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;
19 using System.Globalization;
20 using System.Reflection;
25 /// Tokenizer for C# source code.
28 public class Tokenizer : yyParser.yyInput
32 public readonly int Token;
33 public KeywordEntry Next;
34 public readonly char[] Value;
36 public KeywordEntry (string value, int token)
38 this.Value = value.ToCharArray ();
43 sealed class IdentifiersComparer : IEqualityComparer<char[]>
47 public IdentifiersComparer (int length)
52 public bool Equals (char[] x, char[] y)
54 for (int i = 0; i < length; ++i)
61 public int GetHashCode (char[] obj)
64 for (int i = 0; i < length; ++i)
65 h = (h << 5) - h + obj [i];
72 // This class has to be used in the parser only, it reuses token
73 // details after each parse
75 public class LocatedToken
80 static LocatedToken[] buffer;
83 private LocatedToken ()
87 public static LocatedToken Create (int row, int column)
89 return Create (null, row, column);
92 public static LocatedToken Create (string value, int row, int column)
95 // TODO: I am not very happy about the logic but it's the best
96 // what I could come up with for now.
97 // Ideally we should be using just tiny buffer (256 elements) which
98 // is enough to hold all details for currect stack and recycle elements
99 // poped from the stack but there is a trick needed to recycle
103 if (pos >= buffer.Length) {
104 entry = new LocatedToken ();
106 entry = buffer [pos];
108 entry = new LocatedToken ();
109 buffer [pos] = entry;
116 entry.column = column;
121 // Used for token not required by expression evaluator
123 public static LocatedToken CreateOptional (int row, int col)
126 return Create (row, col);
131 public static void Initialize ()
134 buffer = new LocatedToken [10000];
138 public Location Location {
139 get { return new Location (row, column); }
142 public string Value {
143 get { return value; }
147 SeekableStreamReader reader;
149 CompilationUnit file_name;
150 CompilerContext context;
157 bool handle_get_set = false;
158 bool handle_remove_add = false;
159 bool handle_where = false;
160 bool handle_typeof = false;
161 bool lambda_arguments_parsing;
162 Location current_comment_location = Location.Null;
163 List<Location> escaped_identifiers;
164 int parsing_generic_less_than;
167 // Used mainly for parser optimizations. Some expressions for instance
168 // can appear only in block (including initializer, base initializer)
171 public int parsing_block;
172 internal bool query_parsing;
175 // When parsing type only, useful for ambiguous nullable types
177 public int parsing_type;
180 // Set when parsing generic declaration (type or method header)
182 public bool parsing_generic_declaration;
185 // The value indicates that we have not reach any declaration or
188 public int parsing_declaration;
191 // The special character to inject on streams to trigger the EXPRESSION_PARSE
192 // token to be returned. It just happens to be a Unicode character that
193 // would never be part of a program (can not be an identifier).
195 // This character is only tested just before the tokenizer is about to report
196 // an error; So on the regular operation mode, this addition will have no
197 // impact on the tokenizer's performance.
200 public const int EvalStatementParserCharacter = 0x2190; // Unicode Left Arrow
201 public const int EvalCompilationUnitParserCharacter = 0x2191; // Unicode Arrow
202 public const int EvalUsingDeclarationsParserCharacter = 0x2192; // Unicode Arrow
205 // XML documentation buffer. The save point is used to divide
206 // comments on types and comments on members.
208 StringBuilder xml_comment_buffer;
211 // See comment on XmlCommentState enumeration.
213 XmlCommentState xml_doc_state = XmlCommentState.Allowed;
216 // Whether tokens have been seen on this line
218 bool tokens_seen = false;
221 // Set to true once the GENERATE_COMPLETION token has bee
222 // returned. This helps produce one GENERATE_COMPLETION,
223 // as many COMPLETE_COMPLETION as necessary to complete the
224 // AST tree and one final EOF.
229 // Whether a token has been seen on the file
230 // This is needed because `define' is not allowed to be used
231 // after a token has been seen.
233 bool any_token_seen = false;
235 static readonly char[] simple_whitespaces = new char[] { ' ', '\t' };
237 public bool PropertyParsing {
238 get { return handle_get_set; }
239 set { handle_get_set = value; }
242 public bool EventParsing {
243 get { return handle_remove_add; }
244 set { handle_remove_add = value; }
247 public bool ConstraintsParsing {
248 get { return handle_where; }
249 set { handle_where = value; }
252 public bool TypeOfParsing {
253 get { return handle_typeof; }
254 set { handle_typeof = value; }
257 public XmlCommentState doc_state {
258 get { return xml_doc_state; }
260 if (value == XmlCommentState.Allowed) {
261 check_incorrect_doc_comment ();
262 reset_doc_comment ();
264 xml_doc_state = value;
269 // This is used to trigger completion generation on the parser
270 public bool CompleteOnEOF;
272 void AddEscapedIdentifier (Location loc)
274 if (escaped_identifiers == null)
275 escaped_identifiers = new List<Location> ();
277 escaped_identifiers.Add (loc);
280 public bool IsEscapedIdentifier (Location loc)
282 if (escaped_identifiers != null) {
283 foreach (Location lt in escaped_identifiers)
294 static KeywordEntry[][] keywords;
295 static Dictionary<string, object> keyword_strings; // TODO: HashSet
296 static NumberStyles styles;
297 static NumberFormatInfo csharp_format_info;
300 // Values for the associated token returned
302 internal int putback_char; // Used by repl only
308 const int TAKING = 1;
309 const int ELSE_SEEN = 4;
310 const int PARENT_TAKING = 8;
311 const int REGION = 16;
314 // pre-processor if stack state:
318 static System.Text.StringBuilder string_builder;
320 const int max_id_size = 512;
321 static char [] id_builder = new char [max_id_size];
323 public static Dictionary<char[], string>[] identifiers = new Dictionary<char[], string>[max_id_size + 1];
325 const int max_number_size = 512;
326 static char [] number_builder = new char [max_number_size];
327 static int number_pos;
329 static StringBuilder static_cmd_arg = new System.Text.StringBuilder ();
332 // Details about the error encoutered by the tokenizer
334 string error_details;
336 public string error {
338 return error_details;
349 // This is used when the tokenizer needs to save
350 // the current position as it needs to do some parsing
351 // on its own to deamiguate a token in behalf of the
354 Stack<Position> position_stack = new Stack<Position> (2);
362 public int putback_char;
363 public int previous_col;
364 public Stack<int> ifstack;
365 public int parsing_generic_less_than;
366 public int current_token;
369 public Position (Tokenizer t)
371 position = t.reader.Position;
373 ref_line = t.ref_line;
376 putback_char = t.putback_char;
377 previous_col = t.previous_col;
378 if (t.ifstack != null && t.ifstack.Count != 0) {
379 // There is no simple way to clone Stack<T> all
380 // methods reverse the order
381 var clone = t.ifstack.ToArray ();
382 Array.Reverse (clone);
383 ifstack = new Stack<int> (clone);
385 parsing_generic_less_than = t.parsing_generic_less_than;
386 current_token = t.current_token;
391 public void PushPosition ()
393 position_stack.Push (new Position (this));
396 public void PopPosition ()
398 Position p = position_stack.Pop ();
400 reader.Position = p.position;
401 ref_line = p.ref_line;
405 putback_char = p.putback_char;
406 previous_col = p.previous_col;
408 parsing_generic_less_than = p.parsing_generic_less_than;
409 current_token = p.current_token;
413 // Do not reset the position, ignore it.
414 public void DiscardPosition ()
416 position_stack.Pop ();
419 static void AddKeyword (string kw, int token)
421 keyword_strings.Add (kw, null);
423 int length = kw.Length;
424 if (keywords [length] == null) {
425 keywords [length] = new KeywordEntry ['z' - '_' + 1];
428 int char_index = kw [0] - '_';
429 KeywordEntry kwe = keywords [length] [char_index];
431 keywords [length] [char_index] = new KeywordEntry (kw, token);
435 while (kwe.Next != null) {
439 kwe.Next = new KeywordEntry (kw, token);
442 static void InitTokens ()
444 keyword_strings = new Dictionary<string, object> ();
446 // 11 is the length of the longest keyword for now
447 keywords = new KeywordEntry [11] [];
449 AddKeyword ("__arglist", Token.ARGLIST);
450 AddKeyword ("abstract", Token.ABSTRACT);
451 AddKeyword ("as", Token.AS);
452 AddKeyword ("add", Token.ADD);
453 AddKeyword ("base", Token.BASE);
454 AddKeyword ("bool", Token.BOOL);
455 AddKeyword ("break", Token.BREAK);
456 AddKeyword ("byte", Token.BYTE);
457 AddKeyword ("case", Token.CASE);
458 AddKeyword ("catch", Token.CATCH);
459 AddKeyword ("char", Token.CHAR);
460 AddKeyword ("checked", Token.CHECKED);
461 AddKeyword ("class", Token.CLASS);
462 AddKeyword ("const", Token.CONST);
463 AddKeyword ("continue", Token.CONTINUE);
464 AddKeyword ("decimal", Token.DECIMAL);
465 AddKeyword ("default", Token.DEFAULT);
466 AddKeyword ("delegate", Token.DELEGATE);
467 AddKeyword ("do", Token.DO);
468 AddKeyword ("double", Token.DOUBLE);
469 AddKeyword ("else", Token.ELSE);
470 AddKeyword ("enum", Token.ENUM);
471 AddKeyword ("event", Token.EVENT);
472 AddKeyword ("explicit", Token.EXPLICIT);
473 AddKeyword ("extern", Token.EXTERN);
474 AddKeyword ("false", Token.FALSE);
475 AddKeyword ("finally", Token.FINALLY);
476 AddKeyword ("fixed", Token.FIXED);
477 AddKeyword ("float", Token.FLOAT);
478 AddKeyword ("for", Token.FOR);
479 AddKeyword ("foreach", Token.FOREACH);
480 AddKeyword ("goto", Token.GOTO);
481 AddKeyword ("get", Token.GET);
482 AddKeyword ("if", Token.IF);
483 AddKeyword ("implicit", Token.IMPLICIT);
484 AddKeyword ("in", Token.IN);
485 AddKeyword ("int", Token.INT);
486 AddKeyword ("interface", Token.INTERFACE);
487 AddKeyword ("internal", Token.INTERNAL);
488 AddKeyword ("is", Token.IS);
489 AddKeyword ("lock", Token.LOCK);
490 AddKeyword ("long", Token.LONG);
491 AddKeyword ("namespace", Token.NAMESPACE);
492 AddKeyword ("new", Token.NEW);
493 AddKeyword ("null", Token.NULL);
494 AddKeyword ("object", Token.OBJECT);
495 AddKeyword ("operator", Token.OPERATOR);
496 AddKeyword ("out", Token.OUT);
497 AddKeyword ("override", Token.OVERRIDE);
498 AddKeyword ("params", Token.PARAMS);
499 AddKeyword ("private", Token.PRIVATE);
500 AddKeyword ("protected", Token.PROTECTED);
501 AddKeyword ("public", Token.PUBLIC);
502 AddKeyword ("readonly", Token.READONLY);
503 AddKeyword ("ref", Token.REF);
504 AddKeyword ("remove", Token.REMOVE);
505 AddKeyword ("return", Token.RETURN);
506 AddKeyword ("sbyte", Token.SBYTE);
507 AddKeyword ("sealed", Token.SEALED);
508 AddKeyword ("set", Token.SET);
509 AddKeyword ("short", Token.SHORT);
510 AddKeyword ("sizeof", Token.SIZEOF);
511 AddKeyword ("stackalloc", Token.STACKALLOC);
512 AddKeyword ("static", Token.STATIC);
513 AddKeyword ("string", Token.STRING);
514 AddKeyword ("struct", Token.STRUCT);
515 AddKeyword ("switch", Token.SWITCH);
516 AddKeyword ("this", Token.THIS);
517 AddKeyword ("throw", Token.THROW);
518 AddKeyword ("true", Token.TRUE);
519 AddKeyword ("try", Token.TRY);
520 AddKeyword ("typeof", Token.TYPEOF);
521 AddKeyword ("uint", Token.UINT);
522 AddKeyword ("ulong", Token.ULONG);
523 AddKeyword ("unchecked", Token.UNCHECKED);
524 AddKeyword ("unsafe", Token.UNSAFE);
525 AddKeyword ("ushort", Token.USHORT);
526 AddKeyword ("using", Token.USING);
527 AddKeyword ("virtual", Token.VIRTUAL);
528 AddKeyword ("void", Token.VOID);
529 AddKeyword ("volatile", Token.VOLATILE);
530 AddKeyword ("while", Token.WHILE);
531 AddKeyword ("partial", Token.PARTIAL);
532 AddKeyword ("where", Token.WHERE);
535 AddKeyword ("from", Token.FROM);
536 AddKeyword ("join", Token.JOIN);
537 AddKeyword ("on", Token.ON);
538 AddKeyword ("equals", Token.EQUALS);
539 AddKeyword ("select", Token.SELECT);
540 AddKeyword ("group", Token.GROUP);
541 AddKeyword ("by", Token.BY);
542 AddKeyword ("let", Token.LET);
543 AddKeyword ("orderby", Token.ORDERBY);
544 AddKeyword ("ascending", Token.ASCENDING);
545 AddKeyword ("descending", Token.DESCENDING);
546 AddKeyword ("into", Token.INTO);
555 csharp_format_info = NumberFormatInfo.InvariantInfo;
556 styles = NumberStyles.Float;
558 string_builder = new System.Text.StringBuilder ();
561 int GetKeyword (char[] id, int id_len)
564 // Keywords are stored in an array of arrays grouped by their
565 // length and then by the first character
567 if (id_len >= keywords.Length || keywords [id_len] == null)
570 int first_index = id [0] - '_';
571 if (first_index > 'z')
574 KeywordEntry kwe = keywords [id_len] [first_index];
581 for (int i = 1; i < id_len; ++i) {
582 if (id [i] != kwe.Value [i]) {
588 } while (res == 0 && kwe != null);
602 if (!handle_remove_add)
606 if (parsing_declaration == 0)
607 res = Token.EXTERN_ALIAS;
610 if (peek_token () == Token.COLON) {
612 res = Token.DEFAULT_COLON;
616 if (!handle_where && !query_parsing)
621 // A query expression is any expression that starts with `from identifier'
622 // followed by any token except ; , =
624 if (!query_parsing) {
625 if (lambda_arguments_parsing) {
631 // HACK: to disable generics micro-parser, because PushPosition does not
632 // store identifiers array
633 parsing_generic_less_than = 1;
635 case Token.IDENTIFIER:
647 next_token = xtoken ();
648 if (next_token == Token.SEMICOLON || next_token == Token.COMMA || next_token == Token.EQUALS)
651 res = Token.FROM_FIRST;
652 query_parsing = true;
653 if (RootContext.Version <= LanguageVersion.ISO_2)
654 Report.FeatureIsNotAvailable (Location, "query expressions");
657 Expression.Error_VoidInvalidInTheContext (Location, Report);
661 // HACK: A token is not a keyword so we need to restore identifiers buffer
662 // which has been overwritten before we grabbed the identifier
663 id_builder [0] = 'f'; id_builder [1] = 'r'; id_builder [2] = 'o'; id_builder [3] = 'm';
677 case Token.ASCENDING:
678 case Token.DESCENDING:
685 case Token.NAMESPACE:
686 // TODO: some explanation needed
687 check_incorrect_doc_comment ();
691 if (parsing_block > 0) {
696 // Save current position and parse next token.
699 next_token = token ();
700 bool ok = (next_token == Token.CLASS) ||
701 (next_token == Token.STRUCT) ||
702 (next_token == Token.INTERFACE) ||
703 (next_token == Token.VOID);
708 if (next_token == Token.VOID) {
709 if (RootContext.Version == LanguageVersion.ISO_1 ||
710 RootContext.Version == LanguageVersion.ISO_2)
711 Report.FeatureIsNotAvailable (Location, "partial methods");
712 } else if (RootContext.Version == LanguageVersion.ISO_1)
713 Report.FeatureIsNotAvailable (Location, "partial types");
718 if (next_token < Token.LAST_KEYWORD) {
719 Report.Error (267, Location,
720 "The `partial' modifier can be used only immediately before `class', `struct', `interface', or `void' keyword");
731 public Location Location {
733 return new Location (ref_line, hidden ? -1 : col);
737 public Tokenizer (SeekableStreamReader input, CompilationUnit file, CompilerContext ctx)
739 this.ref_name = file;
740 this.file_name = file;
746 xml_comment_buffer = new StringBuilder ();
749 // FIXME: This could be `Location.Push' but we have to
750 // find out why the MS compiler allows this
752 Mono.CSharp.Location.Push (file, file);
755 static bool is_identifier_start_character (int c)
757 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_' || Char.IsLetter ((char)c);
760 static bool is_identifier_part_character (char c)
762 if (c >= 'a' && c <= 'z')
765 if (c >= 'A' && c <= 'Z')
768 if (c == '_' || (c >= '0' && c <= '9'))
774 return Char.IsLetter (c) || Char.GetUnicodeCategory (c) == UnicodeCategory.ConnectorPunctuation;
777 public static bool IsKeyword (string s)
779 return keyword_strings.ContainsKey (s);
783 // Open parens micro parser. Detects both lambda and cast ambiguity.
785 int TokenizeOpenParens ()
790 int bracket_level = 0;
791 bool is_type = false;
792 bool can_be_type = false;
795 ptoken = current_token;
798 switch (current_token) {
799 case Token.CLOSE_PARENS:
803 // Expression inside parens is lambda, (int i) =>
805 if (current_token == Token.ARROW)
806 return Token.OPEN_PARENS_LAMBDA;
809 // Expression inside parens is single type, (int[])
812 return Token.OPEN_PARENS_CAST;
815 // Expression is possible cast, look at next token, (T)null
818 switch (current_token) {
819 case Token.OPEN_PARENS:
822 case Token.IDENTIFIER:
836 case Token.UNCHECKED:
841 // These can be part of a member access
855 return Token.OPEN_PARENS_CAST;
858 return Token.OPEN_PARENS;
861 case Token.DOUBLE_COLON:
862 if (ptoken != Token.IDENTIFIER && ptoken != Token.OP_GENERICS_GT)
867 case Token.IDENTIFIER:
870 case Token.OP_GENERICS_LT:
872 case Token.DOUBLE_COLON:
874 if (bracket_level == 0)
878 can_be_type = is_type = false;
898 if (bracket_level == 0)
903 if (bracket_level == 0) {
905 can_be_type = is_type = false;
909 case Token.OP_GENERICS_LT:
910 case Token.OPEN_BRACKET:
911 if (bracket_level++ == 0)
915 case Token.OP_GENERICS_GT:
916 case Token.CLOSE_BRACKET:
920 case Token.INTERR_NULLABLE:
922 if (bracket_level == 0)
928 can_be_type = is_type = false;
932 return Token.OPEN_PARENS;
937 public static bool IsValidIdentifier (string s)
939 if (s == null || s.Length == 0)
942 if (!is_identifier_start_character (s [0]))
945 for (int i = 1; i < s.Length; i ++)
946 if (! is_identifier_part_character (s [i]))
952 bool parse_less_than ()
955 int the_token = token ();
956 if (the_token == Token.OPEN_BRACKET) {
958 the_token = token ();
959 } while (the_token != Token.CLOSE_BRACKET);
960 the_token = token ();
961 } else if (the_token == Token.IN || the_token == Token.OUT) {
962 the_token = token ();
965 case Token.IDENTIFIER:
983 case Token.OP_GENERICS_GT:
990 the_token = token ();
992 if (the_token == Token.OP_GENERICS_GT)
994 else if (the_token == Token.COMMA || the_token == Token.DOT || the_token == Token.DOUBLE_COLON)
996 else if (the_token == Token.INTERR_NULLABLE || the_token == Token.STAR)
998 else if (the_token == Token.OP_GENERICS_LT) {
999 if (!parse_less_than ())
1002 } else if (the_token == Token.OPEN_BRACKET) {
1004 the_token = token ();
1005 if (the_token == Token.CLOSE_BRACKET)
1007 else if (the_token == Token.COMMA)
1008 goto rank_specifiers;
1015 bool parse_generic_dimension (out int dimension)
1020 int the_token = token ();
1021 if (the_token == Token.OP_GENERICS_GT)
1023 else if (the_token == Token.COMMA) {
1031 public int peek_token ()
1036 the_token = token ();
1043 // Tonizes `?' using custom disambiguous rules to return one
1044 // of following tokens: INTERR_NULLABLE, OP_COALESCING, INTERR
1046 // Tricky expression look like:
1048 // Foo ? a = x ? b : c;
1050 int TokenizePossibleNullableType ()
1052 if (parsing_block == 0 || parsing_type > 0)
1053 return Token.INTERR_NULLABLE;
1055 int d = peek_char ();
1058 return Token.OP_COALESCING;
1061 switch (current_token) {
1062 case Token.CLOSE_PARENS:
1067 return Token.INTERR;
1071 if (d == ',' || d == ';' || d == '>')
1072 return Token.INTERR_NULLABLE;
1073 if (d == '*' || (d >= '0' && d <= '9'))
1074 return Token.INTERR;
1078 current_token = Token.NONE;
1080 switch (xtoken ()) {
1087 next_token = Token.INTERR;
1090 case Token.SEMICOLON:
1092 case Token.CLOSE_PARENS:
1093 case Token.OPEN_BRACKET:
1094 case Token.OP_GENERICS_GT:
1095 next_token = Token.INTERR_NULLABLE;
1103 if (next_token == -1) {
1104 switch (xtoken ()) {
1106 case Token.SEMICOLON:
1107 case Token.OPEN_BRACE:
1108 case Token.CLOSE_PARENS:
1110 next_token = Token.INTERR_NULLABLE;
1114 next_token = Token.INTERR;
1122 // All shorcuts failed, do it hard way
1124 while ((ntoken = xtoken ()) != Token.EOF) {
1125 if (ntoken == Token.SEMICOLON)
1128 if (ntoken == Token.COLON) {
1129 if (++colons == interrs)
1134 if (ntoken == Token.INTERR) {
1140 next_token = colons != interrs ? Token.INTERR_NULLABLE : Token.INTERR;
1149 bool decimal_digits (int c)
1152 bool seen_digits = false;
1155 if (number_pos == max_number_size)
1156 Error_NumericConstantTooLong ();
1157 number_builder [number_pos++] = (char) c;
1161 // We use peek_char2, because decimal_digits needs to do a
1162 // 2-character look-ahead (5.ToString for example).
1164 while ((d = peek_char2 ()) != -1){
1165 if (d >= '0' && d <= '9'){
1166 if (number_pos == max_number_size)
1167 Error_NumericConstantTooLong ();
1168 number_builder [number_pos++] = (char) d;
1178 static bool is_hex (int e)
1180 return (e >= '0' && e <= '9') || (e >= 'A' && e <= 'F') || (e >= 'a' && e <= 'f');
1183 static TypeCode real_type_suffix (int c)
1187 return TypeCode.Single;
1189 return TypeCode.Double;
1191 return TypeCode.Decimal;
1193 return TypeCode.Empty;
1197 int integer_type_suffix (ulong ul, int c)
1199 bool is_unsigned = false;
1200 bool is_long = false;
1203 bool scanning = true;
1216 // if we have not seen anything in between
1217 // report this error
1219 Report.Warning (78, 4, Location, "The 'l' suffix is easily confused with the digit '1' (use 'L' for clarity)");
1239 if (is_long && is_unsigned){
1240 val = new ULongLiteral (ul, Location);
1241 return Token.LITERAL;
1245 // uint if possible, or ulong else.
1247 if ((ul & 0xffffffff00000000) == 0)
1248 val = new UIntLiteral ((uint) ul, Location);
1250 val = new ULongLiteral (ul, Location);
1251 } else if (is_long){
1252 // long if possible, ulong otherwise
1253 if ((ul & 0x8000000000000000) != 0)
1254 val = new ULongLiteral (ul, Location);
1256 val = new LongLiteral ((long) ul, Location);
1258 // int, uint, long or ulong in that order
1259 if ((ul & 0xffffffff00000000) == 0){
1260 uint ui = (uint) ul;
1262 if ((ui & 0x80000000) != 0)
1263 val = new UIntLiteral (ui, Location);
1265 val = new IntLiteral ((int) ui, Location);
1267 if ((ul & 0x8000000000000000) != 0)
1268 val = new ULongLiteral (ul, Location);
1270 val = new LongLiteral ((long) ul, Location);
1273 return Token.LITERAL;
1277 // given `c' as the next char in the input decide whether
1278 // we need to convert to a special type, and then choose
1279 // the best representation for the integer
1281 int adjust_int (int c)
1284 if (number_pos > 9){
1285 ulong ul = (uint) (number_builder [0] - '0');
1287 for (int i = 1; i < number_pos; i++){
1288 ul = checked ((ul * 10) + ((uint)(number_builder [i] - '0')));
1290 return integer_type_suffix (ul, c);
1292 uint ui = (uint) (number_builder [0] - '0');
1294 for (int i = 1; i < number_pos; i++){
1295 ui = checked ((ui * 10) + ((uint)(number_builder [i] - '0')));
1297 return integer_type_suffix (ui, c);
1299 } catch (OverflowException) {
1300 error_details = "Integral constant is too large";
1301 Report.Error (1021, Location, error_details);
1302 val = new IntLiteral (0, Location);
1303 return Token.LITERAL;
1305 catch (FormatException) {
1306 Report.Error (1013, Location, "Invalid number");
1307 val = new IntLiteral (0, Location);
1308 return Token.LITERAL;
1312 int adjust_real (TypeCode t)
1314 string s = new String (number_builder, 0, number_pos);
1315 const string error_details = "Floating-point constant is outside the range of type `{0}'";
1318 case TypeCode.Decimal:
1320 val = new DecimalLiteral (decimal.Parse (s, styles, csharp_format_info), Location);
1321 } catch (OverflowException) {
1322 val = new DecimalLiteral (0, Location);
1323 Report.Error (594, Location, error_details, "decimal");
1326 case TypeCode.Single:
1328 val = new FloatLiteral (float.Parse (s, styles, csharp_format_info), Location);
1329 } catch (OverflowException) {
1330 val = new FloatLiteral (0, Location);
1331 Report.Error (594, Location, error_details, "float");
1336 val = new DoubleLiteral (double.Parse (s, styles, csharp_format_info), Location);
1337 } catch (OverflowException) {
1338 val = new DoubleLiteral (0, Location);
1339 Report.Error (594, Location, error_details, "double");
1344 return Token.LITERAL;
1353 while ((d = peek_char ()) != -1){
1355 number_builder [number_pos++] = (char) d;
1361 string s = new String (number_builder, 0, number_pos);
1363 if (number_pos <= 8)
1364 ul = System.UInt32.Parse (s, NumberStyles.HexNumber);
1366 ul = System.UInt64.Parse (s, NumberStyles.HexNumber);
1367 } catch (OverflowException){
1368 error_details = "Integral constant is too large";
1369 Report.Error (1021, Location, error_details);
1370 val = new IntLiteral (0, Location);
1371 return Token.LITERAL;
1373 catch (FormatException) {
1374 Report.Error (1013, Location, "Invalid number");
1375 val = new IntLiteral (0, Location);
1376 return Token.LITERAL;
1379 return integer_type_suffix (ul, peek_char ());
1383 // Invoked if we know we have .digits or digits
1385 int is_number (int c)
1387 bool is_real = false;
1391 if (c >= '0' && c <= '9'){
1393 int peek = peek_char ();
1395 if (peek == 'x' || peek == 'X')
1396 return handle_hex ();
1403 // We need to handle the case of
1404 // "1.1" vs "1.string" (LITERAL_FLOAT vs NUMBER DOT IDENTIFIER)
1407 if (decimal_digits ('.')){
1413 return adjust_int (-1);
1417 if (c == 'e' || c == 'E'){
1419 if (number_pos == max_number_size)
1420 Error_NumericConstantTooLong ();
1421 number_builder [number_pos++] = 'e';
1425 if (number_pos == max_number_size)
1426 Error_NumericConstantTooLong ();
1427 number_builder [number_pos++] = '+';
1429 } else if (c == '-') {
1430 if (number_pos == max_number_size)
1431 Error_NumericConstantTooLong ();
1432 number_builder [number_pos++] = '-';
1435 if (number_pos == max_number_size)
1436 Error_NumericConstantTooLong ();
1437 number_builder [number_pos++] = '+';
1444 var type = real_type_suffix (c);
1445 if (type == TypeCode.Empty && !is_real){
1447 return adjust_int (c);
1452 if (type == TypeCode.Empty){
1457 return adjust_real (type);
1459 throw new Exception ("Is Number should never reach this point");
1463 // Accepts exactly count (4 or 8) hex, no more no less
1465 int getHex (int count, out int surrogate, out bool error)
1470 int top = count != -1 ? count : 4;
1475 for (i = 0; i < top; i++){
1478 if (c >= '0' && c <= '9')
1479 c = (int) c - (int) '0';
1480 else if (c >= 'A' && c <= 'F')
1481 c = (int) c - (int) 'A' + 10;
1482 else if (c >= 'a' && c <= 'f')
1483 c = (int) c - (int) 'a' + 10;
1489 total = (total * 16) + c;
1491 int p = peek_char ();
1494 if (!is_hex ((char)p))
1500 if (total > 0x0010FFFF) {
1505 if (total >= 0x00010000) {
1506 surrogate = ((total - 0x00010000) % 0x0400 + 0xDC00);
1507 total = ((total - 0x00010000) / 0x0400 + 0xD800);
1514 int escape (int c, out int surrogate)
1550 v = getHex (-1, out surrogate, out error);
1556 return EscapeUnicode (d, out surrogate);
1559 Report.Error (1009, Location, "Unrecognized escape sequence `\\{0}'", ((char)d).ToString ());
1568 int EscapeUnicode (int ch, out int surrogate)
1572 ch = getHex (8, out surrogate, out error);
1574 ch = getHex (4, out surrogate, out error);
1578 Report.Error (1009, Location, "Unrecognized escape sequence");
1586 if (putback_char != -1) {
1599 void advance_line ()
1609 if (putback_char == -1)
1610 putback_char = reader.Read ();
1611 return putback_char;
1616 if (putback_char != -1)
1617 return putback_char;
1618 return reader.Peek ();
1621 void putback (int c)
1623 if (putback_char != -1){
1624 Console.WriteLine ("Col: " + col);
1625 Console.WriteLine ("Row: " + line);
1626 Console.WriteLine ("Name: " + ref_name.Name);
1627 Console.WriteLine ("Current [{0}] putting back [{1}] ", putback_char, c);
1628 throw new Exception ("This should not happen putback on putback");
1630 if (c == '\n' || col == 0) {
1631 // It won't happen though.
1641 public bool advance ()
1643 return peek_char () != -1 || CompleteOnEOF;
1646 public Object Value {
1652 public Object value ()
1659 current_token = xtoken ();
1660 return current_token;
1663 void get_cmd_arg (out string cmd, out string arg)
1667 tokens_seen = false;
1670 // skip over white space
1673 } while (c == '\r' || c == ' ' || c == '\t');
1675 static_cmd_arg.Length = 0;
1676 while (c != -1 && is_identifier_part_character ((char)c)) {
1677 static_cmd_arg.Append ((char)c);
1680 int peek = peek_char ();
1681 if (peek == 'U' || peek == 'u') {
1683 c = EscapeUnicode (c, out surrogate);
1684 if (surrogate != 0) {
1685 if (is_identifier_part_character ((char) c))
1686 static_cmd_arg.Append ((char) c);
1693 cmd = static_cmd_arg.ToString ();
1695 // skip over white space
1696 while (c == '\r' || c == ' ' || c == '\t')
1699 static_cmd_arg.Length = 0;
1700 int has_identifier_argument = 0;
1702 while (c != -1 && c != '\n' && c != '\r') {
1703 if (c == '\\' && has_identifier_argument >= 0) {
1704 if (has_identifier_argument != 0 || (cmd == "define" || cmd == "if" || cmd == "elif" || cmd == "undef")) {
1705 has_identifier_argument = 1;
1707 int peek = peek_char ();
1708 if (peek == 'U' || peek == 'u') {
1710 c = EscapeUnicode (c, out surrogate);
1711 if (surrogate != 0) {
1712 if (is_identifier_part_character ((char) c))
1713 static_cmd_arg.Append ((char) c);
1718 has_identifier_argument = -1;
1721 static_cmd_arg.Append ((char) c);
1725 if (static_cmd_arg.Length != 0)
1726 arg = static_cmd_arg.ToString ();
1730 // Handles the #line directive
1732 bool PreProcessLine (string arg)
1734 if (arg.Length == 0)
1737 if (arg == "default"){
1739 ref_name = file_name;
1741 Location.Push (file_name, ref_name);
1743 } else if (arg == "hidden"){
1751 if ((pos = arg.IndexOf (' ')) != -1 && pos != 0){
1752 ref_line = System.Int32.Parse (arg.Substring (0, pos));
1755 char [] quotes = { '\"' };
1757 string name = arg.Substring (pos). Trim (quotes);
1758 ref_name = Location.LookupFile (file_name, name);
1759 file_name.AddFile (ref_name);
1761 Location.Push (file_name, ref_name);
1763 ref_line = System.Int32.Parse (arg);
1774 // Handles #define and #undef
1776 void PreProcessDefinition (bool is_define, string ident, bool caller_is_taking)
1778 if (ident.Length == 0 || ident == "true" || ident == "false"){
1779 Report.Error (1001, Location, "Missing identifier to pre-processor directive");
1783 if (ident.IndexOfAny (simple_whitespaces) != -1){
1784 Error_EndLineExpected ();
1788 if (!is_identifier_start_character (ident [0]))
1789 Report.Error (1001, Location, "Identifier expected: {0}", ident);
1791 foreach (char c in ident.Substring (1)){
1792 if (!is_identifier_part_character (c)){
1793 Report.Error (1001, Location, "Identifier expected: {0}", ident);
1798 if (!caller_is_taking)
1805 if (RootContext.IsConditionalDefined (ident))
1808 file_name.AddDefine (ident);
1813 file_name.AddUndefine (ident);
1817 static byte read_hex (string arg, int pos, out bool error)
1824 if ((c >= '0') && (c <= '9'))
1825 total = (int) c - (int) '0';
1826 else if ((c >= 'A') && (c <= 'F'))
1827 total = (int) c - (int) 'A' + 10;
1828 else if ((c >= 'a') && (c <= 'f'))
1829 total = (int) c - (int) 'a' + 10;
1838 if ((c >= '0') && (c <= '9'))
1839 total += (int) c - (int) '0';
1840 else if ((c >= 'A') && (c <= 'F'))
1841 total += (int) c - (int) 'A' + 10;
1842 else if ((c >= 'a') && (c <= 'f'))
1843 total += (int) c - (int) 'a' + 10;
1849 return (byte) total;
1853 /// Handles #pragma checksum
1855 bool PreProcessPragmaChecksum (string arg)
1857 if ((arg [0] != ' ') && (arg [0] != '\t'))
1860 arg = arg.Trim (simple_whitespaces);
1861 if ((arg.Length < 2) || (arg [0] != '"'))
1864 StringBuilder file_sb = new StringBuilder ();
1868 while ((ch = arg [pos++]) != '"') {
1869 if (pos >= arg.Length)
1873 if (pos+1 >= arg.Length)
1878 file_sb.Append (ch);
1881 if ((pos+2 >= arg.Length) || ((arg [pos] != ' ') && (arg [pos] != '\t')))
1884 arg = arg.Substring (pos).Trim (simple_whitespaces);
1885 if ((arg.Length < 42) || (arg [0] != '"') || (arg [1] != '{') ||
1886 (arg [10] != '-') || (arg [15] != '-') || (arg [20] != '-') ||
1887 (arg [25] != '-') || (arg [38] != '}') || (arg [39] != '"'))
1891 byte[] guid_bytes = new byte [16];
1893 for (int i = 0; i < 4; i++) {
1894 guid_bytes [i] = read_hex (arg, 2+2*i, out error);
1898 for (int i = 0; i < 2; i++) {
1899 guid_bytes [i+4] = read_hex (arg, 11+2*i, out error);
1902 guid_bytes [i+6] = read_hex (arg, 16+2*i, out error);
1905 guid_bytes [i+8] = read_hex (arg, 21+2*i, out error);
1910 for (int i = 0; i < 6; i++) {
1911 guid_bytes [i+10] = read_hex (arg, 26+2*i, out error);
1916 arg = arg.Substring (40).Trim (simple_whitespaces);
1917 if ((arg.Length < 34) || (arg [0] != '"') || (arg [33] != '"'))
1920 byte[] checksum_bytes = new byte [16];
1921 for (int i = 0; i < 16; i++) {
1922 checksum_bytes [i] = read_hex (arg, 1+2*i, out error);
1927 arg = arg.Substring (34).Trim (simple_whitespaces);
1931 SourceFile file = Location.LookupFile (file_name, file_sb.ToString ());
1932 file.SetChecksum (guid_bytes, checksum_bytes);
1933 ref_name.AutoGenerated = true;
1938 /// Handles #pragma directive
1940 void PreProcessPragma (string arg)
1942 const string warning = "warning";
1943 const string w_disable = "warning disable";
1944 const string w_restore = "warning restore";
1945 const string checksum = "checksum";
1947 if (arg == w_disable) {
1948 Report.RegisterWarningRegion (Location).WarningDisable (Location.Row);
1952 if (arg == w_restore) {
1953 Report.RegisterWarningRegion (Location).WarningEnable (Location.Row);
1957 if (arg.StartsWith (w_disable)) {
1958 int[] codes = ParseNumbers (arg.Substring (w_disable.Length));
1959 foreach (int code in codes) {
1961 Report.RegisterWarningRegion (Location).WarningDisable (Location, code, Report);
1966 if (arg.StartsWith (w_restore)) {
1967 int[] codes = ParseNumbers (arg.Substring (w_restore.Length));
1968 var w_table = Report.warning_ignore_table;
1969 foreach (int code in codes) {
1970 if (w_table != null && w_table.ContainsKey (code))
1971 Report.Warning (1635, 1, Location, "Cannot restore warning `CS{0:0000}' because it was disabled globally", code);
1972 Report.RegisterWarningRegion (Location).WarningEnable (Location, code, Report);
1977 if (arg.StartsWith (warning)) {
1978 Report.Warning (1634, 1, Location, "Expected disable or restore");
1982 if (arg.StartsWith (checksum)) {
1983 if (!PreProcessPragmaChecksum (arg.Substring (checksum.Length)))
1984 Warning_InvalidPragmaChecksum ();
1988 Report.Warning (1633, 1, Location, "Unrecognized #pragma directive");
1991 int[] ParseNumbers (string text)
1993 string[] string_array = text.Split (',');
1994 int[] values = new int [string_array.Length];
1996 foreach (string string_code in string_array) {
1998 values[index++] = int.Parse (string_code, System.Globalization.CultureInfo.InvariantCulture);
2000 catch (FormatException) {
2001 Report.Warning (1692, 1, Location, "Invalid number");
2007 bool eval_val (string s)
2014 return file_name.IsConditionalDefined (s);
2017 bool pp_primary (ref string s)
2026 s = s.Substring (1);
2027 bool val = pp_expr (ref s, false);
2028 if (s.Length > 0 && s [0] == ')'){
2029 s = s.Substring (1);
2032 Error_InvalidDirective ();
2036 if (is_identifier_start_character (c)){
2042 if (is_identifier_part_character (c)){
2046 bool v = eval_val (s.Substring (0, j));
2047 s = s.Substring (j);
2050 bool vv = eval_val (s);
2055 Error_InvalidDirective ();
2059 bool pp_unary (ref string s)
2066 if (len > 1 && s [1] == '='){
2067 Error_InvalidDirective ();
2070 s = s.Substring (1);
2071 return ! pp_primary (ref s);
2073 return pp_primary (ref s);
2075 Error_InvalidDirective ();
2080 bool pp_eq (ref string s)
2082 bool va = pp_unary (ref s);
2088 if (len > 2 && s [1] == '='){
2089 s = s.Substring (2);
2090 return va == pp_unary (ref s);
2092 Error_InvalidDirective ();
2095 } else if (s [0] == '!' && len > 1 && s [1] == '='){
2096 s = s.Substring (2);
2098 return va != pp_unary (ref s);
2107 bool pp_and (ref string s)
2109 bool va = pp_eq (ref s);
2115 if (len > 2 && s [1] == '&'){
2116 s = s.Substring (2);
2117 return (va & pp_and (ref s));
2119 Error_InvalidDirective ();
2128 // Evaluates an expression for `#if' or `#elif'
2130 bool pp_expr (ref string s, bool isTerm)
2132 bool va = pp_and (ref s);
2139 if (len > 2 && s [1] == '|'){
2140 s = s.Substring (2);
2141 return va | pp_expr (ref s, isTerm);
2143 Error_InvalidDirective ();
2148 Error_EndLineExpected ();
2156 bool eval (string s)
2158 bool v = pp_expr (ref s, true);
2167 void Error_NumericConstantTooLong ()
2169 Report.Error (1021, Location, "Numeric constant too long");
2172 void Error_InvalidDirective ()
2174 Report.Error (1517, Location, "Invalid preprocessor directive");
2177 void Error_UnexpectedDirective (string extra)
2181 "Unexpected processor directive ({0})", extra);
2184 void Error_TokensSeen ()
2186 Report.Error (1032, Location,
2187 "Cannot define or undefine preprocessor symbols after first token in file");
2190 void Eror_WrongPreprocessorLocation ()
2192 Report.Error (1040, Location,
2193 "Preprocessor directives must appear as the first non-whitespace character on a line");
2196 void Error_EndLineExpected ()
2198 Report.Error (1025, Location, "Single-line comment or end-of-line expected");
2201 void Warning_InvalidPragmaChecksum ()
2203 Report.Warning (1695, 1, Location,
2204 "Invalid #pragma checksum syntax; should be " +
2205 "#pragma checksum \"filename\" " +
2206 "\"{XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX}\" \"XXXX...\"");
2209 // if true, then the code continues processing the code
2210 // if false, the code stays in a loop until another directive is
2212 // When caller_is_taking is false we ignore all directives except the ones
2213 // which can help us to identify where the #if block ends
2214 bool handle_preprocessing_directive (bool caller_is_taking)
2217 bool region_directive = false;
2219 get_cmd_arg (out cmd, out arg);
2221 // Eat any trailing whitespaces and single-line comments
2222 if (arg.IndexOf ("//") != -1)
2223 arg = arg.Substring (0, arg.IndexOf ("//"));
2224 arg = arg.Trim (simple_whitespaces);
2227 // The first group of pre-processing instructions is always processed
2231 region_directive = true;
2236 if (ifstack == null || ifstack.Count == 0){
2237 Error_UnexpectedDirective ("no #region for this #endregion");
2240 int pop = ifstack.Pop ();
2242 if ((pop & REGION) == 0)
2243 Report.Error (1027, Location, "Expected `#endif' directive");
2245 return caller_is_taking;
2248 if (ifstack == null)
2249 ifstack = new Stack<int> (2);
2251 int flags = region_directive ? REGION : 0;
2252 if (ifstack.Count == 0){
2253 flags |= PARENT_TAKING;
2255 int state = ifstack.Peek ();
2256 if ((state & TAKING) != 0) {
2257 flags |= PARENT_TAKING;
2261 if (caller_is_taking && eval (arg)) {
2262 ifstack.Push (flags | TAKING);
2265 ifstack.Push (flags);
2269 if (ifstack == null || ifstack.Count == 0){
2270 Error_UnexpectedDirective ("no #if for this #endif");
2273 pop = ifstack.Pop ();
2275 if ((pop & REGION) != 0)
2276 Report.Error (1038, Location, "#endregion directive expected");
2278 if (arg.Length != 0) {
2279 Error_EndLineExpected ();
2282 if (ifstack.Count == 0)
2285 int state = ifstack.Peek ();
2286 return (state & TAKING) != 0;
2290 if (ifstack == null || ifstack.Count == 0){
2291 Error_UnexpectedDirective ("no #if for this #elif");
2294 int state = ifstack.Pop ();
2296 if ((state & REGION) != 0) {
2297 Report.Error (1038, Location, "#endregion directive expected");
2301 if ((state & ELSE_SEEN) != 0){
2302 Error_UnexpectedDirective ("#elif not valid after #else");
2306 if ((state & TAKING) != 0) {
2311 if (eval (arg) && ((state & PARENT_TAKING) != 0)){
2312 ifstack.Push (state | TAKING);
2316 ifstack.Push (state);
2321 if (ifstack == null || ifstack.Count == 0){
2322 Error_UnexpectedDirective ("no #if for this #else");
2325 int state = ifstack.Peek ();
2327 if ((state & REGION) != 0) {
2328 Report.Error (1038, Location, "#endregion directive expected");
2332 if ((state & ELSE_SEEN) != 0){
2333 Error_UnexpectedDirective ("#else within #else");
2339 if (arg.Length != 0) {
2340 Error_EndLineExpected ();
2345 if ((state & PARENT_TAKING) != 0) {
2346 ret = (state & TAKING) == 0;
2354 ifstack.Push (state | ELSE_SEEN);
2359 if (any_token_seen){
2360 Error_TokensSeen ();
2361 return caller_is_taking;
2363 PreProcessDefinition (true, arg, caller_is_taking);
2364 return caller_is_taking;
2367 if (any_token_seen){
2368 Error_TokensSeen ();
2369 return caller_is_taking;
2371 PreProcessDefinition (false, arg, caller_is_taking);
2372 return caller_is_taking;
2376 // These are only processed if we are in a `taking' block
2378 if (!caller_is_taking)
2383 Report.Error (1029, Location, "#error: '{0}'", arg);
2387 Report.Warning (1030, 1, Location, "#warning: `{0}'", arg);
2391 if (RootContext.Version == LanguageVersion.ISO_1) {
2392 Report.FeatureIsNotAvailable (Location, "#pragma");
2396 PreProcessPragma (arg);
2400 if (!PreProcessLine (arg))
2403 "The line number specified for #line directive is missing or invalid");
2404 return caller_is_taking;
2407 Report.Error (1024, Location, "Wrong preprocessor directive");
2412 private int consume_string (bool quoted)
2415 string_builder.Length = 0;
2417 while ((c = get_char ()) != -1){
2419 if (quoted && peek_char () == '"'){
2420 string_builder.Append ((char) c);
2424 val = new StringLiteral (string_builder.ToString (), Location);
2425 return Token.LITERAL;
2431 Report.Error (1010, Location, "Newline in constant");
2436 c = escape (c, out surrogate);
2439 if (surrogate != 0) {
2440 string_builder.Append ((char) c);
2444 string_builder.Append ((char) c);
2447 Report.Error (1039, Location, "Unterminated string literal");
2451 private int consume_identifier (int s)
2453 int res = consume_identifier (s, false);
2455 if (doc_state == XmlCommentState.Allowed)
2456 doc_state = XmlCommentState.NotAllowed;
2461 int consume_identifier (int c, bool quoted)
2464 // This method is very performance sensitive. It accounts
2465 // for approximately 25% of all parser time
2473 c = escape (c, out surrogate);
2474 if (surrogate != 0) {
2475 id_builder [pos++] = (char) c;
2480 id_builder [pos++] = (char) c;
2486 if ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_' || (c >= '0' && c <= '9')) {
2487 id_builder [pos++] = (char) c;
2494 c = escape (c, out surrogate);
2495 if (surrogate != 0) {
2496 if (is_identifier_part_character ((char) c))
2497 id_builder[pos++] = (char) c;
2503 } else if (Char.IsLetter ((char) c) || Char.GetUnicodeCategory ((char) c) == UnicodeCategory.ConnectorPunctuation) {
2504 id_builder [pos++] = (char) c;
2511 } catch (IndexOutOfRangeException) {
2512 Report.Error (645, Location, "Identifier too long (limit is 512 chars)");
2520 // Optimization: avoids doing the keyword lookup
2521 // on uppercase letters
2523 if (id_builder [0] >= '_' && !quoted) {
2524 int keyword = GetKeyword (id_builder, pos);
2525 if (keyword != -1) {
2526 val = LocatedToken.Create (null, ref_line, column);
2532 // Keep identifiers in an array of hashtables to avoid needless
2535 var identifiers_group = identifiers [pos];
2537 if (identifiers_group != null) {
2538 if (identifiers_group.TryGetValue (id_builder, out s)) {
2539 val = LocatedToken.Create (s, ref_line, column);
2541 AddEscapedIdentifier (((LocatedToken) val).Location);
2542 return Token.IDENTIFIER;
2545 // TODO: this should be number of files dependant
2546 // corlib compilation peaks at 1000 and System.Core at 150
2547 int capacity = pos > 20 ? 10 : 100;
2548 identifiers_group = new Dictionary<char[],string> (capacity, new IdentifiersComparer (pos));
2549 identifiers [pos] = identifiers_group;
2552 char [] chars = new char [pos];
2553 Array.Copy (id_builder, chars, pos);
2555 s = new string (id_builder, 0, pos);
2556 identifiers_group.Add (chars, s);
2558 val = LocatedToken.Create (s, ref_line, column);
2560 AddEscapedIdentifier (((LocatedToken) val).Location);
2562 return Token.IDENTIFIER;
2565 public int xtoken ()
2569 // Whether we have seen comments on the current line
2570 bool comments_seen = false;
2571 while ((c = get_char ()) != -1) {
2574 col = ((col + 8) / 8) * 8;
2582 case 0xFEFF: // Ignore BOM anywhere in the file
2585 /* This is required for compatibility with .NET
2587 if (peek_char () == 0xBB) {
2590 if (get_char () == 0xBF)
2597 if (peek_char () != '\n')
2602 any_token_seen |= tokens_seen;
2603 tokens_seen = false;
2604 comments_seen = false;
2609 return consume_identifier (c);
2612 val = LocatedToken.Create (ref_line, col);
2613 return Token.OPEN_BRACE;
2615 val = LocatedToken.Create (ref_line, col);
2616 return Token.CLOSE_BRACE;
2618 // To block doccomment inside attribute declaration.
2619 if (doc_state == XmlCommentState.Allowed)
2620 doc_state = XmlCommentState.NotAllowed;
2621 val = LocatedToken.CreateOptional (ref_line, col);
2622 return Token.OPEN_BRACKET;
2624 val = LocatedToken.CreateOptional (ref_line, col);
2625 return Token.CLOSE_BRACKET;
2627 val = LocatedToken.Create (ref_line, col);
2629 // An expression versions of parens can appear in block context only
2631 if (parsing_block != 0 && !lambda_arguments_parsing) {
2634 // Optmize most common case where we know that parens
2637 switch (current_token) {
2638 case Token.IDENTIFIER:
2646 case Token.DELEGATE:
2647 case Token.OP_GENERICS_GT:
2648 return Token.OPEN_PARENS;
2651 // Optimize using peek
2652 int xx = peek_char ();
2659 return Token.OPEN_PARENS;
2662 lambda_arguments_parsing = true;
2664 d = TokenizeOpenParens ();
2666 lambda_arguments_parsing = false;
2670 return Token.OPEN_PARENS;
2672 val = LocatedToken.CreateOptional (ref_line, col);
2673 return Token.CLOSE_PARENS;
2675 val = LocatedToken.CreateOptional (ref_line, col);
2678 val = LocatedToken.CreateOptional (ref_line, col);
2679 return Token.SEMICOLON;
2681 val = LocatedToken.Create (ref_line, col);
2684 val = LocatedToken.Create (ref_line, col);
2685 return TokenizePossibleNullableType ();
2687 val = LocatedToken.Create (ref_line, col);
2688 if (parsing_generic_less_than++ > 0)
2689 return Token.OP_GENERICS_LT;
2691 return TokenizeLessThan ();
2694 val = LocatedToken.Create (ref_line, col);
2702 if (parsing_generic_less_than > 1 || (parsing_generic_less_than == 1 && d != '>')) {
2703 parsing_generic_less_than--;
2704 return Token.OP_GENERICS_GT;
2713 return Token.OP_SHIFT_RIGHT_ASSIGN;
2715 return Token.OP_SHIFT_RIGHT;
2721 val = LocatedToken.Create (ref_line, col);
2725 } else if (d == '=') {
2726 d = Token.OP_ADD_ASSIGN;
2734 val = LocatedToken.Create (ref_line, col);
2738 } else if (d == '=')
2739 d = Token.OP_SUB_ASSIGN;
2749 val = LocatedToken.Create (ref_line, col);
2750 if (peek_char () == '='){
2757 val = LocatedToken.Create (ref_line, col);
2768 return Token.ASSIGN;
2771 val = LocatedToken.Create (ref_line, col);
2775 return Token.OP_AND;
2779 return Token.OP_AND_ASSIGN;
2781 return Token.BITWISE_AND;
2784 val = LocatedToken.Create (ref_line, col);
2792 return Token.OP_OR_ASSIGN;
2794 return Token.BITWISE_OR;
2797 val = LocatedToken.Create (ref_line, col);
2798 if (peek_char () == '='){
2800 return Token.OP_MULT_ASSIGN;
2807 val = LocatedToken.Create (ref_line, col);
2809 return Token.OP_DIV_ASSIGN;
2812 // Handle double-slash comments.
2815 if (RootContext.Documentation != null && peek_char () == '/') {
2817 // Don't allow ////.
2818 if ((d = peek_char ()) != '/') {
2819 update_comment_location ();
2820 if (doc_state == XmlCommentState.Allowed)
2821 handle_one_line_xml_comment ();
2822 else if (doc_state == XmlCommentState.NotAllowed)
2823 warn_incorrect_doc_comment ();
2826 while ((d = get_char ()) != -1 && (d != '\n') && d != '\r');
2828 any_token_seen |= tokens_seen;
2829 tokens_seen = false;
2830 comments_seen = false;
2832 } else if (d == '*'){
2834 bool docAppend = false;
2835 if (RootContext.Documentation != null && peek_char () == '*') {
2837 update_comment_location ();
2838 // But when it is /**/, just do nothing.
2839 if (peek_char () == '/') {
2843 if (doc_state == XmlCommentState.Allowed)
2845 else if (doc_state == XmlCommentState.NotAllowed)
2846 warn_incorrect_doc_comment ();
2849 int current_comment_start = 0;
2851 current_comment_start = xml_comment_buffer.Length;
2852 xml_comment_buffer.Append (Environment.NewLine);
2855 while ((d = get_char ()) != -1){
2856 if (d == '*' && peek_char () == '/'){
2858 comments_seen = true;
2862 xml_comment_buffer.Append ((char) d);
2865 any_token_seen |= tokens_seen;
2866 tokens_seen = false;
2868 // Reset 'comments_seen' just to be consistent.
2869 // It doesn't matter either way, here.
2871 comments_seen = false;
2875 Report.Error (1035, Location, "End-of-file found, '*/' expected");
2878 update_formatted_doc_comment (current_comment_start);
2884 val = LocatedToken.Create (ref_line, col);
2885 if (peek_char () == '='){
2887 return Token.OP_MOD_ASSIGN;
2889 return Token.PERCENT;
2892 val = LocatedToken.Create (ref_line, col);
2893 if (peek_char () == '='){
2895 return Token.OP_XOR_ASSIGN;
2897 return Token.CARRET;
2900 val = LocatedToken.Create (ref_line, col);
2901 if (peek_char () == ':') {
2903 return Token.DOUBLE_COLON;
2907 case '0': case '1': case '2': case '3': case '4':
2908 case '5': case '6': case '7': case '8': case '9':
2910 return is_number (c);
2912 case '\n': // white space
2913 any_token_seen |= tokens_seen;
2914 tokens_seen = false;
2915 comments_seen = false;
2921 if (d >= '0' && d <= '9')
2922 return is_number (c);
2926 if (tokens_seen || comments_seen) {
2927 Eror_WrongPreprocessorLocation ();
2931 if (handle_preprocessing_directive (true))
2934 bool directive_expected = false;
2935 while ((c = get_char ()) != -1) {
2937 directive_expected = true;
2938 } else if (!directive_expected) {
2939 // TODO: Implement comment support for disabled code and uncomment this code
2941 // Eror_WrongPreprocessorLocation ();
2942 // return Token.ERROR;
2947 if (c == ' ' || c == '\t' || c == '\r' || c == '\n' || c == '\f' || c == '\v' )
2951 if (handle_preprocessing_directive (false))
2954 directive_expected = false;
2958 tokens_seen = false;
2965 return consume_string (false);
2968 return TokenizeBackslash ();
2974 return consume_string (true);
2977 if (is_identifier_start_character (c)){
2978 return consume_identifier (c, true);
2981 Report.Error (1646, Location, "Keyword, identifier, or string expected after verbatim specifier: @");
2984 case EvalStatementParserCharacter:
2985 return Token.EVAL_STATEMENT_PARSER;
2986 case EvalCompilationUnitParserCharacter:
2987 return Token.EVAL_COMPILATION_UNIT_PARSER;
2988 case EvalUsingDeclarationsParserCharacter:
2989 return Token.EVAL_USING_DECLARATIONS_UNIT_PARSER;
2992 if (is_identifier_start_character (c)) {
2994 return consume_identifier (c);
2997 error_details = ((char)c).ToString ();
3003 return Token.COMPLETE_COMPLETION;
3006 return Token.GENERATE_COMPLETION;
3013 int TokenizeBackslash ()
3015 int c = get_char ();
3018 error_details = "Empty character literal";
3019 Report.Error (1011, Location, error_details);
3022 if (c == '\r' || c == '\n') {
3023 Report.Error (1010, Location, "Newline in constant");
3028 c = escape (c, out d);
3032 throw new NotImplementedException ();
3034 val = new CharLiteral ((char) c, Location);
3038 Report.Error (1012, Location, "Too many characters in character literal");
3040 // Try to recover, read until newline or next "'"
3041 while ((c = get_char ()) != -1) {
3042 if (c == '\n' || c == '\'')
3048 return Token.LITERAL;
3051 int TokenizeLessThan ()
3054 if (handle_typeof) {
3056 if (parse_generic_dimension (out d)) {
3059 return Token.GENERIC_DIMENSION;
3064 // Save current position and parse next token.
3066 if (parse_less_than ()) {
3067 if (parsing_generic_declaration && token () != Token.DOT) {
3068 d = Token.OP_GENERICS_LT_DECL;
3070 d = Token.OP_GENERICS_LT;
3077 parsing_generic_less_than = 0;
3086 return Token.OP_SHIFT_LEFT_ASSIGN;
3088 return Token.OP_SHIFT_LEFT;
3099 // Handles one line xml comment
3101 private void handle_one_line_xml_comment ()
3104 while ((c = peek_char ()) == ' ')
3105 get_char (); // skip heading whitespaces.
3106 while ((c = peek_char ()) != -1 && c != '\n' && c != '\r') {
3107 xml_comment_buffer.Append ((char) get_char ());
3109 if (c == '\r' || c == '\n')
3110 xml_comment_buffer.Append (Environment.NewLine);
3114 // Remove heading "*" in Javadoc-like xml documentation.
3116 private void update_formatted_doc_comment (int current_comment_start)
3118 int length = xml_comment_buffer.Length - current_comment_start;
3119 string [] lines = xml_comment_buffer.ToString (
3120 current_comment_start,
3121 length).Replace ("\r", "").Split ('\n');
3123 // The first line starts with /**, thus it is not target
3124 // for the format check.
3125 for (int i = 1; i < lines.Length; i++) {
3126 string s = lines [i];
3127 int idx = s.IndexOf ('*');
3130 if (i < lines.Length - 1)
3134 head = s.Substring (0, idx);
3135 foreach (char c in head)
3138 lines [i] = s.Substring (idx + 1);
3140 xml_comment_buffer.Remove (current_comment_start, length);
3141 xml_comment_buffer.Insert (current_comment_start, String.Join (Environment.NewLine, lines));
3145 // Updates current comment location.
3147 private void update_comment_location ()
3149 if (current_comment_location.IsNull) {
3150 // "-2" is for heading "//" or "/*"
3151 current_comment_location =
3152 new Location (ref_line, hidden ? -1 : col - 2);
3157 // Checks if there was incorrect doc comments and raise
3160 public void check_incorrect_doc_comment ()
3162 if (xml_comment_buffer.Length > 0)
3163 warn_incorrect_doc_comment ();
3167 // Raises a warning when tokenizer found incorrect doccomment
3170 private void warn_incorrect_doc_comment ()
3172 if (doc_state != XmlCommentState.Error) {
3173 doc_state = XmlCommentState.Error;
3174 // in csc, it is 'XML comment is not placed on
3175 // a valid language element'. But that does not
3177 Report.Warning (1587, 2, Location, "XML comment is not placed on a valid language element");
3182 // Consumes the saved xml comment lines (if any)
3183 // as for current target member or type.
3185 public string consume_doc_comment ()
3187 if (xml_comment_buffer.Length > 0) {
3188 string ret = xml_comment_buffer.ToString ();
3189 reset_doc_comment ();
3196 get { return context.Report; }
3199 void reset_doc_comment ()
3201 xml_comment_buffer.Length = 0;
3202 current_comment_location = Location.Null;
3205 public void cleanup ()
3207 if (ifstack != null && ifstack.Count >= 1) {
3208 int state = ifstack.Pop ();
3209 if ((state & REGION) != 0)
3210 Report.Error (1038, Location, "#endregion directive expected");
3212 Report.Error (1027, Location, "Expected `#endif' directive");
3218 // Indicates whether it accepts XML documentation or not.
3220 public enum XmlCommentState {
3221 // comment is allowed in this state.
3223 // comment is not allowed in this state.
3225 // once comments appeared when it is NotAllowed, then the
3226 // state is changed to it, until the state is changed to