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 #include "support/types.h"
  23 #include "support/unicode.h"
  24
  25 #include <functional>
  26
  27
  28 namespace lyx {
  29
  30 using support::compare_ascii_no_case;
  31 using support::getFormatFromContents;
  32 using support::makeDisplayPath;
  33 using support::split;
  34 using support::subst;
  35
  36 using std::endl;
  37 using std::getline;
  38 using std::lower_bound;
  39 using std::sort;
  40 using std::string;
  41 using std::ios;
  42 using std::istream;
  43 using std::ostream;
  44
  45 namespace {
  46
  47 class compare_tags
  48         : public std::binary_function<keyword_item, keyword_item, bool> {
  49 public:
  50         // used by lower_bound, sort and sorted
  51         bool operator()(keyword_item const & a, keyword_item const & b) const
  52         {
  53                 // we use the ascii version, because in turkish, 'i'
  54                 // is not the lowercase version of 'I', and thus
  55                 // turkish locale breaks parsing of tags.
  56                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  57         }
  58 };
  59
  60 } // end of anon namespace
  61
  62
  63 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  64         : is(&fb_), table(tab), no_items(num),
  65           status(0), lineno(0), commentChar('#')
  66 {
  67         verifyTable();
  68 }
  69
  70
  71 string const LyXLex::Pimpl::getString() const
  72 {
  73         return string(buff.begin(), buff.end());
  74 }
  75
  76
  77 docstring const LyXLex::Pimpl::getDocString() const
  78 {
  79         std::vector<char_type> res = utf8_to_ucs4(buff);
  80         docstring dstr(res.begin(), res.end());
  81         return dstr;
  82 }
  83
  84
  85 void LyXLex::Pimpl::printError(string const & message) const
  86 {
  87         string const tmpmsg = subst(message, "$$Token", getString());
  88         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  89                 << " of file " << to_utf8(makeDisplayPath(name)) << ']' << endl;
  90 }
  91
  92
  93 void LyXLex::Pimpl::printTable(ostream & os)
  94 {
  95         os << "\nNumber of tags: " << no_items << endl;
  96         for (int i= 0; i < no_items; ++i)
  97                 os << "table[" << i
  98                    << "]:  tag: `" << table[i].tag
  99                    << "'  code:" << table[i].code << '\n';
 100         os.flush();
 101 }
 102
 103
 104 void LyXLex::Pimpl::verifyTable()
 105 {
 106         // Check if the table is sorted and if not, sort it.
 107         if (table
 108             && !lyx::sorted(table, table + no_items, compare_tags())) {
 109                 lyxerr << "The table passed to LyXLex is not sorted!\n"
 110                        << "Tell the developers to fix it!" << endl;
 111                 // We sort it anyway to avoid problems.
 112                 lyxerr << "\nUnsorted:" << endl;
 113                 printTable(lyxerr);
 114
 115                 sort(table, table + no_items, compare_tags());
 116                 lyxerr << "\nSorted:" << endl;
 117                 printTable(lyxerr);
 118         }
 119 }
 120
 121
 122 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
 123 {
 124         pushed_table tmppu(table, no_items);
 125         pushed.push(tmppu);
 126
 127         table = tab;
 128         no_items = num;
 129
 130         verifyTable();
 131 }
 132
 133
 134 void LyXLex::Pimpl::popTable()
 135 {
 136         if (pushed.empty()) {
 137                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 138                 return;
 139         }
 140
 141         pushed_table tmp = pushed.top();
 142         pushed.pop();
 143         table = tmp.table_elem;
 144         no_items = tmp.table_siz;
 145 }
 146
 147
 148 bool LyXLex::Pimpl::setFile(string const & filename)
 149 {
 150         // Check the format of the file.
 151         string const format = getFormatFromContents(filename);
 152
 153         if (format == "gzip" || format == "zip" || format == "compress") {
 154                 lyxerr[Debug::LYXLEX] << "lyxlex: compressed" << 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 (!gz_.empty() || istream::off_type(is.tellg()) > -1)
 160                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 161                                 "file or stream already set." << endl;
 162                 gz_.push(io::gzip_decompressor());
 163                 gz_.push(io::file_source(filename));
 164                 is.rdbuf(&gz_);
 165                 name = filename;
 166                 lineno = 0;
 167                 return gz_.component<io::file_source>(1)->is_open() && is.good();
 168         } else {
 169                 lyxerr[Debug::LYXLEX] << "lyxlex: UNcompressed" << endl;
 170
 171                 // The check only outputs a debug message, because it triggers
 172                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 173                 // a fresh new filebuf.  (JMarc)
 174                 if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 175                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 176                                 "file or stream already set." << endl;
 177                 fb_.open(filename.c_str(), ios::in);
 178                 is.rdbuf(&fb_);
 179                 name = filename;
 180                 lineno = 0;
 181                 return fb_.is_open() && is.good();
 182         }
 183 }
 184
 185
 186 void LyXLex::Pimpl::setStream(istream & i)
 187 {
 188         if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 189                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 190                         "file or stream already set." << endl;
 191         is.rdbuf(i.rdbuf());
 192         lineno = 0;
 193 }
 194
 195
 196 void LyXLex::Pimpl::setCommentChar(char c)
 197 {
 198         commentChar = c;
 199 }
 200
 201
 202 bool LyXLex::Pimpl::next(bool esc /* = false */)
 203 {
 204         if (!pushTok.empty()) {
 205                 // There can have been a whole line pushed so
 206                 // we extract the first word and leaves the rest
 207                 // in pushTok. (Lgb)
 208                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 209                         string tmp;
 210                         pushTok = split(pushTok, tmp, ' ');
 211                         buff.assign(tmp.begin(), tmp.end());
 212                         return true;
 213                 } else {
 214                         buff.assign(pushTok.begin(), pushTok.end());
 215                         pushTok.erase();
 216                         return true;
 217                 }
 218         }
 219         if (!esc) {
 220                 unsigned char c = 0; // getc() returns an int
 221                 char cc = 0;
 222                 status = 0;
 223                 while (is && !status) {
 224                         is.get(cc);
 225                         c = cc;
 226                         if (c == commentChar) {
 227                                 // Read rest of line (fast :-)
 228 #if 1
 229                                 // That is not fast... (Lgb)
 230                                 string dummy;
 231                                 getline(is, dummy);
 232
 233                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 234                                                       << dummy << '\'' << endl;
 235 #else
 236                                 // unfortunately ignore is buggy (Lgb)
 237                                 is.ignore(100, '\n');
 238 #endif
 239                                 ++lineno;
 240                                 continue;
 241                         }
 242
 243                         if (c == '\"') {
 244                                 buff.clear();
 245
 246                                 do {
 247                                         is.get(cc);
 248                                         c = cc;
 249                                         if (c != '\r')
 250                                                 buff.push_back(c);
 251                                 } while (c != '\"' && c != '\n' && is);
 252
 253                                 if (c != '\"') {
 254                                         printError("Missing quote");
 255                                         if (c == '\n')
 256                                                 ++lineno;
 257                                 }
 258
 259                                 buff.pop_back();
 260                                 status = LEX_DATA;
 261                                 break;
 262                         }
 263
 264                         if (c == ',')
 265                                 continue;              /* Skip ','s */
 266
 267                                 // using relational operators with chars other
 268                                 // than == and != is not safe. And if it is done
 269                                 // the type _have_ to be unsigned. It usually a
 270                                 // lot better to use the functions from cctype
 271                         if (c > ' ' && is)  {
 272                                 buff.clear();
 273
 274                                 do {
 275                                         buff.push_back(c);
 276                                         is.get(cc);
 277                                         c = cc;
 278                                 } while (c > ' ' && c != ',' && is);
 279
 280                                 status = LEX_TOKEN;
 281                         }
 282
 283                         if (c == '\r' && is) {
 284                                 // The Windows support has lead to the
 285                                 // possibility of "\r\n" at the end of
 286                                 // a line.  This will stop LyX choking
 287                                 // when it expected to find a '\n'
 288                                 is.get(cc);
 289                                 c = cc;
 290                         }
 291
 292                         if (c == '\n')
 293                                 ++lineno;
 294
 295                 }
 296                 if (status) return true;
 297
 298                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 299                 buff.clear();
 300                 return false;
 301         } else {
 302                 unsigned char c = 0; // getc() returns an int
 303                 char cc = 0;
 304
 305                 status = 0;
 306                 while (is && !status) {
 307                         is.get(cc);
 308                         c = cc;
 309
 310                         // skip ','s
 311                         if (c == ',') continue;
 312
 313                         if (c == '\\') {
 314                                 // escape
 315                                 buff.clear();
 316
 317                                 do {
 318                                         if (c == '\\') {
 319                                                 // escape the next char
 320                                                 is.get(cc);
 321                                                 c = cc;
 322                                         }
 323                                         buff.push_back(c);
 324                                         is.get(cc);
 325                                         c = cc;
 326                                 } while (c > ' ' && c != ',' && is);
 327
 328                                 status = LEX_TOKEN;
 329                                 continue;
 330                         }
 331
 332                         if (c == commentChar) {
 333                                 // Read rest of line (fast :-)
 334 #if 1
 335                                 // That is still not fast... (Lgb)
 336                                 string dummy;
 337                                 getline(is, dummy);
 338
 339                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 340                                                       << dummy << '\'' << endl;
 341 #else
 342                                 // but ignore is also still buggy (Lgb)
 343                                 // This is fast (Lgb)
 344                                 is.ignore(100, '\n');
 345 #endif
 346                                 ++lineno;
 347                                 continue;
 348                         }
 349
 350                         // string
 351                         if (c == '\"') {
 352                                 buff.clear();
 353
 354                                 bool escaped = false;
 355                                 do {
 356                                         escaped = false;
 357                                         is.get(cc);
 358                                         c = cc;
 359                                         if (c == '\r') continue;
 360                                         if (c == '\\') {
 361                                                 // escape the next char
 362                                                 is.get(cc);
 363                                                 c = cc;
 364                                                 if (c == '\"' || c == '\\')
 365                                                         escaped = true;
 366                                                 else
 367                                                         buff.push_back('\\');
 368                                         }
 369                                         buff.push_back(c);
 370
 371                                         if (!escaped && c == '\"') break;
 372                                 } while (c != '\n' && is);
 373
 374                                 if (c != '\"') {
 375                                         printError("Missing quote");
 376                                         if (c == '\n')
 377                                                 ++lineno;
 378                                 }
 379
 380                                 buff.pop_back();
 381                                 status = LEX_DATA;
 382                                 break;
 383                         }
 384
 385                         if (c > ' ' && is) {
 386                                 buff.clear();
 387
 388                                 do {
 389                                         if (c == '\\') {
 390                                                 // escape the next char
 391                                                 is.get(cc);
 392                                                 c = cc;
 393                                                 //escaped = true;
 394                                         }
 395                                         buff.push_back(c);
 396                                         is.get(cc);
 397                                         c = cc;
 398                                 } while (c > ' ' && c != ',' && is);
 399
 400                                 status = LEX_TOKEN;
 401                         }
 402                         // new line
 403                         if (c == '\n')
 404                                 ++lineno;
 405                 }
 406
 407                 if (status) return true;
 408
 409                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 410                 buff.clear();
 411                 return false;
 412         }
 413 }
 414
 415
 416 int LyXLex::Pimpl::search_kw(char const * const tag) const
 417 {
 418         keyword_item search_tag = { tag, 0 };
 419         keyword_item * res =
 420                 lower_bound(table, table + no_items,
 421                             search_tag, compare_tags());
 422         // use the compare_ascii_no_case instead of compare_no_case,
 423         // because in turkish, 'i' is not the lowercase version of 'I',
 424         // and thus turkish locale breaks parsing of tags.
 425         if (res != table + no_items
 426             && !compare_ascii_no_case(res->tag, tag))
 427                 return res->code;
 428         return LEX_UNDEF;
 429 }
 430
 431
 432 int LyXLex::Pimpl::lex()
 433 {
 434         //NOTE: possible bug.
 435         if (next() && status == LEX_TOKEN) {
 436                 return search_kw(getString().c_str());
 437         } else
 438                 return status;
 439 }
 440
 441
 442 bool LyXLex::Pimpl::eatLine()
 443 {
 444         buff.clear();
 445
 446         unsigned char c = '\0';
 447         char cc = 0;
 448         while (is && c != '\n') {
 449                 is.get(cc);
 450                 c = cc;
 451                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 452                 //                    << c << '\'' << endl;
 453                 if (c != '\r')
 454                         buff.push_back(c);
 455         }
 456
 457         if (c == '\n') {
 458                 ++lineno;
 459                 buff.pop_back();
 460                 status = LEX_DATA;
 461                 return true;
 462         } else {
 463                 return false;
 464         }
 465 }
 466
 467
 468 bool LyXLex::Pimpl::nextToken()
 469 {
 470         if (!pushTok.empty()) {
 471                 // There can have been a whole line pushed so
 472                 // we extract the first word and leaves the rest
 473                 // in pushTok. (Lgb)
 474                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 475                         string tmp;
 476                         pushTok = split(pushTok, tmp, ' ');
 477                         buff.assign(tmp.begin(), tmp.end());
 478                         return true;
 479                 } else {
 480                         buff.assign(pushTok.begin(), pushTok.end());
 481                         pushTok.erase();
 482                         return true;
 483                 }
 484         }
 485
 486         status = 0;
 487         while (is && !status) {
 488                 unsigned char c = 0;
 489                 char cc = 0;
 490                 is.get(cc);
 491                 c = cc;
 492                 if (c >= ' ' && is) {
 493                         buff.clear();
 494
 495                         if (c == '\\') { // first char == '\\'
 496                                 do {
 497                                         buff.push_back(c);
 498                                         is.get(cc);
 499                                         c = cc;
 500                                 } while (c > ' ' && c != '\\' && is);
 501                         } else {
 502                                 do {
 503                                         buff.push_back(c);
 504                                         is.get(cc);
 505                                         c = cc;
 506                                 } while (c >= ' ' && c != '\\' && is);
 507                         }
 508
 509                         if (c == '\\') is.putback(c); // put it back
 510                         status = LEX_TOKEN;
 511                 }
 512
 513                 if (c == '\n')
 514                         ++lineno;
 515
 516         }
 517         if (status)  return true;
 518
 519         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 520         buff.clear();
 521         return false;
 522 }
 523
 524
 525 void LyXLex::Pimpl::pushToken(string const & pt)
 526 {
 527         pushTok = pt;
 528 }
 529
 530
 531 } // namespace lyx