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@gmail.com)
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
12 // Copyright 2011 Xamarin, Inc (http://www.xamarin.com)
17 using System.Collections.Generic;
18 using System.Globalization;
19 using System.Diagnostics;
20 using System.Collections;
25 // This class has to be used by parser only, it reuses token
26 // details once a file is parsed
28 public class LocatedToken
30 public int row, column;
32 public SourceFile file;
34 public LocatedToken ()
38 public LocatedToken (string value, Location loc)
41 file = loc.SourceFile;
46 public override string ToString ()
48 return string.Format ("Token '{0}' at {1},{2}", Value, row, column);
51 public Location Location
53 get { return new Location (file, row, column); }
63 /// Tokenizer for C# source code.
65 public class Tokenizer : yyParser.yyInput
69 public readonly T Token;
70 public KeywordEntry<T> Next;
71 public readonly char[] Value;
73 public KeywordEntry (string value, T token)
75 this.Value = value.ToCharArray ();
80 sealed class IdentifiersComparer : IEqualityComparer<char[]>
84 public IdentifiersComparer (int length)
89 public bool Equals (char[] x, char[] y)
91 for (int i = 0; i < length; ++i)
98 public int GetHashCode (char[] obj)
101 for (int i = 0; i < length; ++i)
102 h = (h << 5) - h + obj [i];
108 public class LocatedTokenBuffer
110 readonly LocatedToken[] buffer;
113 public LocatedTokenBuffer ()
115 buffer = new LocatedToken[0];
118 public LocatedTokenBuffer (LocatedToken[] buffer)
120 this.buffer = buffer ?? new LocatedToken[0];
123 public LocatedToken Create (SourceFile file, int row, int column)
125 return Create (null, file, row, column);
128 public LocatedToken Create (string value, SourceFile file, int row, int column)
131 // TODO: I am not very happy about the logic but it's the best
132 // what I could come up with for now.
133 // Ideally we should be using just tiny buffer (256 elements) which
134 // is enough to hold all details for currect stack and recycle elements
135 // poped from the stack but there is a trick needed to recycle
139 if (pos >= buffer.Length) {
140 entry = new LocatedToken ();
144 entry = new LocatedToken ();
153 entry.column = column;
158 // Used for token not required by expression evaluator
160 [Conditional ("FULL_AST")]
161 public void CreateOptional (SourceFile file, int row, int col, ref object token)
163 token = Create (file, row, col);
167 public enum PreprocessorDirective
173 If = 3 | RequiresArgument,
175 Elif = 5 | RequiresArgument,
177 Define = 7 | RequiresArgument,
178 Undef = 8 | RequiresArgument,
181 Pragma = 11 | CustomArgumentsParsing,
182 Line = 12 | CustomArgumentsParsing,
184 CustomArgumentsParsing = 1 << 10,
185 RequiresArgument = 1 << 11
188 readonly SeekableStreamReader reader;
189 readonly CompilationSourceFile source_file;
190 readonly CompilerContext context;
191 readonly Report Report;
194 SourceFile current_source;
195 Location hidden_block_start;
201 readonly int tab_size;
202 bool handle_get_set = false;
203 bool handle_remove_add = false;
205 bool lambda_arguments_parsing;
206 List<Location> escaped_identifiers;
207 int parsing_generic_less_than;
208 readonly bool doc_processing;
209 readonly LocatedTokenBuffer ltb;
212 // Used mainly for parser optimizations. Some expressions for instance
213 // can appear only in block (including initializer, base initializer)
216 public int parsing_block;
217 internal bool query_parsing;
220 // When parsing type only, useful for ambiguous nullable types
222 public int parsing_type;
225 // Set when parsing generic declaration (type or method header)
227 public bool parsing_generic_declaration;
228 public bool parsing_generic_declaration_doc;
231 // The value indicates that we have not reach any declaration or
234 public int parsing_declaration;
236 public bool parsing_attribute_section;
238 public bool parsing_modifiers;
240 public bool parsing_catch_when;
242 int parsing_string_interpolation;
243 int string_interpolation_section;
244 Stack<bool> parsing_string_interpolation_quoted;
246 public bool parsing_interpolation_format;
249 // The special characters to inject on streams to run the unit parser
250 // in the special expression mode. Using private characters from
251 // Plane Sixteen (U+100000 to U+10FFFD)
253 // This character is only tested just before the tokenizer is about to report
254 // an error; So on the regular operation mode, this addition will have no
255 // impact on the tokenizer's performance.
258 public const int EvalStatementParserCharacter = 0x100000;
259 public const int EvalCompilationUnitParserCharacter = 0x100001;
260 public const int EvalUsingDeclarationsParserCharacter = 0x100002;
261 public const int DocumentationXref = 0x100003;
263 const int UnicodeLS = 0x2028;
264 const int UnicodePS = 0x2029;
267 // XML documentation buffer. The save point is used to divide
268 // comments on types and comments on members.
270 StringBuilder xml_comment_buffer;
273 // See comment on XmlCommentState enumeration.
275 XmlCommentState xml_doc_state = XmlCommentState.Allowed;
278 // Whether tokens have been seen on this line
280 bool tokens_seen = false;
283 // Set to true once the GENERATE_COMPLETION token has bee
284 // returned. This helps produce one GENERATE_COMPLETION,
285 // as many COMPLETE_COMPLETION as necessary to complete the
286 // AST tree and one final EOF.
291 // Whether a token has been seen on the file
292 // This is needed because `define' is not allowed to be used
293 // after a token has been seen.
300 static readonly KeywordEntry<int>[][] keywords;
301 static readonly KeywordEntry<PreprocessorDirective>[][] keywords_preprocessor;
302 static readonly HashSet<string> keyword_strings;
303 static readonly NumberStyles styles;
304 static readonly NumberFormatInfo csharp_format_info;
307 static readonly char[] pragma_warning = "warning".ToCharArray ();
308 static readonly char[] pragma_warning_disable = "disable".ToCharArray ();
309 static readonly char[] pragma_warning_restore = "restore".ToCharArray ();
310 static readonly char[] pragma_checksum = "checksum".ToCharArray ();
311 static readonly char[] line_hidden = "hidden".ToCharArray ();
312 static readonly char[] line_default = "default".ToCharArray ();
314 static readonly char[] simple_whitespaces = new char[] { ' ', '\t' };
316 public bool PropertyParsing {
317 get { return handle_get_set; }
318 set { handle_get_set = value; }
321 public bool EventParsing {
322 get { return handle_remove_add; }
323 set { handle_remove_add = value; }
326 public bool ConstraintsParsing {
327 get { return handle_where; }
328 set { handle_where = value; }
331 public XmlCommentState doc_state {
332 get { return xml_doc_state; }
334 if (value == XmlCommentState.Allowed) {
335 check_incorrect_doc_comment ();
336 reset_doc_comment ();
338 xml_doc_state = value;
343 // This is used to trigger completion generation on the parser
344 public bool CompleteOnEOF;
346 void AddEscapedIdentifier (Location loc)
348 if (escaped_identifiers == null)
349 escaped_identifiers = new List<Location> ();
351 escaped_identifiers.Add (loc);
354 public bool IsEscapedIdentifier (ATypeNameExpression name)
356 return escaped_identifiers != null && escaped_identifiers.Contains (name.Location);
360 // Values for the associated token returned
362 internal int putback_char; // Used by repl only
368 const int TAKING = 1;
369 const int ELSE_SEEN = 4;
370 const int PARENT_TAKING = 8;
371 const int REGION = 16;
374 // pre-processor if stack state:
378 public const int MaxIdentifierLength = 512;
379 public const int MaxNumberLength = 512;
381 readonly char[] id_builder;
382 readonly Dictionary<char[], string>[] identifiers;
383 readonly char[] number_builder;
386 char[] value_builder = new char[64];
395 // This is used when the tokenizer needs to save
396 // the current position as it needs to do some parsing
397 // on its own to deamiguate a token in behalf of the
400 Stack<Position> position_stack = new Stack<Position> (2);
407 public Location hidden;
408 public int putback_char;
409 public int previous_col;
410 public Stack<int> ifstack;
411 public int parsing_generic_less_than;
412 public int current_token;
414 public int parsing_string_interpolation;
415 public int string_interpolation_section;
416 public Stack<bool> parsing_string_interpolation_quoted;
418 public Position (Tokenizer t)
420 position = t.reader.Position;
422 ref_line = t.ref_line;
424 hidden = t.hidden_block_start;
425 putback_char = t.putback_char;
426 previous_col = t.previous_col;
427 if (t.ifstack != null && t.ifstack.Count != 0) {
428 // There is no simple way to clone Stack<T> all
429 // methods reverse the order
430 var clone = t.ifstack.ToArray ();
431 Array.Reverse (clone);
432 ifstack = new Stack<int> (clone);
434 parsing_generic_less_than = t.parsing_generic_less_than;
435 string_interpolation_section = t.string_interpolation_section;
436 current_token = t.current_token;
438 parsing_string_interpolation = t.parsing_string_interpolation;
439 string_interpolation_section = t.string_interpolation_section;
440 if (t.parsing_string_interpolation_quoted != null && t.parsing_string_interpolation_quoted.Count != 0) {
441 var clone = t.parsing_string_interpolation_quoted.ToArray ();
442 Array.Reverse (clone);
443 parsing_string_interpolation_quoted = new Stack<bool> (clone);
448 public Tokenizer (SeekableStreamReader input, CompilationSourceFile file, ParserSession session, Report report)
450 this.source_file = file;
451 this.context = file.Compiler;
452 this.current_source = file.SourceFile;
453 this.identifiers = session.Identifiers;
454 this.id_builder = session.IDBuilder;
455 this.number_builder = session.NumberBuilder;
456 this.ltb = new LocatedTokenBuffer (session.LocatedTokens);
457 this.Report = report;
463 xml_comment_buffer = new StringBuilder ();
464 doc_processing = context.Settings.DocumentationFile != null;
466 tab_size = context.Settings.TabSize;
469 public void PushPosition ()
471 position_stack.Push (new Position (this));
474 public void PopPosition ()
476 Position p = position_stack.Pop ();
478 reader.Position = p.position;
479 ref_line = p.ref_line;
482 hidden_block_start = p.hidden;
483 putback_char = p.putback_char;
484 previous_col = p.previous_col;
486 parsing_generic_less_than = p.parsing_generic_less_than;
487 parsing_string_interpolation = p.parsing_string_interpolation;
488 parsing_string_interpolation_quoted = p.parsing_string_interpolation_quoted;
489 current_token = p.current_token;
493 // Do not reset the position, ignore it.
494 public void DiscardPosition ()
496 position_stack.Pop ();
499 static void AddKeyword (string kw, int token)
501 keyword_strings.Add (kw);
503 AddKeyword (keywords, kw, token);
506 static void AddPreprocessorKeyword (string kw, PreprocessorDirective directive)
508 AddKeyword (keywords_preprocessor, kw, directive);
511 static void AddKeyword<T> (KeywordEntry<T>[][] keywords, string kw, T token)
513 int length = kw.Length;
514 if (keywords[length] == null) {
515 keywords[length] = new KeywordEntry<T>['z' - '_' + 1];
518 int char_index = kw[0] - '_';
519 var kwe = keywords[length][char_index];
521 keywords[length][char_index] = new KeywordEntry<T> (kw, token);
525 while (kwe.Next != null) {
529 kwe.Next = new KeywordEntry<T> (kw, token);
537 keyword_strings = new HashSet<string> ();
539 // 11 is the length of the longest keyword for now
540 keywords = new KeywordEntry<int>[11][];
542 AddKeyword ("__arglist", Token.ARGLIST);
543 AddKeyword ("__makeref", Token.MAKEREF);
544 AddKeyword ("__reftype", Token.REFTYPE);
545 AddKeyword ("__refvalue", Token.REFVALUE);
546 AddKeyword ("abstract", Token.ABSTRACT);
547 AddKeyword ("as", Token.AS);
548 AddKeyword ("add", Token.ADD);
549 AddKeyword ("base", Token.BASE);
550 AddKeyword ("bool", Token.BOOL);
551 AddKeyword ("break", Token.BREAK);
552 AddKeyword ("byte", Token.BYTE);
553 AddKeyword ("case", Token.CASE);
554 AddKeyword ("catch", Token.CATCH);
555 AddKeyword ("char", Token.CHAR);
556 AddKeyword ("checked", Token.CHECKED);
557 AddKeyword ("class", Token.CLASS);
558 AddKeyword ("const", Token.CONST);
559 AddKeyword ("continue", Token.CONTINUE);
560 AddKeyword ("decimal", Token.DECIMAL);
561 AddKeyword ("default", Token.DEFAULT);
562 AddKeyword ("delegate", Token.DELEGATE);
563 AddKeyword ("do", Token.DO);
564 AddKeyword ("double", Token.DOUBLE);
565 AddKeyword ("else", Token.ELSE);
566 AddKeyword ("enum", Token.ENUM);
567 AddKeyword ("event", Token.EVENT);
568 AddKeyword ("explicit", Token.EXPLICIT);
569 AddKeyword ("extern", Token.EXTERN);
570 AddKeyword ("false", Token.FALSE);
571 AddKeyword ("finally", Token.FINALLY);
572 AddKeyword ("fixed", Token.FIXED);
573 AddKeyword ("float", Token.FLOAT);
574 AddKeyword ("for", Token.FOR);
575 AddKeyword ("foreach", Token.FOREACH);
576 AddKeyword ("goto", Token.GOTO);
577 AddKeyword ("get", Token.GET);
578 AddKeyword ("if", Token.IF);
579 AddKeyword ("implicit", Token.IMPLICIT);
580 AddKeyword ("in", Token.IN);
581 AddKeyword ("int", Token.INT);
582 AddKeyword ("interface", Token.INTERFACE);
583 AddKeyword ("internal", Token.INTERNAL);
584 AddKeyword ("is", Token.IS);
585 AddKeyword ("lock", Token.LOCK);
586 AddKeyword ("long", Token.LONG);
587 AddKeyword ("namespace", Token.NAMESPACE);
588 AddKeyword ("new", Token.NEW);
589 AddKeyword ("null", Token.NULL);
590 AddKeyword ("object", Token.OBJECT);
591 AddKeyword ("operator", Token.OPERATOR);
592 AddKeyword ("out", Token.OUT);
593 AddKeyword ("override", Token.OVERRIDE);
594 AddKeyword ("params", Token.PARAMS);
595 AddKeyword ("private", Token.PRIVATE);
596 AddKeyword ("protected", Token.PROTECTED);
597 AddKeyword ("public", Token.PUBLIC);
598 AddKeyword ("readonly", Token.READONLY);
599 AddKeyword ("ref", Token.REF);
600 AddKeyword ("remove", Token.REMOVE);
601 AddKeyword ("return", Token.RETURN);
602 AddKeyword ("sbyte", Token.SBYTE);
603 AddKeyword ("sealed", Token.SEALED);
604 AddKeyword ("set", Token.SET);
605 AddKeyword ("short", Token.SHORT);
606 AddKeyword ("sizeof", Token.SIZEOF);
607 AddKeyword ("stackalloc", Token.STACKALLOC);
608 AddKeyword ("static", Token.STATIC);
609 AddKeyword ("string", Token.STRING);
610 AddKeyword ("struct", Token.STRUCT);
611 AddKeyword ("switch", Token.SWITCH);
612 AddKeyword ("this", Token.THIS);
613 AddKeyword ("throw", Token.THROW);
614 AddKeyword ("true", Token.TRUE);
615 AddKeyword ("try", Token.TRY);
616 AddKeyword ("typeof", Token.TYPEOF);
617 AddKeyword ("uint", Token.UINT);
618 AddKeyword ("ulong", Token.ULONG);
619 AddKeyword ("unchecked", Token.UNCHECKED);
620 AddKeyword ("unsafe", Token.UNSAFE);
621 AddKeyword ("ushort", Token.USHORT);
622 AddKeyword ("using", Token.USING);
623 AddKeyword ("virtual", Token.VIRTUAL);
624 AddKeyword ("void", Token.VOID);
625 AddKeyword ("volatile", Token.VOLATILE);
626 AddKeyword ("while", Token.WHILE);
627 AddKeyword ("partial", Token.PARTIAL);
628 AddKeyword ("where", Token.WHERE);
631 AddKeyword ("from", Token.FROM);
632 AddKeyword ("join", Token.JOIN);
633 AddKeyword ("on", Token.ON);
634 AddKeyword ("equals", Token.EQUALS);
635 AddKeyword ("select", Token.SELECT);
636 AddKeyword ("group", Token.GROUP);
637 AddKeyword ("by", Token.BY);
638 AddKeyword ("let", Token.LET);
639 AddKeyword ("orderby", Token.ORDERBY);
640 AddKeyword ("ascending", Token.ASCENDING);
641 AddKeyword ("descending", Token.DESCENDING);
642 AddKeyword ("into", Token.INTO);
644 // Contextual async keywords
645 AddKeyword ("async", Token.ASYNC);
646 AddKeyword ("await", Token.AWAIT);
648 // Contextual filter catch keyword
649 AddKeyword ("when", Token.WHEN);
651 keywords_preprocessor = new KeywordEntry<PreprocessorDirective>[10][];
653 AddPreprocessorKeyword ("region", PreprocessorDirective.Region);
654 AddPreprocessorKeyword ("endregion", PreprocessorDirective.Endregion);
655 AddPreprocessorKeyword ("if", PreprocessorDirective.If);
656 AddPreprocessorKeyword ("endif", PreprocessorDirective.Endif);
657 AddPreprocessorKeyword ("elif", PreprocessorDirective.Elif);
658 AddPreprocessorKeyword ("else", PreprocessorDirective.Else);
659 AddPreprocessorKeyword ("define", PreprocessorDirective.Define);
660 AddPreprocessorKeyword ("undef", PreprocessorDirective.Undef);
661 AddPreprocessorKeyword ("error", PreprocessorDirective.Error);
662 AddPreprocessorKeyword ("warning", PreprocessorDirective.Warning);
663 AddPreprocessorKeyword ("pragma", PreprocessorDirective.Pragma);
664 AddPreprocessorKeyword ("line", PreprocessorDirective.Line);
666 csharp_format_info = NumberFormatInfo.InvariantInfo;
667 styles = NumberStyles.Float;
670 int GetKeyword (char[] id, int id_len)
673 // Keywords are stored in an array of arrays grouped by their
674 // length and then by the first character
676 if (id_len >= keywords.Length || keywords [id_len] == null)
679 int first_index = id [0] - '_';
680 if (first_index > 'z' - '_')
683 var kwe = keywords [id_len] [first_index];
690 for (int i = 1; i < id_len; ++i) {
691 if (id [i] != kwe.Value [i]) {
697 } while (res == 0 && kwe != null);
711 if (!handle_remove_add)
715 if (parsing_declaration == 0)
716 res = Token.EXTERN_ALIAS;
719 if (peek_token () == Token.COLON) {
721 res = Token.DEFAULT_COLON;
725 if (current_token != Token.CATCH && !parsing_catch_when)
729 if (!(handle_where && current_token != Token.COLON) && !query_parsing)
734 // A query expression is any expression that starts with `from identifier'
735 // followed by any token except ; , =
737 if (!query_parsing) {
738 if (lambda_arguments_parsing || parsing_block == 0) {
744 // HACK: to disable generics micro-parser, because PushPosition does not
745 // store identifiers array
746 parsing_generic_less_than = 1;
748 case Token.IDENTIFIER:
761 next_token = xtoken ();
762 if (next_token == Token.SEMICOLON || next_token == Token.COMMA || next_token == Token.EQUALS || next_token == Token.ASSIGN)
765 res = Token.FROM_FIRST;
766 query_parsing = true;
767 if (context.Settings.Version <= LanguageVersion.ISO_2)
768 Report.FeatureIsNotAvailable (context, Location, "query expressions");
771 Expression.Error_VoidInvalidInTheContext (Location, Report);
775 // HACK: A token is not a keyword so we need to restore identifiers buffer
776 // which has been overwritten before we grabbed the identifier
777 id_builder [0] = 'f'; id_builder [1] = 'r'; id_builder [2] = 'o'; id_builder [3] = 'm';
791 case Token.ASCENDING:
792 case Token.DESCENDING:
799 case Token.NAMESPACE:
800 // TODO: some explanation needed
801 check_incorrect_doc_comment ();
802 parsing_modifiers = false;
806 if (parsing_block > 0) {
811 // Save current position and parse next token.
814 next_token = token ();
815 bool ok = (next_token == Token.CLASS) ||
816 (next_token == Token.STRUCT) ||
817 (next_token == Token.INTERFACE) ||
818 (next_token == Token.VOID);
823 if (next_token == Token.VOID) {
824 if (context.Settings.Version <= LanguageVersion.ISO_2)
825 Report.FeatureIsNotAvailable (context, Location, "partial methods");
826 } else if (context.Settings.Version == LanguageVersion.ISO_1)
827 Report.FeatureIsNotAvailable (context, Location, "partial types");
832 if (next_token < Token.LAST_KEYWORD) {
833 Report.Error (267, Location,
834 "The `partial' modifier can be used only immediately before `class', `struct', `interface', or `void' keyword");
838 // HACK: A token is not a keyword so we need to restore identifiers buffer
839 // which has been overwritten before we grabbed the identifier
851 if (parsing_modifiers) {
853 // Skip attributes section or constructor called async
855 if (parsing_attribute_section || peek_token () == Token.OPEN_PARENS) {
860 } else if (parsing_block > 0) {
861 switch (peek_token ()) {
863 case Token.OPEN_PARENS_LAMBDA:
866 case Token.IDENTIFIER:
869 if (xtoken () != Token.ARROW) {
877 // peek_token could overwrite id_buffer
878 id_builder [0] = 'a'; id_builder [1] = 's'; id_builder [2] = 'y'; id_builder [3] = 'n'; id_builder [4] = 'c';
886 if (res == Token.ASYNC && context.Settings.Version <= LanguageVersion.V_4) {
887 Report.FeatureIsNotAvailable (context, Location, "asynchronous functions");
893 if (parsing_block == 0)
903 static PreprocessorDirective GetPreprocessorDirective (char[] id, int id_len)
906 // Keywords are stored in an array of arrays grouped by their
907 // length and then by the first character
909 if (id_len >= keywords_preprocessor.Length || keywords_preprocessor[id_len] == null)
910 return PreprocessorDirective.Invalid;
912 int first_index = id[0] - '_';
913 if (first_index > 'z' - '_')
914 return PreprocessorDirective.Invalid;
916 var kwe = keywords_preprocessor[id_len][first_index];
918 return PreprocessorDirective.Invalid;
920 PreprocessorDirective res = PreprocessorDirective.Invalid;
923 for (int i = 1; i < id_len; ++i) {
924 if (id[i] != kwe.Value[i]) {
930 } while (res == PreprocessorDirective.Invalid && kwe != null);
935 public Location Location {
937 return new Location (current_source, ref_line, col);
941 static bool is_identifier_start_character (int c)
943 if ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_')
949 return is_identifier_start_character_slow_part ((char) c);
952 static bool is_identifier_part_character (char c)
954 if (c >= 'a' && c <= 'z')
957 if (c >= 'A' && c <= 'Z')
960 if (c == '_' || (c >= '0' && c <= '9'))
966 return is_identifier_part_character_slow_part (c);
969 static bool is_identifier_start_character_slow_part (char c)
971 switch (Char.GetUnicodeCategory (c)) {
972 case UnicodeCategory.LetterNumber:
973 case UnicodeCategory.UppercaseLetter:
974 case UnicodeCategory.LowercaseLetter:
975 case UnicodeCategory.TitlecaseLetter:
976 case UnicodeCategory.ModifierLetter:
977 case UnicodeCategory.OtherLetter:
983 static bool is_identifier_part_character_slow_part (char c)
985 switch (Char.GetUnicodeCategory (c)) {
986 // connecting-character: A Unicode character of the class Pc
987 case UnicodeCategory.ConnectorPunctuation:
989 // combining-character: A Unicode character of classes Mn or Mc
990 case UnicodeCategory.NonSpacingMark:
991 case UnicodeCategory.SpacingCombiningMark:
993 // decimal-digit-character: A Unicode character of the class Nd
994 case UnicodeCategory.DecimalDigitNumber:
996 // plus is_identifier_start_character_slow_part
997 case UnicodeCategory.LetterNumber:
998 case UnicodeCategory.UppercaseLetter:
999 case UnicodeCategory.LowercaseLetter:
1000 case UnicodeCategory.TitlecaseLetter:
1001 case UnicodeCategory.ModifierLetter:
1002 case UnicodeCategory.OtherLetter:
1005 // formatting-character: A Unicode character of the class Cf
1006 case UnicodeCategory.Format:
1007 // csc bug compatibility which recognizes it as a whitespace
1014 public static bool IsKeyword (string s)
1016 return keyword_strings.Contains (s);
1020 // Open parens micro parser. Detects both lambda and cast ambiguity.
1022 int TokenizeOpenParens ()
1027 int bracket_level = 0;
1028 bool is_type = false;
1029 bool can_be_type = false;
1032 ptoken = current_token;
1035 switch (current_token) {
1036 case Token.CLOSE_PARENS:
1040 // Expression inside parens is lambda, (int i) =>
1042 if (current_token == Token.ARROW)
1043 return Token.OPEN_PARENS_LAMBDA;
1046 // Expression inside parens is single type, (int[])
1049 if (current_token == Token.SEMICOLON)
1050 return Token.OPEN_PARENS;
1052 return Token.OPEN_PARENS_CAST;
1056 // Expression is possible cast, look at next token, (T)null
1059 switch (current_token) {
1060 case Token.OPEN_PARENS:
1063 case Token.IDENTIFIER:
1067 case Token.DELEGATE:
1077 case Token.UNCHECKED:
1083 // These can be part of a member access
1098 return Token.OPEN_PARENS_CAST;
1101 return Token.OPEN_PARENS;
1104 case Token.DOUBLE_COLON:
1105 if (ptoken != Token.IDENTIFIER && ptoken != Token.OP_GENERICS_GT)
1110 case Token.IDENTIFIER:
1114 if (bracket_level == 0) {
1120 case Token.OP_GENERICS_LT:
1122 case Token.DOUBLE_COLON:
1124 if (bracket_level == 0)
1128 can_be_type = is_type = false;
1148 if (bracket_level == 0)
1153 if (bracket_level == 0) {
1154 bracket_level = 100;
1155 can_be_type = is_type = false;
1159 case Token.OP_GENERICS_LT:
1160 case Token.OPEN_BRACKET:
1161 if (bracket_level++ == 0)
1165 case Token.OP_GENERICS_GT:
1166 case Token.CLOSE_BRACKET:
1170 case Token.INTERR_NULLABLE:
1172 if (bracket_level == 0)
1178 can_be_type = is_type = false;
1182 return Token.OPEN_PARENS;
1187 public static bool IsValidIdentifier (string s)
1189 if (s == null || s.Length == 0)
1192 if (!is_identifier_start_character (s [0]))
1195 for (int i = 1; i < s.Length; i ++)
1196 if (! is_identifier_part_character (s [i]))
1202 bool parse_less_than (ref int genericDimension)
1205 int the_token = token ();
1206 if (the_token == Token.OPEN_BRACKET) {
1208 the_token = token ();
1209 if (the_token == Token.EOF)
1212 if (the_token == Token.CLOSE_BRACKET)
1215 the_token = token ();
1216 } else if (the_token == Token.IN || the_token == Token.OUT) {
1217 the_token = token ();
1219 switch (the_token) {
1220 case Token.IDENTIFIER:
1238 case Token.OP_GENERICS_GT:
1239 genericDimension = 1;
1247 the_token = token ();
1248 } while (the_token == Token.COMMA);
1250 if (the_token == Token.OP_GENERICS_GT) {
1260 the_token = token ();
1262 if (the_token == Token.OP_GENERICS_GT)
1264 else if (the_token == Token.COMMA || the_token == Token.DOT || the_token == Token.DOUBLE_COLON)
1266 else if (the_token == Token.INTERR_NULLABLE || the_token == Token.STAR)
1268 else if (the_token == Token.OP_GENERICS_LT) {
1269 if (!parse_less_than (ref genericDimension))
1272 } else if (the_token == Token.OPEN_BRACKET) {
1274 the_token = token ();
1275 if (the_token == Token.CLOSE_BRACKET)
1277 else if (the_token == Token.COMMA)
1278 goto rank_specifiers;
1285 public int peek_token ()
1290 the_token = token ();
1297 // Tonizes `?' using custom disambiguous rules to return one
1298 // of following tokens: INTERR_NULLABLE, OP_COALESCING, INTERR
1300 // Tricky expression looks like:
1302 // Foo ? a = x ? b : c;
1304 int TokenizePossibleNullableType ()
1306 if (parsing_block == 0 || parsing_type > 0)
1307 return Token.INTERR_NULLABLE;
1309 int d = peek_char ();
1312 return Token.OP_COALESCING;
1316 return Token.INTERR_OPERATOR;
1320 if (d == ',' || d == ';' || d == '>')
1321 return Token.INTERR_NULLABLE;
1322 if (d == '*' || (d >= '0' && d <= '9'))
1323 return Token.INTERR;
1327 current_token = Token.NONE;
1336 case Token.OPEN_BRACKET_EXPR:
1337 next_token = Token.INTERR_OPERATOR;
1345 case Token.INTERPOLATED_STRING:
1346 next_token = Token.INTERR;
1349 case Token.SEMICOLON:
1351 case Token.CLOSE_PARENS:
1352 case Token.OPEN_BRACKET:
1353 case Token.OP_GENERICS_GT:
1355 case Token.OP_COALESCING:
1357 next_token = Token.INTERR_NULLABLE;
1360 case Token.OPEN_PARENS:
1361 case Token.OPEN_PARENS_CAST:
1362 case Token.OPEN_PARENS_LAMBDA:
1367 case Token.OP_GENERICS_LT:
1368 case Token.OP_GENERICS_LT_DECL:
1369 case Token.GENERIC_DIMENSION:
1379 if (next_token == -1) {
1380 switch (xtoken ()) {
1382 case Token.SEMICOLON:
1383 case Token.OPEN_BRACE:
1385 next_token = Token.INTERR_NULLABLE;
1389 next_token = Token.INTERR;
1392 case Token.OPEN_PARENS:
1393 case Token.OPEN_PARENS_CAST:
1394 case Token.OPEN_PARENS_LAMBDA:
1398 case Token.OPEN_BRACKET:
1399 case Token.OPEN_BRACKET_EXPR:
1403 case Token.CLOSE_PARENS:
1407 case Token.OP_GENERICS_LT:
1408 case Token.OP_GENERICS_LT_DECL:
1409 case Token.GENERIC_DIMENSION:
1419 // All shorcuts failed, do it hard way
1421 while ((ntoken = xtoken ()) != Token.EOF) {
1423 case Token.OPEN_BRACE:
1426 case Token.OPEN_PARENS:
1427 case Token.OPEN_PARENS_CAST:
1428 case Token.OPEN_PARENS_LAMBDA:
1431 case Token.CLOSE_BRACE:
1434 case Token.OP_GENERICS_LT:
1435 case Token.OP_GENERICS_LT_DECL:
1436 case Token.GENERIC_DIMENSION:
1439 case Token.OPEN_BRACKET:
1440 case Token.OPEN_BRACKET_EXPR:
1443 case Token.CLOSE_BRACKET:
1446 case Token.CLOSE_PARENS:
1453 return Token.INTERR_NULLABLE;
1455 case Token.OP_GENERICS_GT:
1462 return Token.INTERR_NULLABLE;
1468 if (ntoken == Token.SEMICOLON)
1474 if (ntoken == Token.COMMA) {
1475 if (generics != 0 || brackets != 0)
1479 return Token.INTERR_NULLABLE;
1482 if (ntoken == Token.COLON) {
1483 if (++colons == interrs)
1488 if (ntoken == Token.INTERR) {
1494 next_token = colons != interrs && braces == 0 ? Token.INTERR_NULLABLE : Token.INTERR;
1503 bool decimal_digits (int c)
1506 bool seen_digits = false;
1509 if (number_pos == MaxNumberLength)
1510 Error_NumericConstantTooLong ();
1511 number_builder [number_pos++] = (char) c;
1515 // We use peek_char2, because decimal_digits needs to do a
1516 // 2-character look-ahead (5.ToString for example).
1518 while ((d = peek_char2 ()) != -1){
1519 if (d >= '0' && d <= '9'){
1520 if (number_pos == MaxNumberLength)
1521 Error_NumericConstantTooLong ();
1522 number_builder [number_pos++] = (char) d;
1532 static bool is_hex (int e)
1534 return (e >= '0' && e <= '9') || (e >= 'A' && e <= 'F') || (e >= 'a' && e <= 'f');
1537 static TypeCode real_type_suffix (int c)
1541 return TypeCode.Single;
1543 return TypeCode.Double;
1545 return TypeCode.Decimal;
1547 return TypeCode.Empty;
1551 ILiteralConstant integer_type_suffix (ulong ul, int c, Location loc)
1553 bool is_unsigned = false;
1554 bool is_long = false;
1557 bool scanning = true;
1570 // if we have not seen anything in between
1571 // report this error
1573 Report.Warning (78, 4, Location, "The `l' suffix is easily confused with the digit `1' (use `L' for clarity)");
1593 if (is_long && is_unsigned){
1594 return new ULongLiteral (context.BuiltinTypes, ul, loc);
1598 // uint if possible, or ulong else.
1600 if ((ul & 0xffffffff00000000) == 0)
1601 return new UIntLiteral (context.BuiltinTypes, (uint) ul, loc);
1603 return new ULongLiteral (context.BuiltinTypes, ul, loc);
1604 } else if (is_long){
1605 // long if possible, ulong otherwise
1606 if ((ul & 0x8000000000000000) != 0)
1607 return new ULongLiteral (context.BuiltinTypes, ul, loc);
1609 return new LongLiteral (context.BuiltinTypes, (long) ul, loc);
1611 // int, uint, long or ulong in that order
1612 if ((ul & 0xffffffff00000000) == 0){
1613 uint ui = (uint) ul;
1615 if ((ui & 0x80000000) != 0)
1616 return new UIntLiteral (context.BuiltinTypes, ui, loc);
1618 return new IntLiteral (context.BuiltinTypes, (int) ui, loc);
1620 if ((ul & 0x8000000000000000) != 0)
1621 return new ULongLiteral (context.BuiltinTypes, ul, loc);
1623 return new LongLiteral (context.BuiltinTypes, (long) ul, loc);
1629 // given `c' as the next char in the input decide whether
1630 // we need to convert to a special type, and then choose
1631 // the best representation for the integer
1633 ILiteralConstant adjust_int (int c, Location loc)
1636 if (number_pos > 9){
1637 ulong ul = (uint) (number_builder [0] - '0');
1639 for (int i = 1; i < number_pos; i++){
1640 ul = checked ((ul * 10) + ((uint)(number_builder [i] - '0')));
1643 return integer_type_suffix (ul, c, loc);
1645 uint ui = (uint) (number_builder [0] - '0');
1647 for (int i = 1; i < number_pos; i++){
1648 ui = checked ((ui * 10) + ((uint)(number_builder [i] - '0')));
1651 return integer_type_suffix (ui, c, loc);
1653 } catch (OverflowException) {
1654 Error_NumericConstantTooLong ();
1655 return new IntLiteral (context.BuiltinTypes, 0, loc);
1657 catch (FormatException) {
1658 Report.Error (1013, Location, "Invalid number");
1659 return new IntLiteral (context.BuiltinTypes, 0, loc);
1663 ILiteralConstant adjust_real (TypeCode t, Location loc)
1665 string s = new string (number_builder, 0, number_pos);
1666 const string error_details = "Floating-point constant is outside the range of type `{0}'";
1669 case TypeCode.Decimal:
1671 return new DecimalLiteral (context.BuiltinTypes, decimal.Parse (s, styles, csharp_format_info), loc);
1672 } catch (OverflowException) {
1673 Report.Error (594, Location, error_details, "decimal");
1674 return new DecimalLiteral (context.BuiltinTypes, 0, loc);
1676 case TypeCode.Single:
1678 return new FloatLiteral (context.BuiltinTypes, float.Parse (s, styles, csharp_format_info), loc);
1679 } catch (OverflowException) {
1680 Report.Error (594, Location, error_details, "float");
1681 return new FloatLiteral (context.BuiltinTypes, 0, loc);
1685 return new DoubleLiteral (context.BuiltinTypes, double.Parse (s, styles, csharp_format_info), loc);
1686 } catch (OverflowException) {
1687 Report.Error (594, loc, error_details, "double");
1688 return new DoubleLiteral (context.BuiltinTypes, 0, loc);
1693 ILiteralConstant handle_hex (Location loc)
1699 while ((d = peek_char ()) != -1){
1701 number_builder [number_pos++] = (char) d;
1707 string s = new String (number_builder, 0, number_pos);
1710 if (number_pos <= 8)
1711 ul = System.UInt32.Parse (s, NumberStyles.HexNumber);
1713 ul = System.UInt64.Parse (s, NumberStyles.HexNumber);
1715 return integer_type_suffix (ul, peek_char (), loc);
1716 } catch (OverflowException){
1717 Error_NumericConstantTooLong ();
1718 return new IntLiteral (context.BuiltinTypes, 0, loc);
1720 catch (FormatException) {
1721 Report.Error (1013, Location, "Invalid number");
1722 return new IntLiteral (context.BuiltinTypes, 0, loc);
1727 // Invoked if we know we have .digits or digits
1729 int is_number (int c, bool dotLead)
1731 ILiteralConstant res;
1734 int read_start = reader.Position - 1;
1737 // Caller did peek_char
1747 int peek = peek_char ();
1749 if (peek == 'x' || peek == 'X') {
1750 val = res = handle_hex (loc);
1752 res.ParsedValue = reader.ReadChars (read_start, reader.Position - 1);
1755 return Token.LITERAL;
1763 // We need to handle the case of
1764 // "1.1" vs "1.string" (LITERAL_FLOAT vs NUMBER DOT IDENTIFIER)
1766 bool is_real = false;
1771 if (decimal_digits ('.')){
1777 val = res = adjust_int (-1, loc);
1780 res.ParsedValue = reader.ReadChars (read_start, reader.Position - 1);
1782 return Token.LITERAL;
1786 if (c == 'e' || c == 'E'){
1789 if (number_pos == MaxNumberLength)
1790 Error_NumericConstantTooLong ();
1791 number_builder [number_pos++] = (char) c;
1795 if (number_pos == MaxNumberLength)
1796 Error_NumericConstantTooLong ();
1797 number_builder [number_pos++] = '+';
1799 } else if (c == '-') {
1800 if (number_pos == MaxNumberLength)
1801 Error_NumericConstantTooLong ();
1802 number_builder [number_pos++] = '-';
1805 if (number_pos == MaxNumberLength)
1806 Error_NumericConstantTooLong ();
1807 number_builder [number_pos++] = '+';
1814 var type = real_type_suffix (c);
1815 if (type == TypeCode.Empty && !is_real) {
1816 res = adjust_int (c, loc);
1820 if (type != TypeCode.Empty) {
1824 res = adjust_real (type, loc);
1830 var chars = reader.ReadChars (read_start, reader.Position - (type == TypeCode.Empty && c > 0 ? 1 : 0));
1831 if (chars[chars.Length - 1] == '\r')
1832 Array.Resize (ref chars, chars.Length - 1);
1833 res.ParsedValue = chars;
1836 return Token.LITERAL;
1840 // Accepts exactly count (4 or 8) hex, no more no less
1842 int getHex (int count, out int surrogate, out bool error)
1847 int top = count != -1 ? count : 4;
1852 for (i = 0; i < top; i++){
1855 if (c >= '0' && c <= '9')
1856 c = (int) c - (int) '0';
1857 else if (c >= 'A' && c <= 'F')
1858 c = (int) c - (int) 'A' + 10;
1859 else if (c >= 'a' && c <= 'f')
1860 c = (int) c - (int) 'a' + 10;
1866 total = (total * 16) + c;
1868 int p = peek_char ();
1871 if (!is_hex ((char)p))
1877 if (total > 0x0010FFFF) {
1882 if (total >= 0x00010000) {
1883 surrogate = ((total - 0x00010000) % 0x0400 + 0xDC00);
1884 total = ((total - 0x00010000) / 0x0400 + 0xD800);
1891 int escape (int c, out int surrogate)
1927 v = getHex (-1, out surrogate, out error);
1933 return EscapeUnicode (d, out surrogate);
1936 Report.Error (1009, Location, "Unrecognized escape sequence `\\{0}'", ((char)d).ToString ());
1945 int EscapeUnicode (int ch, out int surrogate)
1949 ch = getHex (8, out surrogate, out error);
1951 ch = getHex (4, out surrogate, out error);
1955 Report.Error (1009, Location, "Unrecognized escape sequence");
1963 if (putback_char != -1) {
1972 if (peek_char () == '\n') {
1978 } else if (x == '\n') {
1983 } else if (x >= UnicodeLS && x <= UnicodePS) {
1992 void advance_line ()
2002 if (putback_char == -1)
2003 putback_char = reader.Read ();
2004 return putback_char;
2009 if (putback_char != -1)
2010 return putback_char;
2011 return reader.Peek ();
2014 public void putback (int c)
2016 if (putback_char != -1) {
2017 throw new InternalErrorException (string.Format ("Secondary putback [{0}] putting back [{1}] is not allowed", (char)putback_char, (char) c), Location);
2020 if (c == '\n' || col == 0 || (c >= UnicodeLS && c <= UnicodePS)) {
2021 // It won't happen though.
2031 public bool advance ()
2033 return peek_char () != -1 || CompleteOnEOF;
2036 public Object Value {
2042 public Object value ()
2049 current_token = xtoken ();
2050 return current_token;
2053 int TokenizePreprocessorKeyword (out int c)
2055 // skip over white space
2058 } while (c == ' ' || c == '\t');
2062 while (c != -1 && c >= 'a' && c <= 'z') {
2063 id_builder[pos++] = (char) c;
2066 int peek = peek_char ();
2067 if (peek == 'U' || peek == 'u') {
2069 c = EscapeUnicode (c, out surrogate);
2070 if (surrogate != 0) {
2071 if (is_identifier_part_character ((char) c)) {
2072 id_builder[pos++] = (char) c;
2083 PreprocessorDirective get_cmd_arg (out string arg)
2087 tokens_seen = false;
2090 var cmd = GetPreprocessorDirective (id_builder, TokenizePreprocessorKeyword (out c));
2092 if ((cmd & PreprocessorDirective.CustomArgumentsParsing) != 0)
2095 // skip over white space
2096 while (c == ' ' || c == '\t')
2099 int has_identifier_argument = (int)(cmd & PreprocessorDirective.RequiresArgument);
2102 while (c != -1 && c != '\n' && c != UnicodeLS && c != UnicodePS) {
2103 if (c == '\\' && has_identifier_argument >= 0) {
2104 if (has_identifier_argument != 0) {
2105 has_identifier_argument = 1;
2107 int peek = peek_char ();
2108 if (peek == 'U' || peek == 'u') {
2110 c = EscapeUnicode (c, out surrogate);
2111 if (surrogate != 0) {
2112 if (is_identifier_part_character ((char) c)) {
2113 if (pos == value_builder.Length)
2114 Array.Resize (ref value_builder, pos * 2);
2116 value_builder[pos++] = (char) c;
2122 has_identifier_argument = -1;
2124 } else if (c == '/' && peek_char () == '/') {
2126 // Eat single-line comments
2133 if (pos == value_builder.Length)
2134 Array.Resize (ref value_builder, pos * 2);
2136 value_builder[pos++] = (char) c;
2141 if (pos > MaxIdentifierLength)
2142 arg = new string (value_builder, 0, pos);
2144 arg = InternIdentifier (value_builder, pos);
2146 // Eat any trailing whitespaces
2147 arg = arg.Trim (simple_whitespaces);
2154 // Handles the #line directive
2156 bool PreProcessLine ()
2158 Location loc = Location;
2162 int length = TokenizePreprocessorKeyword (out c);
2163 if (length == line_default.Length) {
2164 if (!IsTokenIdentifierEqual (line_default))
2167 current_source = source_file.SourceFile;
2168 if (!hidden_block_start.IsNull) {
2169 current_source.RegisterHiddenScope (hidden_block_start, loc);
2170 hidden_block_start = Location.Null;
2177 if (length == line_hidden.Length) {
2178 if (!IsTokenIdentifierEqual (line_hidden))
2181 if (hidden_block_start.IsNull)
2182 hidden_block_start = loc;
2187 if (length != 0 || c < '0' || c > '9') {
2189 // Eat any remaining characters to continue parsing on next line
2195 int new_line = TokenizeNumber (c);
2198 // Eat any remaining characters to continue parsing on next line
2201 return new_line != 0;
2206 // skip over white space
2209 } while (c == ' ' || c == '\t');
2210 } else if (c == '"') {
2214 if (c != '\n' && c != '/' && c != '"' && c != UnicodeLS && c != UnicodePS) {
2216 // Eat any remaining characters to continue parsing on next line
2220 Report.Error (1578, loc, "Filename, single-line comment or end-of-line expected");
2224 string new_file_name = null;
2226 new_file_name = TokenizeFileName (ref c);
2228 // skip over white space
2229 while (c == ' ' || c == '\t') {
2234 if (c == '\n' || c == UnicodeLS || c == UnicodePS) {
2236 } else if (c == '/') {
2237 ReadSingleLineComment ();
2240 // Eat any remaining characters to continue parsing on next line
2244 Error_EndLineExpected ();
2248 if (new_file_name != null) {
2249 current_source = context.LookupFile (source_file, new_file_name);
2250 source_file.AddIncludeFile (current_source);
2253 if (!hidden_block_start.IsNull) {
2254 current_source.RegisterHiddenScope (hidden_block_start, loc);
2255 hidden_block_start = Location.Null;
2258 ref_line = new_line;
2263 // Handles #define and #undef
2265 void PreProcessDefinition (bool is_define, string ident, bool caller_is_taking)
2267 if (ident.Length == 0 || ident == "true" || ident == "false"){
2268 Report.Error (1001, Location, "Missing identifier to pre-processor directive");
2272 if (ident.IndexOfAny (simple_whitespaces) != -1){
2273 Error_EndLineExpected ();
2277 if (!is_identifier_start_character (ident [0]))
2278 Report.Error (1001, Location, "Identifier expected: {0}", ident);
2280 foreach (char c in ident.Substring (1)){
2281 if (!is_identifier_part_character (c)){
2282 Report.Error (1001, Location, "Identifier expected: {0}", ident);
2287 if (!caller_is_taking)
2294 if (context.Settings.IsConditionalSymbolDefined (ident))
2297 source_file.AddDefine (ident);
2302 source_file.AddUndefine (ident);
2306 byte read_hex (out bool error)
2309 int c = get_char ();
2311 if ((c >= '0') && (c <= '9'))
2312 total = (int) c - (int) '0';
2313 else if ((c >= 'A') && (c <= 'F'))
2314 total = (int) c - (int) 'A' + 10;
2315 else if ((c >= 'a') && (c <= 'f'))
2316 total = (int) c - (int) 'a' + 10;
2325 if ((c >= '0') && (c <= '9'))
2326 total += (int) c - (int) '0';
2327 else if ((c >= 'A') && (c <= 'F'))
2328 total += (int) c - (int) 'A' + 10;
2329 else if ((c >= 'a') && (c <= 'f'))
2330 total += (int) c - (int) 'a' + 10;
2337 return (byte) total;
2341 // Parses #pragma checksum
2343 bool ParsePragmaChecksum ()
2346 // The syntax is ` "foo.txt" "{guid}" "hash"'
2348 // guid is predefined hash algorithm guid {406ea660-64cf-4c82-b6f0-42d48172a799} for md5
2350 int c = get_char ();
2355 string file_name = TokenizeFileName (ref c);
2357 // TODO: Any white-spaces count
2361 SourceFile file = context.LookupFile (source_file, file_name);
2363 if (get_char () != '"' || get_char () != '{')
2367 byte[] guid_bytes = new byte [16];
2370 for (; i < 4; i++) {
2371 guid_bytes [i] = read_hex (out error);
2376 if (get_char () != '-')
2379 for (; i < 10; i++) {
2380 guid_bytes [i] = read_hex (out error);
2384 guid_bytes [i++] = read_hex (out error);
2388 if (get_char () != '-')
2392 for (; i < 16; i++) {
2393 guid_bytes [i] = read_hex (out error);
2398 if (get_char () != '}' || get_char () != '"')
2401 // TODO: Any white-spaces count
2406 if (get_char () != '"')
2409 // Any length of checksum
2410 List<byte> checksum_bytes = new List<byte> (16);
2412 var checksum_location = Location;
2414 while (c != '"' && c != -1) {
2415 checksum_bytes.Add (read_hex (out error));
2423 ReadSingleLineComment ();
2424 } else if (get_char () != '"') {
2428 if (context.Settings.GenerateDebugInfo) {
2429 var chsum = checksum_bytes.ToArray ();
2431 if (file.HasChecksum) {
2432 if (!ArrayComparer.IsEqual (file.Checksum, chsum)) {
2433 // TODO: Report.SymbolRelatedToPreviousError
2434 Report.Warning (1697, 1, checksum_location, "Different checksum values specified for file `{0}'", file.Name);
2438 file.SetChecksum (guid_bytes, chsum);
2439 current_source.AutoGenerated = true;
2445 bool IsTokenIdentifierEqual (char[] identifier)
2447 for (int i = 0; i < identifier.Length; ++i) {
2448 if (identifier[i] != id_builder[i])
2455 bool ScanClosingInterpolationBrace ()
2462 var c = reader.Read ();
2471 if (str_quote % 2 == 1) {
2477 } while (res == null);
2483 int TokenizeNumber (int value)
2487 decimal_digits (value);
2488 uint ui = (uint) (number_builder[0] - '0');
2491 for (int i = 1; i < number_pos; i++) {
2492 ui = checked ((ui * 10) + ((uint) (number_builder[i] - '0')));
2496 } catch (OverflowException) {
2497 Error_NumericConstantTooLong ();
2502 string TokenizeFileName (ref int c)
2504 var string_builder = new StringBuilder ();
2505 while (c != -1 && c != '\n' && c != UnicodeLS && c != UnicodePS) {
2512 string_builder.Append ((char) c);
2515 if (string_builder.Length == 0) {
2516 Report.Warning (1709, 1, Location, "Filename specified for preprocessor directive is empty");
2520 return string_builder.ToString ();
2523 int TokenizePragmaWarningIdentifier (ref int c, ref bool identifier)
2526 if ((c >= '0' && c <= '9') || is_identifier_start_character (c)) {
2529 if (c >= '0' && c <= '9') {
2531 number = TokenizeNumber (c);
2535 if (c != ' ' && c != '\t' && c != ',' && c != '\n' && c != -1 && c != UnicodeLS && c != UnicodePS) {
2536 return ReadPragmaWarningComment (c);
2540 // LAMESPEC v6: No spec what identifier really is in this context, it seems keywords are allowed too
2544 id_builder [pos++] = (char)c;
2545 while (c < MaxIdentifierLength) {
2547 id_builder [pos] = (char)c;
2549 if (c >= '0' && c <= '9') {
2550 if (pos == 6 && id_builder [0] == 'C' && id_builder [1] == 'S') {
2551 // Recognize CSXXXX as C# XXXX warning
2554 for (int i = 0; i < 4; ++i) {
2555 var ch = id_builder [i + 2];
2556 if (ch < '0' || ch > '9') {
2561 number += (ch - '0') * pow;
2565 } else if ((c < 'a' || c > 'z') && (c < 'A' || c > 'Z') && c != '_') {
2578 // skip over white space
2579 while (c == ' ' || c == '\t')
2586 // skip over white space
2587 while (c == ' ' || c == '\t')
2593 return ReadPragmaWarningComment (c);
2596 int ReadPragmaWarningComment (int c)
2599 ReadSingleLineComment ();
2601 Report.Warning (1692, 1, Location, "Invalid number");
2603 // Read everything till the end of the line or file
2610 void ReadToEndOfLine ()
2615 } while (c != -1 && c != '\n' && c != UnicodeLS && c != UnicodePS);
2618 void ReadSingleLineComment ()
2620 if (peek_char () != '/')
2621 Report.Warning (1696, 1, Location, "Single-line comment or end-of-line expected");
2623 // Read everything till the end of the line or file
2628 /// Handles #pragma directive
2630 void ParsePragmaDirective ()
2633 int length = TokenizePreprocessorKeyword (out c);
2634 if (length == pragma_warning.Length && IsTokenIdentifierEqual (pragma_warning)) {
2635 length = TokenizePreprocessorKeyword (out c);
2638 // #pragma warning disable
2639 // #pragma warning restore
2641 if (length == pragma_warning_disable.Length) {
2642 bool disable = IsTokenIdentifierEqual (pragma_warning_disable);
2643 if (disable || IsTokenIdentifierEqual (pragma_warning_restore)) {
2644 // skip over white space
2645 while (c == ' ' || c == '\t')
2650 if (c == '\n' || c == '/' || c == UnicodeLS || c == UnicodePS) {
2652 ReadSingleLineComment ();
2655 // Disable/Restore all warnings
2658 Report.RegisterWarningRegion (loc).WarningDisable (loc.Row);
2660 Report.RegisterWarningRegion (loc).WarningEnable (loc.Row);
2664 // Disable/Restore a warning or group of warnings
2668 bool identifier = false;
2669 code = TokenizePragmaWarningIdentifier (ref c, ref identifier);
2672 // no-op, custom warnings cannot occur in mcs
2673 } else if (disable) {
2674 Report.RegisterWarningRegion (loc).WarningDisable (loc, code, context.Report);
2676 Report.RegisterWarningRegion (loc).WarningEnable (loc, code, context);
2679 } while (code >= 0 && c != '\n' && c != -1 && c != UnicodeLS && c != UnicodePS);
2686 Report.Warning (1634, 1, Location, "Expected disable or restore");
2688 // Eat any remaining characters on the line
2697 if (length == pragma_checksum.Length && IsTokenIdentifierEqual (pragma_checksum)) {
2698 if (c != ' ' || !ParsePragmaChecksum ()) {
2699 Report.Warning (1695, 1, Location,
2700 "Invalid #pragma checksum syntax. Expected \"filename\" \"{XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX}\" \"XXXX...\"");
2706 Report.Warning (1633, 1, Location, "Unrecognized #pragma directive");
2708 // Eat any remaining characters on the line
2712 bool eval_val (string s)
2719 return source_file.IsConditionalDefined (s);
2722 bool pp_primary (ref string s)
2731 s = s.Substring (1);
2732 bool val = pp_expr (ref s, false);
2733 if (s.Length > 0 && s [0] == ')'){
2734 s = s.Substring (1);
2737 Error_InvalidDirective ();
2741 if (is_identifier_start_character (c)){
2747 if (is_identifier_part_character (c)){
2751 bool v = eval_val (s.Substring (0, j));
2752 s = s.Substring (j);
2755 bool vv = eval_val (s);
2760 Error_InvalidDirective ();
2764 bool pp_unary (ref string s)
2771 if (len > 1 && s [1] == '='){
2772 Error_InvalidDirective ();
2775 s = s.Substring (1);
2776 return ! pp_primary (ref s);
2778 return pp_primary (ref s);
2780 Error_InvalidDirective ();
2785 bool pp_eq (ref string s)
2787 bool va = pp_unary (ref s);
2793 if (len > 2 && s [1] == '='){
2794 s = s.Substring (2);
2795 return va == pp_unary (ref s);
2797 Error_InvalidDirective ();
2800 } else if (s [0] == '!' && len > 1 && s [1] == '='){
2801 s = s.Substring (2);
2803 return va != pp_unary (ref s);
2812 bool pp_and (ref string s)
2814 bool va = pp_eq (ref s);
2820 if (len > 2 && s [1] == '&'){
2821 s = s.Substring (2);
2822 return (va & pp_and (ref s));
2824 Error_InvalidDirective ();
2833 // Evaluates an expression for `#if' or `#elif'
2835 bool pp_expr (ref string s, bool isTerm)
2837 bool va = pp_and (ref s);
2844 if (len > 2 && s [1] == '|'){
2845 s = s.Substring (2);
2846 return va | pp_expr (ref s, isTerm);
2848 Error_InvalidDirective ();
2853 Error_EndLineExpected ();
2861 bool eval (string s)
2863 bool v = pp_expr (ref s, true);
2872 void Error_NumericConstantTooLong ()
2874 Report.Error (1021, Location, "Integral constant is too large");
2877 void Error_InvalidDirective ()
2879 Report.Error (1517, Location, "Invalid preprocessor directive");
2882 void Error_UnexpectedDirective (string extra)
2886 "Unexpected processor directive ({0})", extra);
2889 void Error_TokensSeen ()
2891 Report.Error (1032, Location,
2892 "Cannot define or undefine preprocessor symbols after first token in file");
2895 void Eror_WrongPreprocessorLocation ()
2897 Report.Error (1040, Location,
2898 "Preprocessor directives must appear as the first non-whitespace character on a line");
2901 void Error_EndLineExpected ()
2903 Report.Error (1025, Location, "Single-line comment or end-of-line expected");
2907 // Raises a warning when tokenizer found documentation comment
2908 // on unexpected place
2910 void WarningMisplacedComment (Location loc)
2912 if (doc_state != XmlCommentState.Error) {
2913 doc_state = XmlCommentState.Error;
2914 Report.Warning (1587, 2, loc, "XML comment is not placed on a valid language element");
2919 // if true, then the code continues processing the code
2920 // if false, the code stays in a loop until another directive is
2922 // When caller_is_taking is false we ignore all directives except the ones
2923 // which can help us to identify where the #if block ends
2924 bool ParsePreprocessingDirective (bool caller_is_taking)
2927 bool region_directive = false;
2929 var directive = get_cmd_arg (out arg);
2932 // The first group of pre-processing instructions is always processed
2934 switch (directive) {
2935 case PreprocessorDirective.Region:
2936 region_directive = true;
2938 goto case PreprocessorDirective.If;
2940 case PreprocessorDirective.Endregion:
2941 if (ifstack == null || ifstack.Count == 0){
2942 Error_UnexpectedDirective ("no #region for this #endregion");
2945 int pop = ifstack.Pop ();
2947 if ((pop & REGION) == 0)
2948 Report.Error (1027, Location, "Expected `#endif' directive");
2950 return caller_is_taking;
2952 case PreprocessorDirective.If:
2953 if (ifstack == null)
2954 ifstack = new Stack<int> (2);
2956 int flags = region_directive ? REGION : 0;
2957 if (ifstack.Count == 0){
2958 flags |= PARENT_TAKING;
2960 int state = ifstack.Peek ();
2961 if ((state & TAKING) != 0) {
2962 flags |= PARENT_TAKING;
2966 if (eval (arg) && caller_is_taking) {
2967 ifstack.Push (flags | TAKING);
2970 ifstack.Push (flags);
2973 case PreprocessorDirective.Endif:
2974 if (ifstack == null || ifstack.Count == 0){
2975 Error_UnexpectedDirective ("no #if for this #endif");
2978 pop = ifstack.Pop ();
2980 if ((pop & REGION) != 0)
2981 Report.Error (1038, Location, "#endregion directive expected");
2983 if (arg.Length != 0) {
2984 Error_EndLineExpected ();
2987 if (ifstack.Count == 0)
2990 int state = ifstack.Peek ();
2991 return (state & TAKING) != 0;
2994 case PreprocessorDirective.Elif:
2995 if (ifstack == null || ifstack.Count == 0){
2996 Error_UnexpectedDirective ("no #if for this #elif");
2999 int state = ifstack.Pop ();
3001 if ((state & REGION) != 0) {
3002 Report.Error (1038, Location, "#endregion directive expected");
3006 if ((state & ELSE_SEEN) != 0){
3007 Error_UnexpectedDirective ("#elif not valid after #else");
3011 if ((state & TAKING) != 0) {
3016 if (eval (arg) && ((state & PARENT_TAKING) != 0)){
3017 ifstack.Push (state | TAKING);
3021 ifstack.Push (state);
3025 case PreprocessorDirective.Else:
3026 if (ifstack == null || ifstack.Count == 0){
3027 Error_UnexpectedDirective ("no #if for this #else");
3030 int state = ifstack.Peek ();
3032 if ((state & REGION) != 0) {
3033 Report.Error (1038, Location, "#endregion directive expected");
3037 if ((state & ELSE_SEEN) != 0){
3038 Error_UnexpectedDirective ("#else within #else");
3044 if (arg.Length != 0) {
3045 Error_EndLineExpected ();
3050 if ((state & PARENT_TAKING) != 0) {
3051 ret = (state & TAKING) == 0;
3059 ifstack.Push (state | ELSE_SEEN);
3063 case PreprocessorDirective.Define:
3064 if (any_token_seen){
3065 if (caller_is_taking)
3066 Error_TokensSeen ();
3067 return caller_is_taking;
3069 PreProcessDefinition (true, arg, caller_is_taking);
3070 return caller_is_taking;
3072 case PreprocessorDirective.Undef:
3073 if (any_token_seen){
3074 if (caller_is_taking)
3075 Error_TokensSeen ();
3076 return caller_is_taking;
3078 PreProcessDefinition (false, arg, caller_is_taking);
3079 return caller_is_taking;
3081 case PreprocessorDirective.Invalid:
3082 Report.Error (1024, Location, "Wrong preprocessor directive");
3087 // These are only processed if we are in a `taking' block
3089 if (!caller_is_taking)
3093 case PreprocessorDirective.Error:
3094 Report.Error (1029, Location, "#error: '{0}'", arg);
3097 case PreprocessorDirective.Warning:
3098 Report.Warning (1030, 1, Location, "#warning: `{0}'", arg);
3101 case PreprocessorDirective.Pragma:
3102 if (context.Settings.Version == LanguageVersion.ISO_1) {
3103 Report.FeatureIsNotAvailable (context, Location, "#pragma");
3106 ParsePragmaDirective ();
3109 case PreprocessorDirective.Line:
3110 Location loc = Location;
3111 if (!PreProcessLine ())
3112 Report.Error (1576, loc, "The line number specified for #line directive is missing or invalid");
3114 return caller_is_taking;
3117 throw new NotImplementedException (directive.ToString ());
3120 int consume_string (bool quoted)
3124 Location start_location = Location;
3126 start_location = start_location - 1;
3129 int reader_pos = reader.Position;
3133 // Cannot use get_char because of \r in quoted strings
3134 if (putback_char != -1) {
3144 if (quoted && peek_char () == '"') {
3145 if (pos == value_builder.Length)
3146 Array.Resize (ref value_builder, pos * 2);
3148 value_builder[pos++] = (char) c;
3153 ILiteralConstant res = new StringLiteral (context.BuiltinTypes, CreateStringFromBuilder (pos), start_location);
3156 res.ParsedValue = quoted ?
3157 reader.ReadChars (reader_pos - 2, reader.Position - 1) :
3158 reader.ReadChars (reader_pos - 1, reader.Position);
3161 return Token.LITERAL;
3164 if (c == '\n' || c == UnicodeLS || c == UnicodePS) {
3166 Report.Error (1010, Location, "Newline in constant");
3170 // Don't add \r to string literal
3171 if (pos > 1 && value_builder [pos - 1] == '\r')
3174 val = new StringLiteral (context.BuiltinTypes, new string (value_builder, 0, pos), start_location);
3175 return Token.LITERAL;
3179 } else if (c == '\\' && !quoted) {
3182 c = escape (c, out surrogate);
3185 if (surrogate != 0) {
3186 if (pos == value_builder.Length)
3187 Array.Resize (ref value_builder, pos * 2);
3189 value_builder[pos++] = (char) c;
3192 } else if (c == -1) {
3193 Report.Error (1039, Location, "Unterminated string literal");
3199 if (pos == value_builder.Length)
3200 Array.Resize (ref value_builder, pos * 2);
3202 value_builder[pos++] = (char) c;
3206 private int consume_identifier (int s)
3208 int res = consume_identifier (s, false);
3210 if (doc_state == XmlCommentState.Allowed)
3211 doc_state = XmlCommentState.NotAllowed;
3216 int consume_identifier (int c, bool quoted)
3219 // This method is very performance sensitive. It accounts
3220 // for approximately 25% of all parser time
3230 c = escape (c, out surrogate);
3231 if (quoted || is_identifier_start_character (c)) {
3232 // it's added bellow
3233 } else if (surrogate != 0) {
3234 id_builder [pos++] = (char)c;
3237 Report.Error (1056, Location, "Unexpected character `\\{0}'", c.ToString ("x4"));
3242 id_builder [pos++] = (char) c;
3248 if ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_' || (c >= '0' && c <= '9')) {
3249 id_builder [pos++] = (char) c;
3256 c = escape (c, out surrogate);
3257 if (is_identifier_part_character ((char) c))
3258 id_builder[pos++] = (char) c;
3259 else if (surrogate != 0) {
3263 // TODO: Probably need more whitespace characters
3268 Report.Error (1056, Location, "Unexpected character `\\{0}'", c.ToString ("x4"));
3275 } else if (is_identifier_part_character_slow_part ((char) c)) {
3276 id_builder [pos++] = (char) c;
3283 } catch (IndexOutOfRangeException) {
3284 Report.Error (645, Location, "Identifier too long (limit is 512 chars)");
3292 // Optimization: avoids doing the keyword lookup
3293 // on uppercase letters
3295 if (id_builder [0] >= '_' && !quoted) {
3296 int keyword = GetKeyword (id_builder, pos);
3297 if (keyword != -1) {
3298 val = ltb.Create (keyword == Token.AWAIT ? "await" : null, current_source, ref_line, column);
3303 string s = InternIdentifier (id_builder, pos);
3304 val = ltb.Create (s, current_source, ref_line, column);
3305 if (quoted && parsing_attribute_section)
3306 AddEscapedIdentifier (((LocatedToken) val).Location);
3308 return Token.IDENTIFIER;
3311 string InternIdentifier (char[] charBuffer, int length)
3314 // Keep identifiers in an array of hashtables to avoid needless
3317 var identifiers_group = identifiers[length];
3319 if (identifiers_group != null) {
3320 if (identifiers_group.TryGetValue (charBuffer, out s)) {
3324 // TODO: this should be number of files dependant
3325 // corlib compilation peaks at 1000 and System.Core at 150
3326 int capacity = length > 20 ? 10 : 100;
3327 identifiers_group = new Dictionary<char[], string> (capacity, new IdentifiersComparer (length));
3328 identifiers[length] = identifiers_group;
3331 char[] chars = new char[length];
3332 Array.Copy (charBuffer, chars, length);
3334 s = new string (charBuffer, 0, length);
3335 identifiers_group.Add (chars, s);
3339 public int xtoken ()
3341 if (parsing_interpolation_format) {
3342 return TokenizeInterpolationFormat ();
3347 // Whether we have seen comments on the current line
3348 bool comments_seen = false;
3349 while ((c = get_char ()) != -1) {
3352 col = ((col - 1 + tab_size) / tab_size) * tab_size;
3360 case 0xFEFF: // Ignore BOM anywhere in the file
3363 /* This is required for compatibility with .NET
3365 if (peek_char () == 0xBB) {
3368 if (get_char () == 0xBF)
3376 return consume_identifier (c);
3379 val = ltb.Create (current_source, ref_line, col);
3381 if (parsing_string_interpolation > 0)
3382 ++string_interpolation_section;
3384 return Token.OPEN_BRACE;
3386 if (parsing_string_interpolation > 0) {
3387 if (string_interpolation_section == 0) {
3388 --parsing_string_interpolation;
3390 if (parsing_string_interpolation_quoted != null && parsing_string_interpolation_quoted.Count > 0) {
3391 quoted = parsing_string_interpolation_quoted.Pop ();
3396 return TokenizeInterpolatedString (quoted);
3399 --string_interpolation_section;
3402 val = ltb.Create (current_source, ref_line, col);
3403 return Token.CLOSE_BRACE;
3405 // To block doccomment inside attribute declaration.
3406 if (doc_state == XmlCommentState.Allowed)
3407 doc_state = XmlCommentState.NotAllowed;
3409 val = ltb.Create (current_source, ref_line, col);
3411 if (parsing_block == 0 || lambda_arguments_parsing)
3412 return Token.OPEN_BRACKET;
3414 int next = peek_char ();
3418 return Token.OPEN_BRACKET;
3428 next = peek_token ();
3429 if (next == Token.COMMA || next == Token.CLOSE_BRACKET)
3430 return Token.OPEN_BRACKET;
3432 return Token.OPEN_BRACKET_EXPR;
3434 return Token.OPEN_BRACKET_EXPR;
3437 ltb.CreateOptional (current_source, ref_line, col, ref val);
3438 return Token.CLOSE_BRACKET;
3440 val = ltb.Create (current_source, ref_line, col);
3442 // An expression versions of parens can appear in block context only
3444 if (parsing_block != 0 && !lambda_arguments_parsing) {
3447 // Optmize most common case where we know that parens
3450 switch (current_token) {
3451 case Token.IDENTIFIER:
3460 case Token.DELEGATE:
3461 case Token.OP_GENERICS_GT:
3462 return Token.OPEN_PARENS;
3465 // Optimize using peek
3466 int xx = peek_char ();
3473 return Token.OPEN_PARENS;
3476 lambda_arguments_parsing = true;
3478 d = TokenizeOpenParens ();
3480 lambda_arguments_parsing = false;
3484 return Token.OPEN_PARENS;
3486 ltb.CreateOptional (current_source, ref_line, col, ref val);
3487 return Token.CLOSE_PARENS;
3489 ltb.CreateOptional (current_source, ref_line, col, ref val);
3492 ltb.CreateOptional (current_source, ref_line, col, ref val);
3493 return Token.SEMICOLON;
3495 val = ltb.Create (current_source, ref_line, col);
3498 val = ltb.Create (current_source, ref_line, col);
3499 return TokenizePossibleNullableType ();
3501 val = ltb.Create (current_source, ref_line, col);
3502 if (parsing_generic_less_than++ > 0)
3503 return Token.OP_GENERICS_LT;
3505 return TokenizeLessThan ();
3508 val = ltb.Create (current_source, ref_line, col);
3516 if (parsing_generic_less_than > 1 || (parsing_generic_less_than == 1 && d != '>')) {
3517 parsing_generic_less_than--;
3518 return Token.OP_GENERICS_GT;
3527 return Token.OP_SHIFT_RIGHT_ASSIGN;
3529 return Token.OP_SHIFT_RIGHT;
3535 val = ltb.Create (current_source, ref_line, col);
3539 } else if (d == '=') {
3540 d = Token.OP_ADD_ASSIGN;
3548 val = ltb.Create (current_source, ref_line, col);
3552 } else if (d == '=')
3553 d = Token.OP_SUB_ASSIGN;
3563 val = ltb.Create (current_source, ref_line, col);
3564 if (peek_char () == '='){
3571 val = ltb.Create (current_source, ref_line, col);
3582 return Token.ASSIGN;
3585 val = ltb.Create (current_source, ref_line, col);
3589 return Token.OP_AND;
3593 return Token.OP_AND_ASSIGN;
3595 return Token.BITWISE_AND;
3598 val = ltb.Create (current_source, ref_line, col);
3606 return Token.OP_OR_ASSIGN;
3608 return Token.BITWISE_OR;
3611 val = ltb.Create (current_source, ref_line, col);
3612 if (peek_char () == '='){
3614 return Token.OP_MULT_ASSIGN;
3621 val = ltb.Create (current_source, ref_line, col);
3623 return Token.OP_DIV_ASSIGN;
3626 // Handle double-slash comments.
3628 if (parsing_string_interpolation > 0) {
3629 Report.Error (8077, Location, "A single-line comment may not be used in an interpolated string");
3634 if (doc_processing) {
3635 if (peek_char () == '/') {
3637 // Don't allow ////.
3638 if ((d = peek_char ()) != '/') {
3639 if (doc_state == XmlCommentState.Allowed)
3640 handle_one_line_xml_comment ();
3641 else if (doc_state == XmlCommentState.NotAllowed)
3642 WarningMisplacedComment (Location - 3);
3645 if (xml_comment_buffer.Length > 0)
3646 doc_state = XmlCommentState.NotAllowed;
3652 any_token_seen |= tokens_seen;
3653 tokens_seen = false;
3654 comments_seen = false;
3656 } else if (d == '*'){
3658 bool docAppend = false;
3659 if (doc_processing && peek_char () == '*') {
3661 // But when it is /**/, just do nothing.
3662 if (peek_char () == '/') {
3666 if (doc_state == XmlCommentState.Allowed)
3668 else if (doc_state == XmlCommentState.NotAllowed) {
3669 WarningMisplacedComment (Location - 2);
3673 int current_comment_start = 0;
3675 current_comment_start = xml_comment_buffer.Length;
3676 xml_comment_buffer.Append (Environment.NewLine);
3679 while ((d = get_char ()) != -1){
3680 if (d == '*' && peek_char () == '/'){
3682 comments_seen = true;
3686 xml_comment_buffer.Append ((char) d);
3688 if (d == '\n' || d == UnicodeLS || d == UnicodePS){
3689 any_token_seen |= tokens_seen;
3690 tokens_seen = false;
3692 // Reset 'comments_seen' just to be consistent.
3693 // It doesn't matter either way, here.
3695 comments_seen = false;
3699 Report.Error (1035, Location, "End-of-file found, '*/' expected");
3702 update_formatted_doc_comment (current_comment_start);
3705 val = ltb.Create (current_source, ref_line, col);
3709 val = ltb.Create (current_source, ref_line, col);
3710 if (peek_char () == '='){
3712 return Token.OP_MOD_ASSIGN;
3714 return Token.PERCENT;
3717 val = ltb.Create (current_source, ref_line, col);
3718 if (peek_char () == '='){
3720 return Token.OP_XOR_ASSIGN;
3722 return Token.CARRET;
3725 val = ltb.Create (current_source, ref_line, col);
3726 if (peek_char () == ':') {
3728 return Token.DOUBLE_COLON;
3732 case '0': case '1': case '2': case '3': case '4':
3733 case '5': case '6': case '7': case '8': case '9':
3735 return is_number (c, false);
3737 case '\n': // white space
3740 any_token_seen |= tokens_seen;
3741 tokens_seen = false;
3742 comments_seen = false;
3748 if (d >= '0' && d <= '9')
3749 return is_number (c, true);
3751 ltb.CreateOptional (current_source, ref_line, col, ref val);
3755 if (tokens_seen || comments_seen) {
3756 Eror_WrongPreprocessorLocation ();
3760 if (ParsePreprocessingDirective (true))
3763 bool directive_expected = false;
3764 while ((c = get_char ()) != -1) {
3766 directive_expected = true;
3767 } else if (!directive_expected) {
3768 // TODO: Implement comment support for disabled code and uncomment this code
3770 // Eror_WrongPreprocessorLocation ();
3771 // return Token.ERROR;
3776 if (c == ' ' || c == '\t' || c == '\n' || c == '\f' || c == '\v' || c == UnicodeLS || c == UnicodePS)
3780 if (ParsePreprocessingDirective (false))
3783 directive_expected = false;
3787 tokens_seen = false;
3794 if (parsing_string_interpolation > 0 && !ScanClosingInterpolationBrace ()) {
3795 parsing_string_interpolation = 0;
3796 Report.Error (8076, Location, "Missing close delimiter `}' for interpolated expression");
3797 val = new StringLiteral (context.BuiltinTypes, "", Location);
3798 return Token.INTERPOLATED_STRING_END;
3801 return consume_string (false);
3804 return TokenizeBackslash ();
3810 return consume_string (true);
3813 if (is_identifier_start_character (c)){
3814 return consume_identifier (c, true);
3817 Report.Error (1646, Location, "Keyword, identifier, or string expected after verbatim specifier: @");
3821 switch (peek_char ()) {
3824 return TokenizeInterpolatedString (false);
3827 if (peek_char () == '"') {
3829 return TokenizeInterpolatedString (true);
3836 case EvalStatementParserCharacter:
3837 return Token.EVAL_STATEMENT_PARSER;
3838 case EvalCompilationUnitParserCharacter:
3839 return Token.EVAL_COMPILATION_UNIT_PARSER;
3840 case EvalUsingDeclarationsParserCharacter:
3841 return Token.EVAL_USING_DECLARATIONS_UNIT_PARSER;
3842 case DocumentationXref:
3843 return Token.DOC_SEE;
3846 if (is_identifier_start_character (c)) {
3848 return consume_identifier (c);
3851 if (char.IsWhiteSpace ((char) c))
3854 Report.Error (1056, Location, "Unexpected character `{0}'", ((char) c).ToString ());
3859 return Token.COMPLETE_COMPLETION;
3862 return Token.GENERATE_COMPLETION;
3869 int TokenizeBackslash ()
3872 int read_start = reader.Position;
3874 Location start_location = Location;
3875 int c = get_char ();
3878 val = new CharLiteral (context.BuiltinTypes, (char) c, start_location);
3879 Report.Error (1011, start_location, "Empty character literal");
3880 return Token.LITERAL;
3883 if (c == '\n' || c == UnicodeLS || c == UnicodePS) {
3884 Report.Error (1010, start_location, "Newline in constant");
3889 c = escape (c, out d);
3893 throw new NotImplementedException ();
3895 ILiteralConstant res = new CharLiteral (context.BuiltinTypes, (char) c, start_location);
3900 Report.Error (1012, start_location, "Too many characters in character literal");
3902 // Try to recover, read until newline or next "'"
3903 while ((c = get_char ()) != -1) {
3904 if (c == '\n' || c == '\'' || c == UnicodeLS || c == UnicodePS)
3910 res.ParsedValue = reader.ReadChars (read_start - 1, reader.Position);
3913 return Token.LITERAL;
3916 int TokenizeLessThan ()
3920 // Save current position and parse next token.
3922 int generic_dimension = 0;
3923 if (parse_less_than (ref generic_dimension)) {
3924 if (parsing_generic_declaration && (parsing_generic_declaration_doc || token () != Token.DOT)) {
3925 d = Token.OP_GENERICS_LT_DECL;
3927 if (generic_dimension > 0) {
3928 val = generic_dimension;
3930 return Token.GENERIC_DIMENSION;
3933 d = Token.OP_GENERICS_LT;
3940 parsing_generic_less_than = 0;
3949 return Token.OP_SHIFT_LEFT_ASSIGN;
3951 return Token.OP_SHIFT_LEFT;
3961 int TokenizeInterpolatedString (bool quoted)
3964 var start_location = Location;
3967 var ch = get_char ();
3970 if (quoted && peek_char () == '"') {
3975 val = new StringLiteral (context.BuiltinTypes, CreateStringFromBuilder (pos), start_location);
3976 return Token.INTERPOLATED_STRING_END;
3978 if (peek_char () == '{') {
3979 value_builder [pos++] = (char)ch;
3984 ++parsing_string_interpolation;
3986 if (parsing_string_interpolation_quoted == null)
3987 parsing_string_interpolation_quoted = new Stack<bool> ();
3990 if (parsing_string_interpolation_quoted != null) {
3991 parsing_string_interpolation_quoted.Push (quoted);
3994 val = new StringLiteral (context.BuiltinTypes, CreateStringFromBuilder (pos), start_location);
3995 return Token.INTERPOLATED_STRING;
4002 ch = escape (ch, out surrogate);
4006 if (ch == '{' || ch == '}') {
4007 Report.Error (8087, Location, "A `{0}' character may only be escaped by doubling `{0}{0}' in an interpolated string", ((char) ch).ToString ());
4010 if (surrogate != 0) {
4011 if (pos == value_builder.Length)
4012 Array.Resize (ref value_builder, pos * 2);
4014 if (pos == value_builder.Length)
4015 Array.Resize (ref value_builder, pos * 2);
4017 value_builder [pos++] = (char)ch;
4027 if (pos == value_builder.Length)
4028 Array.Resize (ref value_builder, pos * 2);
4030 value_builder[pos++] = (char) ch;
4034 int TokenizeInterpolationFormat ()
4039 var ch = get_char ();
4048 Report.Error (8089, Location, "Empty interpolated expression format specifier");
4049 } else if (Array.IndexOf (simple_whitespaces, value_builder [pos - 1]) >= 0) {
4050 Report.Error (8088, Location, "A interpolated expression format specifier may not contain trailing whitespace");
4053 val = CreateStringFromBuilder (pos);
4054 return Token.LITERAL;
4062 ch = escape (ch, out surrogate);
4066 if (ch == '{' || ch == '}') {
4067 Report.Error (8087, Location, "A `{0}' character may only be escaped by doubling `{0}{0}' in an interpolated string", ((char) ch).ToString ());
4070 if (surrogate != 0) {
4071 if (pos == value_builder.Length)
4072 Array.Resize (ref value_builder, pos * 2);
4074 value_builder [pos++] = (char)ch;
4084 value_builder[pos++] = (char) ch;
4088 string CreateStringFromBuilder (int pos)
4091 return string.Empty;
4093 return InternIdentifier (value_builder, pos);
4095 return new string (value_builder, 0, pos);
4099 // Handles one line xml comment
4101 private void handle_one_line_xml_comment ()
4104 while ((c = peek_char ()) != -1 && c != '\n' && c != '\r') {
4105 xml_comment_buffer.Append ((char) get_char ());
4107 if (c == '\r' || c == '\n')
4108 xml_comment_buffer.Append (Environment.NewLine);
4112 // Remove heading "*" in Javadoc-like xml documentation.
4114 private void update_formatted_doc_comment (int current_comment_start)
4116 int length = xml_comment_buffer.Length - current_comment_start;
4117 string [] lines = xml_comment_buffer.ToString (
4118 current_comment_start,
4119 length).Replace ("\r", "").Split ('\n');
4121 // The first line starts with /**, thus it is not target
4122 // for the format check.
4123 for (int i = 1; i < lines.Length; i++) {
4124 string s = lines [i];
4125 int idx = s.IndexOf ('*');
4128 if (i < lines.Length - 1)
4132 head = s.Substring (0, idx);
4133 foreach (char c in head)
4136 lines [i] = s.Substring (idx + 1);
4138 xml_comment_buffer.Remove (current_comment_start, length);
4139 xml_comment_buffer.Insert (current_comment_start, String.Join (Environment.NewLine, lines));
4143 // Checks if there was incorrect doc comments and raise
4146 public void check_incorrect_doc_comment ()
4148 if (xml_comment_buffer.Length > 0)
4149 WarningMisplacedComment (Location);
4153 // Consumes the saved xml comment lines (if any)
4154 // as for current target member or type.
4156 public string consume_doc_comment ()
4158 if (xml_comment_buffer.Length > 0) {
4159 string ret = xml_comment_buffer.ToString ();
4160 reset_doc_comment ();
4166 void reset_doc_comment ()
4168 xml_comment_buffer.Length = 0;
4171 public void cleanup ()
4173 if (ifstack != null && ifstack.Count >= 1) {
4174 int state = ifstack.Pop ();
4175 if ((state & REGION) != 0)
4176 Report.Error (1038, Location, "#endregion directive expected");
4178 Report.Error (1027, Location, "Expected `#endif' directive");
4184 // Indicates whether it accepts XML documentation or not.
4186 public enum XmlCommentState {
4187 // comment is allowed in this state.
4189 // comment is not allowed in this state.
4191 // once comments appeared when it is NotAllowed, then the
4192 // state is changed to it, until the state is changed to