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         // The check only outputs a debug message, because it triggers
 117         // a bug in compaq cxx 6.2, where is_open() returns 'true' for a
 118         // fresh new filebuf.  (JMarc)
 119         if (fb__.is_open() || is.tellg() > 0)
 120                 lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 121                         "file or stream already set." << endl;
 122         fb__.open(filename.c_str(), ios::in);
 123         is.rdbuf(&fb__);
 124         name = filename;
 125         lineno = 0;
 126         return fb__.is_open() && is.good();
 127 }
 128
 129
 130 void LyXLex::Pimpl::setStream(istream & i)
 131 {
 132         if (fb__.is_open() || is.tellg() > 0)
 133                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 134                         "file or stream already set." << endl;
 135         is.rdbuf(i.rdbuf());
 136         lineno = 0;
 137 }
 138
 139
 140 void LyXLex::Pimpl::setCommentChar(char c)
 141 {
 142         commentChar = c;
 143 }
 144
 145
 146 bool LyXLex::Pimpl::next(bool esc /* = false */)
 147 {
 148         if (!pushTok.empty()) {
 149                 // There can have been a whole line pushed so
 150                 // we extract the first word and leaves the rest
 151                 // in pushTok. (Lgb)
 152                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 153                         string tmp;
 154                         pushTok = split(pushTok, tmp, ' ');
 155                         buff.assign(tmp.begin(), tmp.end());
 156                         return true;
 157                 } else {
 158                         buff.assign(pushTok.begin(), pushTok.end());
 159                         pushTok.erase();
 160                         return true;
 161                 }
 162         }
 163         if (!esc) {
 164                 unsigned char c = 0; // getc() returns an int
 165                 char cc = 0;
 166                 status = 0;
 167                 while (is && !status) {
 168                         is.get(cc);
 169                         c = cc;
 170                         if (c == commentChar) {
 171                                 // Read rest of line (fast :-)
 172 #if 1
 173                                 // That is not fast... (Lgb)
 174                                 string dummy;
 175                                 getline(is, dummy);
 176
 177                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 178                                                       << dummy << '\'' << endl;
 179 #else
 180                                 // unfortunately ignore is buggy (Lgb)
 181                                 is.ignore(100, '\n');
 182 #endif
 183                                 ++lineno;
 184                                 continue;
 185                         }
 186
 187                         if (c == '\"') {
 188                                 buff.clear();
 189
 190                                 do {
 191                                         is.get(cc);
 192                                         c = cc;
 193                                         if (c != '\r')
 194                                                 buff.push_back(c);
 195                                 } while (c != '\"' && c != '\n' && is);
 196
 197                                 if (c != '\"') {
 198                                         printError("Missing quote");
 199                                         if (c == '\n')
 200                                                 ++lineno;
 201                                 }
 202
 203                                 buff.pop_back();
 204                                 status = LEX_DATA;
 205                                 break;
 206                         }
 207
 208                         if (c == ',')
 209                                 continue;              /* Skip ','s */
 210
 211                                 // using relational operators with chars other
 212                                 // than == and != is not safe. And if it is done
 213                                 // the type _have_ to be unsigned. It usually a
 214                                 // lot better to use the functions from cctype
 215                         if (c > ' ' && is)  {
 216                                 buff.clear();
 217
 218                                 do {
 219                                         buff.push_back(c);
 220                                         is.get(cc);
 221                                         c = cc;
 222                                 } while (c > ' ' && c != ',' && is);
 223
 224                                 status = LEX_TOKEN;
 225                         }
 226
 227                         if (c == '\r' && is) {
 228                                 // The Windows support has lead to the
 229                                 // possibility of "\r\n" at the end of
 230                                 // a line.  This will stop LyX choking
 231                                 // when it expected to find a '\n'
 232                                 is.get(cc);
 233                                 c = cc;
 234                         }
 235
 236                         if (c == '\n')
 237                                 ++lineno;
 238
 239                 }
 240                 if (status) return true;
 241
 242                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 243                 buff.clear();
 244                 return false;
 245         } else {
 246                 unsigned char c = 0; // getc() returns an int
 247                 char cc = 0;
 248
 249                 status = 0;
 250                 while (is && !status) {
 251                         is.get(cc);
 252                         c = cc;
 253
 254                         // skip ','s
 255                         if (c == ',') continue;
 256
 257                         if (c == '\\') {
 258                                 // escape
 259                                 buff.clear();
 260
 261                                 do {
 262                                         if (c == '\\') {
 263                                                 // escape the next char
 264                                                 is.get(cc);
 265                                                 c = cc;
 266                                         }
 267                                         buff.push_back(c);
 268                                         is.get(cc);
 269                                         c = cc;
 270                                 } while (c > ' ' && c != ',' && is);
 271
 272                                 status = LEX_TOKEN;
 273                                 continue;
 274                         }
 275
 276                         if (c == commentChar) {
 277                                 // Read rest of line (fast :-)
 278 #if 1
 279                                 // That is still not fast... (Lgb)
 280                                 string dummy;
 281                                 getline(is, dummy);
 282
 283                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 284                                                       << dummy << '\'' << endl;
 285 #else
 286                                 // but ignore is also still buggy (Lgb)
 287                                 // This is fast (Lgb)
 288                                 is.ignore(100, '\n');
 289 #endif
 290                                 ++lineno;
 291                                 continue;
 292                         }
 293
 294                         // string
 295                         if (c == '\"') {
 296                                 buff.clear();
 297
 298                                 bool escaped = false;
 299                                 do {
 300                                         escaped = false;
 301                                         is.get(cc);
 302                                         c = cc;
 303                                         if (c == '\r') continue;
 304                                         if (c == '\\') {
 305                                                 // escape the next char
 306                                                 is.get(cc);
 307                                                 c = cc;
 308                                                 if (c == '\"' || c == '\\')
 309                                                         escaped = true;
 310                                                 else
 311                                                         buff.push_back('\\');
 312                                         }
 313                                         buff.push_back(c);
 314
 315                                         if (!escaped && c == '\"') break;
 316                                 } while (c != '\n' && is);
 317
 318                                 if (c != '\"') {
 319                                         printError("Missing quote");
 320                                         if (c == '\n')
 321                                                 ++lineno;
 322                                 }
 323
 324                                 buff.pop_back();
 325                                 status = LEX_DATA;
 326                                 break;
 327                         }
 328
 329                         if (c > ' ' && is) {
 330                                 buff.clear();
 331
 332                                 do {
 333                                         if (c == '\\') {
 334                                                 // escape the next char
 335                                                 is.get(cc);
 336                                                 c = cc;
 337                                                 //escaped = true;
 338                                         }
 339                                         buff.push_back(c);
 340                                         is.get(cc);
 341                                         c = cc;
 342                                 } while (c > ' ' && c != ',' && is);
 343
 344                                 status = LEX_TOKEN;
 345                         }
 346                         // new line
 347                         if (c == '\n')
 348                                 ++lineno;
 349                 }
 350
 351                 if (status) return true;
 352
 353                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 354                 buff.clear();
 355                 return false;
 356         }
 357 }
 358
 359
 360 int LyXLex::Pimpl::search_kw(char const * const tag) const
 361 {
 362         keyword_item search_tag = { tag, 0 };
 363         keyword_item * res =
 364                 lower_bound(table, table + no_items,
 365                             search_tag, compare_tags());
 366         // use the compare_ascii_no_case instead of compare_no_case,
 367         // because in turkish, 'i' is not the lowercase version of 'I',
 368         // and thus turkish locale breaks parsing of tags.
 369         if (res != table + no_items
 370             && !compare_ascii_no_case(res->tag, tag))
 371                 return res->code;
 372         return LEX_UNDEF;
 373 }
 374
 375
 376 int LyXLex::Pimpl::lex()
 377 {
 378         //NOTE: possible bug.
 379         if (next() && status == LEX_TOKEN) {
 380                 return search_kw(getString().c_str());
 381         } else
 382                 return status;
 383 }
 384
 385
 386 bool LyXLex::Pimpl::eatLine()
 387 {
 388         buff.clear();
 389
 390         unsigned char c = '\0';
 391         char cc = 0;
 392         while (is && c != '\n') {
 393                 is.get(cc);
 394                 c = cc;
 395                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 396                 //                    << c << '\'' << endl;
 397                 if (c != '\r')
 398                         buff.push_back(c);
 399         }
 400
 401         if (c == '\n') {
 402                 ++lineno;
 403                 buff.pop_back();
 404                 status = LEX_DATA;
 405                 return true;
 406         } else {
 407                 return false;
 408         }
 409 }
 410
 411
 412 bool LyXLex::Pimpl::nextToken()
 413 {
 414         if (!pushTok.empty()) {
 415                 // There can have been a whole line pushed so
 416                 // we extract the first word and leaves the rest
 417                 // in pushTok. (Lgb)
 418                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 419                         string tmp;
 420                         pushTok = split(pushTok, tmp, ' ');
 421                         buff.assign(tmp.begin(), tmp.end());
 422                         return true;
 423                 } else {
 424                         buff.assign(pushTok.begin(), pushTok.end());
 425                         pushTok.erase();
 426                         return true;
 427                 }
 428         }
 429
 430         status = 0;
 431         while (is && !status) {
 432                 unsigned char c = 0;
 433                 char cc = 0;
 434                 is.get(cc);
 435                 c = cc;
 436                 if (c >= ' ' && is) {
 437                         buff.clear();
 438
 439                         if (c == '\\') { // first char == '\\'
 440                                 do {
 441                                         buff.push_back(c);
 442                                         is.get(cc);
 443                                         c = cc;
 444                                 } while (c > ' ' && c != '\\' && is);
 445                         } else {
 446                                 do {
 447                                         buff.push_back(c);
 448                                         is.get(cc);
 449                                         c = cc;
 450                                 } while (c >= ' ' && c != '\\' && is);
 451                         }
 452
 453                         if (c == '\\') is.putback(c); // put it back
 454                         status = LEX_TOKEN;
 455                 }
 456
 457                 if (c == '\n')
 458                         ++lineno;
 459
 460         }
 461         if (status)  return true;
 462
 463         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 464         buff.clear();
 465         return false;
 466 }
 467
 468
 469 void LyXLex::Pimpl::pushToken(string const & pt)
 470 {
 471         pushTok = pt;
 472 }