2 // cs-tokenizer.cs: The Tokenizer for the C# compiler
\r
3 // This also implements the preprocessor
\r
5 // Author: Miguel de Icaza (miguel@gnu.org)
\r
7 // Licensed under the terms of the GNU GPL
\r
9 // (C) 2001 Ximian, Inc (http://www.ximian.com)
\r
14 * Make sure we accept the proper Unicode ranges, per the spec.
\r
20 using System.Collections;
\r
22 using System.Globalization;
\r
24 namespace Mono.CSharp
\r
27 /// Tokenizer for C# source code.
\r
30 public class Tokenizer : yyParser.yyInput
\r
32 StreamReader reader;
\r
33 public string ref_name;
\r
34 public int ref_line = 1;
\r
35 public int line = 1;
\r
37 public int current_token;
\r
38 bool handle_get_set = false;
\r
39 bool handle_remove_add = false;
\r
40 bool handle_assembly = false;
\r
43 // Whether tokens have been seen on this line
\r
45 bool tokens_seen = false;
\r
48 // Whether a token has been seen on the file
\r
49 // This is needed because `define' is not allowed to be used
\r
50 // after a token has been seen.
\r
52 bool any_token_seen = false;
\r
55 // Returns a verbose representation of the current location
\r
57 public string location {
\r
61 if (current_token == Token.ERROR)
\r
62 det = "detail: " + error_details;
\r
66 // return "Line: "+line+" Col: "+col + "\n" +
\r
67 // "VirtLine: "+ref_line +
\r
68 // " Token: "+current_token + " " + det;
\r
70 return ref_name + " " + "(" + line + "," + col + "), Token:" + current_token + " " + det;
\r
74 public bool PropertyParsing {
\r
76 return handle_get_set;
\r
80 handle_get_set = value;
\r
84 public bool AssemblyTargetParsing {
\r
86 return handle_assembly;
\r
90 handle_assembly = value;
\r
94 public bool EventParsing {
\r
96 return handle_remove_add;
\r
100 handle_remove_add = value;
\r
107 static Hashtable keywords;
\r
108 static NumberStyles styles;
\r
109 static NumberFormatInfo csharp_format_info;
\r
112 // Values for the associated token returned
\r
122 const int TAKING = 1;
\r
123 const int TAKEN_BEFORE = 2;
\r
124 const int ELSE_SEEN = 4;
\r
125 const int PARENT_TAKING = 8;
\r
128 // pre-processor if stack state:
\r
132 static System.Text.StringBuilder id_builder;
\r
133 static System.Text.StringBuilder string_builder;
\r
134 static System.Text.StringBuilder number_builder;
\r
137 // Details about the error encoutered by the tokenizer
\r
139 string error_details;
\r
141 public string error {
\r
143 return error_details;
\r
159 static void InitTokens ()
\r
161 keywords = new Hashtable ();
\r
163 keywords.Add ("abstract", Token.ABSTRACT);
\r
164 keywords.Add ("as", Token.AS);
\r
165 keywords.Add ("add", Token.ADD);
\r
166 keywords.Add ("assembly", Token.ASSEMBLY);
\r
167 keywords.Add ("base", Token.BASE);
\r
168 keywords.Add ("bool", Token.BOOL);
\r
169 keywords.Add ("break", Token.BREAK);
\r
170 keywords.Add ("byte", Token.BYTE);
\r
171 keywords.Add ("case", Token.CASE);
\r
172 keywords.Add ("catch", Token.CATCH);
\r
173 keywords.Add ("char", Token.CHAR);
\r
174 keywords.Add ("checked", Token.CHECKED);
\r
175 keywords.Add ("class", Token.CLASS);
\r
176 keywords.Add ("const", Token.CONST);
\r
177 keywords.Add ("continue", Token.CONTINUE);
\r
178 keywords.Add ("decimal", Token.DECIMAL);
\r
179 keywords.Add ("default", Token.DEFAULT);
\r
180 keywords.Add ("delegate", Token.DELEGATE);
\r
181 keywords.Add ("do", Token.DO);
\r
182 keywords.Add ("double", Token.DOUBLE);
\r
183 keywords.Add ("else", Token.ELSE);
\r
184 keywords.Add ("enum", Token.ENUM);
\r
185 keywords.Add ("event", Token.EVENT);
\r
186 keywords.Add ("explicit", Token.EXPLICIT);
\r
187 keywords.Add ("extern", Token.EXTERN);
\r
188 keywords.Add ("false", Token.FALSE);
\r
189 keywords.Add ("finally", Token.FINALLY);
\r
190 keywords.Add ("fixed", Token.FIXED);
\r
191 keywords.Add ("float", Token.FLOAT);
\r
192 keywords.Add ("for", Token.FOR);
\r
193 keywords.Add ("foreach", Token.FOREACH);
\r
194 keywords.Add ("goto", Token.GOTO);
\r
195 keywords.Add ("get", Token.GET);
\r
196 keywords.Add ("if", Token.IF);
\r
197 keywords.Add ("implicit", Token.IMPLICIT);
\r
198 keywords.Add ("in", Token.IN);
\r
199 keywords.Add ("int", Token.INT);
\r
200 keywords.Add ("interface", Token.INTERFACE);
\r
201 keywords.Add ("internal", Token.INTERNAL);
\r
202 keywords.Add ("is", Token.IS);
\r
203 keywords.Add ("lock", Token.LOCK);
\r
204 keywords.Add ("long", Token.LONG);
\r
205 keywords.Add ("namespace", Token.NAMESPACE);
\r
206 keywords.Add ("new", Token.NEW);
\r
207 keywords.Add ("null", Token.NULL);
\r
208 keywords.Add ("object", Token.OBJECT);
\r
209 keywords.Add ("operator", Token.OPERATOR);
\r
210 keywords.Add ("out", Token.OUT);
\r
211 keywords.Add ("override", Token.OVERRIDE);
\r
212 keywords.Add ("params", Token.PARAMS);
\r
213 keywords.Add ("private", Token.PRIVATE);
\r
214 keywords.Add ("protected", Token.PROTECTED);
\r
215 keywords.Add ("public", Token.PUBLIC);
\r
216 keywords.Add ("readonly", Token.READONLY);
\r
217 keywords.Add ("ref", Token.REF);
\r
218 keywords.Add ("remove", Token.REMOVE);
\r
219 keywords.Add ("return", Token.RETURN);
\r
220 keywords.Add ("sbyte", Token.SBYTE);
\r
221 keywords.Add ("sealed", Token.SEALED);
\r
222 keywords.Add ("set", Token.SET);
\r
223 keywords.Add ("short", Token.SHORT);
\r
224 keywords.Add ("sizeof", Token.SIZEOF);
\r
225 keywords.Add ("stackalloc", Token.STACKALLOC);
\r
226 keywords.Add ("static", Token.STATIC);
\r
227 keywords.Add ("string", Token.STRING);
\r
228 keywords.Add ("struct", Token.STRUCT);
\r
229 keywords.Add ("switch", Token.SWITCH);
\r
230 keywords.Add ("this", Token.THIS);
\r
231 keywords.Add ("throw", Token.THROW);
\r
232 keywords.Add ("true", Token.TRUE);
\r
233 keywords.Add ("try", Token.TRY);
\r
234 keywords.Add ("typeof", Token.TYPEOF);
\r
235 keywords.Add ("uint", Token.UINT);
\r
236 keywords.Add ("ulong", Token.ULONG);
\r
237 keywords.Add ("unchecked", Token.UNCHECKED);
\r
238 keywords.Add ("unsafe", Token.UNSAFE);
\r
239 keywords.Add ("ushort", Token.USHORT);
\r
240 keywords.Add ("using", Token.USING);
\r
241 keywords.Add ("virtual", Token.VIRTUAL);
\r
242 keywords.Add ("void", Token.VOID);
\r
243 keywords.Add ("volatile", Token.VOLATILE);
\r
244 keywords.Add ("while", Token.WHILE);
\r
248 // Class initializer
\r
250 static Tokenizer ()
\r
253 csharp_format_info = NumberFormatInfo.InvariantInfo;
\r
254 styles = NumberStyles.Float;
\r
256 id_builder = new System.Text.StringBuilder ();
\r
257 string_builder = new System.Text.StringBuilder ();
\r
258 number_builder = new System.Text.StringBuilder ();
\r
261 bool is_keyword (string name)
\r
265 res = keywords.Contains (name);
\r
266 if (handle_get_set == false && (name == "get" || name == "set"))
\r
268 if (handle_remove_add == false && (name == "remove" || name == "add"))
\r
270 if (handle_assembly == false && (name == "assembly"))
\r
275 int GetKeyword (string name)
\r
277 return (int) (keywords [name]);
\r
280 public Location Location {
\r
282 return new Location (ref_line);
\r
286 void define (string def)
\r
288 if (!RootContext.AllDefines.Contains (def)){
\r
289 RootContext.AllDefines [def] = true;
\r
291 if (defines.Contains (def))
\r
293 defines [def] = true;
\r
296 public Tokenizer (System.IO.Stream input, string fname, ArrayList defs)
\r
298 this.ref_name = fname;
\r
299 reader = new System.IO.StreamReader (input);
\r
303 defines = new Hashtable ();
\r
304 foreach (string def in defs)
\r
309 // FIXME: This could be `Location.Push' but we have to
\r
310 // find out why the MS compiler allows this
\r
312 Mono.CSharp.Location.Push (fname);
\r
315 bool is_identifier_start_character (char c)
\r
317 return Char.IsLetter (c) || c == '_' ;
\r
320 bool is_identifier_part_character (char c)
\r
322 return (Char.IsLetter (c) || Char.IsDigit (c) || c == '_');
\r
325 int is_punct (char c, ref bool doread)
\r
334 return Token.OPEN_BRACE;
\r
336 return Token.CLOSE_BRACE;
\r
338 return Token.OPEN_BRACKET;
\r
340 return Token.CLOSE_BRACKET;
\r
342 return Token.OPEN_PARENS;
\r
344 return Token.CLOSE_PARENS;
\r
346 return Token.COMMA;
\r
348 return Token.COLON;
\r
350 return Token.SEMICOLON;
\r
352 return Token.TILDE;
\r
354 return Token.INTERR;
\r
363 t = Token.OP_ADD_ASSIGN;
\r
373 t = Token.OP_SUB_ASSIGN;
\r
377 return Token.MINUS;
\r
385 return Token.OP_NE;
\r
393 return Token.OP_EQ;
\r
395 return Token.ASSIGN;
\r
401 return Token.OP_AND;
\r
402 } else if (d == '='){
\r
404 return Token.OP_AND_ASSIGN;
\r
406 return Token.BITWISE_AND;
\r
412 return Token.OP_OR;
\r
413 } else if (d == '='){
\r
415 return Token.OP_OR_ASSIGN;
\r
417 return Token.BITWISE_OR;
\r
423 return Token.OP_MULT_ASSIGN;
\r
431 return Token.OP_DIV_ASSIGN;
\r
439 return Token.OP_MOD_ASSIGN;
\r
441 return Token.PERCENT;
\r
447 return Token.OP_XOR_ASSIGN;
\r
449 return Token.CARRET;
\r
459 return Token.OP_SHIFT_LEFT_ASSIGN;
\r
461 return Token.OP_SHIFT_LEFT;
\r
462 } else if (d == '='){
\r
464 return Token.OP_LE;
\r
466 return Token.OP_LT;
\r
476 return Token.OP_SHIFT_RIGHT_ASSIGN;
\r
478 return Token.OP_SHIFT_RIGHT;
\r
479 } else if (d == '='){
\r
481 return Token.OP_GE;
\r
483 return Token.OP_GT;
\r
485 return Token.ERROR;
\r
488 bool decimal_digits (int c)
\r
491 bool seen_digits = false;
\r
494 number_builder.Append ((char) c);
\r
496 while ((d = peekChar ()) != -1){
\r
497 if (Char.IsDigit ((char)d)){
\r
498 number_builder.Append ((char) d);
\r
500 seen_digits = true;
\r
505 return seen_digits;
\r
508 bool is_hex (char e)
\r
510 return Char.IsDigit (e) || (e >= 'A' && e <= 'F') || (e >= 'a' && e <= 'f');
\r
513 void hex_digits (int c)
\r
518 number_builder.Append ((char) c);
\r
519 while ((d = peekChar ()) != -1){
\r
520 char e = Char.ToUpper ((char) d);
\r
523 number_builder.Append ((char) e);
\r
530 int real_type_suffix (int c)
\r
535 case 'F': case 'f':
\r
536 t = Token.LITERAL_FLOAT;
\r
538 case 'D': case 'd':
\r
539 t = Token.LITERAL_DOUBLE;
\r
541 case 'M': case 'm':
\r
542 t= Token.LITERAL_DECIMAL;
\r
550 int integer_type_suffix (ulong ul, int c)
\r
552 bool is_unsigned = false;
\r
553 bool is_long = false;
\r
556 bool scanning = true;
\r
559 case 'U': case 'u':
\r
562 is_unsigned = true;
\r
569 // if we have not seen anything in between
\r
570 // report this error
\r
574 "the 'l' suffix is easily confused with digit `1'," +
\r
575 " use 'L' for clarity");
\r
591 } while (scanning);
\r
594 if (is_long && is_unsigned){
\r
596 return Token.LITERAL_INTEGER;
\r
597 } else if (is_unsigned){
\r
598 // uint if possible, or ulong else.
\r
600 if ((ul & 0xffffffff00000000) == 0)
\r
604 } else if (is_long){
\r
605 // long if possible, ulong otherwise
\r
606 if ((ul & 0x8000000000000000) != 0)
\r
611 // int, uint, long or ulong in that order
\r
612 if ((ul & 0xffffffff00000000) == 0){
\r
613 uint ui = (uint) ul;
\r
615 if ((ui & 0x80000000) != 0)
\r
620 if ((ul & 0x8000000000000000) != 0)
\r
626 return Token.LITERAL_INTEGER;
\r
630 // given `c' as the next char in the input decide whether
\r
631 // we need to convert to a special type, and then choose
\r
632 // the best representation for the integer
\r
634 int adjust_int (int c)
\r
637 ulong ul = System.UInt64.Parse (number_builder.ToString ());
\r
638 return integer_type_suffix (ul, c);
\r
639 } catch (OverflowException) {
\r
640 error_details = "Integral constant is too large";
\r
641 Report.Error (1021, Location, error_details);
\r
643 return Token.LITERAL_INTEGER;
\r
647 int adjust_real (int t)
\r
649 string s = number_builder.ToString ();
\r
652 case Token.LITERAL_DECIMAL:
\r
654 val = System.Decimal.Parse (s, styles, csharp_format_info);
\r
655 } catch (OverflowException) {
\r
657 error_details = "Floating-point constant is outside the range of the type 'decimal'";
\r
658 Report.Error(594, Location, error_details);
\r
661 case Token.LITERAL_FLOAT:
\r
663 val = (float) System.Double.Parse (s, styles, csharp_format_info);
\r
664 } catch (OverflowException) {
\r
666 error_details = "Floating-point constant is outside the range of the type 'float'";
\r
667 Report.Error(594, Location, error_details);
\r
671 case Token.LITERAL_DOUBLE:
\r
673 t = Token.LITERAL_DOUBLE;
\r
675 val = System.Double.Parse (s, styles, csharp_format_info);
\r
676 } catch (OverflowException) {
\r
678 error_details = "Floating-point constant is outside the range of the type 'double'";
\r
679 Report.Error(594, Location, error_details);
\r
687 // Invoked if we know we have .digits or digits
\r
689 int is_number (int c)
\r
691 bool is_real = false;
\r
694 number_builder.Length = 0;
\r
696 if (Char.IsDigit ((char)c)){
\r
697 if (c == '0' && peekChar () == 'x' || peekChar () == 'X'){
\r
702 string s = number_builder.ToString ();
\r
704 ul = System.UInt64.Parse (s, NumberStyles.HexNumber);
\r
705 return integer_type_suffix (ul, peekChar ());
\r
707 decimal_digits (c);
\r
712 // We need to handle the case of
\r
713 // "1.1" vs "1.string" (LITERAL_FLOAT vs NUMBER DOT IDENTIFIER)
\r
716 if (decimal_digits ('.')){
\r
721 number_builder.Length -= 1;
\r
722 return adjust_int (-1);
\r
726 if (c == 'e' || c == 'E'){
\r
728 number_builder.Append ("e");
\r
732 number_builder.Append ((char) c);
\r
734 number_builder.Append ((char) c);
\r
735 decimal_digits (-1);
\r
739 type = real_type_suffix (c);
\r
740 if (type == Token.NONE && !is_real){
\r
742 return adjust_int (c);
\r
746 if (type == Token.NONE){
\r
751 return adjust_real (type);
\r
753 Console.WriteLine ("This should not be reached");
\r
754 throw new Exception ("Is Number should never reach this point");
\r
758 // Accepts exactly count (4 or 8) hex, no more no less
\r
760 int getHex (int count, out bool error)
\r
762 int [] buffer = new int [8];
\r
767 int top = count != -1 ? count : 4;
\r
771 for (i = 0; i < top; i++){
\r
773 e = Char.ToUpper ((char) c);
\r
779 if (Char.IsDigit (e))
\r
780 c = (int) e - (int) '0';
\r
782 c = (int) e - (int) 'A' + 10;
\r
783 total = (total * 16) + c;
\r
785 int p = peekChar ();
\r
788 if (!is_hex ((char)p))
\r
829 v = getHex (-1, out error);
\r
834 v = getHex (4, out error);
\r
839 v = getHex (8, out error);
\r
844 Report.Error (1009, Location, "Unrecognized escape sequence in " + (char)d);
\r
853 if (putback_char != -1){
\r
854 int x = putback_char;
\r
859 return reader.Read ();
\r
864 if (putback_char != -1)
\r
865 return putback_char;
\r
866 return reader.Peek ();
\r
869 void putback (int c)
\r
871 if (putback_char != -1)
\r
872 throw new Exception ("This should not happen putback on putback");
\r
876 public bool advance ()
\r
878 return peekChar () != -1;
\r
881 public Object Value {
\r
887 public Object value ()
\r
892 public int token ()
\r
894 current_token = xtoken ();
\r
895 return current_token;
\r
898 static StringBuilder static_cmd_arg = new System.Text.StringBuilder ();
\r
900 void get_cmd_arg (out string cmd, out string arg)
\r
904 tokens_seen = false;
\r
906 static_cmd_arg.Length = 0;
\r
908 while ((c = getChar ()) != -1 && (c != '\n') && (c != ' ') && (c != '\t')){
\r
911 static_cmd_arg.Append ((char) c);
\r
914 cmd = static_cmd_arg.ToString ();
\r
922 // skip over white space
\r
923 while ((c = getChar ()) != -1 && (c != '\n') && ((c == ' ') || (c == '\t')))
\r
932 static_cmd_arg.Length = 0;
\r
933 static_cmd_arg.Append ((char) c);
\r
935 while ((c = getChar ()) != -1 && (c != '\n')){
\r
938 static_cmd_arg.Append ((char) c);
\r
945 arg = static_cmd_arg.ToString ().Trim ();
\r
949 // Handles the #line directive
\r
951 bool PreProcessLine (string arg)
\r
956 if (arg == "default"){
\r
957 line = ref_line = line;
\r
964 if ((pos = arg.IndexOf (' ')) != -1 && pos != 0){
\r
965 ref_line = System.Int32.Parse (arg.Substring (0, pos));
\r
968 char [] quotes = { '\"' };
\r
970 ref_name = arg.Substring (pos). Trim(quotes);
\r
972 ref_line = System.Int32.Parse (arg);
\r
982 // Handles #define and #undef
\r
984 void PreProcessDefinition (bool is_define, string arg)
\r
986 if (arg == "" || arg == "true" || arg == "false"){
\r
987 Report.Error(1001, Location, "Missing identifer to pre-processor directive");
\r
991 char[] whitespace = { ' ', '\t' };
\r
992 if (arg.IndexOfAny (whitespace) != -1){
\r
993 Report.Error(1025, Location, "Single-line comment or end-of-line expected");
\r
997 foreach (char c in arg){
\r
998 if (!Char.IsLetter (c) && (c != '_')){
\r
999 Report.Error(1001, Location, "Identifier expected");
\r
1005 if (defines == null)
\r
1006 defines = new Hashtable ();
\r
1009 if (defines == null)
\r
1011 if (defines.Contains (arg))
\r
1012 defines.Remove (arg);
\r
1016 bool eval_val (string s)
\r
1023 if (defines == null)
\r
1025 if (defines.Contains (s))
\r
1031 bool pp_primary (ref string s)
\r
1034 int len = s.Length;
\r
1040 s = s.Substring (1);
\r
1041 bool val = pp_expr (ref s);
\r
1042 if (s.Length > 0 && s [0] == ')'){
\r
1043 s = s.Substring (1);
\r
1046 Error_InvalidDirective ();
\r
1050 if (is_identifier_start_character(c)){
\r
1056 if (is_identifier_part_character(c)){
\r
1060 bool v = eval_val (s.Substring (0, j));
\r
1061 s = s.Substring (j);
\r
1064 bool vv = eval_val (s);
\r
1069 Error_InvalidDirective ();
\r
1073 bool pp_unary (ref string s)
\r
1076 int len = s.Length;
\r
1079 if (s [0] == '!'){
\r
1080 if (len > 1 && s [1] == '='){
\r
1081 Error_InvalidDirective ();
\r
1084 s = s.Substring (1);
\r
1085 return ! pp_primary (ref s);
\r
1087 return pp_primary (ref s);
\r
1089 Error_InvalidDirective ();
\r
1094 bool pp_eq (ref string s)
\r
1096 bool va = pp_unary (ref s);
\r
1099 int len = s.Length;
\r
1101 if (s [0] == '='){
\r
1102 if (len > 2 && s [1] == '='){
\r
1103 s = s.Substring (2);
\r
1104 return va == pp_unary (ref s);
\r
1106 Error_InvalidDirective ();
\r
1109 } else if (s [0] == '!' && len > 1 && s [1] == '='){
\r
1110 s = s.Substring (2);
\r
1112 return va != pp_unary (ref s);
\r
1121 bool pp_and (ref string s)
\r
1123 bool va = pp_eq (ref s);
\r
1126 int len = s.Length;
\r
1128 if (s [0] == '&'){
\r
1129 if (len > 2 && s [1] == '&'){
\r
1130 s = s.Substring (2);
\r
1131 return va && pp_eq (ref s);
\r
1133 Error_InvalidDirective ();
\r
1142 // Evaluates an expression for `#if' or `#elif'
\r
1144 bool pp_expr (ref string s)
\r
1146 bool va = pp_and (ref s);
\r
1149 int len = s.Length;
\r
1154 if (len > 2 && s [1] == '|'){
\r
1155 s = s.Substring (2);
\r
1156 return va || pp_and (ref s);
\r
1158 Error_InvalidDirective ();
\r
1167 bool eval (string s)
\r
1169 bool v = pp_expr (ref s);
\r
1171 if (s.Length != 0){
\r
1172 Error_InvalidDirective ();
\r
1179 void Error_InvalidDirective ()
\r
1181 Report.Error (1517, Location, "Invalid pre-processor directive");
\r
1184 void Error_UnexpectedDirective (string extra)
\r
1188 "Unexpected processor directive (" + extra + ")");
\r
1191 void Error_TokensSeen ()
\r
1195 "Cannot define or undefine pre-processor symbols after a token in the file");
\r
1199 // if true, then the code continues processing the code
\r
1200 // if false, the code stays in a loop until another directive is
\r
1203 bool handle_preprocessing_directive (bool caller_is_taking)
\r
1205 char [] blank = { ' ', '\t' };
\r
1208 get_cmd_arg (out cmd, out arg);
\r
1210 // Eat any trailing whitespaces and single-line comments
\r
1211 if (arg.IndexOf ("//") != -1)
\r
1212 arg = arg.Substring (0, arg.IndexOf ("//"));
\r
1213 arg = arg.TrimEnd (' ', '\t');
\r
1216 // The first group of pre-processing instructions is always processed
\r
1220 if (!PreProcessLine (arg))
\r
1223 "Argument to #line directive is missing or invalid");
\r
1231 goto case "endif";
\r
1235 Error_InvalidDirective ();
\r
1238 bool taking = false;
\r
1239 if (ifstack == null)
\r
1240 ifstack = new Stack ();
\r
1242 if (ifstack.Count == 0){
\r
1245 int state = (int) ifstack.Peek ();
\r
1246 if ((state & TAKING) != 0)
\r
1250 if (eval (arg) && taking){
\r
1251 ifstack.Push (TAKING | TAKEN_BEFORE | PARENT_TAKING);
\r
1254 ifstack.Push (taking ? PARENT_TAKING : 0);
\r
1259 if (ifstack == null || ifstack.Count == 0){
\r
1260 Error_UnexpectedDirective ("no #if for this #endif");
\r
1264 if (ifstack.Count == 0)
\r
1267 int state = (int) ifstack.Peek ();
\r
1269 if ((state & TAKING) != 0)
\r
1277 if (ifstack == null || ifstack.Count == 0){
\r
1278 Error_UnexpectedDirective ("no #if for this #elif");
\r
1281 int state = (int) ifstack.Peek ();
\r
1283 if ((state & ELSE_SEEN) != 0){
\r
1284 Error_UnexpectedDirective ("#elif not valid after #else");
\r
1288 if ((state & (TAKEN_BEFORE | TAKING)) != 0)
\r
1291 if (eval (arg) && ((state & PARENT_TAKING) != 0)){
\r
1292 state = (int) ifstack.Pop ();
\r
1293 ifstack.Push (state | TAKING | TAKEN_BEFORE);
\r
1300 if (ifstack == null || ifstack.Count == 0){
\r
1303 "Unexpected processor directive (no #if for this #else)");
\r
1306 int state = (int) ifstack.Peek ();
\r
1308 if ((state & ELSE_SEEN) != 0){
\r
1309 Error_UnexpectedDirective ("#else within #else");
\r
1314 ifstack.Push (state | ELSE_SEEN);
\r
1316 if ((state & TAKEN_BEFORE) == 0){
\r
1317 if ((state & PARENT_TAKING) != 0)
\r
1327 // These are only processed if we are in a `taking' block
\r
1329 if (!caller_is_taking)
\r
1334 if (any_token_seen){
\r
1335 Error_TokensSeen ();
\r
1338 PreProcessDefinition (true, arg);
\r
1342 if (any_token_seen){
\r
1343 Error_TokensSeen ();
\r
1346 PreProcessDefinition (false, arg);
\r
1350 Report.Error (1029, Location, "#error: '" + arg + "'");
\r
1354 Report.Warning (1030, Location, "#warning: '" + arg + "'");
\r
1358 Report.Error (1024, Location, "Preprocessor directive expected (got: " + cmd + ")");
\r
1363 private int consume_string(bool quoted)
\r
1366 string_builder.Length = 0;
\r
1368 while ((c = getChar ()) != -1){
\r
1370 if (quoted && peekChar () == '"'){
\r
1371 string_builder.Append ((char) c);
\r
1375 val = string_builder.ToString ();
\r
1376 return Token.LITERAL_STRING;
\r
1380 if (c == '\n' && !quoted) {
\r
1381 Report.Error(1010, Location, "Newline in constant");
\r
1387 return Token.ERROR;
\r
1389 string_builder.Append ((char) c);
\r
1392 Report.Error (1039, Location, "Unterminated string literal");
\r
1396 private int consume_identifier(int c, bool quoted)
\r
1398 id_builder.Length = 0;
\r
1400 id_builder.Append ((char) c);
\r
1402 while ((c = peekChar ()) != -1) {
\r
1403 if (is_identifier_part_character ((char) c)){
\r
1404 id_builder.Append ((char)getChar ());
\r
1410 string ids = id_builder.ToString ();
\r
1412 if (!is_keyword (ids) || quoted) {
\r
1414 if (ids.Length > 512){
\r
1417 "Identifier too long (limit is 512 chars)");
\r
1419 return Token.IDENTIFIER;
\r
1422 // true, false and null are in the hash anyway.
\r
1423 return GetKeyword (ids);
\r
1426 public int xtoken ()
\r
1429 bool doread = false;
\r
1433 // optimization: eliminate col and implement #directive semantic correctly.
\r
1434 for (;(c = getChar ()) != -1; col++) {
\r
1435 if (is_identifier_start_character((char)c)){
\r
1436 tokens_seen = true;
\r
1437 return consume_identifier(c, false);
\r
1441 tokens_seen = true;
\r
1442 if (Char.IsDigit ((char) peekChar ()))
\r
1443 return is_number (c);
\r
1447 if (Char.IsDigit ((char) c)){
\r
1448 tokens_seen = true;
\r
1449 return is_number (c);
\r
1452 // Handle double-slash comments.
\r
1454 int d = peekChar ();
\r
1458 while ((d = getChar ()) != -1 && (d != '\n'))
\r
1463 any_token_seen |= tokens_seen;
\r
1464 tokens_seen = false;
\r
1466 } else if (d == '*'){
\r
1469 while ((d = getChar ()) != -1){
\r
1470 if (d == '*' && peekChar () == '/'){
\r
1479 any_token_seen |= tokens_seen;
\r
1480 tokens_seen = false;
\r
1487 /* For now, ignore pre-processor commands */
\r
1488 // FIXME: In C# the '#' is not limited to appear
\r
1489 // on the first column.
\r
1490 if (c == '#' && !tokens_seen){
\r
1495 cont = handle_preprocessing_directive (cont);
\r
1503 bool skipping = false;
\r
1504 for (;(c = getChar ()) != -1; col++){
\r
1510 } else if (c == ' ' || c == '\t' || c == '\v' || c == '\r')
\r
1512 else if (c != '#')
\r
1514 if (c == '#' && !skipping)
\r
1517 any_token_seen |= tokens_seen;
\r
1518 tokens_seen = false;
\r
1520 Report.Error (1027, Location, "#endif/#endregion expected");
\r
1524 if ((t = is_punct ((char)c, ref doread)) != Token.ERROR){
\r
1525 tokens_seen = true;
\r
1534 return consume_string(false);
\r
1539 tokens_seen = true;
\r
1541 error_details = "Empty character literal";
\r
1542 Report.Error (1011, Location, error_details);
\r
1543 return Token.ERROR;
\r
1547 return Token.ERROR;
\r
1548 val = new System.Char ();
\r
1553 error_details = "Too many characters in character literal";
\r
1554 Report.Error (1012, Location, error_details);
\r
1556 // Try to recover, read until newline or next "'"
\r
1557 while ((c = getChar ()) != -1){
\r
1558 if (c == '\n' || c == '\'')
\r
1562 return Token.ERROR;
\r
1564 return Token.LITERAL_CHARACTER;
\r
1572 any_token_seen |= tokens_seen;
\r
1573 tokens_seen = false;
\r
1577 if (c == ' ' || c == '\t' || c == '\f' || c == '\v' || c == '\r'){
\r
1579 col = (((col + 8) / 8) * 8) - 1;
\r
1586 tokens_seen = true;
\r
1587 return consume_string(true);
\r
1588 } else if (is_identifier_start_character((char) c)){
\r
1589 return consume_identifier(c, true);
\r
1591 Report.Error(1033, Location, "'@' must be followed by string constant or identifier");
\r
1595 error_details = ((char)c).ToString ();
\r
1597 return Token.ERROR;
\r
1600 if (ifstack != null && ifstack.Count >= 1)
\r
1601 Report.Error (1027, Location, "#endif/#endregion expected");
\r