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