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