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