2007-03-19 Chris Toshok <toshok@ximian.com>
[mcs.git] / mcs / cs-tokenizer.cs
blobed10088f42f6983fef2bc7a6ca663b43d6e59d84
1 //
2 // cs-tokenizer.cs: The Tokenizer for the C# compiler
3 // This also implements the preprocessor
4 //
5 // Author: Miguel de Icaza (miguel@gnu.org)
6 // Marek Safar (marek.safar@seznam.cz)
7 //
8 // Licensed under the terms of the GNU GPL
9 //
10 // (C) 2001, 2002 Ximian, Inc (http://www.ximian.com)
11 // (C) 2004 Novell, Inc
15 using System;
16 using System.Text;
17 using System.Collections;
18 using System.IO;
19 using System.Globalization;
20 using System.Reflection;
22 namespace Mono.CSharp
24 /// <summary>
25 /// Tokenizer for C# source code.
26 /// </summary>
28 public class Tokenizer : yyParser.yyInput
30 SeekableStreamReader reader;
31 SourceFile ref_name;
32 SourceFile file_name;
33 int ref_line = 1;
34 int line = 1;
35 int col = 0;
36 int previous_col;
37 int current_token;
38 bool handle_get_set = false;
39 bool handle_remove_add = false;
40 bool handle_assembly = false;
41 bool handle_constraints = false;
42 bool handle_typeof = false;
43 bool query_parsing;
44 Location current_location;
45 Location current_comment_location = Location.Null;
46 ArrayList escapedIdentifiers = new ArrayList ();
48 static bool linq;
51 // XML documentation buffer. The save point is used to divide
52 // comments on types and comments on members.
54 StringBuilder xml_comment_buffer;
57 // See comment on XmlCommentState enumeration.
59 XmlCommentState xmlDocState = XmlCommentState.Allowed;
62 // Whether tokens have been seen on this line
64 bool tokens_seen = false;
67 // Whether a token has been seen on the file
68 // This is needed because `define' is not allowed to be used
69 // after a token has been seen.
71 bool any_token_seen = false;
73 static Hashtable tokenValues;
74 static readonly char[] simple_whitespaces = new char[] { ' ', '\t' };
76 private static Hashtable TokenValueName
78 get {
79 if (tokenValues == null)
80 tokenValues = GetTokenValueNameHash ();
82 return tokenValues;
86 private static Hashtable GetTokenValueNameHash ()
88 Type t = typeof (Token);
89 FieldInfo [] fields = t.GetFields ();
90 Hashtable hash = new Hashtable ();
91 foreach (FieldInfo field in fields) {
92 if (field.IsLiteral && field.IsStatic && field.FieldType == typeof (int))
93 hash.Add (field.GetValue (null), field.Name);
95 return hash;
99 // Returns a verbose representation of the current location
101 public string location {
102 get {
103 string det;
105 if (current_token == Token.ERROR)
106 det = "detail: " + error_details;
107 else
108 det = "";
110 // return "Line: "+line+" Col: "+col + "\n" +
111 // "VirtLine: "+ref_line +
112 // " Token: "+current_token + " " + det;
113 string current_token_name = TokenValueName [current_token] as string;
114 if (current_token_name == null)
115 current_token_name = current_token.ToString ();
117 return String.Format ("{0} ({1},{2}), Token: {3} {4}", ref_name.Name,
118 ref_line,
119 col,
120 current_token_name,
121 det);
125 public bool PropertyParsing {
126 get {
127 return handle_get_set;
130 set {
131 handle_get_set = value;
135 public bool AssemblyTargetParsing {
136 get {
137 return handle_assembly;
140 set {
141 handle_assembly = value;
145 public bool EventParsing {
146 get {
147 return handle_remove_add;
150 set {
151 handle_remove_add = value;
155 public bool ConstraintsParsing {
156 get {
157 return handle_constraints;
160 set {
161 handle_constraints = value;
165 public static bool LinqEnabled {
166 set {
167 linq = value;
171 public bool TypeOfParsing {
172 get {
173 return handle_typeof;
176 set {
177 handle_typeof = value;
181 public bool QueryParsing {
182 set {
183 query_parsing = value;
187 public XmlCommentState doc_state {
188 get { return xmlDocState; }
189 set {
190 if (value == XmlCommentState.Allowed) {
191 check_incorrect_doc_comment ();
192 reset_doc_comment ();
194 xmlDocState = value;
198 public bool IsEscapedIdentifier (Location loc)
200 foreach (LocatedToken lt in escapedIdentifiers)
201 if (lt.Location.Equals (loc))
202 return true;
203 return false;
207 // Class variables
209 static CharArrayHashtable[] keywords;
210 static Hashtable keywordStrings;
211 static NumberStyles styles;
212 static NumberFormatInfo csharp_format_info;
215 // Values for the associated token returned
217 int putback_char;
218 Object val;
221 // Pre-processor
223 Hashtable defines;
225 const int TAKING = 1;
226 const int ELSE_SEEN = 4;
227 const int PARENT_TAKING = 8;
228 const int REGION = 16;
231 // pre-processor if stack state:
233 Stack ifstack;
235 static System.Text.StringBuilder string_builder;
237 const int max_id_size = 512;
238 static char [] id_builder = new char [max_id_size];
240 static CharArrayHashtable [] identifiers = new CharArrayHashtable [max_id_size + 1];
242 const int max_number_size = 512;
243 static char [] number_builder = new char [max_number_size];
244 static int number_pos;
247 // Details about the error encoutered by the tokenizer
249 string error_details;
251 public string error {
252 get {
253 return error_details;
257 public int Line {
258 get {
259 return ref_line;
263 public int Col {
264 get {
265 return col;
270 // This is used when the tokenizer needs to save
271 // the current position as it needs to do some parsing
272 // on its own to deamiguate a token in behalf of the
273 // parser.
275 Stack position_stack = new Stack (2);
276 class Position {
277 public int position;
278 public int ref_line;
279 public int col;
280 public int putback_char;
281 public int previous_col;
282 public Stack ifstack;
283 public int parsing_generic_less_than;
284 public int current_token;
286 public Position (Tokenizer t)
288 position = t.reader.Position;
289 ref_line = t.ref_line;
290 col = t.col;
291 putback_char = t.putback_char;
292 previous_col = t.previous_col;
293 if (t.ifstack != null && t.ifstack.Count != 0)
294 ifstack = (Stack)t.ifstack.Clone ();
295 parsing_generic_less_than = t.parsing_generic_less_than;
296 current_token = t.current_token;
300 public void PushPosition ()
302 position_stack.Push (new Position (this));
305 public void PopPosition ()
307 Position p = (Position) position_stack.Pop ();
309 reader.Position = p.position;
310 ref_line = p.ref_line;
311 col = p.col;
312 putback_char = p.putback_char;
313 previous_col = p.previous_col;
314 ifstack = p.ifstack;
315 parsing_generic_less_than = p.parsing_generic_less_than;
316 current_token = p.current_token;
319 // Do not reset the position, ignore it.
320 public void DiscardPosition ()
322 position_stack.Pop ();
325 static void AddKeyword (string kw, int token)
327 keywordStrings.Add (kw, kw);
328 if (keywords [kw.Length] == null) {
329 keywords [kw.Length] = new CharArrayHashtable (kw.Length);
331 keywords [kw.Length] [kw.ToCharArray ()] = token;
334 static void InitTokens ()
336 keywordStrings = new Hashtable ();
337 keywords = new CharArrayHashtable [64];
339 AddKeyword ("__arglist", Token.ARGLIST);
340 AddKeyword ("abstract", Token.ABSTRACT);
341 AddKeyword ("as", Token.AS);
342 AddKeyword ("add", Token.ADD);
343 AddKeyword ("assembly", Token.ASSEMBLY);
344 AddKeyword ("base", Token.BASE);
345 AddKeyword ("bool", Token.BOOL);
346 AddKeyword ("break", Token.BREAK);
347 AddKeyword ("byte", Token.BYTE);
348 AddKeyword ("case", Token.CASE);
349 AddKeyword ("catch", Token.CATCH);
350 AddKeyword ("char", Token.CHAR);
351 AddKeyword ("checked", Token.CHECKED);
352 AddKeyword ("class", Token.CLASS);
353 AddKeyword ("const", Token.CONST);
354 AddKeyword ("continue", Token.CONTINUE);
355 AddKeyword ("decimal", Token.DECIMAL);
356 AddKeyword ("default", Token.DEFAULT);
357 AddKeyword ("delegate", Token.DELEGATE);
358 AddKeyword ("do", Token.DO);
359 AddKeyword ("double", Token.DOUBLE);
360 AddKeyword ("else", Token.ELSE);
361 AddKeyword ("enum", Token.ENUM);
362 AddKeyword ("event", Token.EVENT);
363 AddKeyword ("explicit", Token.EXPLICIT);
364 AddKeyword ("extern", Token.EXTERN);
365 AddKeyword ("false", Token.FALSE);
366 AddKeyword ("finally", Token.FINALLY);
367 AddKeyword ("fixed", Token.FIXED);
368 AddKeyword ("float", Token.FLOAT);
369 AddKeyword ("for", Token.FOR);
370 AddKeyword ("foreach", Token.FOREACH);
371 AddKeyword ("goto", Token.GOTO);
372 AddKeyword ("get", Token.GET);
373 AddKeyword ("if", Token.IF);
374 AddKeyword ("implicit", Token.IMPLICIT);
375 AddKeyword ("in", Token.IN);
376 AddKeyword ("int", Token.INT);
377 AddKeyword ("interface", Token.INTERFACE);
378 AddKeyword ("internal", Token.INTERNAL);
379 AddKeyword ("is", Token.IS);
380 AddKeyword ("lock", Token.LOCK);
381 AddKeyword ("long", Token.LONG);
382 AddKeyword ("namespace", Token.NAMESPACE);
383 AddKeyword ("new", Token.NEW);
384 AddKeyword ("null", Token.NULL);
385 AddKeyword ("object", Token.OBJECT);
386 AddKeyword ("operator", Token.OPERATOR);
387 AddKeyword ("out", Token.OUT);
388 AddKeyword ("override", Token.OVERRIDE);
389 AddKeyword ("params", Token.PARAMS);
390 AddKeyword ("private", Token.PRIVATE);
391 AddKeyword ("protected", Token.PROTECTED);
392 AddKeyword ("public", Token.PUBLIC);
393 AddKeyword ("readonly", Token.READONLY);
394 AddKeyword ("ref", Token.REF);
395 AddKeyword ("remove", Token.REMOVE);
396 AddKeyword ("return", Token.RETURN);
397 AddKeyword ("sbyte", Token.SBYTE);
398 AddKeyword ("sealed", Token.SEALED);
399 AddKeyword ("set", Token.SET);
400 AddKeyword ("short", Token.SHORT);
401 AddKeyword ("sizeof", Token.SIZEOF);
402 AddKeyword ("stackalloc", Token.STACKALLOC);
403 AddKeyword ("static", Token.STATIC);
404 AddKeyword ("string", Token.STRING);
405 AddKeyword ("struct", Token.STRUCT);
406 AddKeyword ("switch", Token.SWITCH);
407 AddKeyword ("this", Token.THIS);
408 AddKeyword ("throw", Token.THROW);
409 AddKeyword ("true", Token.TRUE);
410 AddKeyword ("try", Token.TRY);
411 AddKeyword ("typeof", Token.TYPEOF);
412 AddKeyword ("uint", Token.UINT);
413 AddKeyword ("ulong", Token.ULONG);
414 AddKeyword ("unchecked", Token.UNCHECKED);
415 AddKeyword ("unsafe", Token.UNSAFE);
416 AddKeyword ("ushort", Token.USHORT);
417 AddKeyword ("using", Token.USING);
418 AddKeyword ("virtual", Token.VIRTUAL);
419 AddKeyword ("void", Token.VOID);
420 AddKeyword ("volatile", Token.VOLATILE);
421 AddKeyword ("while", Token.WHILE);
422 AddKeyword ("partial", Token.PARTIAL);
423 #if GMCS_SOURCE
424 AddKeyword ("where", Token.WHERE);
426 if (linq) {
427 AddKeyword ("from", Token.FROM);
428 AddKeyword ("join", Token.JOIN);
429 AddKeyword ("on", Token.ON);
430 AddKeyword ("equals", Token.EQUALS);
431 AddKeyword ("select", Token.SELECT);
432 AddKeyword ("group", Token.GROUP);
433 AddKeyword ("by", Token.BY);
434 AddKeyword ("let", Token.LET);
435 AddKeyword ("orderby", Token.ORDERBY);
436 AddKeyword ("ascending", Token.ASCENDING);
437 AddKeyword ("descending", Token.DESCENDING);
438 AddKeyword ("into", Token.INTO);
440 #endif
444 // Class initializer
446 static Tokenizer ()
448 Reset ();
451 public static void Reset ()
453 InitTokens ();
454 csharp_format_info = NumberFormatInfo.InvariantInfo;
455 styles = NumberStyles.Float;
457 string_builder = new System.Text.StringBuilder ();
460 int GetKeyword (char[] id, int id_len)
463 * Keywords are stored in an array of hashtables grouped by their
464 * length.
467 if ((id_len >= keywords.Length) || (keywords [id_len] == null))
468 return -1;
469 object o = keywords [id_len] [id];
471 if (o == null)
472 return -1;
474 int res = (int) o;
476 if (handle_get_set == false && (res == Token.GET || res == Token.SET))
477 return -1;
478 if (handle_remove_add == false && (res == Token.REMOVE || res == Token.ADD))
479 return -1;
480 if (handle_assembly == false && res == Token.ASSEMBLY)
481 return -1;
482 #if GMCS_SOURCE
483 if (linq) {
484 if (res == Token.FROM &&
485 (current_token == Token.ASSIGN || current_token == Token.OPEN_BRACKET ||
486 current_token == Token.RETURN || current_token == Token.IN)) {
487 query_parsing = true;
488 return res;
491 if (!query_parsing && res > Token.QUERY_FIRST_TOKEN && res < Token.QUERY_LAST_TOKEN)
492 return -1;
494 return res;
497 if (!handle_constraints && res == Token.WHERE)
498 return -1;
499 #endif
500 return res;
504 public Location Location {
505 get { return current_location; }
508 void define (string def)
510 if (!RootContext.AllDefines.Contains (def)){
511 RootContext.AllDefines [def] = true;
513 if (defines.Contains (def))
514 return;
515 defines [def] = true;
518 public Tokenizer (SeekableStreamReader input, SourceFile file, ArrayList defs)
520 this.ref_name = file;
521 this.file_name = file;
522 reader = input;
524 putback_char = -1;
526 if (defs != null){
527 defines = new Hashtable ();
528 foreach (string def in defs)
529 define (def);
532 xml_comment_buffer = new StringBuilder ();
535 // FIXME: This could be `Location.Push' but we have to
536 // find out why the MS compiler allows this
538 Mono.CSharp.Location.Push (file);
541 static bool is_identifier_start_character (char c)
543 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_' || Char.IsLetter (c);
546 static bool is_identifier_part_character (char c)
548 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_' || (c >= '0' && c <= '9') ||
549 Char.IsLetter (c) || Char.GetUnicodeCategory (c) == UnicodeCategory.ConnectorPunctuation;
552 public static bool IsKeyword (string s)
554 return keywordStrings [s] != null;
557 public static bool IsValidIdentifier (string s)
559 if (s == null || s.Length == 0)
560 return false;
562 if (!is_identifier_start_character (s [0]))
563 return false;
565 for (int i = 1; i < s.Length; i ++)
566 if (! is_identifier_part_character (s [i]))
567 return false;
569 return true;
572 bool parse_generic_dimension (out int dimension)
574 dimension = 1;
576 again:
577 int the_token = token ();
578 if (the_token == Token.OP_GENERICS_GT)
579 return true;
580 else if (the_token == Token.COMMA) {
581 dimension++;
582 goto again;
585 return false;
588 bool parse_less_than ()
590 start:
591 int the_token = token ();
592 if (the_token == Token.OPEN_BRACKET) {
593 do {
594 the_token = token ();
595 } while (the_token != Token.CLOSE_BRACKET);
596 the_token = token ();
598 switch (the_token) {
599 case Token.IDENTIFIER:
600 case Token.OBJECT:
601 case Token.STRING:
602 case Token.BOOL:
603 case Token.DECIMAL:
604 case Token.FLOAT:
605 case Token.DOUBLE:
606 case Token.SBYTE:
607 case Token.BYTE:
608 case Token.SHORT:
609 case Token.USHORT:
610 case Token.INT:
611 case Token.UINT:
612 case Token.LONG:
613 case Token.ULONG:
614 case Token.CHAR:
615 case Token.VOID:
616 break;
618 default:
619 return false;
621 again:
622 the_token = token ();
624 if (the_token == Token.OP_GENERICS_GT)
625 return true;
626 else if (the_token == Token.COMMA || the_token == Token.DOT || the_token == Token.DOUBLE_COLON)
627 goto start;
628 else if (the_token == Token.INTERR || the_token == Token.STAR)
629 goto again;
630 else if (the_token == Token.OP_GENERICS_LT) {
631 if (!parse_less_than ())
632 return false;
633 goto again;
634 } else if (the_token == Token.OPEN_BRACKET) {
635 rank_specifiers:
636 the_token = token ();
637 if (the_token == Token.CLOSE_BRACKET)
638 goto again;
639 else if (the_token == Token.COMMA)
640 goto rank_specifiers;
641 return false;
644 return false;
647 #if GMCS_SOURCE
648 public void PutbackNullable ()
650 if (nullable_pos < 0)
651 throw new Exception ();
653 current_token = -1;
654 val = null;
655 reader.Position = nullable_pos;
657 putback_char = '?';
660 public void PutbackCloseParens ()
662 putback_char = ')';
666 int nullable_pos = -1;
668 public void CheckNullable (bool is_nullable)
670 if (is_nullable)
671 nullable_pos = reader.Position;
672 else
673 nullable_pos = -1;
675 #endif
677 public int peek_token ()
679 int the_token;
681 PushPosition ();
682 the_token = token ();
683 PopPosition ();
685 return the_token;
688 bool parse_namespace_or_typename (int next)
690 if (next == -1)
691 next = peek_token ();
692 while (next == Token.IDENTIFIER){
693 token ();
694 again:
695 next = peek_token ();
696 if (next == Token.DOT || next == Token.DOUBLE_COLON){
697 token ();
698 next = peek_token ();
699 continue;
701 if (next == Token.OP_GENERICS_LT){
702 token ();
703 if (!parse_less_than ())
704 return false;
705 goto again;
707 return true;
710 return false;
713 bool is_simple_type (int token)
715 return (token == Token.BOOL ||
716 token == Token.DECIMAL ||
717 token == Token.SBYTE ||
718 token == Token.BYTE ||
719 token == Token.SHORT ||
720 token == Token.USHORT ||
721 token == Token.INT ||
722 token == Token.UINT ||
723 token == Token.LONG ||
724 token == Token.ULONG ||
725 token == Token.CHAR ||
726 token == Token.FLOAT ||
727 token == Token.DOUBLE);
730 bool is_builtin_reference_type (int token)
732 return (token == Token.OBJECT || token == Token.STRING);
735 bool parse_opt_rank (int next)
737 while (true){
738 if (next != Token.OPEN_BRACKET)
739 return true;
741 token ();
742 while (true){
743 next = token ();
744 if (next == Token.CLOSE_BRACKET){
745 next = peek_token ();
746 break;
748 if (next == Token.COMMA)
749 continue;
751 return false;
756 bool parse_type ()
758 int next = peek_token ();
760 if (is_simple_type (next)){
761 token ();
762 next = peek_token ();
763 if (next == Token.INTERR)
764 token ();
765 return parse_opt_rank (peek_token ());
767 if (parse_namespace_or_typename (next)){
768 next = peek_token ();
769 if (next == Token.INTERR)
770 token ();
771 return parse_opt_rank (peek_token ());
772 } else if (is_builtin_reference_type (next)){
773 token ();
774 return parse_opt_rank (peek_token ());
777 return false;
781 // Invoked after '(' has been seen and tries to parse:
782 // type identifier [, type identifier]*
784 // if this is the case, instead of returning an
785 // OPEN_PARENS token we return a special token that
786 // triggers lambda parsing.
788 // This is needed because we can not introduce the
789 // explicitly_typed_lambda_parameter_list after a '(' in the
790 // grammar without introducing reduce/reduce conflicts.
792 // We need to parse a type and if it is followed by an
793 // identifier, we know it has to be parsed as a lambda
794 // expression.
796 // the type expression can be prefixed with `ref' or `out'
798 public bool parse_lambda_parameters ()
800 while (true){
801 int next = peek_token ();
803 if (next == Token.REF || next == Token.OUT)
804 token ();
806 if (parse_type ()){
807 next = peek_token ();
808 if (next == Token.IDENTIFIER){
809 token ();
810 next = peek_token ();
811 if (next == Token.COMMA){
812 token ();
813 continue;
815 if (next == Token.CLOSE_PARENS)
816 return true;
819 return false;
823 int parsing_generic_less_than = 0;
825 int is_punct (char c, ref bool doread)
827 int d;
828 int t;
830 doread = false;
832 switch (c){
833 case '{':
834 val = Location;
835 return Token.OPEN_BRACE;
836 case '}':
837 val = Location;
838 return Token.CLOSE_BRACE;
839 case '[':
840 // To block doccomment inside attribute declaration.
841 if (doc_state == XmlCommentState.Allowed)
842 doc_state = XmlCommentState.NotAllowed;
843 return Token.OPEN_BRACKET;
844 case ']':
845 return Token.CLOSE_BRACKET;
846 case '(':
847 if (linq){
848 PushPosition ();
849 bool have_lambda_parameter = parse_lambda_parameters ();
850 PopPosition ();
852 if (have_lambda_parameter)
853 return Token.OPEN_PARENS_LAMBDA;
854 else
855 return Token.OPEN_PARENS;
856 } else
857 return Token.OPEN_PARENS;
858 case ')': {
859 if (deambiguate_close_parens == 0)
860 return Token.CLOSE_PARENS;
862 --deambiguate_close_parens;
864 PushPosition ();
866 int new_token = xtoken ();
868 PopPosition ();
870 if (new_token == Token.OPEN_PARENS)
871 return Token.CLOSE_PARENS_OPEN_PARENS;
872 else if (new_token == Token.MINUS)
873 return Token.CLOSE_PARENS_MINUS;
874 else if (IsCastToken (new_token))
875 return Token.CLOSE_PARENS_CAST;
876 else
877 return Token.CLOSE_PARENS_NO_CAST;
880 case ',':
881 return Token.COMMA;
882 case ';':
883 val = Location;
884 return Token.SEMICOLON;
885 case '~':
886 val = Location;
887 return Token.TILDE;
888 case '?':
889 return Token.INTERR;
891 #if GMCS_SOURCE
892 if (c == '<') {
893 if (parsing_generic_less_than++ > 0)
894 return Token.OP_GENERICS_LT;
896 if (handle_typeof) {
897 int dimension;
898 PushPosition ();
899 if (parse_generic_dimension (out dimension)) {
900 val = dimension;
901 DiscardPosition ();
902 return Token.GENERIC_DIMENSION;
904 PopPosition ();
907 // Save current position and parse next token.
908 PushPosition ();
909 bool is_generic_lt = parse_less_than ();
910 PopPosition ();
912 if (is_generic_lt) {
913 return Token.OP_GENERICS_LT;
914 } else
915 parsing_generic_less_than = 0;
917 d = peekChar ();
918 if (d == '<'){
919 getChar ();
920 d = peekChar ();
922 if (d == '='){
923 doread = true;
924 return Token.OP_SHIFT_LEFT_ASSIGN;
926 return Token.OP_SHIFT_LEFT;
927 } else if (d == '='){
928 doread = true;
929 return Token.OP_LE;
931 return Token.OP_LT;
932 } else if (c == '>') {
933 if (parsing_generic_less_than > 0) {
934 parsing_generic_less_than--;
935 return Token.OP_GENERICS_GT;
938 d = peekChar ();
939 if (d == '>'){
940 getChar ();
941 d = peekChar ();
943 if (d == '='){
944 doread = true;
945 return Token.OP_SHIFT_RIGHT_ASSIGN;
947 return Token.OP_SHIFT_RIGHT;
948 } else if (d == '='){
949 doread = true;
950 return Token.OP_GE;
952 return Token.OP_GT;
954 #endif
955 d = peekChar ();
956 if (c == '+'){
958 if (d == '+') {
959 val = Location;
960 t = Token.OP_INC;
962 else if (d == '=')
963 t = Token.OP_ADD_ASSIGN;
964 else {
965 val = Location;
966 return Token.PLUS;
968 doread = true;
969 return t;
971 if (c == '-'){
972 if (d == '-') {
973 val = Location;
974 t = Token.OP_DEC;
976 else if (d == '=')
977 t = Token.OP_SUB_ASSIGN;
978 else if (d == '>')
979 t = Token.OP_PTR;
980 else {
981 val = Location;
982 return Token.MINUS;
984 doread = true;
985 return t;
988 if (c == '!'){
989 if (d == '='){
990 doread = true;
991 return Token.OP_NE;
993 val = Location;
994 return Token.BANG;
997 if (c == '='){
998 if (d == '='){
999 doread = true;
1000 return Token.OP_EQ;
1002 if (d == '>'){
1003 doread = true;
1004 val = Location;
1005 return Token.ARROW;
1007 return Token.ASSIGN;
1010 if (c == '&'){
1011 if (d == '&'){
1012 doread = true;
1013 return Token.OP_AND;
1014 } else if (d == '='){
1015 doread = true;
1016 return Token.OP_AND_ASSIGN;
1018 val = Location;
1019 return Token.BITWISE_AND;
1022 if (c == '|'){
1023 if (d == '|'){
1024 doread = true;
1025 return Token.OP_OR;
1026 } else if (d == '='){
1027 doread = true;
1028 return Token.OP_OR_ASSIGN;
1030 return Token.BITWISE_OR;
1033 if (c == '*'){
1034 if (d == '='){
1035 doread = true;
1036 return Token.OP_MULT_ASSIGN;
1038 val = Location;
1039 return Token.STAR;
1042 if (c == '/'){
1043 if (d == '='){
1044 doread = true;
1045 return Token.OP_DIV_ASSIGN;
1047 return Token.DIV;
1050 if (c == '%'){
1051 if (d == '='){
1052 doread = true;
1053 return Token.OP_MOD_ASSIGN;
1055 return Token.PERCENT;
1058 if (c == '^'){
1059 if (d == '='){
1060 doread = true;
1061 return Token.OP_XOR_ASSIGN;
1063 return Token.CARRET;
1066 #if !GMCS_SOURCE
1067 if (c == '<'){
1068 if (d == '<'){
1069 getChar ();
1070 d = peekChar ();
1072 if (d == '='){
1073 doread = true;
1074 return Token.OP_SHIFT_LEFT_ASSIGN;
1076 return Token.OP_SHIFT_LEFT;
1077 } else if (d == '='){
1078 doread = true;
1079 return Token.OP_LE;
1081 return Token.OP_LT;
1084 if (c == '>'){
1085 if (d == '>'){
1086 getChar ();
1087 d = peekChar ();
1089 if (d == '='){
1090 doread = true;
1091 return Token.OP_SHIFT_RIGHT_ASSIGN;
1093 return Token.OP_SHIFT_RIGHT;
1094 } else if (d == '='){
1095 doread = true;
1096 return Token.OP_GE;
1098 return Token.OP_GT;
1100 #endif
1101 if (c == ':'){
1102 if (d == ':'){
1103 doread = true;
1104 return Token.DOUBLE_COLON;
1106 val = Location;
1107 return Token.COLON;
1110 return Token.ERROR;
1113 int deambiguate_close_parens = 0;
1115 public void Deambiguate_CloseParens (object expression)
1117 putback (')');
1119 // When any binary operation is used we are sure it is not a cast
1120 if (expression is Binary)
1121 return;
1123 deambiguate_close_parens++;
1126 bool decimal_digits (int c)
1128 int d;
1129 bool seen_digits = false;
1131 if (c != -1){
1132 if (number_pos == max_number_size)
1133 Error_NumericConstantTooLong ();
1134 number_builder [number_pos++] = (char) c;
1138 // We use peekChar2, because decimal_digits needs to do a
1139 // 2-character look-ahead (5.ToString for example).
1141 while ((d = peekChar2 ()) != -1){
1142 if (d >= '0' && d <= '9'){
1143 if (number_pos == max_number_size)
1144 Error_NumericConstantTooLong ();
1145 number_builder [number_pos++] = (char) d;
1146 getChar ();
1147 seen_digits = true;
1148 } else
1149 break;
1152 return seen_digits;
1155 static bool is_hex (int e)
1157 return (e >= '0' && e <= '9') || (e >= 'A' && e <= 'F') || (e >= 'a' && e <= 'f');
1160 static int real_type_suffix (int c)
1162 int t;
1164 switch (c){
1165 case 'F': case 'f':
1166 t = Token.LITERAL_FLOAT;
1167 break;
1168 case 'D': case 'd':
1169 t = Token.LITERAL_DOUBLE;
1170 break;
1171 case 'M': case 'm':
1172 t= Token.LITERAL_DECIMAL;
1173 break;
1174 default:
1175 return Token.NONE;
1177 return t;
1180 int integer_type_suffix (ulong ul, int c)
1182 bool is_unsigned = false;
1183 bool is_long = false;
1185 if (c != -1){
1186 bool scanning = true;
1187 do {
1188 switch (c){
1189 case 'U': case 'u':
1190 if (is_unsigned)
1191 scanning = false;
1192 is_unsigned = true;
1193 getChar ();
1194 break;
1196 case 'l':
1197 if (!is_unsigned && (RootContext.WarningLevel >= 4)){
1199 // if we have not seen anything in between
1200 // report this error
1202 Report.Warning (78, 4, Location, "The 'l' suffix is easily confused with the digit '1' (use 'L' for clarity)");
1205 // This goto statement causes the MS CLR 2.0 beta 1 csc to report an error, so
1206 // work around that.
1208 //goto case 'L';
1209 if (is_long)
1210 scanning = false;
1211 is_long = true;
1212 getChar ();
1213 break;
1215 case 'L':
1216 if (is_long)
1217 scanning = false;
1218 is_long = true;
1219 getChar ();
1220 break;
1222 default:
1223 scanning = false;
1224 break;
1226 c = peekChar ();
1227 } while (scanning);
1230 if (is_long && is_unsigned){
1231 val = ul;
1232 return Token.LITERAL_INTEGER;
1233 } else if (is_unsigned){
1234 // uint if possible, or ulong else.
1236 if ((ul & 0xffffffff00000000) == 0)
1237 val = (uint) ul;
1238 else
1239 val = ul;
1240 } else if (is_long){
1241 // long if possible, ulong otherwise
1242 if ((ul & 0x8000000000000000) != 0)
1243 val = ul;
1244 else
1245 val = (long) ul;
1246 } else {
1247 // int, uint, long or ulong in that order
1248 if ((ul & 0xffffffff00000000) == 0){
1249 uint ui = (uint) ul;
1251 if ((ui & 0x80000000) != 0)
1252 val = ui;
1253 else
1254 val = (int) ui;
1255 } else {
1256 if ((ul & 0x8000000000000000) != 0)
1257 val = ul;
1258 else
1259 val = (long) ul;
1262 return Token.LITERAL_INTEGER;
1266 // given `c' as the next char in the input decide whether
1267 // we need to convert to a special type, and then choose
1268 // the best representation for the integer
1270 int adjust_int (int c)
1272 try {
1273 if (number_pos > 9){
1274 ulong ul = (uint) (number_builder [0] - '0');
1276 for (int i = 1; i < number_pos; i++){
1277 ul = checked ((ul * 10) + ((uint)(number_builder [i] - '0')));
1279 return integer_type_suffix (ul, c);
1280 } else {
1281 uint ui = (uint) (number_builder [0] - '0');
1283 for (int i = 1; i < number_pos; i++){
1284 ui = checked ((ui * 10) + ((uint)(number_builder [i] - '0')));
1286 return integer_type_suffix (ui, c);
1288 } catch (OverflowException) {
1289 error_details = "Integral constant is too large";
1290 Report.Error (1021, Location, error_details);
1291 val = 0ul;
1292 return Token.LITERAL_INTEGER;
1294 catch (FormatException) {
1295 Report.Error (1013, Location, "Invalid number");
1296 val = 0ul;
1297 return Token.LITERAL_INTEGER;
1301 int adjust_real (int t)
1303 string s = new String (number_builder, 0, number_pos);
1304 const string error_details = "Floating-point constant is outside the range of type `{0}'";
1306 switch (t){
1307 case Token.LITERAL_DECIMAL:
1308 try {
1309 val = System.Decimal.Parse (s, styles, csharp_format_info);
1310 } catch (OverflowException) {
1311 val = 0m;
1312 Report.Error (594, Location, error_details, "decimal");
1314 break;
1315 case Token.LITERAL_FLOAT:
1316 try {
1317 val = float.Parse (s, styles, csharp_format_info);
1318 } catch (OverflowException) {
1319 val = 0.0f;
1320 Report.Error (594, Location, error_details, "float");
1322 break;
1324 case Token.LITERAL_DOUBLE:
1325 case Token.NONE:
1326 t = Token.LITERAL_DOUBLE;
1327 try {
1328 val = System.Double.Parse (s, styles, csharp_format_info);
1329 } catch (OverflowException) {
1330 val = 0.0;
1331 Report.Error (594, Location, error_details, "double");
1333 break;
1335 return t;
1338 int handle_hex ()
1340 int d;
1341 ulong ul;
1343 getChar ();
1344 while ((d = peekChar ()) != -1){
1345 if (is_hex (d)){
1346 number_builder [number_pos++] = (char) d;
1347 getChar ();
1348 } else
1349 break;
1352 string s = new String (number_builder, 0, number_pos);
1353 try {
1354 if (number_pos <= 8)
1355 ul = System.UInt32.Parse (s, NumberStyles.HexNumber);
1356 else
1357 ul = System.UInt64.Parse (s, NumberStyles.HexNumber);
1358 } catch (OverflowException){
1359 error_details = "Integral constant is too large";
1360 Report.Error (1021, Location, error_details);
1361 val = 0ul;
1362 return Token.LITERAL_INTEGER;
1364 catch (FormatException) {
1365 Report.Error (1013, Location, "Invalid number");
1366 val = 0ul;
1367 return Token.LITERAL_INTEGER;
1370 return integer_type_suffix (ul, peekChar ());
1374 // Invoked if we know we have .digits or digits
1376 int is_number (int c)
1378 bool is_real = false;
1379 int type;
1381 number_pos = 0;
1383 if (c >= '0' && c <= '9'){
1384 if (c == '0'){
1385 int peek = peekChar ();
1387 if (peek == 'x' || peek == 'X')
1388 return handle_hex ();
1390 decimal_digits (c);
1391 c = getChar ();
1395 // We need to handle the case of
1396 // "1.1" vs "1.string" (LITERAL_FLOAT vs NUMBER DOT IDENTIFIER)
1398 if (c == '.'){
1399 if (decimal_digits ('.')){
1400 is_real = true;
1401 c = getChar ();
1402 } else {
1403 putback ('.');
1404 number_pos--;
1405 return adjust_int (-1);
1409 if (c == 'e' || c == 'E'){
1410 is_real = true;
1411 if (number_pos == max_number_size)
1412 Error_NumericConstantTooLong ();
1413 number_builder [number_pos++] = 'e';
1414 c = getChar ();
1416 if (c == '+'){
1417 if (number_pos == max_number_size)
1418 Error_NumericConstantTooLong ();
1419 number_builder [number_pos++] = '+';
1420 c = -1;
1421 } else if (c == '-') {
1422 if (number_pos == max_number_size)
1423 Error_NumericConstantTooLong ();
1424 number_builder [number_pos++] = '-';
1425 c = -1;
1426 } else {
1427 if (number_pos == max_number_size)
1428 Error_NumericConstantTooLong ();
1429 number_builder [number_pos++] = '+';
1432 decimal_digits (c);
1433 c = getChar ();
1436 type = real_type_suffix (c);
1437 if (type == Token.NONE && !is_real){
1438 putback (c);
1439 return adjust_int (c);
1440 } else
1441 is_real = true;
1443 if (type == Token.NONE){
1444 putback (c);
1447 if (is_real)
1448 return adjust_real (type);
1450 Console.WriteLine ("This should not be reached");
1451 throw new Exception ("Is Number should never reach this point");
1455 // Accepts exactly count (4 or 8) hex, no more no less
1457 int getHex (int count, out bool error)
1459 int i;
1460 int total = 0;
1461 int c;
1462 int top = count != -1 ? count : 4;
1464 getChar ();
1465 error = false;
1466 for (i = 0; i < top; i++){
1467 c = getChar ();
1469 if (c >= '0' && c <= '9')
1470 c = (int) c - (int) '0';
1471 else if (c >= 'A' && c <= 'F')
1472 c = (int) c - (int) 'A' + 10;
1473 else if (c >= 'a' && c <= 'f')
1474 c = (int) c - (int) 'a' + 10;
1475 else {
1476 error = true;
1477 return 0;
1480 total = (total * 16) + c;
1481 if (count == -1){
1482 int p = peekChar ();
1483 if (p == -1)
1484 break;
1485 if (!is_hex ((char)p))
1486 break;
1489 return total;
1492 int escape (int c)
1494 bool error;
1495 int d;
1496 int v;
1498 d = peekChar ();
1499 if (c != '\\')
1500 return c;
1502 switch (d){
1503 case 'a':
1504 v = '\a'; break;
1505 case 'b':
1506 v = '\b'; break;
1507 case 'n':
1508 v = '\n'; break;
1509 case 't':
1510 v = '\t'; break;
1511 case 'v':
1512 v = '\v'; break;
1513 case 'r':
1514 v = '\r'; break;
1515 case '\\':
1516 v = '\\'; break;
1517 case 'f':
1518 v = '\f'; break;
1519 case '0':
1520 v = 0; break;
1521 case '"':
1522 v = '"'; break;
1523 case '\'':
1524 v = '\''; break;
1525 case 'x':
1526 v = getHex (-1, out error);
1527 if (error)
1528 goto default;
1529 return v;
1530 case 'u':
1531 v = getHex (4, out error);
1532 if (error)
1533 goto default;
1534 return v;
1535 case 'U':
1536 v = getHex (8, out error);
1537 if (error)
1538 goto default;
1539 return v;
1540 default:
1541 Report.Error (1009, Location, "Unrecognized escape sequence `\\{0}'", ((char)d).ToString ());
1542 return d;
1544 getChar ();
1545 return v;
1548 int getChar ()
1550 int x;
1551 if (putback_char != -1) {
1552 x = putback_char;
1553 putback_char = -1;
1554 } else
1555 x = reader.Read ();
1556 if (x == '\n') {
1557 line++;
1558 ref_line++;
1559 previous_col = col;
1560 col = 0;
1562 else
1563 col++;
1564 return x;
1567 int peekChar ()
1569 if (putback_char != -1)
1570 return putback_char;
1571 putback_char = reader.Read ();
1572 return putback_char;
1575 int peekChar2 ()
1577 if (putback_char != -1)
1578 return putback_char;
1579 return reader.Peek ();
1582 void putback (int c)
1584 if (putback_char != -1){
1585 Console.WriteLine ("Col: " + col);
1586 Console.WriteLine ("Row: " + line);
1587 Console.WriteLine ("Name: " + ref_name.Name);
1588 Console.WriteLine ("Current [{0}] putting back [{1}] ", putback_char, c);
1589 throw new Exception ("This should not happen putback on putback");
1591 if (c == '\n' || col == 0) {
1592 // It won't happen though.
1593 line--;
1594 ref_line--;
1595 col = previous_col;
1597 else
1598 col--;
1599 putback_char = c;
1602 public bool advance ()
1604 return peekChar () != -1;
1607 public Object Value {
1608 get {
1609 return val;
1613 public Object value ()
1615 return val;
1618 static bool IsCastToken (int token)
1620 switch (token) {
1621 case Token.BANG:
1622 case Token.TILDE:
1623 case Token.IDENTIFIER:
1624 case Token.LITERAL_INTEGER:
1625 case Token.LITERAL_FLOAT:
1626 case Token.LITERAL_DOUBLE:
1627 case Token.LITERAL_DECIMAL:
1628 case Token.LITERAL_CHARACTER:
1629 case Token.LITERAL_STRING:
1630 case Token.BASE:
1631 case Token.CHECKED:
1632 case Token.DELEGATE:
1633 case Token.FALSE:
1634 case Token.FIXED:
1635 case Token.NEW:
1636 case Token.NULL:
1637 case Token.SIZEOF:
1638 case Token.THIS:
1639 case Token.THROW:
1640 case Token.TRUE:
1641 case Token.TYPEOF:
1642 case Token.UNCHECKED:
1643 case Token.UNSAFE:
1644 #if GMCS_SOURCE
1645 case Token.DEFAULT:
1646 #endif
1649 // These can be part of a member access
1651 case Token.INT:
1652 case Token.UINT:
1653 case Token.SHORT:
1654 case Token.USHORT:
1655 case Token.LONG:
1656 case Token.ULONG:
1657 case Token.DOUBLE:
1658 case Token.FLOAT:
1659 case Token.CHAR:
1660 return true;
1662 default:
1663 return false;
1667 public int token ()
1669 current_token = xtoken ();
1671 #if GMCS_SOURCE
1672 if (current_token != Token.DEFAULT)
1673 return current_token;
1675 PushPosition();
1676 int c = xtoken();
1677 if (c == -1)
1678 current_token = Token.ERROR;
1679 else if (c == Token.OPEN_PARENS)
1680 current_token = Token.DEFAULT_OPEN_PARENS;
1681 else if (c == Token.COLON)
1682 current_token = Token.DEFAULT_COLON;
1683 else
1684 PopPosition();
1685 #endif
1686 return current_token;
1689 static StringBuilder static_cmd_arg = new System.Text.StringBuilder ();
1691 void get_cmd_arg (out string cmd, out string arg)
1693 int c;
1695 tokens_seen = false;
1696 arg = "";
1697 static_cmd_arg.Length = 0;
1699 // skip over white space
1700 while ((c = getChar ()) != -1 && (c != '\n') && ((c == '\r') || (c == ' ') || (c == '\t')))
1703 while ((c != -1) && (c != '\n') && (c != ' ') && (c != '\t') && (c != '\r')){
1704 if (is_identifier_part_character ((char) c)){
1705 static_cmd_arg.Append ((char) c);
1706 c = getChar ();
1707 } else {
1708 putback (c);
1709 break;
1713 cmd = static_cmd_arg.ToString ();
1715 if (c == '\n' || c == '\r'){
1716 return;
1719 // skip over white space
1720 while ((c = getChar ()) != -1 && (c != '\n') && ((c == '\r') || (c == ' ') || (c == '\t')))
1723 if (c == '\n'){
1724 return;
1725 } else if (c == '\r'){
1726 return;
1727 } else if (c == -1){
1728 arg = "";
1729 return;
1732 static_cmd_arg.Length = 0;
1733 static_cmd_arg.Append ((char) c);
1735 while ((c = getChar ()) != -1 && (c != '\n') && (c != '\r')){
1736 static_cmd_arg.Append ((char) c);
1739 arg = static_cmd_arg.ToString ();
1743 // Handles the #line directive
1745 bool PreProcessLine (string arg)
1747 if (arg.Length == 0)
1748 return false;
1750 if (arg == "default"){
1751 ref_line = line;
1752 ref_name = file_name;
1753 Location.Push (ref_name);
1754 return true;
1755 } else if (arg == "hidden"){
1757 // We ignore #line hidden
1759 return true;
1762 try {
1763 int pos;
1765 if ((pos = arg.IndexOf (' ')) != -1 && pos != 0){
1766 ref_line = System.Int32.Parse (arg.Substring (0, pos));
1767 pos++;
1769 char [] quotes = { '\"' };
1771 string name = arg.Substring (pos). Trim (quotes);
1772 ref_name = Location.LookupFile (name);
1773 file_name.HasLineDirective = true;
1774 ref_name.HasLineDirective = true;
1775 Location.Push (ref_name);
1776 } else {
1777 ref_line = System.Int32.Parse (arg);
1779 } catch {
1780 return false;
1783 return true;
1787 // Handles #define and #undef
1789 void PreProcessDefinition (bool is_define, string arg, bool caller_is_taking)
1791 if (arg.Length == 0 || arg == "true" || arg == "false"){
1792 Report.Error (1001, Location, "Missing identifer to pre-processor directive");
1793 return;
1796 if (arg.IndexOfAny (simple_whitespaces) != -1){
1797 Error_EndLineExpected ();
1798 return;
1801 if (!is_identifier_start_character (arg [0]))
1802 Report.Error (1001, Location, "Identifier expected: " + arg);
1804 foreach (char c in arg.Substring (1)){
1805 if (!is_identifier_part_character (c)){
1806 Report.Error (1001, Location, "Identifier expected: " + arg);
1807 return;
1811 if (!caller_is_taking)
1812 return;
1814 if (is_define){
1815 if (defines == null)
1816 defines = new Hashtable ();
1817 define (arg);
1818 } else {
1819 if (defines == null)
1820 return;
1821 if (defines.Contains (arg))
1822 defines.Remove (arg);
1826 /// <summary>
1827 /// Handles #pragma directive
1828 /// </summary>
1829 void PreProcessPragma (string arg)
1831 const string warning = "warning";
1832 const string w_disable = "warning disable";
1833 const string w_restore = "warning restore";
1835 if (arg == w_disable) {
1836 Report.RegisterWarningRegion (Location).WarningDisable (line);
1837 return;
1840 if (arg == w_restore) {
1841 Report.RegisterWarningRegion (Location).WarningEnable (line);
1842 return;
1845 if (arg.StartsWith (w_disable)) {
1846 int[] codes = ParseNumbers (arg.Substring (w_disable.Length));
1847 foreach (int code in codes) {
1848 if (code != 0)
1849 Report.RegisterWarningRegion (Location).WarningDisable (Location, code);
1851 return;
1854 if (arg.StartsWith (w_restore)) {
1855 int[] codes = ParseNumbers (arg.Substring (w_restore.Length));
1856 Hashtable w_table = Report.warning_ignore_table;
1857 foreach (int code in codes) {
1858 if (w_table != null && w_table.Contains (code))
1859 Report.Warning (1635, 1, Location, String.Format ("Cannot restore warning `CS{0:0000}' because it was disabled globally", code));
1860 Report.RegisterWarningRegion (Location).WarningEnable (Location, code);
1862 return;
1865 if (arg.StartsWith (warning)) {
1866 Report.Warning (1634, 1, Location, "Expected disable or restore");
1867 return;
1870 Report.Warning (1633, 1, Location, "Unrecognized #pragma directive");
1873 int[] ParseNumbers (string text)
1875 string[] string_array = text.Split (',');
1876 int[] values = new int [string_array.Length];
1877 int index = 0;
1878 foreach (string string_code in string_array) {
1879 try {
1880 values[index++] = int.Parse (string_code, System.Globalization.CultureInfo.InvariantCulture);
1882 catch (FormatException) {
1883 Report.Warning (1692, 1, Location, "Invalid number");
1886 return values;
1889 bool eval_val (string s)
1891 if (s == "true")
1892 return true;
1893 if (s == "false")
1894 return false;
1896 if (defines == null)
1897 return false;
1898 if (defines.Contains (s))
1899 return true;
1901 return false;
1904 bool pp_primary (ref string s)
1906 s = s.Trim ();
1907 int len = s.Length;
1909 if (len > 0){
1910 char c = s [0];
1912 if (c == '('){
1913 s = s.Substring (1);
1914 bool val = pp_expr (ref s, false);
1915 if (s.Length > 0 && s [0] == ')'){
1916 s = s.Substring (1);
1917 return val;
1919 Error_InvalidDirective ();
1920 return false;
1923 if (is_identifier_start_character (c)){
1924 int j = 1;
1926 while (j < len){
1927 c = s [j];
1929 if (is_identifier_part_character (c)){
1930 j++;
1931 continue;
1933 bool v = eval_val (s.Substring (0, j));
1934 s = s.Substring (j);
1935 return v;
1937 bool vv = eval_val (s);
1938 s = "";
1939 return vv;
1942 Error_InvalidDirective ();
1943 return false;
1946 bool pp_unary (ref string s)
1948 s = s.Trim ();
1949 int len = s.Length;
1951 if (len > 0){
1952 if (s [0] == '!'){
1953 if (len > 1 && s [1] == '='){
1954 Error_InvalidDirective ();
1955 return false;
1957 s = s.Substring (1);
1958 return ! pp_primary (ref s);
1959 } else
1960 return pp_primary (ref s);
1961 } else {
1962 Error_InvalidDirective ();
1963 return false;
1967 bool pp_eq (ref string s)
1969 bool va = pp_unary (ref s);
1971 s = s.Trim ();
1972 int len = s.Length;
1973 if (len > 0){
1974 if (s [0] == '='){
1975 if (len > 2 && s [1] == '='){
1976 s = s.Substring (2);
1977 return va == pp_unary (ref s);
1978 } else {
1979 Error_InvalidDirective ();
1980 return false;
1982 } else if (s [0] == '!' && len > 1 && s [1] == '='){
1983 s = s.Substring (2);
1985 return va != pp_unary (ref s);
1990 return va;
1994 bool pp_and (ref string s)
1996 bool va = pp_eq (ref s);
1998 s = s.Trim ();
1999 int len = s.Length;
2000 if (len > 0){
2001 if (s [0] == '&'){
2002 if (len > 2 && s [1] == '&'){
2003 s = s.Substring (2);
2004 return (va & pp_and (ref s));
2005 } else {
2006 Error_InvalidDirective ();
2007 return false;
2011 return va;
2015 // Evaluates an expression for `#if' or `#elif'
2017 bool pp_expr (ref string s, bool isTerm)
2019 bool va = pp_and (ref s);
2020 s = s.Trim ();
2021 int len = s.Length;
2022 if (len > 0){
2023 char c = s [0];
2025 if (c == '|'){
2026 if (len > 2 && s [1] == '|'){
2027 s = s.Substring (2);
2028 return va | pp_expr (ref s, isTerm);
2029 } else {
2030 Error_InvalidDirective ();
2031 return false;
2034 if (isTerm) {
2035 Error_EndLineExpected ();
2036 return false;
2040 return va;
2043 bool eval (string s)
2045 bool v = pp_expr (ref s, true);
2046 s = s.Trim ();
2047 if (s.Length != 0){
2048 return false;
2051 return v;
2054 void Error_NumericConstantTooLong ()
2056 Report.Error (1021, Location, "Numeric constant too long");
2059 void Error_InvalidDirective ()
2061 Report.Error (1517, Location, "Invalid preprocessor directive");
2064 void Error_UnexpectedDirective (string extra)
2066 Report.Error (
2067 1028, Location,
2068 "Unexpected processor directive (" + extra + ")");
2071 void Error_TokensSeen ()
2073 Report.Error (1032, Location,
2074 "Cannot define or undefine preprocessor symbols after first token in file");
2077 void Eror_WrongPreprocessorLocation ()
2079 Report.Error (1040, Location,
2080 "Preprocessor directives must appear as the first non-whitespace character on a line");
2083 void Error_EndLineExpected ()
2085 Report.Error (1025, Location, "Single-line comment or end-of-line expected");
2089 // if true, then the code continues processing the code
2090 // if false, the code stays in a loop until another directive is
2091 // reached.
2092 // When caller_is_taking is false we ignore all directives except the ones
2093 // which can help us to identify where the #if block ends
2094 bool handle_preprocessing_directive (bool caller_is_taking)
2096 string cmd, arg;
2097 bool region_directive = false;
2099 get_cmd_arg (out cmd, out arg);
2101 // Eat any trailing whitespaces and single-line comments
2102 if (arg.IndexOf ("//") != -1)
2103 arg = arg.Substring (0, arg.IndexOf ("//"));
2104 arg = arg.Trim (simple_whitespaces);
2107 // The first group of pre-processing instructions is always processed
2109 switch (cmd){
2110 case "region":
2111 region_directive = true;
2112 arg = "true";
2113 goto case "if";
2115 case "endregion":
2116 if (ifstack == null || ifstack.Count == 0){
2117 Error_UnexpectedDirective ("no #region for this #endregion");
2118 return true;
2120 int pop = (int) ifstack.Pop ();
2122 if ((pop & REGION) == 0)
2123 Report.Error (1027, Location, "Expected `#endif' directive");
2125 return caller_is_taking;
2127 case "if":
2128 if (ifstack == null)
2129 ifstack = new Stack (2);
2131 int flags = region_directive ? REGION : 0;
2132 if (ifstack.Count == 0){
2133 flags |= PARENT_TAKING;
2134 } else {
2135 int state = (int) ifstack.Peek ();
2136 if ((state & TAKING) != 0) {
2137 flags |= PARENT_TAKING;
2141 if (caller_is_taking && eval (arg)) {
2142 ifstack.Push (flags | TAKING);
2143 return true;
2145 ifstack.Push (flags);
2146 return false;
2148 case "endif":
2149 if (ifstack == null || ifstack.Count == 0){
2150 Error_UnexpectedDirective ("no #if for this #endif");
2151 return true;
2152 } else {
2153 pop = (int) ifstack.Pop ();
2155 if ((pop & REGION) != 0)
2156 Report.Error (1038, Location, "#endregion directive expected");
2158 if (arg.Length != 0) {
2159 Error_EndLineExpected ();
2162 if (ifstack.Count == 0)
2163 return true;
2165 int state = (int) ifstack.Peek ();
2166 return (state & TAKING) != 0;
2169 case "elif":
2170 if (ifstack == null || ifstack.Count == 0){
2171 Error_UnexpectedDirective ("no #if for this #elif");
2172 return true;
2173 } else {
2174 int state = (int) ifstack.Pop ();
2176 if ((state & REGION) != 0) {
2177 Report.Error (1038, Location, "#endregion directive expected");
2178 return true;
2181 if ((state & ELSE_SEEN) != 0){
2182 Error_UnexpectedDirective ("#elif not valid after #else");
2183 return true;
2186 if ((state & TAKING) != 0) {
2187 ifstack.Push (0);
2188 return false;
2191 if (eval (arg) && ((state & PARENT_TAKING) != 0)){
2192 ifstack.Push (state | TAKING);
2193 return true;
2196 ifstack.Push (state);
2197 return false;
2200 case "else":
2201 if (ifstack == null || ifstack.Count == 0){
2202 Error_UnexpectedDirective ("no #if for this #else");
2203 return true;
2204 } else {
2205 int state = (int) ifstack.Peek ();
2207 if ((state & REGION) != 0) {
2208 Report.Error (1038, Location, "#endregion directive expected");
2209 return true;
2212 if ((state & ELSE_SEEN) != 0){
2213 Error_UnexpectedDirective ("#else within #else");
2214 return true;
2217 ifstack.Pop ();
2219 if (arg.Length != 0) {
2220 Error_EndLineExpected ();
2221 return true;
2224 bool ret = false;
2225 if ((state & PARENT_TAKING) != 0) {
2226 ret = (state & TAKING) == 0;
2228 if (ret)
2229 state |= TAKING;
2230 else
2231 state &= ~TAKING;
2234 ifstack.Push (state | ELSE_SEEN);
2236 return ret;
2238 case "define":
2239 if (any_token_seen){
2240 Error_TokensSeen ();
2241 return caller_is_taking;
2243 PreProcessDefinition (true, arg, caller_is_taking);
2244 return caller_is_taking;
2246 case "undef":
2247 if (any_token_seen){
2248 Error_TokensSeen ();
2249 return caller_is_taking;
2251 PreProcessDefinition (false, arg, caller_is_taking);
2252 return caller_is_taking;
2256 // These are only processed if we are in a `taking' block
2258 if (!caller_is_taking)
2259 return false;
2261 switch (cmd){
2262 case "error":
2263 Report.Error (1029, Location, "#error: '" + arg + "'");
2264 return true;
2266 case "warning":
2267 Report.Warning (1030, 1, Location, "#warning: `{0}'", arg);
2268 return true;
2270 case "pragma":
2271 if (RootContext.Version == LanguageVersion.ISO_1) {
2272 Report.FeatureIsNotISO1 (Location, "#pragma");
2273 return true;
2276 PreProcessPragma (arg);
2277 return true;
2279 case "line":
2280 if (!PreProcessLine (arg))
2281 Report.Error (
2282 1576, Location,
2283 "The line number specified for #line directive is missing or invalid");
2284 return caller_is_taking;
2287 Report.Error (1024, Location, "Wrong preprocessor directive");
2288 return true;
2292 private int consume_string (bool quoted)
2294 int c;
2295 string_builder.Length = 0;
2297 while ((c = getChar ()) != -1){
2298 if (c == '"'){
2299 if (quoted && peekChar () == '"'){
2300 string_builder.Append ((char) c);
2301 getChar ();
2302 continue;
2303 } else {
2304 val = string_builder.ToString ();
2305 return Token.LITERAL_STRING;
2309 if (c == '\n'){
2310 if (!quoted)
2311 Report.Error (1010, Location, "Newline in constant");
2314 if (!quoted){
2315 c = escape (c);
2316 if (c == -1)
2317 return Token.ERROR;
2319 string_builder.Append ((char) c);
2322 Report.Error (1039, Location, "Unterminated string literal");
2323 return Token.EOF;
2326 private int consume_identifier (int s)
2328 int res = consume_identifier (s, false);
2330 if (doc_state == XmlCommentState.Allowed)
2331 doc_state = XmlCommentState.NotAllowed;
2332 switch (res) {
2333 case Token.USING:
2334 case Token.NAMESPACE:
2335 check_incorrect_doc_comment ();
2336 break;
2339 if (res == Token.PARTIAL) {
2340 // Save current position and parse next token.
2341 PushPosition ();
2343 int next_token = token ();
2344 bool ok = (next_token == Token.CLASS) ||
2345 (next_token == Token.STRUCT) ||
2346 (next_token == Token.INTERFACE);
2348 PopPosition ();
2350 if (ok)
2351 return res;
2353 if (next_token < Token.LAST_KEYWORD)
2354 Report.Error (267, Location, "The `partial' modifier can be used only immediately before keyword `class', `struct', or `interface'");
2356 val = new LocatedToken (Location, "partial");
2357 return Token.IDENTIFIER;
2360 return res;
2363 private int consume_identifier (int s, bool quoted)
2365 int pos = 1;
2366 int c = -1;
2368 id_builder [0] = (char) s;
2370 current_location = new Location (ref_line, Col);
2372 while ((c = getChar ()) != -1) {
2373 loop:
2374 if (is_identifier_part_character ((char) c)){
2375 if (pos == max_id_size){
2376 Report.Error (645, Location, "Identifier too long (limit is 512 chars)");
2377 return Token.ERROR;
2380 id_builder [pos++] = (char) c;
2381 // putback_char = -1;
2382 } else if (c == '\\') {
2383 c = escape (c);
2384 goto loop;
2385 } else {
2386 // putback_char = c;
2387 putback (c);
2388 break;
2393 // Optimization: avoids doing the keyword lookup
2394 // on uppercase letters and _
2396 if (!quoted && (s >= 'a' || s == '_')){
2397 int keyword = GetKeyword (id_builder, pos);
2398 if (keyword != -1) {
2399 val = Location;
2400 return keyword;
2405 // Keep identifiers in an array of hashtables to avoid needless
2406 // allocations
2409 if (identifiers [pos] != null) {
2410 val = identifiers [pos][id_builder];
2411 if (val != null) {
2412 val = new LocatedToken (Location, (string) val);
2413 if (quoted)
2414 escapedIdentifiers.Add (val);
2415 return Token.IDENTIFIER;
2418 else
2419 identifiers [pos] = new CharArrayHashtable (pos);
2421 val = new String (id_builder, 0, pos);
2422 if (RootContext.Version == LanguageVersion.ISO_1) {
2423 for (int i = 1; i < id_builder.Length; i += 3) {
2424 if (id_builder [i] == '_' && (id_builder [i - 1] == '_' || id_builder [i + 1] == '_')) {
2425 Report.Error (1638, Location,
2426 "`{0}': Any identifier with double underscores cannot be used when ISO language version mode is specified", val.ToString ());
2427 break;
2432 char [] chars = new char [pos];
2433 Array.Copy (id_builder, chars, pos);
2435 identifiers [pos] [chars] = val;
2437 val = new LocatedToken (Location, (string) val);
2438 if (quoted)
2439 escapedIdentifiers.Add (val);
2440 return Token.IDENTIFIER;
2443 public int xtoken ()
2445 int t;
2446 bool doread = false;
2447 int c;
2449 // Whether we have seen comments on the current line
2450 bool comments_seen = false;
2451 val = null;
2452 for (;(c = getChar ()) != -1;) {
2453 if (c == '\t'){
2454 col = ((col + 8) / 8) * 8;
2455 continue;
2458 if (c == ' ' || c == '\f' || c == '\v' || c == 0xa0 || c == 0)
2459 continue;
2461 if (c == '\r') {
2462 if (peekChar () == '\n')
2463 getChar ();
2465 any_token_seen |= tokens_seen;
2466 tokens_seen = false;
2467 comments_seen = false;
2468 continue;
2471 // Handle double-slash comments.
2472 if (c == '/'){
2473 int d = peekChar ();
2475 if (d == '/'){
2476 getChar ();
2477 if (RootContext.Documentation != null && peekChar () == '/') {
2478 getChar ();
2479 // Don't allow ////.
2480 if ((d = peekChar ()) != '/') {
2481 update_comment_location ();
2482 if (doc_state == XmlCommentState.Allowed)
2483 handle_one_line_xml_comment ();
2484 else if (doc_state == XmlCommentState.NotAllowed)
2485 warn_incorrect_doc_comment ();
2488 while ((d = getChar ()) != -1 && (d != '\n') && d != '\r')
2489 if (d == '\n'){
2491 any_token_seen |= tokens_seen;
2492 tokens_seen = false;
2493 comments_seen = false;
2494 continue;
2495 } else if (d == '*'){
2496 getChar ();
2497 bool docAppend = false;
2498 if (RootContext.Documentation != null && peekChar () == '*') {
2499 getChar ();
2500 update_comment_location ();
2501 // But when it is /**/, just do nothing.
2502 if (peekChar () == '/') {
2503 getChar ();
2504 continue;
2506 if (doc_state == XmlCommentState.Allowed)
2507 docAppend = true;
2508 else if (doc_state == XmlCommentState.NotAllowed)
2509 warn_incorrect_doc_comment ();
2512 int current_comment_start = 0;
2513 if (docAppend) {
2514 current_comment_start = xml_comment_buffer.Length;
2515 xml_comment_buffer.Append (Environment.NewLine);
2518 Location start_location = Location;
2520 while ((d = getChar ()) != -1){
2521 if (d == '*' && peekChar () == '/'){
2522 getChar ();
2523 comments_seen = true;
2524 break;
2526 if (docAppend)
2527 xml_comment_buffer.Append ((char) d);
2529 if (d == '\n'){
2530 any_token_seen |= tokens_seen;
2531 tokens_seen = false;
2533 // Reset 'comments_seen' just to be consistent.
2534 // It doesn't matter either way, here.
2536 comments_seen = false;
2539 if (!comments_seen)
2540 Report.Error (1035, start_location, "End-of-file found, '*/' expected");
2542 if (docAppend)
2543 update_formatted_doc_comment (current_comment_start);
2544 continue;
2546 goto is_punct_label;
2550 if (c == '\\' || is_identifier_start_character ((char)c)){
2551 tokens_seen = true;
2552 return consume_identifier (c);
2555 is_punct_label:
2556 current_location = new Location (ref_line, Col);
2557 if ((t = is_punct ((char)c, ref doread)) != Token.ERROR){
2558 tokens_seen = true;
2559 if (doread){
2560 getChar ();
2562 return t;
2565 // white space
2566 if (c == '\n'){
2567 any_token_seen |= tokens_seen;
2568 tokens_seen = false;
2569 comments_seen = false;
2570 continue;
2573 if (c >= '0' && c <= '9'){
2574 tokens_seen = true;
2575 return is_number (c);
2578 if (c == '.'){
2579 tokens_seen = true;
2580 int peek = peekChar ();
2581 if (peek >= '0' && peek <= '9')
2582 return is_number (c);
2583 return Token.DOT;
2586 if (c == '#') {
2587 if (tokens_seen || comments_seen) {
2588 Eror_WrongPreprocessorLocation ();
2589 return Token.ERROR;
2592 if (handle_preprocessing_directive (true))
2593 continue;
2595 bool directive_expected = false;
2596 while ((c = getChar ()) != -1) {
2597 if (col == 1) {
2598 directive_expected = true;
2599 } else if (!directive_expected) {
2600 // TODO: Implement comment support for disabled code and uncomment this code
2601 // if (c == '#') {
2602 // Eror_WrongPreprocessorLocation ();
2603 // return Token.ERROR;
2604 // }
2605 continue;
2608 if (c == ' ' || c == '\t' || c == '\r' || c == '\n' || c == '\f' || c == '\v' )
2609 continue;
2611 if (c == '#') {
2612 if (handle_preprocessing_directive (false))
2613 break;
2615 directive_expected = false;
2618 if (c != -1) {
2619 tokens_seen = false;
2620 continue;
2623 return Token.EOF;
2626 if (c == '"')
2627 return consume_string (false);
2629 if (c == '\''){
2630 c = getChar ();
2631 tokens_seen = true;
2632 if (c == '\''){
2633 error_details = "Empty character literal";
2634 Report.Error (1011, Location, error_details);
2635 return Token.ERROR;
2637 if (c == '\r' || c == '\n') {
2638 Report.Error (1010, Location, "Newline in constant");
2639 return Token.ERROR;
2641 c = escape (c);
2642 if (c == -1)
2643 return Token.ERROR;
2644 val = new System.Char ();
2645 val = (char) c;
2646 c = getChar ();
2648 if (c != '\''){
2649 error_details = "Too many characters in character literal";
2650 Report.Error (1012, Location, error_details);
2652 // Try to recover, read until newline or next "'"
2653 while ((c = getChar ()) != -1){
2654 if (c == '\n'){
2655 break;
2657 else if (c == '\'')
2658 break;
2660 return Token.ERROR;
2662 return Token.LITERAL_CHARACTER;
2665 if (c == '@') {
2666 c = getChar ();
2667 if (c == '"') {
2668 tokens_seen = true;
2669 return consume_string (true);
2670 } else if (is_identifier_start_character ((char) c)){
2671 return consume_identifier (c, true);
2672 } else {
2673 Report.Error (1646, Location, "Keyword, identifier, or string expected after verbatim specifier: @");
2677 error_details = ((char)c).ToString ();
2679 return Token.ERROR;
2682 return Token.EOF;
2686 // Handles one line xml comment
2688 private void handle_one_line_xml_comment ()
2690 int c;
2691 while ((c = peekChar ()) == ' ')
2692 getChar (); // skip heading whitespaces.
2693 while ((c = peekChar ()) != -1 && c != '\n' && c != '\r') {
2694 xml_comment_buffer.Append ((char) getChar ());
2696 if (c == '\r' || c == '\n')
2697 xml_comment_buffer.Append (Environment.NewLine);
2701 // Remove heading "*" in Javadoc-like xml documentation.
2703 private void update_formatted_doc_comment (int current_comment_start)
2705 int length = xml_comment_buffer.Length - current_comment_start;
2706 string [] lines = xml_comment_buffer.ToString (
2707 current_comment_start,
2708 length).Replace ("\r", "").Split ('\n');
2710 // The first line starts with /**, thus it is not target
2711 // for the format check.
2712 for (int i = 1; i < lines.Length; i++) {
2713 string s = lines [i];
2714 int idx = s.IndexOf ('*');
2715 string head = null;
2716 if (idx < 0) {
2717 if (i < lines.Length - 1)
2718 return;
2719 head = s;
2720 } else
2721 head = s.Substring (0, idx);
2722 foreach (char c in head)
2723 if (c != ' ')
2724 return;
2725 lines [i] = s.Substring (idx + 1);
2727 xml_comment_buffer.Remove (current_comment_start, length);
2728 xml_comment_buffer.Insert (current_comment_start, String.Join (Environment.NewLine, lines));
2732 // Updates current comment location.
2734 private void update_comment_location ()
2736 if (current_comment_location.IsNull) {
2737 // "-2" is for heading "//" or "/*"
2738 current_comment_location =
2739 new Location (ref_line, col - 2);
2744 // Checks if there was incorrect doc comments and raise
2745 // warnings.
2747 public void check_incorrect_doc_comment ()
2749 if (xml_comment_buffer.Length > 0)
2750 warn_incorrect_doc_comment ();
2754 // Raises a warning when tokenizer found incorrect doccomment
2755 // markup.
2757 private void warn_incorrect_doc_comment ()
2759 if (doc_state != XmlCommentState.Error) {
2760 doc_state = XmlCommentState.Error;
2761 // in csc, it is 'XML comment is not placed on
2762 // a valid language element'. But that does not
2763 // make sense.
2764 Report.Warning (1587, 2, Location, "XML comment is not placed on a valid language element");
2769 // Consumes the saved xml comment lines (if any)
2770 // as for current target member or type.
2772 public string consume_doc_comment ()
2774 if (xml_comment_buffer.Length > 0) {
2775 string ret = xml_comment_buffer.ToString ();
2776 reset_doc_comment ();
2777 return ret;
2779 return null;
2782 void reset_doc_comment ()
2784 xml_comment_buffer.Length = 0;
2785 current_comment_location = Location.Null;
2788 public void cleanup ()
2790 if (ifstack != null && ifstack.Count >= 1) {
2791 current_location = new Location (ref_line, Col);
2792 int state = (int) ifstack.Pop ();
2793 if ((state & REGION) != 0)
2794 Report.Error (1038, Location, "#endregion directive expected");
2795 else
2796 Report.Error (1027, Location, "Expected `#endif' directive");
2802 // Indicates whether it accepts XML documentation or not.
2804 public enum XmlCommentState {
2805 // comment is allowed in this state.
2806 Allowed,
2807 // comment is not allowed in this state.
2808 NotAllowed,
2809 // once comments appeared when it is NotAllowed, then the
2810 // state is changed to it, until the state is changed to
2811 // .Allowed.
2812 Error