mcs/mbas/mb-tokenizer.cs

   1 //
   2 // Mono.MonoBASIC.Tokenizer.cs: The Tokenizer for the MonoBASIC compiler
   3 //
   4 // Author: A Rafael D Teixeira (rafaelteixeirabr@hotmail.com)
   5 //       : Manjula GHM (mmanjula@novell.com)
   6 // Based on cs-tokenizer.cs by Miguel de Icaza (miguel@gnu.org)
   7 //
   8 // Licensed under the terms of the GNU GPL
   9 //
  10 // Copyright (C) 2001 A Rafael D Teixeira
  11 //
  12
  13 namespace Mono.MonoBASIC
  14 {
  15         using System;
  16         using System.Text;
  17         using System.Collections;
  18         using System.IO;
  19         using System.Globalization;
  20         using Mono.Languages;
  21         using Mono.MonoBASIC;
  22
  23         /// <summary>
  24         ///    Tokenizer for MonoBASIC source code.
  25         /// </summary>
  26
  27         public class Tokenizer : yyParser.yyInput
  28         {
  29                 TextReader reader;
  30                 string file_name;
  31                 string ref_name;
  32                 int ref_line = 0;
  33                 int line = 0;
  34                 int col = 1;
  35                 public int current_token = Token.ERROR;
  36                 bool handle_get_set = false;
  37                 bool cant_have_a_type_character = false;
  38
  39                 public int ExpandedTabsSize = 4;
  40
  41                 public string location {
  42                         get {
  43                                 string det;
  44
  45                                 if (current_token == Token.ERROR)
  46                                         det = "detail: " + error_details;
  47                                 else
  48                                         det = "";
  49
  50                                 return "Line:     "+line+" Col: "+col + "\n" +
  51                                        "VirtLine: "+ref_line +
  52                                        " Token: "+current_token + " " + det;
  53                         }
  54                 }
  55
  56                 public bool properties {
  57                         get {
  58                                 return handle_get_set;
  59                         }
  60
  61                         set {
  62                                 handle_get_set = value;
  63                         }
  64                 }
  65
  66                 //
  67                 // Class variables
  68                 //
  69                 static Hashtable keywords;
  70                 static NumberStyles styles;
  71                 static NumberFormatInfo csharp_format_info;
  72
  73                 //
  74                 // Values for the associated token returned
  75                 //
  76                 StringBuilder number;
  77                 int putback_char = -1;
  78                 Object val;
  79
  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                         while ((d = peekChar ()) != -1){
 502                                 if (Char.IsDigit ((char)d)){
 503                                         number.Append ((char) d);
 504                                         getChar ();
 505                                         seen_digits = true;
 506                                 } else
 507                                         break;
 508                         }
 509                         return seen_digits;
 510                 }
 511
 512
 513                 int real_type_suffix (int c)
 514                 {
 515                         int t;
 516
 517                         switch (c){
 518                         case 'F': case 'f':
 519                                 t =  Token.LITERAL_SINGLE;
 520                                 break;
 521                         case 'R': case 'r':
 522                                 t = Token.LITERAL_DOUBLE;
 523                                 break;
 524                         case 'D': case 'd':
 525                                  t= Token.LITERAL_DECIMAL;
 526                                 break;
 527                         default:
 528                                 return Token.NONE;
 529                         }
 530                         getChar ();
 531                         return t;
 532                 }
 533
 534                 int integer_type_suffix (int c)
 535                 {
 536                         int t;
 537
 538                         try {
 539
 540                                 switch (c){
 541                                 case 'S': case 's':
 542                                         t =  Token.LITERAL_INTEGER; // SHORT ?
 543                                         val = ((IConvertible)val).ToInt16(null);
 544                                         break;
 545                                 case 'I': case 'i':
 546                                         t = Token.LITERAL_INTEGER;
 547                                         val = ((IConvertible)val).ToInt32(null);
 548                                         break;
 549                                 case 'L': case 'l':
 550                                          t= Token.LITERAL_INTEGER; // LONG ?
 551                                          val = ((IConvertible)val).ToInt64(null);
 552                                         break;
 553                                 default:
 554                                         if ((long)val <= System.Int32.MaxValue &&
 555                                                 (long)val >= System.Int32.MinValue) {
 556                                                 val = ((IConvertible)val).ToInt32(null);
 557                                                 return Token.LITERAL_INTEGER;
 558                                         } else {
 559                                                 val = ((IConvertible)val).ToInt64(null);
 560                                                 return Token.LITERAL_INTEGER; // LONG ?
 561                                         }
 562                                 }
 563                                 getChar ();
 564                                 return t;
 565                         } catch (Exception e) {
 566                                 val = e.ToString();
 567                                 return Token.ERROR;
 568                         }
 569                 }
 570
 571                 int adjust_real (int t)
 572                 {
 573                         string s = number.ToString ();
 574
 575                         switch (t){
 576                         case Token.LITERAL_DECIMAL:
 577                                 val = new System.Decimal ();
 578                                 val = System.Decimal.Parse (
 579                                         s, styles, csharp_format_info);
 580                                 break;
 581                         case Token.LITERAL_DOUBLE:
 582                                 val = new System.Double ();
 583                                 val = System.Double.Parse (
 584                                         s, styles, csharp_format_info);
 585                                 break;
 586                         case Token.LITERAL_SINGLE:
 587                                 val = new System.Double ();
 588                                 val = (float) System.Double.Parse (
 589                                         s, styles, csharp_format_info);
 590                                 break;
 591
 592                         case Token.NONE:
 593                                 val = new System.Double ();
 594                                 val = System.Double.Parse (
 595                                         s, styles, csharp_format_info);
 596                                 t = Token.LITERAL_DOUBLE;
 597                                 break;
 598                         }
 599                         return t;
 600                 }
 601
 602                 long hex_digits ()
 603                 {
 604                         StringBuilder hexNumber = new StringBuilder ();
 605
 606                         int d;
 607
 608                         while ((d = peekChar ()) != -1){
 609                                 char e = Char.ToUpper ((char) d);
 610
 611                                 if (Char.IsDigit (e) || (e >= 'A' && e <= 'F')){
 612                                         hexNumber.Append (e);
 613                                         getChar ();
 614                                 } else
 615                                         break;
 616                         }
 617                         return System.Int64.Parse (hexNumber.ToString(), NumberStyles.HexNumber);
 618                 }
 619
 620                 long octal_digits ()
 621                 {
 622                         long valueToReturn = 0;
 623
 624                         int d;
 625
 626                         while ((d = peekChar ()) != -1){
 627                                 char e = (char)d;
 628                                 if (Char.IsDigit (e) && (e < '8')){
 629                                         valueToReturn *= 8;
 630                                         valueToReturn += (d - (int)'0');
 631                                         getChar ();
 632                                 } else
 633                                         break;
 634                         }
 635
 636                         return valueToReturn;
 637                 }
 638
 639                 int handle_integer_literal_in_other_bases(int peek)
 640                 {
 641                         if (peek == 'h' || peek == 'H'){
 642                                 getChar ();
 643                                 val = hex_digits ();
 644                                 return integer_type_suffix (peekChar ());
 645                         }
 646
 647                         if (peek == 'o' || peek == 'O'){
 648                                 getChar ();
 649                                 val = octal_digits ();
 650                                 return integer_type_suffix (peekChar ());
 651                         }
 652
 653                         return Token.NONE;
 654                 }
 655
 656                 //
 657                 // Invoked if we know we have .digits or digits
 658                 //
 659                 int is_number (int c)
 660                 {
 661                         bool is_real = false;
 662                         number = new StringBuilder ();
 663                         int type;
 664                         bool non_prefixdecimal = false; //To capture decimals like .50
 665
 666                         number.Length = 0;
 667
 668                         if (Char.IsDigit ((char)c)){
 669                                 decimal_digits (c);
 670                                 c = peekChar ();
 671                                 non_prefixdecimal = true;
 672                         }
 673
 674                         //
 675                         // We need to handle the case of
 676                         // "1.1" vs "1.ToString()" (LITERAL_SINGLE vs NUMBER DOT IDENTIFIER)
 677                         //
 678                         if (c == '.'){
 679                                 if (non_prefixdecimal == false)
 680                                          putback ('.');
 681                                 if (decimal_digits (getChar())){
 682                                         is_real = true;
 683                                         c = peekChar ();
 684                                 } else {
 685                                         putback ('.');
 686                                         number.Length -= 1;
 687                                         val = System.Int64.Parse(number.ToString());
 688                                         return integer_type_suffix('.');
 689                                 }
 690                         }
 691
 692                         if (c == 'e' || c == 'E'){
 693                                 is_real = true;
 694                                 number.Append ("e");
 695                                 getChar ();
 696
 697                                 c = peekChar ();
 698                                 if (c == '+'){
 699                                         number.Append ((char) c);
 700                                         getChar ();
 701                                         c = peekChar ();
 702                                 } else if (c == '-'){
 703                                         number.Append ((char) c);
 704                                         getChar ();
 705                                         c = peekChar ();
 706                                 }
 707                                 decimal_digits (-1);
 708                                 c = peekChar ();
 709                         }
 710
 711                         type = real_type_suffix (c);
 712                         if (type == Token.NONE && !is_real){
 713                                 val = System.Int64.Parse(number.ToString());
 714                                 return integer_type_suffix(c);
 715                         }
 716
 717                         return adjust_real (type);
 718                 }
 719
 720                 int getChar ()
 721                 {
 722                         if (putback_char != -1){
 723                                 int x = putback_char;
 724                                 putback_char = -1;
 725
 726                                 return x;
 727                         }
 728                         return reader.Read ();
 729                 }
 730
 731                 int peekChar ()
 732                 {
 733                         if (putback_char != -1)
 734                                 return putback_char;
 735                         return reader.Peek ();
 736                 }
 737
 738
 739                 void putback (int c)
 740                 {
 741                         if (putback_char != -1)
 742                                 throw new Exception ("This should not happen putback on putback");
 743                         putback_char = c;
 744                 }
 745
 746                 public bool advance ()
 747                 {
 748                         return current_token != Token.EOF ;
 749                 }
 750
 751                 public Object Value {
 752                         get {
 753                                 return val;
 754                         }
 755                 }
 756
 757                 public Object value ()
 758                 {
 759                         return val;
 760                 }
 761
 762                 private bool IsEOL(int currentChar)
 763                 {
 764                         bool retVal;
 765
 766                         if (currentChar ==  0x0D) {
 767                                 if (peekChar() ==  0x0A) // if it is a CR-LF pair consume LF also
 768                                         getChar();
 769
 770                                 retVal = true;
 771                         }
 772                         else {
 773                                 retVal = (currentChar ==  -1 || currentChar ==  0x0A || currentChar ==  0x2028 || currentChar ==  0x2029);
 774                         }
 775
 776                         if(retVal) {
 777                                 nextLine();
 778                         }
 779
 780                         return retVal;
 781                 }
 782
 783                 private int DropComments()
 784                 {
 785                         //int d;
 786                         while (!IsEOL(/*d =*/ getChar ()))
 787                                 col++;
 788
 789                         return Token.EOL;
 790                 }
 791
 792                 public bool putbacktoken = false;
 793                 public bool flag = false;
 794                 int next_token;
 795
 796                 public int token ()
 797                 {
 798                         int lastToken = current_token;
 799                         do
 800                         {
 801                                 current_token = xtoken ();
 802                                 if(current_token == Token.END) {
 803                                         next_token = xtoken();
 804                                         putbacktoken = true;
 805                                         if (next_token == Token.EOL)
 806                                                 return Token.END_EOL;
 807                                          else
 808                                                 return Token.END;
 809                                 }
 810                                 if (current_token == 0)
 811                                         return Token.EOF;
 812                                 if (current_token == Token.REM)
 813                                         current_token = DropComments();
 814                         } while (lastToken == Token.EOL && current_token == Token.EOL);
 815
 816                         return current_token;
 817                 }
 818
 819                 private string GetIdentifier()
 820                 {
 821                         int c = getChar();
 822                         if (is_identifier_start_character ((char) c))
 823                                 return GetIdentifier(c);
 824                         else
 825                                 return null;
 826                 }
 827
 828                 private string GetIdentifier(int c)
 829                 {
 830                         StringBuilder id = new StringBuilder ();
 831
 832                         id.Append ((char) c);
 833
 834                         while ((c = peekChar ()) != -1)
 835                         {
 836                                 if (is_identifier_part_character ((char) c))
 837                                 {
 838                                         id.Append ((char)getChar ());
 839                                         col++;
 840                                 }
 841                                 else
 842                                         break;
 843                         }
 844
 845                         cant_have_a_type_character = false;
 846
 847                         return id.ToString();
 848                 }
 849
 850                 private bool is_doublequote(int currentChar)
 851                 {
 852                         return (currentChar == '"' ||
 853                                         currentChar == 0x201C || // unicode left double-quote character
 854                                         currentChar == 0x201D);  // unicode right double-quote character
 855                 }
 856
 857                 private bool is_whitespace(int c)
 858                 {
 859                         return (c == ' ' || c == '\t' || c == '\v' || c == '\r' || c == 0xa0);
 860                 }
 861
 862                 private bool tokens_seen = false;
 863
 864                 private void nextLine()
 865                 {
 866                         cant_have_a_type_character = true;
 867                         line++;
 868                         ref_line++;
 869                         col = 0;
 870                         tokens_seen = false;
 871                 }
 872
 873                 public int xtoken ()
 874                 {
 875                         int t;
 876                         bool doread = false;
 877                         int c;
 878
 879                         if (putbacktoken == true) {
 880                                 putbacktoken = false;
 881                                 return next_token;
 882                         }
 883
 884                         val = null;
 885                         for (;(c = getChar ()) != -1; col++) {
 886
 887                                 // Handle line continuation character
 888                                 if (c == '_')
 889                                 {
 890                                         int d = peekChar();
 891                                         if (!is_identifier_part_character((char)d)) {
 892                                                 while ((c = getChar ()) != -1 && !IsEOL(c)) {}
 893                                                 c = getChar ();
 894                                         }
 895                                 }
 896
 897
 898                                 // white space
 899                                 if (is_whitespace(c)) {
 900                                         // expand tabs for location
 901                                         if (c == '\t')
 902                                                 col = (((col + ExpandedTabsSize) / ExpandedTabsSize) * ExpandedTabsSize) - 1;
 903                                         cant_have_a_type_character = true;
 904                                         continue;
 905                                 }
 906
 907                                 // Handle line comments.
 908                                 if (c == '\'')
 909                                         return Token.REM;
 910
 911                                 // Handle EOL.
 912                                 if (IsEOL(c))
 913                                 {
 914                                         if (current_token == Token.EOL) // if last token was also EOL keep skipping
 915                                                 continue;
 916                                         return Token.EOL;
 917                                 }
 918
 919                                 // Handle escaped identifiers
 920                                 if (c == '[')
 921                                 {
 922                                         if ((val = GetIdentifier()) == null)
 923                                                 break;
 924                                         if ((c = getChar()) != ']')
 925                                                 break;
 926                                         tokens_seen = true;
 927                                         return Token.IDENTIFIER;
 928                                 }
 929
 930                                 // Handle unescaped identifiers
 931                                 if (is_identifier_start_character ((char) c))
 932                                 {
 933                                         string id;
 934                                         if ((id = GetIdentifier(c)) == null)
 935                                                 break;
 936                                         val = id;
 937                                         tokens_seen = true;
 938                                         if (is_keyword(id) && (current_token != Token.DOT))
 939                                                 return getKeyword(id);
 940                                         return Token.IDENTIFIER;
 941                                 }
 942
 943                                 // Treat string literals
 944                                 if (is_doublequote(c)) {
 945                                         cant_have_a_type_character = true;
 946                                         return ExtractStringOrCharLiteral(c);
 947                                 }
 948
 949                                 // handle numeric literals
 950
 951                                 if (Char.IsDigit ((char) c))
 952                                 {
 953                                         cant_have_a_type_character = true;
 954                                         tokens_seen = true;
 955                                         return is_number (c);
 956                                 }
 957
 958                                 if (c == '.')
 959                                 {
 960                                         cant_have_a_type_character = true;
 961                                         tokens_seen = true;
 962                                         if (Char.IsDigit ((char) peekChar ()))
 963                                                 return is_number (c);
 964                                         return Token.DOT;
 965                                 }
 966                                 if ((t = is_punct ((char)c, ref doread)) != Token.ERROR) {
 967                                         cant_have_a_type_character = true;
 968
 969                                         if (t == Token.NONE)
 970                                                 continue;
 971
 972                                         if (doread){
 973                                                 getChar ();
 974                                                 col++;
 975                                         }
 976                                         tokens_seen = true;
 977                                         return t;
 978                                 }
 979
 980                                 error_details = ((char)c).ToString ();
 981                                 return Token.ERROR;
 982                         }
 983
 984                         if (current_token != Token.EOL) // if last token wasn't EOL send it before EOF
 985                                 return Token.EOL;
 986
 987                         return Token.EOF;
 988                 }
 989
 990                 private int ExtractDateTimeLiteral()
 991                 {
 992                         int c;
 993
 994                         StringBuilder sb = new StringBuilder();
 995                         for (;(c = getChar ()) != -1; col++)
 996                         {
 997                                 if (c == '#') {
 998                                         val = ParseDateLiteral(sb);
 999                                         return Token.LITERAL_DATE;
1000                                 }
1001                                 if (IsEOL(c)) {
1002                                         break;
1003                                 }
1004                                 if (c == '-')
1005                                         c = '/';
1006                                 sb.Append((char)c);
1007                         }
1008                         return Token.ERROR;
1009                 }
1010
1011                 private int ExtractStringOrCharLiteral(int c)
1012                 {
1013                         StringBuilder s = new StringBuilder ();
1014
1015                         tokens_seen = true;
1016
1017                         while ((c = getChar ()) != -1){
1018                                 if (is_doublequote(c)){
1019                                         if (is_doublequote(peekChar()))
1020                                                 getChar();
1021                                         else {
1022                                                 //handle Char Literals
1023                                                 if (peekChar() == 'C' || peekChar() == 'c') {
1024                                                         getChar();
1025                                                         if (s.Length == 1) {
1026                                                                 val = s[0];
1027                                                                 return Token.LITERAL_CHARACTER;
1028                                                         } else {
1029                                                                 val = "Incorrect length for a character literal";
1030                                                                 return Token.ERROR;
1031                                                         }
1032                                                 } else {
1033                                                         val = s.ToString ();
1034                                                         return Token.LITERAL_STRING;
1035                                                 }
1036                                         }
1037                                 }
1038
1039                                 if (IsEOL(c)) {
1040                                         return Token.ERROR;
1041                                 }
1042
1043                                 s.Append ((char) c);
1044                         }
1045
1046                         return Token.ERROR;
1047                 }
1048
1049                 static IFormatProvider enUSculture = new CultureInfo("en-US", true);
1050
1051                 private DateTime ParseDateLiteral(StringBuilder value)
1052                 {
1053                         try
1054                         {
1055                                 return DateTime.Parse(value.ToString(),
1056                                                   enUSculture,
1057                                                   DateTimeStyles.NoCurrentDateDefault | DateTimeStyles.AllowWhiteSpaces);
1058                         }
1059                         catch (FormatException ex)
1060                         {
1061                                 //TODO: What is the correct error number and message?
1062                                 Report.Error (1, Location, string.Format("Invalid date literal '{0}'", value.ToString())
1063                                         + Environment.NewLine + ex.ToString());
1064                         }
1065                         catch (Exception)
1066                         {
1067                                 Report.Error (1, Location, "Error parsing date literal");       //TODO: What is the correct error number and message?
1068                         }
1069                         return new DateTime();
1070                 }
1071
1072                 public void PositionCursorAtNextPreProcessorDirective()
1073                 {
1074                         int t;
1075
1076                         for(t = token(); t != Token.HASH && t != Token.EOF ; t = token());
1077
1078                         if(t == Token.EOF)
1079                                 throw new ApplicationException("Unexpected EOF while looking for a pre-processor directive");
1080
1081                         if(t == Token.HASH) {
1082                                 tokens_seen = false;
1083                                 putback('#');
1084                         }
1085                 }
1086
1087         }
1088 }