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 #include "debug.h"
  17
  18 #include "support/lyxalgo.h"
  19 #include "support/filetools.h"
  20 #include "support/lstrings.h"
  21
  22 #include <algorithm>
  23
  24 using namespace lyx::support;
  25
  26 using std::sort;
  27 using std::ostream;
  28 using std::ios;
  29 using std::istream;
  30 using std::endl;
  31 using std::lower_bound;
  32 using std::vector;
  33 using std::getline;
  34
  35 // namespace {
  36 struct compare_tags {
  37         // used by lower_bound, sort and sorted
  38         inline
  39         int operator()(keyword_item const & a, keyword_item const & b) const {
  40                 // we use the ascii version, because in turkish, 'i'
  41                 // is not the lowercase version of 'I', and thus
  42                 // turkish locale breaks parsing of tags.
  43                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  44         }
  45 };
  46 // } // end of anon namespace
  47
  48
  49 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  50         : is(&fb__), table(tab), no_items(num),
  51           status(0), lineno(0), commentChar('#')
  52 {
  53         verifyTable();
  54 }
  55
  56
  57 string const LyXLex::Pimpl::getString() const
  58 {
  59         return string(buff.begin(), buff.end());
  60 }
  61
  62
  63 void LyXLex::Pimpl::printError(string const & message) const
  64 {
  65         string const tmpmsg = subst(message, "$$Token", getString());
  66         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  67                << " of file " << MakeDisplayPath(name) << ']' << endl;
  68 }
  69
  70
  71 void LyXLex::Pimpl::printTable(ostream & os)
  72 {
  73         os << "\nNumber of tags: " << no_items << endl;
  74         for (int i= 0; i < no_items; ++i)
  75                 os << "table[" << i
  76                    << "]:  tag: `" << table[i].tag
  77                    << "'  code:" << table[i].code << '\n';
  78         os.flush();
  79 }
  80
  81
  82 void LyXLex::Pimpl::verifyTable()
  83 {
  84         // Check if the table is sorted and if not, sort it.
  85         if (table
  86             && !lyx::sorted(table, table + no_items, compare_tags())) {
  87                 lyxerr << "The table passed to LyXLex is not sorted!\n"
  88                        << "Tell the developers to fix it!" << endl;
  89                 // We sort it anyway to avoid problems.
  90                 lyxerr << "\nUnsorted:" << endl;
  91                 printTable(lyxerr);
  92
  93                 sort(table, table + no_items, compare_tags());
  94                 lyxerr << "\nSorted:" << endl;
  95                 printTable(lyxerr);
  96         }
  97 }
  98
  99
 100 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
 101 {
 102         pushed_table tmppu(table, no_items);
 103         pushed.push(tmppu);
 104
 105         table = tab;
 106         no_items = num;
 107
 108         verifyTable();
 109 }
 110
 111
 112 void LyXLex::Pimpl::popTable()
 113 {
 114         if (pushed.empty()) {
 115                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 116                 return;
 117         }
 118
 119         pushed_table tmp = pushed.top();
 120         pushed.pop();
 121         table = tmp.table_elem;
 122         no_items = tmp.table_siz;
 123 }
 124
 125
 126 bool LyXLex::Pimpl::setFile(string const & filename)
 127 {
 128
 129         // Check the format of the file.
 130         string const format = getExtFromContents(filename);
 131
 132         if (format == "gzip" || format == "zip" || format == "compress") {
 133                 lyxerr << "lyxlex: compressed" << endl;
 134
 135                 // The check only outputs a debug message, because it triggers
 136                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 137                 // a fresh new filebuf.  (JMarc)
 138                 if (gz__.is_open() || is.tellg() > 0)
 139                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 140                                 "file or stream already set." << endl;
 141                 gz__.open(filename.c_str(), ios::in);
 142                 is.rdbuf(&gz__);
 143                 name = filename;
 144                 lineno = 0;
 145                 return gz__.is_open() && is.good();
 146         } else {
 147                 lyxerr << "lyxlex: UNcompressed" << endl;
 148
 149                 // The check only outputs a debug message, because it triggers
 150                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 151                 // a fresh new filebuf.  (JMarc)
 152                 if (fb__.is_open() || is.tellg() > 0)
 153                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 154                                 "file or stream already set." << endl;
 155                 fb__.open(filename.c_str(), ios::in);
 156                 is.rdbuf(&fb__);
 157                 name = filename;
 158                 lineno = 0;
 159                 return fb__.is_open() && is.good();
 160         }
 161 }
 162
 163
 164 void LyXLex::Pimpl::setStream(istream & i)
 165 {
 166         if (fb__.is_open() || is.tellg() > 0)
 167                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 168                         "file or stream already set." << endl;
 169         is.rdbuf(i.rdbuf());
 170         lineno = 0;
 171 }
 172
 173
 174 void LyXLex::Pimpl::setCommentChar(char c)
 175 {
 176         commentChar = c;
 177 }
 178
 179
 180 bool LyXLex::Pimpl::next(bool esc /* = false */)
 181 {
 182         if (!pushTok.empty()) {
 183                 // There can have been a whole line pushed so
 184                 // we extract the first word and leaves the rest
 185                 // in pushTok. (Lgb)
 186                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 187                         string tmp;
 188                         pushTok = split(pushTok, tmp, ' ');
 189                         buff.assign(tmp.begin(), tmp.end());
 190                         return true;
 191                 } else {
 192                         buff.assign(pushTok.begin(), pushTok.end());
 193                         pushTok.erase();
 194                         return true;
 195                 }
 196         }
 197         if (!esc) {
 198                 unsigned char c = 0; // getc() returns an int
 199                 char cc = 0;
 200                 status = 0;
 201                 while (is && !status) {
 202                         is.get(cc);
 203                         c = cc;
 204                         if (c == commentChar) {
 205                                 // Read rest of line (fast :-)
 206 #if 1
 207                                 // That is not fast... (Lgb)
 208                                 string dummy;
 209                                 getline(is, dummy);
 210
 211                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 212                                                       << dummy << '\'' << endl;
 213 #else
 214                                 // unfortunately ignore is buggy (Lgb)
 215                                 is.ignore(100, '\n');
 216 #endif
 217                                 ++lineno;
 218                                 continue;
 219                         }
 220
 221                         if (c == '\"') {
 222                                 buff.clear();
 223
 224                                 do {
 225                                         is.get(cc);
 226                                         c = cc;
 227                                         if (c != '\r')
 228                                                 buff.push_back(c);
 229                                 } while (c != '\"' && c != '\n' && is);
 230
 231                                 if (c != '\"') {
 232                                         printError("Missing quote");
 233                                         if (c == '\n')
 234                                                 ++lineno;
 235                                 }
 236
 237                                 buff.pop_back();
 238                                 status = LEX_DATA;
 239                                 break;
 240                         }
 241
 242                         if (c == ',')
 243                                 continue;              /* Skip ','s */
 244
 245                                 // using relational operators with chars other
 246                                 // than == and != is not safe. And if it is done
 247                                 // the type _have_ to be unsigned. It usually a
 248                                 // lot better to use the functions from cctype
 249                         if (c > ' ' && is)  {
 250                                 buff.clear();
 251
 252                                 do {
 253                                         buff.push_back(c);
 254                                         is.get(cc);
 255                                         c = cc;
 256                                 } while (c > ' ' && c != ',' && is);
 257
 258                                 status = LEX_TOKEN;
 259                         }
 260
 261                         if (c == '\r' && is) {
 262                                 // The Windows support has lead to the
 263                                 // possibility of "\r\n" at the end of
 264                                 // a line.  This will stop LyX choking
 265                                 // when it expected to find a '\n'
 266                                 is.get(cc);
 267                                 c = cc;
 268                         }
 269
 270                         if (c == '\n')
 271                                 ++lineno;
 272
 273                 }
 274                 if (status) return true;
 275
 276                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 277                 buff.clear();
 278                 return false;
 279         } else {
 280                 unsigned char c = 0; // getc() returns an int
 281                 char cc = 0;
 282
 283                 status = 0;
 284                 while (is && !status) {
 285                         is.get(cc);
 286                         c = cc;
 287
 288                         // skip ','s
 289                         if (c == ',') continue;
 290
 291                         if (c == '\\') {
 292                                 // escape
 293                                 buff.clear();
 294
 295                                 do {
 296                                         if (c == '\\') {
 297                                                 // escape the next char
 298                                                 is.get(cc);
 299                                                 c = cc;
 300                                         }
 301                                         buff.push_back(c);
 302                                         is.get(cc);
 303                                         c = cc;
 304                                 } while (c > ' ' && c != ',' && is);
 305
 306                                 status = LEX_TOKEN;
 307                                 continue;
 308                         }
 309
 310                         if (c == commentChar) {
 311                                 // Read rest of line (fast :-)
 312 #if 1
 313                                 // That is still not fast... (Lgb)
 314                                 string dummy;
 315                                 getline(is, dummy);
 316
 317                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 318                                                       << dummy << '\'' << endl;
 319 #else
 320                                 // but ignore is also still buggy (Lgb)
 321                                 // This is fast (Lgb)
 322                                 is.ignore(100, '\n');
 323 #endif
 324                                 ++lineno;
 325                                 continue;
 326                         }
 327
 328                         // string
 329                         if (c == '\"') {
 330                                 buff.clear();
 331
 332                                 bool escaped = false;
 333                                 do {
 334                                         escaped = false;
 335                                         is.get(cc);
 336                                         c = cc;
 337                                         if (c == '\r') continue;
 338                                         if (c == '\\') {
 339                                                 // escape the next char
 340                                                 is.get(cc);
 341                                                 c = cc;
 342                                                 if (c == '\"' || c == '\\')
 343                                                         escaped = true;
 344                                                 else
 345                                                         buff.push_back('\\');
 346                                         }
 347                                         buff.push_back(c);
 348
 349                                         if (!escaped && c == '\"') break;
 350                                 } while (c != '\n' && is);
 351
 352                                 if (c != '\"') {
 353                                         printError("Missing quote");
 354                                         if (c == '\n')
 355                                                 ++lineno;
 356                                 }
 357
 358                                 buff.pop_back();
 359                                 status = LEX_DATA;
 360                                 break;
 361                         }
 362
 363                         if (c > ' ' && is) {
 364                                 buff.clear();
 365
 366                                 do {
 367                                         if (c == '\\') {
 368                                                 // escape the next char
 369                                                 is.get(cc);
 370                                                 c = cc;
 371                                                 //escaped = true;
 372                                         }
 373                                         buff.push_back(c);
 374                                         is.get(cc);
 375                                         c = cc;
 376                                 } while (c > ' ' && c != ',' && is);
 377
 378                                 status = LEX_TOKEN;
 379                         }
 380                         // new line
 381                         if (c == '\n')
 382                                 ++lineno;
 383                 }
 384
 385                 if (status) return true;
 386
 387                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 388                 buff.clear();
 389                 return false;
 390         }
 391 }
 392
 393
 394 int LyXLex::Pimpl::search_kw(char const * const tag) const
 395 {
 396         keyword_item search_tag = { tag, 0 };
 397         keyword_item * res =
 398                 lower_bound(table, table + no_items,
 399                             search_tag, compare_tags());
 400         // use the compare_ascii_no_case instead of compare_no_case,
 401         // because in turkish, 'i' is not the lowercase version of 'I',
 402         // and thus turkish locale breaks parsing of tags.
 403         if (res != table + no_items
 404             && !compare_ascii_no_case(res->tag, tag))
 405                 return res->code;
 406         return LEX_UNDEF;
 407 }
 408
 409
 410 int LyXLex::Pimpl::lex()
 411 {
 412         //NOTE: possible bug.
 413         if (next() && status == LEX_TOKEN) {
 414                 return search_kw(getString().c_str());
 415         } else
 416                 return status;
 417 }
 418
 419
 420 bool LyXLex::Pimpl::eatLine()
 421 {
 422         buff.clear();
 423
 424         unsigned char c = '\0';
 425         char cc = 0;
 426         while (is && c != '\n') {
 427                 is.get(cc);
 428                 c = cc;
 429                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 430                 //                    << c << '\'' << endl;
 431                 if (c != '\r')
 432                         buff.push_back(c);
 433         }
 434
 435         if (c == '\n') {
 436                 ++lineno;
 437                 buff.pop_back();
 438                 status = LEX_DATA;
 439                 return true;
 440         } else {
 441                 return false;
 442         }
 443 }
 444
 445
 446 bool LyXLex::Pimpl::nextToken()
 447 {
 448         if (!pushTok.empty()) {
 449                 // There can have been a whole line pushed so
 450                 // we extract the first word and leaves the rest
 451                 // in pushTok. (Lgb)
 452                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 453                         string tmp;
 454                         pushTok = split(pushTok, tmp, ' ');
 455                         buff.assign(tmp.begin(), tmp.end());
 456                         return true;
 457                 } else {
 458                         buff.assign(pushTok.begin(), pushTok.end());
 459                         pushTok.erase();
 460                         return true;
 461                 }
 462         }
 463
 464         status = 0;
 465         while (is && !status) {
 466                 unsigned char c = 0;
 467                 char cc = 0;
 468                 is.get(cc);
 469                 c = cc;
 470                 if (c >= ' ' && is) {
 471                         buff.clear();
 472
 473                         if (c == '\\') { // first char == '\\'
 474                                 do {
 475                                         buff.push_back(c);
 476                                         is.get(cc);
 477                                         c = cc;
 478                                 } while (c > ' ' && c != '\\' && is);
 479                         } else {
 480                                 do {
 481                                         buff.push_back(c);
 482                                         is.get(cc);
 483                                         c = cc;
 484                                 } while (c >= ' ' && c != '\\' && is);
 485                         }
 486
 487                         if (c == '\\') is.putback(c); // put it back
 488                         status = LEX_TOKEN;
 489                 }
 490
 491                 if (c == '\n')
 492                         ++lineno;
 493
 494         }
 495         if (status)  return true;
 496
 497         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 498         buff.clear();
 499         return false;
 500 }
 501
 502
 503 void LyXLex::Pimpl::pushToken(string const & pt)
 504 {
 505         pushTok = pt;
 506 }