2 // cs-tokenizer.cs: The Tokenizer for the C# compiler
3 // This also implements the preprocessor
5 // Author: Miguel de Icaza (miguel@gnu.org)
6 // Marek Safar (marek.safar@seznam.cz)
8 // Dual licensed under the terms of the MIT X11 or GNU GPL
10 // Copyright 2001, 2002 Ximian, Inc (http://www.ximian.com)
11 // Copyright 2004-2008 Novell, Inc
17 using System.Collections.Generic;
18 using System.Globalization;
19 using System.Diagnostics;
24 /// Tokenizer for C# source code.
27 public class Tokenizer : yyParser.yyInput
31 public readonly T Token;
32 public KeywordEntry<T> Next;
33 public readonly char[] Value;
35 public KeywordEntry (string value, T token)
37 this.Value = value.ToCharArray ();
42 sealed class IdentifiersComparer : IEqualityComparer<char[]>
46 public IdentifiersComparer (int length)
51 public bool Equals (char[] x, char[] y)
53 for (int i = 0; i < length; ++i)
60 public int GetHashCode (char[] obj)
63 for (int i = 0; i < length; ++i)
64 h = (h << 5) - h + obj [i];
71 // This class has to be used in the parser only, it reuses token
72 // details after each parse
74 public class LocatedToken
79 static LocatedToken[] buffer;
82 private LocatedToken ()
86 public static LocatedToken Create (int row, int column)
88 return Create (null, row, column);
91 public static LocatedToken Create (string value, int row, int column)
94 // TODO: I am not very happy about the logic but it's the best
95 // what I could come up with for now.
96 // Ideally we should be using just tiny buffer (256 elements) which
97 // is enough to hold all details for currect stack and recycle elements
98 // poped from the stack but there is a trick needed to recycle
102 if (pos >= buffer.Length) {
103 entry = new LocatedToken ();
105 entry = buffer [pos];
107 entry = new LocatedToken ();
108 buffer [pos] = entry;
115 entry.column = column;
120 // Used for token not required by expression evaluator
122 [Conditional ("FULL_AST")]
123 public static void CreateOptional (int row, int col, ref object token)
125 token = Create (row, col);
128 public static void Initialize ()
131 buffer = new LocatedToken [10000];
135 public Location Location {
136 get { return new Location (row, column); }
139 public string Value {
140 get { return value; }
144 enum PreprocessorDirective
150 If = 3 | RequiresArgument,
152 Elif = 5 | RequiresArgument,
154 Define = 7 | RequiresArgument,
155 Undef = 8 | RequiresArgument,
158 Pragma = 11 | CustomArgumentsParsing,
161 CustomArgumentsParsing = 1 << 10,
162 RequiresArgument = 1 << 11
165 SeekableStreamReader reader;
167 CompilationSourceFile file_name;
168 CompilerContext context;
176 bool handle_get_set = false;
177 bool handle_remove_add = false;
178 bool handle_where = false;
179 bool handle_typeof = false;
180 bool lambda_arguments_parsing;
181 List<Location> escaped_identifiers;
182 int parsing_generic_less_than;
183 readonly bool doc_processing;
186 // Used mainly for parser optimizations. Some expressions for instance
187 // can appear only in block (including initializer, base initializer)
190 public int parsing_block;
191 internal bool query_parsing;
194 // When parsing type only, useful for ambiguous nullable types
196 public int parsing_type;
199 // Set when parsing generic declaration (type or method header)
201 public bool parsing_generic_declaration;
202 public bool parsing_generic_declaration_doc;
205 // The value indicates that we have not reach any declaration or
208 public int parsing_declaration;
210 public bool parsing_attribute_section;
213 // The special characters to inject on streams to run the unit parser
214 // in the special expression mode. Using private characters from
215 // Plane Sixteen (U+100000 to U+10FFFD)
217 // This character is only tested just before the tokenizer is about to report
218 // an error; So on the regular operation mode, this addition will have no
219 // impact on the tokenizer's performance.
222 public const int EvalStatementParserCharacter = 0x100000;
223 public const int EvalCompilationUnitParserCharacter = 0x100001;
224 public const int EvalUsingDeclarationsParserCharacter = 0x100002;
225 public const int DocumentationXref = 0x100003;
228 // XML documentation buffer. The save point is used to divide
229 // comments on types and comments on members.
231 StringBuilder xml_comment_buffer;
234 // See comment on XmlCommentState enumeration.
236 XmlCommentState xml_doc_state = XmlCommentState.Allowed;
239 // Whether tokens have been seen on this line
241 bool tokens_seen = false;
244 // Set to true once the GENERATE_COMPLETION token has bee
245 // returned. This helps produce one GENERATE_COMPLETION,
246 // as many COMPLETE_COMPLETION as necessary to complete the
247 // AST tree and one final EOF.
252 // Whether a token has been seen on the file
253 // This is needed because `define' is not allowed to be used
254 // after a token has been seen.
261 static readonly KeywordEntry<int>[][] keywords;
262 static readonly KeywordEntry<PreprocessorDirective>[][] keywords_preprocessor;
263 static readonly Dictionary<string, object> keyword_strings; // TODO: HashSet
264 static readonly NumberStyles styles;
265 static readonly NumberFormatInfo csharp_format_info;
268 static readonly char[] pragma_warning = "warning".ToCharArray ();
269 static readonly char[] pragma_warning_disable = "disable".ToCharArray ();
270 static readonly char[] pragma_warning_restore = "restore".ToCharArray ();
271 static readonly char[] pragma_checksum = "checksum".ToCharArray ();
273 static readonly char[] simple_whitespaces = new char[] { ' ', '\t' };
275 public bool PropertyParsing {
276 get { return handle_get_set; }
277 set { handle_get_set = value; }
280 public bool EventParsing {
281 get { return handle_remove_add; }
282 set { handle_remove_add = value; }
285 public bool ConstraintsParsing {
286 get { return handle_where; }
287 set { handle_where = value; }
290 public bool TypeOfParsing {
291 get { return handle_typeof; }
292 set { handle_typeof = value; }
296 get { return tab_size; }
297 set { tab_size = value; }
300 public XmlCommentState doc_state {
301 get { return xml_doc_state; }
303 if (value == XmlCommentState.Allowed) {
304 check_incorrect_doc_comment ();
305 reset_doc_comment ();
307 xml_doc_state = value;
312 // This is used to trigger completion generation on the parser
313 public bool CompleteOnEOF;
315 void AddEscapedIdentifier (Location loc)
317 if (escaped_identifiers == null)
318 escaped_identifiers = new List<Location> ();
320 escaped_identifiers.Add (loc);
323 public bool IsEscapedIdentifier (MemberName name)
325 return escaped_identifiers != null && escaped_identifiers.Contains (name.Location);
329 // Values for the associated token returned
331 internal int putback_char; // Used by repl only
337 const int TAKING = 1;
338 const int ELSE_SEEN = 4;
339 const int PARENT_TAKING = 8;
340 const int REGION = 16;
343 // pre-processor if stack state:
347 static System.Text.StringBuilder string_builder;
349 const int max_id_size = 512;
350 static readonly char [] id_builder = new char [max_id_size];
352 public static Dictionary<char[], string>[] identifiers = new Dictionary<char[], string>[max_id_size + 1];
354 const int max_number_size = 512;
355 static char [] number_builder = new char [max_number_size];
356 static int number_pos;
358 static char[] value_builder = new char[256];
367 // This is used when the tokenizer needs to save
368 // the current position as it needs to do some parsing
369 // on its own to deamiguate a token in behalf of the
372 Stack<Position> position_stack = new Stack<Position> (2);
380 public int putback_char;
381 public int previous_col;
382 public Stack<int> ifstack;
383 public int parsing_generic_less_than;
384 public int current_token;
387 public Position (Tokenizer t)
389 position = t.reader.Position;
391 ref_line = t.ref_line;
394 putback_char = t.putback_char;
395 previous_col = t.previous_col;
396 if (t.ifstack != null && t.ifstack.Count != 0) {
397 // There is no simple way to clone Stack<T> all
398 // methods reverse the order
399 var clone = t.ifstack.ToArray ();
400 Array.Reverse (clone);
401 ifstack = new Stack<int> (clone);
403 parsing_generic_less_than = t.parsing_generic_less_than;
404 current_token = t.current_token;
409 public Tokenizer (SeekableStreamReader input, CompilationSourceFile file, CompilerContext ctx)
411 this.ref_name = file;
412 this.file_name = file;
418 xml_comment_buffer = new StringBuilder ();
419 doc_processing = ctx.Settings.DocumentationFile != null;
421 if (Environment.OSVersion.Platform == PlatformID.Win32NT)
427 // FIXME: This could be `Location.Push' but we have to
428 // find out why the MS compiler allows this
430 Mono.CSharp.Location.Push (file, file);
433 public void PushPosition ()
435 position_stack.Push (new Position (this));
438 public void PopPosition ()
440 Position p = position_stack.Pop ();
442 reader.Position = p.position;
443 ref_line = p.ref_line;
447 putback_char = p.putback_char;
448 previous_col = p.previous_col;
450 parsing_generic_less_than = p.parsing_generic_less_than;
451 current_token = p.current_token;
455 // Do not reset the position, ignore it.
456 public void DiscardPosition ()
458 position_stack.Pop ();
461 static void AddKeyword (string kw, int token)
463 keyword_strings.Add (kw, null);
465 AddKeyword (keywords, kw, token);
468 static void AddPreprocessorKeyword (string kw, PreprocessorDirective directive)
470 AddKeyword (keywords_preprocessor, kw, directive);
473 static void AddKeyword<T> (KeywordEntry<T>[][] keywords, string kw, T token)
475 int length = kw.Length;
476 if (keywords[length] == null) {
477 keywords[length] = new KeywordEntry<T>['z' - '_' + 1];
480 int char_index = kw[0] - '_';
481 var kwe = keywords[length][char_index];
483 keywords[length][char_index] = new KeywordEntry<T> (kw, token);
487 while (kwe.Next != null) {
491 kwe.Next = new KeywordEntry<T> (kw, token);
499 keyword_strings = new Dictionary<string, object> ();
501 // 11 is the length of the longest keyword for now
502 keywords = new KeywordEntry<int>[11][];
504 AddKeyword ("__arglist", Token.ARGLIST);
505 AddKeyword ("__makeref", Token.MAKEREF);
506 AddKeyword ("__reftype", Token.REFTYPE);
507 AddKeyword ("__refvalue", Token.REFVALUE);
508 AddKeyword ("abstract", Token.ABSTRACT);
509 AddKeyword ("as", Token.AS);
510 AddKeyword ("add", Token.ADD);
511 AddKeyword ("base", Token.BASE);
512 AddKeyword ("bool", Token.BOOL);
513 AddKeyword ("break", Token.BREAK);
514 AddKeyword ("byte", Token.BYTE);
515 AddKeyword ("case", Token.CASE);
516 AddKeyword ("catch", Token.CATCH);
517 AddKeyword ("char", Token.CHAR);
518 AddKeyword ("checked", Token.CHECKED);
519 AddKeyword ("class", Token.CLASS);
520 AddKeyword ("const", Token.CONST);
521 AddKeyword ("continue", Token.CONTINUE);
522 AddKeyword ("decimal", Token.DECIMAL);
523 AddKeyword ("default", Token.DEFAULT);
524 AddKeyword ("delegate", Token.DELEGATE);
525 AddKeyword ("do", Token.DO);
526 AddKeyword ("double", Token.DOUBLE);
527 AddKeyword ("else", Token.ELSE);
528 AddKeyword ("enum", Token.ENUM);
529 AddKeyword ("event", Token.EVENT);
530 AddKeyword ("explicit", Token.EXPLICIT);
531 AddKeyword ("extern", Token.EXTERN);
532 AddKeyword ("false", Token.FALSE);
533 AddKeyword ("finally", Token.FINALLY);
534 AddKeyword ("fixed", Token.FIXED);
535 AddKeyword ("float", Token.FLOAT);
536 AddKeyword ("for", Token.FOR);
537 AddKeyword ("foreach", Token.FOREACH);
538 AddKeyword ("goto", Token.GOTO);
539 AddKeyword ("get", Token.GET);
540 AddKeyword ("if", Token.IF);
541 AddKeyword ("implicit", Token.IMPLICIT);
542 AddKeyword ("in", Token.IN);
543 AddKeyword ("int", Token.INT);
544 AddKeyword ("interface", Token.INTERFACE);
545 AddKeyword ("internal", Token.INTERNAL);
546 AddKeyword ("is", Token.IS);
547 AddKeyword ("lock", Token.LOCK);
548 AddKeyword ("long", Token.LONG);
549 AddKeyword ("namespace", Token.NAMESPACE);
550 AddKeyword ("new", Token.NEW);
551 AddKeyword ("null", Token.NULL);
552 AddKeyword ("object", Token.OBJECT);
553 AddKeyword ("operator", Token.OPERATOR);
554 AddKeyword ("out", Token.OUT);
555 AddKeyword ("override", Token.OVERRIDE);
556 AddKeyword ("params", Token.PARAMS);
557 AddKeyword ("private", Token.PRIVATE);
558 AddKeyword ("protected", Token.PROTECTED);
559 AddKeyword ("public", Token.PUBLIC);
560 AddKeyword ("readonly", Token.READONLY);
561 AddKeyword ("ref", Token.REF);
562 AddKeyword ("remove", Token.REMOVE);
563 AddKeyword ("return", Token.RETURN);
564 AddKeyword ("sbyte", Token.SBYTE);
565 AddKeyword ("sealed", Token.SEALED);
566 AddKeyword ("set", Token.SET);
567 AddKeyword ("short", Token.SHORT);
568 AddKeyword ("sizeof", Token.SIZEOF);
569 AddKeyword ("stackalloc", Token.STACKALLOC);
570 AddKeyword ("static", Token.STATIC);
571 AddKeyword ("string", Token.STRING);
572 AddKeyword ("struct", Token.STRUCT);
573 AddKeyword ("switch", Token.SWITCH);
574 AddKeyword ("this", Token.THIS);
575 AddKeyword ("throw", Token.THROW);
576 AddKeyword ("true", Token.TRUE);
577 AddKeyword ("try", Token.TRY);
578 AddKeyword ("typeof", Token.TYPEOF);
579 AddKeyword ("uint", Token.UINT);
580 AddKeyword ("ulong", Token.ULONG);
581 AddKeyword ("unchecked", Token.UNCHECKED);
582 AddKeyword ("unsafe", Token.UNSAFE);
583 AddKeyword ("ushort", Token.USHORT);
584 AddKeyword ("using", Token.USING);
585 AddKeyword ("virtual", Token.VIRTUAL);
586 AddKeyword ("void", Token.VOID);
587 AddKeyword ("volatile", Token.VOLATILE);
588 AddKeyword ("while", Token.WHILE);
589 AddKeyword ("partial", Token.PARTIAL);
590 AddKeyword ("where", Token.WHERE);
591 AddKeyword ("async", Token.ASYNC);
594 AddKeyword ("from", Token.FROM);
595 AddKeyword ("join", Token.JOIN);
596 AddKeyword ("on", Token.ON);
597 AddKeyword ("equals", Token.EQUALS);
598 AddKeyword ("select", Token.SELECT);
599 AddKeyword ("group", Token.GROUP);
600 AddKeyword ("by", Token.BY);
601 AddKeyword ("let", Token.LET);
602 AddKeyword ("orderby", Token.ORDERBY);
603 AddKeyword ("ascending", Token.ASCENDING);
604 AddKeyword ("descending", Token.DESCENDING);
605 AddKeyword ("into", Token.INTO);
607 keywords_preprocessor = new KeywordEntry<PreprocessorDirective>[10][];
609 AddPreprocessorKeyword ("region", PreprocessorDirective.Region);
610 AddPreprocessorKeyword ("endregion", PreprocessorDirective.Endregion);
611 AddPreprocessorKeyword ("if", PreprocessorDirective.If);
612 AddPreprocessorKeyword ("endif", PreprocessorDirective.Endif);
613 AddPreprocessorKeyword ("elif", PreprocessorDirective.Elif);
614 AddPreprocessorKeyword ("else", PreprocessorDirective.Else);
615 AddPreprocessorKeyword ("define", PreprocessorDirective.Define);
616 AddPreprocessorKeyword ("undef", PreprocessorDirective.Undef);
617 AddPreprocessorKeyword ("error", PreprocessorDirective.Error);
618 AddPreprocessorKeyword ("warning", PreprocessorDirective.Warning);
619 AddPreprocessorKeyword ("pragma", PreprocessorDirective.Pragma);
620 AddPreprocessorKeyword ("line", PreprocessorDirective.Line);
622 csharp_format_info = NumberFormatInfo.InvariantInfo;
623 styles = NumberStyles.Float;
625 string_builder = new System.Text.StringBuilder ();
628 int GetKeyword (char[] id, int id_len)
631 // Keywords are stored in an array of arrays grouped by their
632 // length and then by the first character
634 if (id_len >= keywords.Length || keywords [id_len] == null)
637 int first_index = id [0] - '_';
638 if (first_index > 'z' - '_')
641 var kwe = keywords [id_len] [first_index];
648 for (int i = 1; i < id_len; ++i) {
649 if (id [i] != kwe.Value [i]) {
655 } while (res == 0 && kwe != null);
669 if (!handle_remove_add)
673 if (parsing_declaration == 0)
674 res = Token.EXTERN_ALIAS;
677 if (peek_token () == Token.COLON) {
679 res = Token.DEFAULT_COLON;
683 if (!handle_where && !query_parsing)
688 // A query expression is any expression that starts with `from identifier'
689 // followed by any token except ; , =
691 if (!query_parsing) {
692 if (lambda_arguments_parsing) {
698 // HACK: to disable generics micro-parser, because PushPosition does not
699 // store identifiers array
700 parsing_generic_less_than = 1;
702 case Token.IDENTIFIER:
714 next_token = xtoken ();
715 if (next_token == Token.SEMICOLON || next_token == Token.COMMA || next_token == Token.EQUALS)
718 res = Token.FROM_FIRST;
719 query_parsing = true;
720 if (context.Settings.Version <= LanguageVersion.ISO_2)
721 Report.FeatureIsNotAvailable (context, Location, "query expressions");
724 Expression.Error_VoidInvalidInTheContext (Location, Report);
728 // HACK: A token is not a keyword so we need to restore identifiers buffer
729 // which has been overwritten before we grabbed the identifier
730 id_builder [0] = 'f'; id_builder [1] = 'r'; id_builder [2] = 'o'; id_builder [3] = 'm';
744 case Token.ASCENDING:
745 case Token.DESCENDING:
752 case Token.NAMESPACE:
753 // TODO: some explanation needed
754 check_incorrect_doc_comment ();
758 if (parsing_block > 0) {
763 // Save current position and parse next token.
766 next_token = token ();
767 bool ok = (next_token == Token.CLASS) ||
768 (next_token == Token.STRUCT) ||
769 (next_token == Token.INTERFACE) ||
770 (next_token == Token.VOID);
775 if (next_token == Token.VOID) {
776 if (context.Settings.Version <= LanguageVersion.ISO_2)
777 Report.FeatureIsNotAvailable (context, Location, "partial methods");
778 } else if (context.Settings.Version == LanguageVersion.ISO_1)
779 Report.FeatureIsNotAvailable (context, Location, "partial types");
784 if (next_token < Token.LAST_KEYWORD) {
785 Report.Error (267, Location,
786 "The `partial' modifier can be used only immediately before `class', `struct', `interface', or `void' keyword");
794 if (parsing_block > 0 || context.Settings.Version != LanguageVersion.Future) {
804 static PreprocessorDirective GetPreprocessorDirective (char[] id, int id_len)
807 // Keywords are stored in an array of arrays grouped by their
808 // length and then by the first character
810 if (id_len >= keywords_preprocessor.Length || keywords_preprocessor[id_len] == null)
811 return PreprocessorDirective.Invalid;
813 int first_index = id[0] - '_';
814 if (first_index > 'z' - '_')
815 return PreprocessorDirective.Invalid;
817 var kwe = keywords_preprocessor[id_len][first_index];
819 return PreprocessorDirective.Invalid;
821 PreprocessorDirective res = PreprocessorDirective.Invalid;
824 for (int i = 1; i < id_len; ++i) {
825 if (id[i] != kwe.Value[i]) {
831 } while (res == PreprocessorDirective.Invalid && kwe != null);
836 public Location Location {
838 return new Location (ref_line, hidden ? -1 : col);
842 static bool is_identifier_start_character (int c)
844 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_' || Char.IsLetter ((char)c);
847 static bool is_identifier_part_character (char c)
849 if (c >= 'a' && c <= 'z')
852 if (c >= 'A' && c <= 'Z')
855 if (c == '_' || (c >= '0' && c <= '9'))
861 return Char.IsLetter (c) || Char.GetUnicodeCategory (c) == UnicodeCategory.ConnectorPunctuation;
864 public static bool IsKeyword (string s)
866 return keyword_strings.ContainsKey (s);
870 // Open parens micro parser. Detects both lambda and cast ambiguity.
872 int TokenizeOpenParens ()
877 int bracket_level = 0;
878 bool is_type = false;
879 bool can_be_type = false;
882 ptoken = current_token;
885 switch (current_token) {
886 case Token.CLOSE_PARENS:
890 // Expression inside parens is lambda, (int i) =>
892 if (current_token == Token.ARROW)
893 return Token.OPEN_PARENS_LAMBDA;
896 // Expression inside parens is single type, (int[])
899 return Token.OPEN_PARENS_CAST;
902 // Expression is possible cast, look at next token, (T)null
905 switch (current_token) {
906 case Token.OPEN_PARENS:
909 case Token.IDENTIFIER:
923 case Token.UNCHECKED:
928 // These can be part of a member access
942 return Token.OPEN_PARENS_CAST;
945 return Token.OPEN_PARENS;
948 case Token.DOUBLE_COLON:
949 if (ptoken != Token.IDENTIFIER && ptoken != Token.OP_GENERICS_GT)
954 case Token.IDENTIFIER:
957 if (bracket_level == 0) {
963 case Token.OP_GENERICS_LT:
965 case Token.DOUBLE_COLON:
967 if (bracket_level == 0)
971 can_be_type = is_type = false;
991 if (bracket_level == 0)
996 if (bracket_level == 0) {
998 can_be_type = is_type = false;
1002 case Token.OP_GENERICS_LT:
1003 case Token.OPEN_BRACKET:
1004 if (bracket_level++ == 0)
1008 case Token.OP_GENERICS_GT:
1009 case Token.CLOSE_BRACKET:
1013 case Token.INTERR_NULLABLE:
1015 if (bracket_level == 0)
1021 can_be_type = is_type = false;
1025 return Token.OPEN_PARENS;
1030 public static bool IsValidIdentifier (string s)
1032 if (s == null || s.Length == 0)
1035 if (!is_identifier_start_character (s [0]))
1038 for (int i = 1; i < s.Length; i ++)
1039 if (! is_identifier_part_character (s [i]))
1045 bool parse_less_than ()
1048 int the_token = token ();
1049 if (the_token == Token.OPEN_BRACKET) {
1051 the_token = token ();
1052 } while (the_token != Token.CLOSE_BRACKET);
1053 the_token = token ();
1054 } else if (the_token == Token.IN || the_token == Token.OUT) {
1055 the_token = token ();
1057 switch (the_token) {
1058 case Token.IDENTIFIER:
1076 case Token.OP_GENERICS_GT:
1083 the_token = token ();
1085 if (the_token == Token.OP_GENERICS_GT)
1087 else if (the_token == Token.COMMA || the_token == Token.DOT || the_token == Token.DOUBLE_COLON)
1089 else if (the_token == Token.INTERR_NULLABLE || the_token == Token.STAR)
1091 else if (the_token == Token.OP_GENERICS_LT) {
1092 if (!parse_less_than ())
1095 } else if (the_token == Token.OPEN_BRACKET) {
1097 the_token = token ();
1098 if (the_token == Token.CLOSE_BRACKET)
1100 else if (the_token == Token.COMMA)
1101 goto rank_specifiers;
1108 bool parse_generic_dimension (out int dimension)
1113 int the_token = token ();
1114 if (the_token == Token.OP_GENERICS_GT)
1116 else if (the_token == Token.COMMA) {
1124 public int peek_token ()
1129 the_token = token ();
1136 // Tonizes `?' using custom disambiguous rules to return one
1137 // of following tokens: INTERR_NULLABLE, OP_COALESCING, INTERR
1139 // Tricky expression look like:
1141 // Foo ? a = x ? b : c;
1143 int TokenizePossibleNullableType ()
1145 if (parsing_block == 0 || parsing_type > 0)
1146 return Token.INTERR_NULLABLE;
1148 int d = peek_char ();
1151 return Token.OP_COALESCING;
1154 switch (current_token) {
1155 case Token.CLOSE_PARENS:
1160 return Token.INTERR;
1164 if (d == ',' || d == ';' || d == '>')
1165 return Token.INTERR_NULLABLE;
1166 if (d == '*' || (d >= '0' && d <= '9'))
1167 return Token.INTERR;
1171 current_token = Token.NONE;
1173 switch (xtoken ()) {
1180 next_token = Token.INTERR;
1183 case Token.SEMICOLON:
1185 case Token.CLOSE_PARENS:
1186 case Token.OPEN_BRACKET:
1187 case Token.OP_GENERICS_GT:
1188 next_token = Token.INTERR_NULLABLE;
1196 if (next_token == -1) {
1197 switch (xtoken ()) {
1199 case Token.SEMICOLON:
1200 case Token.OPEN_BRACE:
1201 case Token.CLOSE_PARENS:
1203 next_token = Token.INTERR_NULLABLE;
1207 next_token = Token.INTERR;
1215 // All shorcuts failed, do it hard way
1217 while ((ntoken = xtoken ()) != Token.EOF) {
1218 if (ntoken == Token.SEMICOLON)
1221 if (ntoken == Token.COLON) {
1222 if (++colons == interrs)
1227 if (ntoken == Token.INTERR) {
1233 next_token = colons != interrs ? Token.INTERR_NULLABLE : Token.INTERR;
1242 bool decimal_digits (int c)
1245 bool seen_digits = false;
1248 if (number_pos == max_number_size)
1249 Error_NumericConstantTooLong ();
1250 number_builder [number_pos++] = (char) c;
1254 // We use peek_char2, because decimal_digits needs to do a
1255 // 2-character look-ahead (5.ToString for example).
1257 while ((d = peek_char2 ()) != -1){
1258 if (d >= '0' && d <= '9'){
1259 if (number_pos == max_number_size)
1260 Error_NumericConstantTooLong ();
1261 number_builder [number_pos++] = (char) d;
1271 static bool is_hex (int e)
1273 return (e >= '0' && e <= '9') || (e >= 'A' && e <= 'F') || (e >= 'a' && e <= 'f');
1276 static TypeCode real_type_suffix (int c)
1280 return TypeCode.Single;
1282 return TypeCode.Double;
1284 return TypeCode.Decimal;
1286 return TypeCode.Empty;
1290 int integer_type_suffix (ulong ul, int c)
1292 bool is_unsigned = false;
1293 bool is_long = false;
1296 bool scanning = true;
1309 // if we have not seen anything in between
1310 // report this error
1312 Report.Warning (78, 4, Location, "The 'l' suffix is easily confused with the digit '1' (use 'L' for clarity)");
1332 if (is_long && is_unsigned){
1333 val = new ULongLiteral (context.BuiltinTypes, ul, Location);
1334 return Token.LITERAL;
1338 // uint if possible, or ulong else.
1340 if ((ul & 0xffffffff00000000) == 0)
1341 val = new UIntLiteral (context.BuiltinTypes, (uint) ul, Location);
1343 val = new ULongLiteral (context.BuiltinTypes, ul, Location);
1344 } else if (is_long){
1345 // long if possible, ulong otherwise
1346 if ((ul & 0x8000000000000000) != 0)
1347 val = new ULongLiteral (context.BuiltinTypes, ul, Location);
1349 val = new LongLiteral (context.BuiltinTypes, (long) ul, Location);
1351 // int, uint, long or ulong in that order
1352 if ((ul & 0xffffffff00000000) == 0){
1353 uint ui = (uint) ul;
1355 if ((ui & 0x80000000) != 0)
1356 val = new UIntLiteral (context.BuiltinTypes, ui, Location);
1358 val = new IntLiteral (context.BuiltinTypes, (int) ui, Location);
1360 if ((ul & 0x8000000000000000) != 0)
1361 val = new ULongLiteral (context.BuiltinTypes, ul, Location);
1363 val = new LongLiteral (context.BuiltinTypes, (long) ul, Location);
1366 return Token.LITERAL;
1370 // given `c' as the next char in the input decide whether
1371 // we need to convert to a special type, and then choose
1372 // the best representation for the integer
1374 int adjust_int (int c)
1377 if (number_pos > 9){
1378 ulong ul = (uint) (number_builder [0] - '0');
1380 for (int i = 1; i < number_pos; i++){
1381 ul = checked ((ul * 10) + ((uint)(number_builder [i] - '0')));
1383 return integer_type_suffix (ul, c);
1385 uint ui = (uint) (number_builder [0] - '0');
1387 for (int i = 1; i < number_pos; i++){
1388 ui = checked ((ui * 10) + ((uint)(number_builder [i] - '0')));
1390 return integer_type_suffix (ui, c);
1392 } catch (OverflowException) {
1393 Error_NumericConstantTooLong ();
1394 val = new IntLiteral (context.BuiltinTypes, 0, Location);
1395 return Token.LITERAL;
1397 catch (FormatException) {
1398 Report.Error (1013, Location, "Invalid number");
1399 val = new IntLiteral (context.BuiltinTypes, 0, Location);
1400 return Token.LITERAL;
1404 int adjust_real (TypeCode t)
1406 string s = new String (number_builder, 0, number_pos);
1407 const string error_details = "Floating-point constant is outside the range of type `{0}'";
1410 case TypeCode.Decimal:
1412 val = new DecimalLiteral (context.BuiltinTypes, decimal.Parse (s, styles, csharp_format_info), Location);
1413 } catch (OverflowException) {
1414 val = new DecimalLiteral (context.BuiltinTypes, 0, Location);
1415 Report.Error (594, Location, error_details, "decimal");
1418 case TypeCode.Single:
1420 val = new FloatLiteral (context.BuiltinTypes, float.Parse (s, styles, csharp_format_info), Location);
1421 } catch (OverflowException) {
1422 val = new FloatLiteral (context.BuiltinTypes, 0, Location);
1423 Report.Error (594, Location, error_details, "float");
1428 val = new DoubleLiteral (context.BuiltinTypes, double.Parse (s, styles, csharp_format_info), Location);
1429 } catch (OverflowException) {
1430 val = new DoubleLiteral (context.BuiltinTypes, 0, Location);
1431 Report.Error (594, Location, error_details, "double");
1436 return Token.LITERAL;
1445 while ((d = peek_char ()) != -1){
1447 number_builder [number_pos++] = (char) d;
1453 string s = new String (number_builder, 0, number_pos);
1455 if (number_pos <= 8)
1456 ul = System.UInt32.Parse (s, NumberStyles.HexNumber);
1458 ul = System.UInt64.Parse (s, NumberStyles.HexNumber);
1459 } catch (OverflowException){
1460 Error_NumericConstantTooLong ();
1461 val = new IntLiteral (context.BuiltinTypes, 0, Location);
1462 return Token.LITERAL;
1464 catch (FormatException) {
1465 Report.Error (1013, Location, "Invalid number");
1466 val = new IntLiteral (context.BuiltinTypes, 0, Location);
1467 return Token.LITERAL;
1470 return integer_type_suffix (ul, peek_char ());
1474 // Invoked if we know we have .digits or digits
1476 int is_number (int c)
1478 bool is_real = false;
1482 if (c >= '0' && c <= '9'){
1484 int peek = peek_char ();
1486 if (peek == 'x' || peek == 'X')
1487 return handle_hex ();
1494 // We need to handle the case of
1495 // "1.1" vs "1.string" (LITERAL_FLOAT vs NUMBER DOT IDENTIFIER)
1498 if (decimal_digits ('.')){
1504 return adjust_int (-1);
1508 if (c == 'e' || c == 'E'){
1510 if (number_pos == max_number_size)
1511 Error_NumericConstantTooLong ();
1512 number_builder [number_pos++] = 'e';
1516 if (number_pos == max_number_size)
1517 Error_NumericConstantTooLong ();
1518 number_builder [number_pos++] = '+';
1520 } else if (c == '-') {
1521 if (number_pos == max_number_size)
1522 Error_NumericConstantTooLong ();
1523 number_builder [number_pos++] = '-';
1526 if (number_pos == max_number_size)
1527 Error_NumericConstantTooLong ();
1528 number_builder [number_pos++] = '+';
1535 var type = real_type_suffix (c);
1536 if (type == TypeCode.Empty && !is_real){
1538 return adjust_int (c);
1543 if (type == TypeCode.Empty){
1548 return adjust_real (type);
1550 throw new Exception ("Is Number should never reach this point");
1554 // Accepts exactly count (4 or 8) hex, no more no less
1556 int getHex (int count, out int surrogate, out bool error)
1561 int top = count != -1 ? count : 4;
1566 for (i = 0; i < top; i++){
1569 if (c >= '0' && c <= '9')
1570 c = (int) c - (int) '0';
1571 else if (c >= 'A' && c <= 'F')
1572 c = (int) c - (int) 'A' + 10;
1573 else if (c >= 'a' && c <= 'f')
1574 c = (int) c - (int) 'a' + 10;
1580 total = (total * 16) + c;
1582 int p = peek_char ();
1585 if (!is_hex ((char)p))
1591 if (total > 0x0010FFFF) {
1596 if (total >= 0x00010000) {
1597 surrogate = ((total - 0x00010000) % 0x0400 + 0xDC00);
1598 total = ((total - 0x00010000) / 0x0400 + 0xD800);
1605 int escape (int c, out int surrogate)
1641 v = getHex (-1, out surrogate, out error);
1647 return EscapeUnicode (d, out surrogate);
1650 Report.Error (1009, Location, "Unrecognized escape sequence `\\{0}'", ((char)d).ToString ());
1659 int EscapeUnicode (int ch, out int surrogate)
1663 ch = getHex (8, out surrogate, out error);
1665 ch = getHex (4, out surrogate, out error);
1669 Report.Error (1009, Location, "Unrecognized escape sequence");
1677 if (putback_char != -1) {
1685 if (peek_char () == '\n') {
1691 } else if (x == '\n') {
1699 void advance_line ()
1709 if (putback_char == -1)
1710 putback_char = reader.Read ();
1711 return putback_char;
1716 if (putback_char != -1)
1717 return putback_char;
1718 return reader.Peek ();
1721 void putback (int c)
1723 if (putback_char != -1){
1724 Console.WriteLine ("Col: " + col);
1725 Console.WriteLine ("Row: " + line);
1726 Console.WriteLine ("Name: " + ref_name.Name);
1727 Console.WriteLine ("Current [{0}] putting back [{1}] ", putback_char, c);
1728 throw new Exception ("This should not happen putback on putback");
1730 if (c == '\n' || col == 0) {
1731 // It won't happen though.
1741 public bool advance ()
1743 return peek_char () != -1 || CompleteOnEOF;
1746 public Object Value {
1752 public Object value ()
1759 current_token = xtoken ();
1760 return current_token;
1763 int TokenizePreprocessorIdentifier (out int c)
1765 // skip over white space
1768 } while (c == ' ' || c == '\t');
1772 while (c != -1 && c >= 'a' && c <= 'z') {
1773 id_builder[pos++] = (char) c;
1776 int peek = peek_char ();
1777 if (peek == 'U' || peek == 'u') {
1779 c = EscapeUnicode (c, out surrogate);
1780 if (surrogate != 0) {
1781 if (is_identifier_part_character ((char) c)) {
1782 id_builder[pos++] = (char) c;
1793 PreprocessorDirective get_cmd_arg (out string arg)
1797 tokens_seen = false;
1800 var cmd = GetPreprocessorDirective (id_builder, TokenizePreprocessorIdentifier (out c));
1802 if ((cmd & PreprocessorDirective.CustomArgumentsParsing) != 0)
1805 // skip over white space
1806 while (c == ' ' || c == '\t')
1809 int has_identifier_argument = (int)(cmd & PreprocessorDirective.RequiresArgument);
1812 while (c != -1 && c != '\n') {
1813 if (c == '\\' && has_identifier_argument >= 0) {
1814 if (has_identifier_argument != 0) {
1815 has_identifier_argument = 1;
1817 int peek = peek_char ();
1818 if (peek == 'U' || peek == 'u') {
1820 c = EscapeUnicode (c, out surrogate);
1821 if (surrogate != 0) {
1822 if (is_identifier_part_character ((char) c)) {
1823 if (pos == value_builder.Length)
1824 Array.Resize (ref value_builder, pos * 2);
1826 value_builder[pos++] = (char) c;
1832 has_identifier_argument = -1;
1834 } else if (c == '/' && peek_char () == '/') {
1836 // Eat single-line comments
1841 } while (c != -1 && c != '\n');
1846 if (pos == value_builder.Length)
1847 Array.Resize (ref value_builder, pos * 2);
1849 value_builder[pos++] = (char) c;
1854 if (pos > max_id_size)
1855 arg = new string (value_builder, 0, pos);
1857 arg = InternIdentifier (value_builder, pos);
1859 // Eat any trailing whitespaces
1860 arg = arg.Trim (simple_whitespaces);
1867 // Handles the #line directive
1869 bool PreProcessLine (string arg)
1871 if (arg.Length == 0)
1874 if (arg == "default"){
1876 ref_name = file_name;
1878 Location.Push (file_name, ref_name);
1880 } else if (arg == "hidden"){
1888 if ((pos = arg.IndexOf (' ')) != -1 && pos != 0){
1889 ref_line = System.Int32.Parse (arg.Substring (0, pos));
1892 char [] quotes = { '\"' };
1894 string name = arg.Substring (pos). Trim (quotes);
1895 ref_name = context.LookupFile (file_name, name);
1896 file_name.AddIncludeFile (ref_name);
1898 Location.Push (file_name, ref_name);
1900 ref_line = System.Int32.Parse (arg);
1911 // Handles #define and #undef
1913 void PreProcessDefinition (bool is_define, string ident, bool caller_is_taking)
1915 if (ident.Length == 0 || ident == "true" || ident == "false"){
1916 Report.Error (1001, Location, "Missing identifier to pre-processor directive");
1920 if (ident.IndexOfAny (simple_whitespaces) != -1){
1921 Error_EndLineExpected ();
1925 if (!is_identifier_start_character (ident [0]))
1926 Report.Error (1001, Location, "Identifier expected: {0}", ident);
1928 foreach (char c in ident.Substring (1)){
1929 if (!is_identifier_part_character (c)){
1930 Report.Error (1001, Location, "Identifier expected: {0}", ident);
1935 if (!caller_is_taking)
1942 if (context.Settings.IsConditionalSymbolDefined (ident))
1945 file_name.AddDefine (ident);
1950 file_name.AddUndefine (ident);
1954 byte read_hex (out bool error)
1957 int c = get_char ();
1959 if ((c >= '0') && (c <= '9'))
1960 total = (int) c - (int) '0';
1961 else if ((c >= 'A') && (c <= 'F'))
1962 total = (int) c - (int) 'A' + 10;
1963 else if ((c >= 'a') && (c <= 'f'))
1964 total = (int) c - (int) 'a' + 10;
1973 if ((c >= '0') && (c <= '9'))
1974 total += (int) c - (int) '0';
1975 else if ((c >= 'A') && (c <= 'F'))
1976 total += (int) c - (int) 'A' + 10;
1977 else if ((c >= 'a') && (c <= 'f'))
1978 total += (int) c - (int) 'a' + 10;
1985 return (byte) total;
1989 // Parses #pragma checksum
1991 bool ParsePragmaChecksum ()
1994 // The syntax is ` "foo.txt" "{guid}" "hash"'
1996 int c = get_char ();
2001 string_builder.Length = 0;
2002 while (c != -1 && c != '\n') {
2009 string_builder.Append ((char) c);
2012 if (string_builder.Length == 0) {
2013 Report.Warning (1709, 1, Location, "Filename specified for preprocessor directive is empty");
2016 // TODO: Any white-spaces count
2020 SourceFile file = context.LookupFile (file_name, string_builder.ToString ());
2022 if (get_char () != '"' || get_char () != '{')
2026 byte[] guid_bytes = new byte [16];
2029 for (; i < 4; i++) {
2030 guid_bytes [i] = read_hex (out error);
2035 if (get_char () != '-')
2038 for (; i < 10; i++) {
2039 guid_bytes [i] = read_hex (out error);
2043 guid_bytes [i++] = read_hex (out error);
2047 if (get_char () != '-')
2051 for (; i < 16; i++) {
2052 guid_bytes [i] = read_hex (out error);
2057 if (get_char () != '}' || get_char () != '"')
2060 // TODO: Any white-spaces count
2065 if (get_char () != '"')
2068 // Any length of checksum
2069 List<byte> checksum_bytes = new List<byte> (16);
2072 while (c != '"' && c != -1) {
2073 checksum_bytes.Add (read_hex (out error));
2081 ReadSingleLineComment ();
2082 } else if (get_char () != '"') {
2086 file.SetChecksum (guid_bytes, checksum_bytes.ToArray ());
2087 ref_name.AutoGenerated = true;
2091 bool IsTokenIdentifierEqual (char[] identifier)
2093 for (int i = 0; i < identifier.Length; ++i) {
2094 if (identifier[i] != id_builder[i])
2101 int TokenizePragmaNumber (ref int c)
2107 if (c >= '0' && c <= '9') {
2109 uint ui = (uint) (number_builder[0] - '0');
2112 for (int i = 1; i < number_pos; i++) {
2113 ui = checked ((ui * 10) + ((uint) (number_builder[i] - '0')));
2117 } catch (OverflowException) {
2118 Error_NumericConstantTooLong ();
2125 // skip over white space
2126 while (c == ' ' || c == '\t')
2133 // skip over white space
2134 while (c == ' ' || c == '\t')
2139 ReadSingleLineComment ();
2141 Report.Warning (1692, 1, Location, "Invalid number");
2143 // Read everything till the end of the line or file
2146 } while (c != -1 && c != '\n');
2153 void ReadSingleLineComment ()
2155 if (peek_char () != '/')
2156 Report.Warning (1696, 1, Location, "Single-line comment or end-of-line expected");
2158 // Read everything till the end of the line or file
2162 } while (c != -1 && c != '\n');
2166 /// Handles #pragma directive
2168 void ParsePragmaDirective (string arg)
2171 int length = TokenizePreprocessorIdentifier (out c);
2172 if (length == pragma_warning.Length && IsTokenIdentifierEqual (pragma_warning)) {
2173 length = TokenizePreprocessorIdentifier (out c);
2176 // #pragma warning disable
2177 // #pragma warning restore
2179 if (length == pragma_warning_disable.Length) {
2180 bool disable = IsTokenIdentifierEqual (pragma_warning_disable);
2181 if (disable || IsTokenIdentifierEqual (pragma_warning_restore)) {
2182 // skip over white space
2183 while (c == ' ' || c == '\t')
2188 if (c == '\n' || c == '/') {
2190 ReadSingleLineComment ();
2193 // Disable/Restore all warnings
2196 Report.RegisterWarningRegion (loc).WarningDisable (loc.Row);
2198 Report.RegisterWarningRegion (loc).WarningEnable (loc.Row);
2202 // Disable/Restore a warning or group of warnings
2206 code = TokenizePragmaNumber (ref c);
2209 Report.RegisterWarningRegion (loc).WarningDisable (loc, code, Report);
2211 Report.RegisterWarningRegion (loc).WarningEnable (loc, code, Report);
2214 } while (code >= 0 && c != '\n' && c != -1);
2221 Report.Warning (1634, 1, Location, "Expected disable or restore");
2228 if (length == pragma_checksum.Length && IsTokenIdentifierEqual (pragma_checksum)) {
2229 if (c != ' ' || !ParsePragmaChecksum ()) {
2230 Report.Warning (1695, 1, Location,
2231 "Invalid #pragma checksum syntax. Expected \"filename\" \"{XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX}\" \"XXXX...\"");
2237 Report.Warning (1633, 1, Location, "Unrecognized #pragma directive");
2240 bool eval_val (string s)
2247 return file_name.IsConditionalDefined (context, s);
2250 bool pp_primary (ref string s)
2259 s = s.Substring (1);
2260 bool val = pp_expr (ref s, false);
2261 if (s.Length > 0 && s [0] == ')'){
2262 s = s.Substring (1);
2265 Error_InvalidDirective ();
2269 if (is_identifier_start_character (c)){
2275 if (is_identifier_part_character (c)){
2279 bool v = eval_val (s.Substring (0, j));
2280 s = s.Substring (j);
2283 bool vv = eval_val (s);
2288 Error_InvalidDirective ();
2292 bool pp_unary (ref string s)
2299 if (len > 1 && s [1] == '='){
2300 Error_InvalidDirective ();
2303 s = s.Substring (1);
2304 return ! pp_primary (ref s);
2306 return pp_primary (ref s);
2308 Error_InvalidDirective ();
2313 bool pp_eq (ref string s)
2315 bool va = pp_unary (ref s);
2321 if (len > 2 && s [1] == '='){
2322 s = s.Substring (2);
2323 return va == pp_unary (ref s);
2325 Error_InvalidDirective ();
2328 } else if (s [0] == '!' && len > 1 && s [1] == '='){
2329 s = s.Substring (2);
2331 return va != pp_unary (ref s);
2340 bool pp_and (ref string s)
2342 bool va = pp_eq (ref s);
2348 if (len > 2 && s [1] == '&'){
2349 s = s.Substring (2);
2350 return (va & pp_and (ref s));
2352 Error_InvalidDirective ();
2361 // Evaluates an expression for `#if' or `#elif'
2363 bool pp_expr (ref string s, bool isTerm)
2365 bool va = pp_and (ref s);
2372 if (len > 2 && s [1] == '|'){
2373 s = s.Substring (2);
2374 return va | pp_expr (ref s, isTerm);
2376 Error_InvalidDirective ();
2381 Error_EndLineExpected ();
2389 bool eval (string s)
2391 bool v = pp_expr (ref s, true);
2400 void Error_NumericConstantTooLong ()
2402 Report.Error (1021, Location, "Integral constant is too large");
2405 void Error_InvalidDirective ()
2407 Report.Error (1517, Location, "Invalid preprocessor directive");
2410 void Error_UnexpectedDirective (string extra)
2414 "Unexpected processor directive ({0})", extra);
2417 void Error_TokensSeen ()
2419 Report.Error (1032, Location,
2420 "Cannot define or undefine preprocessor symbols after first token in file");
2423 void Eror_WrongPreprocessorLocation ()
2425 Report.Error (1040, Location,
2426 "Preprocessor directives must appear as the first non-whitespace character on a line");
2429 void Error_EndLineExpected ()
2431 Report.Error (1025, Location, "Single-line comment or end-of-line expected");
2435 // Raises a warning when tokenizer found documentation comment
2436 // on unexpected place
2438 void WarningMisplacedComment (Location loc)
2440 if (doc_state != XmlCommentState.Error) {
2441 doc_state = XmlCommentState.Error;
2442 Report.Warning (1587, 2, loc, "XML comment is not placed on a valid language element");
2447 // if true, then the code continues processing the code
2448 // if false, the code stays in a loop until another directive is
2450 // When caller_is_taking is false we ignore all directives except the ones
2451 // which can help us to identify where the #if block ends
2452 bool ParsePreprocessingDirective (bool caller_is_taking)
2455 bool region_directive = false;
2457 var directive = get_cmd_arg (out arg);
2460 // The first group of pre-processing instructions is always processed
2462 switch (directive) {
2463 case PreprocessorDirective.Region:
2464 region_directive = true;
2466 goto case PreprocessorDirective.If;
2468 case PreprocessorDirective.Endregion:
2469 if (ifstack == null || ifstack.Count == 0){
2470 Error_UnexpectedDirective ("no #region for this #endregion");
2473 int pop = ifstack.Pop ();
2475 if ((pop & REGION) == 0)
2476 Report.Error (1027, Location, "Expected `#endif' directive");
2478 return caller_is_taking;
2480 case PreprocessorDirective.If:
2481 if (ifstack == null)
2482 ifstack = new Stack<int> (2);
2484 int flags = region_directive ? REGION : 0;
2485 if (ifstack.Count == 0){
2486 flags |= PARENT_TAKING;
2488 int state = ifstack.Peek ();
2489 if ((state & TAKING) != 0) {
2490 flags |= PARENT_TAKING;
2494 if (eval (arg) && caller_is_taking) {
2495 ifstack.Push (flags | TAKING);
2498 ifstack.Push (flags);
2501 case PreprocessorDirective.Endif:
2502 if (ifstack == null || ifstack.Count == 0){
2503 Error_UnexpectedDirective ("no #if for this #endif");
2506 pop = ifstack.Pop ();
2508 if ((pop & REGION) != 0)
2509 Report.Error (1038, Location, "#endregion directive expected");
2511 if (arg.Length != 0) {
2512 Error_EndLineExpected ();
2515 if (ifstack.Count == 0)
2518 int state = ifstack.Peek ();
2519 return (state & TAKING) != 0;
2522 case PreprocessorDirective.Elif:
2523 if (ifstack == null || ifstack.Count == 0){
2524 Error_UnexpectedDirective ("no #if for this #elif");
2527 int state = ifstack.Pop ();
2529 if ((state & REGION) != 0) {
2530 Report.Error (1038, Location, "#endregion directive expected");
2534 if ((state & ELSE_SEEN) != 0){
2535 Error_UnexpectedDirective ("#elif not valid after #else");
2539 if ((state & TAKING) != 0) {
2544 if (eval (arg) && ((state & PARENT_TAKING) != 0)){
2545 ifstack.Push (state | TAKING);
2549 ifstack.Push (state);
2553 case PreprocessorDirective.Else:
2554 if (ifstack == null || ifstack.Count == 0){
2555 Error_UnexpectedDirective ("no #if for this #else");
2558 int state = ifstack.Peek ();
2560 if ((state & REGION) != 0) {
2561 Report.Error (1038, Location, "#endregion directive expected");
2565 if ((state & ELSE_SEEN) != 0){
2566 Error_UnexpectedDirective ("#else within #else");
2572 if (arg.Length != 0) {
2573 Error_EndLineExpected ();
2578 if ((state & PARENT_TAKING) != 0) {
2579 ret = (state & TAKING) == 0;
2587 ifstack.Push (state | ELSE_SEEN);
2591 case PreprocessorDirective.Define:
2592 if (any_token_seen){
2593 Error_TokensSeen ();
2594 return caller_is_taking;
2596 PreProcessDefinition (true, arg, caller_is_taking);
2597 return caller_is_taking;
2599 case PreprocessorDirective.Undef:
2600 if (any_token_seen){
2601 Error_TokensSeen ();
2602 return caller_is_taking;
2604 PreProcessDefinition (false, arg, caller_is_taking);
2605 return caller_is_taking;
2607 case PreprocessorDirective.Invalid:
2608 Report.Error (1024, Location, "Wrong preprocessor directive");
2613 // These are only processed if we are in a `taking' block
2615 if (!caller_is_taking)
2619 case PreprocessorDirective.Error:
2620 Report.Error (1029, Location, "#error: '{0}'", arg);
2623 case PreprocessorDirective.Warning:
2624 Report.Warning (1030, 1, Location, "#warning: `{0}'", arg);
2627 case PreprocessorDirective.Pragma:
2628 if (context.Settings.Version == LanguageVersion.ISO_1) {
2629 Report.FeatureIsNotAvailable (context, Location, "#pragma");
2632 ParsePragmaDirective (arg);
2635 case PreprocessorDirective.Line:
2636 if (!PreProcessLine (arg))
2639 "The line number specified for #line directive is missing or invalid");
2640 return caller_is_taking;
2643 throw new NotImplementedException (directive.ToString ());
2646 private int consume_string (bool quoted)
2650 Location start_location = Location;
2654 if (quoted && peek_char () == '"') {
2655 if (pos == value_builder.Length)
2656 Array.Resize (ref value_builder, pos * 2);
2658 value_builder[pos++] = (char) c;
2667 s = InternIdentifier (value_builder, pos);
2669 s = new string (value_builder, 0, pos);
2671 val = new StringLiteral (context.BuiltinTypes, s, start_location);
2672 return Token.LITERAL;
2677 Report.Error (1010, Location, "Newline in constant");
2678 } else if (c == '\\' && !quoted) {
2680 c = escape (c, out surrogate);
2683 if (surrogate != 0) {
2684 if (pos == value_builder.Length)
2685 Array.Resize (ref value_builder, pos * 2);
2687 value_builder[pos++] = (char) c;
2690 } else if (c == -1) {
2691 Report.Error (1039, Location, "Unterminated string literal");
2695 if (pos == value_builder.Length)
2696 Array.Resize (ref value_builder, pos * 2);
2698 value_builder[pos++] = (char) c;
2702 private int consume_identifier (int s)
2704 int res = consume_identifier (s, false);
2706 if (doc_state == XmlCommentState.Allowed)
2707 doc_state = XmlCommentState.NotAllowed;
2712 int consume_identifier (int c, bool quoted)
2715 // This method is very performance sensitive. It accounts
2716 // for approximately 25% of all parser time
2724 c = escape (c, out surrogate);
2725 if (surrogate != 0) {
2726 id_builder [pos++] = (char) c;
2731 id_builder [pos++] = (char) c;
2737 if ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_' || (c >= '0' && c <= '9')) {
2738 id_builder [pos++] = (char) c;
2745 c = escape (c, out surrogate);
2746 if (surrogate != 0) {
2747 if (is_identifier_part_character ((char) c))
2748 id_builder[pos++] = (char) c;
2754 } else if (Char.IsLetter ((char) c) || Char.GetUnicodeCategory ((char) c) == UnicodeCategory.ConnectorPunctuation) {
2755 id_builder [pos++] = (char) c;
2762 } catch (IndexOutOfRangeException) {
2763 Report.Error (645, Location, "Identifier too long (limit is 512 chars)");
2771 // Optimization: avoids doing the keyword lookup
2772 // on uppercase letters
2774 if (id_builder [0] >= '_' && !quoted) {
2775 int keyword = GetKeyword (id_builder, pos);
2776 if (keyword != -1) {
2777 val = LocatedToken.Create (null, ref_line, column);
2782 string s = InternIdentifier (id_builder, pos);
2783 val = LocatedToken.Create (s, ref_line, column);
2784 if (quoted && parsing_attribute_section)
2785 AddEscapedIdentifier (((LocatedToken) val).Location);
2787 return Token.IDENTIFIER;
2790 static string InternIdentifier (char[] charBuffer, int length)
2793 // Keep identifiers in an array of hashtables to avoid needless
2796 var identifiers_group = identifiers[length];
2798 if (identifiers_group != null) {
2799 if (identifiers_group.TryGetValue (charBuffer, out s)) {
2803 // TODO: this should be number of files dependant
2804 // corlib compilation peaks at 1000 and System.Core at 150
2805 int capacity = length > 20 ? 10 : 100;
2806 identifiers_group = new Dictionary<char[], string> (capacity, new IdentifiersComparer (length));
2807 identifiers[length] = identifiers_group;
2810 char[] chars = new char[length];
2811 Array.Copy (charBuffer, chars, length);
2813 s = new string (charBuffer, 0, length);
2814 identifiers_group.Add (chars, s);
2818 public int xtoken ()
2822 // Whether we have seen comments on the current line
2823 bool comments_seen = false;
2824 while ((c = get_char ()) != -1) {
2827 col = ((col - 1 + tab_size) / tab_size) * tab_size;
2835 case 0xFEFF: // Ignore BOM anywhere in the file
2838 /* This is required for compatibility with .NET
2840 if (peek_char () == 0xBB) {
2843 if (get_char () == 0xBF)
2851 return consume_identifier (c);
2854 val = LocatedToken.Create (ref_line, col);
2855 return Token.OPEN_BRACE;
2857 val = LocatedToken.Create (ref_line, col);
2858 return Token.CLOSE_BRACE;
2860 // To block doccomment inside attribute declaration.
2861 if (doc_state == XmlCommentState.Allowed)
2862 doc_state = XmlCommentState.NotAllowed;
2864 val = LocatedToken.Create (ref_line, col);
2866 if (parsing_block == 0 || lambda_arguments_parsing)
2867 return Token.OPEN_BRACKET;
2869 int next = peek_char ();
2873 return Token.OPEN_BRACKET;
2881 next = peek_token ();
2882 if (next == Token.COMMA || next == Token.CLOSE_BRACKET)
2883 return Token.OPEN_BRACKET;
2885 return Token.OPEN_BRACKET_EXPR;
2887 return Token.OPEN_BRACKET_EXPR;
2890 LocatedToken.CreateOptional (ref_line, col, ref val);
2891 return Token.CLOSE_BRACKET;
2893 val = LocatedToken.Create (ref_line, col);
2895 // An expression versions of parens can appear in block context only
2897 if (parsing_block != 0 && !lambda_arguments_parsing) {
2900 // Optmize most common case where we know that parens
2903 switch (current_token) {
2904 case Token.IDENTIFIER:
2912 case Token.DELEGATE:
2913 case Token.OP_GENERICS_GT:
2914 return Token.OPEN_PARENS;
2917 // Optimize using peek
2918 int xx = peek_char ();
2925 return Token.OPEN_PARENS;
2928 lambda_arguments_parsing = true;
2930 d = TokenizeOpenParens ();
2932 lambda_arguments_parsing = false;
2936 return Token.OPEN_PARENS;
2938 LocatedToken.CreateOptional (ref_line, col, ref val);
2939 return Token.CLOSE_PARENS;
2941 LocatedToken.CreateOptional (ref_line, col, ref val);
2944 LocatedToken.CreateOptional (ref_line, col, ref val);
2945 return Token.SEMICOLON;
2947 val = LocatedToken.Create (ref_line, col);
2950 val = LocatedToken.Create (ref_line, col);
2951 return TokenizePossibleNullableType ();
2953 val = LocatedToken.Create (ref_line, col);
2954 if (parsing_generic_less_than++ > 0)
2955 return Token.OP_GENERICS_LT;
2957 return TokenizeLessThan ();
2960 val = LocatedToken.Create (ref_line, col);
2968 if (parsing_generic_less_than > 1 || (parsing_generic_less_than == 1 && d != '>')) {
2969 parsing_generic_less_than--;
2970 return Token.OP_GENERICS_GT;
2979 return Token.OP_SHIFT_RIGHT_ASSIGN;
2981 return Token.OP_SHIFT_RIGHT;
2987 val = LocatedToken.Create (ref_line, col);
2991 } else if (d == '=') {
2992 d = Token.OP_ADD_ASSIGN;
3000 val = LocatedToken.Create (ref_line, col);
3004 } else if (d == '=')
3005 d = Token.OP_SUB_ASSIGN;
3015 val = LocatedToken.Create (ref_line, col);
3016 if (peek_char () == '='){
3023 val = LocatedToken.Create (ref_line, col);
3034 return Token.ASSIGN;
3037 val = LocatedToken.Create (ref_line, col);
3041 return Token.OP_AND;
3045 return Token.OP_AND_ASSIGN;
3047 return Token.BITWISE_AND;
3050 val = LocatedToken.Create (ref_line, col);
3058 return Token.OP_OR_ASSIGN;
3060 return Token.BITWISE_OR;
3063 val = LocatedToken.Create (ref_line, col);
3064 if (peek_char () == '='){
3066 return Token.OP_MULT_ASSIGN;
3073 val = LocatedToken.Create (ref_line, col);
3075 return Token.OP_DIV_ASSIGN;
3078 // Handle double-slash comments.
3081 if (doc_processing) {
3082 if (peek_char () == '/') {
3084 // Don't allow ////.
3085 if ((d = peek_char ()) != '/') {
3086 if (doc_state == XmlCommentState.Allowed)
3087 handle_one_line_xml_comment ();
3088 else if (doc_state == XmlCommentState.NotAllowed)
3089 WarningMisplacedComment (Location - 3);
3092 if (xml_comment_buffer.Length > 0)
3093 doc_state = XmlCommentState.NotAllowed;
3097 while ((d = get_char ()) != -1 && d != '\n');
3099 any_token_seen |= tokens_seen;
3100 tokens_seen = false;
3101 comments_seen = false;
3103 } else if (d == '*'){
3105 bool docAppend = false;
3106 if (doc_processing && peek_char () == '*') {
3108 // But when it is /**/, just do nothing.
3109 if (peek_char () == '/') {
3113 if (doc_state == XmlCommentState.Allowed)
3115 else if (doc_state == XmlCommentState.NotAllowed) {
3116 WarningMisplacedComment (Location - 2);
3120 int current_comment_start = 0;
3122 current_comment_start = xml_comment_buffer.Length;
3123 xml_comment_buffer.Append (Environment.NewLine);
3126 while ((d = get_char ()) != -1){
3127 if (d == '*' && peek_char () == '/'){
3129 comments_seen = true;
3133 xml_comment_buffer.Append ((char) d);
3136 any_token_seen |= tokens_seen;
3137 tokens_seen = false;
3139 // Reset 'comments_seen' just to be consistent.
3140 // It doesn't matter either way, here.
3142 comments_seen = false;
3146 Report.Error (1035, Location, "End-of-file found, '*/' expected");
3149 update_formatted_doc_comment (current_comment_start);
3152 val = LocatedToken.Create (ref_line, col);
3156 val = LocatedToken.Create (ref_line, col);
3157 if (peek_char () == '='){
3159 return Token.OP_MOD_ASSIGN;
3161 return Token.PERCENT;
3164 val = LocatedToken.Create (ref_line, col);
3165 if (peek_char () == '='){
3167 return Token.OP_XOR_ASSIGN;
3169 return Token.CARRET;
3172 val = LocatedToken.Create (ref_line, col);
3173 if (peek_char () == ':') {
3175 return Token.DOUBLE_COLON;
3179 case '0': case '1': case '2': case '3': case '4':
3180 case '5': case '6': case '7': case '8': case '9':
3182 return is_number (c);
3184 case '\n': // white space
3185 any_token_seen |= tokens_seen;
3186 tokens_seen = false;
3187 comments_seen = false;
3193 if (d >= '0' && d <= '9')
3194 return is_number (c);
3196 LocatedToken.CreateOptional (ref_line, col, ref val);
3200 if (tokens_seen || comments_seen) {
3201 Eror_WrongPreprocessorLocation ();
3205 if (ParsePreprocessingDirective (true))
3208 bool directive_expected = false;
3209 while ((c = get_char ()) != -1) {
3211 directive_expected = true;
3212 } else if (!directive_expected) {
3213 // TODO: Implement comment support for disabled code and uncomment this code
3215 // Eror_WrongPreprocessorLocation ();
3216 // return Token.ERROR;
3221 if (c == ' ' || c == '\t' || c == '\n' || c == '\f' || c == '\v' )
3225 if (ParsePreprocessingDirective (false))
3228 directive_expected = false;
3232 tokens_seen = false;
3239 return consume_string (false);
3242 return TokenizeBackslash ();
3248 return consume_string (true);
3251 if (is_identifier_start_character (c)){
3252 return consume_identifier (c, true);
3255 Report.Error (1646, Location, "Keyword, identifier, or string expected after verbatim specifier: @");
3258 case EvalStatementParserCharacter:
3259 return Token.EVAL_STATEMENT_PARSER;
3260 case EvalCompilationUnitParserCharacter:
3261 return Token.EVAL_COMPILATION_UNIT_PARSER;
3262 case EvalUsingDeclarationsParserCharacter:
3263 return Token.EVAL_USING_DECLARATIONS_UNIT_PARSER;
3264 case DocumentationXref:
3265 return Token.DOC_SEE;
3268 if (is_identifier_start_character (c)) {
3270 return consume_identifier (c);
3273 if (char.IsWhiteSpace ((char) c))
3276 Report.Error (1056, Location, "Unexpected character `{0}'", ((char) c).ToString ());
3281 return Token.COMPLETE_COMPLETION;
3284 return Token.GENERATE_COMPLETION;
3291 int TokenizeBackslash ()
3293 int c = get_char ();
3296 val = new CharLiteral (context.BuiltinTypes, (char) c, Location);
3297 Report.Error (1011, Location, "Empty character literal");
3298 return Token.LITERAL;
3302 Report.Error (1010, Location, "Newline in constant");
3307 c = escape (c, out d);
3311 throw new NotImplementedException ();
3313 val = new CharLiteral (context.BuiltinTypes, (char) c, Location);
3317 Report.Error (1012, Location, "Too many characters in character literal");
3319 // Try to recover, read until newline or next "'"
3320 while ((c = get_char ()) != -1) {
3321 if (c == '\n' || c == '\'')
3326 return Token.LITERAL;
3329 int TokenizeLessThan ()
3332 if (handle_typeof) {
3334 if (parse_generic_dimension (out d)) {
3337 return Token.GENERIC_DIMENSION;
3342 // Save current position and parse next token.
3344 if (parse_less_than ()) {
3345 if (parsing_generic_declaration && (parsing_generic_declaration_doc || token () != Token.DOT)) {
3346 d = Token.OP_GENERICS_LT_DECL;
3348 d = Token.OP_GENERICS_LT;
3355 parsing_generic_less_than = 0;
3364 return Token.OP_SHIFT_LEFT_ASSIGN;
3366 return Token.OP_SHIFT_LEFT;
3377 // Handles one line xml comment
3379 private void handle_one_line_xml_comment ()
3382 while ((c = peek_char ()) == ' ')
3383 get_char (); // skip heading whitespaces.
3384 while ((c = peek_char ()) != -1 && c != '\n' && c != '\r') {
3385 xml_comment_buffer.Append ((char) get_char ());
3387 if (c == '\r' || c == '\n')
3388 xml_comment_buffer.Append (Environment.NewLine);
3392 // Remove heading "*" in Javadoc-like xml documentation.
3394 private void update_formatted_doc_comment (int current_comment_start)
3396 int length = xml_comment_buffer.Length - current_comment_start;
3397 string [] lines = xml_comment_buffer.ToString (
3398 current_comment_start,
3399 length).Replace ("\r", "").Split ('\n');
3401 // The first line starts with /**, thus it is not target
3402 // for the format check.
3403 for (int i = 1; i < lines.Length; i++) {
3404 string s = lines [i];
3405 int idx = s.IndexOf ('*');
3408 if (i < lines.Length - 1)
3412 head = s.Substring (0, idx);
3413 foreach (char c in head)
3416 lines [i] = s.Substring (idx + 1);
3418 xml_comment_buffer.Remove (current_comment_start, length);
3419 xml_comment_buffer.Insert (current_comment_start, String.Join (Environment.NewLine, lines));
3423 // Checks if there was incorrect doc comments and raise
3426 public void check_incorrect_doc_comment ()
3428 if (xml_comment_buffer.Length > 0)
3429 WarningMisplacedComment (Location);
3433 // Consumes the saved xml comment lines (if any)
3434 // as for current target member or type.
3436 public string consume_doc_comment ()
3438 if (xml_comment_buffer.Length > 0) {
3439 string ret = xml_comment_buffer.ToString ();
3440 reset_doc_comment ();
3447 get { return context.Report; }
3450 void reset_doc_comment ()
3452 xml_comment_buffer.Length = 0;
3455 public void cleanup ()
3457 if (ifstack != null && ifstack.Count >= 1) {
3458 int state = ifstack.Pop ();
3459 if ((state & REGION) != 0)
3460 Report.Error (1038, Location, "#endregion directive expected");
3462 Report.Error (1027, Location, "Expected `#endif' directive");
3468 // Indicates whether it accepts XML documentation or not.
3470 public enum XmlCommentState {
3471 // comment is allowed in this state.
3473 // comment is not allowed in this state.
3475 // once comments appeared when it is NotAllowed, then the
3476 // state is changed to it, until the state is changed to