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