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 using lyx::support::compare_ascii_no_case;
  28 using lyx::support::getFormatFromContents;
  29 using lyx::support::makeDisplayPath;
  30 using lyx::support::split;
  31 using lyx::support::subst;
  32
  33 using std::endl;
  34 using std::getline;
  35 using std::lower_bound;
  36 using std::sort;
  37 using std::string;
  38 using std::ios;
  39 using std::istream;
  40 using std::ostream;
  41
  42 namespace {
  43
  44 class compare_tags
  45         : public std::binary_function<keyword_item, keyword_item, bool> {
  46 public:
  47         // used by lower_bound, sort and sorted
  48         bool operator()(keyword_item const & a, keyword_item const & b) const
  49         {
  50                 // we use the ascii version, because in turkish, 'i'
  51                 // is not the lowercase version of 'I', and thus
  52                 // turkish locale breaks parsing of tags.
  53                 return compare_ascii_no_case(a.tag, b.tag) < 0;
  54         }
  55 };
  56
  57 } // end of anon namespace
  58
  59
  60 LyXLex::Pimpl::Pimpl(keyword_item * tab, int num)
  61         : is(&fb_), table(tab), no_items(num),
  62           status(0), lineno(0), commentChar('#')
  63 {
  64         verifyTable();
  65 }
  66
  67
  68 string const LyXLex::Pimpl::getString() const
  69 {
  70         return string(buff.begin(), buff.end());
  71 }
  72
  73
  74 lyx::docstring const LyXLex::Pimpl::getDocString() const
  75 {
  76         std::vector<boost::uint32_t> res = utf8_to_ucs4(buff);
  77         lyx::docstring dstr(res.begin(), res.end());
  78         return dstr;
  79 }
  80
  81
  82 void LyXLex::Pimpl::printError(string const & message) const
  83 {
  84         string const tmpmsg = subst(message, "$$Token", getString());
  85         lyxerr << "LyX: " << tmpmsg << " [around line " << lineno
  86                << " of file " << makeDisplayPath(name) << ']' << endl;
  87 }
  88
  89
  90 void LyXLex::Pimpl::printTable(ostream & os)
  91 {
  92         os << "\nNumber of tags: " << no_items << endl;
  93         for (int i= 0; i < no_items; ++i)
  94                 os << "table[" << i
  95                    << "]:  tag: `" << table[i].tag
  96                    << "'  code:" << table[i].code << '\n';
  97         os.flush();
  98 }
  99
 100
 101 void LyXLex::Pimpl::verifyTable()
 102 {
 103         // Check if the table is sorted and if not, sort it.
 104         if (table
 105             && !lyx::sorted(table, table + no_items, compare_tags())) {
 106                 lyxerr << "The table passed to LyXLex is not sorted!\n"
 107                        << "Tell the developers to fix it!" << endl;
 108                 // We sort it anyway to avoid problems.
 109                 lyxerr << "\nUnsorted:" << endl;
 110                 printTable(lyxerr);
 111
 112                 sort(table, table + no_items, compare_tags());
 113                 lyxerr << "\nSorted:" << endl;
 114                 printTable(lyxerr);
 115         }
 116 }
 117
 118
 119 void LyXLex::Pimpl::pushTable(keyword_item * tab, int num)
 120 {
 121         pushed_table tmppu(table, no_items);
 122         pushed.push(tmppu);
 123
 124         table = tab;
 125         no_items = num;
 126
 127         verifyTable();
 128 }
 129
 130
 131 void LyXLex::Pimpl::popTable()
 132 {
 133         if (pushed.empty()) {
 134                 lyxerr << "LyXLex error: nothing to pop!" << endl;
 135                 return;
 136         }
 137
 138         pushed_table tmp = pushed.top();
 139         pushed.pop();
 140         table = tmp.table_elem;
 141         no_items = tmp.table_siz;
 142 }
 143
 144
 145 bool LyXLex::Pimpl::setFile(string const & filename)
 146 {
 147         // Check the format of the file.
 148         string const format = getFormatFromContents(filename);
 149
 150         if (format == "gzip" || format == "zip" || format == "compress") {
 151                 lyxerr[Debug::LYXLEX] << "lyxlex: compressed" << endl;
 152
 153                 // The check only outputs a debug message, because it triggers
 154                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 155                 // a fresh new filebuf.  (JMarc)
 156                 if (!gz_.empty() || istream::off_type(is.tellg()) > -1)
 157                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 158                                 "file or stream already set." << endl;
 159                 gz_.push(io::gzip_decompressor());
 160                 gz_.push(io::file_source(filename));
 161                 is.rdbuf(&gz_);
 162                 name = filename;
 163                 lineno = 0;
 164                 return gz_.component<io::file_source>(1)->is_open() && is.good();
 165         } else {
 166                 lyxerr[Debug::LYXLEX] << "lyxlex: UNcompressed" << endl;
 167
 168                 // The check only outputs a debug message, because it triggers
 169                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 170                 // a fresh new filebuf.  (JMarc)
 171                 if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 172                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 173                                 "file or stream already set." << endl;
 174                 fb_.open(filename.c_str(), ios::in);
 175                 is.rdbuf(&fb_);
 176                 name = filename;
 177                 lineno = 0;
 178                 return fb_.is_open() && is.good();
 179         }
 180 }
 181
 182
 183 void LyXLex::Pimpl::setStream(istream & i)
 184 {
 185         if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 186                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 187                         "file or stream already set." << endl;
 188         is.rdbuf(i.rdbuf());
 189         lineno = 0;
 190 }
 191
 192
 193 void LyXLex::Pimpl::setCommentChar(char c)
 194 {
 195         commentChar = c;
 196 }
 197
 198
 199 bool LyXLex::Pimpl::next(bool esc /* = false */)
 200 {
 201         if (!pushTok.empty()) {
 202                 // There can have been a whole line pushed so
 203                 // we extract the first word and leaves the rest
 204                 // in pushTok. (Lgb)
 205                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 206                         string tmp;
 207                         pushTok = split(pushTok, tmp, ' ');
 208                         buff.assign(tmp.begin(), tmp.end());
 209                         return true;
 210                 } else {
 211                         buff.assign(pushTok.begin(), pushTok.end());
 212                         pushTok.erase();
 213                         return true;
 214                 }
 215         }
 216         if (!esc) {
 217                 unsigned char c = 0; // getc() returns an int
 218                 char cc = 0;
 219                 status = 0;
 220                 while (is && !status) {
 221                         is.get(cc);
 222                         c = cc;
 223                         if (c == commentChar) {
 224                                 // Read rest of line (fast :-)
 225 #if 1
 226                                 // That is not fast... (Lgb)
 227                                 string dummy;
 228                                 getline(is, dummy);
 229
 230                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 231                                                       << dummy << '\'' << endl;
 232 #else
 233                                 // unfortunately ignore is buggy (Lgb)
 234                                 is.ignore(100, '\n');
 235 #endif
 236                                 ++lineno;
 237                                 continue;
 238                         }
 239
 240                         if (c == '\"') {
 241                                 buff.clear();
 242
 243                                 do {
 244                                         is.get(cc);
 245                                         c = cc;
 246                                         if (c != '\r')
 247                                                 buff.push_back(c);
 248                                 } while (c != '\"' && c != '\n' && is);
 249
 250                                 if (c != '\"') {
 251                                         printError("Missing quote");
 252                                         if (c == '\n')
 253                                                 ++lineno;
 254                                 }
 255
 256                                 buff.pop_back();
 257                                 status = LEX_DATA;
 258                                 break;
 259                         }
 260
 261                         if (c == ',')
 262                                 continue;              /* Skip ','s */
 263
 264                                 // using relational operators with chars other
 265                                 // than == and != is not safe. And if it is done
 266                                 // the type _have_ to be unsigned. It usually a
 267                                 // lot better to use the functions from cctype
 268                         if (c > ' ' && is)  {
 269                                 buff.clear();
 270
 271                                 do {
 272                                         buff.push_back(c);
 273                                         is.get(cc);
 274                                         c = cc;
 275                                 } while (c > ' ' && c != ',' && is);
 276
 277                                 status = LEX_TOKEN;
 278                         }
 279
 280                         if (c == '\r' && is) {
 281                                 // The Windows support has lead to the
 282                                 // possibility of "\r\n" at the end of
 283                                 // a line.  This will stop LyX choking
 284                                 // when it expected to find a '\n'
 285                                 is.get(cc);
 286                                 c = cc;
 287                         }
 288
 289                         if (c == '\n')
 290                                 ++lineno;
 291
 292                 }
 293                 if (status) return true;
 294
 295                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 296                 buff.clear();
 297                 return false;
 298         } else {
 299                 unsigned char c = 0; // getc() returns an int
 300                 char cc = 0;
 301
 302                 status = 0;
 303                 while (is && !status) {
 304                         is.get(cc);
 305                         c = cc;
 306
 307                         // skip ','s
 308                         if (c == ',') continue;
 309
 310                         if (c == '\\') {
 311                                 // escape
 312                                 buff.clear();
 313
 314                                 do {
 315                                         if (c == '\\') {
 316                                                 // escape the next char
 317                                                 is.get(cc);
 318                                                 c = cc;
 319                                         }
 320                                         buff.push_back(c);
 321                                         is.get(cc);
 322                                         c = cc;
 323                                 } while (c > ' ' && c != ',' && is);
 324
 325                                 status = LEX_TOKEN;
 326                                 continue;
 327                         }
 328
 329                         if (c == commentChar) {
 330                                 // Read rest of line (fast :-)
 331 #if 1
 332                                 // That is still not fast... (Lgb)
 333                                 string dummy;
 334                                 getline(is, dummy);
 335
 336                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 337                                                       << dummy << '\'' << endl;
 338 #else
 339                                 // but ignore is also still buggy (Lgb)
 340                                 // This is fast (Lgb)
 341                                 is.ignore(100, '\n');
 342 #endif
 343                                 ++lineno;
 344                                 continue;
 345                         }
 346
 347                         // string
 348                         if (c == '\"') {
 349                                 buff.clear();
 350
 351                                 bool escaped = false;
 352                                 do {
 353                                         escaped = false;
 354                                         is.get(cc);
 355                                         c = cc;
 356                                         if (c == '\r') continue;
 357                                         if (c == '\\') {
 358                                                 // escape the next char
 359                                                 is.get(cc);
 360                                                 c = cc;
 361                                                 if (c == '\"' || c == '\\')
 362                                                         escaped = true;
 363                                                 else
 364                                                         buff.push_back('\\');
 365                                         }
 366                                         buff.push_back(c);
 367
 368                                         if (!escaped && c == '\"') break;
 369                                 } while (c != '\n' && is);
 370
 371                                 if (c != '\"') {
 372                                         printError("Missing quote");
 373                                         if (c == '\n')
 374                                                 ++lineno;
 375                                 }
 376
 377                                 buff.pop_back();
 378                                 status = LEX_DATA;
 379                                 break;
 380                         }
 381
 382                         if (c > ' ' && is) {
 383                                 buff.clear();
 384
 385                                 do {
 386                                         if (c == '\\') {
 387                                                 // escape the next char
 388                                                 is.get(cc);
 389                                                 c = cc;
 390                                                 //escaped = true;
 391                                         }
 392                                         buff.push_back(c);
 393                                         is.get(cc);
 394                                         c = cc;
 395                                 } while (c > ' ' && c != ',' && is);
 396
 397                                 status = LEX_TOKEN;
 398                         }
 399                         // new line
 400                         if (c == '\n')
 401                                 ++lineno;
 402                 }
 403
 404                 if (status) return true;
 405
 406                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 407                 buff.clear();
 408                 return false;
 409         }
 410 }
 411
 412
 413 int LyXLex::Pimpl::search_kw(char const * const tag) const
 414 {
 415         keyword_item search_tag = { tag, 0 };
 416         keyword_item * res =
 417                 lower_bound(table, table + no_items,
 418                             search_tag, compare_tags());
 419         // use the compare_ascii_no_case instead of compare_no_case,
 420         // because in turkish, 'i' is not the lowercase version of 'I',
 421         // and thus turkish locale breaks parsing of tags.
 422         if (res != table + no_items
 423             && !compare_ascii_no_case(res->tag, tag))
 424                 return res->code;
 425         return LEX_UNDEF;
 426 }
 427
 428
 429 int LyXLex::Pimpl::lex()
 430 {
 431         //NOTE: possible bug.
 432         if (next() && status == LEX_TOKEN) {
 433                 return search_kw(getString().c_str());
 434         } else
 435                 return status;
 436 }
 437
 438
 439 bool LyXLex::Pimpl::eatLine()
 440 {
 441         buff.clear();
 442
 443         unsigned char c = '\0';
 444         char cc = 0;
 445         while (is && c != '\n') {
 446                 is.get(cc);
 447                 c = cc;
 448                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 449                 //                    << c << '\'' << endl;
 450                 if (c != '\r')
 451                         buff.push_back(c);
 452         }
 453
 454         if (c == '\n') {
 455                 ++lineno;
 456                 buff.pop_back();
 457                 status = LEX_DATA;
 458                 return true;
 459         } else {
 460                 return false;
 461         }
 462 }
 463
 464
 465 bool LyXLex::Pimpl::nextToken()
 466 {
 467         if (!pushTok.empty()) {
 468                 // There can have been a whole line pushed so
 469                 // we extract the first word and leaves the rest
 470                 // in pushTok. (Lgb)
 471                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 472                         string tmp;
 473                         pushTok = split(pushTok, tmp, ' ');
 474                         buff.assign(tmp.begin(), tmp.end());
 475                         return true;
 476                 } else {
 477                         buff.assign(pushTok.begin(), pushTok.end());
 478                         pushTok.erase();
 479                         return true;
 480                 }
 481         }
 482
 483         status = 0;
 484         while (is && !status) {
 485                 unsigned char c = 0;
 486                 char cc = 0;
 487                 is.get(cc);
 488                 c = cc;
 489                 if (c >= ' ' && is) {
 490                         buff.clear();
 491
 492                         if (c == '\\') { // first char == '\\'
 493                                 do {
 494                                         buff.push_back(c);
 495                                         is.get(cc);
 496                                         c = cc;
 497                                 } while (c > ' ' && c != '\\' && is);
 498                         } else {
 499                                 do {
 500                                         buff.push_back(c);
 501                                         is.get(cc);
 502                                         c = cc;
 503                                 } while (c >= ' ' && c != '\\' && is);
 504                         }
 505
 506                         if (c == '\\') is.putback(c); // put it back
 507                         status = LEX_TOKEN;
 508                 }
 509
 510                 if (c == '\n')
 511                         ++lineno;
 512
 513         }
 514         if (status)  return true;
 515
 516         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 517         buff.clear();
 518         return false;
 519 }
 520
 521
 522 void LyXLex::Pimpl::pushToken(string const & pt)
 523 {
 524         pushTok = pt;
 525 }