Update copyright year in gdbarch.sh doc/gdb.texinfo and doc/refcard.tex
[binutils-gdb.git] / gdb / cp-name-parser.y
bloba209b350674bcba29a5514432bee9f9e45ddd916
1 /* YACC parser for C++ names, for GDB.
3 Copyright (C) 2003-2020 Free Software Foundation, Inc.
5 Parts of the lexer are based on c-exp.y from GDB.
7 This file is part of GDB.
9 This program is free software; you can redistribute it and/or modify
10 it under the terms of the GNU General Public License as published by
11 the Free Software Foundation; either version 3 of the License, or
12 (at your option) any later version.
14 This program is distributed in the hope that it will be useful,
15 but WITHOUT ANY WARRANTY; without even the implied warranty of
16 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 GNU General Public License for more details.
19 You should have received a copy of the GNU General Public License
20 along with this program. If not, see <http://www.gnu.org/licenses/>. */
22 /* Note that malloc's and realloc's in this file are transformed to
23 xmalloc and xrealloc respectively by the same sed command in the
24 makefile that remaps any other malloc/realloc inserted by the parser
25 generator. Doing this with #defines and trying to control the interaction
26 with include files (<malloc.h> and <stdlib.h> for example) just became
27 too messy, particularly when such includes can be inserted at random
28 times by the parser generator. */
30 /* The Bison manual says that %pure-parser is deprecated, but we use
31 it anyway because it also works with Byacc. That is also why
32 this uses %lex-param and %parse-param rather than the simpler
33 %param -- Byacc does not support the latter. */
34 %pure-parser
35 %lex-param {struct cpname_state *state}
36 %parse-param {struct cpname_state *state}
40 #include "defs.h"
42 #include <unistd.h>
43 #include "safe-ctype.h"
44 #include "demangle.h"
45 #include "cp-support.h"
46 #include "c-support.h"
47 #include "parser-defs.h"
49 #define GDB_YY_REMAP_PREFIX cpname
50 #include "yy-remap.h"
52 /* The components built by the parser are allocated ahead of time,
53 and cached in this structure. */
55 #define ALLOC_CHUNK 100
57 struct demangle_info {
58 int used;
59 struct demangle_info *next;
60 struct demangle_component comps[ALLOC_CHUNK];
65 %union
67 struct demangle_component *comp;
68 struct nested {
69 struct demangle_component *comp;
70 struct demangle_component **last;
71 } nested;
72 struct {
73 struct demangle_component *comp, *last;
74 } nested1;
75 struct {
76 struct demangle_component *comp, **last;
77 struct nested fn;
78 struct demangle_component *start;
79 int fold_flag;
80 } abstract;
81 int lval;
82 const char *opname;
87 struct cpname_state
89 /* LEXPTR is the current pointer into our lex buffer. PREV_LEXPTR
90 is the start of the last token lexed, only used for diagnostics.
91 ERROR_LEXPTR is the first place an error occurred. GLOBAL_ERRMSG
92 is the first error message encountered. */
94 const char *lexptr, *prev_lexptr, *error_lexptr, *global_errmsg;
96 struct demangle_info *demangle_info;
98 /* The parse tree created by the parser is stored here after a
99 successful parse. */
101 struct demangle_component *global_result;
103 struct demangle_component *d_grab ();
105 /* Helper functions. These wrap the demangler tree interface,
106 handle allocation from our global store, and return the allocated
107 component. */
109 struct demangle_component *fill_comp (enum demangle_component_type d_type,
110 struct demangle_component *lhs,
111 struct demangle_component *rhs);
113 struct demangle_component *make_operator (const char *name, int args);
115 struct demangle_component *make_dtor (enum gnu_v3_dtor_kinds kind,
116 struct demangle_component *name);
118 struct demangle_component *make_builtin_type (const char *name);
120 struct demangle_component *make_name (const char *name, int len);
122 struct demangle_component *d_qualify (struct demangle_component *lhs,
123 int qualifiers, int is_method);
125 struct demangle_component *d_int_type (int flags);
127 struct demangle_component *d_unary (const char *name,
128 struct demangle_component *lhs);
130 struct demangle_component *d_binary (const char *name,
131 struct demangle_component *lhs,
132 struct demangle_component *rhs);
134 int parse_number (const char *p, int len, int parsed_float, YYSTYPE *lvalp);
137 struct demangle_component *
138 cpname_state::d_grab ()
140 struct demangle_info *more;
142 if (demangle_info->used >= ALLOC_CHUNK)
144 if (demangle_info->next == NULL)
146 more = XNEW (struct demangle_info);
147 more->next = NULL;
148 demangle_info->next = more;
150 else
151 more = demangle_info->next;
153 more->used = 0;
154 demangle_info = more;
156 return &demangle_info->comps[demangle_info->used++];
159 /* Flags passed to d_qualify. */
161 #define QUAL_CONST 1
162 #define QUAL_RESTRICT 2
163 #define QUAL_VOLATILE 4
165 /* Flags passed to d_int_type. */
167 #define INT_CHAR (1 << 0)
168 #define INT_SHORT (1 << 1)
169 #define INT_LONG (1 << 2)
170 #define INT_LLONG (1 << 3)
172 #define INT_SIGNED (1 << 4)
173 #define INT_UNSIGNED (1 << 5)
175 /* Enable yydebug for the stand-alone parser. */
176 #ifdef TEST_CPNAMES
177 # define YYDEBUG 1
178 #endif
180 /* Helper functions. These wrap the demangler tree interface, handle
181 allocation from our global store, and return the allocated component. */
183 struct demangle_component *
184 cpname_state::fill_comp (enum demangle_component_type d_type,
185 struct demangle_component *lhs,
186 struct demangle_component *rhs)
188 struct demangle_component *ret = d_grab ();
189 int i;
191 i = cplus_demangle_fill_component (ret, d_type, lhs, rhs);
192 gdb_assert (i);
194 return ret;
197 struct demangle_component *
198 cpname_state::make_operator (const char *name, int args)
200 struct demangle_component *ret = d_grab ();
201 int i;
203 i = cplus_demangle_fill_operator (ret, name, args);
204 gdb_assert (i);
206 return ret;
209 struct demangle_component *
210 cpname_state::make_dtor (enum gnu_v3_dtor_kinds kind,
211 struct demangle_component *name)
213 struct demangle_component *ret = d_grab ();
214 int i;
216 i = cplus_demangle_fill_dtor (ret, kind, name);
217 gdb_assert (i);
219 return ret;
222 struct demangle_component *
223 cpname_state::make_builtin_type (const char *name)
225 struct demangle_component *ret = d_grab ();
226 int i;
228 i = cplus_demangle_fill_builtin_type (ret, name);
229 gdb_assert (i);
231 return ret;
234 struct demangle_component *
235 cpname_state::make_name (const char *name, int len)
237 struct demangle_component *ret = d_grab ();
238 int i;
240 i = cplus_demangle_fill_name (ret, name, len);
241 gdb_assert (i);
243 return ret;
246 #define d_left(dc) (dc)->u.s_binary.left
247 #define d_right(dc) (dc)->u.s_binary.right
249 static int yylex (YYSTYPE *, cpname_state *);
250 static void yyerror (cpname_state *, const char *);
253 %type <comp> exp exp1 type start start_opt oper colon_name
254 %type <comp> unqualified_name colon_ext_name
255 %type <comp> templ template_arg
256 %type <comp> builtin_type
257 %type <comp> typespec_2 array_indicator
258 %type <comp> colon_ext_only ext_only_name
260 %type <comp> demangler_special function conversion_op
261 %type <nested> conversion_op_name
263 %type <abstract> abstract_declarator direct_abstract_declarator
264 %type <abstract> abstract_declarator_fn
265 %type <nested> declarator direct_declarator function_arglist
267 %type <nested> declarator_1 direct_declarator_1
269 %type <nested> template_params function_args
270 %type <nested> ptr_operator
272 %type <nested1> nested_name
274 %type <lval> qualifier qualifiers qualifiers_opt
276 %type <lval> int_part int_seq
278 %token <comp> INT
279 %token <comp> FLOAT
281 %token <comp> NAME
282 %type <comp> name
284 %token STRUCT CLASS UNION ENUM SIZEOF UNSIGNED COLONCOLON
285 %token TEMPLATE
286 %token ERROR
287 %token NEW DELETE OPERATOR
288 %token STATIC_CAST REINTERPRET_CAST DYNAMIC_CAST
290 /* Special type cases, put in to allow the parser to distinguish different
291 legal basetypes. */
292 %token SIGNED_KEYWORD LONG SHORT INT_KEYWORD CONST_KEYWORD VOLATILE_KEYWORD DOUBLE_KEYWORD BOOL
293 %token ELLIPSIS RESTRICT VOID FLOAT_KEYWORD CHAR WCHAR_T
295 %token <opname> ASSIGN_MODIFY
297 /* C++ */
298 %token TRUEKEYWORD
299 %token FALSEKEYWORD
301 /* Non-C++ things we get from the demangler. */
302 %token <lval> DEMANGLER_SPECIAL
303 %token CONSTRUCTION_VTABLE CONSTRUCTION_IN
305 /* Precedence declarations. */
307 /* Give NAME lower precedence than COLONCOLON, so that nested_name will
308 associate greedily. */
309 %nonassoc NAME
311 /* Give NEW and DELETE lower precedence than ']', because we can not
312 have an array of type operator new. This causes NEW '[' to be
313 parsed as operator new[]. */
314 %nonassoc NEW DELETE
316 /* Give VOID higher precedence than NAME. Then we can use %prec NAME
317 to prefer (VOID) to (function_args). */
318 %nonassoc VOID
320 /* Give VOID lower precedence than ')' for similar reasons. */
321 %nonassoc ')'
323 %left ','
324 %right '=' ASSIGN_MODIFY
325 %right '?'
326 %left OROR
327 %left ANDAND
328 %left '|'
329 %left '^'
330 %left '&'
331 %left EQUAL NOTEQUAL
332 %left '<' '>' LEQ GEQ
333 %left LSH RSH
334 %left '@'
335 %left '+' '-'
336 %left '*' '/' '%'
337 %right UNARY INCREMENT DECREMENT
339 /* We don't need a precedence for '(' in this reduced grammar, and it
340 can mask some unpleasant bugs, so disable it for now. */
342 %right ARROW '.' '[' /* '(' */
343 %left COLONCOLON
348 result : start
349 { state->global_result = $1; }
352 start : type
354 | demangler_special
356 | function
360 start_opt : /* */
361 { $$ = NULL; }
362 | COLONCOLON start
363 { $$ = $2; }
366 function
367 /* Function with a return type. declarator_1 is used to prevent
368 ambiguity with the next rule. */
369 : typespec_2 declarator_1
370 { $$ = $2.comp;
371 *$2.last = $1;
374 /* Function without a return type. We need to use typespec_2
375 to prevent conflicts from qualifiers_opt - harmless. The
376 start_opt is used to handle "function-local" variables and
377 types. */
378 | typespec_2 function_arglist start_opt
379 { $$ = state->fill_comp (DEMANGLE_COMPONENT_TYPED_NAME,
380 $1, $2.comp);
381 if ($3)
382 $$ = state->fill_comp (DEMANGLE_COMPONENT_LOCAL_NAME,
383 $$, $3);
385 | colon_ext_only function_arglist start_opt
386 { $$ = state->fill_comp (DEMANGLE_COMPONENT_TYPED_NAME, $1, $2.comp);
387 if ($3) $$ = state->fill_comp (DEMANGLE_COMPONENT_LOCAL_NAME, $$, $3); }
389 | conversion_op_name start_opt
390 { $$ = $1.comp;
391 if ($2) $$ = state->fill_comp (DEMANGLE_COMPONENT_LOCAL_NAME, $$, $2); }
392 | conversion_op_name abstract_declarator_fn
393 { if ($2.last)
395 /* First complete the abstract_declarator's type using
396 the typespec from the conversion_op_name. */
397 *$2.last = *$1.last;
398 /* Then complete the conversion_op_name with the type. */
399 *$1.last = $2.comp;
401 /* If we have an arglist, build a function type. */
402 if ($2.fn.comp)
403 $$ = state->fill_comp (DEMANGLE_COMPONENT_TYPED_NAME, $1.comp, $2.fn.comp);
404 else
405 $$ = $1.comp;
406 if ($2.start) $$ = state->fill_comp (DEMANGLE_COMPONENT_LOCAL_NAME, $$, $2.start);
410 demangler_special
411 : DEMANGLER_SPECIAL start
412 { $$ = state->fill_comp ((enum demangle_component_type) $1, $2, NULL); }
413 | CONSTRUCTION_VTABLE start CONSTRUCTION_IN start
414 { $$ = state->fill_comp (DEMANGLE_COMPONENT_CONSTRUCTION_VTABLE, $2, $4); }
417 oper : OPERATOR NEW
419 /* Match the whitespacing of cplus_demangle_operators.
420 It would abort on unrecognized string otherwise. */
421 $$ = state->make_operator ("new", 3);
423 | OPERATOR DELETE
425 /* Match the whitespacing of cplus_demangle_operators.
426 It would abort on unrecognized string otherwise. */
427 $$ = state->make_operator ("delete ", 1);
429 | OPERATOR NEW '[' ']'
431 /* Match the whitespacing of cplus_demangle_operators.
432 It would abort on unrecognized string otherwise. */
433 $$ = state->make_operator ("new[]", 3);
435 | OPERATOR DELETE '[' ']'
437 /* Match the whitespacing of cplus_demangle_operators.
438 It would abort on unrecognized string otherwise. */
439 $$ = state->make_operator ("delete[] ", 1);
441 | OPERATOR '+'
442 { $$ = state->make_operator ("+", 2); }
443 | OPERATOR '-'
444 { $$ = state->make_operator ("-", 2); }
445 | OPERATOR '*'
446 { $$ = state->make_operator ("*", 2); }
447 | OPERATOR '/'
448 { $$ = state->make_operator ("/", 2); }
449 | OPERATOR '%'
450 { $$ = state->make_operator ("%", 2); }
451 | OPERATOR '^'
452 { $$ = state->make_operator ("^", 2); }
453 | OPERATOR '&'
454 { $$ = state->make_operator ("&", 2); }
455 | OPERATOR '|'
456 { $$ = state->make_operator ("|", 2); }
457 | OPERATOR '~'
458 { $$ = state->make_operator ("~", 1); }
459 | OPERATOR '!'
460 { $$ = state->make_operator ("!", 1); }
461 | OPERATOR '='
462 { $$ = state->make_operator ("=", 2); }
463 | OPERATOR '<'
464 { $$ = state->make_operator ("<", 2); }
465 | OPERATOR '>'
466 { $$ = state->make_operator (">", 2); }
467 | OPERATOR ASSIGN_MODIFY
468 { $$ = state->make_operator ($2, 2); }
469 | OPERATOR LSH
470 { $$ = state->make_operator ("<<", 2); }
471 | OPERATOR RSH
472 { $$ = state->make_operator (">>", 2); }
473 | OPERATOR EQUAL
474 { $$ = state->make_operator ("==", 2); }
475 | OPERATOR NOTEQUAL
476 { $$ = state->make_operator ("!=", 2); }
477 | OPERATOR LEQ
478 { $$ = state->make_operator ("<=", 2); }
479 | OPERATOR GEQ
480 { $$ = state->make_operator (">=", 2); }
481 | OPERATOR ANDAND
482 { $$ = state->make_operator ("&&", 2); }
483 | OPERATOR OROR
484 { $$ = state->make_operator ("||", 2); }
485 | OPERATOR INCREMENT
486 { $$ = state->make_operator ("++", 1); }
487 | OPERATOR DECREMENT
488 { $$ = state->make_operator ("--", 1); }
489 | OPERATOR ','
490 { $$ = state->make_operator (",", 2); }
491 | OPERATOR ARROW '*'
492 { $$ = state->make_operator ("->*", 2); }
493 | OPERATOR ARROW
494 { $$ = state->make_operator ("->", 2); }
495 | OPERATOR '(' ')'
496 { $$ = state->make_operator ("()", 2); }
497 | OPERATOR '[' ']'
498 { $$ = state->make_operator ("[]", 2); }
501 /* Conversion operators. We don't try to handle some of
502 the wackier demangler output for function pointers,
503 since it's not clear that it's parseable. */
504 conversion_op
505 : OPERATOR typespec_2
506 { $$ = state->fill_comp (DEMANGLE_COMPONENT_CONVERSION, $2, NULL); }
509 conversion_op_name
510 : nested_name conversion_op
511 { $$.comp = $1.comp;
512 d_right ($1.last) = $2;
513 $$.last = &d_left ($2);
515 | conversion_op
516 { $$.comp = $1;
517 $$.last = &d_left ($1);
519 | COLONCOLON nested_name conversion_op
520 { $$.comp = $2.comp;
521 d_right ($2.last) = $3;
522 $$.last = &d_left ($3);
524 | COLONCOLON conversion_op
525 { $$.comp = $2;
526 $$.last = &d_left ($2);
530 /* DEMANGLE_COMPONENT_NAME */
531 /* This accepts certain invalid placements of '~'. */
532 unqualified_name: oper
533 | oper '<' template_params '>'
534 { $$ = state->fill_comp (DEMANGLE_COMPONENT_TEMPLATE, $1, $3.comp); }
535 | '~' NAME
536 { $$ = state->make_dtor (gnu_v3_complete_object_dtor, $2); }
539 /* This rule is used in name and nested_name, and expanded inline there
540 for efficiency. */
542 scope_id : NAME
543 | template
547 colon_name : name
548 | COLONCOLON name
549 { $$ = $2; }
552 /* DEMANGLE_COMPONENT_QUAL_NAME */
553 /* DEMANGLE_COMPONENT_CTOR / DEMANGLE_COMPONENT_DTOR ? */
554 name : nested_name NAME %prec NAME
555 { $$ = $1.comp; d_right ($1.last) = $2; }
556 | NAME %prec NAME
557 | nested_name templ %prec NAME
558 { $$ = $1.comp; d_right ($1.last) = $2; }
559 | templ %prec NAME
562 colon_ext_name : colon_name
563 | colon_ext_only
566 colon_ext_only : ext_only_name
567 | COLONCOLON ext_only_name
568 { $$ = $2; }
571 ext_only_name : nested_name unqualified_name
572 { $$ = $1.comp; d_right ($1.last) = $2; }
573 | unqualified_name
576 nested_name : NAME COLONCOLON
577 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_QUAL_NAME, $1, NULL);
578 $$.last = $$.comp;
580 | nested_name NAME COLONCOLON
581 { $$.comp = $1.comp;
582 d_right ($1.last) = state->fill_comp (DEMANGLE_COMPONENT_QUAL_NAME, $2, NULL);
583 $$.last = d_right ($1.last);
585 | templ COLONCOLON
586 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_QUAL_NAME, $1, NULL);
587 $$.last = $$.comp;
589 | nested_name templ COLONCOLON
590 { $$.comp = $1.comp;
591 d_right ($1.last) = state->fill_comp (DEMANGLE_COMPONENT_QUAL_NAME, $2, NULL);
592 $$.last = d_right ($1.last);
596 /* DEMANGLE_COMPONENT_TEMPLATE */
597 /* DEMANGLE_COMPONENT_TEMPLATE_ARGLIST */
598 templ : NAME '<' template_params '>'
599 { $$ = state->fill_comp (DEMANGLE_COMPONENT_TEMPLATE, $1, $3.comp); }
602 template_params : template_arg
603 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_TEMPLATE_ARGLIST, $1, NULL);
604 $$.last = &d_right ($$.comp); }
605 | template_params ',' template_arg
606 { $$.comp = $1.comp;
607 *$1.last = state->fill_comp (DEMANGLE_COMPONENT_TEMPLATE_ARGLIST, $3, NULL);
608 $$.last = &d_right (*$1.last);
612 /* "type" is inlined into template_arg and function_args. */
614 /* Also an integral constant-expression of integral type, and a
615 pointer to member (?) */
616 template_arg : typespec_2
617 | typespec_2 abstract_declarator
618 { $$ = $2.comp;
619 *$2.last = $1;
621 | '&' start
622 { $$ = state->fill_comp (DEMANGLE_COMPONENT_UNARY, state->make_operator ("&", 1), $2); }
623 | '&' '(' start ')'
624 { $$ = state->fill_comp (DEMANGLE_COMPONENT_UNARY, state->make_operator ("&", 1), $3); }
625 | exp
628 function_args : typespec_2
629 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_ARGLIST, $1, NULL);
630 $$.last = &d_right ($$.comp);
632 | typespec_2 abstract_declarator
633 { *$2.last = $1;
634 $$.comp = state->fill_comp (DEMANGLE_COMPONENT_ARGLIST, $2.comp, NULL);
635 $$.last = &d_right ($$.comp);
637 | function_args ',' typespec_2
638 { *$1.last = state->fill_comp (DEMANGLE_COMPONENT_ARGLIST, $3, NULL);
639 $$.comp = $1.comp;
640 $$.last = &d_right (*$1.last);
642 | function_args ',' typespec_2 abstract_declarator
643 { *$4.last = $3;
644 *$1.last = state->fill_comp (DEMANGLE_COMPONENT_ARGLIST, $4.comp, NULL);
645 $$.comp = $1.comp;
646 $$.last = &d_right (*$1.last);
648 | function_args ',' ELLIPSIS
649 { *$1.last
650 = state->fill_comp (DEMANGLE_COMPONENT_ARGLIST,
651 state->make_builtin_type ("..."),
652 NULL);
653 $$.comp = $1.comp;
654 $$.last = &d_right (*$1.last);
658 function_arglist: '(' function_args ')' qualifiers_opt %prec NAME
659 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_FUNCTION_TYPE, NULL, $2.comp);
660 $$.last = &d_left ($$.comp);
661 $$.comp = state->d_qualify ($$.comp, $4, 1); }
662 | '(' VOID ')' qualifiers_opt
663 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_FUNCTION_TYPE, NULL, NULL);
664 $$.last = &d_left ($$.comp);
665 $$.comp = state->d_qualify ($$.comp, $4, 1); }
666 | '(' ')' qualifiers_opt
667 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_FUNCTION_TYPE, NULL, NULL);
668 $$.last = &d_left ($$.comp);
669 $$.comp = state->d_qualify ($$.comp, $3, 1); }
672 /* Should do something about DEMANGLE_COMPONENT_VENDOR_TYPE_QUAL */
673 qualifiers_opt : /* epsilon */
674 { $$ = 0; }
675 | qualifiers
678 qualifier : RESTRICT
679 { $$ = QUAL_RESTRICT; }
680 | VOLATILE_KEYWORD
681 { $$ = QUAL_VOLATILE; }
682 | CONST_KEYWORD
683 { $$ = QUAL_CONST; }
686 qualifiers : qualifier
687 | qualifier qualifiers
688 { $$ = $1 | $2; }
691 /* This accepts all sorts of invalid constructions and produces
692 invalid output for them - an error would be better. */
694 int_part : INT_KEYWORD
695 { $$ = 0; }
696 | SIGNED_KEYWORD
697 { $$ = INT_SIGNED; }
698 | UNSIGNED
699 { $$ = INT_UNSIGNED; }
700 | CHAR
701 { $$ = INT_CHAR; }
702 | LONG
703 { $$ = INT_LONG; }
704 | SHORT
705 { $$ = INT_SHORT; }
708 int_seq : int_part
709 | int_seq int_part
710 { $$ = $1 | $2; if ($1 & $2 & INT_LONG) $$ = $1 | INT_LLONG; }
713 builtin_type : int_seq
714 { $$ = state->d_int_type ($1); }
715 | FLOAT_KEYWORD
716 { $$ = state->make_builtin_type ("float"); }
717 | DOUBLE_KEYWORD
718 { $$ = state->make_builtin_type ("double"); }
719 | LONG DOUBLE_KEYWORD
720 { $$ = state->make_builtin_type ("long double"); }
721 | BOOL
722 { $$ = state->make_builtin_type ("bool"); }
723 | WCHAR_T
724 { $$ = state->make_builtin_type ("wchar_t"); }
725 | VOID
726 { $$ = state->make_builtin_type ("void"); }
729 ptr_operator : '*' qualifiers_opt
730 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_POINTER, NULL, NULL);
731 $$.last = &d_left ($$.comp);
732 $$.comp = state->d_qualify ($$.comp, $2, 0); }
733 /* g++ seems to allow qualifiers after the reference? */
734 | '&'
735 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_REFERENCE, NULL, NULL);
736 $$.last = &d_left ($$.comp); }
737 | ANDAND
738 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_RVALUE_REFERENCE, NULL, NULL);
739 $$.last = &d_left ($$.comp); }
740 | nested_name '*' qualifiers_opt
741 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_PTRMEM_TYPE, $1.comp, NULL);
742 /* Convert the innermost DEMANGLE_COMPONENT_QUAL_NAME to a DEMANGLE_COMPONENT_NAME. */
743 *$1.last = *d_left ($1.last);
744 $$.last = &d_right ($$.comp);
745 $$.comp = state->d_qualify ($$.comp, $3, 0); }
746 | COLONCOLON nested_name '*' qualifiers_opt
747 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_PTRMEM_TYPE, $2.comp, NULL);
748 /* Convert the innermost DEMANGLE_COMPONENT_QUAL_NAME to a DEMANGLE_COMPONENT_NAME. */
749 *$2.last = *d_left ($2.last);
750 $$.last = &d_right ($$.comp);
751 $$.comp = state->d_qualify ($$.comp, $4, 0); }
754 array_indicator : '[' ']'
755 { $$ = state->fill_comp (DEMANGLE_COMPONENT_ARRAY_TYPE, NULL, NULL); }
756 | '[' INT ']'
757 { $$ = state->fill_comp (DEMANGLE_COMPONENT_ARRAY_TYPE, $2, NULL); }
760 /* Details of this approach inspired by the G++ < 3.4 parser. */
762 /* This rule is only used in typespec_2, and expanded inline there for
763 efficiency. */
765 typespec : builtin_type
766 | colon_name
770 typespec_2 : builtin_type qualifiers
771 { $$ = state->d_qualify ($1, $2, 0); }
772 | builtin_type
773 | qualifiers builtin_type qualifiers
774 { $$ = state->d_qualify ($2, $1 | $3, 0); }
775 | qualifiers builtin_type
776 { $$ = state->d_qualify ($2, $1, 0); }
778 | name qualifiers
779 { $$ = state->d_qualify ($1, $2, 0); }
780 | name
781 | qualifiers name qualifiers
782 { $$ = state->d_qualify ($2, $1 | $3, 0); }
783 | qualifiers name
784 { $$ = state->d_qualify ($2, $1, 0); }
786 | COLONCOLON name qualifiers
787 { $$ = state->d_qualify ($2, $3, 0); }
788 | COLONCOLON name
789 { $$ = $2; }
790 | qualifiers COLONCOLON name qualifiers
791 { $$ = state->d_qualify ($3, $1 | $4, 0); }
792 | qualifiers COLONCOLON name
793 { $$ = state->d_qualify ($3, $1, 0); }
796 abstract_declarator
797 : ptr_operator
798 { $$.comp = $1.comp; $$.last = $1.last;
799 $$.fn.comp = NULL; $$.fn.last = NULL; }
800 | ptr_operator abstract_declarator
801 { $$ = $2; $$.fn.comp = NULL; $$.fn.last = NULL;
802 if ($2.fn.comp) { $$.last = $2.fn.last; *$2.last = $2.fn.comp; }
803 *$$.last = $1.comp;
804 $$.last = $1.last; }
805 | direct_abstract_declarator
806 { $$.fn.comp = NULL; $$.fn.last = NULL;
807 if ($1.fn.comp) { $$.last = $1.fn.last; *$1.last = $1.fn.comp; }
811 direct_abstract_declarator
812 : '(' abstract_declarator ')'
813 { $$ = $2; $$.fn.comp = NULL; $$.fn.last = NULL; $$.fold_flag = 1;
814 if ($2.fn.comp) { $$.last = $2.fn.last; *$2.last = $2.fn.comp; }
816 | direct_abstract_declarator function_arglist
817 { $$.fold_flag = 0;
818 if ($1.fn.comp) { $$.last = $1.fn.last; *$1.last = $1.fn.comp; }
819 if ($1.fold_flag)
821 *$$.last = $2.comp;
822 $$.last = $2.last;
824 else
825 $$.fn = $2;
827 | direct_abstract_declarator array_indicator
828 { $$.fn.comp = NULL; $$.fn.last = NULL; $$.fold_flag = 0;
829 if ($1.fn.comp) { $$.last = $1.fn.last; *$1.last = $1.fn.comp; }
830 *$1.last = $2;
831 $$.last = &d_right ($2);
833 | array_indicator
834 { $$.fn.comp = NULL; $$.fn.last = NULL; $$.fold_flag = 0;
835 $$.comp = $1;
836 $$.last = &d_right ($1);
838 /* G++ has the following except for () and (type). Then
839 (type) is handled in regcast_or_absdcl and () is handled
840 in fcast_or_absdcl.
842 However, this is only useful for function types, and
843 generates reduce/reduce conflicts with direct_declarator.
844 We're interested in pointer-to-function types, and in
845 functions, but not in function types - so leave this
846 out. */
847 /* | function_arglist */
850 abstract_declarator_fn
851 : ptr_operator
852 { $$.comp = $1.comp; $$.last = $1.last;
853 $$.fn.comp = NULL; $$.fn.last = NULL; $$.start = NULL; }
854 | ptr_operator abstract_declarator_fn
855 { $$ = $2;
856 if ($2.last)
857 *$$.last = $1.comp;
858 else
859 $$.comp = $1.comp;
860 $$.last = $1.last;
862 | direct_abstract_declarator
863 { $$.comp = $1.comp; $$.last = $1.last; $$.fn = $1.fn; $$.start = NULL; }
864 | direct_abstract_declarator function_arglist COLONCOLON start
865 { $$.start = $4;
866 if ($1.fn.comp) { $$.last = $1.fn.last; *$1.last = $1.fn.comp; }
867 if ($1.fold_flag)
869 *$$.last = $2.comp;
870 $$.last = $2.last;
872 else
873 $$.fn = $2;
875 | function_arglist start_opt
876 { $$.fn = $1;
877 $$.start = $2;
878 $$.comp = NULL; $$.last = NULL;
882 type : typespec_2
883 | typespec_2 abstract_declarator
884 { $$ = $2.comp;
885 *$2.last = $1;
889 declarator : ptr_operator declarator
890 { $$.comp = $2.comp;
891 $$.last = $1.last;
892 *$2.last = $1.comp; }
893 | direct_declarator
896 direct_declarator
897 : '(' declarator ')'
898 { $$ = $2; }
899 | direct_declarator function_arglist
900 { $$.comp = $1.comp;
901 *$1.last = $2.comp;
902 $$.last = $2.last;
904 | direct_declarator array_indicator
905 { $$.comp = $1.comp;
906 *$1.last = $2;
907 $$.last = &d_right ($2);
909 | colon_ext_name
910 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_TYPED_NAME, $1, NULL);
911 $$.last = &d_right ($$.comp);
915 /* These are similar to declarator and direct_declarator except that they
916 do not permit ( colon_ext_name ), which is ambiguous with a function
917 argument list. They also don't permit a few other forms with redundant
918 parentheses around the colon_ext_name; any colon_ext_name in parentheses
919 must be followed by an argument list or an array indicator, or preceded
920 by a pointer. */
921 declarator_1 : ptr_operator declarator_1
922 { $$.comp = $2.comp;
923 $$.last = $1.last;
924 *$2.last = $1.comp; }
925 | colon_ext_name
926 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_TYPED_NAME, $1, NULL);
927 $$.last = &d_right ($$.comp);
929 | direct_declarator_1
931 /* Function local variable or type. The typespec to
932 our left is the type of the containing function.
933 This should be OK, because function local types
934 can not be templates, so the return types of their
935 members will not be mangled. If they are hopefully
936 they'll end up to the right of the ::. */
937 | colon_ext_name function_arglist COLONCOLON start
938 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_TYPED_NAME, $1, $2.comp);
939 $$.last = $2.last;
940 $$.comp = state->fill_comp (DEMANGLE_COMPONENT_LOCAL_NAME, $$.comp, $4);
942 | direct_declarator_1 function_arglist COLONCOLON start
943 { $$.comp = $1.comp;
944 *$1.last = $2.comp;
945 $$.last = $2.last;
946 $$.comp = state->fill_comp (DEMANGLE_COMPONENT_LOCAL_NAME, $$.comp, $4);
950 direct_declarator_1
951 : '(' ptr_operator declarator ')'
952 { $$.comp = $3.comp;
953 $$.last = $2.last;
954 *$3.last = $2.comp; }
955 | direct_declarator_1 function_arglist
956 { $$.comp = $1.comp;
957 *$1.last = $2.comp;
958 $$.last = $2.last;
960 | direct_declarator_1 array_indicator
961 { $$.comp = $1.comp;
962 *$1.last = $2;
963 $$.last = &d_right ($2);
965 | colon_ext_name function_arglist
966 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_TYPED_NAME, $1, $2.comp);
967 $$.last = $2.last;
969 | colon_ext_name array_indicator
970 { $$.comp = state->fill_comp (DEMANGLE_COMPONENT_TYPED_NAME, $1, $2);
971 $$.last = &d_right ($2);
975 exp : '(' exp1 ')'
976 { $$ = $2; }
979 /* Silly trick. Only allow '>' when parenthesized, in order to
980 handle conflict with templates. */
981 exp1 : exp
984 exp1 : exp '>' exp
985 { $$ = state->d_binary (">", $1, $3); }
988 /* References. Not allowed everywhere in template parameters, only
989 at the top level, but treat them as expressions in case they are wrapped
990 in parentheses. */
991 exp1 : '&' start
992 { $$ = state->fill_comp (DEMANGLE_COMPONENT_UNARY, state->make_operator ("&", 1), $2); }
993 | '&' '(' start ')'
994 { $$ = state->fill_comp (DEMANGLE_COMPONENT_UNARY, state->make_operator ("&", 1), $3); }
997 /* Expressions, not including the comma operator. */
998 exp : '-' exp %prec UNARY
999 { $$ = state->d_unary ("-", $2); }
1002 exp : '!' exp %prec UNARY
1003 { $$ = state->d_unary ("!", $2); }
1006 exp : '~' exp %prec UNARY
1007 { $$ = state->d_unary ("~", $2); }
1010 /* Casts. First your normal C-style cast. If exp is a LITERAL, just change
1011 its type. */
1013 exp : '(' type ')' exp %prec UNARY
1014 { if ($4->type == DEMANGLE_COMPONENT_LITERAL
1015 || $4->type == DEMANGLE_COMPONENT_LITERAL_NEG)
1017 $$ = $4;
1018 d_left ($4) = $2;
1020 else
1021 $$ = state->fill_comp (DEMANGLE_COMPONENT_UNARY,
1022 state->fill_comp (DEMANGLE_COMPONENT_CAST, $2, NULL),
1023 $4);
1027 /* Mangling does not differentiate between these, so we don't need to
1028 either. */
1029 exp : STATIC_CAST '<' type '>' '(' exp1 ')' %prec UNARY
1030 { $$ = state->fill_comp (DEMANGLE_COMPONENT_UNARY,
1031 state->fill_comp (DEMANGLE_COMPONENT_CAST, $3, NULL),
1032 $6);
1036 exp : DYNAMIC_CAST '<' type '>' '(' exp1 ')' %prec UNARY
1037 { $$ = state->fill_comp (DEMANGLE_COMPONENT_UNARY,
1038 state->fill_comp (DEMANGLE_COMPONENT_CAST, $3, NULL),
1039 $6);
1043 exp : REINTERPRET_CAST '<' type '>' '(' exp1 ')' %prec UNARY
1044 { $$ = state->fill_comp (DEMANGLE_COMPONENT_UNARY,
1045 state->fill_comp (DEMANGLE_COMPONENT_CAST, $3, NULL),
1046 $6);
1050 /* Another form of C++-style cast is "type ( exp1 )". This creates too many
1051 conflicts to support. For a while we supported the simpler
1052 "typespec_2 ( exp1 )", but that conflicts with "& ( start )" as a
1053 reference, deep within the wilderness of abstract declarators:
1054 Qux<int(&(*))> vs Qux<int(&(var))>, a shift-reduce conflict at the
1055 innermost left parenthesis. So we do not support function-like casts.
1056 Fortunately they never appear in demangler output. */
1058 /* TO INVESTIGATE: ._0 style anonymous names; anonymous namespaces */
1060 /* Binary operators in order of decreasing precedence. */
1062 exp : exp '*' exp
1063 { $$ = state->d_binary ("*", $1, $3); }
1066 exp : exp '/' exp
1067 { $$ = state->d_binary ("/", $1, $3); }
1070 exp : exp '%' exp
1071 { $$ = state->d_binary ("%", $1, $3); }
1074 exp : exp '+' exp
1075 { $$ = state->d_binary ("+", $1, $3); }
1078 exp : exp '-' exp
1079 { $$ = state->d_binary ("-", $1, $3); }
1082 exp : exp LSH exp
1083 { $$ = state->d_binary ("<<", $1, $3); }
1086 exp : exp RSH exp
1087 { $$ = state->d_binary (">>", $1, $3); }
1090 exp : exp EQUAL exp
1091 { $$ = state->d_binary ("==", $1, $3); }
1094 exp : exp NOTEQUAL exp
1095 { $$ = state->d_binary ("!=", $1, $3); }
1098 exp : exp LEQ exp
1099 { $$ = state->d_binary ("<=", $1, $3); }
1102 exp : exp GEQ exp
1103 { $$ = state->d_binary (">=", $1, $3); }
1106 exp : exp '<' exp
1107 { $$ = state->d_binary ("<", $1, $3); }
1110 exp : exp '&' exp
1111 { $$ = state->d_binary ("&", $1, $3); }
1114 exp : exp '^' exp
1115 { $$ = state->d_binary ("^", $1, $3); }
1118 exp : exp '|' exp
1119 { $$ = state->d_binary ("|", $1, $3); }
1122 exp : exp ANDAND exp
1123 { $$ = state->d_binary ("&&", $1, $3); }
1126 exp : exp OROR exp
1127 { $$ = state->d_binary ("||", $1, $3); }
1130 /* Not 100% sure these are necessary, but they're harmless. */
1131 exp : exp ARROW NAME
1132 { $$ = state->d_binary ("->", $1, $3); }
1135 exp : exp '.' NAME
1136 { $$ = state->d_binary (".", $1, $3); }
1139 exp : exp '?' exp ':' exp %prec '?'
1140 { $$ = state->fill_comp (DEMANGLE_COMPONENT_TRINARY, state->make_operator ("?", 3),
1141 state->fill_comp (DEMANGLE_COMPONENT_TRINARY_ARG1, $1,
1142 state->fill_comp (DEMANGLE_COMPONENT_TRINARY_ARG2, $3, $5)));
1146 exp : INT
1149 /* Not generally allowed. */
1150 exp : FLOAT
1153 exp : SIZEOF '(' type ')' %prec UNARY
1155 /* Match the whitespacing of cplus_demangle_operators.
1156 It would abort on unrecognized string otherwise. */
1157 $$ = state->d_unary ("sizeof ", $3);
1161 /* C++. */
1162 exp : TRUEKEYWORD
1163 { struct demangle_component *i;
1164 i = state->make_name ("1", 1);
1165 $$ = state->fill_comp (DEMANGLE_COMPONENT_LITERAL,
1166 state->make_builtin_type ( "bool"),
1171 exp : FALSEKEYWORD
1172 { struct demangle_component *i;
1173 i = state->make_name ("0", 1);
1174 $$ = state->fill_comp (DEMANGLE_COMPONENT_LITERAL,
1175 state->make_builtin_type ("bool"),
1180 /* end of C++. */
1184 /* Apply QUALIFIERS to LHS and return a qualified component. IS_METHOD
1185 is set if LHS is a method, in which case the qualifiers are logically
1186 applied to "this". We apply qualifiers in a consistent order; LHS
1187 may already be qualified; duplicate qualifiers are not created. */
1189 struct demangle_component *
1190 cpname_state::d_qualify (struct demangle_component *lhs, int qualifiers,
1191 int is_method)
1193 struct demangle_component **inner_p;
1194 enum demangle_component_type type;
1196 /* For now the order is CONST (innermost), VOLATILE, RESTRICT. */
1198 #define HANDLE_QUAL(TYPE, MTYPE, QUAL) \
1199 if ((qualifiers & QUAL) && (type != TYPE) && (type != MTYPE)) \
1201 *inner_p = fill_comp (is_method ? MTYPE : TYPE, \
1202 *inner_p, NULL); \
1203 inner_p = &d_left (*inner_p); \
1204 type = (*inner_p)->type; \
1206 else if (type == TYPE || type == MTYPE) \
1208 inner_p = &d_left (*inner_p); \
1209 type = (*inner_p)->type; \
1212 inner_p = &lhs;
1214 type = (*inner_p)->type;
1216 HANDLE_QUAL (DEMANGLE_COMPONENT_RESTRICT, DEMANGLE_COMPONENT_RESTRICT_THIS, QUAL_RESTRICT);
1217 HANDLE_QUAL (DEMANGLE_COMPONENT_VOLATILE, DEMANGLE_COMPONENT_VOLATILE_THIS, QUAL_VOLATILE);
1218 HANDLE_QUAL (DEMANGLE_COMPONENT_CONST, DEMANGLE_COMPONENT_CONST_THIS, QUAL_CONST);
1220 return lhs;
1223 /* Return a builtin type corresponding to FLAGS. */
1225 struct demangle_component *
1226 cpname_state::d_int_type (int flags)
1228 const char *name;
1230 switch (flags)
1232 case INT_SIGNED | INT_CHAR:
1233 name = "signed char";
1234 break;
1235 case INT_CHAR:
1236 name = "char";
1237 break;
1238 case INT_UNSIGNED | INT_CHAR:
1239 name = "unsigned char";
1240 break;
1241 case 0:
1242 case INT_SIGNED:
1243 name = "int";
1244 break;
1245 case INT_UNSIGNED:
1246 name = "unsigned int";
1247 break;
1248 case INT_LONG:
1249 case INT_SIGNED | INT_LONG:
1250 name = "long";
1251 break;
1252 case INT_UNSIGNED | INT_LONG:
1253 name = "unsigned long";
1254 break;
1255 case INT_SHORT:
1256 case INT_SIGNED | INT_SHORT:
1257 name = "short";
1258 break;
1259 case INT_UNSIGNED | INT_SHORT:
1260 name = "unsigned short";
1261 break;
1262 case INT_LLONG | INT_LONG:
1263 case INT_SIGNED | INT_LLONG | INT_LONG:
1264 name = "long long";
1265 break;
1266 case INT_UNSIGNED | INT_LLONG | INT_LONG:
1267 name = "unsigned long long";
1268 break;
1269 default:
1270 return NULL;
1273 return make_builtin_type (name);
1276 /* Wrapper to create a unary operation. */
1278 struct demangle_component *
1279 cpname_state::d_unary (const char *name, struct demangle_component *lhs)
1281 return fill_comp (DEMANGLE_COMPONENT_UNARY, make_operator (name, 1), lhs);
1284 /* Wrapper to create a binary operation. */
1286 struct demangle_component *
1287 cpname_state::d_binary (const char *name, struct demangle_component *lhs,
1288 struct demangle_component *rhs)
1290 return fill_comp (DEMANGLE_COMPONENT_BINARY, make_operator (name, 2),
1291 fill_comp (DEMANGLE_COMPONENT_BINARY_ARGS, lhs, rhs));
1294 /* Find the end of a symbol name starting at LEXPTR. */
1296 static const char *
1297 symbol_end (const char *lexptr)
1299 const char *p = lexptr;
1301 while (*p && (c_ident_is_alnum (*p) || *p == '_' || *p == '$' || *p == '.'))
1302 p++;
1304 return p;
1307 /* Take care of parsing a number (anything that starts with a digit).
1308 The number starts at P and contains LEN characters. Store the result in
1309 YYLVAL. */
1312 cpname_state::parse_number (const char *p, int len, int parsed_float,
1313 YYSTYPE *lvalp)
1315 int unsigned_p = 0;
1317 /* Number of "L" suffixes encountered. */
1318 int long_p = 0;
1320 struct demangle_component *signed_type;
1321 struct demangle_component *unsigned_type;
1322 struct demangle_component *type, *name;
1323 enum demangle_component_type literal_type;
1325 if (p[0] == '-')
1327 literal_type = DEMANGLE_COMPONENT_LITERAL_NEG;
1328 p++;
1329 len--;
1331 else
1332 literal_type = DEMANGLE_COMPONENT_LITERAL;
1334 if (parsed_float)
1336 /* It's a float since it contains a point or an exponent. */
1337 char c;
1339 /* The GDB lexer checks the result of scanf at this point. Not doing
1340 this leaves our error checking slightly weaker but only for invalid
1341 data. */
1343 /* See if it has `f' or `l' suffix (float or long double). */
1345 c = TOLOWER (p[len - 1]);
1347 if (c == 'f')
1349 len--;
1350 type = make_builtin_type ("float");
1352 else if (c == 'l')
1354 len--;
1355 type = make_builtin_type ("long double");
1357 else if (ISDIGIT (c) || c == '.')
1358 type = make_builtin_type ("double");
1359 else
1360 return ERROR;
1362 name = make_name (p, len);
1363 lvalp->comp = fill_comp (literal_type, type, name);
1365 return FLOAT;
1368 /* This treats 0x1 and 1 as different literals. We also do not
1369 automatically generate unsigned types. */
1371 long_p = 0;
1372 unsigned_p = 0;
1373 while (len > 0)
1375 if (p[len - 1] == 'l' || p[len - 1] == 'L')
1377 len--;
1378 long_p++;
1379 continue;
1381 if (p[len - 1] == 'u' || p[len - 1] == 'U')
1383 len--;
1384 unsigned_p++;
1385 continue;
1387 break;
1390 if (long_p == 0)
1392 unsigned_type = make_builtin_type ("unsigned int");
1393 signed_type = make_builtin_type ("int");
1395 else if (long_p == 1)
1397 unsigned_type = make_builtin_type ("unsigned long");
1398 signed_type = make_builtin_type ("long");
1400 else
1402 unsigned_type = make_builtin_type ("unsigned long long");
1403 signed_type = make_builtin_type ("long long");
1406 if (unsigned_p)
1407 type = unsigned_type;
1408 else
1409 type = signed_type;
1411 name = make_name (p, len);
1412 lvalp->comp = fill_comp (literal_type, type, name);
1414 return INT;
1417 static const char backslashable[] = "abefnrtv";
1418 static const char represented[] = "\a\b\e\f\n\r\t\v";
1420 /* Translate the backslash the way we would in the host character set. */
1421 static int
1422 c_parse_backslash (int host_char, int *target_char)
1424 const char *ix;
1425 ix = strchr (backslashable, host_char);
1426 if (! ix)
1427 return 0;
1428 else
1429 *target_char = represented[ix - backslashable];
1430 return 1;
1433 /* Parse a C escape sequence. STRING_PTR points to a variable
1434 containing a pointer to the string to parse. That pointer
1435 should point to the character after the \. That pointer
1436 is updated past the characters we use. The value of the
1437 escape sequence is returned.
1439 A negative value means the sequence \ newline was seen,
1440 which is supposed to be equivalent to nothing at all.
1442 If \ is followed by a null character, we return a negative
1443 value and leave the string pointer pointing at the null character.
1445 If \ is followed by 000, we return 0 and leave the string pointer
1446 after the zeros. A value of 0 does not mean end of string. */
1448 static int
1449 cp_parse_escape (const char **string_ptr)
1451 int target_char;
1452 int c = *(*string_ptr)++;
1453 if (c_parse_backslash (c, &target_char))
1454 return target_char;
1455 else
1456 switch (c)
1458 case '\n':
1459 return -2;
1460 case 0:
1461 (*string_ptr)--;
1462 return 0;
1463 case '^':
1465 c = *(*string_ptr)++;
1467 if (c == '?')
1468 return 0177;
1469 else if (c == '\\')
1470 target_char = cp_parse_escape (string_ptr);
1471 else
1472 target_char = c;
1474 /* Now target_char is something like `c', and we want to find
1475 its control-character equivalent. */
1476 target_char = target_char & 037;
1478 return target_char;
1481 case '0':
1482 case '1':
1483 case '2':
1484 case '3':
1485 case '4':
1486 case '5':
1487 case '6':
1488 case '7':
1490 int i = c - '0';
1491 int count = 0;
1492 while (++count < 3)
1494 c = (**string_ptr);
1495 if (c >= '0' && c <= '7')
1497 (*string_ptr)++;
1498 i *= 8;
1499 i += c - '0';
1501 else
1503 break;
1506 return i;
1508 default:
1509 return c;
1513 #define HANDLE_SPECIAL(string, comp) \
1514 if (strncmp (tokstart, string, sizeof (string) - 1) == 0) \
1516 state->lexptr = tokstart + sizeof (string) - 1; \
1517 lvalp->lval = comp; \
1518 return DEMANGLER_SPECIAL; \
1521 #define HANDLE_TOKEN2(string, token) \
1522 if (state->lexptr[1] == string[1]) \
1524 state->lexptr += 2; \
1525 lvalp->opname = string; \
1526 return token; \
1529 #define HANDLE_TOKEN3(string, token) \
1530 if (state->lexptr[1] == string[1] && state->lexptr[2] == string[2]) \
1532 state->lexptr += 3; \
1533 lvalp->opname = string; \
1534 return token; \
1537 /* Read one token, getting characters through LEXPTR. */
1539 static int
1540 yylex (YYSTYPE *lvalp, cpname_state *state)
1542 int c;
1543 int namelen;
1544 const char *tokstart;
1546 retry:
1547 state->prev_lexptr = state->lexptr;
1548 tokstart = state->lexptr;
1550 switch (c = *tokstart)
1552 case 0:
1553 return 0;
1555 case ' ':
1556 case '\t':
1557 case '\n':
1558 state->lexptr++;
1559 goto retry;
1561 case '\'':
1562 /* We either have a character constant ('0' or '\177' for example)
1563 or we have a quoted symbol reference ('foo(int,int)' in C++
1564 for example). */
1565 state->lexptr++;
1566 c = *state->lexptr++;
1567 if (c == '\\')
1568 c = cp_parse_escape (&state->lexptr);
1569 else if (c == '\'')
1571 yyerror (state, _("empty character constant"));
1572 return ERROR;
1575 c = *state->lexptr++;
1576 if (c != '\'')
1578 yyerror (state, _("invalid character constant"));
1579 return ERROR;
1582 /* FIXME: We should refer to a canonical form of the character,
1583 presumably the same one that appears in manglings - the decimal
1584 representation. But if that isn't in our input then we have to
1585 allocate memory for it somewhere. */
1586 lvalp->comp
1587 = state->fill_comp (DEMANGLE_COMPONENT_LITERAL,
1588 state->make_builtin_type ("char"),
1589 state->make_name (tokstart,
1590 state->lexptr - tokstart));
1592 return INT;
1594 case '(':
1595 if (strncmp (tokstart, "(anonymous namespace)", 21) == 0)
1597 state->lexptr += 21;
1598 lvalp->comp = state->make_name ("(anonymous namespace)",
1599 sizeof "(anonymous namespace)" - 1);
1600 return NAME;
1602 /* FALL THROUGH */
1604 case ')':
1605 case ',':
1606 state->lexptr++;
1607 return c;
1609 case '.':
1610 if (state->lexptr[1] == '.' && state->lexptr[2] == '.')
1612 state->lexptr += 3;
1613 return ELLIPSIS;
1616 /* Might be a floating point number. */
1617 if (state->lexptr[1] < '0' || state->lexptr[1] > '9')
1618 goto symbol; /* Nope, must be a symbol. */
1620 goto try_number;
1622 case '-':
1623 HANDLE_TOKEN2 ("-=", ASSIGN_MODIFY);
1624 HANDLE_TOKEN2 ("--", DECREMENT);
1625 HANDLE_TOKEN2 ("->", ARROW);
1627 /* For construction vtables. This is kind of hokey. */
1628 if (strncmp (tokstart, "-in-", 4) == 0)
1630 state->lexptr += 4;
1631 return CONSTRUCTION_IN;
1634 if (state->lexptr[1] < '0' || state->lexptr[1] > '9')
1636 state->lexptr++;
1637 return '-';
1639 /* FALL THRU. */
1641 try_number:
1642 case '0':
1643 case '1':
1644 case '2':
1645 case '3':
1646 case '4':
1647 case '5':
1648 case '6':
1649 case '7':
1650 case '8':
1651 case '9':
1653 /* It's a number. */
1654 int got_dot = 0, got_e = 0, toktype;
1655 const char *p = tokstart;
1656 int hex = 0;
1658 if (c == '-')
1659 p++;
1661 if (c == '0' && (p[1] == 'x' || p[1] == 'X'))
1663 p += 2;
1664 hex = 1;
1666 else if (c == '0' && (p[1]=='t' || p[1]=='T' || p[1]=='d' || p[1]=='D'))
1668 p += 2;
1669 hex = 0;
1672 for (;; ++p)
1674 /* This test includes !hex because 'e' is a valid hex digit
1675 and thus does not indicate a floating point number when
1676 the radix is hex. */
1677 if (!hex && !got_e && (*p == 'e' || *p == 'E'))
1678 got_dot = got_e = 1;
1679 /* This test does not include !hex, because a '.' always indicates
1680 a decimal floating point number regardless of the radix.
1682 NOTE drow/2005-03-09: This comment is not accurate in C99;
1683 however, it's not clear that all the floating point support
1684 in this file is doing any good here. */
1685 else if (!got_dot && *p == '.')
1686 got_dot = 1;
1687 else if (got_e && (p[-1] == 'e' || p[-1] == 'E')
1688 && (*p == '-' || *p == '+'))
1689 /* This is the sign of the exponent, not the end of the
1690 number. */
1691 continue;
1692 /* We will take any letters or digits. parse_number will
1693 complain if past the radix, or if L or U are not final. */
1694 else if (! ISALNUM (*p))
1695 break;
1697 toktype = state->parse_number (tokstart, p - tokstart, got_dot|got_e,
1698 lvalp);
1699 if (toktype == ERROR)
1701 char *err_copy = (char *) alloca (p - tokstart + 1);
1703 memcpy (err_copy, tokstart, p - tokstart);
1704 err_copy[p - tokstart] = 0;
1705 yyerror (state, _("invalid number"));
1706 return ERROR;
1708 state->lexptr = p;
1709 return toktype;
1712 case '+':
1713 HANDLE_TOKEN2 ("+=", ASSIGN_MODIFY);
1714 HANDLE_TOKEN2 ("++", INCREMENT);
1715 state->lexptr++;
1716 return c;
1717 case '*':
1718 HANDLE_TOKEN2 ("*=", ASSIGN_MODIFY);
1719 state->lexptr++;
1720 return c;
1721 case '/':
1722 HANDLE_TOKEN2 ("/=", ASSIGN_MODIFY);
1723 state->lexptr++;
1724 return c;
1725 case '%':
1726 HANDLE_TOKEN2 ("%=", ASSIGN_MODIFY);
1727 state->lexptr++;
1728 return c;
1729 case '|':
1730 HANDLE_TOKEN2 ("|=", ASSIGN_MODIFY);
1731 HANDLE_TOKEN2 ("||", OROR);
1732 state->lexptr++;
1733 return c;
1734 case '&':
1735 HANDLE_TOKEN2 ("&=", ASSIGN_MODIFY);
1736 HANDLE_TOKEN2 ("&&", ANDAND);
1737 state->lexptr++;
1738 return c;
1739 case '^':
1740 HANDLE_TOKEN2 ("^=", ASSIGN_MODIFY);
1741 state->lexptr++;
1742 return c;
1743 case '!':
1744 HANDLE_TOKEN2 ("!=", NOTEQUAL);
1745 state->lexptr++;
1746 return c;
1747 case '<':
1748 HANDLE_TOKEN3 ("<<=", ASSIGN_MODIFY);
1749 HANDLE_TOKEN2 ("<=", LEQ);
1750 HANDLE_TOKEN2 ("<<", LSH);
1751 state->lexptr++;
1752 return c;
1753 case '>':
1754 HANDLE_TOKEN3 (">>=", ASSIGN_MODIFY);
1755 HANDLE_TOKEN2 (">=", GEQ);
1756 HANDLE_TOKEN2 (">>", RSH);
1757 state->lexptr++;
1758 return c;
1759 case '=':
1760 HANDLE_TOKEN2 ("==", EQUAL);
1761 state->lexptr++;
1762 return c;
1763 case ':':
1764 HANDLE_TOKEN2 ("::", COLONCOLON);
1765 state->lexptr++;
1766 return c;
1768 case '[':
1769 case ']':
1770 case '?':
1771 case '@':
1772 case '~':
1773 case '{':
1774 case '}':
1775 symbol:
1776 state->lexptr++;
1777 return c;
1779 case '"':
1780 /* These can't occur in C++ names. */
1781 yyerror (state, _("unexpected string literal"));
1782 return ERROR;
1785 if (!(c == '_' || c == '$' || c_ident_is_alpha (c)))
1787 /* We must have come across a bad character (e.g. ';'). */
1788 yyerror (state, _("invalid character"));
1789 return ERROR;
1792 /* It's a name. See how long it is. */
1793 namelen = 0;
1795 c = tokstart[++namelen];
1796 while (c_ident_is_alnum (c) || c == '_' || c == '$');
1798 state->lexptr += namelen;
1800 /* Catch specific keywords. Notice that some of the keywords contain
1801 spaces, and are sorted by the length of the first word. They must
1802 all include a trailing space in the string comparison. */
1803 switch (namelen)
1805 case 16:
1806 if (strncmp (tokstart, "reinterpret_cast", 16) == 0)
1807 return REINTERPRET_CAST;
1808 break;
1809 case 12:
1810 if (strncmp (tokstart, "construction vtable for ", 24) == 0)
1812 state->lexptr = tokstart + 24;
1813 return CONSTRUCTION_VTABLE;
1815 if (strncmp (tokstart, "dynamic_cast", 12) == 0)
1816 return DYNAMIC_CAST;
1817 break;
1818 case 11:
1819 if (strncmp (tokstart, "static_cast", 11) == 0)
1820 return STATIC_CAST;
1821 break;
1822 case 9:
1823 HANDLE_SPECIAL ("covariant return thunk to ", DEMANGLE_COMPONENT_COVARIANT_THUNK);
1824 HANDLE_SPECIAL ("reference temporary for ", DEMANGLE_COMPONENT_REFTEMP);
1825 break;
1826 case 8:
1827 HANDLE_SPECIAL ("typeinfo for ", DEMANGLE_COMPONENT_TYPEINFO);
1828 HANDLE_SPECIAL ("typeinfo fn for ", DEMANGLE_COMPONENT_TYPEINFO_FN);
1829 HANDLE_SPECIAL ("typeinfo name for ", DEMANGLE_COMPONENT_TYPEINFO_NAME);
1830 if (strncmp (tokstart, "operator", 8) == 0)
1831 return OPERATOR;
1832 if (strncmp (tokstart, "restrict", 8) == 0)
1833 return RESTRICT;
1834 if (strncmp (tokstart, "unsigned", 8) == 0)
1835 return UNSIGNED;
1836 if (strncmp (tokstart, "template", 8) == 0)
1837 return TEMPLATE;
1838 if (strncmp (tokstart, "volatile", 8) == 0)
1839 return VOLATILE_KEYWORD;
1840 break;
1841 case 7:
1842 HANDLE_SPECIAL ("virtual thunk to ", DEMANGLE_COMPONENT_VIRTUAL_THUNK);
1843 if (strncmp (tokstart, "wchar_t", 7) == 0)
1844 return WCHAR_T;
1845 break;
1846 case 6:
1847 if (strncmp (tokstart, "global constructors keyed to ", 29) == 0)
1849 const char *p;
1850 state->lexptr = tokstart + 29;
1851 lvalp->lval = DEMANGLE_COMPONENT_GLOBAL_CONSTRUCTORS;
1852 /* Find the end of the symbol. */
1853 p = symbol_end (state->lexptr);
1854 lvalp->comp = state->make_name (state->lexptr, p - state->lexptr);
1855 state->lexptr = p;
1856 return DEMANGLER_SPECIAL;
1858 if (strncmp (tokstart, "global destructors keyed to ", 28) == 0)
1860 const char *p;
1861 state->lexptr = tokstart + 28;
1862 lvalp->lval = DEMANGLE_COMPONENT_GLOBAL_DESTRUCTORS;
1863 /* Find the end of the symbol. */
1864 p = symbol_end (state->lexptr);
1865 lvalp->comp = state->make_name (state->lexptr, p - state->lexptr);
1866 state->lexptr = p;
1867 return DEMANGLER_SPECIAL;
1870 HANDLE_SPECIAL ("vtable for ", DEMANGLE_COMPONENT_VTABLE);
1871 if (strncmp (tokstart, "delete", 6) == 0)
1872 return DELETE;
1873 if (strncmp (tokstart, "struct", 6) == 0)
1874 return STRUCT;
1875 if (strncmp (tokstart, "signed", 6) == 0)
1876 return SIGNED_KEYWORD;
1877 if (strncmp (tokstart, "sizeof", 6) == 0)
1878 return SIZEOF;
1879 if (strncmp (tokstart, "double", 6) == 0)
1880 return DOUBLE_KEYWORD;
1881 break;
1882 case 5:
1883 HANDLE_SPECIAL ("guard variable for ", DEMANGLE_COMPONENT_GUARD);
1884 if (strncmp (tokstart, "false", 5) == 0)
1885 return FALSEKEYWORD;
1886 if (strncmp (tokstart, "class", 5) == 0)
1887 return CLASS;
1888 if (strncmp (tokstart, "union", 5) == 0)
1889 return UNION;
1890 if (strncmp (tokstart, "float", 5) == 0)
1891 return FLOAT_KEYWORD;
1892 if (strncmp (tokstart, "short", 5) == 0)
1893 return SHORT;
1894 if (strncmp (tokstart, "const", 5) == 0)
1895 return CONST_KEYWORD;
1896 break;
1897 case 4:
1898 if (strncmp (tokstart, "void", 4) == 0)
1899 return VOID;
1900 if (strncmp (tokstart, "bool", 4) == 0)
1901 return BOOL;
1902 if (strncmp (tokstart, "char", 4) == 0)
1903 return CHAR;
1904 if (strncmp (tokstart, "enum", 4) == 0)
1905 return ENUM;
1906 if (strncmp (tokstart, "long", 4) == 0)
1907 return LONG;
1908 if (strncmp (tokstart, "true", 4) == 0)
1909 return TRUEKEYWORD;
1910 break;
1911 case 3:
1912 HANDLE_SPECIAL ("VTT for ", DEMANGLE_COMPONENT_VTT);
1913 HANDLE_SPECIAL ("non-virtual thunk to ", DEMANGLE_COMPONENT_THUNK);
1914 if (strncmp (tokstart, "new", 3) == 0)
1915 return NEW;
1916 if (strncmp (tokstart, "int", 3) == 0)
1917 return INT_KEYWORD;
1918 break;
1919 default:
1920 break;
1923 lvalp->comp = state->make_name (tokstart, namelen);
1924 return NAME;
1927 static void
1928 yyerror (cpname_state *state, const char *msg)
1930 if (state->global_errmsg)
1931 return;
1933 state->error_lexptr = state->prev_lexptr;
1934 state->global_errmsg = msg ? msg : "parse error";
1937 /* Allocate a chunk of the components we'll need to build a tree. We
1938 generally allocate too many components, but the extra memory usage
1939 doesn't hurt because the trees are temporary and the storage is
1940 reused. More may be allocated later, by d_grab. */
1941 static struct demangle_info *
1942 allocate_info (void)
1944 struct demangle_info *info = XNEW (struct demangle_info);
1946 info->next = NULL;
1947 info->used = 0;
1948 return info;
1951 /* Convert RESULT to a string. The return value is allocated
1952 using xmalloc. ESTIMATED_LEN is used only as a guide to the
1953 length of the result. This functions handles a few cases that
1954 cplus_demangle_print does not, specifically the global destructor
1955 and constructor labels. */
1957 gdb::unique_xmalloc_ptr<char>
1958 cp_comp_to_string (struct demangle_component *result, int estimated_len)
1960 size_t err;
1962 char *res = cplus_demangle_print (DMGL_PARAMS | DMGL_ANSI,
1963 result, estimated_len, &err);
1964 return gdb::unique_xmalloc_ptr<char> (res);
1967 /* Constructor for demangle_parse_info. */
1969 demangle_parse_info::demangle_parse_info ()
1970 : info (NULL),
1971 tree (NULL)
1973 obstack_init (&obstack);
1976 /* Destructor for demangle_parse_info. */
1978 demangle_parse_info::~demangle_parse_info ()
1980 /* Free any allocated chunks of memory for the parse. */
1981 while (info != NULL)
1983 struct demangle_info *next = info->next;
1985 free (info);
1986 info = next;
1989 /* Free any memory allocated during typedef replacement. */
1990 obstack_free (&obstack, NULL);
1993 /* Merge the two parse trees given by DEST and SRC. The parse tree
1994 in SRC is attached to DEST at the node represented by TARGET.
1996 NOTE 1: Since there is no API to merge obstacks, this function does
1997 even attempt to try it. Fortunately, we do not (yet?) need this ability.
1998 The code will assert if SRC->obstack is not empty.
2000 NOTE 2: The string from which SRC was parsed must not be freed, since
2001 this function will place pointers to that string into DEST. */
2003 void
2004 cp_merge_demangle_parse_infos (struct demangle_parse_info *dest,
2005 struct demangle_component *target,
2006 struct demangle_parse_info *src)
2009 struct demangle_info *di;
2011 /* Copy the SRC's parse data into DEST. */
2012 *target = *src->tree;
2013 di = dest->info;
2014 while (di->next != NULL)
2015 di = di->next;
2016 di->next = src->info;
2018 /* Clear the (pointer to) SRC's parse data so that it is not freed when
2019 cp_demangled_parse_info_free is called. */
2020 src->info = NULL;
2023 /* Convert a demangled name to a demangle_component tree. On success,
2024 a structure containing the root of the new tree is returned. On
2025 error, NULL is returned, and an error message will be set in
2026 *ERRMSG. */
2028 struct std::unique_ptr<demangle_parse_info>
2029 cp_demangled_name_to_comp (const char *demangled_name,
2030 std::string *errmsg)
2032 cpname_state state;
2034 state.prev_lexptr = state.lexptr = demangled_name;
2035 state.error_lexptr = NULL;
2036 state.global_errmsg = NULL;
2038 state.demangle_info = allocate_info ();
2040 std::unique_ptr<demangle_parse_info> result (new demangle_parse_info);
2041 result->info = state.demangle_info;
2043 if (yyparse (&state))
2045 if (state.global_errmsg && errmsg)
2046 *errmsg = state.global_errmsg;
2047 return NULL;
2050 result->tree = state.global_result;
2052 return result;
2055 #ifdef TEST_CPNAMES
2057 static void
2058 cp_print (struct demangle_component *result)
2060 char *str;
2061 size_t err = 0;
2063 str = cplus_demangle_print (DMGL_PARAMS | DMGL_ANSI, result, 64, &err);
2064 if (str == NULL)
2065 return;
2067 fputs (str, stdout);
2069 free (str);
2072 static char
2073 trim_chars (char *lexptr, char **extra_chars)
2075 char *p = (char *) symbol_end (lexptr);
2076 char c = 0;
2078 if (*p)
2080 c = *p;
2081 *p = 0;
2082 *extra_chars = p + 1;
2085 return c;
2088 /* When this file is built as a standalone program, xmalloc comes from
2089 libiberty --- in which case we have to provide xfree ourselves. */
2091 void
2092 xfree (void *ptr)
2094 if (ptr != NULL)
2096 /* Literal `free' would get translated back to xfree again. */
2097 CONCAT2 (fr,ee) (ptr);
2101 /* GDB normally defines internal_error itself, but when this file is built
2102 as a standalone program, we must also provide an implementation. */
2104 void
2105 internal_error (const char *file, int line, const char *fmt, ...)
2107 va_list ap;
2109 va_start (ap, fmt);
2110 fprintf (stderr, "%s:%d: internal error: ", file, line);
2111 vfprintf (stderr, fmt, ap);
2112 exit (1);
2116 main (int argc, char **argv)
2118 char *str2, *extra_chars, c;
2119 char buf[65536];
2120 int arg;
2122 arg = 1;
2123 if (argv[arg] && strcmp (argv[arg], "--debug") == 0)
2125 yydebug = 1;
2126 arg++;
2129 if (argv[arg] == NULL)
2130 while (fgets (buf, 65536, stdin) != NULL)
2132 buf[strlen (buf) - 1] = 0;
2133 /* Use DMGL_VERBOSE to get expanded standard substitutions. */
2134 c = trim_chars (buf, &extra_chars);
2135 str2 = cplus_demangle (buf, DMGL_PARAMS | DMGL_ANSI | DMGL_VERBOSE);
2136 if (str2 == NULL)
2138 printf ("Demangling error\n");
2139 if (c)
2140 printf ("%s%c%s\n", buf, c, extra_chars);
2141 else
2142 printf ("%s\n", buf);
2143 continue;
2146 std::string errmsg;
2147 std::unique_ptr<demangle_parse_info> result
2148 = cp_demangled_name_to_comp (str2, &errmsg);
2149 if (result == NULL)
2151 fputs (errmsg.c_str (), stderr);
2152 fputc ('\n', stderr);
2153 continue;
2156 cp_print (result->tree);
2158 free (str2);
2159 if (c)
2161 putchar (c);
2162 fputs (extra_chars, stdout);
2164 putchar ('\n');
2166 else
2168 std::string errmsg;
2169 std::unique_ptr<demangle_parse_info> result
2170 = cp_demangled_name_to_comp (argv[arg], &errmsg);
2171 if (result == NULL)
2173 fputs (errmsg.c_str (), stderr);
2174 fputc ('\n', stderr);
2175 return 0;
2177 cp_print (result->tree);
2178 putchar ('\n');
2180 return 0;
2183 #endif