This commit was manufactured by cvs2svn to create branch 'mono-1-0'.
[mono.git] / mcs / gmcs / cs-tokenizer.cs
1 // -*- coding: dos -*-\r
2 //\r
3 // cs-tokenizer.cs: The Tokenizer for the C# compiler\r
4 //                  This also implements the preprocessor\r
5 //\r
6 // Author: Miguel de Icaza (miguel@gnu.org)\r
7 //\r
8 // Licensed under the terms of the GNU GPL\r
9 //\r
10 // (C) 2001, 2002 Ximian, Inc (http://www.ximian.com)\r
11 //\r
12 \r
13 /*\r
14  * TODO:\r
15  *   Make sure we accept the proper Unicode ranges, per the spec.\r
16  *   Report error 1032\r
17 */\r
18 \r
19 using System;\r
20 using System.Text;\r
21 using System.Collections;\r
22 using System.IO;\r
23 using System.Globalization;\r
24 using System.Reflection;\r
25 \r
26 namespace Mono.CSharp\r
27 {\r
28         /// <summary>\r
29         ///    Tokenizer for C# source code. \r
30         /// </summary>\r
31 \r
32         public class Tokenizer : yyParser.yyInput\r
33         {\r
34                 SeekableStreamReader reader;\r
35                 public SourceFile ref_name;\r
36                 public SourceFile file_name;\r
37                 public int ref_line = 1;\r
38                 public int line = 1;\r
39                 public int col = 1;\r
40                 public int current_token;\r
41                 bool handle_get_set = false;\r
42                 bool handle_remove_add = false;\r
43                 bool handle_assembly = false;\r
44                 bool handle_constraints = false;\r
45 \r
46                 //\r
47                 // Whether tokens have been seen on this line\r
48                 //\r
49                 bool tokens_seen = false;\r
50 \r
51                 //\r
52                 // Whether a token has been seen on the file\r
53                 // This is needed because `define' is not allowed to be used\r
54                 // after a token has been seen.\r
55                 //\r
56                 bool any_token_seen = false;\r
57                 static Hashtable tokenValues;\r
58                 \r
59                 private static Hashtable TokenValueName\r
60                 {\r
61                         get {\r
62                                 if (tokenValues == null)\r
63                                         tokenValues = GetTokenValueNameHash ();\r
64 \r
65                                 return tokenValues;\r
66                         }\r
67                 }\r
68 \r
69                 private static Hashtable GetTokenValueNameHash ()\r
70                 {\r
71                         Type t = typeof (Token);\r
72                         FieldInfo [] fields = t.GetFields ();\r
73                         Hashtable hash = new Hashtable ();\r
74                         foreach (FieldInfo field in fields) {\r
75                                 if (field.IsLiteral && field.IsStatic && field.FieldType == typeof (int))\r
76                                         hash.Add (field.GetValue (null), field.Name);\r
77                         }\r
78                         return hash;\r
79                 }\r
80                 \r
81                 //\r
82                 // Returns a verbose representation of the current location\r
83                 //\r
84                 public string location {\r
85                         get {\r
86                                 string det;\r
87 \r
88                                 if (current_token == Token.ERROR)\r
89                                         det = "detail: " + error_details;\r
90                                 else\r
91                                         det = "";\r
92                                 \r
93                                 // return "Line:     "+line+" Col: "+col + "\n" +\r
94                                 //       "VirtLine: "+ref_line +\r
95                                 //       " Token: "+current_token + " " + det;\r
96                                 string current_token_name = TokenValueName [current_token] as string;\r
97                                 if (current_token_name == null)\r
98                                         current_token_name = current_token.ToString ();\r
99 \r
100                                 return String.Format ("{0} ({1},{2}), Token: {3} {4}", ref_name.Name,\r
101                                                                                        ref_line,\r
102                                                                                        col,\r
103                                                                                        current_token_name,\r
104                                                                                        det);\r
105                         }\r
106                 }\r
107 \r
108                 public bool PropertyParsing {\r
109                         get {\r
110                                 return handle_get_set;\r
111                         }\r
112 \r
113                         set {\r
114                                 handle_get_set = value;\r
115                         }\r
116                 }\r
117 \r
118                 public bool AssemblyTargetParsing {\r
119                         get {\r
120                                 return handle_assembly;\r
121                         }\r
122 \r
123                         set {\r
124                                 handle_assembly = value;\r
125                         }\r
126                 }\r
127 \r
128                 public bool EventParsing {\r
129                         get {\r
130                                 return handle_remove_add;\r
131                         }\r
132 \r
133                         set {\r
134                                 handle_remove_add = value;\r
135                         }\r
136                 }\r
137 \r
138                 public bool ConstraintsParsing {\r
139                         get {\r
140                                 return handle_constraints;\r
141                         }\r
142 \r
143                         set {\r
144                                 handle_constraints = value;\r
145                         }\r
146                 }\r
147                 \r
148                 //\r
149                 // Class variables\r
150                 // \r
151                 static CharArrayHashtable[] keywords;\r
152                 static NumberStyles styles;\r
153                 static NumberFormatInfo csharp_format_info;\r
154                 \r
155                 //\r
156                 // Values for the associated token returned\r
157                 //\r
158                 int putback_char;\r
159                 Object val;\r
160 \r
161                 //\r
162                 // Pre-processor\r
163                 //\r
164                 Hashtable defines;\r
165 \r
166                 const int TAKING        = 1;\r
167                 const int TAKEN_BEFORE  = 2;\r
168                 const int ELSE_SEEN     = 4;\r
169                 const int PARENT_TAKING = 8;\r
170                 const int REGION        = 16;           \r
171 \r
172                 //\r
173                 // pre-processor if stack state:\r
174                 //\r
175                 Stack ifstack;\r
176 \r
177                 static System.Text.StringBuilder string_builder;\r
178 \r
179                 const int max_id_size = 512;\r
180                 static char [] id_builder = new char [max_id_size];\r
181 \r
182                 static CharArrayHashtable [] identifiers = new CharArrayHashtable [max_id_size + 1];\r
183 \r
184                 const int max_number_size = 128;\r
185                 static char [] number_builder = new char [max_number_size];\r
186                 static int number_pos;\r
187                 \r
188                 //\r
189                 // Details about the error encoutered by the tokenizer\r
190                 //\r
191                 string error_details;\r
192                 \r
193                 public string error {\r
194                         get {\r
195                                 return error_details;\r
196                         }\r
197                 }\r
198                 \r
199                 public int Line {\r
200                         get {\r
201                                 return ref_line;\r
202                         }\r
203                 }\r
204 \r
205                 public int Col {\r
206                         get {\r
207                                 return col;\r
208                         }\r
209                 }\r
210 \r
211                 static void AddKeyword (string kw, int token) {\r
212                         if (keywords [kw.Length] == null) {\r
213                                 keywords [kw.Length] = new CharArrayHashtable (kw.Length);\r
214                         }\r
215                         keywords [kw.Length] [kw.ToCharArray ()] = token;\r
216                 }\r
217 \r
218                 static void InitTokens ()\r
219                 {\r
220                         keywords = new CharArrayHashtable [64];\r
221 \r
222                         AddKeyword ("__arglist", Token.ARGLIST);\r
223                         AddKeyword ("abstract", Token.ABSTRACT);\r
224                         AddKeyword ("as", Token.AS);\r
225                         AddKeyword ("add", Token.ADD);\r
226                         AddKeyword ("assembly", Token.ASSEMBLY);\r
227                         AddKeyword ("base", Token.BASE);\r
228                         AddKeyword ("bool", Token.BOOL);\r
229                         AddKeyword ("break", Token.BREAK);\r
230                         AddKeyword ("byte", Token.BYTE);\r
231                         AddKeyword ("case", Token.CASE);\r
232                         AddKeyword ("catch", Token.CATCH);\r
233                         AddKeyword ("char", Token.CHAR);\r
234                         AddKeyword ("checked", Token.CHECKED);\r
235                         AddKeyword ("class", Token.CLASS);\r
236                         AddKeyword ("const", Token.CONST);\r
237                         AddKeyword ("continue", Token.CONTINUE);\r
238                         AddKeyword ("decimal", Token.DECIMAL);\r
239                         AddKeyword ("default", Token.DEFAULT);\r
240                         AddKeyword ("delegate", Token.DELEGATE);\r
241                         AddKeyword ("do", Token.DO);\r
242                         AddKeyword ("double", Token.DOUBLE);\r
243                         AddKeyword ("else", Token.ELSE);\r
244                         AddKeyword ("enum", Token.ENUM);\r
245                         AddKeyword ("event", Token.EVENT);\r
246                         AddKeyword ("explicit", Token.EXPLICIT);\r
247                         AddKeyword ("extern", Token.EXTERN);\r
248                         AddKeyword ("false", Token.FALSE);\r
249                         AddKeyword ("finally", Token.FINALLY);\r
250                         AddKeyword ("fixed", Token.FIXED);\r
251                         AddKeyword ("float", Token.FLOAT);\r
252                         AddKeyword ("for", Token.FOR);\r
253                         AddKeyword ("foreach", Token.FOREACH);\r
254                         AddKeyword ("goto", Token.GOTO);\r
255                         AddKeyword ("get", Token.GET);\r
256                         AddKeyword ("if", Token.IF);\r
257                         AddKeyword ("implicit", Token.IMPLICIT);\r
258                         AddKeyword ("in", Token.IN);\r
259                         AddKeyword ("int", Token.INT);\r
260                         AddKeyword ("interface", Token.INTERFACE);\r
261                         AddKeyword ("internal", Token.INTERNAL);\r
262                         AddKeyword ("is", Token.IS);\r
263                         AddKeyword ("lock", Token.LOCK);\r
264                         AddKeyword ("long", Token.LONG);\r
265                         AddKeyword ("namespace", Token.NAMESPACE);\r
266                         AddKeyword ("new", Token.NEW);\r
267                         AddKeyword ("null", Token.NULL);\r
268                         AddKeyword ("object", Token.OBJECT);\r
269                         AddKeyword ("operator", Token.OPERATOR);\r
270                         AddKeyword ("out", Token.OUT);\r
271                         AddKeyword ("override", Token.OVERRIDE);\r
272                         AddKeyword ("params", Token.PARAMS);\r
273                         AddKeyword ("private", Token.PRIVATE);\r
274                         AddKeyword ("protected", Token.PROTECTED);\r
275                         AddKeyword ("public", Token.PUBLIC);\r
276                         AddKeyword ("readonly", Token.READONLY);\r
277                         AddKeyword ("ref", Token.REF);\r
278                         AddKeyword ("remove", Token.REMOVE);\r
279                         AddKeyword ("return", Token.RETURN);\r
280                         AddKeyword ("sbyte", Token.SBYTE);\r
281                         AddKeyword ("sealed", Token.SEALED);\r
282                         AddKeyword ("set", Token.SET);\r
283                         AddKeyword ("short", Token.SHORT);\r
284                         AddKeyword ("sizeof", Token.SIZEOF);\r
285                         AddKeyword ("stackalloc", Token.STACKALLOC);\r
286                         AddKeyword ("static", Token.STATIC);\r
287                         AddKeyword ("string", Token.STRING);\r
288                         AddKeyword ("struct", Token.STRUCT);\r
289                         AddKeyword ("switch", Token.SWITCH);\r
290                         AddKeyword ("this", Token.THIS);\r
291                         AddKeyword ("throw", Token.THROW);\r
292                         AddKeyword ("true", Token.TRUE);\r
293                         AddKeyword ("try", Token.TRY);\r
294                         AddKeyword ("typeof", Token.TYPEOF);\r
295                         AddKeyword ("uint", Token.UINT);\r
296                         AddKeyword ("ulong", Token.ULONG);\r
297                         AddKeyword ("unchecked", Token.UNCHECKED);\r
298                         AddKeyword ("unsafe", Token.UNSAFE);\r
299                         AddKeyword ("ushort", Token.USHORT);\r
300                         AddKeyword ("using", Token.USING);\r
301                         AddKeyword ("virtual", Token.VIRTUAL);\r
302                         AddKeyword ("void", Token.VOID);\r
303                         AddKeyword ("volatile", Token.VOLATILE);\r
304                         AddKeyword ("where", Token.WHERE);\r
305                         AddKeyword ("while", Token.WHILE);\r
306                 }\r
307 \r
308                 //\r
309                 // Class initializer\r
310                 // \r
311                 static Tokenizer ()\r
312                 {\r
313                         InitTokens ();\r
314                         csharp_format_info = NumberFormatInfo.InvariantInfo;\r
315                         styles = NumberStyles.Float;\r
316                         \r
317                         string_builder = new System.Text.StringBuilder ();\r
318                 }\r
319 \r
320                 int GetKeyword (char[] id, int id_len)\r
321                 {\r
322                         /*\r
323                          * Keywords are stored in an array of hashtables grouped by their\r
324                          * length.\r
325                          */\r
326 \r
327                         if ((id_len >= keywords.Length) || (keywords [id_len] == null))\r
328                                 return -1;\r
329                         object o = keywords [id_len] [id];\r
330 \r
331                         if (o == null)\r
332                                 return -1;\r
333                         \r
334                         int res = (int) o;\r
335 \r
336                         if (handle_get_set == false && (res == Token.GET || res == Token.SET))\r
337                                 return -1;\r
338                         if (handle_remove_add == false && (res == Token.REMOVE || res == Token.ADD))\r
339                                 return -1;\r
340                         if (handle_assembly == false && res == Token.ASSEMBLY)\r
341                                 return -1;\r
342                         if (handle_constraints == false && res == Token.WHERE)\r
343                                 return -1;\r
344                         return res;\r
345                         \r
346                 }\r
347 \r
348                 public Location Location {\r
349                         get {\r
350                                 return new Location (ref_line);\r
351                         }\r
352                 }\r
353 \r
354                 void define (string def)\r
355                 {\r
356                         if (!RootContext.AllDefines.Contains (def)){\r
357                                 RootContext.AllDefines [def] = true;\r
358                         }\r
359                         if (defines.Contains (def))\r
360                                 return;\r
361                         defines [def] = true;\r
362                 }\r
363                 \r
364                 public Tokenizer (SeekableStreamReader input, SourceFile file, ArrayList defs)\r
365                 {\r
366                         this.ref_name = file;\r
367                         this.file_name = file;\r
368                         reader = input;\r
369                         \r
370                         putback_char = -1;\r
371 \r
372                         if (defs != null){\r
373                                 defines = new Hashtable ();\r
374                                 foreach (string def in defs)\r
375                                         define (def);\r
376                         }\r
377 \r
378                         //\r
379                         // FIXME: This could be `Location.Push' but we have to\r
380                         // find out why the MS compiler allows this\r
381                         //\r
382                         Mono.CSharp.Location.Push (file);\r
383                 }\r
384 \r
385                 public static void Cleanup () {\r
386                         identifiers = null;\r
387                 }\r
388 \r
389                 static bool is_identifier_start_character (char c)\r
390                 {\r
391                         return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') || c == '_' || Char.IsLetter (c);\r
392                 }\r
393 \r
394                 static bool is_identifier_part_character (char c)\r
395                 {\r
396                         return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_' || (c >= '0' && c <= '9') || Char.IsLetter (c);\r
397                 }
398                 
399                 public static bool IsValidIdentifier (string s)
400                 {
401                         if (s == null || s.Length == 0)
402                                 return false;
403                         
404                         if (!is_identifier_start_character (s [0]))
405                                 return false;
406                         
407                         for (int i = 1; i < s.Length; i ++)
408                                 if (! is_identifier_part_character (s [i]))
409                                         return false;
410                         
411                         return true;
412                 }\r
413 \r
414                 bool parse_less_than ()\r
415                 {\r
416                 start:\r
417                         int the_token = token ();\r
418                         switch (the_token) {\r
419                         case Token.IDENTIFIER:\r
420                         case Token.OBJECT:\r
421                         case Token.STRING:\r
422                         case Token.BOOL:\r
423                         case Token.DECIMAL:\r
424                         case Token.FLOAT:\r
425                         case Token.DOUBLE:\r
426                         case Token.SBYTE:\r
427                         case Token.BYTE:\r
428                         case Token.SHORT:\r
429                         case Token.USHORT:\r
430                         case Token.INT:\r
431                         case Token.UINT:\r
432                         case Token.LONG:\r
433                         case Token.ULONG:\r
434                         case Token.CHAR:\r
435                         case Token.VOID:\r
436                                 break;\r
437 \r
438                         default:\r
439                                 return false;\r
440                         }\r
441                 again:\r
442                         the_token = token ();\r
443 \r
444                         if (the_token == Token.OP_GENERICS_GT)\r
445                                 return true;\r
446                         else if ((the_token == Token.COMMA) || (the_token == Token.DOT))\r
447                                 goto start;\r
448                         else if (the_token == Token.OP_GENERICS_LT) {\r
449                                 if (!parse_less_than ())\r
450                                         return false;\r
451                                 goto again;\r
452                         }\r
453 \r
454                         return false;\r
455                 }\r
456 \r
457                 bool parsing_less_than = false;\r
458                 int parsing_generic_less_than = 0;\r
459 \r
460                 int is_punct (char c, ref bool doread)\r
461                 {\r
462                         int d;\r
463                         int t;\r
464 \r
465                         doread = false;\r
466 \r
467                         switch (c){\r
468                         case '{':\r
469                                 return Token.OPEN_BRACE;\r
470                         case '}':\r
471                                 return Token.CLOSE_BRACE;\r
472                         case '[':\r
473                                 return Token.OPEN_BRACKET;\r
474                         case ']':\r
475                                 return Token.CLOSE_BRACKET;\r
476                         case '(':\r
477                                 return Token.OPEN_PARENS;\r
478                         case ')': {\r
479                                 if (deambiguate_close_parens == 0)\r
480                                         return Token.CLOSE_PARENS;\r
481 \r
482                                 --deambiguate_close_parens;\r
483 \r
484                                 // Save current position and parse next token.\r
485                                 int old = reader.Position;\r
486                                 int new_token = token ();\r
487                                 reader.Position = old;\r
488                                 putback_char = -1;\r
489 \r
490                                 if (new_token == Token.OPEN_PARENS)\r
491                                         return Token.CLOSE_PARENS_OPEN_PARENS;\r
492                                 else if (new_token == Token.MINUS)\r
493                                         return Token.CLOSE_PARENS_MINUS;\r
494                                 else if (IsCastToken (new_token))\r
495                                         return Token.CLOSE_PARENS_CAST;\r
496                                 else\r
497                                         return Token.CLOSE_PARENS_NO_CAST;\r
498                         }\r
499 \r
500                         case ',':\r
501                                 return Token.COMMA;\r
502                         case ':':\r
503                                 return Token.COLON;\r
504                         case ';':\r
505                                 return Token.SEMICOLON;\r
506                         case '~':\r
507                                 return Token.TILDE;\r
508                         case '?':\r
509                                 return Token.INTERR;\r
510                         }\r
511 \r
512                         if (c == '<') {\r
513                                 if (parsing_generic_less_than++ > 0)\r
514                                         return Token.OP_GENERICS_LT;\r
515 \r
516                                 // Save current position and parse next token.\r
517                                 int old = reader.Position;\r
518                                 bool is_generic_lt = parse_less_than ();\r
519                                 reader.Position = old;\r
520                                 putback_char = -1;\r
521 \r
522                                 if (is_generic_lt) {\r
523                                         parsing_generic_less_than++;\r
524                                         return Token.OP_GENERICS_LT;\r
525                                 } else\r
526                                         parsing_generic_less_than = 0;\r
527 \r
528                                 d = peekChar ();\r
529                                 if (d == '<'){\r
530                                         getChar ();\r
531                                         d = peekChar ();\r
532 \r
533                                         if (d == '='){\r
534                                                 doread = true;\r
535                                                 return Token.OP_SHIFT_LEFT_ASSIGN;\r
536                                         }\r
537                                         return Token.OP_SHIFT_LEFT;\r
538                                 } else if (d == '='){\r
539                                         doread = true;\r
540                                         return Token.OP_LE;\r
541                                 }\r
542                                 return Token.OP_LT;\r
543                         } else if (c == '>') {\r
544                                 if (parsing_generic_less_than > 0) {\r
545                                         parsing_generic_less_than--;\r
546                                         return Token.OP_GENERICS_GT;\r
547                                 }\r
548 \r
549                                 d = peekChar ();\r
550                                 if (d == '>'){\r
551                                         getChar ();\r
552                                         d = peekChar ();\r
553 \r
554                                         if (d == '='){\r
555                                                 doread = true;\r
556                                                 return Token.OP_SHIFT_RIGHT_ASSIGN;\r
557                                         }\r
558                                         return Token.OP_SHIFT_RIGHT;\r
559                                 } else if (d == '='){\r
560                                         doread = true;\r
561                                         return Token.OP_GE;\r
562                                 }\r
563                                 return Token.OP_GT;\r
564                         }\r
565 \r
566                         d = peekChar ();\r
567                         if (c == '+'){\r
568                                 \r
569                                 if (d == '+')\r
570                                         t = Token.OP_INC;\r
571                                 else if (d == '=')\r
572                                         t = Token.OP_ADD_ASSIGN;\r
573                                 else\r
574                                         return Token.PLUS;\r
575                                 doread = true;\r
576                                 return t;\r
577                         }\r
578                         if (c == '-'){\r
579                                 if (d == '-')\r
580                                         t = Token.OP_DEC;\r
581                                 else if (d == '=')\r
582                                         t = Token.OP_SUB_ASSIGN;\r
583                                 else if (d == '>')\r
584                                         t = Token.OP_PTR;\r
585                                 else\r
586                                         return Token.MINUS;\r
587                                 doread = true;\r
588                                 return t;\r
589                         }\r
590 \r
591                         if (c == '!'){\r
592                                 if (d == '='){\r
593                                         doread = true;\r
594                                         return Token.OP_NE;\r
595                                 }\r
596                                 return Token.BANG;\r
597                         }\r
598 \r
599                         if (c == '='){\r
600                                 if (d == '='){\r
601                                         doread = true;\r
602                                         return Token.OP_EQ;\r
603                                 }\r
604                                 return Token.ASSIGN;\r
605                         }\r
606 \r
607                         if (c == '&'){\r
608                                 if (d == '&'){\r
609                                         doread = true;\r
610                                         return Token.OP_AND;\r
611                                 } else if (d == '='){\r
612                                         doread = true;\r
613                                         return Token.OP_AND_ASSIGN;\r
614                                 }\r
615                                 return Token.BITWISE_AND;\r
616                         }\r
617 \r
618                         if (c == '|'){\r
619                                 if (d == '|'){\r
620                                         doread = true;\r
621                                         return Token.OP_OR;\r
622                                 } else if (d == '='){\r
623                                         doread = true;\r
624                                         return Token.OP_OR_ASSIGN;\r
625                                 }\r
626                                 return Token.BITWISE_OR;\r
627                         }\r
628 \r
629                         if (c == '*'){\r
630                                 if (d == '='){\r
631                                         doread = true;\r
632                                         return Token.OP_MULT_ASSIGN;\r
633                                 }\r
634                                 return Token.STAR;\r
635                         }\r
636 \r
637                         if (c == '/'){\r
638                                 if (d == '='){\r
639                                         doread = true;\r
640                                         return Token.OP_DIV_ASSIGN;\r
641                                 }\r
642                                 return Token.DIV;\r
643                         }\r
644 \r
645                         if (c == '%'){\r
646                                 if (d == '='){\r
647                                         doread = true;\r
648                                         return Token.OP_MOD_ASSIGN;\r
649                                 }\r
650                                 return Token.PERCENT;\r
651                         }\r
652 \r
653                         if (c == '^'){\r
654                                 if (d == '='){\r
655                                         doread = true;\r
656                                         return Token.OP_XOR_ASSIGN;\r
657                                 }\r
658                                 return Token.CARRET;\r
659                         }\r
660 \r
661                         return Token.ERROR;\r
662                 }\r
663 \r
664                 int deambiguate_close_parens = 0;\r
665 \r
666                 public void Deambiguate_CloseParens ()\r
667                 {\r
668                         putback (')');\r
669                         deambiguate_close_parens++;\r
670                 }\r
671 \r
672                 void Error_NumericConstantTooLong ()\r
673                 {\r
674                         Report.Error (1021, Location, "Numeric constant too long");                     \r
675                 }\r
676                 \r
677                 bool decimal_digits (int c)\r
678                 {\r
679                         int d;\r
680                         bool seen_digits = false;\r
681                         \r
682                         if (c != -1){\r
683                                 if (number_pos == max_number_size)\r
684                                         Error_NumericConstantTooLong ();\r
685                                 number_builder [number_pos++] = (char) c;\r
686                         }\r
687                         \r
688                         //\r
689                         // We use peekChar2, because decimal_digits needs to do a \r
690                         // 2-character look-ahead (5.ToString for example).\r
691                         //\r
692                         while ((d = peekChar2 ()) != -1){\r
693                                 if (d >= '0' && d <= '9'){\r
694                                         if (number_pos == max_number_size)\r
695                                                 Error_NumericConstantTooLong ();\r
696                                         number_builder [number_pos++] = (char) d;\r
697                                         getChar ();\r
698                                         seen_digits = true;\r
699                                 } else\r
700                                         break;\r
701                         }\r
702                         \r
703                         return seen_digits;\r
704                 }\r
705 \r
706                 bool is_hex (int e)\r
707                 {\r
708                         return (e >= '0' && e <= '9') || (e >= 'A' && e <= 'F') || (e >= 'a' && e <= 'f');\r
709                 }\r
710                 \r
711                 void hex_digits (int c)\r
712                 {\r
713                         if (c != -1)\r
714                                 number_builder [number_pos++] = (char) c;\r
715                         \r
716                 }\r
717                 \r
718                 int real_type_suffix (int c)\r
719                 {\r
720                         int t;\r
721 \r
722                         switch (c){\r
723                         case 'F': case 'f':\r
724                                 t =  Token.LITERAL_FLOAT;\r
725                                 break;\r
726                         case 'D': case 'd':\r
727                                 t = Token.LITERAL_DOUBLE;\r
728                                 break;\r
729                         case 'M': case 'm':\r
730                                  t= Token.LITERAL_DECIMAL;\r
731                                 break;\r
732                         default:\r
733                                 return Token.NONE;\r
734                         }\r
735                         return t;\r
736                 }\r
737 \r
738                 int integer_type_suffix (ulong ul, int c)\r
739                 {\r
740                         bool is_unsigned = false;\r
741                         bool is_long = false;\r
742 \r
743                         if (c != -1){\r
744                                 bool scanning = true;\r
745                                 do {\r
746                                         switch (c){\r
747                                         case 'U': case 'u':\r
748                                                 if (is_unsigned)\r
749                                                         scanning = false;\r
750                                                 is_unsigned = true;\r
751                                                 getChar ();\r
752                                                 break;\r
753 \r
754                                         case 'l':\r
755                                                 if (!is_unsigned){\r
756                                                         //\r
757                                                         // if we have not seen anything in between\r
758                                                         // report this error\r
759                                                         //\r
760                                                         Report.Warning (\r
761                                                                 78, Location,\r
762                                                         "the 'l' suffix is easily confused with digit `1'," +\r
763                                                         " use 'L' for clarity");\r
764                                                 }\r
765                                                 goto case 'L';\r
766                                                 \r
767                                         case 'L': \r
768                                                 if (is_long)\r
769                                                         scanning = false;\r
770                                                 is_long = true;\r
771                                                 getChar ();\r
772                                                 break;\r
773                                                 \r
774                                         default:\r
775                                                 scanning = false;\r
776                                                 break;\r
777                                         }\r
778                                         c = peekChar ();\r
779                                 } while (scanning);\r
780                         }\r
781 \r
782                         if (is_long && is_unsigned){\r
783                                 val = ul;\r
784                                 return Token.LITERAL_INTEGER;\r
785                         } else if (is_unsigned){\r
786                                 // uint if possible, or ulong else.\r
787 \r
788                                 if ((ul & 0xffffffff00000000) == 0)\r
789                                         val = (uint) ul;\r
790                                 else\r
791                                         val = ul;\r
792                         } else if (is_long){\r
793                                 // long if possible, ulong otherwise\r
794                                 if ((ul & 0x8000000000000000) != 0)\r
795                                         val = ul;\r
796                                 else\r
797                                         val = (long) ul;\r
798                         } else {\r
799                                 // int, uint, long or ulong in that order\r
800                                 if ((ul & 0xffffffff00000000) == 0){\r
801                                         uint ui = (uint) ul;\r
802                                         \r
803                                         if ((ui & 0x80000000) != 0)\r
804                                                 val = ui;\r
805                                         else\r
806                                                 val = (int) ui;\r
807                                 } else {\r
808                                         if ((ul & 0x8000000000000000) != 0)\r
809                                                 val = ul;\r
810                                         else\r
811                                                 val = (long) ul;\r
812                                 }\r
813                         }\r
814                         return Token.LITERAL_INTEGER;\r
815                 }\r
816                                 \r
817                 //\r
818                 // given `c' as the next char in the input decide whether\r
819                 // we need to convert to a special type, and then choose\r
820                 // the best representation for the integer\r
821                 //\r
822                 int adjust_int (int c)\r
823                 {\r
824                         try {\r
825                                 if (number_pos > 9){\r
826                                         ulong ul = (uint) (number_builder [0] - '0');\r
827 \r
828                                         for (int i = 1; i < number_pos; i++){\r
829                                                 ul = checked ((ul * 10) + ((uint)(number_builder [i] - '0')));\r
830                                         }\r
831                                         return integer_type_suffix (ul, c);\r
832                                 } else {\r
833                                         uint ui = (uint) (number_builder [0] - '0');\r
834 \r
835                                         for (int i = 1; i < number_pos; i++){\r
836                                                 ui = checked ((ui * 10) + ((uint)(number_builder [i] - '0')));\r
837                                         }\r
838                                         return integer_type_suffix (ui, c);\r
839                                 }\r
840                         } catch (OverflowException) {\r
841                                 error_details = "Integral constant is too large";\r
842                                 Report.Error (1021, Location, error_details);\r
843                                 val = 0ul;\r
844                                 return Token.LITERAL_INTEGER;\r
845                         }\r
846                 }\r
847                 \r
848                 int adjust_real (int t)\r
849                 {\r
850                         string s = new String (number_builder, 0, number_pos);\r
851 \r
852                         switch (t){\r
853                         case Token.LITERAL_DECIMAL:\r
854                                 try {\r
855                                         val = System.Decimal.Parse (s, styles, csharp_format_info);\r
856                                 } catch (OverflowException) {\r
857                                         val = 0m;     \r
858                                         error_details = "Floating-point constant is outside the range of the type 'decimal'";\r
859                                         Report.Error (594, Location, error_details);\r
860                                 }\r
861                                 break;\r
862                         case Token.LITERAL_FLOAT:\r
863                                 try {\r
864                                         val = (float) System.Double.Parse (s, styles, csharp_format_info);\r
865                                 } catch (OverflowException) {\r
866                                         val = 0.0f;     \r
867                                         error_details = "Floating-point constant is outside the range of the type 'float'";\r
868                                         Report.Error (594, Location, error_details);\r
869                                 }\r
870                                 break;\r
871                                 \r
872                         case Token.LITERAL_DOUBLE:\r
873                         case Token.NONE:\r
874                                 t = Token.LITERAL_DOUBLE;\r
875                                 try {\r
876                                         val = System.Double.Parse (s, styles, csharp_format_info);\r
877                                 } catch (OverflowException) {\r
878                                         val = 0.0;     \r
879                                         error_details = "Floating-point constant is outside the range of the type 'double'";\r
880                                         Report.Error (594, Location, error_details);\r
881                                 }\r
882                                 break;\r
883                         }\r
884                         return t;\r
885                 }\r
886 \r
887                 int handle_hex ()\r
888                 {\r
889                         int d;\r
890                         ulong ul;\r
891                         \r
892                         getChar ();\r
893                         while ((d = peekChar ()) != -1){\r
894                                 if (is_hex (d)){\r
895                                         number_builder [number_pos++] = (char) d;\r
896                                         getChar ();\r
897                                 } else\r
898                                         break;\r
899                         }\r
900                         \r
901                         string s = new String (number_builder, 0, number_pos);\r
902                         try {\r
903                                 if (number_pos <= 8)\r
904                                         ul = System.UInt32.Parse (s, NumberStyles.HexNumber);\r
905                                 else\r
906                                         ul = System.UInt64.Parse (s, NumberStyles.HexNumber);\r
907                         } catch (OverflowException){\r
908                                 error_details = "Integral constant is too large";\r
909                                 Report.Error (1021, Location, error_details);\r
910                                 val = 0ul;\r
911                                 return Token.LITERAL_INTEGER;\r
912                         }\r
913                         \r
914                         return integer_type_suffix (ul, peekChar ());\r
915                 }\r
916 \r
917                 //\r
918                 // Invoked if we know we have .digits or digits\r
919                 //\r
920                 int is_number (int c)\r
921                 {\r
922                         bool is_real = false;\r
923                         int type;\r
924 \r
925                         number_pos = 0;\r
926 \r
927                         if (c >= '0' && c <= '9'){\r
928                                 if (c == '0'){\r
929                                         int peek = peekChar ();\r
930 \r
931                                         if (peek == 'x' || peek == 'X')\r
932                                                 return handle_hex ();\r
933                                 }\r
934                                 decimal_digits (c);\r
935                                 c = getChar ();\r
936                         }\r
937 \r
938                         //\r
939                         // We need to handle the case of\r
940                         // "1.1" vs "1.string" (LITERAL_FLOAT vs NUMBER DOT IDENTIFIER)\r
941                         //\r
942                         if (c == '.'){\r
943                                 if (decimal_digits ('.')){\r
944                                         is_real = true;\r
945                                         c = getChar ();\r
946                                 } else {\r
947                                         putback ('.');\r
948                                         number_pos--;\r
949                                         return adjust_int (-1);\r
950                                 }\r
951                         }\r
952                         \r
953                         if (c == 'e' || c == 'E'){\r
954                                 is_real = true;\r
955                                 if (number_pos == max_number_size)\r
956                                         Error_NumericConstantTooLong ();\r
957                                 number_builder [number_pos++] = 'e';\r
958                                 c = getChar ();\r
959                                 \r
960                                 if (c == '+'){\r
961                                         if (number_pos == max_number_size)\r
962                                                 Error_NumericConstantTooLong ();\r
963                                         number_builder [number_pos++] = '+';\r
964                                         c = -1;\r
965                                 } else if (c == '-') {\r
966                                         if (number_pos == max_number_size)\r
967                                                 Error_NumericConstantTooLong ();\r
968                                         number_builder [number_pos++] = '-';\r
969                                         c = -1;\r
970                                 } else {\r
971                                         if (number_pos == max_number_size)\r
972                                                 Error_NumericConstantTooLong ();\r
973                                         number_builder [number_pos++] = '+';\r
974                                 }\r
975                                         \r
976                                 decimal_digits (c);\r
977                                 c = getChar ();\r
978                         }\r
979 \r
980                         type = real_type_suffix (c);\r
981                         if (type == Token.NONE && !is_real){\r
982                                 putback (c);\r
983                                 return adjust_int (c);\r
984                         } else \r
985                                 is_real = true;\r
986 \r
987                         if (type == Token.NONE){\r
988                                 putback (c);\r
989                         }\r
990                         \r
991                         if (is_real)\r
992                                 return adjust_real (type);\r
993 \r
994                         Console.WriteLine ("This should not be reached");\r
995                         throw new Exception ("Is Number should never reach this point");\r
996                 }\r
997 \r
998                 //\r
999                 // Accepts exactly count (4 or 8) hex, no more no less\r
1000                 //\r
1001                 int getHex (int count, out bool error)\r
1002                 {\r
1003                         int i;\r
1004                         int total = 0;\r
1005                         int c;\r
1006                         int top = count != -1 ? count : 4;\r
1007                         \r
1008                         getChar ();\r
1009                         error = false;\r
1010                         for (i = 0; i < top; i++){\r
1011                                 c = getChar ();\r
1012                                 \r
1013                                 if (c >= '0' && c <= '9')\r
1014                                         c = (int) c - (int) '0';\r
1015                                 else if (c >= 'A' && c <= 'F')\r
1016                                         c = (int) c - (int) 'A' + 10;\r
1017                                 else if (c >= 'a' && c <= 'f')\r
1018                                         c = (int) c - (int) 'a' + 10;\r
1019                                 else {\r
1020                                         error = true;\r
1021                                         return 0;\r
1022                                 }\r
1023                                 \r
1024                                 total = (total * 16) + c;\r
1025                                 if (count == -1){\r
1026                                         int p = peekChar ();\r
1027                                         if (p == -1)\r
1028                                                 break;\r
1029                                         if (!is_hex ((char)p))\r
1030                                                 break;\r
1031                                 }\r
1032                         }\r
1033                         return total;\r
1034                 }\r
1035 \r
1036                 int escape (int c)\r
1037                 {\r
1038                         bool error;\r
1039                         int d;\r
1040                         int v;\r
1041 \r
1042                         d = peekChar ();\r
1043                         if (c != '\\')\r
1044                                 return c;\r
1045                         \r
1046                         switch (d){\r
1047                         case 'a':\r
1048                                 v = '\a'; break;\r
1049                         case 'b':\r
1050                                 v = '\b'; break;\r
1051                         case 'n':\r
1052                                 v = '\n'; break;\r
1053                         case 't':\r
1054                                 v = '\t'; break;\r
1055                         case 'v':\r
1056                                 v = '\v'; break;\r
1057                         case 'r':\r
1058                                 v = '\r'; break;\r
1059                         case '\\':\r
1060                                 v = '\\'; break;\r
1061                         case 'f':\r
1062                                 v = '\f'; break;\r
1063                         case '0':\r
1064                                 v = 0; break;\r
1065                         case '"':\r
1066                                 v = '"'; break;\r
1067                         case '\'':\r
1068                                 v = '\''; break;\r
1069                         case 'x':\r
1070                                 v = getHex (-1, out error);\r
1071                                 if (error)\r
1072                                         goto default;\r
1073                                 return v;\r
1074                         case 'u':\r
1075                                 v = getHex (4, out error);\r
1076                                 if (error)\r
1077                                         goto default;\r
1078                                 return v;\r
1079                         case 'U':\r
1080                                 v = getHex (8, out error);\r
1081                                 if (error)\r
1082                                         goto default;\r
1083                                 return v;\r
1084                         default:\r
1085                                 Report.Error (1009, Location, "Unrecognized escape sequence in " + (char)d);\r
1086                                 return d;\r
1087                         }\r
1088                         getChar ();\r
1089                         return v;\r
1090                 }\r
1091 \r
1092                 int getChar ()\r
1093                 {\r
1094                         if (putback_char != -1){\r
1095                                 int x = putback_char;\r
1096                                 putback_char = -1;\r
1097 \r
1098                                 return x;\r
1099                         }\r
1100                         return reader.Read ();\r
1101                 }\r
1102 \r
1103                 int peekChar ()\r
1104                 {\r
1105                         if (putback_char != -1)\r
1106                                 return putback_char;\r
1107                         putback_char = reader.Read ();\r
1108                         return putback_char;\r
1109                 }\r
1110 \r
1111                 int peekChar2 ()\r
1112                 {\r
1113                         if (putback_char != -1)\r
1114                                 return putback_char;\r
1115                         return reader.Peek ();\r
1116                 }\r
1117                 \r
1118                 void putback (int c)\r
1119                 {\r
1120                         if (putback_char != -1){\r
1121                                 Console.WriteLine ("Col: " + col);\r
1122                                 Console.WriteLine ("Row: " + line);\r
1123                                 Console.WriteLine ("Name: " + ref_name.Name);\r
1124                                 Console.WriteLine ("Current [{0}] putting back [{1}]  ", putback_char, c);\r
1125                                 throw new Exception ("This should not happen putback on putback");\r
1126                         }\r
1127                         putback_char = c;\r
1128                 }\r
1129 \r
1130                 public bool advance ()\r
1131                 {\r
1132                         return peekChar () != -1;\r
1133                 }\r
1134 \r
1135                 public Object Value {\r
1136                         get {\r
1137                                 return val;\r
1138                         }\r
1139                 }\r
1140 \r
1141                 public Object value ()\r
1142                 {\r
1143                         return val;\r
1144                 }\r
1145 \r
1146                 bool IsCastToken (int token)\r
1147                 {\r
1148                         switch (token) {\r
1149                         case Token.BANG:\r
1150                         case Token.TILDE:\r
1151                         case Token.IDENTIFIER:\r
1152                         case Token.LITERAL_INTEGER:\r
1153                         case Token.LITERAL_FLOAT:\r
1154                         case Token.LITERAL_DOUBLE:\r
1155                         case Token.LITERAL_DECIMAL:\r
1156                         case Token.LITERAL_CHARACTER:\r
1157                         case Token.LITERAL_STRING:\r
1158                         case Token.BASE:\r
1159                         case Token.CHECKED:\r
1160                         case Token.FALSE:\r
1161                         case Token.FIXED:\r
1162                         case Token.NEW:\r
1163                         case Token.NULL:\r
1164                         case Token.SIZEOF:\r
1165                         case Token.THIS:\r
1166                         case Token.THROW:\r
1167                         case Token.TRUE:\r
1168                         case Token.TYPEOF:\r
1169                         case Token.UNCHECKED:\r
1170                         case Token.UNSAFE:\r
1171 \r
1172                                 //\r
1173                                 // These can be part of a member access\r
1174                                 //\r
1175                         case Token.INT:\r
1176                         case Token.UINT:\r
1177                         case Token.SHORT:\r
1178                         case Token.USHORT:\r
1179                         case Token.LONG:\r
1180                         case Token.ULONG:\r
1181                         case Token.DOUBLE:\r
1182                         case Token.FLOAT:\r
1183                         case Token.CHAR:\r
1184                                 return true;\r
1185 \r
1186                         default:\r
1187                                 return false;\r
1188                         }\r
1189                 }\r
1190 \r
1191                 public int token ()\r
1192                 {\r
1193                         current_token = xtoken ();\r
1194 \r
1195                         if (current_token != Token.DEFAULT)\r
1196                                 return current_token;\r
1197 \r
1198                         int c = consume_whitespace ();\r
1199                         if (c == -1)\r
1200                                 current_token = Token.ERROR;\r
1201                         else if (c == '(')\r
1202                                 current_token = Token.DEFAULT_OPEN_PARENS;\r
1203                         else\r
1204                                 putback (c);\r
1205 \r
1206                         return current_token;\r
1207                 }\r
1208 \r
1209                 static StringBuilder static_cmd_arg = new System.Text.StringBuilder ();\r
1210                 \r
1211                 void get_cmd_arg (out string cmd, out string arg)\r
1212                 {\r
1213                         int c;\r
1214                         \r
1215                         tokens_seen = false;\r
1216                         arg = "";\r
1217                         static_cmd_arg.Length = 0;\r
1218 \r
1219                         // skip over white space\r
1220                         while ((c = getChar ()) != -1 && (c != '\n') && ((c == '\r') || (c == ' ') || (c == '\t')))\r
1221                                 ;\r
1222                                 \r
1223                         while ((c != -1) && (c != '\n') && (c != ' ') && (c != '\t') && (c != '\r')){\r
1224                                 if (is_identifier_part_character ((char) c)){\r
1225                                         static_cmd_arg.Append ((char) c);\r
1226                                         c = getChar ();\r
1227                                 } else {\r
1228                                         putback (c);\r
1229                                         break;\r
1230                                 }\r
1231                         }\r
1232 \r
1233                         cmd = static_cmd_arg.ToString ();\r
1234 \r
1235                         if (c == '\n'){\r
1236                                 line++;\r
1237                                 ref_line++;\r
1238                                 return;\r
1239                         } else if (c == '\r')\r
1240                                 col = 0;\r
1241 \r
1242                         // skip over white space\r
1243                         while ((c = getChar ()) != -1 && (c != '\n') && ((c == '\r') || (c == ' ') || (c == '\t')))\r
1244                                 ;\r
1245 \r
1246                         if (c == '\n'){\r
1247                                 line++;\r
1248                                 ref_line++;\r
1249                                 return;\r
1250                         } else if (c == '\r'){\r
1251                                 col = 0;\r
1252                                 return;\r
1253                         }\r
1254                         \r
1255                         static_cmd_arg.Length = 0;\r
1256                         static_cmd_arg.Append ((char) c);\r
1257                         \r
1258                         while ((c = getChar ()) != -1 && (c != '\n') && (c != '\r')){\r
1259                                 static_cmd_arg.Append ((char) c);\r
1260                         }\r
1261 \r
1262                         if (c == '\n'){\r
1263                                 line++;\r
1264                                 ref_line++;\r
1265                         } else if (c == '\r')\r
1266                                 col = 0;\r
1267                         arg = static_cmd_arg.ToString ().Trim ();\r
1268                 }\r
1269 \r
1270                 //\r
1271                 // Handles the #line directive\r
1272                 //\r
1273                 bool PreProcessLine (string arg)\r
1274                 {\r
1275                         if (arg == "")\r
1276                                 return false;\r
1277 \r
1278                         if (arg == "default"){\r
1279                                 ref_line = line;\r
1280                                 ref_name = file_name;\r
1281                                 Location.Push (ref_name);\r
1282                                 return true;\r
1283                         } else if (arg == "hidden"){\r
1284                                 //\r
1285                                 // We ignore #line hidden\r
1286                                 //\r
1287                                 return true;\r
1288                         }\r
1289                         \r
1290                         try {\r
1291                                 int pos;\r
1292 \r
1293                                 if ((pos = arg.IndexOf (' ')) != -1 && pos != 0){\r
1294                                         ref_line = System.Int32.Parse (arg.Substring (0, pos));\r
1295                                         pos++;\r
1296                                         \r
1297                                         char [] quotes = { '\"' };\r
1298                                         \r
1299                                         string name = arg.Substring (pos). Trim (quotes);\r
1300                                         ref_name = Location.LookupFile (name);\r
1301                                         file_name.HasLineDirective = true;\r
1302                                         ref_name.HasLineDirective = true;\r
1303                                         Location.Push (ref_name);\r
1304                                 } else {\r
1305                                         ref_line = System.Int32.Parse (arg);\r
1306                                 }\r
1307                         } catch {\r
1308                                 return false;\r
1309                         }\r
1310                         \r
1311                         return true;\r
1312                 }\r
1313 \r
1314                 //\r
1315                 // Handles #define and #undef\r
1316                 //\r
1317                 void PreProcessDefinition (bool is_define, string arg)\r
1318                 {\r
1319                         if (arg == "" || arg == "true" || arg == "false"){\r
1320                                 Report.Error (1001, Location, "Missing identifer to pre-processor directive");\r
1321                                 return;\r
1322                         }\r
1323 \r
1324                         char[] whitespace = { ' ', '\t' };\r
1325                         if (arg.IndexOfAny (whitespace) != -1){\r
1326                                 Report.Error (1025, Location, "Single-line comment or end-of-line expected");\r
1327                                 return;\r
1328                         }\r
1329 \r
1330                         if (!is_identifier_start_character (arg [0]))\r
1331                                 Report.Error (1001, Location, "Identifier expected: " + arg);\r
1332                         \r
1333                         foreach (char c in arg.Substring (1)){\r
1334                                 if (!is_identifier_part_character (c)){\r
1335                                         Report.Error (1001, Location, "Identifier expected: " + arg);\r
1336                                         return;\r
1337                                 }\r
1338                         }\r
1339 \r
1340                         if (is_define){\r
1341                                 if (defines == null)\r
1342                                         defines = new Hashtable ();\r
1343                                 define (arg);\r
1344                         } else {\r
1345                                 if (defines == null)\r
1346                                         return;\r
1347                                 if (defines.Contains (arg))\r
1348                                         defines.Remove (arg);\r
1349                         }\r
1350                 }\r
1351 \r
1352                 bool eval_val (string s)\r
1353                 {\r
1354                         if (s == "true")\r
1355                                 return true;\r
1356                         if (s == "false")\r
1357                                 return false;\r
1358                         \r
1359                         if (defines == null)\r
1360                                 return false;\r
1361                         if (defines.Contains (s))\r
1362                                 return true;\r
1363 \r
1364                         return false;\r
1365                 }\r
1366 \r
1367                 bool pp_primary (ref string s)\r
1368                 {\r
1369                         s = s.Trim ();\r
1370                         int len = s.Length;\r
1371 \r
1372                         if (len > 0){\r
1373                                 char c = s [0];\r
1374                                 \r
1375                                 if (c == '('){\r
1376                                         s = s.Substring (1);\r
1377                                         bool val = pp_expr (ref s);\r
1378                                         if (s.Length > 0 && s [0] == ')'){\r
1379                                                 s = s.Substring (1);\r
1380                                                 return val;\r
1381                                         }\r
1382                                         Error_InvalidDirective ();\r
1383                                         return false;\r
1384                                 }\r
1385                                 \r
1386                                 if (is_identifier_start_character (c)){\r
1387                                         int j = 1;\r
1388 \r
1389                                         while (j < len){\r
1390                                                 c = s [j];\r
1391                                                 \r
1392                                                 if (is_identifier_part_character (c)){\r
1393                                                         j++;\r
1394                                                         continue;\r
1395                                                 }\r
1396                                                 bool v = eval_val (s.Substring (0, j));\r
1397                                                 s = s.Substring (j);\r
1398                                                 return v;\r
1399                                         }\r
1400                                         bool vv = eval_val (s);\r
1401                                         s = "";\r
1402                                         return vv;\r
1403                                 }\r
1404                         }\r
1405                         Error_InvalidDirective ();\r
1406                         return false;\r
1407                 }\r
1408                 \r
1409                 bool pp_unary (ref string s)\r
1410                 {\r
1411                         s = s.Trim ();\r
1412                         int len = s.Length;\r
1413 \r
1414                         if (len > 0){\r
1415                                 if (s [0] == '!'){\r
1416                                         if (len > 1 && s [1] == '='){\r
1417                                                 Error_InvalidDirective ();\r
1418                                                 return false;\r
1419                                         }\r
1420                                         s = s.Substring (1);\r
1421                                         return ! pp_primary (ref s);\r
1422                                 } else\r
1423                                         return pp_primary (ref s);\r
1424                         } else {\r
1425                                 Error_InvalidDirective ();\r
1426                                 return false;\r
1427                         }\r
1428                 }\r
1429                 \r
1430                 bool pp_eq (ref string s)\r
1431                 {\r
1432                         bool va = pp_unary (ref s);\r
1433 \r
1434                         s = s.Trim ();\r
1435                         int len = s.Length;\r
1436                         if (len > 0){\r
1437                                 if (s [0] == '='){\r
1438                                         if (len > 2 && s [1] == '='){\r
1439                                                 s = s.Substring (2);\r
1440                                                 return va == pp_unary (ref s);\r
1441                                         } else {\r
1442                                                 Error_InvalidDirective ();\r
1443                                                 return false;\r
1444                                         }\r
1445                                 } else if (s [0] == '!' && len > 1 && s [1] == '='){\r
1446                                         s = s.Substring (2);\r
1447 \r
1448                                         return va != pp_unary (ref s);\r
1449 \r
1450                                 } \r
1451                         }\r
1452 \r
1453                         return va;\r
1454                                 \r
1455                 }\r
1456                 \r
1457                 bool pp_and (ref string s)\r
1458                 {\r
1459                         bool va = pp_eq (ref s);\r
1460 \r
1461                         s = s.Trim ();\r
1462                         int len = s.Length;\r
1463                         if (len > 0){\r
1464                                 if (s [0] == '&'){\r
1465                                         if (len > 2 && s [1] == '&'){\r
1466                                                 s = s.Substring (2);\r
1467                                                 return (va & pp_eq (ref s));\r
1468                                         } else {\r
1469                                                 Error_InvalidDirective ();\r
1470                                                 return false;\r
1471                                         }\r
1472                                 } \r
1473                         }\r
1474                         return va;\r
1475                 }\r
1476                 \r
1477                 //\r
1478                 // Evaluates an expression for `#if' or `#elif'\r
1479                 //\r
1480                 bool pp_expr (ref string s)\r
1481                 {\r
1482                         bool va = pp_and (ref s);\r
1483                         s = s.Trim ();\r
1484                         int len = s.Length;\r
1485                         if (len > 0){\r
1486                                 char c = s [0];\r
1487                                 \r
1488                                 if (c == '|'){\r
1489                                         if (len > 2 && s [1] == '|'){\r
1490                                                 s = s.Substring (2);\r
1491                                                 return va | pp_expr (ref s);\r
1492                                         } else {\r
1493                                                 Error_InvalidDirective ();\r
1494                                                 return false;\r
1495                                         }\r
1496                                 } \r
1497                         }\r
1498                         \r
1499                         return va;\r
1500                 }\r
1501 \r
1502                 bool eval (string s)\r
1503                 {\r
1504                         bool v = pp_expr (ref s);\r
1505                         s = s.Trim ();\r
1506                         if (s.Length != 0){\r
1507                                 Error_InvalidDirective ();\r
1508                                 return false;\r
1509                         }\r
1510 \r
1511                         return v;\r
1512                 }\r
1513                 \r
1514                 void Error_InvalidDirective ()\r
1515                 {\r
1516                         Report.Error (1517, Location, "Invalid pre-processor directive");\r
1517                 }\r
1518 \r
1519                 void Error_UnexpectedDirective (string extra)\r
1520                 {\r
1521                         Report.Error (\r
1522                                 1028, Location,\r
1523                                 "Unexpected processor directive (" + extra + ")");\r
1524                 }\r
1525 \r
1526                 void Error_TokensSeen ()\r
1527                 {\r
1528                         Report.Error (\r
1529                                 1032, Location,\r
1530                                 "Cannot define or undefine pre-processor symbols after a token in the file");\r
1531                 }\r
1532                 \r
1533                 //\r
1534                 // if true, then the code continues processing the code\r
1535                 // if false, the code stays in a loop until another directive is\r
1536                 // reached.\r
1537                 //\r
1538                 bool handle_preprocessing_directive (bool caller_is_taking)\r
1539                 {\r
1540                         string cmd, arg;\r
1541                         bool region_directive = false;\r
1542 \r
1543                         get_cmd_arg (out cmd, out arg);\r
1544 \r
1545                         // Eat any trailing whitespaces and single-line comments\r
1546                         if (arg.IndexOf ("//") != -1)\r
1547                                 arg = arg.Substring (0, arg.IndexOf ("//"));\r
1548                         arg = arg.TrimEnd (' ', '\t');\r
1549 \r
1550                         //\r
1551                         // The first group of pre-processing instructions is always processed\r
1552                         //\r
1553                         switch (cmd){\r
1554                         case "pragma":\r
1555                                 if (RootContext.V2)\r
1556                                         return true;\r
1557                                 break;\r
1558                                 \r
1559                         case "line":\r
1560                                 if (!PreProcessLine (arg))\r
1561                                         Report.Error (\r
1562                                                 1576, Location,\r
1563                                                 "Argument to #line directive is missing or invalid");\r
1564                                 return true;\r
1565 \r
1566                         case "region":\r
1567                                 region_directive = true;\r
1568                                 arg = "true";\r
1569                                 goto case "if";\r
1570 \r
1571                         case "endregion":\r
1572                                 region_directive = true;\r
1573                                 goto case "endif";\r
1574                                 \r
1575                         case "if":\r
1576                                 if (arg == ""){\r
1577                                         Error_InvalidDirective ();\r
1578                                         return true;\r
1579                                 }\r
1580                                 bool taking = false;\r
1581                                 if (ifstack == null)\r
1582                                         ifstack = new Stack ();\r
1583 \r
1584                                 if (ifstack.Count == 0){\r
1585                                         taking = true;\r
1586                                 } else {\r
1587                                         int state = (int) ifstack.Peek ();\r
1588                                         if ((state & TAKING) != 0)\r
1589                                                 taking = true;\r
1590                                 }\r
1591 \r
1592                                 if (eval (arg) && taking){\r
1593                                         int push = TAKING | TAKEN_BEFORE | PARENT_TAKING;\r
1594                                         if (region_directive)\r
1595                                                 push |= REGION;\r
1596                                         ifstack.Push (push);\r
1597                                         return true;\r
1598                                 } else {\r
1599                                         int push = (taking ? PARENT_TAKING : 0);\r
1600                                         if (region_directive)\r
1601                                                 push |= REGION;\r
1602                                         ifstack.Push (push);\r
1603                                         return false;\r
1604                                 }\r
1605                                 \r
1606                         case "endif":\r
1607                                 if (ifstack == null || ifstack.Count == 0){\r
1608                                         Error_UnexpectedDirective ("no #if for this #endif");\r
1609                                         return true;\r
1610                                 } else {\r
1611                                         int pop = (int) ifstack.Pop ();\r
1612                                         \r
1613                                         if (region_directive && ((pop & REGION) == 0))\r
1614                                                 Report.Error (1027, Location, "#endif directive expected");\r
1615                                         else if (!region_directive && ((pop & REGION) != 0))\r
1616                                                 Report.Error (1038, Location, "#endregion directive expected");\r
1617                                         \r
1618                                         if (ifstack.Count == 0)\r
1619                                                 return true;\r
1620                                         else {\r
1621                                                 int state = (int) ifstack.Peek ();\r
1622 \r
1623                                                 if ((state & TAKING) != 0)\r
1624                                                         return true;\r
1625                                                 else\r
1626                                                         return false;\r
1627                                         }\r
1628                                 }\r
1629 \r
1630                         case "elif":\r
1631                                 if (ifstack == null || ifstack.Count == 0){\r
1632                                         Error_UnexpectedDirective ("no #if for this #elif");\r
1633                                         return true;\r
1634                                 } else {\r
1635                                         int state = (int) ifstack.Peek ();\r
1636 \r
1637                                         if ((state & REGION) != 0) {\r
1638                                                 Report.Error (1038, Location, "#endregion directive expected");\r
1639                                                 return true;\r
1640                                         }\r
1641 \r
1642                                         if ((state & ELSE_SEEN) != 0){\r
1643                                                 Error_UnexpectedDirective ("#elif not valid after #else");\r
1644                                                 return true;\r
1645                                         }\r
1646 \r
1647                                         if ((state & (TAKEN_BEFORE | TAKING)) != 0)\r
1648                                                 return false;\r
1649 \r
1650                                         if (eval (arg) && ((state & PARENT_TAKING) != 0)){\r
1651                                                 state = (int) ifstack.Pop ();\r
1652                                                 ifstack.Push (state | TAKING | TAKEN_BEFORE);\r
1653                                                 return true;\r
1654                                         } else \r
1655                                                 return false;\r
1656                                 }\r
1657 \r
1658                         case "else":\r
1659                                 if (ifstack == null || ifstack.Count == 0){\r
1660                                         Report.Error (\r
1661                                                 1028, Location,\r
1662                                                 "Unexpected processor directive (no #if for this #else)");\r
1663                                         return true;\r
1664                                 } else {\r
1665                                         int state = (int) ifstack.Peek ();\r
1666 \r
1667                                         if ((state & REGION) != 0) {\r
1668                                                 Report.Error (1038, Location, "#endregion directive expected");\r
1669                                                 return true;\r
1670                                         }\r
1671 \r
1672                                         if ((state & ELSE_SEEN) != 0){\r
1673                                                 Error_UnexpectedDirective ("#else within #else");\r
1674                                                 return true;\r
1675                                         }\r
1676 \r
1677                                         ifstack.Pop ();\r
1678 \r
1679                                         bool ret;\r
1680                                         if ((state & TAKEN_BEFORE) == 0){\r
1681                                                 ret = ((state & PARENT_TAKING) != 0);\r
1682                                         } else\r
1683                                                 ret = false;\r
1684                                         \r
1685                                         if (ret)\r
1686                                                 state |= TAKING;\r
1687                                         else\r
1688                                                 state &= ~TAKING;\r
1689                                         \r
1690                                         ifstack.Push (state | ELSE_SEEN);\r
1691                                         \r
1692                                         return ret;\r
1693                                 }\r
1694                         }\r
1695 \r
1696                         //\r
1697                         // These are only processed if we are in a `taking' block\r
1698                         //\r
1699                         if (!caller_is_taking)\r
1700                                 return false;\r
1701                                         \r
1702                         switch (cmd){\r
1703                         case "define":\r
1704                                 if (any_token_seen){\r
1705                                         Error_TokensSeen ();\r
1706                                         return true;\r
1707                                 }\r
1708                                 PreProcessDefinition (true, arg);\r
1709                                 return true;\r
1710 \r
1711                         case "undef":\r
1712                                 if (any_token_seen){\r
1713                                         Error_TokensSeen ();\r
1714                                         return true;\r
1715                                 }\r
1716                                 PreProcessDefinition (false, arg);\r
1717                                 return true;\r
1718 \r
1719                         case "error":\r
1720                                 Report.Error (1029, Location, "#error: '" + arg + "'");\r
1721                                 return true;\r
1722 \r
1723                         case "warning":\r
1724                                 Report.Warning (1030, Location, "#warning: '" + arg + "'");\r
1725                                 return true;\r
1726                         }\r
1727 \r
1728                         Report.Error (1024, Location, "Preprocessor directive expected (got: " + cmd + ")");\r
1729                         return true;\r
1730 \r
1731                 }\r
1732 \r
1733                 private int consume_string (bool quoted) \r
1734                 {\r
1735                         int c;\r
1736                         string_builder.Length = 0;\r
1737                                                                 \r
1738                         while ((c = getChar ()) != -1){\r
1739                                 if (c == '"'){\r
1740                                         if (quoted && peekChar () == '"'){\r
1741                                                 string_builder.Append ((char) c);\r
1742                                                 getChar ();\r
1743                                                 continue;\r
1744                                         } else {\r
1745                                                 val = string_builder.ToString ();\r
1746                                                 return Token.LITERAL_STRING;\r
1747                                         }\r
1748                                 }\r
1749 \r
1750                                 if (c == '\n'){\r
1751                                         if (!quoted)\r
1752                                                 Report.Error (1010, Location, "Newline in constant");\r
1753                                         line++;\r
1754                                         ref_line++;\r
1755                                         col = 0;\r
1756                                 } else\r
1757                                         col++;\r
1758 \r
1759                                 if (!quoted){\r
1760                                         c = escape (c);\r
1761                                         if (c == -1)\r
1762                                                 return Token.ERROR;\r
1763                                 }\r
1764                                 string_builder.Append ((char) c);\r
1765                         }\r
1766 \r
1767                         Report.Error (1039, Location, "Unterminated string literal");\r
1768                         return Token.EOF;\r
1769                 }\r
1770 \r
1771                 private int consume_identifier (int s, bool quoted) \r
1772                 {\r
1773                         int pos = 1;\r
1774                         int c;\r
1775                         \r
1776                         id_builder [0] = (char) s;\r
1777                                         \r
1778                         while ((c = reader.Read ()) != -1) {\r
1779                                 if (is_identifier_part_character ((char) c)){\r
1780                                         if (pos == max_id_size){\r
1781                                                 Report.Error (645, Location, "Identifier too long (limit is 512 chars)");\r
1782                                                 return Token.ERROR;\r
1783                                         }\r
1784                                         \r
1785                                         id_builder [pos++] = (char) c;\r
1786                                         putback_char = -1;\r
1787                                         col++;\r
1788                                 } else {\r
1789                                         putback_char = c;\r
1790                                         break;\r
1791                                 }\r
1792                         }\r
1793 \r
1794                         //\r
1795                         // Optimization: avoids doing the keyword lookup\r
1796                         // on uppercase letters and _\r
1797                         //\r
1798                         if (s >= 'a' || s == '_'){\r
1799                                 int keyword = GetKeyword (id_builder, pos);\r
1800                                 if (keyword != -1 && !quoted)\r
1801                                 return keyword;\r
1802                         }\r
1803 \r
1804                         //\r
1805                         // Keep identifiers in an array of hashtables to avoid needless\r
1806                         // allocations\r
1807                         //\r
1808 \r
1809                         if (identifiers [pos] != null) {\r
1810                                 val = identifiers [pos][id_builder];\r
1811                                 if (val != null) {\r
1812                                         return Token.IDENTIFIER;\r
1813                                 }\r
1814                         }\r
1815                         else\r
1816                                 identifiers [pos] = new CharArrayHashtable (pos);\r
1817 \r
1818                         val = new String (id_builder, 0, pos);\r
1819 \r
1820                         char [] chars = new char [pos];\r
1821                         Array.Copy (id_builder, chars, pos);\r
1822 \r
1823                         identifiers [pos] [chars] = val;\r
1824 \r
1825                         return Token.IDENTIFIER;\r
1826                 }\r
1827 \r
1828                 int consume_whitespace ()\r
1829                 {\r
1830                         int t;\r
1831                         bool doread = false;\r
1832                         int c;\r
1833 \r
1834                         val = null;\r
1835                         // optimization: eliminate col and implement #directive semantic correctly.\r
1836                         for (;(c = getChar ()) != -1; col++) {\r
1837                                 if (c == ' ')\r
1838                                         continue;\r
1839                                 \r
1840                                 if (c == '\t') {\r
1841                                         col = (((col + 8) / 8) * 8) - 1;\r
1842                                         continue;\r
1843                                 }\r
1844                                 \r
1845                                 if (c == ' ' || c == '\f' || c == '\v' || c == 0xa0)\r
1846                                         continue;\r
1847 \r
1848                                 if (c == '\r') {\r
1849                                         if (peekChar () == '\n')\r
1850                                                 getChar ();\r
1851 \r
1852                                         line++;\r
1853                                         ref_line++;\r
1854                                         col = 0;\r
1855                                         any_token_seen |= tokens_seen;\r
1856                                         tokens_seen = false;\r
1857                                         continue;\r
1858                                 }\r
1859 \r
1860                                 // Handle double-slash comments.\r
1861                                 if (c == '/'){\r
1862                                         int d = peekChar ();\r
1863                                 \r
1864                                         if (d == '/'){\r
1865                                                 getChar ();\r
1866                                                 while ((d = getChar ()) != -1 && (d != '\n') && d != '\r')\r
1867                                                         col++;\r
1868                                                 if (d == '\n'){\r
1869                                                         line++;\r
1870                                                         ref_line++;\r
1871                                                         col = 0;\r
1872                                                 }\r
1873                                                 any_token_seen |= tokens_seen;\r
1874                                                 tokens_seen = false;\r
1875                                                 continue;\r
1876                                         } else if (d == '*'){\r
1877                                                 getChar ();\r
1878 \r
1879                                                 while ((d = getChar ()) != -1){\r
1880                                                         if (d == '*' && peekChar () == '/'){\r
1881                                                                 getChar ();\r
1882                                                                 col++;\r
1883                                                                 break;\r
1884                                                         }\r
1885                                                         if (d == '\n'){\r
1886                                                                 line++;\r
1887                                                                 ref_line++;\r
1888                                                                 col = 0;\r
1889                                                                 any_token_seen |= tokens_seen;\r
1890                                                                 tokens_seen = false;\r
1891                                                         }\r
1892                                                 }\r
1893                                                 continue;\r
1894                                         }\r
1895                                         goto is_punct_label;\r
1896                                 }\r
1897 \r
1898                         is_punct_label:\r
1899                                 // white space\r
1900                                 if (c == '\n'){\r
1901                                         line++;\r
1902                                         ref_line++;\r
1903                                         col = 0;\r
1904                                         any_token_seen |= tokens_seen;\r
1905                                         tokens_seen = false;\r
1906                                         continue;\r
1907                                 }\r
1908 \r
1909                                 /* For now, ignore pre-processor commands */\r
1910                                 // FIXME: In C# the '#' is not limited to appear\r
1911                                 // on the first column.\r
1912                                 if (c == '#' && !tokens_seen){\r
1913                                         bool cont = true;\r
1914                                         \r
1915                                 start_again:\r
1916                                         \r
1917                                         cont = handle_preprocessing_directive (cont);\r
1918 \r
1919                                         if (cont){\r
1920                                                 col = 0;\r
1921                                                 continue;\r
1922                                         }\r
1923                                         col = 1;\r
1924 \r
1925                                         bool skipping = false;\r
1926                                         for (;(c = getChar ()) != -1; col++){\r
1927                                                 if (c == '\n'){\r
1928                                                         col = 0;\r
1929                                                         line++;\r
1930                                                         ref_line++;\r
1931                                                         skipping = false;\r
1932                                                 } else if (c == ' ' || c == '\t' || c == '\v' || c == '\r' || c == 0xa0)\r
1933                                                         continue;\r
1934                                                 else if (c != '#')\r
1935                                                         skipping = true;\r
1936                                                 if (c == '#' && !skipping)\r
1937                                                         goto start_again;\r
1938                                         }\r
1939                                         any_token_seen |= tokens_seen;\r
1940                                         tokens_seen = false;\r
1941                                         if (c == -1)\r
1942                                                 Report.Error (1027, Location, "#endif/#endregion expected");\r
1943                                         continue;\r
1944                                 }\r
1945 \r
1946                                 return c;\r
1947                         }\r
1948 \r
1949                         return -1;\r
1950                 }\r
1951                 \r
1952                 public int xtoken ()\r
1953                 {\r
1954                         int t;\r
1955                         bool doread = false;\r
1956                         int c;\r
1957 \r
1958                         val = null;\r
1959                         // optimization: eliminate col and implement #directive semantic correctly.\r
1960 \r
1961                         c = consume_whitespace ();\r
1962                         if (c == -1)\r
1963                                 return Token.EOF;\r
1964 \r
1965                         if (is_identifier_start_character ((char)c)){\r
1966                                 tokens_seen = true;\r
1967                                 return consume_identifier (c, false);\r
1968                         }\r
1969 \r
1970                 is_punct_label:\r
1971                         if ((t = is_punct ((char)c, ref doread)) != Token.ERROR){\r
1972                                 tokens_seen = true;\r
1973                                 if (doread){\r
1974                                         getChar ();\r
1975                                         col++;\r
1976                                 }\r
1977                                 return t;\r
1978                         }\r
1979 \r
1980                         if (c >= '0' && c <= '9'){\r
1981                                 tokens_seen = true;\r
1982                                 return is_number (c);\r
1983                         }\r
1984 \r
1985                         if (c == '.'){\r
1986                                 tokens_seen = true;\r
1987                                 int peek = peekChar ();\r
1988                                 if (peek >= '0' && peek <= '9')\r
1989                                         return is_number (c);\r
1990                                 return Token.DOT;\r
1991                         }\r
1992 \r
1993                         if (c == '"') \r
1994                                 return consume_string (false);\r
1995 \r
1996                         if (c == '\''){\r
1997                                 c = getChar ();\r
1998                                 tokens_seen = true;\r
1999                                 if (c == '\''){\r
2000                                         error_details = "Empty character literal";\r
2001                                         Report.Error (1011, Location, error_details);\r
2002                                         return Token.ERROR;\r
2003                                 }\r
2004                                 c = escape (c);\r
2005                                 if (c == -1)\r
2006                                         return Token.ERROR;\r
2007                                 val = new System.Char ();\r
2008                                 val = (char) c;\r
2009                                 c = getChar ();\r
2010 \r
2011                                 if (c != '\''){\r
2012                                         error_details = "Too many characters in character literal";\r
2013                                         Report.Error (1012, Location, error_details);\r
2014 \r
2015                                         // Try to recover, read until newline or next "'"\r
2016                                         while ((c = getChar ()) != -1){\r
2017                                                 if (c == '\n' || c == '\''){\r
2018                                                         line++;\r
2019                                                         ref_line++;\r
2020                                                         col = 0;\r
2021                                                         break;\r
2022                                                 } else\r
2023                                                         col++;\r
2024                                         }\r
2025                                         return Token.ERROR;\r
2026                                 }\r
2027                                 return Token.LITERAL_CHARACTER;\r
2028                         }\r
2029                                 \r
2030                         if (c == '@') {\r
2031                                 c = getChar ();\r
2032                                 if (c == '"') {\r
2033                                         tokens_seen = true;\r
2034                                         return consume_string (true);\r
2035                                 } else if (is_identifier_start_character ((char) c)){\r
2036                                         return consume_identifier (c, true);\r
2037                                 } else {\r
2038                                         Report.Error (1033, Location, "'@' must be followed by string constant or identifier");\r
2039                                 }\r
2040                         }\r
2041 \r
2042                         if (c == '#') {\r
2043                                 error_details = "Preprocessor directives must appear as the first non-whitespace " +\r
2044                                         "character on a line.";\r
2045 \r
2046                                 Report.Error (1040, Location, error_details);\r
2047 \r
2048                                 return Token.ERROR;\r
2049                         }\r
2050 \r
2051                         error_details = ((char)c).ToString ();\r
2052 \r
2053                         return Token.ERROR;\r
2054                 }\r
2055 \r
2056                 public void cleanup ()\r
2057                 {\r
2058                         if (ifstack != null && ifstack.Count >= 1) {\r
2059                                 int state = (int) ifstack.Pop ();\r
2060                                 if ((state & REGION) != 0)\r
2061                                         Report.Error (1038, "#endregion directive expected");\r
2062                                 else \r
2063                                         Report.Error (1027, "#endif directive expected");\r
2064                         }\r
2065                                 \r
2066                 }\r
2067         }\r
2068 }\r
2069 \r