* Makefile.in: Update license to GPLv3.
[deliverable/binutils-gdb.git] / gdb / p-exp.y
1 /* YACC parser for Pascal expressions, for GDB.
2 Copyright (C) 2000, 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
3
4 This file is part of GDB.
5
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.
10
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.
15
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/>. */
18
19 /* This file is derived from c-exp.y */
20
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.
29
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. */
37
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 */
44 %{
45
46 #include "defs.h"
47 #include "gdb_string.h"
48 #include <ctype.h>
49 #include "expression.h"
50 #include "value.h"
51 #include "parser-defs.h"
52 #include "language.h"
53 #include "p-lang.h"
54 #include "bfd.h" /* Required by objfiles.h. */
55 #include "symfile.h" /* Required by objfiles.h. */
56 #include "objfiles.h" /* For have_full_symbols and have_partial_symbols */
57 #include "block.h"
58
59 #define parse_type builtin_type (parse_gdbarch)
60
61 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, etc),
62 as well as gratuitiously global symbol names, so we can have multiple
63 yacc generated parsers in gdb. Note that these are only the variables
64 produced by yacc. If other parser generators (bison, byacc, etc) produce
65 additional global names that conflict at link time, then those parser
66 generators need to be fixed instead of adding those names to this list. */
67
68 #define yymaxdepth pascal_maxdepth
69 #define yyparse pascal_parse
70 #define yylex pascal_lex
71 #define yyerror pascal_error
72 #define yylval pascal_lval
73 #define yychar pascal_char
74 #define yydebug pascal_debug
75 #define yypact pascal_pact
76 #define yyr1 pascal_r1
77 #define yyr2 pascal_r2
78 #define yydef pascal_def
79 #define yychk pascal_chk
80 #define yypgo pascal_pgo
81 #define yyact pascal_act
82 #define yyexca pascal_exca
83 #define yyerrflag pascal_errflag
84 #define yynerrs pascal_nerrs
85 #define yyps pascal_ps
86 #define yypv pascal_pv
87 #define yys pascal_s
88 #define yy_yys pascal_yys
89 #define yystate pascal_state
90 #define yytmp pascal_tmp
91 #define yyv pascal_v
92 #define yy_yyv pascal_yyv
93 #define yyval pascal_val
94 #define yylloc pascal_lloc
95 #define yyreds pascal_reds /* With YYDEBUG defined */
96 #define yytoks pascal_toks /* With YYDEBUG defined */
97 #define yyname pascal_name /* With YYDEBUG defined */
98 #define yyrule pascal_rule /* With YYDEBUG defined */
99 #define yylhs pascal_yylhs
100 #define yylen pascal_yylen
101 #define yydefred pascal_yydefred
102 #define yydgoto pascal_yydgoto
103 #define yysindex pascal_yysindex
104 #define yyrindex pascal_yyrindex
105 #define yygindex pascal_yygindex
106 #define yytable pascal_yytable
107 #define yycheck pascal_yycheck
108
109 #ifndef YYDEBUG
110 #define YYDEBUG 1 /* Default to yydebug support */
111 #endif
112
113 #define YYFPRINTF parser_fprintf
114
115 int yyparse (void);
116
117 static int yylex (void);
118
119 void
120 yyerror (char *);
121
122 static char * uptok (char *, int);
123 %}
124
125 /* Although the yacc "value" of an expression is not used,
126 since the result is stored in the structure being created,
127 other node types do have values. */
128
129 %union
130 {
131 LONGEST lval;
132 struct {
133 LONGEST val;
134 struct type *type;
135 } typed_val_int;
136 struct {
137 DOUBLEST dval;
138 struct type *type;
139 } typed_val_float;
140 struct symbol *sym;
141 struct type *tval;
142 struct stoken sval;
143 struct ttype tsym;
144 struct symtoken ssym;
145 int voidval;
146 struct block *bval;
147 enum exp_opcode opcode;
148 struct internalvar *ivar;
149
150 struct type **tvec;
151 int *ivec;
152 }
153
154 %{
155 /* YYSTYPE gets defined by %union */
156 static int
157 parse_number (char *, int, int, YYSTYPE *);
158
159 static struct type *current_type;
160 static int leftdiv_is_integer;
161 static void push_current_type (void);
162 static void pop_current_type (void);
163 static int search_field;
164 %}
165
166 %type <voidval> exp exp1 type_exp start normal_start variable qualified_name
167 %type <tval> type typebase
168 /* %type <bval> block */
169
170 /* Fancy type parsing. */
171 %type <tval> ptype
172
173 %token <typed_val_int> INT
174 %token <typed_val_float> FLOAT
175
176 /* Both NAME and TYPENAME tokens represent symbols in the input,
177 and both convey their data as strings.
178 But a TYPENAME is a string that happens to be defined as a typedef
179 or builtin type name (such as int or char)
180 and a NAME is any other symbol.
181 Contexts where this distinction is not important can use the
182 nonterminal "name", which matches either NAME or TYPENAME. */
183
184 %token <sval> STRING
185 %token <sval> FIELDNAME
186 %token <ssym> NAME /* BLOCKNAME defined below to give it higher precedence. */
187 %token <tsym> TYPENAME
188 %type <sval> name
189 %type <ssym> name_not_typename
190
191 /* A NAME_OR_INT is a symbol which is not known in the symbol table,
192 but which would parse as a valid number in the current input radix.
193 E.g. "c" when input_radix==16. Depending on the parse, it will be
194 turned into a name or into a number. */
195
196 %token <ssym> NAME_OR_INT
197
198 %token STRUCT CLASS SIZEOF COLONCOLON
199 %token ERROR
200
201 /* Special type cases, put in to allow the parser to distinguish different
202 legal basetypes. */
203
204 %token <voidval> VARIABLE
205
206
207 /* Object pascal */
208 %token THIS
209 %token <lval> TRUEKEYWORD FALSEKEYWORD
210
211 %left ','
212 %left ABOVE_COMMA
213 %right ASSIGN
214 %left NOT
215 %left OR
216 %left XOR
217 %left ANDAND
218 %left '=' NOTEQUAL
219 %left '<' '>' LEQ GEQ
220 %left LSH RSH DIV MOD
221 %left '@'
222 %left '+' '-'
223 %left '*' '/'
224 %right UNARY INCREMENT DECREMENT
225 %right ARROW '.' '[' '('
226 %left '^'
227 %token <ssym> BLOCKNAME
228 %type <bval> block
229 %left COLONCOLON
230
231 \f
232 %%
233
234 start : { current_type = NULL;
235 search_field = 0;
236 leftdiv_is_integer = 0;
237 }
238 normal_start {}
239 ;
240
241 normal_start :
242 exp1
243 | type_exp
244 ;
245
246 type_exp: type
247 { write_exp_elt_opcode(OP_TYPE);
248 write_exp_elt_type($1);
249 write_exp_elt_opcode(OP_TYPE);
250 current_type = $1; } ;
251
252 /* Expressions, including the comma operator. */
253 exp1 : exp
254 | exp1 ',' exp
255 { write_exp_elt_opcode (BINOP_COMMA); }
256 ;
257
258 /* Expressions, not including the comma operator. */
259 exp : exp '^' %prec UNARY
260 { write_exp_elt_opcode (UNOP_IND);
261 if (current_type)
262 current_type = TYPE_TARGET_TYPE (current_type); }
263 ;
264
265 exp : '@' exp %prec UNARY
266 { write_exp_elt_opcode (UNOP_ADDR);
267 if (current_type)
268 current_type = TYPE_POINTER_TYPE (current_type); }
269 ;
270
271 exp : '-' exp %prec UNARY
272 { write_exp_elt_opcode (UNOP_NEG); }
273 ;
274
275 exp : NOT exp %prec UNARY
276 { write_exp_elt_opcode (UNOP_LOGICAL_NOT); }
277 ;
278
279 exp : INCREMENT '(' exp ')' %prec UNARY
280 { write_exp_elt_opcode (UNOP_PREINCREMENT); }
281 ;
282
283 exp : DECREMENT '(' exp ')' %prec UNARY
284 { write_exp_elt_opcode (UNOP_PREDECREMENT); }
285 ;
286
287 exp : exp '.' { search_field = 1; }
288 FIELDNAME
289 /* name */
290 { write_exp_elt_opcode (STRUCTOP_STRUCT);
291 write_exp_string ($4);
292 write_exp_elt_opcode (STRUCTOP_STRUCT);
293 search_field = 0;
294 if (current_type)
295 { while (TYPE_CODE (current_type) == TYPE_CODE_PTR)
296 current_type = TYPE_TARGET_TYPE (current_type);
297 current_type = lookup_struct_elt_type (
298 current_type, $4.ptr, 0); };
299 } ;
300 exp : exp '['
301 /* We need to save the current_type value */
302 { char *arrayname;
303 int arrayfieldindex;
304 arrayfieldindex = is_pascal_string_type (
305 current_type, NULL, NULL,
306 NULL, NULL, &arrayname);
307 if (arrayfieldindex)
308 {
309 struct stoken stringsval;
310 stringsval.ptr = alloca (strlen (arrayname) + 1);
311 stringsval.length = strlen (arrayname);
312 strcpy (stringsval.ptr, arrayname);
313 current_type = TYPE_FIELD_TYPE (current_type,
314 arrayfieldindex - 1);
315 write_exp_elt_opcode (STRUCTOP_STRUCT);
316 write_exp_string (stringsval);
317 write_exp_elt_opcode (STRUCTOP_STRUCT);
318 }
319 push_current_type (); }
320 exp1 ']'
321 { pop_current_type ();
322 write_exp_elt_opcode (BINOP_SUBSCRIPT);
323 if (current_type)
324 current_type = TYPE_TARGET_TYPE (current_type); }
325 ;
326
327 exp : exp '('
328 /* This is to save the value of arglist_len
329 being accumulated by an outer function call. */
330 { push_current_type ();
331 start_arglist (); }
332 arglist ')' %prec ARROW
333 { write_exp_elt_opcode (OP_FUNCALL);
334 write_exp_elt_longcst ((LONGEST) end_arglist ());
335 write_exp_elt_opcode (OP_FUNCALL);
336 pop_current_type ();
337 if (current_type)
338 current_type = TYPE_TARGET_TYPE (current_type);
339 }
340 ;
341
342 arglist :
343 | exp
344 { arglist_len = 1; }
345 | arglist ',' exp %prec ABOVE_COMMA
346 { arglist_len++; }
347 ;
348
349 exp : type '(' exp ')' %prec UNARY
350 { if (current_type)
351 {
352 /* Allow automatic dereference of classes. */
353 if ((TYPE_CODE (current_type) == TYPE_CODE_PTR)
354 && (TYPE_CODE (TYPE_TARGET_TYPE (current_type)) == TYPE_CODE_CLASS)
355 && (TYPE_CODE ($1) == TYPE_CODE_CLASS))
356 write_exp_elt_opcode (UNOP_IND);
357 }
358 write_exp_elt_opcode (UNOP_CAST);
359 write_exp_elt_type ($1);
360 write_exp_elt_opcode (UNOP_CAST);
361 current_type = $1; }
362 ;
363
364 exp : '(' exp1 ')'
365 { }
366 ;
367
368 /* Binary operators in order of decreasing precedence. */
369
370 exp : exp '*' exp
371 { write_exp_elt_opcode (BINOP_MUL); }
372 ;
373
374 exp : exp '/' {
375 if (current_type && is_integral_type (current_type))
376 leftdiv_is_integer = 1;
377 }
378 exp
379 {
380 if (leftdiv_is_integer && current_type
381 && is_integral_type (current_type))
382 {
383 write_exp_elt_opcode (UNOP_CAST);
384 write_exp_elt_type (parse_type->builtin_long_double);
385 current_type = parse_type->builtin_long_double;
386 write_exp_elt_opcode (UNOP_CAST);
387 leftdiv_is_integer = 0;
388 }
389
390 write_exp_elt_opcode (BINOP_DIV);
391 }
392 ;
393
394 exp : exp DIV exp
395 { write_exp_elt_opcode (BINOP_INTDIV); }
396 ;
397
398 exp : exp MOD exp
399 { write_exp_elt_opcode (BINOP_REM); }
400 ;
401
402 exp : exp '+' exp
403 { write_exp_elt_opcode (BINOP_ADD); }
404 ;
405
406 exp : exp '-' exp
407 { write_exp_elt_opcode (BINOP_SUB); }
408 ;
409
410 exp : exp LSH exp
411 { write_exp_elt_opcode (BINOP_LSH); }
412 ;
413
414 exp : exp RSH exp
415 { write_exp_elt_opcode (BINOP_RSH); }
416 ;
417
418 exp : exp '=' exp
419 { write_exp_elt_opcode (BINOP_EQUAL);
420 current_type = parse_type->builtin_bool;
421 }
422 ;
423
424 exp : exp NOTEQUAL exp
425 { write_exp_elt_opcode (BINOP_NOTEQUAL);
426 current_type = parse_type->builtin_bool;
427 }
428 ;
429
430 exp : exp LEQ exp
431 { write_exp_elt_opcode (BINOP_LEQ);
432 current_type = parse_type->builtin_bool;
433 }
434 ;
435
436 exp : exp GEQ exp
437 { write_exp_elt_opcode (BINOP_GEQ);
438 current_type = parse_type->builtin_bool;
439 }
440 ;
441
442 exp : exp '<' exp
443 { write_exp_elt_opcode (BINOP_LESS);
444 current_type = parse_type->builtin_bool;
445 }
446 ;
447
448 exp : exp '>' exp
449 { write_exp_elt_opcode (BINOP_GTR);
450 current_type = parse_type->builtin_bool;
451 }
452 ;
453
454 exp : exp ANDAND exp
455 { write_exp_elt_opcode (BINOP_BITWISE_AND); }
456 ;
457
458 exp : exp XOR exp
459 { write_exp_elt_opcode (BINOP_BITWISE_XOR); }
460 ;
461
462 exp : exp OR exp
463 { write_exp_elt_opcode (BINOP_BITWISE_IOR); }
464 ;
465
466 exp : exp ASSIGN exp
467 { write_exp_elt_opcode (BINOP_ASSIGN); }
468 ;
469
470 exp : TRUEKEYWORD
471 { write_exp_elt_opcode (OP_BOOL);
472 write_exp_elt_longcst ((LONGEST) $1);
473 current_type = parse_type->builtin_bool;
474 write_exp_elt_opcode (OP_BOOL); }
475 ;
476
477 exp : FALSEKEYWORD
478 { write_exp_elt_opcode (OP_BOOL);
479 write_exp_elt_longcst ((LONGEST) $1);
480 current_type = parse_type->builtin_bool;
481 write_exp_elt_opcode (OP_BOOL); }
482 ;
483
484 exp : INT
485 { write_exp_elt_opcode (OP_LONG);
486 write_exp_elt_type ($1.type);
487 current_type = $1.type;
488 write_exp_elt_longcst ((LONGEST)($1.val));
489 write_exp_elt_opcode (OP_LONG); }
490 ;
491
492 exp : NAME_OR_INT
493 { YYSTYPE val;
494 parse_number ($1.stoken.ptr, $1.stoken.length, 0, &val);
495 write_exp_elt_opcode (OP_LONG);
496 write_exp_elt_type (val.typed_val_int.type);
497 current_type = val.typed_val_int.type;
498 write_exp_elt_longcst ((LONGEST)val.typed_val_int.val);
499 write_exp_elt_opcode (OP_LONG);
500 }
501 ;
502
503
504 exp : FLOAT
505 { write_exp_elt_opcode (OP_DOUBLE);
506 write_exp_elt_type ($1.type);
507 current_type = $1.type;
508 write_exp_elt_dblcst ($1.dval);
509 write_exp_elt_opcode (OP_DOUBLE); }
510 ;
511
512 exp : variable
513 ;
514
515 exp : VARIABLE
516 /* Already written by write_dollar_variable. */
517 ;
518
519 exp : SIZEOF '(' type ')' %prec UNARY
520 { write_exp_elt_opcode (OP_LONG);
521 write_exp_elt_type (parse_type->builtin_int);
522 CHECK_TYPEDEF ($3);
523 write_exp_elt_longcst ((LONGEST) TYPE_LENGTH ($3));
524 write_exp_elt_opcode (OP_LONG); }
525 ;
526
527 exp : STRING
528 { /* C strings are converted into array constants with
529 an explicit null byte added at the end. Thus
530 the array upper bound is the string length.
531 There is no such thing in C as a completely empty
532 string. */
533 char *sp = $1.ptr; int count = $1.length;
534 while (count-- > 0)
535 {
536 write_exp_elt_opcode (OP_LONG);
537 write_exp_elt_type (parse_type->builtin_char);
538 write_exp_elt_longcst ((LONGEST)(*sp++));
539 write_exp_elt_opcode (OP_LONG);
540 }
541 write_exp_elt_opcode (OP_LONG);
542 write_exp_elt_type (parse_type->builtin_char);
543 write_exp_elt_longcst ((LONGEST)'\0');
544 write_exp_elt_opcode (OP_LONG);
545 write_exp_elt_opcode (OP_ARRAY);
546 write_exp_elt_longcst ((LONGEST) 0);
547 write_exp_elt_longcst ((LONGEST) ($1.length));
548 write_exp_elt_opcode (OP_ARRAY); }
549 ;
550
551 /* Object pascal */
552 exp : THIS
553 {
554 struct value * this_val;
555 struct type * this_type;
556 write_exp_elt_opcode (OP_THIS);
557 write_exp_elt_opcode (OP_THIS);
558 /* we need type of this */
559 this_val = value_of_this (0);
560 if (this_val)
561 this_type = value_type (this_val);
562 else
563 this_type = NULL;
564 if (this_type)
565 {
566 if (TYPE_CODE (this_type) == TYPE_CODE_PTR)
567 {
568 this_type = TYPE_TARGET_TYPE (this_type);
569 write_exp_elt_opcode (UNOP_IND);
570 }
571 }
572
573 current_type = this_type;
574 }
575 ;
576
577 /* end of object pascal. */
578
579 block : BLOCKNAME
580 {
581 if ($1.sym != 0)
582 $$ = SYMBOL_BLOCK_VALUE ($1.sym);
583 else
584 {
585 struct symtab *tem =
586 lookup_symtab (copy_name ($1.stoken));
587 if (tem)
588 $$ = BLOCKVECTOR_BLOCK (BLOCKVECTOR (tem), STATIC_BLOCK);
589 else
590 error ("No file or function \"%s\".",
591 copy_name ($1.stoken));
592 }
593 }
594 ;
595
596 block : block COLONCOLON name
597 { struct symbol *tem
598 = lookup_symbol (copy_name ($3), $1,
599 VAR_DOMAIN, (int *) NULL);
600 if (!tem || SYMBOL_CLASS (tem) != LOC_BLOCK)
601 error ("No function \"%s\" in specified context.",
602 copy_name ($3));
603 $$ = SYMBOL_BLOCK_VALUE (tem); }
604 ;
605
606 variable: block COLONCOLON name
607 { struct symbol *sym;
608 sym = lookup_symbol (copy_name ($3), $1,
609 VAR_DOMAIN, (int *) NULL);
610 if (sym == 0)
611 error ("No symbol \"%s\" in specified context.",
612 copy_name ($3));
613
614 write_exp_elt_opcode (OP_VAR_VALUE);
615 /* block_found is set by lookup_symbol. */
616 write_exp_elt_block (block_found);
617 write_exp_elt_sym (sym);
618 write_exp_elt_opcode (OP_VAR_VALUE); }
619 ;
620
621 qualified_name: typebase COLONCOLON name
622 {
623 struct type *type = $1;
624 if (TYPE_CODE (type) != TYPE_CODE_STRUCT
625 && TYPE_CODE (type) != TYPE_CODE_UNION)
626 error ("`%s' is not defined as an aggregate type.",
627 TYPE_NAME (type));
628
629 write_exp_elt_opcode (OP_SCOPE);
630 write_exp_elt_type (type);
631 write_exp_string ($3);
632 write_exp_elt_opcode (OP_SCOPE);
633 }
634 ;
635
636 variable: qualified_name
637 | COLONCOLON name
638 {
639 char *name = copy_name ($2);
640 struct symbol *sym;
641 struct minimal_symbol *msymbol;
642
643 sym =
644 lookup_symbol (name, (const struct block *) NULL,
645 VAR_DOMAIN, (int *) NULL);
646 if (sym)
647 {
648 write_exp_elt_opcode (OP_VAR_VALUE);
649 write_exp_elt_block (NULL);
650 write_exp_elt_sym (sym);
651 write_exp_elt_opcode (OP_VAR_VALUE);
652 break;
653 }
654
655 msymbol = lookup_minimal_symbol (name, NULL, NULL);
656 if (msymbol != NULL)
657 write_exp_msymbol (msymbol);
658 else if (!have_full_symbols () && !have_partial_symbols ())
659 error ("No symbol table is loaded. Use the \"file\" command.");
660 else
661 error ("No symbol \"%s\" in current context.", name);
662 }
663 ;
664
665 variable: name_not_typename
666 { struct symbol *sym = $1.sym;
667
668 if (sym)
669 {
670 if (symbol_read_needs_frame (sym))
671 {
672 if (innermost_block == 0
673 || contained_in (block_found,
674 innermost_block))
675 innermost_block = block_found;
676 }
677
678 write_exp_elt_opcode (OP_VAR_VALUE);
679 /* We want to use the selected frame, not
680 another more inner frame which happens to
681 be in the same block. */
682 write_exp_elt_block (NULL);
683 write_exp_elt_sym (sym);
684 write_exp_elt_opcode (OP_VAR_VALUE);
685 current_type = sym->type; }
686 else if ($1.is_a_field_of_this)
687 {
688 struct value * this_val;
689 struct type * this_type;
690 /* Object pascal: it hangs off of `this'. Must
691 not inadvertently convert from a method call
692 to data ref. */
693 if (innermost_block == 0
694 || contained_in (block_found,
695 innermost_block))
696 innermost_block = block_found;
697 write_exp_elt_opcode (OP_THIS);
698 write_exp_elt_opcode (OP_THIS);
699 write_exp_elt_opcode (STRUCTOP_PTR);
700 write_exp_string ($1.stoken);
701 write_exp_elt_opcode (STRUCTOP_PTR);
702 /* we need type of this */
703 this_val = value_of_this (0);
704 if (this_val)
705 this_type = value_type (this_val);
706 else
707 this_type = NULL;
708 if (this_type)
709 current_type = lookup_struct_elt_type (
710 this_type,
711 copy_name ($1.stoken), 0);
712 else
713 current_type = NULL;
714 }
715 else
716 {
717 struct minimal_symbol *msymbol;
718 char *arg = copy_name ($1.stoken);
719
720 msymbol =
721 lookup_minimal_symbol (arg, NULL, NULL);
722 if (msymbol != NULL)
723 write_exp_msymbol (msymbol);
724 else if (!have_full_symbols () && !have_partial_symbols ())
725 error ("No symbol table is loaded. Use the \"file\" command.");
726 else
727 error ("No symbol \"%s\" in current context.",
728 copy_name ($1.stoken));
729 }
730 }
731 ;
732
733
734 ptype : typebase
735 ;
736
737 /* We used to try to recognize more pointer to member types here, but
738 that didn't work (shift/reduce conflicts meant that these rules never
739 got executed). The problem is that
740 int (foo::bar::baz::bizzle)
741 is a function type but
742 int (foo::bar::baz::bizzle::*)
743 is a pointer to member type. Stroustrup loses again! */
744
745 type : ptype
746 ;
747
748 typebase /* Implements (approximately): (type-qualifier)* type-specifier */
749 : '^' typebase
750 { $$ = lookup_pointer_type ($2); }
751 | TYPENAME
752 { $$ = $1.type; }
753 | STRUCT name
754 { $$ = lookup_struct (copy_name ($2),
755 expression_context_block); }
756 | CLASS name
757 { $$ = lookup_struct (copy_name ($2),
758 expression_context_block); }
759 /* "const" and "volatile" are curently ignored. A type qualifier
760 after the type is handled in the ptype rule. I think these could
761 be too. */
762 ;
763
764 name : NAME { $$ = $1.stoken; }
765 | BLOCKNAME { $$ = $1.stoken; }
766 | TYPENAME { $$ = $1.stoken; }
767 | NAME_OR_INT { $$ = $1.stoken; }
768 ;
769
770 name_not_typename : NAME
771 | BLOCKNAME
772 /* These would be useful if name_not_typename was useful, but it is just
773 a fake for "variable", so these cause reduce/reduce conflicts because
774 the parser can't tell whether NAME_OR_INT is a name_not_typename (=variable,
775 =exp) or just an exp. If name_not_typename was ever used in an lvalue
776 context where only a name could occur, this might be useful.
777 | NAME_OR_INT
778 */
779 ;
780
781 %%
782
783 /* Take care of parsing a number (anything that starts with a digit).
784 Set yylval and return the token type; update lexptr.
785 LEN is the number of characters in it. */
786
787 /*** Needs some error checking for the float case ***/
788
789 static int
790 parse_number (p, len, parsed_float, putithere)
791 char *p;
792 int len;
793 int parsed_float;
794 YYSTYPE *putithere;
795 {
796 /* FIXME: Shouldn't these be unsigned? We don't deal with negative values
797 here, and we do kind of silly things like cast to unsigned. */
798 LONGEST n = 0;
799 LONGEST prevn = 0;
800 ULONGEST un;
801
802 int i = 0;
803 int c;
804 int base = input_radix;
805 int unsigned_p = 0;
806
807 /* Number of "L" suffixes encountered. */
808 int long_p = 0;
809
810 /* We have found a "L" or "U" suffix. */
811 int found_suffix = 0;
812
813 ULONGEST high_bit;
814 struct type *signed_type;
815 struct type *unsigned_type;
816
817 if (parsed_float)
818 {
819 /* It's a float since it contains a point or an exponent. */
820 char c;
821 int num = 0; /* number of tokens scanned by scanf */
822 char saved_char = p[len];
823
824 p[len] = 0; /* null-terminate the token */
825 num = sscanf (p, "%" DOUBLEST_SCAN_FORMAT "%c",
826 &putithere->typed_val_float.dval, &c);
827 p[len] = saved_char; /* restore the input stream */
828 if (num != 1) /* check scanf found ONLY a float ... */
829 return ERROR;
830 /* See if it has `f' or `l' suffix (float or long double). */
831
832 c = tolower (p[len - 1]);
833
834 if (c == 'f')
835 putithere->typed_val_float.type = parse_type->builtin_float;
836 else if (c == 'l')
837 putithere->typed_val_float.type = parse_type->builtin_long_double;
838 else if (isdigit (c) || c == '.')
839 putithere->typed_val_float.type = parse_type->builtin_double;
840 else
841 return ERROR;
842
843 return FLOAT;
844 }
845
846 /* Handle base-switching prefixes 0x, 0t, 0d, 0 */
847 if (p[0] == '0')
848 switch (p[1])
849 {
850 case 'x':
851 case 'X':
852 if (len >= 3)
853 {
854 p += 2;
855 base = 16;
856 len -= 2;
857 }
858 break;
859
860 case 't':
861 case 'T':
862 case 'd':
863 case 'D':
864 if (len >= 3)
865 {
866 p += 2;
867 base = 10;
868 len -= 2;
869 }
870 break;
871
872 default:
873 base = 8;
874 break;
875 }
876
877 while (len-- > 0)
878 {
879 c = *p++;
880 if (c >= 'A' && c <= 'Z')
881 c += 'a' - 'A';
882 if (c != 'l' && c != 'u')
883 n *= base;
884 if (c >= '0' && c <= '9')
885 {
886 if (found_suffix)
887 return ERROR;
888 n += i = c - '0';
889 }
890 else
891 {
892 if (base > 10 && c >= 'a' && c <= 'f')
893 {
894 if (found_suffix)
895 return ERROR;
896 n += i = c - 'a' + 10;
897 }
898 else if (c == 'l')
899 {
900 ++long_p;
901 found_suffix = 1;
902 }
903 else if (c == 'u')
904 {
905 unsigned_p = 1;
906 found_suffix = 1;
907 }
908 else
909 return ERROR; /* Char not a digit */
910 }
911 if (i >= base)
912 return ERROR; /* Invalid digit in this base */
913
914 /* Portably test for overflow (only works for nonzero values, so make
915 a second check for zero). FIXME: Can't we just make n and prevn
916 unsigned and avoid this? */
917 if (c != 'l' && c != 'u' && (prevn >= n) && n != 0)
918 unsigned_p = 1; /* Try something unsigned */
919
920 /* Portably test for unsigned overflow.
921 FIXME: This check is wrong; for example it doesn't find overflow
922 on 0x123456789 when LONGEST is 32 bits. */
923 if (c != 'l' && c != 'u' && n != 0)
924 {
925 if ((unsigned_p && (ULONGEST) prevn >= (ULONGEST) n))
926 error ("Numeric constant too large.");
927 }
928 prevn = n;
929 }
930
931 /* An integer constant is an int, a long, or a long long. An L
932 suffix forces it to be long; an LL suffix forces it to be long
933 long. If not forced to a larger size, it gets the first type of
934 the above that it fits in. To figure out whether it fits, we
935 shift it right and see whether anything remains. Note that we
936 can't shift sizeof (LONGEST) * HOST_CHAR_BIT bits or more in one
937 operation, because many compilers will warn about such a shift
938 (which always produces a zero result). Sometimes gdbarch_int_bit
939 or gdbarch_long_bit will be that big, sometimes not. To deal with
940 the case where it is we just always shift the value more than
941 once, with fewer bits each time. */
942
943 un = (ULONGEST)n >> 2;
944 if (long_p == 0
945 && (un >> (gdbarch_int_bit (parse_gdbarch) - 2)) == 0)
946 {
947 high_bit = ((ULONGEST)1) << (gdbarch_int_bit (parse_gdbarch) - 1);
948
949 /* A large decimal (not hex or octal) constant (between INT_MAX
950 and UINT_MAX) is a long or unsigned long, according to ANSI,
951 never an unsigned int, but this code treats it as unsigned
952 int. This probably should be fixed. GCC gives a warning on
953 such constants. */
954
955 unsigned_type = parse_type->builtin_unsigned_int;
956 signed_type = parse_type->builtin_int;
957 }
958 else if (long_p <= 1
959 && (un >> (gdbarch_long_bit (parse_gdbarch) - 2)) == 0)
960 {
961 high_bit = ((ULONGEST)1) << (gdbarch_long_bit (parse_gdbarch) - 1);
962 unsigned_type = parse_type->builtin_unsigned_long;
963 signed_type = parse_type->builtin_long;
964 }
965 else
966 {
967 int shift;
968 if (sizeof (ULONGEST) * HOST_CHAR_BIT
969 < gdbarch_long_long_bit (parse_gdbarch))
970 /* A long long does not fit in a LONGEST. */
971 shift = (sizeof (ULONGEST) * HOST_CHAR_BIT - 1);
972 else
973 shift = (gdbarch_long_long_bit (parse_gdbarch) - 1);
974 high_bit = (ULONGEST) 1 << shift;
975 unsigned_type = parse_type->builtin_unsigned_long_long;
976 signed_type = parse_type->builtin_long_long;
977 }
978
979 putithere->typed_val_int.val = n;
980
981 /* If the high bit of the worked out type is set then this number
982 has to be unsigned. */
983
984 if (unsigned_p || (n & high_bit))
985 {
986 putithere->typed_val_int.type = unsigned_type;
987 }
988 else
989 {
990 putithere->typed_val_int.type = signed_type;
991 }
992
993 return INT;
994 }
995
996
997 struct type_push
998 {
999 struct type *stored;
1000 struct type_push *next;
1001 };
1002
1003 static struct type_push *tp_top = NULL;
1004
1005 static void
1006 push_current_type (void)
1007 {
1008 struct type_push *tpnew;
1009 tpnew = (struct type_push *) malloc (sizeof (struct type_push));
1010 tpnew->next = tp_top;
1011 tpnew->stored = current_type;
1012 current_type = NULL;
1013 tp_top = tpnew;
1014 }
1015
1016 static void
1017 pop_current_type (void)
1018 {
1019 struct type_push *tp = tp_top;
1020 if (tp)
1021 {
1022 current_type = tp->stored;
1023 tp_top = tp->next;
1024 free (tp);
1025 }
1026 }
1027
1028 struct token
1029 {
1030 char *operator;
1031 int token;
1032 enum exp_opcode opcode;
1033 };
1034
1035 static const struct token tokentab3[] =
1036 {
1037 {"shr", RSH, BINOP_END},
1038 {"shl", LSH, BINOP_END},
1039 {"and", ANDAND, BINOP_END},
1040 {"div", DIV, BINOP_END},
1041 {"not", NOT, BINOP_END},
1042 {"mod", MOD, BINOP_END},
1043 {"inc", INCREMENT, BINOP_END},
1044 {"dec", DECREMENT, BINOP_END},
1045 {"xor", XOR, BINOP_END}
1046 };
1047
1048 static const struct token tokentab2[] =
1049 {
1050 {"or", OR, BINOP_END},
1051 {"<>", NOTEQUAL, BINOP_END},
1052 {"<=", LEQ, BINOP_END},
1053 {">=", GEQ, BINOP_END},
1054 {":=", ASSIGN, BINOP_END},
1055 {"::", COLONCOLON, BINOP_END} };
1056
1057 /* Allocate uppercased var */
1058 /* make an uppercased copy of tokstart */
1059 static char * uptok (tokstart, namelen)
1060 char *tokstart;
1061 int namelen;
1062 {
1063 int i;
1064 char *uptokstart = (char *)malloc(namelen+1);
1065 for (i = 0;i <= namelen;i++)
1066 {
1067 if ((tokstart[i]>='a' && tokstart[i]<='z'))
1068 uptokstart[i] = tokstart[i]-('a'-'A');
1069 else
1070 uptokstart[i] = tokstart[i];
1071 }
1072 uptokstart[namelen]='\0';
1073 return uptokstart;
1074 }
1075 /* Read one token, getting characters through lexptr. */
1076
1077
1078 static int
1079 yylex ()
1080 {
1081 int c;
1082 int namelen;
1083 unsigned int i;
1084 char *tokstart;
1085 char *uptokstart;
1086 char *tokptr;
1087 char *p;
1088 int explen, tempbufindex;
1089 static char *tempbuf;
1090 static int tempbufsize;
1091
1092 retry:
1093
1094 prev_lexptr = lexptr;
1095
1096 tokstart = lexptr;
1097 explen = strlen (lexptr);
1098 /* See if it is a special token of length 3. */
1099 if (explen > 2)
1100 for (i = 0; i < sizeof (tokentab3) / sizeof (tokentab3[0]); i++)
1101 if (strncasecmp (tokstart, tokentab3[i].operator, 3) == 0
1102 && (!isalpha (tokentab3[i].operator[0]) || explen == 3
1103 || (!isalpha (tokstart[3]) && !isdigit (tokstart[3]) && tokstart[3] != '_')))
1104 {
1105 lexptr += 3;
1106 yylval.opcode = tokentab3[i].opcode;
1107 return tokentab3[i].token;
1108 }
1109
1110 /* See if it is a special token of length 2. */
1111 if (explen > 1)
1112 for (i = 0; i < sizeof (tokentab2) / sizeof (tokentab2[0]); i++)
1113 if (strncasecmp (tokstart, tokentab2[i].operator, 2) == 0
1114 && (!isalpha (tokentab2[i].operator[0]) || explen == 2
1115 || (!isalpha (tokstart[2]) && !isdigit (tokstart[2]) && tokstart[2] != '_')))
1116 {
1117 lexptr += 2;
1118 yylval.opcode = tokentab2[i].opcode;
1119 return tokentab2[i].token;
1120 }
1121
1122 switch (c = *tokstart)
1123 {
1124 case 0:
1125 return 0;
1126
1127 case ' ':
1128 case '\t':
1129 case '\n':
1130 lexptr++;
1131 goto retry;
1132
1133 case '\'':
1134 /* We either have a character constant ('0' or '\177' for example)
1135 or we have a quoted symbol reference ('foo(int,int)' in object pascal
1136 for example). */
1137 lexptr++;
1138 c = *lexptr++;
1139 if (c == '\\')
1140 c = parse_escape (&lexptr);
1141 else if (c == '\'')
1142 error ("Empty character constant.");
1143
1144 yylval.typed_val_int.val = c;
1145 yylval.typed_val_int.type = parse_type->builtin_char;
1146
1147 c = *lexptr++;
1148 if (c != '\'')
1149 {
1150 namelen = skip_quoted (tokstart) - tokstart;
1151 if (namelen > 2)
1152 {
1153 lexptr = tokstart + namelen;
1154 if (lexptr[-1] != '\'')
1155 error ("Unmatched single quote.");
1156 namelen -= 2;
1157 tokstart++;
1158 uptokstart = uptok(tokstart,namelen);
1159 goto tryname;
1160 }
1161 error ("Invalid character constant.");
1162 }
1163 return INT;
1164
1165 case '(':
1166 paren_depth++;
1167 lexptr++;
1168 return c;
1169
1170 case ')':
1171 if (paren_depth == 0)
1172 return 0;
1173 paren_depth--;
1174 lexptr++;
1175 return c;
1176
1177 case ',':
1178 if (comma_terminates && paren_depth == 0)
1179 return 0;
1180 lexptr++;
1181 return c;
1182
1183 case '.':
1184 /* Might be a floating point number. */
1185 if (lexptr[1] < '0' || lexptr[1] > '9')
1186 goto symbol; /* Nope, must be a symbol. */
1187 /* FALL THRU into number case. */
1188
1189 case '0':
1190 case '1':
1191 case '2':
1192 case '3':
1193 case '4':
1194 case '5':
1195 case '6':
1196 case '7':
1197 case '8':
1198 case '9':
1199 {
1200 /* It's a number. */
1201 int got_dot = 0, got_e = 0, toktype;
1202 char *p = tokstart;
1203 int hex = input_radix > 10;
1204
1205 if (c == '0' && (p[1] == 'x' || p[1] == 'X'))
1206 {
1207 p += 2;
1208 hex = 1;
1209 }
1210 else if (c == '0' && (p[1]=='t' || p[1]=='T' || p[1]=='d' || p[1]=='D'))
1211 {
1212 p += 2;
1213 hex = 0;
1214 }
1215
1216 for (;; ++p)
1217 {
1218 /* This test includes !hex because 'e' is a valid hex digit
1219 and thus does not indicate a floating point number when
1220 the radix is hex. */
1221 if (!hex && !got_e && (*p == 'e' || *p == 'E'))
1222 got_dot = got_e = 1;
1223 /* This test does not include !hex, because a '.' always indicates
1224 a decimal floating point number regardless of the radix. */
1225 else if (!got_dot && *p == '.')
1226 got_dot = 1;
1227 else if (got_e && (p[-1] == 'e' || p[-1] == 'E')
1228 && (*p == '-' || *p == '+'))
1229 /* This is the sign of the exponent, not the end of the
1230 number. */
1231 continue;
1232 /* We will take any letters or digits. parse_number will
1233 complain if past the radix, or if L or U are not final. */
1234 else if ((*p < '0' || *p > '9')
1235 && ((*p < 'a' || *p > 'z')
1236 && (*p < 'A' || *p > 'Z')))
1237 break;
1238 }
1239 toktype = parse_number (tokstart, p - tokstart, got_dot|got_e, &yylval);
1240 if (toktype == ERROR)
1241 {
1242 char *err_copy = (char *) alloca (p - tokstart + 1);
1243
1244 memcpy (err_copy, tokstart, p - tokstart);
1245 err_copy[p - tokstart] = 0;
1246 error ("Invalid number \"%s\".", err_copy);
1247 }
1248 lexptr = p;
1249 return toktype;
1250 }
1251
1252 case '+':
1253 case '-':
1254 case '*':
1255 case '/':
1256 case '|':
1257 case '&':
1258 case '^':
1259 case '~':
1260 case '!':
1261 case '@':
1262 case '<':
1263 case '>':
1264 case '[':
1265 case ']':
1266 case '?':
1267 case ':':
1268 case '=':
1269 case '{':
1270 case '}':
1271 symbol:
1272 lexptr++;
1273 return c;
1274
1275 case '"':
1276
1277 /* Build the gdb internal form of the input string in tempbuf,
1278 translating any standard C escape forms seen. Note that the
1279 buffer is null byte terminated *only* for the convenience of
1280 debugging gdb itself and printing the buffer contents when
1281 the buffer contains no embedded nulls. Gdb does not depend
1282 upon the buffer being null byte terminated, it uses the length
1283 string instead. This allows gdb to handle C strings (as well
1284 as strings in other languages) with embedded null bytes */
1285
1286 tokptr = ++tokstart;
1287 tempbufindex = 0;
1288
1289 do {
1290 /* Grow the static temp buffer if necessary, including allocating
1291 the first one on demand. */
1292 if (tempbufindex + 1 >= tempbufsize)
1293 {
1294 tempbuf = (char *) realloc (tempbuf, tempbufsize += 64);
1295 }
1296
1297 switch (*tokptr)
1298 {
1299 case '\0':
1300 case '"':
1301 /* Do nothing, loop will terminate. */
1302 break;
1303 case '\\':
1304 tokptr++;
1305 c = parse_escape (&tokptr);
1306 if (c == -1)
1307 {
1308 continue;
1309 }
1310 tempbuf[tempbufindex++] = c;
1311 break;
1312 default:
1313 tempbuf[tempbufindex++] = *tokptr++;
1314 break;
1315 }
1316 } while ((*tokptr != '"') && (*tokptr != '\0'));
1317 if (*tokptr++ != '"')
1318 {
1319 error ("Unterminated string in expression.");
1320 }
1321 tempbuf[tempbufindex] = '\0'; /* See note above */
1322 yylval.sval.ptr = tempbuf;
1323 yylval.sval.length = tempbufindex;
1324 lexptr = tokptr;
1325 return (STRING);
1326 }
1327
1328 if (!(c == '_' || c == '$'
1329 || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')))
1330 /* We must have come across a bad character (e.g. ';'). */
1331 error ("Invalid character '%c' in expression.", c);
1332
1333 /* It's a name. See how long it is. */
1334 namelen = 0;
1335 for (c = tokstart[namelen];
1336 (c == '_' || c == '$' || (c >= '0' && c <= '9')
1337 || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '<');)
1338 {
1339 /* Template parameter lists are part of the name.
1340 FIXME: This mishandles `print $a<4&&$a>3'. */
1341 if (c == '<')
1342 {
1343 int i = namelen;
1344 int nesting_level = 1;
1345 while (tokstart[++i])
1346 {
1347 if (tokstart[i] == '<')
1348 nesting_level++;
1349 else if (tokstart[i] == '>')
1350 {
1351 if (--nesting_level == 0)
1352 break;
1353 }
1354 }
1355 if (tokstart[i] == '>')
1356 namelen = i;
1357 else
1358 break;
1359 }
1360
1361 /* do NOT uppercase internals because of registers !!! */
1362 c = tokstart[++namelen];
1363 }
1364
1365 uptokstart = uptok(tokstart,namelen);
1366
1367 /* The token "if" terminates the expression and is NOT
1368 removed from the input stream. */
1369 if (namelen == 2 && uptokstart[0] == 'I' && uptokstart[1] == 'F')
1370 {
1371 free (uptokstart);
1372 return 0;
1373 }
1374
1375 lexptr += namelen;
1376
1377 tryname:
1378
1379 /* Catch specific keywords. Should be done with a data structure. */
1380 switch (namelen)
1381 {
1382 case 6:
1383 if (strcmp (uptokstart, "OBJECT") == 0)
1384 {
1385 free (uptokstart);
1386 return CLASS;
1387 }
1388 if (strcmp (uptokstart, "RECORD") == 0)
1389 {
1390 free (uptokstart);
1391 return STRUCT;
1392 }
1393 if (strcmp (uptokstart, "SIZEOF") == 0)
1394 {
1395 free (uptokstart);
1396 return SIZEOF;
1397 }
1398 break;
1399 case 5:
1400 if (strcmp (uptokstart, "CLASS") == 0)
1401 {
1402 free (uptokstart);
1403 return CLASS;
1404 }
1405 if (strcmp (uptokstart, "FALSE") == 0)
1406 {
1407 yylval.lval = 0;
1408 free (uptokstart);
1409 return FALSEKEYWORD;
1410 }
1411 break;
1412 case 4:
1413 if (strcmp (uptokstart, "TRUE") == 0)
1414 {
1415 yylval.lval = 1;
1416 free (uptokstart);
1417 return TRUEKEYWORD;
1418 }
1419 if (strcmp (uptokstart, "SELF") == 0)
1420 {
1421 /* here we search for 'this' like
1422 inserted in FPC stabs debug info */
1423 static const char this_name[] = "this";
1424
1425 if (lookup_symbol (this_name, expression_context_block,
1426 VAR_DOMAIN, (int *) NULL))
1427 {
1428 free (uptokstart);
1429 return THIS;
1430 }
1431 }
1432 break;
1433 default:
1434 break;
1435 }
1436
1437 yylval.sval.ptr = tokstart;
1438 yylval.sval.length = namelen;
1439
1440 if (*tokstart == '$')
1441 {
1442 /* $ is the normal prefix for pascal hexadecimal values
1443 but this conflicts with the GDB use for debugger variables
1444 so in expression to enter hexadecimal values
1445 we still need to use C syntax with 0xff */
1446 write_dollar_variable (yylval.sval);
1447 free (uptokstart);
1448 return VARIABLE;
1449 }
1450
1451 /* Use token-type BLOCKNAME for symbols that happen to be defined as
1452 functions or symtabs. If this is not so, then ...
1453 Use token-type TYPENAME for symbols that happen to be defined
1454 currently as names of types; NAME for other symbols.
1455 The caller is not constrained to care about the distinction. */
1456 {
1457 char *tmp = copy_name (yylval.sval);
1458 struct symbol *sym;
1459 int is_a_field_of_this = 0;
1460 int is_a_field = 0;
1461 int hextype;
1462
1463
1464 if (search_field && current_type)
1465 is_a_field = (lookup_struct_elt_type (current_type, tmp, 1) != NULL);
1466 if (is_a_field)
1467 sym = NULL;
1468 else
1469 sym = lookup_symbol (tmp, expression_context_block,
1470 VAR_DOMAIN, &is_a_field_of_this);
1471 /* second chance uppercased (as Free Pascal does). */
1472 if (!sym && !is_a_field_of_this && !is_a_field)
1473 {
1474 for (i = 0; i <= namelen; i++)
1475 {
1476 if ((tmp[i] >= 'a' && tmp[i] <= 'z'))
1477 tmp[i] -= ('a'-'A');
1478 }
1479 if (search_field && current_type)
1480 is_a_field = (lookup_struct_elt_type (current_type, tmp, 1) != NULL);
1481 if (is_a_field)
1482 sym = NULL;
1483 else
1484 sym = lookup_symbol (tmp, expression_context_block,
1485 VAR_DOMAIN, &is_a_field_of_this);
1486 if (sym || is_a_field_of_this || is_a_field)
1487 for (i = 0; i <= namelen; i++)
1488 {
1489 if ((tokstart[i] >= 'a' && tokstart[i] <= 'z'))
1490 tokstart[i] -= ('a'-'A');
1491 }
1492 }
1493 /* Third chance Capitalized (as GPC does). */
1494 if (!sym && !is_a_field_of_this && !is_a_field)
1495 {
1496 for (i = 0; i <= namelen; i++)
1497 {
1498 if (i == 0)
1499 {
1500 if ((tmp[i] >= 'a' && tmp[i] <= 'z'))
1501 tmp[i] -= ('a'-'A');
1502 }
1503 else
1504 if ((tmp[i] >= 'A' && tmp[i] <= 'Z'))
1505 tmp[i] -= ('A'-'a');
1506 }
1507 if (search_field && current_type)
1508 is_a_field = (lookup_struct_elt_type (current_type, tmp, 1) != NULL);
1509 if (is_a_field)
1510 sym = NULL;
1511 else
1512 sym = lookup_symbol (tmp, expression_context_block,
1513 VAR_DOMAIN, &is_a_field_of_this);
1514 if (sym || is_a_field_of_this || is_a_field)
1515 for (i = 0; i <= namelen; i++)
1516 {
1517 if (i == 0)
1518 {
1519 if ((tokstart[i] >= 'a' && tokstart[i] <= 'z'))
1520 tokstart[i] -= ('a'-'A');
1521 }
1522 else
1523 if ((tokstart[i] >= 'A' && tokstart[i] <= 'Z'))
1524 tokstart[i] -= ('A'-'a');
1525 }
1526 }
1527
1528 if (is_a_field)
1529 {
1530 tempbuf = (char *) realloc (tempbuf, namelen + 1);
1531 strncpy (tempbuf, tokstart, namelen); tempbuf [namelen] = 0;
1532 yylval.sval.ptr = tempbuf;
1533 yylval.sval.length = namelen;
1534 free (uptokstart);
1535 return FIELDNAME;
1536 }
1537 /* Call lookup_symtab, not lookup_partial_symtab, in case there are
1538 no psymtabs (coff, xcoff, or some future change to blow away the
1539 psymtabs once once symbols are read). */
1540 if ((sym && SYMBOL_CLASS (sym) == LOC_BLOCK)
1541 || lookup_symtab (tmp))
1542 {
1543 yylval.ssym.sym = sym;
1544 yylval.ssym.is_a_field_of_this = is_a_field_of_this;
1545 free (uptokstart);
1546 return BLOCKNAME;
1547 }
1548 if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
1549 {
1550 #if 1
1551 /* Despite the following flaw, we need to keep this code enabled.
1552 Because we can get called from check_stub_method, if we don't
1553 handle nested types then it screws many operations in any
1554 program which uses nested types. */
1555 /* In "A::x", if x is a member function of A and there happens
1556 to be a type (nested or not, since the stabs don't make that
1557 distinction) named x, then this code incorrectly thinks we
1558 are dealing with nested types rather than a member function. */
1559
1560 char *p;
1561 char *namestart;
1562 struct symbol *best_sym;
1563
1564 /* Look ahead to detect nested types. This probably should be
1565 done in the grammar, but trying seemed to introduce a lot
1566 of shift/reduce and reduce/reduce conflicts. It's possible
1567 that it could be done, though. Or perhaps a non-grammar, but
1568 less ad hoc, approach would work well. */
1569
1570 /* Since we do not currently have any way of distinguishing
1571 a nested type from a non-nested one (the stabs don't tell
1572 us whether a type is nested), we just ignore the
1573 containing type. */
1574
1575 p = lexptr;
1576 best_sym = sym;
1577 while (1)
1578 {
1579 /* Skip whitespace. */
1580 while (*p == ' ' || *p == '\t' || *p == '\n')
1581 ++p;
1582 if (*p == ':' && p[1] == ':')
1583 {
1584 /* Skip the `::'. */
1585 p += 2;
1586 /* Skip whitespace. */
1587 while (*p == ' ' || *p == '\t' || *p == '\n')
1588 ++p;
1589 namestart = p;
1590 while (*p == '_' || *p == '$' || (*p >= '0' && *p <= '9')
1591 || (*p >= 'a' && *p <= 'z')
1592 || (*p >= 'A' && *p <= 'Z'))
1593 ++p;
1594 if (p != namestart)
1595 {
1596 struct symbol *cur_sym;
1597 /* As big as the whole rest of the expression, which is
1598 at least big enough. */
1599 char *ncopy = alloca (strlen (tmp)+strlen (namestart)+3);
1600 char *tmp1;
1601
1602 tmp1 = ncopy;
1603 memcpy (tmp1, tmp, strlen (tmp));
1604 tmp1 += strlen (tmp);
1605 memcpy (tmp1, "::", 2);
1606 tmp1 += 2;
1607 memcpy (tmp1, namestart, p - namestart);
1608 tmp1[p - namestart] = '\0';
1609 cur_sym = lookup_symbol (ncopy, expression_context_block,
1610 VAR_DOMAIN, (int *) NULL);
1611 if (cur_sym)
1612 {
1613 if (SYMBOL_CLASS (cur_sym) == LOC_TYPEDEF)
1614 {
1615 best_sym = cur_sym;
1616 lexptr = p;
1617 }
1618 else
1619 break;
1620 }
1621 else
1622 break;
1623 }
1624 else
1625 break;
1626 }
1627 else
1628 break;
1629 }
1630
1631 yylval.tsym.type = SYMBOL_TYPE (best_sym);
1632 #else /* not 0 */
1633 yylval.tsym.type = SYMBOL_TYPE (sym);
1634 #endif /* not 0 */
1635 free (uptokstart);
1636 return TYPENAME;
1637 }
1638 yylval.tsym.type
1639 = language_lookup_primitive_type_by_name (parse_language,
1640 parse_gdbarch, tmp);
1641 if (yylval.tsym.type != NULL)
1642 {
1643 free (uptokstart);
1644 return TYPENAME;
1645 }
1646
1647 /* Input names that aren't symbols but ARE valid hex numbers,
1648 when the input radix permits them, can be names or numbers
1649 depending on the parse. Note we support radixes > 16 here. */
1650 if (!sym
1651 && ((tokstart[0] >= 'a' && tokstart[0] < 'a' + input_radix - 10)
1652 || (tokstart[0] >= 'A' && tokstart[0] < 'A' + input_radix - 10)))
1653 {
1654 YYSTYPE newlval; /* Its value is ignored. */
1655 hextype = parse_number (tokstart, namelen, 0, &newlval);
1656 if (hextype == INT)
1657 {
1658 yylval.ssym.sym = sym;
1659 yylval.ssym.is_a_field_of_this = is_a_field_of_this;
1660 free (uptokstart);
1661 return NAME_OR_INT;
1662 }
1663 }
1664
1665 free(uptokstart);
1666 /* Any other kind of symbol */
1667 yylval.ssym.sym = sym;
1668 yylval.ssym.is_a_field_of_this = is_a_field_of_this;
1669 return NAME;
1670 }
1671 }
1672
1673 void
1674 yyerror (msg)
1675 char *msg;
1676 {
1677 if (prev_lexptr)
1678 lexptr = prev_lexptr;
1679
1680 error ("A %s in expression, near `%s'.", (msg ? msg : "error"), lexptr);
1681 }
This page took 0.08915 seconds and 4 git commands to generate.