mcs/mbas/mb-tokenizer.cs

   1 //
   2 // Mono.MonoBASIC.Tokenizer.cs: The Tokenizer for the MonoBASIC compiler
   3 //
   4 // Author: A Rafael D Teixeira (rafaelteixeirabr@hotmail.com)
   5 //       : Manjula GHM (mmanjula@novell.com)
   6 // Based on cs-tokenizer.cs by Miguel de Icaza (miguel@gnu.org)
   7 //
   8 // Licensed under the terms of the GNU GPL
   9 //
  10 // Copyright (C) 2001 A Rafael D Teixeira
  11 //
  12
  13 namespace Mono.MonoBASIC
  14 {
  15         using System;
  16         using System.Text;
  17         using System.Collections;
  18         using System.IO;
  19         using System.Globalization;
  20         using Mono.Languages;
  21         using Mono.MonoBASIC;
  22
  23         /// <summary>
  24         ///    Tokenizer for MonoBASIC source code.
  25         /// </summary>
  26
  27         public class Tokenizer : yyParser.yyInput
  28         {
  29                 TextReader reader;
  30                 string file_name;
  31                 string ref_name;
  32                 int ref_line = 0;
  33                 int line = 0;
  34                 int col = 1;
  35                 public int current_token = Token.ERROR;
  36                 bool handle_get_set = false;
  37                 bool cant_have_a_type_character = false;
  38
  39                 public int ExpandedTabsSize = 4;
  40
  41                 public string location {
  42                         get {
  43                                 string det;
  44
  45                                 if (current_token == Token.ERROR)
  46                                         det = "detail: " + error_details;
  47                                 else
  48                                         det = "";
  49
  50                                 return "Line:     "+line+" Col: "+col + "\n" +
  51                                        "VirtLine: "+ref_line +
  52                                        " Token: "+current_token + " " + det;
  53                         }
  54                 }
  55
  56                 public bool properties {
  57                         get {
  58                                 return handle_get_set;
  59                         }
  60
  61                         set {
  62                                 handle_get_set = value;
  63                         }
  64                 }
  65
  66                 //
  67                 // Class variables
  68                 //
  69                 static Hashtable keywords;
  70                 static NumberStyles styles;
  71                 static NumberFormatInfo csharp_format_info;
  72
  73                 //
  74                 // Values for the associated token returned
  75                 //
  76                 StringBuilder number;
  77                 int putback_char = -1;
  78                 Object val;
  79                 long lon = 0;
  80
  81                 //
  82                 // Details about the error encoutered by the tokenizer
  83                 //
  84                 string error_details;
  85
  86                 public string error {
  87                         get {
  88                                 return error_details;
  89                         }
  90                 }
  91
  92
  93                 public string Source {
  94                         get {
  95                                 return file_name;
  96                         }
  97
  98                         set {
  99                                 file_name = value;
 100                                 ref_name = value;
 101                                 Location.SetCurrentSource(file_name);
 102                         }
 103                 }
 104
 105                 public string EffectiveSource {
 106                         get {
 107                                 return ref_name;
 108                         }
 109                         set {
 110                                 ref_name = value;
 111                                 Location.SetCurrentSource(ref_name);
 112                         }
 113                 }
 114
 115                 public int Line {
 116                         get {
 117                                 return line;
 118                         }
 119                 }
 120
 121                 public int EffectiveLine {
 122                         get {
 123                                 return ref_line;
 124                         }
 125                         set {
 126                                 ref_line = value;
 127                         }
 128                 }
 129
 130                 public int Col {
 131                         get {
 132                                 return col;
 133                         }
 134                 }
 135
 136                 static void initTokens ()
 137                 {
 138                         keywords = new Hashtable ();
 139
 140                         keywords.Add ("addhandler", Token.ADDHANDLER);
 141                         keywords.Add ("addressof", Token.ADDRESSOF);
 142                         keywords.Add ("alias", Token.ALIAS);
 143                         keywords.Add ("and", Token.AND);
 144                         keywords.Add ("andalso", Token.ANDALSO);
 145                         keywords.Add ("ansi", Token.ANSI);
 146                         keywords.Add ("as", Token.AS);
 147                         keywords.Add ("assembly", Token.ASSEMBLY);
 148                         keywords.Add ("auto", Token.AUTO);
 149                         keywords.Add ("binary", Token.BINARY); // Not a VB.NET Keyword
 150                         keywords.Add ("boolean", Token.BOOLEAN);
 151                         keywords.Add ("byref", Token.BYREF);
 152                         keywords.Add ("byte", Token.BYTE);
 153                         keywords.Add ("byval", Token.BYVAL);
 154                         keywords.Add ("call", Token.CALL);
 155                         keywords.Add ("case", Token.CASE);
 156                         keywords.Add ("catch", Token.CATCH);
 157                         keywords.Add ("cbool", Token.CBOOL);
 158                         keywords.Add ("cbyte", Token.CBYTE);
 159                         keywords.Add ("cchar", Token.CCHAR);
 160                         keywords.Add ("cdate", Token.CDATE);
 161                         keywords.Add ("cdec", Token.CDEC);
 162                         keywords.Add ("cdbl", Token.CDBL);
 163                         keywords.Add ("char", Token.CHAR);
 164                         keywords.Add ("cint", Token.CINT);
 165                         keywords.Add ("class", Token.CLASS);
 166                         keywords.Add ("clng", Token.CLNG);
 167                         keywords.Add ("cobj", Token.COBJ);
 168                         keywords.Add ("compare", Token.COMPARE); // Not a VB.NET Keyword
 169                         keywords.Add ("const", Token.CONST);
 170                         keywords.Add ("cshort", Token.CSHORT);
 171                         keywords.Add ("csng", Token.CSNG);
 172                         keywords.Add ("cstr", Token.CSTR);
 173                         keywords.Add ("ctype", Token.CTYPE);
 174                         keywords.Add ("date", Token.DATE);
 175                         keywords.Add ("decimal", Token.DECIMAL);
 176                         keywords.Add ("declare", Token.DECLARE);
 177                         keywords.Add ("default", Token.DEFAULT);
 178                         keywords.Add ("delegate", Token.DELEGATE);
 179                         keywords.Add ("dim", Token.DIM);
 180                         keywords.Add ("directcast", Token.DIRECTCAST);
 181                         keywords.Add ("do", Token.DO);
 182                         keywords.Add ("double", Token.DOUBLE);
 183                         keywords.Add ("each", Token.EACH);
 184                         keywords.Add ("else", Token.ELSE);
 185                         keywords.Add ("elseif", Token.ELSEIF);
 186                         keywords.Add ("end", Token.END);
 187                         keywords.Add ("endif", Token.ENDIF); // An unused VB.NET keyword
 188                         keywords.Add ("enum", Token.ENUM);
 189                         keywords.Add ("erase", Token.ERASE);
 190                         keywords.Add ("error", Token.ERROR);
 191                         keywords.Add ("event", Token.EVENT);
 192                         keywords.Add ("exit", Token.EXIT);
 193                         keywords.Add ("explicit", Token.EXPLICIT); // Not a VB.NET keyword
 194                         keywords.Add ("false", Token.FALSE);
 195                         keywords.Add ("finally", Token.FINALLY);
 196                         keywords.Add ("for", Token.FOR);
 197                         keywords.Add ("friend", Token.FRIEND);
 198                         keywords.Add ("function", Token.FUNCTION);
 199                         keywords.Add ("get", Token.GET);
 200                         keywords.Add ("gettype", Token.GETTYPE);
 201                         keywords.Add ("gosub", Token.GOSUB); // An unused VB.NET keyword
 202                         keywords.Add ("goto", Token.GOTO);
 203                         keywords.Add ("handles", Token.HANDLES);
 204                         keywords.Add ("if", Token.IF);
 205                         keywords.Add ("implements", Token.IMPLEMENTS);
 206                         keywords.Add ("imports", Token.IMPORTS);
 207                         keywords.Add ("in", Token.IN);
 208                         keywords.Add ("inherits", Token.INHERITS);
 209                         keywords.Add ("integer", Token.INTEGER);
 210                         keywords.Add ("interface", Token.INTERFACE);
 211                         keywords.Add ("is", Token.IS);
 212                         keywords.Add ("let ", Token.LET ); // An unused VB.NET keyword
 213                         keywords.Add ("lib ", Token.LIB );
 214                         keywords.Add ("like", Token.LIKE );
 215                         keywords.Add ("long", Token.LONG);
 216                         keywords.Add ("loop", Token.LOOP);
 217                         keywords.Add ("me", Token.ME);
 218                         keywords.Add ("mod", Token.MOD);
 219                         keywords.Add ("module", Token.MODULE);
 220                         keywords.Add ("mustinherit", Token.MUSTINHERIT);
 221                         keywords.Add ("mustoverride", Token.MUSTOVERRIDE);
 222                         keywords.Add ("mybase", Token.MYBASE);
 223                         keywords.Add ("myclass", Token.MYCLASS);
 224                         keywords.Add ("namespace", Token.NAMESPACE);
 225                         keywords.Add ("new", Token.NEW);
 226                         keywords.Add ("next", Token.NEXT);
 227                         keywords.Add ("not", Token.NOT);
 228                         keywords.Add ("nothing", Token.NOTHING);
 229                         keywords.Add ("notinheritable", Token.NOTINHERITABLE);
 230                         keywords.Add ("notoverridable", Token.NOTOVERRIDABLE);
 231                         keywords.Add ("object", Token.OBJECT);
 232                         keywords.Add ("off", Token.OFF); // Not a VB.NET Keyword
 233                         keywords.Add ("on", Token.ON);
 234                         keywords.Add ("option", Token.OPTION);
 235                         keywords.Add ("optional", Token.OPTIONAL);
 236                         keywords.Add ("or", Token.OR);
 237                         keywords.Add ("orelse", Token.ORELSE);
 238                         keywords.Add ("overloads", Token.OVERLOADS);
 239                         keywords.Add ("overridable", Token.OVERRIDABLE);
 240                         keywords.Add ("overrides", Token.OVERRIDES);
 241                         keywords.Add ("paramarray", Token.PARAM_ARRAY);
 242                         keywords.Add ("preserve", Token.PRESERVE);
 243                         keywords.Add ("private", Token.PRIVATE);
 244                         keywords.Add ("property", Token.PROPERTY);
 245                         keywords.Add ("protected", Token.PROTECTED);
 246                         keywords.Add ("public", Token.PUBLIC);
 247                         keywords.Add ("raiseevent", Token.RAISEEVENT);
 248                         keywords.Add ("readonly", Token.READONLY);
 249                         keywords.Add ("redim", Token.REDIM);
 250                         keywords.Add ("rem", Token.REM);
 251                         keywords.Add ("removehandler", Token.REMOVEHANDLER);
 252                         keywords.Add ("resume", Token.RESUME);
 253                         keywords.Add ("return", Token.RETURN);
 254                         keywords.Add ("select", Token.SELECT);
 255                         keywords.Add ("set", Token.SET);
 256                         keywords.Add ("shadows", Token.SHADOWS);
 257                         keywords.Add ("shared", Token.SHARED);
 258                         keywords.Add ("short", Token.SHORT);
 259                         keywords.Add ("single", Token.SINGLE);
 260                         keywords.Add ("sizeof", Token.SIZEOF); // Not a VB.NET Keyword
 261                         keywords.Add ("static", Token.STATIC);
 262                         keywords.Add ("step", Token.STEP);
 263                         keywords.Add ("stop", Token.STOP);
 264                         keywords.Add ("strict", Token.STRICT); // Not a VB.NET Keyword
 265                         keywords.Add ("string", Token.STRING);
 266                         keywords.Add ("structure", Token.STRUCTURE);
 267                         keywords.Add ("sub", Token.SUB);
 268                         keywords.Add ("synclock", Token.SYNCLOCK);
 269                         keywords.Add ("text", Token.TEXT); // Not a VB.NET Keyword
 270                         keywords.Add ("then", Token.THEN);
 271                         keywords.Add ("throw", Token.THROW);
 272                         keywords.Add ("to", Token.TO);
 273                         keywords.Add ("true", Token.TRUE);
 274                         keywords.Add ("try", Token.TRY);
 275                         keywords.Add ("typeof", Token.TYPEOF);
 276                         keywords.Add ("unicode", Token.UNICODE);
 277                         keywords.Add ("until", Token.UNTIL);
 278                         keywords.Add ("variant", Token.VARIANT); // An unused VB.NET keyword
 279                         keywords.Add ("wend", Token.WEND); // An unused VB.NET keyword
 280                         keywords.Add ("when", Token.WHEN);
 281                         keywords.Add ("while", Token.WHILE);
 282                         keywords.Add ("with", Token.WITH);
 283                         keywords.Add ("withevents", Token.WITHEVENTS);
 284                         keywords.Add ("writeonly", Token.WRITEONLY);
 285                         keywords.Add ("xor", Token.XOR);
 286
 287                         if (Parser.UseExtendedSyntax){
 288                                 keywords.Add ("yield", Token.YIELD);
 289                         }
 290
 291                 }
 292
 293                 static Tokenizer ()
 294                 {
 295                         initTokens ();
 296                         csharp_format_info = new NumberFormatInfo ();
 297                         csharp_format_info.CurrencyDecimalSeparator = ".";
 298                         styles = NumberStyles.AllowExponent | NumberStyles.AllowDecimalPoint;
 299                 }
 300
 301                 public Tokenizer (System.IO.TextReader input, string fname, ArrayList defines)
 302                 {
 303                         this.Source = fname;
 304
 305                         reader = input;
 306
 307                         // putback an EOL at the beginning of a stream. This is a convenience that
 308                         // allows pre-processor directives to be added to the beginning of a vb file.
 309                         putback('\n');
 310                 }
 311
 312                 bool is_keyword (string name)
 313                 {
 314                         bool res;
 315                         name = name.ToLower();
 316
 317                         res = keywords.Contains(name);
 318                         if ((name == "GET" || name == "SET") && handle_get_set == false)
 319                                 return false;
 320                         return res;
 321                 }
 322
 323                 int getKeyword (string name)
 324                 {
 325                         return (int) (keywords [name.ToLower()]);
 326                 }
 327
 328                 public Location Location {
 329                         get {
 330                                 return new Location (ref_line, col);
 331                         }
 332                 }
 333
 334                 public bool PropertyParsing {
 335                         get {
 336                                 return handle_get_set;
 337                         }
 338
 339                         set {
 340                                 handle_get_set = value;
 341                         }
 342                 }
 343
 344                 bool is_identifier_start_character (char c)
 345                 {
 346                         return Char.IsLetter (c) || c == '_' ;
 347                 }
 348
 349                 bool is_identifier_part_character (char c)
 350                 {
 351                         return (Char.IsLetter (c) || Char.IsDigit (c) || c == '_');
 352                 }
 353
 354                 int is_punct (char c, ref bool doread)
 355                 {
 356                         int d;
 357                         int t;
 358
 359                         doread = false;
 360
 361                         error_details = c.ToString();
 362
 363                         d = peekChar ();
 364
 365                         switch (c){
 366                         case '[':
 367                                 return Token.OPEN_BRACKET;
 368                         case ']':
 369                                 return Token.CLOSE_BRACKET;
 370                         case '{':
 371                                 return Token.OPEN_BRACE;
 372                         case '}':
 373                                 return Token.CLOSE_BRACE;
 374                         case '(':
 375                                 return Token.OPEN_PARENS;
 376                         case ')':
 377                                 return Token.CLOSE_PARENS;
 378                         case ',':
 379                                 return Token.COMMA;
 380                         case '?':
 381                                 return Token.INTERR;
 382                         case '!':
 383                                 if (is_identifier_start_character((char)d) || cant_have_a_type_character)
 384                                         return Token.EXCLAMATION;
 385                                 return Token.SINGLETYPECHAR;
 386                         case '$':
 387                                 if (cant_have_a_type_character)
 388                                         return Token.ERROR;
 389                                 return Token.DOLAR_SIGN;
 390                         case '@':
 391                                 if (cant_have_a_type_character)
 392                                         return Token.ERROR;
 393                                 return Token.AT_SIGN;
 394                         case '%':
 395                                 if (cant_have_a_type_character)
 396                                         return Token.ERROR;
 397                                 return Token.PERCENT;
 398                         case '#':
 399                                 if(tokens_seen)
 400                                 {
 401                                         if (cant_have_a_type_character)
 402                                                 return ExtractDateTimeLiteral();
 403                                         else
 404                                                 return Token.NUMBER_SIGN;
 405                                 }
 406                                 else
 407                                 {
 408                                         tokens_seen = true;
 409                                         return Token.HASH;
 410                                 }
 411                         case '&':
 412                                 if (!cant_have_a_type_character)
 413                                         return Token.LONGTYPECHAR;
 414                                 t = handle_integer_literal_in_other_bases(d);
 415                                 if (t == Token.NONE) {
 416                                         t = Token.OP_CONCAT;
 417                                 }
 418                                 return t;
 419                         }
 420
 421                         if (c == '+'){
 422                                 if (d == '+')
 423                                         t = Token.OP_INC;
 424                                 else
 425                                         return Token.PLUS;
 426                                 doread = true;
 427                                 return t;
 428                         }
 429                         if (c == '-'){
 430                                 return Token.MINUS;
 431                         }
 432
 433                         if (c == '='){
 434                                 return Token.ASSIGN;
 435                         }
 436
 437                         if (c == '*'){
 438                                 return Token.STAR;
 439                         }
 440
 441                         if (c == '/'){
 442                                 return Token.DIV;
 443                         }
 444
 445                         if (c == '\\'){
 446                                 return Token.OP_IDIV;
 447                         }
 448
 449                         if (c == '^'){
 450                                 return Token.OP_EXP;
 451                         }
 452
 453                         if (c == '<'){
 454                                 if (d == '>')
 455                                 {
 456                                         doread = true;
 457                                         return Token.OP_NE;
 458                                 }
 459                                 if (d == '='){
 460                                         doread = true;
 461                                         return Token.OP_LE;
 462                                 }
 463                                 if (d == '<')
 464                                 {
 465                                         doread = true;
 466                                         return Token.OP_SHIFT_LEFT;
 467                                 }
 468                                 return Token.OP_LT;
 469                         }
 470
 471                         if (c == '>'){
 472                                 if (d == '='){
 473                                         doread = true;
 474                                         return Token.OP_GE;
 475                                 }
 476                                 if (d == '>')
 477                                 {
 478                                         doread = true;
 479                                         return Token.OP_SHIFT_RIGHT;
 480                                 }
 481                                 return Token.OP_GT;
 482                         }
 483
 484                         if (c == ':'){
 485                                 if (d == '='){
 486                                         doread = true;
 487                                         return Token.ATTR_ASSIGN;
 488                                 }
 489                                 return Token.COLON;
 490                         }
 491
 492                         return Token.ERROR;
 493                 }
 494
 495                 bool decimal_digits (int c)
 496                 {
 497                         int d;
 498                         bool seen_digits = false;
 499
 500                         if (c != -1)
 501                                 number.Append ((char) c);
 502                         while ((d = peekChar ()) != -1){
 503                                 if (Char.IsDigit ((char)d)){
 504                                         number.Append ((char) d);
 505                                         getChar ();
 506                                         seen_digits = true;
 507                                 } else
 508                                         break;
 509                         }
 510                         return seen_digits;
 511                 }
 512
 513
 514                 int real_type_suffix (int c)
 515                 {
 516                         int t;
 517
 518                         switch (c){
 519                         case 'F': case 'f':
 520                                 t =  Token.LITERAL_SINGLE;
 521                                 break;
 522                         case 'R': case 'r':
 523                                 t = Token.LITERAL_DOUBLE;
 524                                 break;
 525                         case 'D': case 'd':
 526                                  t= Token.LITERAL_DECIMAL;
 527                                 break;
 528                         default:
 529                                 return Token.NONE;
 530                         }
 531                         getChar ();
 532                         return t;
 533                 }
 534
 535                 int integer_type_suffix (int c)
 536                 {
 537                         int t;
 538
 539                         try {
 540
 541                                 switch (c){
 542                                 case 'S': case 's':
 543                                         t =  Token.LITERAL_INTEGER; // SHORT ?
 544
 545                                 // hexadecimal literals - like &H8000S is "-32768"
 546                                 // and not an overflow exception
 547                                 // Check for other literals ???
 548
 549                                         if(lon == 32768) {
 550                                                 val = (short) lon;
 551                                         }
 552                                         else
 553                                                 val = ((IConvertible)val).ToInt16(null);
 554                                         break;
 555                                 case 'I': case 'i':
 556                                         t = Token.LITERAL_INTEGER;
 557                                         val = ((IConvertible)val).ToInt32(null);
 558                                         break;
 559                                 case 'L': case 'l':
 560                                          t= Token.LITERAL_INTEGER; // LONG ?
 561                                          val = ((IConvertible)val).ToInt64(null);
 562                                         break;
 563                                 default:
 564                                         if ((long)val <= System.Int32.MaxValue &&
 565                                                 (long)val >= System.Int32.MinValue) {
 566                                                 val = ((IConvertible)val).ToInt32(null);
 567                                                 return Token.LITERAL_INTEGER;
 568                                         } else {
 569                                                 val = ((IConvertible)val).ToInt64(null);
 570                                                 return Token.LITERAL_INTEGER; // LONG ?
 571                                         }
 572                                 }
 573                                 getChar ();
 574                                 return t;
 575                         } catch (Exception e) {
 576                                 val = e.ToString();
 577                                 return Token.ERROR;
 578                         }
 579                 }
 580
 581                 int adjust_real (int t)
 582                 {
 583                         string s = number.ToString ();
 584
 585                         switch (t){
 586                         case Token.LITERAL_DECIMAL:
 587                                 val = new System.Decimal ();
 588                                 val = System.Decimal.Parse (
 589                                         s, styles, csharp_format_info);
 590                                 break;
 591                         case Token.LITERAL_DOUBLE:
 592                                 val = new System.Double ();
 593                                 val = System.Double.Parse (
 594                                         s, styles, csharp_format_info);
 595                                 break;
 596                         case Token.LITERAL_SINGLE:
 597                                 val = new System.Double ();
 598                                 val = (float) System.Double.Parse (
 599                                         s, styles, csharp_format_info);
 600                                 break;
 601
 602                         case Token.NONE:
 603                                 val = new System.Double ();
 604                                 val = System.Double.Parse (
 605                                         s, styles, csharp_format_info);
 606                                 t = Token.LITERAL_DOUBLE;
 607                                 break;
 608                         }
 609                         return t;
 610                 }
 611
 612                 long hex_digits ()
 613                 {
 614                         StringBuilder hexNumber = new StringBuilder ();
 615
 616                         int d;
 617
 618                         while ((d = peekChar ()) != -1){
 619                                 char e = Char.ToUpper ((char) d);
 620
 621                                 if (Char.IsDigit (e) || (e >= 'A' && e <= 'F')){
 622                                         hexNumber.Append (e);
 623                                         getChar ();
 624                                 } else
 625                                         break;
 626                         }
 627                         lon = System.Int64.Parse (hexNumber.ToString(), NumberStyles.HexNumber);
 628                         return lon;
 629                 }
 630
 631                 long octal_digits ()
 632                 {
 633                         long valueToReturn = 0;
 634
 635                         int d;
 636
 637                         while ((d = peekChar ()) != -1){
 638                                 char e = (char)d;
 639                                 if (Char.IsDigit (e) && (e < '8')){
 640                                         valueToReturn *= 8;
 641                                         valueToReturn += (d - (int)'0');
 642                                         getChar ();
 643                                 } else
 644                                         break;
 645                         }
 646
 647                         return valueToReturn;
 648                 }
 649
 650                 int handle_integer_literal_in_other_bases(int peek)
 651                 {
 652                         if (peek == 'h' || peek == 'H'){
 653                                 getChar ();
 654                                 val = hex_digits ();
 655                                 return integer_type_suffix (peekChar ());
 656                         }
 657
 658                         if (peek == 'o' || peek == 'O'){
 659                                 getChar ();
 660                                 val = octal_digits ();
 661                                 return integer_type_suffix (peekChar ());
 662                         }
 663
 664                         return Token.NONE;
 665                 }
 666
 667                 //
 668                 // Invoked if we know we have .digits or digits
 669                 //
 670                 int is_number (int c)
 671                 {
 672                         bool is_real = false;
 673                         number = new StringBuilder ();
 674                         int type;
 675                         bool non_prefixdecimal = false; //To capture decimals like .50
 676
 677                         number.Length = 0;
 678
 679                         if (Char.IsDigit ((char)c)){
 680                                 decimal_digits (c);
 681                                 c = peekChar ();
 682                                 non_prefixdecimal = true;
 683                         }
 684
 685                         //
 686                         // We need to handle the case of
 687                         // "1.1" vs "1.ToString()" (LITERAL_SINGLE vs NUMBER DOT IDENTIFIER)
 688                         //
 689                         if (c == '.'){
 690                                 if (non_prefixdecimal == false)
 691                                          putback ('.');
 692                                 if (decimal_digits (getChar())){
 693                                         is_real = true;
 694                                         c = peekChar ();
 695                                 } else {
 696                                         putback ('.');
 697                                         number.Length -= 1;
 698                                         val = System.Int64.Parse(number.ToString());
 699                                         return integer_type_suffix('.');
 700                                 }
 701                         }
 702
 703                         if (c == 'e' || c == 'E'){
 704                                 is_real = true;
 705                                 number.Append ("e");
 706                                 getChar ();
 707
 708                                 c = peekChar ();
 709                                 if (c == '+'){
 710                                         number.Append ((char) c);
 711                                         getChar ();
 712                                         c = peekChar ();
 713                                 } else if (c == '-'){
 714                                         number.Append ((char) c);
 715                                         getChar ();
 716                                         c = peekChar ();
 717                                 }
 718                                 decimal_digits (-1);
 719                                 c = peekChar ();
 720                         }
 721
 722                         type = real_type_suffix (c);
 723                         if (type == Token.NONE && !is_real){
 724                                 val = System.Int64.Parse(number.ToString());
 725                                 return integer_type_suffix(c);
 726                         }
 727
 728                         return adjust_real (type);
 729                 }
 730
 731                 int getChar ()
 732                 {
 733                         if (putback_char != -1){
 734                                 int x = putback_char;
 735                                 putback_char = -1;
 736
 737                                 return x;
 738                         }
 739                         return reader.Read ();
 740                 }
 741
 742                 int peekChar ()
 743                 {
 744                         if (putback_char != -1)
 745                                 return putback_char;
 746                         return reader.Peek ();
 747                 }
 748
 749
 750                 void putback (int c)
 751                 {
 752                         if (putback_char != -1)
 753                                 throw new Exception ("This should not happen putback on putback");
 754                         putback_char = c;
 755                 }
 756
 757                 public bool advance ()
 758                 {
 759                         return current_token != Token.EOF ;
 760                 }
 761
 762                 public Object Value {
 763                         get {
 764                                 return val;
 765                         }
 766                 }
 767
 768                 public Object value ()
 769                 {
 770                         return val;
 771                 }
 772
 773                 private bool IsEOL(int currentChar)
 774                 {
 775                         bool retVal;
 776
 777                         if (currentChar ==  0x0D) {
 778                                 if (peekChar() ==  0x0A) // if it is a CR-LF pair consume LF also
 779                                         getChar();
 780
 781                                 retVal = true;
 782                         }
 783                         else {
 784                                 retVal = (currentChar ==  -1 || currentChar ==  0x0A || currentChar ==  0x2028 || currentChar ==  0x2029);
 785                         }
 786
 787                         if(retVal) {
 788                                 nextLine();
 789                         }
 790
 791                         return retVal;
 792                 }
 793
 794                 private int DropComments()
 795                 {
 796                         //int d;
 797                         while (!IsEOL(/*d =*/ getChar ()))
 798                                 col++;
 799
 800                         return Token.EOL;
 801                 }
 802
 803                 public bool putbacktoken = false;
 804                 public bool flag = false;
 805                 int next_token;
 806
 807                 public int token ()
 808                 {
 809                         int lastToken = current_token;
 810                         do
 811                         {
 812                                 current_token = xtoken ();
 813                                 if(current_token == Token.END) {
 814                                         next_token = xtoken();
 815                                         putbacktoken = true;
 816                                         if (next_token == Token.EOL)
 817                                                 return Token.END_EOL;
 818                                          else
 819                                                 return Token.END;
 820                                 }
 821                                 if (current_token == 0)
 822                                         return Token.EOF;
 823                                 if (current_token == Token.REM)
 824                                         current_token = DropComments();
 825                         } while (lastToken == Token.EOL && current_token == Token.EOL);
 826
 827                         return current_token;
 828                 }
 829
 830                 private string GetIdentifier()
 831                 {
 832                         int c = getChar();
 833                         if (is_identifier_start_character ((char) c))
 834                                 return GetIdentifier(c);
 835                         else
 836                                 return null;
 837                 }
 838
 839                 private string GetIdentifier(int c)
 840                 {
 841                         StringBuilder id = new StringBuilder ();
 842
 843                         id.Append ((char) c);
 844
 845                         while ((c = peekChar ()) != -1)
 846                         {
 847                                 if (is_identifier_part_character ((char) c))
 848                                 {
 849                                         id.Append ((char)getChar ());
 850                                         col++;
 851                                 }
 852                                 else
 853                                         break;
 854                         }
 855
 856                         cant_have_a_type_character = false;
 857
 858                         return id.ToString();
 859                 }
 860
 861                 private bool is_doublequote(int currentChar)
 862                 {
 863                         return (currentChar == '"' ||
 864                                         currentChar == 0x201C || // unicode left double-quote character
 865                                         currentChar == 0x201D);  // unicode right double-quote character
 866                 }
 867
 868                 private bool is_whitespace(int c)
 869                 {
 870                         return (c == ' ' || c == '\t' || c == '\v' || c == '\r' || c == 0xa0);
 871                 }
 872
 873                 private bool tokens_seen = false;
 874
 875                 private void nextLine()
 876                 {
 877                         cant_have_a_type_character = true;
 878                         line++;
 879                         ref_line++;
 880                         col = 0;
 881                         tokens_seen = false;
 882                 }
 883
 884                 public int xtoken ()
 885                 {
 886                         int t;
 887                         bool doread = false;
 888                         int c;
 889
 890                         if (putbacktoken == true) {
 891                                 putbacktoken = false;
 892                                 return next_token;
 893                         }
 894
 895                         val = null;
 896                         for (;(c = getChar ()) != -1; col++) {
 897
 898                                 // Handle line continuation character
 899                                 if (c == '_')
 900                                 {
 901                                         int d = peekChar();
 902                                         if (!is_identifier_part_character((char)d)) {
 903                                                 while ((c = getChar ()) != -1 && !IsEOL(c)) {}
 904                                                 c = getChar ();
 905                                         }
 906                                 }
 907
 908
 909                                 // white space
 910                                 if (is_whitespace(c)) {
 911                                         // expand tabs for location
 912                                         if (c == '\t')
 913                                                 col = (((col + ExpandedTabsSize) / ExpandedTabsSize) * ExpandedTabsSize) - 1;
 914                                         cant_have_a_type_character = true;
 915                                         continue;
 916                                 }
 917
 918                                 // Handle line comments.
 919                                 if (c == '\'')
 920                                         return Token.REM;
 921
 922                                 // Handle EOL.
 923                                 if (IsEOL(c))
 924                                 {
 925                                         if (current_token == Token.EOL) // if last token was also EOL keep skipping
 926                                                 continue;
 927                                         return Token.EOL;
 928                                 }
 929
 930                                 // Handle escaped identifiers
 931                                 if (c == '[')
 932                                 {
 933                                         if ((val = GetIdentifier()) == null)
 934                                                 break;
 935                                         if ((c = getChar()) != ']')
 936                                                 break;
 937                                         tokens_seen = true;
 938                                         return Token.IDENTIFIER;
 939                                 }
 940
 941                                 // Handle unescaped identifiers
 942                                 if (is_identifier_start_character ((char) c))
 943                                 {
 944                                         string id;
 945                                         if ((id = GetIdentifier(c)) == null)
 946                                                 break;
 947                                         val = id;
 948                                         tokens_seen = true;
 949                                         if (is_keyword(id) && (current_token != Token.DOT))
 950                                                 return getKeyword(id);
 951                                         return Token.IDENTIFIER;
 952                                 }
 953
 954                                 // Treat string literals
 955                                 if (is_doublequote(c)) {
 956                                         cant_have_a_type_character = true;
 957                                         return ExtractStringOrCharLiteral(c);
 958                                 }
 959
 960                                 // handle numeric literals
 961
 962                                 if (Char.IsDigit ((char) c))
 963                                 {
 964                                         cant_have_a_type_character = true;
 965                                         tokens_seen = true;
 966                                         return is_number (c);
 967                                 }
 968
 969                                 if (c == '.')
 970                                 {
 971                                         cant_have_a_type_character = true;
 972                                         tokens_seen = true;
 973                                         if (Char.IsDigit ((char) peekChar ()))
 974                                                 return is_number (c);
 975                                         return Token.DOT;
 976                                 }
 977                                 if ((t = is_punct ((char)c, ref doread)) != Token.ERROR) {
 978                                         cant_have_a_type_character = true;
 979
 980                                         if (t == Token.NONE)
 981                                                 continue;
 982
 983                                         if (doread){
 984                                                 getChar ();
 985                                                 col++;
 986                                         }
 987                                         tokens_seen = true;
 988                                         return t;
 989                                 }
 990
 991                                 error_details = ((char)c).ToString ();
 992                                 return Token.ERROR;
 993                         }
 994
 995                         if (current_token != Token.EOL) // if last token wasn't EOL send it before EOF
 996                                 return Token.EOL;
 997
 998                         return Token.EOF;
 999                 }
1000
1001                 private int ExtractDateTimeLiteral()
1002                 {
1003                         int c;
1004
1005                         StringBuilder sb = new StringBuilder();
1006                         for (;(c = getChar ()) != -1; col++)
1007                         {
1008                                 if (c == '#') {
1009                                         val = ParseDateLiteral(sb);
1010                                         return Token.LITERAL_DATE;
1011                                 }
1012                                 if (IsEOL(c)) {
1013                                         break;
1014                                 }
1015                                 if (c == '-')
1016                                         c = '/';
1017                                 sb.Append((char)c);
1018                         }
1019                         return Token.ERROR;
1020                 }
1021
1022                 private int ExtractStringOrCharLiteral(int c)
1023                 {
1024                         StringBuilder s = new StringBuilder ();
1025
1026                         tokens_seen = true;
1027
1028                         while ((c = getChar ()) != -1){
1029                                 if (is_doublequote(c)){
1030                                         if (is_doublequote(peekChar()))
1031                                                 getChar();
1032                                         else {
1033                                                 //handle Char Literals
1034                                                 if (peekChar() == 'C' || peekChar() == 'c') {
1035                                                         getChar();
1036                                                         if (s.Length == 1) {
1037                                                                 val = s[0];
1038                                                                 return Token.LITERAL_CHARACTER;
1039                                                         } else {
1040                                                                 val = "Incorrect length for a character literal";
1041                                                                 return Token.ERROR;
1042                                                         }
1043                                                 } else {
1044                                                         val = s.ToString ();
1045                                                         return Token.LITERAL_STRING;
1046                                                 }
1047                                         }
1048                                 }
1049
1050                                 if (IsEOL(c)) {
1051                                         return Token.ERROR;
1052                                 }
1053
1054                                 s.Append ((char) c);
1055                         }
1056
1057                         return Token.ERROR;
1058                 }
1059
1060                 static IFormatProvider enUSculture = new CultureInfo("en-US", true);
1061
1062                 private DateTime ParseDateLiteral(StringBuilder value)
1063                 {
1064                         try
1065                         {
1066                                 return DateTime.Parse(value.ToString(),
1067                                                   enUSculture,
1068                                                   DateTimeStyles.NoCurrentDateDefault | DateTimeStyles.AllowWhiteSpaces);
1069                         }
1070                         catch (FormatException ex)
1071                         {
1072                                 //TODO: What is the correct error number and message?
1073                                 Report.Error (1, Location, string.Format("Invalid date literal '{0}'", value.ToString())
1074                                         + Environment.NewLine + ex.ToString());
1075                         }
1076                         catch (Exception)
1077                         {
1078                                 Report.Error (1, Location, "Error parsing date literal");       //TODO: What is the correct error number and message?
1079                         }
1080                         return new DateTime();
1081                 }
1082
1083                 public void PositionCursorAtNextPreProcessorDirective()
1084                 {
1085                         int t;
1086
1087                         for(t = token(); t != Token.HASH && t != Token.EOF ; t = token());
1088
1089                         if(t == Token.EOF)
1090                                 throw new ApplicationException("Unexpected EOF while looking for a pre-processor directive");
1091
1092                         if(t == Token.HASH) {
1093                                 tokens_seen = false;
1094                                 putback('#');
1095                         }
1096                 }
1097
1098         }
1099 }