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