1 /* Parser for GNU CHILL (CCITT High-Level Language) -*- C -*-
2 Copyright (C) 1992, 1993, 1995 Free Software Foundation, Inc.
4 This file is part of GDB.
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2 of the License, or
9 (at your option) any later version.
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, write to the Free Software
18 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
20 /* Parse a Chill expression from text in a string,
21 and return the result as a struct expression pointer.
22 That structure contains arithmetic operations in reverse polish,
23 with constants represented by operations that are followed by special data.
24 See expression.h for the details of the format.
25 What is important here is that it can be built up sequentially
26 during the process of parsing; the lower levels of the tree always
27 come first in the result.
29 Note that malloc's and realloc's in this file are transformed to
30 xmalloc and xrealloc respectively by the same sed command in the
31 makefile that remaps any other malloc/realloc inserted by the parser
32 generator. Doing this with #defines and trying to control the interaction
33 with include files (<malloc.h> and <stdlib.h> for example) just became
34 too messy, particularly when such includes can be inserted at random
35 times by the parser generator.
37 Also note that the language accepted by this parser is more liberal
38 than the one accepted by an actual Chill compiler. For example, the
39 language rule that a simple name string can not be one of the reserved
40 simple name strings is not enforced (e.g "case" is not treated as a
41 reserved name). Another example is that Chill is a strongly typed
42 language, and certain expressions that violate the type constraints
43 may still be evaluated if gdb can do so in a meaningful manner, while
44 such expressions would be rejected by the compiler. The reason for
45 this more liberal behavior is the philosophy that the debugger
46 is intended to be a tool that is used by the programmer when things
47 go wrong, and as such, it should provide as few artificial barriers
48 to it's use as possible. If it can do something meaningful, even
49 something that violates language contraints that are enforced by the
50 compiler, it should do so without complaint.
57 #include "expression.h"
60 #include "parser-defs.h"
62 #include "bfd.h" /* Required by objfiles.h. */
63 #include "symfile.h" /* Required by objfiles.h. */
64 #include "objfiles.h" /* For have_full_symbols and have_partial_symbols */
70 unsigned LONGEST ulval
;
85 /* '\001' ... '\xff' come first. */
91 GENERAL_PROCEDURE_NAME
,
94 CHARACTER_STRING_LITERAL
,
140 /* Forward declarations. */
141 static void parse_expr ();
142 static void parse_primval ();
143 static void parse_untyped_expr ();
144 static int parse_opt_untyped_expr ();
145 static void parse_if_expression_body
PARAMS((void));
146 static void write_lower_upper_value
PARAMS ((enum exp_opcode
, struct type
*));
147 static enum ch_terminal
ch_lex ();
149 #define MAX_LOOK_AHEAD 2
150 static enum ch_terminal terminal_buffer
[MAX_LOOK_AHEAD
+1] = {
151 TOKEN_NOT_READ
, TOKEN_NOT_READ
, TOKEN_NOT_READ
};
152 static YYSTYPE yylval
;
153 static YYSTYPE val_buffer
[MAX_LOOK_AHEAD
+1];
155 /*int current_token, lookahead_token;*/
160 static enum ch_terminal
163 if (terminal_buffer
[0] == TOKEN_NOT_READ
)
165 terminal_buffer
[0] = ch_lex ();
166 val_buffer
[0] = yylval
;
168 return terminal_buffer
[0];
170 #define PEEK_LVAL() val_buffer[0]
171 #define PEEK_TOKEN1() peek_token_(1)
172 #define PEEK_TOKEN2() peek_token_(2)
173 static enum ch_terminal
177 if (i
> MAX_LOOK_AHEAD
)
178 fatal ("internal error - too much lookahead");
179 if (terminal_buffer
[i
] == TOKEN_NOT_READ
)
181 terminal_buffer
[i
] = ch_lex ();
182 val_buffer
[i
] = yylval
;
184 return terminal_buffer
[i
];
188 pushback_token (code
, node
)
189 enum ch_terminal code
;
193 if (terminal_buffer
[MAX_LOOK_AHEAD
] != TOKEN_NOT_READ
)
194 fatal ("internal error - cannot pushback token");
195 for (i
= MAX_LOOK_AHEAD
; i
> 0; i
--)
197 terminal_buffer
[i
] = terminal_buffer
[i
- 1];
198 val_buffer
[i
] = val_buffer
[i
- 1];
200 terminal_buffer
[0] = code
;
201 val_buffer
[0] = node
;
208 for (i
= 0; i
< MAX_LOOK_AHEAD
; i
++)
210 terminal_buffer
[i
] = terminal_buffer
[i
+1];
211 val_buffer
[i
] = val_buffer
[i
+1];
213 terminal_buffer
[MAX_LOOK_AHEAD
] = TOKEN_NOT_READ
;
215 #define FORWARD_TOKEN() forward_token_()
217 /* Skip the next token.
218 if it isn't TOKEN, the parser is broken. */
222 enum ch_terminal token
;
224 if (PEEK_TOKEN() != token
)
227 sprintf (buf
, "internal parser error - expected token %d", (int)token
);
235 enum ch_terminal token
;
237 if (PEEK_TOKEN() != token
)
243 /* return 0 if expected token was not found,
247 expect(token
, message
)
248 enum ch_terminal token
;
251 if (PEEK_TOKEN() != token
)
255 else if (token
< 256)
256 error ("syntax error - expected a '%c' here \"%s\"", token
, lexptr
);
258 error ("syntax error");
268 parse_opt_name_string (allow_all
)
269 int allow_all
; /* 1 if ALL is allowed as a postfix */
271 int token
= PEEK_TOKEN();
275 if (token
== ALL
&& allow_all
)
286 token
= PEEK_TOKEN();
290 token
= PEEK_TOKEN();
291 if (token
== ALL
&& allow_all
)
292 return get_identifier3(IDENTIFIER_POINTER (name
), "!", "*");
296 error ("'%s!' is not followed by an identifier",
297 IDENTIFIER_POINTER (name
));
300 name
= get_identifier3(IDENTIFIER_POINTER(name
),
301 "!", IDENTIFIER_POINTER(PEEK_LVAL()));
306 parse_simple_name_string ()
308 int token
= PEEK_TOKEN();
312 error ("expected a name here");
313 return error_mark_node
;
323 tree name
= parse_opt_name_string (0);
327 error ("expected a name string here");
328 return error_mark_node
;
331 /* Matches: <name_string>
332 Returns if pass 1: the identifier.
333 Returns if pass 2: a decl or value for identifier. */
338 tree name
= parse_name_string ();
339 if (pass
== 1 || ignoring
)
343 tree decl
= lookup_name (name
);
344 if (decl
== NULL_TREE
)
346 error ("`%s' undeclared", IDENTIFIER_POINTER (name
));
347 return error_mark_node
;
349 else if (TREE_CODE (TREE_TYPE (decl
)) == ERROR_MARK
)
350 return error_mark_node
;
351 else if (TREE_CODE (decl
) == CONST_DECL
)
352 return DECL_INITIAL (decl
);
353 else if (TREE_CODE (TREE_TYPE (decl
)) == REFERENCE_TYPE
)
354 return convert_from_reference (decl
);
363 pushback_paren_expr (expr
)
366 if (pass
== 1 && !ignoring
)
367 expr
= build1 (PAREN_EXPR
, NULL_TREE
, expr
);
368 pushback_token (EXPR
, expr
);
372 /* Matches: <case label> */
377 if (check_token (ELSE
))
378 error ("ELSE in tuples labels not implemented");
379 /* Does not handle the case of a mode name. FIXME */
381 if (check_token (':'))
384 write_exp_elt_opcode (BINOP_RANGE
);
389 parse_opt_untyped_expr ()
391 switch (PEEK_TOKEN ())
398 parse_untyped_expr ();
412 /* Parse NAME '(' MODENAME ')'. */
420 if (PEEK_TOKEN () != TYPENAME
)
421 error ("expect MODENAME here `%s'", lexptr
);
422 type
= PEEK_LVAL().tsym
.type
;
429 parse_mode_or_normal_call ()
434 if (PEEK_TOKEN () == TYPENAME
)
436 type
= PEEK_LVAL().tsym
.type
;
448 /* Parse something that looks like a function call.
449 Assume we have parsed the function, and are at the '('. */
456 /* This is to save the value of arglist_len
457 being accumulated for each dimension. */
459 if (parse_opt_untyped_expr ())
461 int tok
= PEEK_TOKEN ();
463 if (tok
== UP
|| tok
== ':')
467 expect (')', "expected ')' to terminate slice");
469 write_exp_elt_opcode (tok
== UP
? TERNOP_SLICE_COUNT
473 while (check_token (','))
475 parse_untyped_expr ();
482 arg_count
= end_arglist ();
483 write_exp_elt_opcode (MULTI_SUBSCRIPT
);
484 write_exp_elt_longcst (arg_count
);
485 write_exp_elt_opcode (MULTI_SUBSCRIPT
);
489 parse_named_record_element ()
493 label
= PEEK_LVAL ().sval
;
494 expect (FIELD_NAME
, "expected a field name here `%s'", lexptr
);
495 if (check_token (','))
496 parse_named_record_element ();
497 else if (check_token (':'))
500 error ("syntax error near `%s' in named record tuple element", lexptr
);
501 write_exp_elt_opcode (OP_LABELED
);
502 write_exp_string (label
);
503 write_exp_elt_opcode (OP_LABELED
);
506 /* Returns one or nore TREE_LIST nodes, in reverse order. */
509 parse_tuple_element ()
511 if (PEEK_TOKEN () == FIELD_NAME
)
513 /* Parse a labelled structure tuple. */
514 parse_named_record_element ();
518 if (check_token ('('))
520 if (check_token ('*'))
522 expect (')', "missing ')' after '*' case label list");
523 error ("(*) not implemented in case label list");
528 while (check_token (','))
531 write_exp_elt_opcode (BINOP_COMMA
);
537 parse_untyped_expr ();
538 if (check_token (':'))
540 /* A powerset range or a labeled Array. */
541 parse_untyped_expr ();
542 write_exp_elt_opcode (BINOP_RANGE
);
546 /* Matches: a COMMA-separated list of tuple elements.
547 Returns a list (of TREE_LIST nodes). */
549 parse_opt_element_list ()
552 if (PEEK_TOKEN () == ']')
556 parse_tuple_element ();
558 if (PEEK_TOKEN () == ']')
560 if (!check_token (','))
561 error ("bad syntax in tuple");
565 /* Parses: '[' elements ']'
566 If modename is non-NULL it prefixed the tuple. */
574 parse_opt_element_list ();
575 expect (']', "missing ']' after tuple");
576 write_exp_elt_opcode (OP_ARRAY
);
577 write_exp_elt_longcst ((LONGEST
) 0);
578 write_exp_elt_longcst ((LONGEST
) end_arglist () - 1);
579 write_exp_elt_opcode (OP_ARRAY
);
582 write_exp_elt_opcode (UNOP_CAST
);
583 write_exp_elt_type (mode
);
584 write_exp_elt_opcode (UNOP_CAST
);
594 switch (PEEK_TOKEN ())
596 case INTEGER_LITERAL
:
597 case CHARACTER_LITERAL
:
598 write_exp_elt_opcode (OP_LONG
);
599 write_exp_elt_type (PEEK_LVAL ().typed_val
.type
);
600 write_exp_elt_longcst (PEEK_LVAL ().typed_val
.val
);
601 write_exp_elt_opcode (OP_LONG
);
604 case BOOLEAN_LITERAL
:
605 write_exp_elt_opcode (OP_BOOL
);
606 write_exp_elt_longcst ((LONGEST
) PEEK_LVAL ().ulval
);
607 write_exp_elt_opcode (OP_BOOL
);
611 write_exp_elt_opcode (OP_DOUBLE
);
612 write_exp_elt_type (builtin_type_double
);
613 write_exp_elt_dblcst (PEEK_LVAL ().dval
);
614 write_exp_elt_opcode (OP_DOUBLE
);
617 case EMPTINESS_LITERAL
:
618 write_exp_elt_opcode (OP_LONG
);
619 write_exp_elt_type (lookup_pointer_type (builtin_type_void
));
620 write_exp_elt_longcst (0);
621 write_exp_elt_opcode (OP_LONG
);
624 case CHARACTER_STRING_LITERAL
:
625 write_exp_elt_opcode (OP_STRING
);
626 write_exp_string (PEEK_LVAL ().sval
);
627 write_exp_elt_opcode (OP_STRING
);
630 case BIT_STRING_LITERAL
:
631 write_exp_elt_opcode (OP_BITSTRING
);
632 write_exp_bitstring (PEEK_LVAL ().sval
);
633 write_exp_elt_opcode (OP_BITSTRING
);
638 /* This is pseudo-Chill, similar to C's '(TYPE[])EXPR'
639 which casts to an artificial array. */
642 if (PEEK_TOKEN () != TYPENAME
)
643 error ("missing MODENAME after ARRAY()");
644 type
= PEEK_LVAL().tsym
.type
;
648 expect (')', "missing right parenthesis");
649 type
= create_array_type ((struct type
*) NULL
, type
,
650 create_range_type ((struct type
*) NULL
,
651 builtin_type_int
, 0, 0));
652 TYPE_ARRAY_UPPER_BOUND_TYPE(type
) = BOUND_CANNOT_BE_DETERMINED
;
653 write_exp_elt_opcode (UNOP_CAST
);
654 write_exp_elt_type (type
);
655 write_exp_elt_opcode (UNOP_CAST
);
667 expect (')', "missing right parenthesis");
672 case GENERAL_PROCEDURE_NAME
:
674 write_exp_elt_opcode (OP_VAR_VALUE
);
675 write_exp_elt_block (NULL
);
676 write_exp_elt_sym (PEEK_LVAL ().ssym
.sym
);
677 write_exp_elt_opcode (OP_VAR_VALUE
);
680 case GDB_VARIABLE
: /* gdb specific */
685 write_exp_elt_opcode (UNOP_CAST
);
686 write_exp_elt_type (builtin_type_int
);
687 write_exp_elt_opcode (UNOP_CAST
);
689 case PRED
: op_name
= "PRED"; goto unimplemented_unary_builtin
;
690 case SUCC
: op_name
= "SUCC"; goto unimplemented_unary_builtin
;
691 case ABS
: op_name
= "ABS"; goto unimplemented_unary_builtin
;
692 case CARD
: op_name
= "CARD"; goto unimplemented_unary_builtin
;
693 case MAX_TOKEN
: op_name
= "MAX"; goto unimplemented_unary_builtin
;
694 case MIN_TOKEN
: op_name
= "MIN"; goto unimplemented_unary_builtin
;
695 unimplemented_unary_builtin
:
697 error ("not implemented: %s builtin function", op_name
);
701 write_exp_elt_opcode (UNOP_ADDR
);
704 type
= parse_mode_or_normal_call ();
706 { write_exp_elt_opcode (OP_LONG
);
707 write_exp_elt_type (builtin_type_int
);
708 CHECK_TYPEDEF (type
);
709 write_exp_elt_longcst ((LONGEST
) TYPE_LENGTH (type
));
710 write_exp_elt_opcode (OP_LONG
);
713 write_exp_elt_opcode (UNOP_SIZEOF
);
722 type
= parse_mode_or_normal_call ();
723 write_lower_upper_value (op
, type
);
727 write_exp_elt_opcode (UNOP_LENGTH
);
730 type
= PEEK_LVAL ().tsym
.type
;
732 switch (PEEK_TOKEN())
740 expect (')', "missing right parenthesis");
741 write_exp_elt_opcode (UNOP_CAST
);
742 write_exp_elt_type (type
);
743 write_exp_elt_opcode (UNOP_CAST
);
746 error ("typename in invalid context");
751 error ("invalid expression syntax at `%s'", lexptr
);
755 switch (PEEK_TOKEN ())
758 write_exp_elt_opcode (STRUCTOP_STRUCT
);
759 write_exp_string (PEEK_LVAL ().sval
);
760 write_exp_elt_opcode (STRUCTOP_STRUCT
);
765 if (PEEK_TOKEN () == TYPENAME
)
767 type
= PEEK_LVAL ().tsym
.type
;
768 write_exp_elt_opcode (UNOP_CAST
);
769 write_exp_elt_type (lookup_pointer_type (type
));
770 write_exp_elt_opcode (UNOP_CAST
);
773 write_exp_elt_opcode (UNOP_IND
);
778 case CHARACTER_STRING_LITERAL
:
779 case CHARACTER_LITERAL
:
780 case BIT_STRING_LITERAL
:
781 /* Handle string repetition. (See comment in parse_operand5.) */
783 write_exp_elt_opcode (MULTI_SUBSCRIPT
);
784 write_exp_elt_longcst (1);
785 write_exp_elt_opcode (MULTI_SUBSCRIPT
);
796 if (check_token (RECEIVE
))
799 error ("not implemented: RECEIVE expression");
801 else if (check_token (POINTER
))
804 write_exp_elt_opcode (UNOP_ADDR
);
814 /* We are supposed to be looking for a <string repetition operator>,
815 but in general we can't distinguish that from a parenthesized
816 expression. This is especially difficult if we allow the
817 string operand to be a constant expression (as requested by
818 some users), and not just a string literal.
819 Consider: LPRN expr RPRN LPRN expr RPRN
820 Is that a function call or string repetition?
821 Instead, we handle string repetition in parse_primval,
822 and build_generalized_call. */
823 switch (PEEK_TOKEN())
825 case NOT
: op
= UNOP_LOGICAL_NOT
; break;
826 case '-': op
= UNOP_NEG
; break;
834 write_exp_elt_opcode (op
);
844 switch (PEEK_TOKEN())
846 case '*': op
= BINOP_MUL
; break;
847 case '/': op
= BINOP_DIV
; break;
848 case MOD
: op
= BINOP_MOD
; break;
849 case REM
: op
= BINOP_REM
; break;
855 write_exp_elt_opcode (op
);
866 switch (PEEK_TOKEN())
868 case '+': op
= BINOP_ADD
; break;
869 case '-': op
= BINOP_SUB
; break;
870 case SLASH_SLASH
: op
= BINOP_CONCAT
; break;
876 write_exp_elt_opcode (op
);
887 if (check_token (IN
))
890 write_exp_elt_opcode (BINOP_IN
);
894 switch (PEEK_TOKEN())
896 case '>': op
= BINOP_GTR
; break;
897 case GEQ
: op
= BINOP_GEQ
; break;
898 case '<': op
= BINOP_LESS
; break;
899 case LEQ
: op
= BINOP_LEQ
; break;
900 case '=': op
= BINOP_EQUAL
; break;
901 case NOTEQUAL
: op
= BINOP_NOTEQUAL
; break;
907 write_exp_elt_opcode (op
);
919 switch (PEEK_TOKEN())
921 case LOGAND
: op
= BINOP_BITWISE_AND
; break;
922 case ANDIF
: op
= BINOP_LOGICAL_AND
; break;
928 write_exp_elt_opcode (op
);
939 switch (PEEK_TOKEN())
941 case LOGIOR
: op
= BINOP_BITWISE_IOR
; break;
942 case LOGXOR
: op
= BINOP_BITWISE_XOR
; break;
943 case ORIF
: op
= BINOP_LOGICAL_OR
; break;
949 write_exp_elt_opcode (op
);
957 if (check_token (GDB_ASSIGNMENT
))
960 write_exp_elt_opcode (BINOP_ASSIGN
);
965 parse_then_alternative ()
967 expect (THEN
, "missing 'THEN' in 'IF' expression");
972 parse_else_alternative ()
974 if (check_token (ELSIF
))
975 parse_if_expression_body ();
976 else if (check_token (ELSE
))
979 error ("missing ELSE/ELSIF in IF expression");
982 /* Matches: <boolean expression> <then alternative> <else alternative> */
985 parse_if_expression_body ()
988 parse_then_alternative ();
989 parse_else_alternative ();
990 write_exp_elt_opcode (TERNOP_COND
);
994 parse_if_expression ()
997 parse_if_expression_body ();
998 expect (FI
, "missing 'FI' at end of conditional expression");
1001 /* An <untyped_expr> is a superset of <expr>. It also includes
1002 <conditional expressions> and untyped <tuples>, whose types
1003 are not given by their constituents. Hence, these are only
1004 allowed in certain contexts that expect a certain type.
1005 You should call convert() to fix up the <untyped_expr>. */
1008 parse_untyped_expr ()
1010 switch (PEEK_TOKEN())
1013 parse_if_expression ();
1016 error ("not implemented: CASE expression");
1018 switch (PEEK_TOKEN1())
1026 parse_untyped_expr ();
1027 expect (')', "missing ')'");
1040 terminal_buffer
[0] = TOKEN_NOT_READ
;
1041 if (PEEK_TOKEN () == TYPENAME
&& PEEK_TOKEN1 () == END_TOKEN
)
1043 write_exp_elt_opcode(OP_TYPE
);
1044 write_exp_elt_type(PEEK_LVAL ().tsym
.type
);
1045 write_exp_elt_opcode(OP_TYPE
);
1050 if (terminal_buffer
[0] != END_TOKEN
)
1052 if (comma_terminates
&& terminal_buffer
[0] == ',')
1053 lexptr
--; /* Put the comma back. */
1055 error ("Junk after end of expression.");
1061 /* Implementation of a dynamically expandable buffer for processing input
1062 characters acquired through lexptr and building a value to return in
1065 static char *tempbuf
; /* Current buffer contents */
1066 static int tempbufsize
; /* Size of allocated buffer */
1067 static int tempbufindex
; /* Current index into buffer */
1069 #define GROWBY_MIN_SIZE 64 /* Minimum amount to grow buffer by */
1071 #define CHECKBUF(size) \
1073 if (tempbufindex + (size) >= tempbufsize) \
1075 growbuf_by_size (size); \
1079 /* Grow the static temp buffer if necessary, including allocating the first one
1083 growbuf_by_size (count
)
1088 growby
= max (count
, GROWBY_MIN_SIZE
);
1089 tempbufsize
+= growby
;
1090 if (tempbuf
== NULL
)
1092 tempbuf
= (char *) malloc (tempbufsize
);
1096 tempbuf
= (char *) realloc (tempbuf
, tempbufsize
);
1100 /* Try to consume a simple name string token. If successful, returns
1101 a pointer to a nullbyte terminated copy of the name that can be used
1102 in symbol table lookups. If not successful, returns NULL. */
1105 match_simple_name_string ()
1107 char *tokptr
= lexptr
;
1109 if (isalpha (*tokptr
) || *tokptr
== '_')
1114 } while (isalnum (*tokptr
) || (*tokptr
== '_'));
1115 yylval
.sval
.ptr
= lexptr
;
1116 yylval
.sval
.length
= tokptr
- lexptr
;
1118 result
= copy_name (yylval
.sval
);
1124 /* Start looking for a value composed of valid digits as set by the base
1125 in use. Note that '_' characters are valid anywhere, in any quantity,
1126 and are simply ignored. Since we must find at least one valid digit,
1127 or reject this token as an integer literal, we keep track of how many
1128 digits we have encountered. */
1131 decode_integer_value (base
, tokptrptr
, ivalptr
)
1136 char *tokptr
= *tokptrptr
;
1140 while (*tokptr
!= '\0')
1144 temp
= tolower (temp
);
1150 case '0': case '1': case '2': case '3': case '4':
1151 case '5': case '6': case '7': case '8': case '9':
1154 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1170 /* Found something not in domain for current base. */
1171 tokptr
--; /* Unconsume what gave us indigestion. */
1176 /* If we didn't find any digits, then we don't have a valid integer
1177 value, so reject the entire token. Otherwise, update the lexical
1178 scan pointer, and return non-zero for success. */
1186 *tokptrptr
= tokptr
;
1192 decode_integer_literal (valptr
, tokptrptr
)
1196 char *tokptr
= *tokptrptr
;
1199 int explicit_base
= 0;
1201 /* Look for an explicit base specifier, which is optional. */
1234 /* If we found an explicit base ensure that the character after the
1235 explicit base is a single quote. */
1237 if (explicit_base
&& (*tokptr
++ != '\''))
1242 /* Attempt to decode whatever follows as an integer value in the
1243 indicated base, updating the token pointer in the process and
1244 computing the value into ival. Also, if we have an explicit
1245 base, then the next character must not be a single quote, or we
1246 have a bitstring literal, so reject the entire token in this case.
1247 Otherwise, update the lexical scan pointer, and return non-zero
1250 if (!decode_integer_value (base
, &tokptr
, &ival
))
1254 else if (explicit_base
&& (*tokptr
== '\''))
1261 *tokptrptr
= tokptr
;
1266 /* If it wasn't for the fact that floating point values can contain '_'
1267 characters, we could just let strtod do all the hard work by letting it
1268 try to consume as much of the current token buffer as possible and
1269 find a legal conversion. Unfortunately we need to filter out the '_'
1270 characters before calling strtod, which we do by copying the other
1271 legal chars to a local buffer to be converted. However since we also
1272 need to keep track of where the last unconsumed character in the input
1273 buffer is, we have transfer only as many characters as may compose a
1274 legal floating point value. */
1276 static enum ch_terminal
1277 match_float_literal ()
1279 char *tokptr
= lexptr
;
1283 extern double strtod ();
1285 /* Make local buffer in which to build the string to convert. This is
1286 required because underscores are valid in chill floating point numbers
1287 but not in the string passed to strtod to convert. The string will be
1288 no longer than our input string. */
1290 copy
= buf
= (char *) alloca (strlen (tokptr
) + 1);
1292 /* Transfer all leading digits to the conversion buffer, discarding any
1295 while (isdigit (*tokptr
) || *tokptr
== '_')
1304 /* Now accept either a '.', or one of [eEdD]. Dot is legal regardless
1305 of whether we found any leading digits, and we simply accept it and
1306 continue on to look for the fractional part and/or exponent. One of
1307 [eEdD] is legal only if we have seen digits, and means that there
1308 is no fractional part. If we find neither of these, then this is
1309 not a floating point number, so return failure. */
1314 /* Accept and then look for fractional part and/or exponent. */
1327 goto collect_exponent
;
1335 /* We found a '.', copy any fractional digits to the conversion buffer, up
1336 to the first nondigit, non-underscore character. */
1338 while (isdigit (*tokptr
) || *tokptr
== '_')
1347 /* Look for an exponent, which must start with one of [eEdD]. If none
1348 is found, jump directly to trying to convert what we have collected
1365 /* Accept an optional '-' or '+' following one of [eEdD]. */
1368 if (*tokptr
== '+' || *tokptr
== '-')
1370 *copy
++ = *tokptr
++;
1373 /* Now copy an exponent into the conversion buffer. Note that at the
1374 moment underscores are *not* allowed in exponents. */
1376 while (isdigit (*tokptr
))
1378 *copy
++ = *tokptr
++;
1381 /* If we transfered any chars to the conversion buffer, try to interpret its
1382 contents as a floating point value. If any characters remain, then we
1383 must not have a valid floating point string. */
1389 dval
= strtod (buf
, ©
);
1394 return (FLOAT_LITERAL
);
1400 /* Recognize a string literal. A string literal is a sequence
1401 of characters enclosed in matching single or double quotes, except that
1402 a single character inside single quotes is a character literal, which
1403 we reject as a string literal. To embed the terminator character inside
1404 a string, it is simply doubled (I.E. "this""is""one""string") */
1406 static enum ch_terminal
1407 match_string_literal ()
1409 char *tokptr
= lexptr
;
1411 for (tempbufindex
= 0, tokptr
++; *tokptr
!= '\0'; tokptr
++)
1414 if (*tokptr
== *lexptr
)
1416 if (*(tokptr
+ 1) == *lexptr
)
1425 tempbuf
[tempbufindex
++] = *tokptr
;
1427 if (*tokptr
== '\0' /* no terminator */
1428 || (tempbufindex
== 1 && *tokptr
== '\'')) /* char literal */
1434 tempbuf
[tempbufindex
] = '\0';
1435 yylval
.sval
.ptr
= tempbuf
;
1436 yylval
.sval
.length
= tempbufindex
;
1438 return (CHARACTER_STRING_LITERAL
);
1442 /* Recognize a character literal. A character literal is single character
1443 or a control sequence, enclosed in single quotes. A control sequence
1444 is a comma separated list of one or more integer literals, enclosed
1445 in parenthesis and introduced with a circumflex character.
1447 EX: 'a' '^(7)' '^(7,8)'
1449 As a GNU chill extension, the syntax C'xx' is also recognized as a
1450 character literal, where xx is a hex value for the character.
1452 Note that more than a single character, enclosed in single quotes, is
1455 Also note that the control sequence form is not in GNU Chill since it
1456 is ambiguous with the string literal form using single quotes. I.E.
1457 is '^(7)' a character literal or a string literal. In theory it it
1458 possible to tell by context, but GNU Chill doesn't accept the control
1459 sequence form, so neither do we (for now the code is disabled).
1461 Returns CHARACTER_LITERAL if a match is found.
1464 static enum ch_terminal
1465 match_character_literal ()
1467 char *tokptr
= lexptr
;
1470 if ((*tokptr
== 'c' || *tokptr
== 'C') && (*(tokptr
+ 1) == '\''))
1472 /* We have a GNU chill extension form, so skip the leading "C'",
1473 decode the hex value, and then ensure that we have a trailing
1474 single quote character. */
1476 if (!decode_integer_value (16, &tokptr
, &ival
) || (*tokptr
!= '\''))
1482 else if (*tokptr
== '\'')
1486 /* Determine which form we have, either a control sequence or the
1487 single character form. */
1489 if ((*tokptr
== '^') && (*(tokptr
+ 1) == '('))
1491 #if 0 /* Disable, see note above. -fnf */
1492 /* Match and decode a control sequence. Return zero if we don't
1493 find a valid integer literal, or if the next unconsumed character
1494 after the integer literal is not the trailing ')'.
1495 FIXME: We currently don't handle the multiple integer literal
1498 if (!decode_integer_literal (&ival
, &tokptr
) || (*tokptr
++ != ')'))
1511 /* The trailing quote has not yet been consumed. If we don't find
1512 it, then we have no match. */
1514 if (*tokptr
++ != '\'')
1521 /* Not a character literal. */
1524 yylval
.typed_val
.val
= ival
;
1525 yylval
.typed_val
.type
= builtin_type_chill_char
;
1527 return (CHARACTER_LITERAL
);
1530 /* Recognize an integer literal, as specified in Z.200 sec 5.2.4.2.
1531 Note that according to 5.2.4.2, a single "_" is also a valid integer
1532 literal, however GNU-chill requires there to be at least one "digit"
1533 in any integer literal. */
1535 static enum ch_terminal
1536 match_integer_literal ()
1538 char *tokptr
= lexptr
;
1541 if (!decode_integer_literal (&ival
, &tokptr
))
1547 yylval
.typed_val
.val
= ival
;
1548 #if defined(CC_HAS_LONG_LONG) && defined(__STDC__)
1549 if (ival
> (LONGEST
)2147483647U || ival
< -(LONGEST
)2147483648U)
1550 yylval
.typed_val
.type
= builtin_type_long_long
;
1553 yylval
.typed_val
.type
= builtin_type_int
;
1555 return (INTEGER_LITERAL
);
1559 /* Recognize a bit-string literal, as specified in Z.200 sec 5.2.4.8
1560 Note that according to 5.2.4.8, a single "_" is also a valid bit-string
1561 literal, however GNU-chill requires there to be at least one "digit"
1562 in any bit-string literal. */
1564 static enum ch_terminal
1565 match_bitstring_literal ()
1567 register char *tokptr
= lexptr
;
1577 /* Look for the required explicit base specifier. */
1598 /* Ensure that the character after the explicit base is a single quote. */
1600 if (*tokptr
++ != '\'')
1605 while (*tokptr
!= '\0' && *tokptr
!= '\'')
1608 if (isupper (digit
))
1609 digit
= tolower (digit
);
1615 case '0': case '1': case '2': case '3': case '4':
1616 case '5': case '6': case '7': case '8': case '9':
1619 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
1624 error ("Invalid character in bitstring or integer.");
1626 if (digit
>= 1 << bits_per_char
)
1628 /* Found something not in domain for current base. */
1629 error ("Too-large digit in bitstring or integer.");
1633 /* Extract bits from digit, packing them into the bitstring byte. */
1634 int k
= TARGET_BYTE_ORDER
== BIG_ENDIAN
? bits_per_char
- 1 : 0;
1635 for (; TARGET_BYTE_ORDER
== BIG_ENDIAN
? k
>= 0 : k
< bits_per_char
;
1636 TARGET_BYTE_ORDER
== BIG_ENDIAN
? k
-- : k
++)
1639 if (digit
& (1 << k
))
1641 tempbuf
[tempbufindex
] |=
1642 (TARGET_BYTE_ORDER
== BIG_ENDIAN
)
1643 ? (1 << (HOST_CHAR_BIT
- 1 - bitoffset
))
1647 if (bitoffset
== HOST_CHAR_BIT
)
1652 tempbuf
[tempbufindex
] = 0;
1658 /* Verify that we consumed everything up to the trailing single quote,
1659 and that we found some bits (IE not just underbars). */
1661 if (*tokptr
++ != '\'')
1667 yylval
.sval
.ptr
= tempbuf
;
1668 yylval
.sval
.length
= bitcount
;
1670 return (BIT_STRING_LITERAL
);
1680 static const struct token idtokentab
[] =
1683 { "length", LENGTH
},
1694 { "max", MAX_TOKEN
},
1695 { "min", MIN_TOKEN
},
1704 { "addr", ADDR_TOKEN
},
1705 { "null", EMPTINESS_LITERAL
}
1708 static const struct token tokentab2
[] =
1710 { ":=", GDB_ASSIGNMENT
},
1711 { "//", SLASH_SLASH
},
1718 /* Read one token, getting characters through lexptr. */
1719 /* This is where we will check to make sure that the language and the
1720 operators used are compatible. */
1722 static enum ch_terminal
1726 enum ch_terminal token
;
1730 /* Skip over any leading whitespace. */
1731 while (isspace (*lexptr
))
1735 /* Look for special single character cases which can't be the first
1736 character of some other multicharacter token. */
1753 /* Look for characters which start a particular kind of multicharacter
1754 token, such as a character literal, register name, convenience
1755 variable name, string literal, etc. */
1760 /* First try to match a string literal, which is any
1761 sequence of characters enclosed in matching single or double
1762 quotes, except that a single character inside single quotes
1763 is a character literal, so we have to catch that case also. */
1764 token
= match_string_literal ();
1769 if (*lexptr
== '\'')
1771 token
= match_character_literal ();
1780 token
= match_character_literal ();
1787 yylval
.sval
.ptr
= lexptr
;
1790 } while (isalnum (*lexptr
) || *lexptr
== '_' || *lexptr
== '$');
1791 yylval
.sval
.length
= lexptr
- yylval
.sval
.ptr
;
1792 write_dollar_variable (yylval
.sval
);
1793 return GDB_VARIABLE
;
1796 /* See if it is a special token of length 2. */
1797 for (i
= 0; i
< sizeof (tokentab2
) / sizeof (tokentab2
[0]); i
++)
1799 if (STREQN (lexptr
, tokentab2
[i
].operator, 2))
1802 return (tokentab2
[i
].token
);
1805 /* Look for single character cases which which could be the first
1806 character of some other multicharacter token, but aren't, or we
1807 would already have found it. */
1817 /* Look for a float literal before looking for an integer literal, so
1818 we match as much of the input stream as possible. */
1819 token
= match_float_literal ();
1824 token
= match_bitstring_literal ();
1829 token
= match_integer_literal ();
1835 /* Try to match a simple name string, and if a match is found, then
1836 further classify what sort of name it is and return an appropriate
1837 token. Note that attempting to match a simple name string consumes
1838 the token from lexptr, so we can't back out if we later find that
1839 we can't classify what sort of name it is. */
1841 inputname
= match_simple_name_string ();
1843 if (inputname
!= NULL
)
1845 char *simplename
= (char*) alloca (strlen (inputname
) + 1);
1847 char *dptr
= simplename
, *sptr
= inputname
;
1848 for (; *sptr
; sptr
++)
1849 *dptr
++ = isupper (*sptr
) ? tolower(*sptr
) : *sptr
;
1852 /* See if it is a reserved identifier. */
1853 for (i
= 0; i
< sizeof (idtokentab
) / sizeof (idtokentab
[0]); i
++)
1855 if (STREQ (simplename
, idtokentab
[i
].operator))
1857 return (idtokentab
[i
].token
);
1861 /* Look for other special tokens. */
1862 if (STREQ (simplename
, "true"))
1865 return (BOOLEAN_LITERAL
);
1867 if (STREQ (simplename
, "false"))
1870 return (BOOLEAN_LITERAL
);
1873 sym
= lookup_symbol (inputname
, expression_context_block
,
1874 VAR_NAMESPACE
, (int *) NULL
,
1875 (struct symtab
**) NULL
);
1876 if (sym
== NULL
&& strcmp (inputname
, simplename
) != 0)
1878 sym
= lookup_symbol (simplename
, expression_context_block
,
1879 VAR_NAMESPACE
, (int *) NULL
,
1880 (struct symtab
**) NULL
);
1884 yylval
.ssym
.stoken
.ptr
= NULL
;
1885 yylval
.ssym
.stoken
.length
= 0;
1886 yylval
.ssym
.sym
= sym
;
1887 yylval
.ssym
.is_a_field_of_this
= 0; /* FIXME, C++'ism */
1888 switch (SYMBOL_CLASS (sym
))
1891 /* Found a procedure name. */
1892 return (GENERAL_PROCEDURE_NAME
);
1894 /* Found a global or local static variable. */
1895 return (LOCATION_NAME
);
1900 case LOC_REGPARM_ADDR
:
1904 case LOC_BASEREG_ARG
:
1905 if (innermost_block
== NULL
1906 || contained_in (block_found
, innermost_block
))
1908 innermost_block
= block_found
;
1910 return (LOCATION_NAME
);
1914 return (LOCATION_NAME
);
1917 yylval
.tsym
.type
= SYMBOL_TYPE (sym
);
1920 case LOC_CONST_BYTES
:
1921 case LOC_OPTIMIZED_OUT
:
1922 error ("Symbol \"%s\" names no location.", inputname
);
1926 else if (!have_full_symbols () && !have_partial_symbols ())
1928 error ("No symbol table is loaded. Use the \"file\" command.");
1932 error ("No symbol \"%s\" in current context.", inputname
);
1936 /* Catch single character tokens which are not part of some
1941 case '.': /* Not float for example. */
1943 while (isspace (*lexptr
)) lexptr
++;
1944 inputname
= match_simple_name_string ();
1950 return (ILLEGAL_TOKEN
);
1954 write_lower_upper_value (opcode
, type
)
1955 enum exp_opcode opcode
; /* Either UNOP_LOWER or UNOP_UPPER */
1959 write_exp_elt_opcode (opcode
);
1962 extern LONGEST
type_lower_upper ();
1963 struct type
*result_type
;
1964 LONGEST val
= type_lower_upper (opcode
, type
, &result_type
);
1965 write_exp_elt_opcode (OP_LONG
);
1966 write_exp_elt_type (result_type
);
1967 write_exp_elt_longcst (val
);
1968 write_exp_elt_opcode (OP_LONG
);