1 /* YACC parser for Pascal expressions, for GDB.
2 Copyright (C) 2000-2024 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 3 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, see <http://www.gnu.org/licenses/>. */
19 /* This file is derived from c-exp.y */
21 /* Parse a Pascal expression from text in a string,
22 and return the result as a struct expression pointer.
23 That structure contains arithmetic operations in reverse polish,
24 with constants represented by operations that are followed by special data.
25 See expression.h for the details of the format.
26 What is important here is that it can be built up sequentially
27 during the process of parsing; the lower levels of the tree always
28 come first in the result.
30 Note that malloc's and realloc's in this file are transformed to
31 xmalloc and xrealloc respectively by the same sed command in the
32 makefile that remaps any other malloc/realloc inserted by the parser
33 generator. Doing this with #defines and trying to control the interaction
34 with include files (<malloc.h> and <stdlib.h> for example) just became
35 too messy, particularly when such includes can be inserted at random
36 times by the parser generator. */
38 /* Known bugs or limitations:
39 - pascal string operations are not supported at all.
40 - there are some problems with boolean types.
41 - Pascal type hexadecimal constants are not supported
42 because they conflict with the internal variables format.
43 Probably also lots of other problems, less well defined PM. */
47 #include "expression.h"
49 #include "parser-defs.h"
55 #define parse_type(ps) builtin_type (ps->gdbarch ())
57 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror,
59 #define GDB_YY_REMAP_PREFIX pascal_
62 /* The state of the parser, used internally when we are parsing the
65 static struct parser_state
*pstate
= NULL
;
67 /* Depth of parentheses. */
68 static int paren_depth
;
72 static int yylex (void);
74 static void yyerror (const char *);
76 static char *uptok
(const char *, int);
78 static const char *pascal_skip_string
(const char *str
);
83 /* Although the yacc "value" of an expression is not used,
84 since the result is stored in the structure being created,
85 other node types do have values. */
102 struct symtoken ssym
;
104 const struct block
*bval
;
105 enum exp_opcode opcode
;
106 struct internalvar
*ivar
;
113 /* YYSTYPE gets defined by %union */
114 static int parse_number
(struct parser_state
*,
115 const char *, int, int, YYSTYPE *);
117 static struct type
*current_type
;
118 static int leftdiv_is_integer
;
119 static void push_current_type
(void);
120 static void pop_current_type
(void);
121 static int search_field
;
124 %type
<voidval
> exp exp1 type_exp start normal_start variable qualified_name
125 %type
<tval
> type typebase
126 /* %type <bval> block */
128 /* Fancy type parsing. */
131 %token
<typed_val_int
> INT
132 %token
<typed_val_float
> FLOAT
134 /* Both NAME and TYPENAME tokens represent symbols in the input,
135 and both convey their data as strings.
136 But a TYPENAME is a string that happens to be defined as a typedef
137 or builtin type name (such as int or char)
138 and a NAME is any other symbol.
139 Contexts where this distinction is not important can use the
140 nonterminal "name", which matches either NAME or TYPENAME. */
143 %token
<sval
> FIELDNAME
144 %token
<voidval
> COMPLETE
145 %token
<ssym
> NAME
/* BLOCKNAME defined below to give it higher precedence. */
146 %token
<tsym
> TYPENAME
148 %type
<ssym
> name_not_typename
150 /* A NAME_OR_INT is a symbol which is not known in the symbol table,
151 but which would parse as a valid number in the current input radix.
152 E.g. "c" when input_radix==16. Depending on the parse, it will be
153 turned into a name or into a number. */
155 %token
<ssym
> NAME_OR_INT
157 %token STRUCT CLASS SIZEOF COLONCOLON
160 /* Special type cases, put in to allow the parser to distinguish different
163 %token
<sval
> DOLLAR_VARIABLE
168 %token
<lval
> TRUEKEYWORD FALSEKEYWORD
178 %left
'<' '>' LEQ GEQ
179 %left LSH RSH DIV MOD
183 %right UNARY INCREMENT DECREMENT
184 %right ARROW
'.' '[' '('
186 %token
<ssym
> BLOCKNAME
193 start
: { current_type
= NULL
;
195 leftdiv_is_integer
= 0;
207 pstate
->push_new
<type_operation
> ($1);
208 current_type
= $1; } ;
210 /* Expressions, including the comma operator. */
213 { pstate
->wrap2
<comma_operation
> (); }
216 /* Expressions, not including the comma operator. */
217 exp
: exp
'^' %prec UNARY
218 { pstate
->wrap
<unop_ind_operation
> ();
220 current_type
= current_type
->target_type
(); }
223 exp
: '@' exp %prec UNARY
224 { pstate
->wrap
<unop_addr_operation
> ();
226 current_type
= TYPE_POINTER_TYPE
(current_type
); }
229 exp
: '-' exp %prec UNARY
230 { pstate
->wrap
<unary_neg_operation
> (); }
233 exp
: NOT exp %prec UNARY
234 { pstate
->wrap
<unary_logical_not_operation
> (); }
237 exp
: INCREMENT
'(' exp
')' %prec UNARY
238 { pstate
->wrap
<preinc_operation
> (); }
241 exp
: DECREMENT
'(' exp
')' %prec UNARY
242 { pstate
->wrap
<predec_operation
> (); }
246 field_exp
: exp
'.' %prec UNARY
247 { search_field
= 1; }
250 exp
: field_exp FIELDNAME
252 pstate
->push_new
<structop_operation
>
253 (pstate
->pop
(), copy_name
($2));
257 while
(current_type
->code
()
260 current_type
->target_type
();
261 current_type
= lookup_struct_elt_type
(
262 current_type
, $2.ptr
, 0);
270 pstate
->push_new
<structop_operation
>
271 (pstate
->pop
(), copy_name
($2));
275 while
(current_type
->code
()
278 current_type
->target_type
();
279 current_type
= lookup_struct_elt_type
(
280 current_type
, $2.ptr
, 0);
284 exp
: field_exp name COMPLETE
286 structop_base_operation
*op
287 = new structop_ptr_operation
(pstate
->pop
(),
289 pstate
->mark_struct_expression
(op
);
290 pstate
->push
(operation_up
(op
));
293 exp
: field_exp COMPLETE
295 structop_base_operation
*op
296 = new structop_ptr_operation
(pstate
->pop
(), "");
297 pstate
->mark_struct_expression
(op
);
298 pstate
->push
(operation_up
(op
));
303 /* We need to save the current_type value. */
304 { const char *arrayname
;
306 = pascal_is_string_type
(current_type
, NULL
, NULL
,
307 NULL
, NULL
, &arrayname
);
312 ->field
(arrayfieldindex
- 1).type
());
313 pstate
->push_new
<structop_operation
>
314 (pstate
->pop
(), arrayname
);
316 push_current_type
(); }
318 { pop_current_type
();
319 pstate
->wrap2
<subscript_operation
> ();
321 current_type
= current_type
->target_type
(); }
325 /* This is to save the value of arglist_len
326 being accumulated by an outer function call. */
327 { push_current_type
();
328 pstate
->start_arglist
(); }
329 arglist
')' %prec ARROW
331 std
::vector
<operation_up
> args
332 = pstate
->pop_vector
(pstate
->end_arglist
());
333 pstate
->push_new
<funcall_operation
>
334 (pstate
->pop
(), std
::move
(args
));
337 current_type
= current_type
->target_type
();
343 { pstate
->arglist_len
= 1; }
344 | arglist
',' exp %prec ABOVE_COMMA
345 { pstate
->arglist_len
++; }
348 exp
: type
'(' exp
')' %prec UNARY
351 /* Allow automatic dereference of classes. */
352 if
((current_type
->code
() == TYPE_CODE_PTR
)
353 && (current_type
->target_type
()->code
() == TYPE_CODE_STRUCT
)
354 && (($1)->code
() == TYPE_CODE_STRUCT
))
355 pstate
->wrap
<unop_ind_operation
> ();
357 pstate
->push_new
<unop_cast_operation
>
358 (pstate
->pop
(), $1);
366 /* Binary operators in order of decreasing precedence. */
369 { pstate
->wrap2
<mul_operation
> (); }
373 if
(current_type
&& is_integral_type
(current_type
))
374 leftdiv_is_integer
= 1;
378 if
(leftdiv_is_integer
&& current_type
379 && is_integral_type
(current_type
))
381 pstate
->push_new
<unop_cast_operation
>
383 parse_type
(pstate
)->builtin_long_double
);
385 = parse_type
(pstate
)->builtin_long_double
;
386 leftdiv_is_integer
= 0;
389 pstate
->wrap2
<div_operation
> ();
394 { pstate
->wrap2
<intdiv_operation
> (); }
398 { pstate
->wrap2
<rem_operation
> (); }
402 { pstate
->wrap2
<add_operation
> (); }
406 { pstate
->wrap2
<sub_operation
> (); }
410 { pstate
->wrap2
<lsh_operation
> (); }
414 { pstate
->wrap2
<rsh_operation
> (); }
419 pstate
->wrap2
<equal_operation
> ();
420 current_type
= parse_type
(pstate
)->builtin_bool
;
424 exp
: exp NOTEQUAL exp
426 pstate
->wrap2
<notequal_operation
> ();
427 current_type
= parse_type
(pstate
)->builtin_bool
;
433 pstate
->wrap2
<leq_operation
> ();
434 current_type
= parse_type
(pstate
)->builtin_bool
;
440 pstate
->wrap2
<geq_operation
> ();
441 current_type
= parse_type
(pstate
)->builtin_bool
;
447 pstate
->wrap2
<less_operation
> ();
448 current_type
= parse_type
(pstate
)->builtin_bool
;
454 pstate
->wrap2
<gtr_operation
> ();
455 current_type
= parse_type
(pstate
)->builtin_bool
;
460 { pstate
->wrap2
<bitwise_and_operation
> (); }
464 { pstate
->wrap2
<bitwise_xor_operation
> (); }
468 { pstate
->wrap2
<bitwise_ior_operation
> (); }
472 { pstate
->wrap2
<assign_operation
> (); }
477 pstate
->push_new
<bool_operation
> ($1);
478 current_type
= parse_type
(pstate
)->builtin_bool
;
484 pstate
->push_new
<bool_operation
> ($1);
485 current_type
= parse_type
(pstate
)->builtin_bool
;
491 pstate
->push_new
<long_const_operation
>
493 current_type
= $1.type
;
499 parse_number
(pstate
, $1.stoken.ptr
,
500 $1.stoken.length
, 0, &val
);
501 pstate
->push_new
<long_const_operation
>
502 (val.typed_val_int.type
,
503 val.typed_val_int.val
);
504 current_type
= val.typed_val_int.type
;
512 std
::copy
(std
::begin
($1.val
), std
::end
($1.val
),
514 pstate
->push_new
<float_const_operation
> ($1.type
, data
);
521 exp
: DOLLAR_VARIABLE
523 pstate
->push_dollar
($1);
525 /* $ is the normal prefix for pascal
526 hexadecimal values but this conflicts
527 with the GDB use for debugger variables
528 so in expression to enter hexadecimal
529 values we still need to use C syntax with
531 std
::string tmp
($1.ptr
, $1.length
);
532 /* Handle current_type. */
533 struct internalvar
*intvar
534 = lookup_only_internalvar
(tmp.c_str
() + 1);
535 if
(intvar
!= nullptr
)
537 scoped_value_mark mark
;
540 = value_of_internalvar
(pstate
->gdbarch
(),
542 current_type
= val
->type
();
547 exp
: SIZEOF
'(' type
')' %prec UNARY
549 current_type
= parse_type
(pstate
)->builtin_int
;
550 $3 = check_typedef
($3);
551 pstate
->push_new
<long_const_operation
>
552 (parse_type
(pstate
)->builtin_int
,
556 exp
: SIZEOF
'(' exp
')' %prec UNARY
557 { pstate
->wrap
<unop_sizeof_operation
> ();
558 current_type
= parse_type
(pstate
)->builtin_int
; }
561 { /* C strings are converted into array constants with
562 an explicit null byte added at the end. Thus
563 the array upper bound is the string length.
564 There is no such thing in C as a completely empty
566 const char *sp
= $1.ptr
; int count
= $1.length
;
568 std
::vector
<operation_up
> args
(count
+ 1);
569 for
(int i
= 0; i
< count
; ++i
)
570 args
[i
] = (make_operation
<long_const_operation
>
571 (parse_type
(pstate
)->builtin_char
,
573 args
[count
] = (make_operation
<long_const_operation
>
574 (parse_type
(pstate
)->builtin_char
,
576 pstate
->push_new
<array_operation
>
577 (0, $1.length
, std
::move
(args
));
584 struct value
* this_val
;
585 struct type
* this_type
;
586 pstate
->push_new
<op_this_operation
> ();
587 /* We need type of this. */
589 = value_of_this_silent
(pstate
->language
());
591 this_type
= this_val
->type
();
596 if
(this_type
->code
() == TYPE_CODE_PTR
)
598 this_type
= this_type
->target_type
();
599 pstate
->wrap
<unop_ind_operation
> ();
603 current_type
= this_type
;
607 /* end of object pascal. */
611 if
($1.sym.symbol
!= 0)
612 $$
= $1.sym.symbol
->value_block
();
615 std
::string copy
= copy_name
($1.stoken
);
617 lookup_symtab
(current_program_space
, copy.c_str
());
619 $$
= (tem
->compunit
()->blockvector
()
622 error (_
("No file or function \"%s\"."),
628 block
: block COLONCOLON name
630 std
::string copy
= copy_name
($3);
632 = lookup_symbol
(copy.c_str
(), $1,
633 SEARCH_FUNCTION_DOMAIN
,
637 error (_
("No function \"%s\" in specified context."),
639 $$
= tem
->value_block
(); }
642 variable: block COLONCOLON name
643 { struct block_symbol sym
;
645 std
::string copy
= copy_name
($3);
646 sym
= lookup_symbol
(copy.c_str
(), $1,
649 error (_
("No symbol \"%s\" in specified context."),
652 pstate
->push_new
<var_value_operation
> (sym
);
656 qualified_name: typebase COLONCOLON name
658 struct type
*type
= $1;
660 if
(type
->code
() != TYPE_CODE_STRUCT
661 && type
->code
() != TYPE_CODE_UNION
)
662 error (_
("`%s' is not defined as an aggregate type."),
665 pstate
->push_new
<scope_operation
>
666 (type
, copy_name
($3));
670 variable: qualified_name
673 std
::string name
= copy_name
($2);
675 struct block_symbol sym
676 = lookup_symbol
(name.c_str
(), nullptr
,
677 SEARCH_VFT
, nullptr
);
678 pstate
->push_symbol
(name.c_str
(), sym
);
682 variable: name_not_typename
683 { struct block_symbol sym
= $1.sym
;
687 if
(symbol_read_needs_frame
(sym.symbol
))
688 pstate
->block_tracker
->update
(sym
);
690 pstate
->push_new
<var_value_operation
> (sym
);
691 current_type
= sym.symbol
->type
(); }
692 else if
($1.is_a_field_of_this
)
694 struct value
* this_val
;
695 struct type
* this_type
;
696 /* Object pascal: it hangs off of `this'. Must
697 not inadvertently convert from a method call
699 pstate
->block_tracker
->update
(sym
);
701 = make_operation
<op_this_operation
> ();
702 pstate
->push_new
<structop_operation
>
703 (std
::move
(thisop
), copy_name
($1.stoken
));
704 /* We need type of this. */
706 = value_of_this_silent
(pstate
->language
());
708 this_type
= this_val
->type
();
712 current_type
= lookup_struct_elt_type
(
714 copy_name
($1.stoken
).c_str
(), 0);
720 std
::string arg
= copy_name
($1.stoken
);
722 bound_minimal_symbol msymbol
723 = lookup_minimal_symbol
(current_program_space
, arg.c_str
());
724 if
(msymbol.minsym
!= NULL
)
725 pstate
->push_new
<var_msym_value_operation
>
727 else if
(!have_full_symbols
(current_program_space
)
728 && !have_partial_symbols
(current_program_space
))
729 error (_
("No symbol table is loaded. "
730 "Use the \"file\" command."));
732 error (_
("No symbol \"%s\" in current context."),
742 /* We used to try to recognize more pointer to member types here, but
743 that didn't work (shift/reduce conflicts meant that these rules never
744 got executed). The problem is that
745 int (foo::bar::baz::bizzle)
746 is a function type but
747 int (foo::bar::baz::bizzle::*)
748 is a pointer to member type. Stroustrup loses again! */
753 typebase
/* Implements (approximately): (type-qualifier)* type-specifier */
755 { $$
= lookup_pointer_type
($2); }
760 = lookup_struct
(copy_name
($2).c_str
(),
761 pstate
->expression_context_block
);
765 = lookup_struct
(copy_name
($2).c_str
(),
766 pstate
->expression_context_block
);
768 /* "const" and "volatile" are curently ignored. A type qualifier
769 after the type is handled in the ptype rule. I think these could
773 name
: NAME
{ $$
= $1.stoken
; }
774 | BLOCKNAME
{ $$
= $1.stoken
; }
775 | TYPENAME
{ $$
= $1.stoken
; }
776 | NAME_OR_INT
{ $$
= $1.stoken
; }
779 name_not_typename
: NAME
781 /* These would be useful if name_not_typename was useful, but it is just
782 a fake for "variable", so these cause reduce/reduce conflicts because
783 the parser can't tell whether NAME_OR_INT is a name_not_typename (=variable,
784 =exp) or just an exp. If name_not_typename was ever used in an lvalue
785 context where only a name could occur, this might be useful.
792 /* Take care of parsing a number (anything that starts with a digit).
793 Set yylval and return the token type; update lexptr.
794 LEN is the number of characters in it. */
796 /*** Needs some error checking for the float case ***/
799 parse_number
(struct parser_state
*par_state
,
800 const char *p
, int len
, int parsed_float
, YYSTYPE *putithere
)
807 int base
= input_radix
;
810 /* Number of "L" suffixes encountered. */
813 /* We have found a "L" or "U" suffix. */
814 int found_suffix
= 0;
818 /* Handle suffixes: 'f' for float, 'l' for long double.
819 FIXME: This appears to be an extension -- do we want this? */
820 if
(len
>= 1 && tolower
(p
[len
- 1]) == 'f')
822 putithere
->typed_val_float.type
823 = parse_type
(par_state
)->builtin_float
;
826 else if
(len
>= 1 && tolower
(p
[len
- 1]) == 'l')
828 putithere
->typed_val_float.type
829 = parse_type
(par_state
)->builtin_long_double
;
832 /* Default type for floating-point literals is double. */
835 putithere
->typed_val_float.type
836 = parse_type
(par_state
)->builtin_double
;
839 if
(!parse_float
(p
, len
,
840 putithere
->typed_val_float.type
,
841 putithere
->typed_val_float.val
))
846 /* Handle base-switching prefixes 0x, 0t, 0d, 0. */
847 if
(p
[0] == '0' && len
> 1)
880 if
(c
>= 'A' && c
<= 'Z')
882 if
(c
!= 'l' && c
!= 'u')
884 if
(c
>= '0' && c
<= '9')
892 if
(base
> 10 && c
>= 'a' && c
<= 'f')
896 n
+= i
= c
- 'a' + 10;
909 return ERROR
; /* Char not a digit */
912 return ERROR
; /* Invalid digit in this base. */
914 if
(c
!= 'l' && c
!= 'u')
916 /* Test for overflow. */
917 if
(prevn
== 0 && n
== 0)
920 error (_
("Numeric constant too large."));
925 /* An integer constant is an int, a long, or a long long. An L
926 suffix forces it to be long; an LL suffix forces it to be long
927 long. If not forced to a larger size, it gets the first type of
928 the above that it fits in. To figure out whether it fits, we
929 shift it right and see whether anything remains. Note that we
930 can't shift sizeof (LONGEST) * HOST_CHAR_BIT bits or more in one
931 operation, because many compilers will warn about such a shift
932 (which always produces a zero result). Sometimes gdbarch_int_bit
933 or gdbarch_long_bit will be that big, sometimes not. To deal with
934 the case where it is we just always shift the value more than
935 once, with fewer bits each time. */
937 int int_bits
= gdbarch_int_bit
(par_state
->gdbarch
());
938 int long_bits
= gdbarch_long_bit
(par_state
->gdbarch
());
939 int long_long_bits
= gdbarch_long_long_bit
(par_state
->gdbarch
());
940 bool have_signed
= !unsigned_p
;
941 bool have_int
= long_p
== 0;
942 bool have_long
= long_p
<= 1;
943 if
(have_int
&& have_signed
&& fits_in_type
(1, n
, int_bits
, true
))
944 putithere
->typed_val_int.type
= parse_type
(par_state
)->builtin_int
;
945 else if
(have_int
&& fits_in_type
(1, n
, int_bits
, false
))
946 putithere
->typed_val_int.type
947 = parse_type
(par_state
)->builtin_unsigned_int
;
948 else if
(have_long
&& have_signed
&& fits_in_type
(1, n
, long_bits
, true
))
949 putithere
->typed_val_int.type
= parse_type
(par_state
)->builtin_long
;
950 else if
(have_long
&& fits_in_type
(1, n
, long_bits
, false
))
951 putithere
->typed_val_int.type
952 = parse_type
(par_state
)->builtin_unsigned_long
;
953 else if
(have_signed
&& fits_in_type
(1, n
, long_long_bits
, true
))
954 putithere
->typed_val_int.type
955 = parse_type
(par_state
)->builtin_long_long
;
956 else if
(fits_in_type
(1, n
, long_long_bits
, false
))
957 putithere
->typed_val_int.type
958 = parse_type
(par_state
)->builtin_unsigned_long_long
;
960 error (_
("Numeric constant too large."));
961 putithere
->typed_val_int.val
= n
;
970 struct type_push
*next
;
973 static struct type_push
*tp_top
= NULL
;
976 push_current_type
(void)
978 struct type_push
*tpnew
;
979 tpnew
= (struct type_push
*) malloc
(sizeof
(struct type_push
));
980 tpnew
->next
= tp_top
;
981 tpnew
->stored
= current_type
;
987 pop_current_type
(void)
989 struct type_push
*tp
= tp_top
;
992 current_type
= tp
->stored
;
1002 enum exp_opcode opcode
;
1005 static const struct p_token tokentab3
[] =
1007 {"shr", RSH
, OP_NULL
},
1008 {"shl", LSH
, OP_NULL
},
1009 {"and", ANDAND
, OP_NULL
},
1010 {"div", DIV
, OP_NULL
},
1011 {"not", NOT
, OP_NULL
},
1012 {"mod", MOD
, OP_NULL
},
1013 {"inc", INCREMENT
, OP_NULL
},
1014 {"dec", DECREMENT
, OP_NULL
},
1015 {"xor", XOR
, OP_NULL
}
1018 static const struct p_token tokentab2
[] =
1020 {"or", OR
, OP_NULL
},
1021 {"<>", NOTEQUAL
, OP_NULL
},
1022 {"<=", LEQ
, OP_NULL
},
1023 {">=", GEQ
, OP_NULL
},
1024 {":=", ASSIGN
, OP_NULL
},
1025 {"::", COLONCOLON
, OP_NULL
} };
1027 /* Allocate uppercased var: */
1028 /* make an uppercased copy of tokstart. */
1030 uptok
(const char *tokstart
, int namelen
)
1033 char *uptokstart
= (char *)malloc
(namelen
+1);
1034 for
(i
= 0;i
<= namelen
;i
++)
1036 if
((tokstart
[i
]>='a' && tokstart
[i
]<='z'))
1037 uptokstart
[i
] = tokstart
[i
]-('a'-'A');
1039 uptokstart
[i
] = tokstart
[i
];
1041 uptokstart
[namelen
]='\0';
1045 /* Skip over a Pascal string. STR must point to the opening single quote
1046 character. This function returns a pointer to the character after the
1047 closing single quote character.
1049 This function does not support embedded, escaped single quotes, which
1050 is done by placing two consecutive single quotes into a string.
1051 Support for this would be easy to add, but this function is only used
1052 from the Python expression parser, and if we did skip over escaped
1053 quotes then the rest of the expression parser wouldn't handle them
1056 pascal_skip_string
(const char *str
)
1058 gdb_assert
(*str
== '\'');
1062 while
(*str
!= '\0' && *str
!= '\'');
1067 /* Read one token, getting characters through lexptr. */
1074 const char *tokstart
;
1077 int explen
, tempbufindex
;
1078 static char *tempbuf
;
1079 static int tempbufsize
;
1083 pstate
->prev_lexptr
= pstate
->lexptr
;
1085 tokstart
= pstate
->lexptr
;
1086 explen
= strlen
(pstate
->lexptr
);
1088 /* See if it is a special token of length 3. */
1090 for
(const auto
&token
: tokentab3
)
1091 if
(strncasecmp
(tokstart
, token.oper
, 3) == 0
1092 && (!isalpha
(token.oper
[0]) || explen
== 3
1093 ||
(!isalpha
(tokstart
[3])
1094 && !isdigit
(tokstart
[3]) && tokstart
[3] != '_')))
1096 pstate
->lexptr
+= 3;
1097 yylval.opcode
= token.opcode
;
1101 /* See if it is a special token of length 2. */
1103 for
(const auto
&token
: tokentab2
)
1104 if
(strncasecmp
(tokstart
, token.oper
, 2) == 0
1105 && (!isalpha
(token.oper
[0]) || explen
== 2
1106 ||
(!isalpha
(tokstart
[2])
1107 && !isdigit
(tokstart
[2]) && tokstart
[2] != '_')))
1109 pstate
->lexptr
+= 2;
1110 yylval.opcode
= token.opcode
;
1114 switch
(c
= *tokstart
)
1117 if
(search_field
&& pstate
->parse_completion
)
1129 /* We either have a character constant ('0' or '\177' for example)
1130 or we have a quoted symbol reference ('foo(int,int)' in object pascal
1133 c
= *pstate
->lexptr
++;
1135 c
= parse_escape
(pstate
->gdbarch
(), &pstate
->lexptr
);
1137 error (_
("Empty character constant."));
1139 yylval.typed_val_int.val
= c
;
1140 yylval.typed_val_int.type
= parse_type
(pstate
)->builtin_char
;
1142 c
= *pstate
->lexptr
++;
1145 namelen
= pascal_skip_string
(tokstart
) - tokstart
;
1148 pstate
->lexptr
= tokstart
+ namelen
;
1149 if
(pstate
->lexptr
[-1] != '\'')
1150 error (_
("Unmatched single quote."));
1153 uptokstart
= uptok
(tokstart
,namelen
);
1156 error (_
("Invalid character constant."));
1166 if
(paren_depth
== 0)
1173 if
(pstate
->comma_terminates
&& paren_depth
== 0)
1179 /* Might be a floating point number. */
1180 if
(pstate
->lexptr
[1] < '0' || pstate
->lexptr
[1] > '9')
1182 goto symbol
; /* Nope, must be a symbol. */
1198 /* It's a number. */
1199 int got_dot
= 0, got_e
= 0, toktype
;
1200 const char *p
= tokstart
;
1201 int hex
= input_radix
> 10;
1203 if
(c
== '0' && (p
[1] == 'x' || p
[1] == 'X'))
1208 else if
(c
== '0' && (p
[1]=='t' || p
[1]=='T'
1209 || p
[1]=='d' || p
[1]=='D'))
1217 /* This test includes !hex because 'e' is a valid hex digit
1218 and thus does not indicate a floating point number when
1219 the radix is hex. */
1220 if
(!hex
&& !got_e
&& (*p
== 'e' ||
*p
== 'E'))
1221 got_dot
= got_e
= 1;
1222 /* This test does not include !hex, because a '.' always indicates
1223 a decimal floating point number regardless of the radix. */
1224 else if
(!got_dot
&& *p
== '.')
1226 else if
(got_e
&& (p
[-1] == 'e' || p
[-1] == 'E')
1227 && (*p
== '-' ||
*p
== '+'))
1228 /* This is the sign of the exponent, not the end of the
1231 /* We will take any letters or digits. parse_number will
1232 complain if past the radix, or if L or U are not final. */
1233 else if
((*p
< '0' ||
*p
> '9')
1234 && ((*p
< 'a' ||
*p
> 'z')
1235 && (*p
< 'A' ||
*p
> 'Z')))
1238 toktype
= parse_number
(pstate
, tokstart
,
1239 p
- tokstart
, got_dot | got_e
, &yylval);
1240 if
(toktype
== ERROR
)
1241 error (_
("Invalid number \"%.*s\"."), (int) (p
- tokstart
),
1272 /* Build the gdb internal form of the input string in tempbuf,
1273 translating any standard C escape forms seen. Note that the
1274 buffer is null byte terminated *only* for the convenience of
1275 debugging gdb itself and printing the buffer contents when
1276 the buffer contains no embedded nulls. Gdb does not depend
1277 upon the buffer being null byte terminated, it uses the length
1278 string instead. This allows gdb to handle C strings (as well
1279 as strings in other languages) with embedded null bytes. */
1281 tokptr
= ++tokstart
;
1285 /* Grow the static temp buffer if necessary, including allocating
1286 the first one on demand. */
1287 if
(tempbufindex
+ 1 >= tempbufsize
)
1289 tempbuf
= (char *) realloc
(tempbuf
, tempbufsize
+= 64);
1296 /* Do nothing, loop will terminate. */
1300 c
= parse_escape
(pstate
->gdbarch
(), &tokptr
);
1305 tempbuf
[tempbufindex
++] = c
;
1308 tempbuf
[tempbufindex
++] = *tokptr
++;
1311 } while
((*tokptr
!= '"') && (*tokptr
!= '\0'));
1312 if
(*tokptr
++ != '"')
1314 error (_
("Unterminated string in expression."));
1316 tempbuf
[tempbufindex
] = '\0'; /* See note above. */
1317 yylval.sval.ptr
= tempbuf
;
1318 yylval.sval.length
= tempbufindex
;
1319 pstate
->lexptr
= tokptr
;
1323 if
(!(c
== '_' || c
== '$'
1324 ||
(c
>= 'a' && c
<= 'z') ||
(c
>= 'A' && c
<= 'Z')))
1325 /* We must have come across a bad character (e.g. ';'). */
1326 error (_
("Invalid character '%c' in expression."), c
);
1328 /* It's a name. See how long it is. */
1330 for
(c
= tokstart
[namelen
];
1331 (c
== '_' || c
== '$' ||
(c
>= '0' && c
<= '9')
1332 ||
(c
>= 'a' && c
<= 'z') ||
(c
>= 'A' && c
<= 'Z') || c
== '<');)
1334 /* Template parameter lists are part of the name.
1335 FIXME: This mishandles `print $a<4&&$a>3'. */
1339 int nesting_level
= 1;
1340 while
(tokstart
[++i
])
1342 if
(tokstart
[i
] == '<')
1344 else if
(tokstart
[i
] == '>')
1346 if
(--nesting_level
== 0)
1350 if
(tokstart
[i
] == '>')
1356 /* do NOT uppercase internals because of registers !!! */
1357 c
= tokstart
[++namelen
];
1360 uptokstart
= uptok
(tokstart
,namelen
);
1362 /* The token "if" terminates the expression and is NOT
1363 removed from the input stream. */
1364 if
(namelen
== 2 && uptokstart
[0] == 'I' && uptokstart
[1] == 'F')
1370 pstate
->lexptr
+= namelen
;
1374 /* Catch specific keywords. Should be done with a data structure. */
1378 if
(strcmp
(uptokstart
, "OBJECT") == 0)
1383 if
(strcmp
(uptokstart
, "RECORD") == 0)
1388 if
(strcmp
(uptokstart
, "SIZEOF") == 0)
1395 if
(strcmp
(uptokstart
, "CLASS") == 0)
1400 if
(strcmp
(uptokstart
, "FALSE") == 0)
1404 return FALSEKEYWORD
;
1408 if
(strcmp
(uptokstart
, "TRUE") == 0)
1414 if
(strcmp
(uptokstart
, "SELF") == 0)
1416 /* Here we search for 'this' like
1417 inserted in FPC stabs debug info. */
1418 static const char this_name
[] = "this";
1420 if
(lookup_symbol
(this_name
, pstate
->expression_context_block
,
1421 SEARCH_VFT
, NULL
).symbol
)
1432 yylval.sval.ptr
= tokstart
;
1433 yylval.sval.length
= namelen
;
1435 if
(*tokstart
== '$')
1438 return DOLLAR_VARIABLE
;
1441 /* Use token-type BLOCKNAME for symbols that happen to be defined as
1442 functions or symtabs. If this is not so, then ...
1443 Use token-type TYPENAME for symbols that happen to be defined
1444 currently as names of types; NAME for other symbols.
1445 The caller is not constrained to care about the distinction. */
1447 std
::string tmp
= copy_name
(yylval.sval
);
1449 struct field_of_this_result is_a_field_of_this
;
1453 is_a_field_of_this.type
= NULL
;
1454 if
(search_field
&& current_type
)
1455 is_a_field
= (lookup_struct_elt_type
(current_type
,
1456 tmp.c_str
(), 1) != NULL
);
1460 sym
= lookup_symbol
(tmp.c_str
(), pstate
->expression_context_block
,
1461 SEARCH_VFT
, &is_a_field_of_this
).symbol
;
1462 /* second chance uppercased (as Free Pascal does). */
1463 if
(!sym
&& is_a_field_of_this.type
== NULL
&& !is_a_field
)
1465 for
(int i
= 0; i
<= namelen
; i
++)
1467 if
((tmp
[i
] >= 'a' && tmp
[i
] <= 'z'))
1468 tmp
[i
] -= ('a'-'A');
1470 if
(search_field
&& current_type
)
1471 is_a_field
= (lookup_struct_elt_type
(current_type
,
1472 tmp.c_str
(), 1) != NULL
);
1476 sym
= lookup_symbol
(tmp.c_str
(), pstate
->expression_context_block
,
1477 SEARCH_VFT
, &is_a_field_of_this
).symbol
;
1479 /* Third chance Capitalized (as GPC does). */
1480 if
(!sym
&& is_a_field_of_this.type
== NULL
&& !is_a_field
)
1482 for
(int i
= 0; i
<= namelen
; i
++)
1486 if
((tmp
[i
] >= 'a' && tmp
[i
] <= 'z'))
1487 tmp
[i
] -= ('a'-'A');
1490 if
((tmp
[i
] >= 'A' && tmp
[i
] <= 'Z'))
1491 tmp
[i
] -= ('A'-'a');
1493 if
(search_field
&& current_type
)
1494 is_a_field
= (lookup_struct_elt_type
(current_type
,
1495 tmp.c_str
(), 1) != NULL
);
1499 sym
= lookup_symbol
(tmp.c_str
(), pstate
->expression_context_block
,
1500 SEARCH_VFT
, &is_a_field_of_this
).symbol
;
1503 if
(is_a_field ||
(is_a_field_of_this.type
!= NULL
))
1505 tempbuf
= (char *) realloc
(tempbuf
, namelen
+ 1);
1506 strncpy
(tempbuf
, tmp.c_str
(), namelen
);
1507 tempbuf
[namelen
] = 0;
1508 yylval.sval.ptr
= tempbuf
;
1509 yylval.sval.length
= namelen
;
1510 yylval.ssym.sym.symbol
= NULL
;
1511 yylval.ssym.sym.block
= NULL
;
1513 yylval.ssym.is_a_field_of_this
= is_a_field_of_this.type
!= NULL
;
1519 /* Call lookup_symtab, not lookup_partial_symtab, in case there are
1520 no psymtabs (coff, xcoff, or some future change to blow away the
1521 psymtabs once once symbols are read). */
1522 if
((sym
&& sym
->aclass
() == LOC_BLOCK
)
1523 || lookup_symtab
(current_program_space
, tmp.c_str
()))
1525 yylval.ssym.sym.symbol
= sym
;
1526 yylval.ssym.sym.block
= NULL
;
1527 yylval.ssym.is_a_field_of_this
= is_a_field_of_this.type
!= NULL
;
1531 if
(sym
&& sym
->aclass
() == LOC_TYPEDEF
)
1534 /* Despite the following flaw, we need to keep this code enabled.
1535 Because we can get called from check_stub_method, if we don't
1536 handle nested types then it screws many operations in any
1537 program which uses nested types. */
1538 /* In "A::x", if x is a member function of A and there happens
1539 to be a type (nested or not, since the stabs don't make that
1540 distinction) named x, then this code incorrectly thinks we
1541 are dealing with nested types rather than a member function. */
1544 const char *namestart
;
1545 struct symbol
*best_sym
;
1547 /* Look ahead to detect nested types. This probably should be
1548 done in the grammar, but trying seemed to introduce a lot
1549 of shift/reduce and reduce/reduce conflicts. It's possible
1550 that it could be done, though. Or perhaps a non-grammar, but
1551 less ad hoc, approach would work well. */
1553 /* Since we do not currently have any way of distinguishing
1554 a nested type from a non-nested one (the stabs don't tell
1555 us whether a type is nested), we just ignore the
1562 /* Skip whitespace. */
1563 while
(*p
== ' ' ||
*p
== '\t' ||
*p
== '\n')
1565 if
(*p
== ':' && p
[1] == ':')
1567 /* Skip the `::'. */
1569 /* Skip whitespace. */
1570 while
(*p
== ' ' ||
*p
== '\t' ||
*p
== '\n')
1573 while
(*p
== '_' ||
*p
== '$' ||
(*p
>= '0' && *p
<= '9')
1574 ||
(*p
>= 'a' && *p
<= 'z')
1575 ||
(*p
>= 'A' && *p
<= 'Z'))
1579 struct symbol
*cur_sym
;
1580 /* As big as the whole rest of the expression, which is
1581 at least big enough. */
1583 = (char *) alloca
(tmp.size
() + strlen
(namestart
)
1588 memcpy
(tmp1
, tmp.c_str
(), tmp.size
());
1589 tmp1
+= tmp.size
();
1590 memcpy
(tmp1
, "::", 2);
1592 memcpy
(tmp1
, namestart
, p
- namestart
);
1593 tmp1
[p
- namestart
] = '\0';
1595 = lookup_symbol
(ncopy
,
1596 pstate
->expression_context_block
,
1597 SEARCH_VFT
, NULL
).symbol
;
1600 if
(cur_sym
->aclass
() == LOC_TYPEDEF
)
1618 yylval.tsym.type
= best_sym
->type
();
1620 yylval.tsym.type
= sym
->type
();
1626 = language_lookup_primitive_type
(pstate
->language
(),
1627 pstate
->gdbarch
(), tmp.c_str
());
1628 if
(yylval.tsym.type
!= NULL
)
1634 /* Input names that aren't symbols but ARE valid hex numbers,
1635 when the input radix permits them, can be names or numbers
1636 depending on the parse. Note we support radixes > 16 here. */
1638 && ((tokstart
[0] >= 'a' && tokstart
[0] < 'a' + input_radix
- 10)
1639 ||
(tokstart
[0] >= 'A' && tokstart
[0] < 'A' + input_radix
- 10)))
1641 YYSTYPE newlval
; /* Its value is ignored. */
1642 hextype
= parse_number
(pstate
, tokstart
, namelen
, 0, &newlval
);
1645 yylval.ssym.sym.symbol
= sym
;
1646 yylval.ssym.sym.block
= NULL
;
1647 yylval.ssym.is_a_field_of_this
= is_a_field_of_this.type
!= NULL
;
1654 /* Any other kind of symbol. */
1655 yylval.ssym.sym.symbol
= sym
;
1656 yylval.ssym.sym.block
= NULL
;
1661 /* See language.h. */
1664 pascal_language::parser
(struct parser_state
*par_state
) const
1666 /* Setting up the parser state. */
1667 scoped_restore pstate_restore
= make_scoped_restore
(&pstate
);
1668 gdb_assert
(par_state
!= NULL
);
1672 int result
= yyparse ();
1674 pstate
->set_operation
(pstate
->pop
());
1679 yyerror (const char *msg
)
1681 pstate
->parse_error
(msg
);