src/lyxlex_pimpl.C

   1 #include <config.h>
   2
   3 #include "lyxlex_pimpl.h"
   4 #include "debug.h"
   5
   6 #include "support/lyxalgo.h"
   7 #include "support/filetools.h"
   8 #include "support/lstrings.h"
   9
  10 #include <algorithm>
  11
  12 using namespace lyx::support;
  13
  14 using std::sort;
  15 using std::ostream;
  16 using std::ios;
  17 using std::istream;
  18 using std::endl;
  19 using std::lower_bound;
  20 using std::vector;
  21 using std::getline;
  22
  23 // namespace {
  24 struct compare_tags {
  25         // used by lower_bound, sort and sorted
  26         inline
  27         int operator()(keyword_item const & a, keyword_item const & b) const {
  28                 // we use the ascii version, because in turkish, 'i'
  29                 // is not the lowercase version of 'I', and thus
  30                 // turkish locale breaks parsing of tags.
  31                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  32         }
  33 };
  34 // } // end of anon namespace
  35
  36
  37 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  38         : is(&fb__), table(tab), no_items(num),
  39           status(0), lineno(0), commentChar('#')
  40 {
  41         verifyTable();
  42 }
  43
  44
  45 string const LyXLex::Pimpl::getString() const
  46 {
  47         return string(buff.begin(), buff.end());
  48 }
  49
  50
  51 void LyXLex::Pimpl::printError(string const & message) const
  52 {
  53         string const tmpmsg = subst(message, "$$Token", getString());
  54         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  55                << " of file " << MakeDisplayPath(name) << ']' << endl;
  56 }
  57
  58
  59 void LyXLex::Pimpl::printTable(ostream & os)
  60 {
  61         os << "\nNumber of tags: " << no_items << '\n';
  62         for (int i= 0; i < no_items; ++i)
  63                 os << "table[" << i
  64                    << "]:  tag: `" << table[i].tag
  65                    << "'  code:" << table[i].code << '\n';
  66         os.flush();
  67 }
  68
  69
  70 void LyXLex::Pimpl::verifyTable()
  71 {
  72         // Check if the table is sorted and if not, sort it.
  73         if (table
  74             && !lyx::sorted(table, table + no_items, compare_tags())) {
  75                 lyxerr << "The table passed to LyXLex is not sorted!\n"
  76                        << "Tell the developers to fix it!" << endl;
  77                 // We sort it anyway to avoid problems.
  78                 lyxerr << "\nUnsorted:\n";
  79                 printTable(lyxerr);
  80
  81                 sort(table, table + no_items, compare_tags());
  82                 lyxerr << "\nSorted:\n";
  83                 printTable(lyxerr);
  84         }
  85 }
  86
  87
  88 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
  89 {
  90         pushed_table tmppu(table, no_items);
  91         pushed.push(tmppu);
  92
  93         table = tab;
  94         no_items = num;
  95
  96         verifyTable();
  97 }
  98
  99
 100 void LyXLex::Pimpl::popTable()
 101 {
 102         if (pushed.empty()) {
 103                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 104                 return;
 105         }
 106
 107         pushed_table tmp = pushed.top();
 108         pushed.pop();
 109         table = tmp.table_elem;
 110         no_items = tmp.table_siz;
 111 }
 112
 113
 114 bool LyXLex::Pimpl::setFile(string const & filename)
 115 {
 116         // Is this a compressed file or not?
 117         bool const compressed = (filename.substr(filename.size() - 3, 3) == ".gz");
 118
 119         // The check only outputs a debug message, because it triggers
 120         // a bug in compaq cxx 6.2, where is_open() returns 'true' for a
 121         // fresh new filebuf.  (JMarc)
 122         if (compressed) {
 123                 lyxerr << "lyxlex: compressed" << endl;
 124
 125                 if (gz__.is_open() || is.tellg() > 0)
 126                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 127                                 "file or stream already set." << endl;
 128                 gz__.open(filename.c_str(), ios::in);
 129                 is.rdbuf(&gz__);
 130                 name = filename;
 131                 lineno = 0;
 132                 return gz__.is_open() && is.good();
 133         } else {
 134                 lyxerr << "lyxlex: UNcompressed" << endl;
 135
 136                 if (fb__.is_open() || is.tellg() > 0)
 137                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 138                                 "file or stream already set." << endl;
 139                 fb__.open(filename.c_str(), ios::in);
 140                 is.rdbuf(&fb__);
 141                 name = filename;
 142                 lineno = 0;
 143                 return fb__.is_open() && is.good();
 144         }
 145
 146 }
 147
 148
 149 void LyXLex::Pimpl::setStream(istream & i)
 150 {
 151         if (fb__.is_open() || is.tellg() > 0)
 152                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 153                         "file or stream already set." << endl;
 154         is.rdbuf(i.rdbuf());
 155         lineno = 0;
 156 }
 157
 158
 159 void LyXLex::Pimpl::setCommentChar(char c)
 160 {
 161         commentChar = c;
 162 }
 163
 164
 165 bool LyXLex::Pimpl::next(bool esc /* = false */)
 166 {
 167         if (!pushTok.empty()) {
 168                 // There can have been a whole line pushed so
 169                 // we extract the first word and leaves the rest
 170                 // in pushTok. (Lgb)
 171                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 172                         string tmp;
 173                         pushTok = split(pushTok, tmp, ' ');
 174                         buff.assign(tmp.begin(), tmp.end());
 175                         return true;
 176                 } else {
 177                         buff.assign(pushTok.begin(), pushTok.end());
 178                         pushTok.erase();
 179                         return true;
 180                 }
 181         }
 182         if (!esc) {
 183                 unsigned char c = 0; // getc() returns an int
 184                 char cc = 0;
 185                 status = 0;
 186                 while (is && !status) {
 187                         is.get(cc);
 188                         c = cc;
 189                         if (c == commentChar) {
 190                                 // Read rest of line (fast :-)
 191 #if 1
 192                                 // That is not fast... (Lgb)
 193                                 string dummy;
 194                                 getline(is, dummy);
 195
 196                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 197                                                       << dummy << '\'' << endl;
 198 #else
 199                                 // unfortunately ignore is buggy (Lgb)
 200                                 is.ignore(100, '\n');
 201 #endif
 202                                 ++lineno;
 203                                 continue;
 204                         }
 205
 206                         if (c == '\"') {
 207                                 buff.clear();
 208
 209                                 do {
 210                                         is.get(cc);
 211                                         c = cc;
 212                                         if (c != '\r')
 213                                                 buff.push_back(c);
 214                                 } while (c != '\"' && c != '\n' && is);
 215
 216                                 if (c != '\"') {
 217                                         printError("Missing quote");
 218                                         if (c == '\n')
 219                                                 ++lineno;
 220                                 }
 221
 222                                 buff.pop_back();
 223                                 status = LEX_DATA;
 224                                 break;
 225                         }
 226
 227                         if (c == ',')
 228                                 continue;              /* Skip ','s */
 229
 230                                 // using relational operators with chars other
 231                                 // than == and != is not safe. And if it is done
 232                                 // the type _have_ to be unsigned. It usually a
 233                                 // lot better to use the functions from cctype
 234                         if (c > ' ' && is)  {
 235                                 buff.clear();
 236
 237                                 do {
 238                                         buff.push_back(c);
 239                                         is.get(cc);
 240                                         c = cc;
 241                                 } while (c > ' ' && c != ',' && is);
 242
 243                                 status = LEX_TOKEN;
 244                         }
 245
 246                         if (c == '\r' && is) {
 247                                 // The Windows support has lead to the
 248                                 // possibility of "\r\n" at the end of
 249                                 // a line.  This will stop LyX choking
 250                                 // when it expected to find a '\n'
 251                                 is.get(cc);
 252                                 c = cc;
 253                         }
 254
 255                         if (c == '\n')
 256                                 ++lineno;
 257
 258                 }
 259                 if (status) return true;
 260
 261                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 262                 buff.clear();
 263                 return false;
 264         } else {
 265                 unsigned char c = 0; // getc() returns an int
 266                 char cc = 0;
 267
 268                 status = 0;
 269                 while (is && !status) {
 270                         is.get(cc);
 271                         c = cc;
 272
 273                         // skip ','s
 274                         if (c == ',') continue;
 275
 276                         if (c == '\\') {
 277                                 // escape
 278                                 buff.clear();
 279
 280                                 do {
 281                                         if (c == '\\') {
 282                                                 // escape the next char
 283                                                 is.get(cc);
 284                                                 c = cc;
 285                                         }
 286                                         buff.push_back(c);
 287                                         is.get(cc);
 288                                         c = cc;
 289                                 } while (c > ' ' && c != ',' && is);
 290
 291                                 status = LEX_TOKEN;
 292                                 continue;
 293                         }
 294
 295                         if (c == commentChar) {
 296                                 // Read rest of line (fast :-)
 297 #if 1
 298                                 // That is still not fast... (Lgb)
 299                                 string dummy;
 300                                 getline(is, dummy);
 301
 302                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 303                                                       << dummy << '\'' << endl;
 304 #else
 305                                 // but ignore is also still buggy (Lgb)
 306                                 // This is fast (Lgb)
 307                                 is.ignore(100, '\n');
 308 #endif
 309                                 ++lineno;
 310                                 continue;
 311                         }
 312
 313                         // string
 314                         if (c == '\"') {
 315                                 buff.clear();
 316
 317                                 bool escaped = false;
 318                                 do {
 319                                         escaped = false;
 320                                         is.get(cc);
 321                                         c = cc;
 322                                         if (c == '\r') continue;
 323                                         if (c == '\\') {
 324                                                 // escape the next char
 325                                                 is.get(cc);
 326                                                 c = cc;
 327                                                 if (c == '\"' || c == '\\')
 328                                                         escaped = true;
 329                                                 else
 330                                                         buff.push_back('\\');
 331                                         }
 332                                         buff.push_back(c);
 333
 334                                         if (!escaped && c == '\"') break;
 335                                 } while (c != '\n' && is);
 336
 337                                 if (c != '\"') {
 338                                         printError("Missing quote");
 339                                         if (c == '\n')
 340                                                 ++lineno;
 341                                 }
 342
 343                                 buff.pop_back();
 344                                 status = LEX_DATA;
 345                                 break;
 346                         }
 347
 348                         if (c > ' ' && is) {
 349                                 buff.clear();
 350
 351                                 do {
 352                                         if (c == '\\') {
 353                                                 // escape the next char
 354                                                 is.get(cc);
 355                                                 c = cc;
 356                                                 //escaped = true;
 357                                         }
 358                                         buff.push_back(c);
 359                                         is.get(cc);
 360                                         c = cc;
 361                                 } while (c > ' ' && c != ',' && is);
 362
 363                                 status = LEX_TOKEN;
 364                         }
 365                         // new line
 366                         if (c == '\n')
 367                                 ++lineno;
 368                 }
 369
 370                 if (status) return true;
 371
 372                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 373                 buff.clear();
 374                 return false;
 375         }
 376 }
 377
 378
 379 int LyXLex::Pimpl::search_kw(char const * const tag) const
 380 {
 381         keyword_item search_tag = { tag, 0 };
 382         keyword_item * res =
 383                 lower_bound(table, table + no_items,
 384                             search_tag, compare_tags());
 385         // use the compare_ascii_no_case instead of compare_no_case,
 386         // because in turkish, 'i' is not the lowercase version of 'I',
 387         // and thus turkish locale breaks parsing of tags.
 388         if (res != table + no_items
 389             && !compare_ascii_no_case(res->tag, tag))
 390                 return res->code;
 391         return LEX_UNDEF;
 392 }
 393
 394
 395 int LyXLex::Pimpl::lex()
 396 {
 397         //NOTE: possible bug.
 398         if (next() && status == LEX_TOKEN) {
 399                 return search_kw(getString().c_str());
 400         } else
 401                 return status;
 402 }
 403
 404
 405 bool LyXLex::Pimpl::eatLine()
 406 {
 407         buff.clear();
 408
 409         unsigned char c = '\0';
 410         char cc = 0;
 411         while (is && c != '\n') {
 412                 is.get(cc);
 413                 c = cc;
 414                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 415                 //                    << c << '\'' << endl;
 416                 if (c != '\r')
 417                         buff.push_back(c);
 418         }
 419
 420         if (c == '\n') {
 421                 ++lineno;
 422                 buff.pop_back();
 423                 status = LEX_DATA;
 424                 return true;
 425         } else {
 426                 return false;
 427         }
 428 }
 429
 430
 431 bool LyXLex::Pimpl::nextToken()
 432 {
 433         if (!pushTok.empty()) {
 434                 // There can have been a whole line pushed so
 435                 // we extract the first word and leaves the rest
 436                 // in pushTok. (Lgb)
 437                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 438                         string tmp;
 439                         pushTok = split(pushTok, tmp, ' ');
 440                         buff.assign(tmp.begin(), tmp.end());
 441                         return true;
 442                 } else {
 443                         buff.assign(pushTok.begin(), pushTok.end());
 444                         pushTok.erase();
 445                         return true;
 446                 }
 447         }
 448
 449         status = 0;
 450         while (is && !status) {
 451                 unsigned char c = 0;
 452                 char cc = 0;
 453                 is.get(cc);
 454                 c = cc;
 455                 if (c >= ' ' && is) {
 456                         buff.clear();
 457
 458                         if (c == '\\') { // first char == '\\'
 459                                 do {
 460                                         buff.push_back(c);
 461                                         is.get(cc);
 462                                         c = cc;
 463                                 } while (c > ' ' && c != '\\' && is);
 464                         } else {
 465                                 do {
 466                                         buff.push_back(c);
 467                                         is.get(cc);
 468                                         c = cc;
 469                                 } while (c >= ' ' && c != '\\' && is);
 470                         }
 471
 472                         if (c == '\\') is.putback(c); // put it back
 473                         status = LEX_TOKEN;
 474                 }
 475
 476                 if (c == '\n')
 477                         ++lineno;
 478
 479         }
 480         if (status)  return true;
 481
 482         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 483         buff.clear();
 484         return false;
 485 }
 486
 487
 488 void LyXLex::Pimpl::pushToken(string const & pt)
 489 {
 490         pushTok = pt;
 491 }