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                 // The check only outputs a debug message, because it triggers
 144                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 145                 // a fresh new filebuf.  (JMarc)
 146                 if (!gz_.empty() || istream::off_type(is.tellg()) > -1)
 147                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 148                                 "file or stream already set." << endl;
 149                 gz_.push(io::gzip_decompressor());
 150                 gz_.push(io::file_source(filename));
 151                 is.rdbuf(&gz_);
 152                 name = filename;
 153                 lineno = 0;
 154                 return gz_.component<io::file_source>(1)->is_open() && is.good();
 155         } else {
 156                 lyxerr[Debug::LYXLEX] << "lyxlex: UNcompressed" << endl;
 157
 158                 // The check only outputs a debug message, because it triggers
 159                 // a bug in compaq cxx 6.2, where is_open() returns 'true' for
 160                 // a fresh new filebuf.  (JMarc)
 161                 if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 162                         lyxerr[Debug::LYXLEX] << "Error in LyXLex::setFile: "
 163                                 "file or stream already set." << endl;
 164                 fb_.open(filename.c_str(), ios::in);
 165                 is.rdbuf(&fb_);
 166                 name = filename;
 167                 lineno = 0;
 168                 return fb_.is_open() && is.good();
 169         }
 170 }
 171
 172
 173 void LyXLex::Pimpl::setStream(istream & i)
 174 {
 175         if (fb_.is_open() || istream::off_type(is.tellg()) > 0)
 176                 lyxerr[Debug::LYXLEX]  << "Error in LyXLex::setStream: "
 177                         "file or stream already set." << endl;
 178         is.rdbuf(i.rdbuf());
 179         lineno = 0;
 180 }
 181
 182
 183 void LyXLex::Pimpl::setCommentChar(char c)
 184 {
 185         commentChar = c;
 186 }
 187
 188
 189 bool LyXLex::Pimpl::next(bool esc /* = false */)
 190 {
 191         if (!pushTok.empty()) {
 192                 // There can have been a whole line pushed so
 193                 // we extract the first word and leaves the rest
 194                 // in pushTok. (Lgb)
 195                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 196                         string tmp;
 197                         pushTok = split(pushTok, tmp, ' ');
 198                         buff.assign(tmp.begin(), tmp.end());
 199                         return true;
 200                 } else {
 201                         buff.assign(pushTok.begin(), pushTok.end());
 202                         pushTok.erase();
 203                         return true;
 204                 }
 205         }
 206         if (!esc) {
 207                 unsigned char c = 0; // getc() returns an int
 208                 char cc = 0;
 209                 status = 0;
 210                 while (is && !status) {
 211                         is.get(cc);
 212                         c = cc;
 213                         if (c == commentChar) {
 214                                 // Read rest of line (fast :-)
 215 #if 1
 216                                 // That is not fast... (Lgb)
 217                                 string dummy;
 218                                 getline(is, dummy);
 219
 220                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 221                                                       << dummy << '\'' << endl;
 222 #else
 223                                 // unfortunately ignore is buggy (Lgb)
 224                                 is.ignore(100, '\n');
 225 #endif
 226                                 ++lineno;
 227                                 continue;
 228                         }
 229
 230                         if (c == '\"') {
 231                                 buff.clear();
 232
 233                                 do {
 234                                         is.get(cc);
 235                                         c = cc;
 236                                         if (c != '\r')
 237                                                 buff.push_back(c);
 238                                 } while (c != '\"' && c != '\n' && is);
 239
 240                                 if (c != '\"') {
 241                                         printError("Missing quote");
 242                                         if (c == '\n')
 243                                                 ++lineno;
 244                                 }
 245
 246                                 buff.pop_back();
 247                                 status = LEX_DATA;
 248                                 break;
 249                         }
 250
 251                         if (c == ',')
 252                                 continue;              /* Skip ','s */
 253
 254                                 // using relational operators with chars other
 255                                 // than == and != is not safe. And if it is done
 256                                 // the type _have_ to be unsigned. It usually a
 257                                 // lot better to use the functions from cctype
 258                         if (c > ' ' && is)  {
 259                                 buff.clear();
 260
 261                                 do {
 262                                         buff.push_back(c);
 263                                         is.get(cc);
 264                                         c = cc;
 265                                 } while (c > ' ' && c != ',' && is);
 266
 267                                 status = LEX_TOKEN;
 268                         }
 269
 270                         if (c == '\r' && is) {
 271                                 // The Windows support has lead to the
 272                                 // possibility of "\r\n" at the end of
 273                                 // a line.  This will stop LyX choking
 274                                 // when it expected to find a '\n'
 275                                 is.get(cc);
 276                                 c = cc;
 277                         }
 278
 279                         if (c == '\n')
 280                                 ++lineno;
 281
 282                 }
 283                 if (status) return true;
 284
 285                 status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 286                 buff.clear();
 287                 return false;
 288         } else {
 289                 unsigned char c = 0; // getc() returns an int
 290                 char cc = 0;
 291
 292                 status = 0;
 293                 while (is && !status) {
 294                         is.get(cc);
 295                         c = cc;
 296
 297                         // skip ','s
 298                         if (c == ',') continue;
 299
 300                         if (c == '\\') {
 301                                 // escape
 302                                 buff.clear();
 303
 304                                 do {
 305                                         if (c == '\\') {
 306                                                 // escape the next char
 307                                                 is.get(cc);
 308                                                 c = cc;
 309                                         }
 310                                         buff.push_back(c);
 311                                         is.get(cc);
 312                                         c = cc;
 313                                 } while (c > ' ' && c != ',' && is);
 314
 315                                 status = LEX_TOKEN;
 316                                 continue;
 317                         }
 318
 319                         if (c == commentChar) {
 320                                 // Read rest of line (fast :-)
 321 #if 1
 322                                 // That is still not fast... (Lgb)
 323                                 string dummy;
 324                                 getline(is, dummy);
 325
 326                                 lyxerr[Debug::LYXLEX] << "Comment read: `" << c
 327                                                       << dummy << '\'' << endl;
 328 #else
 329                                 // but ignore is also still buggy (Lgb)
 330                                 // This is fast (Lgb)
 331                                 is.ignore(100, '\n');
 332 #endif
 333                                 ++lineno;
 334                                 continue;
 335                         }
 336
 337                         // string
 338                         if (c == '\"') {
 339                                 buff.clear();
 340
 341                                 bool escaped = false;
 342                                 do {
 343                                         escaped = false;
 344                                         is.get(cc);
 345                                         c = cc;
 346                                         if (c == '\r') continue;
 347                                         if (c == '\\') {
 348                                                 // escape the next char
 349                                                 is.get(cc);
 350                                                 c = cc;
 351                                                 if (c == '\"' || c == '\\')
 352                                                         escaped = true;
 353                                                 else
 354                                                         buff.push_back('\\');
 355                                         }
 356                                         buff.push_back(c);
 357
 358                                         if (!escaped && c == '\"') break;
 359                                 } while (c != '\n' && is);
 360
 361                                 if (c != '\"') {
 362                                         printError("Missing quote");
 363                                         if (c == '\n')
 364                                                 ++lineno;
 365                                 }
 366
 367                                 buff.pop_back();
 368                                 status = LEX_DATA;
 369                                 break;
 370                         }
 371
 372                         if (c > ' ' && is) {
 373                                 buff.clear();
 374
 375                                 do {
 376                                         if (c == '\\') {
 377                                                 // escape the next char
 378                                                 is.get(cc);
 379                                                 c = cc;
 380                                                 //escaped = true;
 381                                         }
 382                                         buff.push_back(c);
 383                                         is.get(cc);
 384                                         c = cc;
 385                                 } while (c > ' ' && c != ',' && is);
 386
 387                                 status = LEX_TOKEN;
 388                         }
 389                         // new line
 390                         if (c == '\n')
 391                                 ++lineno;
 392                 }
 393
 394                 if (status) return true;
 395
 396                 status = is.eof() ? LEX_FEOF : LEX_UNDEF;
 397                 buff.clear();
 398                 return false;
 399         }
 400 }
 401
 402
 403 int LyXLex::Pimpl::search_kw(char const * const tag) const
 404 {
 405         keyword_item search_tag = { tag, 0 };
 406         keyword_item * res =
 407                 lower_bound(table, table + no_items,
 408                             search_tag, compare_tags());
 409         // use the compare_ascii_no_case instead of compare_no_case,
 410         // because in turkish, 'i' is not the lowercase version of 'I',
 411         // and thus turkish locale breaks parsing of tags.
 412         if (res != table + no_items
 413             && !compare_ascii_no_case(res->tag, tag))
 414                 return res->code;
 415         return LEX_UNDEF;
 416 }
 417
 418
 419 int LyXLex::Pimpl::lex()
 420 {
 421         //NOTE: possible bug.
 422         if (next() && status == LEX_TOKEN) {
 423                 return search_kw(getString().c_str());
 424         } else
 425                 return status;
 426 }
 427
 428
 429 bool LyXLex::Pimpl::eatLine()
 430 {
 431         buff.clear();
 432
 433         unsigned char c = '\0';
 434         char cc = 0;
 435         while (is && c != '\n') {
 436                 is.get(cc);
 437                 c = cc;
 438                 //lyxerr[Debug::LYXLEX] << "LyXLex::EatLine read char: `"
 439                 //                    << c << '\'' << endl;
 440                 if (c != '\r')
 441                         buff.push_back(c);
 442         }
 443
 444         if (c == '\n') {
 445                 ++lineno;
 446                 buff.pop_back();
 447                 status = LEX_DATA;
 448                 return true;
 449         } else {
 450                 return false;
 451         }
 452 }
 453
 454
 455 bool LyXLex::Pimpl::nextToken()
 456 {
 457         if (!pushTok.empty()) {
 458                 // There can have been a whole line pushed so
 459                 // we extract the first word and leaves the rest
 460                 // in pushTok. (Lgb)
 461                 if (pushTok.find(' ') != string::npos && pushTok[0] == '\\') {
 462                         string tmp;
 463                         pushTok = split(pushTok, tmp, ' ');
 464                         buff.assign(tmp.begin(), tmp.end());
 465                         return true;
 466                 } else {
 467                         buff.assign(pushTok.begin(), pushTok.end());
 468                         pushTok.erase();
 469                         return true;
 470                 }
 471         }
 472
 473         status = 0;
 474         while (is && !status) {
 475                 unsigned char c = 0;
 476                 char cc = 0;
 477                 is.get(cc);
 478                 c = cc;
 479                 if (c >= ' ' && is) {
 480                         buff.clear();
 481
 482                         if (c == '\\') { // first char == '\\'
 483                                 do {
 484                                         buff.push_back(c);
 485                                         is.get(cc);
 486                                         c = cc;
 487                                 } while (c > ' ' && c != '\\' && is);
 488                         } else {
 489                                 do {
 490                                         buff.push_back(c);
 491                                         is.get(cc);
 492                                         c = cc;
 493                                 } while (c >= ' ' && c != '\\' && is);
 494                         }
 495
 496                         if (c == '\\') is.putback(c); // put it back
 497                         status = LEX_TOKEN;
 498                 }
 499
 500                 if (c == '\n')
 501                         ++lineno;
 502
 503         }
 504         if (status)  return true;
 505
 506         status = is.eof() ? LEX_FEOF: LEX_UNDEF;
 507         buff.clear();
 508         return false;
 509 }
 510
 511
 512 void LyXLex::Pimpl::pushToken(string const & pt)
 513 {
 514         pushTok = pt;
 515 }