while (is && !status) {
 222                         is.get(cc);
 223                         c = cc;
 224                         if (c == commentChar) {
 225                                 // Read rest of line (fast :-)
 226 #if 1
 227                                 // That is not fast... (Lgb)
 228                                 string dummy;
 229                                 getline(is, dummy);
 230
 231                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 232                                                       << dummy << '\'' << endl;
 233 #else
 234                                 // unfortunately ignore is buggy (Lgb)
 235                                 is.ignore(100, '\n');
 236 #endif
 237                                 ++lineno;
 238                                 continue;
 239                         }
 240
 241                         if (c == '\"') {
 242                                 buff.clear();
 243
 244                                 do {
 245                                         is.get(cc);
 246                                         c = cc;
 247                                         if (c != '\r')
 248                                                 buff.push_back(c);
 249                                 } while (c != '\"' && c != '\n' && is);
 250
 251                                 if (c != '\"') {
 252                                         printError("Missing quote");
 253                                         if (c == '\n')
 254                                                 ++lineno;
 255                                 }
 256
 257                                 buff.resize(buff.size()-1);
 258                                 status = LEX_DATA;
 259                                 break;
 260                         }
 261
 262                         if (c == ',')
 263                                 continue;              /* Skip ','s */
 264
 265                                 // using relational operators with chars other
 266                                 // than == and != is not safe. And if it is done
 267                                 // the type _have_ to be unsigned. It usually a
 268                                 // lot better to use the functions from cctype
 269                         if (c > ' ' && is)  {
 270                                 buff.clear();
 271
 272                                 do {
 273                                         buff.push_back(c);
 274                                         is.get(cc);
 275                                         c = cc;
 276                                 } while (c > ' ' && c != ',' && is);
 277
 278                                 status = LEX_TOKEN;
 279                         }
 280
 281                         if (c == '\r' && is) {
 282                                 // The Windows support has lead to the
 283                                 // possibility of "\r\n" at the end of
 284                                 // a line.  This will stop LyX choking
 285                                 // when it expected to find a '\n'
 286                                 is.get(cc);
 287                                 c = cc;
 288                         }
 289
 290                         if (c == '\n')
 291                                 ++lineno;
 292
 293                 }
 294                 if (status) return true;
 295
 296                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 297                 buff.clear();
 298                 return false;
 299         } else {
 300                 unsigned char c = 0; // getc() returns an int
 301                 char cc = 0;
 302
 303                 status = 0;
 304                 while (is && !status) {
 305                         is.get(cc);
 306                         c = cc;
 307
 308                         // skip ','s
 309                         if (c == ',') continue;
 310
 311                         if (c == '\\') {
 312                                 // escape
 313                                 buff.clear();
 314
 315                                 do {
 316                                         if (c == '\\') {
 317                                                 // escape the next char
 318                                                 is.get(cc);
 319                                                 c = cc;
 320                                         }
 321                                         buff.push_back(c);
 322                                         is.get(cc);
 323                                         c = cc;
 324                                 } while (c > ' ' && c != ',' && is);
 325
 326                                 status = LEX_TOKEN;
 327                                 continue;
 328                         }
 329
 330                         if (c == commentChar) {
 331                                 // Read rest of line (fast :-)
 332 #if 1
 333                                 // That is still not fast... (Lgb)
 334                                 string dummy;
 335                                 getline(is, dummy);
 336
 337                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 338                                                       << dummy << '\'' << endl;
 339 #else
 340                                 // but ignore is also still buggy (Lgb)
 341                                 // This is fast (Lgb)
 342                                 is.ignore(100, '\n');
 343 #endif
 344                                 ++lineno;
 345                                 continue;
 346                         }
 347
 348                         // string
 349                         if (c == '\"') {
 350                                 buff.clear();
 351
 352                                 bool escaped = false;
 353                                 do {
 354                                         escaped = false;
 355                                         is.get(cc);
 356                                         c = cc;
 357                                         if (c == '\r') continue;
 358                                         if (c == '\\') {
 359                                                 // escape the next char
 360                                                 is.get(cc);
 361                                                 c = cc;
 362                                                 if (c == '\"' || c == '\\')
 363                                                         escaped = true;
 364                                                 else
 365                                                         buff.push_back('\\');
 366                                         }
 367                                         buff.push_back(c);
 368
 369                                         if (!escaped && c == '\"') break;
 370                                 } while (c != '\n' && is);
 371
 372                                 if (c != '\"') {
 373                                         printError("Missing quote");
 374                                         if (c == '\n')
 375                                                 ++lineno;
 376                                 }
 377
 378                                 buff.resize(buff.size() -1);
 379                                 status = LEX_DATA;
 380                                 break;
 381                         }
 382
 383                         if (c > ' ' && is) {
 384                                 buff.clear();
 385
 386                                 do {
 387                                         if (c == '\\') {
 388                                                 // escape the next char
 389                                                 is.get(cc);
 390                                                 c = cc;
 391                                                 //escaped = true;
 392                                         }
 393                                         buff.push_back(c);
 394                                         is.get(cc);
 395                                         c = cc;
 396                                 } while (c > ' ' && c != ',' && is);
 397
 398                                 status = LEX_TOKEN;
 399                         }
 400                         // new line
 401                         if (c == '\n')
 402                                 ++lineno;
 403                 }
 404
 405                 if (status) return true;
 406
 407                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 408                 buff.clear();
 409                 return false;
 410         }
 411 }
 412
 413
 414 int LyXLex::Pimpl::search_kw(char const * const tag) const
 415 {
 416         keyword_item search_tag = { tag, 0 };
 417         keyword_item * res =
 418                 lower_bound(table, table + no_items,
 419                             search_tag, compare_tags());
 420         // use the compare_ascii_no_case instead of compare_no_case,
 421         // because in turkish, 'i' is not the lowercase version of 'I',
 422         // and thus turkish locale breaks parsing of tags.
 423         if (res != table + no_items
 424             && !compare_ascii_no_case(res->tag, tag))
 425                 return res->code;
 426         return LEX_UNDEF;
 427 }
 428
 429
 430 int LyXLex::Pimpl::lex()
 431 {
 432         //NOTE: possible bug.
 433         if (next() && status == LEX_TOKEN) {
 434                 return search_kw(getString().c_str());
 435         } else
 436                 return status;
 437 }
 438
 439
 440 bool LyXLex::Pimpl::eatLine()
 441 {
 442         buff.clear();
 443
 444         unsigned char c = '\0';
 445         char cc = 0;
 446         while (is && c != '\n') {
 447                 is.get(cc);
 448                 c = cc;
 449                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 450                 //                    << c << '\'' << endl;
 451                 if (c != '\r')
 452                         buff.push_back(c);
 453         }
 454
 455         if (c == '\n') {
 456                 ++lineno;
 457                 buff.resize(buff.size() - 1);
 458                 status = LEX_DATA;
 459                 return true;
 460         } else {
 461                 return false;
 462         }
 463 }
 464
 465
 466 bool LyXLex::Pimpl::nextToken()
 467 {
 468         if (!pushTok.empty()) {
 469                 // There can have been a whole line pushed so
 470                 // we extract the first word and leaves the rest
 471                 // in pushTok. (Lgb)
 472                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 473                         buff.clear();
 474                         pushTok = split(pushTok, buff, ' ');
 475                         return true;
 476                 } else {
 477                         buff = pushTok;
 478                         pushTok.clear();
 479                         return true;
 480                 }
 481         }
 482
 483         status = 0;
 484         while (is && !status) {
 485                 unsigned char c = 0;
 486                 char cc = 0;
 487                 is.get(cc);
 488                 c = cc;
 489                 if (c >= ' ' && is) {
 490                         buff.clear();
 491
 492                         if (c == '\\') { // first char == '\\'
 493                                 do {
 494                                         buff.push_back(c);
 495                                         is.get(cc);
 496                                         c = cc;
 497                                 } while (c > ' ' && c != '\\' && is);
 498                         } else {
 499                                 do {
 500                                         buff.push_back(c);
 501                                         is.get(cc);
 502                                         c = cc;
 503                                 } while (c >= ' ' && c != '\\' && is);
 504                         }
 505
 506                         if (c == '\\') is.putback(c); // put it back
 507                         status = LEX_TOKEN;
 508                 }
 509
 510                 if (c == '\n')
 511                         ++lineno;
 512
 513         }
 514         if (status)  return true;
 515
 516         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 517         buff.clear();
 518         return false;
 519 }
 520
 521
 522 void LyXLex::Pimpl::pushToken(string const & pt)
 523 {
 524         pushTok = pt;
 525 }
 526
 527
 528 } // namespace lyx