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