2 * Demangle VC++ symbols into C function prototypes
4 * Copyright 2000 Jon Griffiths
7 * This library is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * This library is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with this library; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
23 #include "wine/port.h"
30 #include "wine/debug.h"
32 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt
);
35 * - document a bit (grammar + functions)
36 * - back-port this new code into tools/winedump/msmangle.c
39 #define UNDNAME_COMPLETE (0x0000)
40 #define UNDNAME_NO_LEADING_UNDERSCORES (0x0001) /* Don't show __ in calling convention */
41 #define UNDNAME_NO_MS_KEYWORDS (0x0002) /* Don't show calling convention at all */
42 #define UNDNAME_NO_FUNCTION_RETURNS (0x0004) /* Don't show function/method return value */
43 #define UNDNAME_NO_ALLOCATION_MODEL (0x0008)
44 #define UNDNAME_NO_ALLOCATION_LANGUAGE (0x0010)
45 #define UNDNAME_NO_MS_THISTYPE (0x0020)
46 #define UNDNAME_NO_CV_THISTYPE (0x0040)
47 #define UNDNAME_NO_THISTYPE (0x0060)
48 #define UNDNAME_NO_ACCESS_SPECIFIERS (0x0080) /* Don't show access specifier (public/protected/private) */
49 #define UNDNAME_NO_THROW_SIGNATURES (0x0100)
50 #define UNDNAME_NO_MEMBER_TYPE (0x0200) /* Don't show static/virtual specifier */
51 #define UNDNAME_NO_RETURN_UDT_MODEL (0x0400)
52 #define UNDNAME_32_BIT_DECODE (0x0800)
53 #define UNDNAME_NAME_ONLY (0x1000) /* Only report the variable/method name */
54 #define UNDNAME_NO_ARGUMENTS (0x2000) /* Don't show method arguments */
55 #define UNDNAME_NO_SPECIAL_SYMS (0x4000)
56 #define UNDNAME_NO_COMPLEX_TYPE (0x8000)
58 /* How data types modifiers are stored:
59 * M (in the following definitions) is defined for
60 * 'A', 'B', 'C' and 'D' as follows
64 * {<D>}: "const volatile "
71 * same as for arguments and also the following
78 unsigned start
; /* first valid reference in array */
79 unsigned num
; /* total number of used elts */
85 /* Structure holding a parsed symbol */
88 unsigned flags
; /* the UNDNAME_ flags used for demangling */
89 malloc_func_t mem_alloc_ptr
; /* internal allocator */
90 free_func_t mem_free_ptr
; /* internal deallocator */
92 const char* current
; /* pointer in input (mangled) string */
93 char* result
; /* demangled string */
95 struct array names
; /* array of names for back reference */
96 struct array stack
; /* stack of parsed strings */
98 void* alloc_list
; /* linked list of allocated blocks */
99 unsigned avail_in_first
; /* number of available bytes in head block */
102 /* Type for parsing mangled types */
109 static BOOL
symbol_demangle(struct parsed_symbol
* sym
);
111 /******************************************************************
114 * Internal allocator. Uses a simple linked list of large blocks
115 * where we use a poor-man allocator. It's fast, and since all
116 * allocation is pool, memory management is easy (esp. freeing).
118 static void* und_alloc(struct parsed_symbol
* sym
, unsigned int len
)
122 #define BLOCK_SIZE 1024
123 #define AVAIL_SIZE (1024 - sizeof(void*))
125 if (len
> AVAIL_SIZE
)
127 /* allocate a specific block */
128 ptr
= sym
->mem_alloc_ptr(sizeof(void*) + len
);
129 if (!ptr
) return NULL
;
130 *(void**)ptr
= sym
->alloc_list
;
131 sym
->alloc_list
= ptr
;
132 sym
->avail_in_first
= 0;
133 ptr
= (char*)sym
->alloc_list
+ sizeof(void*);
137 if (len
> sym
->avail_in_first
)
139 /* add a new block */
140 ptr
= sym
->mem_alloc_ptr(BLOCK_SIZE
);
141 if (!ptr
) return NULL
;
142 *(void**)ptr
= sym
->alloc_list
;
143 sym
->alloc_list
= ptr
;
144 sym
->avail_in_first
= AVAIL_SIZE
;
146 /* grab memory from head block */
147 ptr
= (char*)sym
->alloc_list
+ BLOCK_SIZE
- sym
->avail_in_first
;
148 sym
->avail_in_first
-= len
;
155 /******************************************************************
157 * Frees all the blocks in the list of large blocks allocated by
160 static void und_free_all(struct parsed_symbol
* sym
)
164 while (sym
->alloc_list
)
166 next
= *(void**)sym
->alloc_list
;
167 if(sym
->mem_free_ptr
) sym
->mem_free_ptr(sym
->alloc_list
);
168 sym
->alloc_list
= next
;
170 sym
->avail_in_first
= 0;
173 /******************************************************************
175 * Initialises an array of strings
177 static void str_array_init(struct array
* a
)
179 a
->start
= a
->num
= a
->max
= a
->alloc
= 0;
183 /******************************************************************
185 * Adding a new string to an array
187 static BOOL
str_array_push(struct parsed_symbol
* sym
, const char* ptr
, int len
,
197 new = und_alloc(sym
, (a
->alloc
= 32) * sizeof(a
->elts
[0]));
198 if (!new) return FALSE
;
201 else if (a
->max
>= a
->alloc
)
203 new = und_alloc(sym
, (a
->alloc
* 2) * sizeof(a
->elts
[0]));
204 if (!new) return FALSE
;
205 memcpy(new, a
->elts
, a
->alloc
* sizeof(a
->elts
[0]));
209 if (len
== -1) len
= strlen(ptr
);
210 a
->elts
[a
->num
] = und_alloc(sym
, len
+ 1);
211 assert(a
->elts
[a
->num
]);
212 memcpy(a
->elts
[a
->num
], ptr
, len
);
213 a
->elts
[a
->num
][len
] = '\0';
214 if (++a
->num
>= a
->max
) a
->max
= a
->num
;
219 for (i
= a
->max
- 1; i
>= 0; i
--)
222 if (i
< a
->start
) c
= '-';
223 else if (i
>= a
->num
) c
= '}';
224 TRACE("%p\t%d%c %s\n", a
, i
, c
, a
->elts
[i
]);
231 /******************************************************************
233 * Extracts a reference from an existing array (doing proper type
236 static char* str_array_get_ref(struct array
* cref
, unsigned idx
)
239 if (cref
->start
+ idx
>= cref
->max
)
241 WARN("Out of bounds: %p %d + %d >= %d\n",
242 cref
, cref
->start
, idx
, cref
->max
);
245 TRACE("Returning %p[%d] => %s\n",
246 cref
, idx
, cref
->elts
[cref
->start
+ idx
]);
247 return cref
->elts
[cref
->start
+ idx
];
250 /******************************************************************
252 * Helper for printf type of command (only %s and %c are implemented)
253 * while dynamically allocating the buffer
255 static char* str_printf(struct parsed_symbol
* sym
, const char* format
, ...)
258 unsigned int len
= 1, i
, sz
;
263 va_start(args
, format
);
264 for (i
= 0; format
[i
]; i
++)
266 if (format
[i
] == '%')
270 case 's': t
= va_arg(args
, char*); if (t
) len
+= strlen(t
); break;
271 case 'c': (void)va_arg(args
, int); len
++; break;
272 default: i
--; /* fall through */
273 case '%': len
++; break;
279 if (!(tmp
= und_alloc(sym
, len
))) return NULL
;
280 va_start(args
, format
);
281 for (p
= tmp
, i
= 0; format
[i
]; i
++)
283 if (format
[i
] == '%')
288 t
= va_arg(args
, char*);
297 *p
++ = (char)va_arg(args
, int);
299 default: i
--; /* fall through */
300 case '%': *p
++ = '%'; break;
303 else *p
++ = format
[i
];
310 /* forward declaration */
311 static BOOL
demangle_datatype(struct parsed_symbol
* sym
, struct datatype_t
* ct
,
312 struct array
* pmt
, BOOL in_args
);
314 static const char* get_number(struct parsed_symbol
* sym
)
319 if (*sym
->current
== '?')
324 if (*sym
->current
>= '0' && *sym
->current
<= '8')
326 ptr
= und_alloc(sym
, 3);
327 if (sgn
) ptr
[0] = '-';
328 ptr
[sgn
? 1 : 0] = *sym
->current
+ 1;
329 ptr
[sgn
? 2 : 1] = '\0';
332 else if (*sym
->current
== '9')
334 ptr
= und_alloc(sym
, 4);
335 if (sgn
) ptr
[0] = '-';
336 ptr
[sgn
? 1 : 0] = '1';
337 ptr
[sgn
? 2 : 1] = '0';
338 ptr
[sgn
? 3 : 2] = '\0';
341 else if (*sym
->current
>= 'A' && *sym
->current
<= 'P')
345 while (*sym
->current
>= 'A' && *sym
->current
<= 'P')
348 ret
+= *sym
->current
++ - 'A';
350 if (*sym
->current
!= '@') return NULL
;
352 ptr
= und_alloc(sym
, 17);
353 sprintf(ptr
, "%s%d", sgn
? "-" : "", ret
);
360 /******************************************************************
362 * Parses a list of function/method arguments, creates a string corresponding
363 * to the arguments' list.
365 static char* get_args(struct parsed_symbol
* sym
, struct array
* pmt_ref
, BOOL z_term
,
366 char open_char
, char close_char
)
369 struct datatype_t ct
;
370 struct array arg_collect
;
371 char* args_str
= NULL
;
375 str_array_init(&arg_collect
);
377 /* Now come the function arguments */
378 while (*sym
->current
)
380 /* Decode each data type and append it to the argument list */
381 if (*sym
->current
== '@')
386 if (!demangle_datatype(sym
, &ct
, pmt_ref
, TRUE
))
388 /* 'void' terminates an argument list in a function */
389 if (z_term
&& !strcmp(ct
.left
, "void")) break;
390 if (!str_array_push(sym
, str_printf(sym
, "%s%s", ct
.left
, ct
.right
), -1,
393 if (!strcmp(ct
.left
, "...")) break;
395 /* Functions are always terminated by 'Z'. If we made it this far and
396 * don't find it, we have incorrectly identified a data type.
398 if (z_term
&& *sym
->current
++ != 'Z') return NULL
;
400 if (arg_collect
.num
== 0 ||
401 (arg_collect
.num
== 1 && !strcmp(arg_collect
.elts
[0], "void")))
402 return str_printf(sym
, "%cvoid%c", open_char
, close_char
);
403 for (i
= 1; i
< arg_collect
.num
; i
++)
405 args_str
= str_printf(sym
, "%s,%s", args_str
, arg_collect
.elts
[i
]);
408 last
= args_str
? args_str
: arg_collect
.elts
[0];
409 if (close_char
== '>' && last
[strlen(last
) - 1] == '>')
410 args_str
= str_printf(sym
, "%c%s%s %c",
411 open_char
, arg_collect
.elts
[0], args_str
, close_char
);
413 args_str
= str_printf(sym
, "%c%s%s%c",
414 open_char
, arg_collect
.elts
[0], args_str
, close_char
);
419 /******************************************************************
421 * Parses the type modifier. Always returns static strings.
423 static BOOL
get_modifier(struct parsed_symbol
*sym
, const char **ret
, const char **ptr_modif
)
426 if (*sym
->current
== 'E')
428 *ptr_modif
= "__ptr64";
431 switch (*sym
->current
++)
433 case 'A': *ret
= NULL
; break;
434 case 'B': *ret
= "const"; break;
435 case 'C': *ret
= "volatile"; break;
436 case 'D': *ret
= "const volatile"; break;
437 default: return FALSE
;
442 static BOOL
get_modified_type(struct datatype_t
*ct
, struct parsed_symbol
* sym
,
443 struct array
*pmt_ref
, char modif
, BOOL in_args
)
445 const char* modifier
;
446 const char* str_modif
;
447 const char *ptr_modif
= "";
449 if (*sym
->current
== 'E')
451 ptr_modif
= " __ptr64";
457 case 'A': str_modif
= str_printf(sym
, " &%s", ptr_modif
); break;
458 case 'B': str_modif
= str_printf(sym
, " &%s volatile", ptr_modif
); break;
459 case 'P': str_modif
= str_printf(sym
, " *%s", ptr_modif
); break;
460 case 'Q': str_modif
= str_printf(sym
, " *%s const", ptr_modif
); break;
461 case 'R': str_modif
= str_printf(sym
, " *%s volatile", ptr_modif
); break;
462 case 'S': str_modif
= str_printf(sym
, " *%s const volatile", ptr_modif
); break;
463 case '?': str_modif
= ""; break;
464 default: return FALSE
;
467 if (get_modifier(sym
, &modifier
, &ptr_modif
))
469 unsigned mark
= sym
->stack
.num
;
470 struct datatype_t sub_ct
;
472 /* multidimensional arrays */
473 if (*sym
->current
== 'Y')
479 if (!(n1
= get_number(sym
))) return FALSE
;
482 if (str_modif
[0] == ' ' && !modifier
)
487 str_modif
= str_printf(sym
, " (%s%s)", modifier
, str_modif
);
491 str_modif
= str_printf(sym
, " (%s)", str_modif
);
494 str_modif
= str_printf(sym
, "%s[%s]", str_modif
, get_number(sym
));
497 /* Recurse to get the referred-to type */
498 if (!demangle_datatype(sym
, &sub_ct
, pmt_ref
, FALSE
))
501 ct
->left
= str_printf(sym
, "%s %s%s", sub_ct
.left
, modifier
, str_modif
);
504 /* don't insert a space between duplicate '*' */
505 if (!in_args
&& str_modif
[0] && str_modif
[1] == '*' && sub_ct
.left
[strlen(sub_ct
.left
)-1] == '*')
507 ct
->left
= str_printf(sym
, "%s%s", sub_ct
.left
, str_modif
);
509 ct
->right
= sub_ct
.right
;
510 sym
->stack
.num
= mark
;
515 /******************************************************************
517 * Gets the literal name from the current position in the mangled
518 * symbol to the first '@' character. It pushes the parsed name to
519 * the symbol names stack and returns a pointer to it or NULL in
522 static char* get_literal_string(struct parsed_symbol
* sym
)
524 const char *ptr
= sym
->current
;
527 if (!((*sym
->current
>= 'A' && *sym
->current
<= 'Z') ||
528 (*sym
->current
>= 'a' && *sym
->current
<= 'z') ||
529 (*sym
->current
>= '0' && *sym
->current
<= '9') ||
530 *sym
->current
== '_' || *sym
->current
== '$')) {
531 TRACE("Failed at '%c' in %s\n", *sym
->current
, ptr
);
534 } while (*++sym
->current
!= '@');
536 if (!str_array_push(sym
, ptr
, sym
->current
- 1 - ptr
, &sym
->names
))
539 return str_array_get_ref(&sym
->names
, sym
->names
.num
- sym
->names
.start
- 1);
542 /******************************************************************
544 * Parses a name with a template argument list and returns it as
546 * In a template argument list the back reference to the names
547 * table is separately created. '0' points to the class component
548 * name with the template arguments. We use the same stack array
549 * to hold the names but save/restore the stack state before/after
550 * parsing the template argument list.
552 static char* get_template_name(struct parsed_symbol
* sym
)
555 unsigned num_mark
= sym
->names
.num
;
556 unsigned start_mark
= sym
->names
.start
;
557 unsigned stack_mark
= sym
->stack
.num
;
558 struct array array_pmt
;
560 sym
->names
.start
= sym
->names
.num
;
561 if (!(name
= get_literal_string(sym
)))
563 str_array_init(&array_pmt
);
564 args
= get_args(sym
, &array_pmt
, FALSE
, '<', '>');
566 name
= str_printf(sym
, "%s%s", name
, args
);
567 sym
->names
.num
= num_mark
;
568 sym
->names
.start
= start_mark
;
569 sym
->stack
.num
= stack_mark
;
573 /******************************************************************
575 * Parses class as a list of parent-classes, terminated by '@' and stores the
576 * result in 'a' array. Each parent-classes, as well as the inner element
577 * (either field/method name or class name), are represented in the mangled
578 * name by a literal name ([a-zA-Z0-9_]+ terminated by '@') or a back reference
579 * ([0-9]) or a name with template arguments ('?$' literal name followed by the
580 * template argument list). The class name components appear in the reverse
581 * order in the mangled name, e.g aaa@bbb@ccc@@ will be demangled to
583 * For each of these class name components a string will be allocated in the
586 static BOOL
get_class(struct parsed_symbol
* sym
)
588 const char* name
= NULL
;
590 while (*sym
->current
!= '@')
592 switch (*sym
->current
)
594 case '\0': return FALSE
;
596 case '0': case '1': case '2': case '3':
597 case '4': case '5': case '6': case '7':
599 name
= str_array_get_ref(&sym
->names
, *sym
->current
++ - '0');
602 switch (*++sym
->current
)
606 if ((name
= get_template_name(sym
)) &&
607 !str_array_push(sym
, name
, -1, &sym
->names
))
612 struct array stack
= sym
->stack
;
613 unsigned int start
= sym
->names
.start
;
614 unsigned int num
= sym
->names
.num
;
616 str_array_init( &sym
->stack
);
617 if (symbol_demangle( sym
)) name
= str_printf( sym
, "`%s'", sym
->result
);
618 sym
->names
.start
= start
;
619 sym
->names
.num
= num
;
624 if (!(name
= get_number( sym
))) return FALSE
;
625 name
= str_printf( sym
, "`%s'", name
);
630 name
= get_literal_string(sym
);
633 if (!name
|| !str_array_push(sym
, name
, -1, &sym
->stack
))
640 /******************************************************************
642 * From an array collected by get_class in sym->stack, constructs the
643 * corresponding (allocated) string
645 static char* get_class_string(struct parsed_symbol
* sym
, int start
)
648 unsigned int len
, sz
;
650 struct array
*a
= &sym
->stack
;
652 for (len
= 0, i
= start
; i
< a
->num
; i
++)
655 len
+= 2 + strlen(a
->elts
[i
]);
657 if (!(ret
= und_alloc(sym
, len
- 1))) return NULL
;
658 for (len
= 0, i
= a
->num
- 1; i
>= start
; i
--)
660 sz
= strlen(a
->elts
[i
]);
661 memcpy(ret
+ len
, a
->elts
[i
], sz
);
673 /******************************************************************
675 * Wrapper around get_class and get_class_string.
677 static char* get_class_name(struct parsed_symbol
* sym
)
679 unsigned mark
= sym
->stack
.num
;
683 s
= get_class_string(sym
, mark
);
684 sym
->stack
.num
= mark
;
688 /******************************************************************
689 * get_calling_convention
690 * Returns a static string corresponding to the calling convention described
691 * by char 'ch'. Sets export to TRUE iff the calling convention is exported.
693 static BOOL
get_calling_convention(char ch
, const char** call_conv
,
694 const char** exported
, unsigned flags
)
696 *call_conv
= *exported
= NULL
;
698 if (!(flags
& (UNDNAME_NO_MS_KEYWORDS
| UNDNAME_NO_ALLOCATION_LANGUAGE
)))
700 if (flags
& UNDNAME_NO_LEADING_UNDERSCORES
)
702 if (((ch
- 'A') % 2) == 1) *exported
= "dll_export ";
705 case 'A': case 'B': *call_conv
= "cdecl"; break;
706 case 'C': case 'D': *call_conv
= "pascal"; break;
707 case 'E': case 'F': *call_conv
= "thiscall"; break;
708 case 'G': case 'H': *call_conv
= "stdcall"; break;
709 case 'I': case 'J': *call_conv
= "fastcall"; break;
710 case 'K': case 'L': break;
711 case 'M': *call_conv
= "clrcall"; break;
712 default: ERR("Unknown calling convention %c\n", ch
); return FALSE
;
717 if (((ch
- 'A') % 2) == 1) *exported
= "__dll_export ";
720 case 'A': case 'B': *call_conv
= "__cdecl"; break;
721 case 'C': case 'D': *call_conv
= "__pascal"; break;
722 case 'E': case 'F': *call_conv
= "__thiscall"; break;
723 case 'G': case 'H': *call_conv
= "__stdcall"; break;
724 case 'I': case 'J': *call_conv
= "__fastcall"; break;
725 case 'K': case 'L': break;
726 case 'M': *call_conv
= "__clrcall"; break;
727 default: ERR("Unknown calling convention %c\n", ch
); return FALSE
;
734 /*******************************************************************
736 * Return a string containing an allocated string for a simple data type
738 static const char* get_simple_type(char c
)
740 const char* type_string
;
744 case 'C': type_string
= "signed char"; break;
745 case 'D': type_string
= "char"; break;
746 case 'E': type_string
= "unsigned char"; break;
747 case 'F': type_string
= "short"; break;
748 case 'G': type_string
= "unsigned short"; break;
749 case 'H': type_string
= "int"; break;
750 case 'I': type_string
= "unsigned int"; break;
751 case 'J': type_string
= "long"; break;
752 case 'K': type_string
= "unsigned long"; break;
753 case 'M': type_string
= "float"; break;
754 case 'N': type_string
= "double"; break;
755 case 'O': type_string
= "long double"; break;
756 case 'X': type_string
= "void"; break;
757 case 'Z': type_string
= "..."; break;
758 default: type_string
= NULL
; break;
763 /*******************************************************************
765 * Return a string containing an allocated string for a simple data type
767 static const char* get_extended_type(char c
)
769 const char* type_string
;
773 case 'D': type_string
= "__int8"; break;
774 case 'E': type_string
= "unsigned __int8"; break;
775 case 'F': type_string
= "__int16"; break;
776 case 'G': type_string
= "unsigned __int16"; break;
777 case 'H': type_string
= "__int32"; break;
778 case 'I': type_string
= "unsigned __int32"; break;
779 case 'J': type_string
= "__int64"; break;
780 case 'K': type_string
= "unsigned __int64"; break;
781 case 'L': type_string
= "__int128"; break;
782 case 'M': type_string
= "unsigned __int128"; break;
783 case 'N': type_string
= "bool"; break;
784 case 'W': type_string
= "wchar_t"; break;
785 default: type_string
= NULL
; break;
790 /*******************************************************************
793 * Attempt to demangle a C++ data type, which may be datatype.
794 * a datatype type is made up of a number of simple types. e.g:
795 * char** = (pointer to (pointer to (char)))
797 static BOOL
demangle_datatype(struct parsed_symbol
* sym
, struct datatype_t
* ct
,
798 struct array
* pmt_ref
, BOOL in_args
)
804 ct
->left
= ct
->right
= NULL
;
806 switch (dt
= *sym
->current
++)
809 /* MS type: __int8,__int16 etc */
810 ct
->left
= get_extended_type(*sym
->current
++);
812 case 'C': case 'D': case 'E': case 'F': case 'G':
813 case 'H': case 'I': case 'J': case 'K': case 'M':
814 case 'N': case 'O': case 'X': case 'Z':
815 /* Simple data types */
816 ct
->left
= get_simple_type(dt
);
819 case 'T': /* union */
820 case 'U': /* struct */
821 case 'V': /* class */
822 case 'Y': /* cointerface */
823 /* Class/struct/union/cointerface */
825 const char* struct_name
= NULL
;
826 const char* type_name
= NULL
;
828 if (!(struct_name
= get_class_name(sym
)))
830 if (!(sym
->flags
& UNDNAME_NO_COMPLEX_TYPE
))
834 case 'T': type_name
= "union "; break;
835 case 'U': type_name
= "struct "; break;
836 case 'V': type_name
= "class "; break;
837 case 'Y': type_name
= "cointerface "; break;
840 ct
->left
= str_printf(sym
, "%s%s", type_name
, struct_name
);
844 /* not all the time is seems */
848 if (!(ptr
= get_number(sym
))) goto done
;
849 ct
->left
= str_printf(sym
, "`template-parameter-%s'", ptr
);
853 if (!get_modified_type(ct
, sym
, pmt_ref
, '?', in_args
)) goto done
;
856 case 'A': /* reference */
857 case 'B': /* volatile reference */
858 if (!get_modified_type(ct
, sym
, pmt_ref
, dt
, in_args
)) goto done
;
860 case 'Q': /* const pointer */
861 case 'R': /* volatile pointer */
862 case 'S': /* const volatile pointer */
863 if (!get_modified_type(ct
, sym
, pmt_ref
, in_args
? dt
: 'P', in_args
)) goto done
;
865 case 'P': /* Pointer */
866 if (isdigit(*sym
->current
))
868 /* FIXME: P6 = Function pointer, others who knows.. */
869 if (*sym
->current
++ == '6')
872 const char* call_conv
;
873 const char* exported
;
874 struct datatype_t sub_ct
;
875 unsigned mark
= sym
->stack
.num
;
877 if (!get_calling_convention(*sym
->current
++,
878 &call_conv
, &exported
,
879 sym
->flags
& ~UNDNAME_NO_ALLOCATION_LANGUAGE
) ||
880 !demangle_datatype(sym
, &sub_ct
, pmt_ref
, FALSE
))
883 args
= get_args(sym
, pmt_ref
, TRUE
, '(', ')');
884 if (!args
) goto done
;
885 sym
->stack
.num
= mark
;
887 ct
->left
= str_printf(sym
, "%s%s (%s*",
888 sub_ct
.left
, sub_ct
.right
, call_conv
);
889 ct
->right
= str_printf(sym
, ")%s", args
);
893 else if (!get_modified_type(ct
, sym
, pmt_ref
, 'P', in_args
)) goto done
;
896 if (*sym
->current
== '4')
900 if (!(enum_name
= get_class_name(sym
)))
902 if (sym
->flags
& UNDNAME_NO_COMPLEX_TYPE
)
903 ct
->left
= enum_name
;
905 ct
->left
= str_printf(sym
, "enum %s", enum_name
);
909 case '0': case '1': case '2': case '3': case '4':
910 case '5': case '6': case '7': case '8': case '9':
911 /* Referring back to previously parsed type */
912 /* left and right are pushed as two separate strings */
913 ct
->left
= str_array_get_ref(pmt_ref
, (dt
- '0') * 2);
914 ct
->right
= str_array_get_ref(pmt_ref
, (dt
- '0') * 2 + 1);
915 if (!ct
->left
) goto done
;
919 switch (*sym
->current
++)
922 if (!(ct
->left
= get_number(sym
))) goto done
;
927 if (!(ptr
= get_number(sym
))) goto done
;
928 ct
->left
= str_printf(sym
, "`template-parameter%s'", ptr
);
935 if (!(p1
= get_number(sym
))) goto done
;
936 if (!(p2
= get_number(sym
))) goto done
;
937 ct
->left
= str_printf(sym
, "{%s,%s}", p1
, p2
);
945 if (!(p1
= get_number(sym
))) goto done
;
946 if (!(p2
= get_number(sym
))) goto done
;
947 if (!(p3
= get_number(sym
))) goto done
;
948 ct
->left
= str_printf(sym
, "{%s,%s,%s}", p1
, p2
, p3
);
954 if (!(ptr
= get_number(sym
))) goto done
;
955 ct
->left
= str_printf(sym
, "`non-type-template-parameter%s'", ptr
);
959 if (*sym
->current
== 'C')
961 const char *ptr
, *ptr_modif
;
964 if (!get_modifier(sym
, &ptr
, &ptr_modif
)) goto done
;
965 if (!demangle_datatype(sym
, ct
, pmt_ref
, in_args
)) goto done
;
966 ct
->left
= str_printf(sym
, "%s %s", ct
->left
, ptr
);
972 ERR("Unknown type %c\n", dt
);
975 if (add_pmt
&& pmt_ref
&& in_args
)
977 /* left and right are pushed as two separate strings */
978 if (!str_array_push(sym
, ct
->left
? ct
->left
: "", -1, pmt_ref
) ||
979 !str_array_push(sym
, ct
->right
? ct
->right
: "", -1, pmt_ref
))
984 return ct
->left
!= NULL
;
987 /******************************************************************
989 * Does the final parsing and handling for a variable or a field in
992 static BOOL
handle_data(struct parsed_symbol
* sym
)
994 const char* access
= NULL
;
995 const char* member_type
= NULL
;
996 const char* modifier
= NULL
;
997 const char* ptr_modif
;
998 struct datatype_t ct
;
1003 * 1 protected static
1005 * 3 private non-static
1006 * 4 protected non-static
1007 * 5 public non-static
1012 if (!(sym
->flags
& UNDNAME_NO_ACCESS_SPECIFIERS
))
1014 /* we only print the access for static members */
1015 switch (*sym
->current
)
1017 case '0': access
= "private: "; break;
1018 case '1': access
= "protected: "; break;
1019 case '2': access
= "public: "; break;
1023 if (!(sym
->flags
& UNDNAME_NO_MEMBER_TYPE
))
1025 if (*sym
->current
>= '0' && *sym
->current
<= '2')
1026 member_type
= "static ";
1029 name
= get_class_string(sym
, 0);
1031 switch (*sym
->current
++)
1033 case '0': case '1': case '2':
1034 case '3': case '4': case '5':
1036 unsigned mark
= sym
->stack
.num
;
1039 str_array_init(&pmt
);
1041 if (!demangle_datatype(sym
, &ct
, &pmt
, FALSE
)) goto done
;
1042 if (!get_modifier(sym
, &modifier
, &ptr_modif
)) goto done
;
1043 if (modifier
&& ptr_modif
) modifier
= str_printf(sym
, "%s %s", modifier
, ptr_modif
);
1044 else if (!modifier
) modifier
= ptr_modif
;
1045 sym
->stack
.num
= mark
;
1048 case '6' : /* compiler generated static */
1049 case '7' : /* compiler generated static */
1050 ct
.left
= ct
.right
= NULL
;
1051 if (!get_modifier(sym
, &modifier
, &ptr_modif
)) goto done
;
1052 if (*sym
->current
!= '@')
1056 if (!(cls
= get_class_name(sym
)))
1058 ct
.right
= str_printf(sym
, "{for `%s'}", cls
);
1063 modifier
= ct
.left
= ct
.right
= NULL
;
1067 if (sym
->flags
& UNDNAME_NAME_ONLY
) ct
.left
= ct
.right
= modifier
= NULL
;
1069 sym
->result
= str_printf(sym
, "%s%s%s%s%s%s%s%s", access
,
1070 member_type
, ct
.left
,
1071 modifier
&& ct
.left
? " " : NULL
, modifier
,
1072 modifier
|| ct
.left
? " " : NULL
, name
, ct
.right
);
1078 /******************************************************************
1080 * Does the final parsing and handling for a function or a method in
1083 static BOOL
handle_method(struct parsed_symbol
* sym
, BOOL cast_op
)
1086 const char* access
= NULL
;
1087 const char* member_type
= NULL
;
1088 struct datatype_t ct_ret
;
1089 const char* call_conv
;
1090 const char* modifier
= NULL
;
1091 const char* exported
;
1092 const char* args_str
= NULL
;
1093 const char* name
= NULL
;
1096 struct array array_pmt
;
1098 /* FIXME: why 2 possible letters for each option?
1101 * 'C' private: static
1102 * 'D' private: static
1103 * 'E' private: virtual
1104 * 'F' private: virtual
1105 * 'G' private: thunk
1106 * 'H' private: thunk
1109 * 'K' protected: static
1110 * 'L' protected: static
1111 * 'M' protected: virtual
1112 * 'N' protected: virtual
1113 * 'O' protected: thunk
1114 * 'P' protected: thunk
1117 * 'S' public: static
1118 * 'T' public: static
1119 * 'U' public: virtual
1120 * 'V' public: virtual
1126 accmem
= *sym
->current
++;
1127 if (accmem
< 'A' || accmem
> 'Z') goto done
;
1129 if (!(sym
->flags
& UNDNAME_NO_ACCESS_SPECIFIERS
))
1131 switch ((accmem
- 'A') / 8)
1133 case 0: access
= "private: "; break;
1134 case 1: access
= "protected: "; break;
1135 case 2: access
= "public: "; break;
1138 if (!(sym
->flags
& UNDNAME_NO_MEMBER_TYPE
))
1142 switch ((accmem
- 'A') % 8)
1144 case 2: case 3: member_type
= "static "; break;
1145 case 4: case 5: member_type
= "virtual "; break;
1147 access
= str_printf(sym
, "[thunk]:%s", access
);
1148 member_type
= "virtual ";
1154 name
= get_class_string(sym
, 0);
1156 if ((accmem
- 'A') % 8 == 6 || (accmem
- '8') % 8 == 7) /* a thunk */
1157 name
= str_printf(sym
, "%s`adjustor{%s}' ", name
, get_number(sym
));
1161 if (((accmem
- 'A') % 8) != 2 && ((accmem
- 'A') % 8) != 3)
1163 const char *ptr_modif
;
1164 /* Implicit 'this' pointer */
1165 /* If there is an implicit this pointer, const modifier follows */
1166 if (!get_modifier(sym
, &modifier
, &ptr_modif
)) goto done
;
1167 if (modifier
|| ptr_modif
) modifier
= str_printf(sym
, "%s %s", modifier
, ptr_modif
);
1171 if (!get_calling_convention(*sym
->current
++, &call_conv
, &exported
,
1175 str_array_init(&array_pmt
);
1177 /* Return type, or @ if 'void' */
1178 if (*sym
->current
== '@')
1180 ct_ret
.left
= "void";
1181 ct_ret
.right
= NULL
;
1186 if (!demangle_datatype(sym
, &ct_ret
, &array_pmt
, FALSE
))
1189 if (sym
->flags
& UNDNAME_NO_FUNCTION_RETURNS
)
1190 ct_ret
.left
= ct_ret
.right
= NULL
;
1193 name
= str_printf(sym
, "%s%s%s", name
, ct_ret
.left
, ct_ret
.right
);
1194 ct_ret
.left
= ct_ret
.right
= NULL
;
1197 mark
= sym
->stack
.num
;
1198 if (!(args_str
= get_args(sym
, &array_pmt
, TRUE
, '(', ')'))) goto done
;
1199 if (sym
->flags
& UNDNAME_NAME_ONLY
) args_str
= modifier
= NULL
;
1200 sym
->stack
.num
= mark
;
1202 /* Note: '()' after 'Z' means 'throws', but we don't care here
1205 sym
->result
= str_printf(sym
, "%s%s%s%s%s%s%s%s%s%s%s",
1206 access
, member_type
, ct_ret
.left
,
1207 (ct_ret
.left
&& !ct_ret
.right
) ? " " : NULL
,
1208 call_conv
, call_conv
? " " : NULL
, exported
,
1209 name
, args_str
, modifier
, ct_ret
.right
);
1215 /******************************************************************
1217 * Does the final parsing and handling for a name with templates
1219 static BOOL
handle_template(struct parsed_symbol
* sym
)
1224 assert(*sym
->current
== '$');
1226 if (!(name
= get_literal_string(sym
))) return FALSE
;
1227 if (!(args
= get_args(sym
, NULL
, FALSE
, '<', '>'))) return FALSE
;
1228 sym
->result
= str_printf(sym
, "%s%s", name
, args
);
1232 /*******************************************************************
1234 * Demangle a C++ linker symbol
1236 static BOOL
symbol_demangle(struct parsed_symbol
* sym
)
1239 unsigned do_after
= 0;
1240 static CHAR dashed_null
[] = "--null--";
1242 /* FIXME seems wrong as name, as it demangles a simple data type */
1243 if (sym
->flags
& UNDNAME_NO_ARGUMENTS
)
1245 struct datatype_t ct
;
1247 if (demangle_datatype(sym
, &ct
, NULL
, FALSE
))
1249 sym
->result
= str_printf(sym
, "%s%s", ct
.left
, ct
.right
);
1255 /* MS mangled names always begin with '?' */
1256 if (*sym
->current
!= '?') return FALSE
;
1259 /* Then function name or operator code */
1260 if (*sym
->current
== '?' && (sym
->current
[1] != '$' || sym
->current
[2] == '?'))
1262 const char* function_name
= NULL
;
1264 if (sym
->current
[1] == '$')
1270 /* C++ operator code (one character, or two if the first is '_') */
1271 switch (*++sym
->current
)
1273 case '0': do_after
= 1; break;
1274 case '1': do_after
= 2; break;
1275 case '2': function_name
= "operator new"; break;
1276 case '3': function_name
= "operator delete"; break;
1277 case '4': function_name
= "operator="; break;
1278 case '5': function_name
= "operator>>"; break;
1279 case '6': function_name
= "operator<<"; break;
1280 case '7': function_name
= "operator!"; break;
1281 case '8': function_name
= "operator=="; break;
1282 case '9': function_name
= "operator!="; break;
1283 case 'A': function_name
= "operator[]"; break;
1284 case 'B': function_name
= "operator "; do_after
= 3; break;
1285 case 'C': function_name
= "operator->"; break;
1286 case 'D': function_name
= "operator*"; break;
1287 case 'E': function_name
= "operator++"; break;
1288 case 'F': function_name
= "operator--"; break;
1289 case 'G': function_name
= "operator-"; break;
1290 case 'H': function_name
= "operator+"; break;
1291 case 'I': function_name
= "operator&"; break;
1292 case 'J': function_name
= "operator->*"; break;
1293 case 'K': function_name
= "operator/"; break;
1294 case 'L': function_name
= "operator%"; break;
1295 case 'M': function_name
= "operator<"; break;
1296 case 'N': function_name
= "operator<="; break;
1297 case 'O': function_name
= "operator>"; break;
1298 case 'P': function_name
= "operator>="; break;
1299 case 'Q': function_name
= "operator,"; break;
1300 case 'R': function_name
= "operator()"; break;
1301 case 'S': function_name
= "operator~"; break;
1302 case 'T': function_name
= "operator^"; break;
1303 case 'U': function_name
= "operator|"; break;
1304 case 'V': function_name
= "operator&&"; break;
1305 case 'W': function_name
= "operator||"; break;
1306 case 'X': function_name
= "operator*="; break;
1307 case 'Y': function_name
= "operator+="; break;
1308 case 'Z': function_name
= "operator-="; break;
1310 switch (*++sym
->current
)
1312 case '0': function_name
= "operator/="; break;
1313 case '1': function_name
= "operator%="; break;
1314 case '2': function_name
= "operator>>="; break;
1315 case '3': function_name
= "operator<<="; break;
1316 case '4': function_name
= "operator&="; break;
1317 case '5': function_name
= "operator|="; break;
1318 case '6': function_name
= "operator^="; break;
1319 case '7': function_name
= "`vftable'"; break;
1320 case '8': function_name
= "`vbtable'"; break;
1321 case '9': function_name
= "`vcall'"; break;
1322 case 'A': function_name
= "`typeof'"; break;
1323 case 'B': function_name
= "`local static guard'"; break;
1324 case 'C': function_name
= "`string'"; do_after
= 4; break;
1325 case 'D': function_name
= "`vbase destructor'"; break;
1326 case 'E': function_name
= "`vector deleting destructor'"; break;
1327 case 'F': function_name
= "`default constructor closure'"; break;
1328 case 'G': function_name
= "`scalar deleting destructor'"; break;
1329 case 'H': function_name
= "`vector constructor iterator'"; break;
1330 case 'I': function_name
= "`vector destructor iterator'"; break;
1331 case 'J': function_name
= "`vector vbase constructor iterator'"; break;
1332 case 'K': function_name
= "`virtual displacement map'"; break;
1333 case 'L': function_name
= "`eh vector constructor iterator'"; break;
1334 case 'M': function_name
= "`eh vector destructor iterator'"; break;
1335 case 'N': function_name
= "`eh vector vbase constructor iterator'"; break;
1336 case 'O': function_name
= "`copy constructor closure'"; break;
1338 sym
->flags
|= UNDNAME_NO_FUNCTION_RETURNS
;
1339 switch (*++sym
->current
)
1343 struct datatype_t ct
;
1347 str_array_init(&pmt
);
1348 demangle_datatype(sym
, &ct
, &pmt
, FALSE
);
1349 function_name
= str_printf(sym
, "%s%s `RTTI Type Descriptor'",
1356 const char* n1
, *n2
, *n3
, *n4
;
1358 n1
= get_number(sym
);
1359 n2
= get_number(sym
);
1360 n3
= get_number(sym
);
1361 n4
= get_number(sym
);
1363 function_name
= str_printf(sym
, "`RTTI Base Class Descriptor at (%s,%s,%s,%s)'",
1367 case '2': function_name
= "`RTTI Base Class Array'"; break;
1368 case '3': function_name
= "`RTTI Class Hierarchy Descriptor'"; break;
1369 case '4': function_name
= "`RTTI Complete Object Locator'"; break;
1371 ERR("Unknown RTTI operator: _R%c\n", *sym
->current
);
1375 case 'S': function_name
= "`local vftable'"; break;
1376 case 'T': function_name
= "`local vftable constructor closure'"; break;
1377 case 'U': function_name
= "operator new[]"; break;
1378 case 'V': function_name
= "operator delete[]"; break;
1379 case 'X': function_name
= "`placement delete closure'"; break;
1380 case 'Y': function_name
= "`placement delete[] closure'"; break;
1382 ERR("Unknown operator: _%c\n", *sym
->current
);
1387 /* FIXME: Other operators */
1388 ERR("Unknown operator: %c\n", *sym
->current
);
1395 if (!str_array_push(sym
, dashed_null
, -1, &sym
->stack
))
1399 sym
->result
= (char*)function_name
;
1405 struct array array_pmt
;
1407 str_array_init(&array_pmt
);
1408 args
= get_args(sym
, &array_pmt
, FALSE
, '<', '>');
1409 if (args
!= NULL
) function_name
= str_printf(sym
, "%s%s", function_name
, args
);
1414 if (!str_array_push(sym
, function_name
, -1, &sym
->stack
))
1419 else if (*sym
->current
== '$')
1421 /* Strange construct, it's a name with a template argument list
1424 ret
= (sym
->result
= get_template_name(sym
)) != NULL
;
1427 else if (*sym
->current
== '?' && sym
->current
[1] == '$')
1430 /* Either a class name, or '@' if the symbol is not a class member */
1431 switch (*sym
->current
)
1433 case '@': sym
->current
++; break;
1436 /* Class the function is associated with, terminated by '@@' */
1437 if (!get_class(sym
)) goto done
;
1443 case 0: default: break;
1445 /* it's time to set the member name for ctor & dtor */
1446 if (sym
->stack
.num
<= 1) goto done
;
1448 sym
->stack
.elts
[0] = sym
->stack
.elts
[1];
1450 sym
->stack
.elts
[0] = str_printf(sym
, "~%s", sym
->stack
.elts
[1]);
1451 /* ctors and dtors don't have return type */
1452 sym
->flags
|= UNDNAME_NO_FUNCTION_RETURNS
;
1455 sym
->flags
&= ~UNDNAME_NO_FUNCTION_RETURNS
;
1462 /* Function/Data type and access level */
1463 if (*sym
->current
>= '0' && *sym
->current
<= '9')
1464 ret
= handle_data(sym
);
1465 else if (*sym
->current
>= 'A' && *sym
->current
<= 'Z')
1466 ret
= handle_method(sym
, do_after
== 3);
1467 else if (*sym
->current
== '$')
1468 ret
= handle_template(sym
);
1471 if (ret
) assert(sym
->result
);
1472 else WARN("Failed at %s\n", sym
->current
);
1477 /*********************************************************************
1478 * __unDNameEx (MSVCRT.@)
1480 * Demangle a C++ identifier.
1483 * buffer [O] If not NULL, the place to put the demangled string
1484 * mangled [I] Mangled name of the function
1485 * buflen [I] Length of buffer
1486 * memget [I] Function to allocate memory with
1487 * memfree [I] Function to free memory with
1488 * unknown [?] Unknown, possibly a call back
1489 * flags [I] Flags determining demangled format
1492 * Success: A string pointing to the unmangled name, allocated with memget.
1495 char* CDECL
__unDNameEx(char* buffer
, const char* mangled
, int buflen
,
1496 malloc_func_t memget
, free_func_t memfree
,
1497 void* unknown
, unsigned short int flags
)
1499 struct parsed_symbol sym
;
1502 TRACE("(%p,%s,%d,%p,%p,%p,%x)\n",
1503 buffer
, mangled
, buflen
, memget
, memfree
, unknown
, flags
);
1505 /* The flags details is not documented by MS. However, it looks exactly
1506 * like the UNDNAME_ manifest constants from imagehlp.h and dbghelp.h
1507 * So, we copied those (on top of the file)
1509 memset(&sym
, 0, sizeof(struct parsed_symbol
));
1510 if (flags
& UNDNAME_NAME_ONLY
)
1511 flags
|= UNDNAME_NO_FUNCTION_RETURNS
| UNDNAME_NO_ACCESS_SPECIFIERS
|
1512 UNDNAME_NO_MEMBER_TYPE
| UNDNAME_NO_ALLOCATION_LANGUAGE
|
1513 UNDNAME_NO_COMPLEX_TYPE
;
1516 sym
.mem_alloc_ptr
= memget
;
1517 sym
.mem_free_ptr
= memfree
;
1518 sym
.current
= mangled
;
1519 str_array_init( &sym
.names
);
1520 str_array_init( &sym
.stack
);
1522 result
= symbol_demangle(&sym
) ? sym
.result
: mangled
;
1523 if (buffer
&& buflen
)
1525 lstrcpynA( buffer
, result
, buflen
);
1529 buffer
= memget(strlen(result
) + 1);
1530 if (buffer
) strcpy(buffer
, result
);
1539 /*********************************************************************
1540 * __unDName (MSVCRT.@)
1542 char* CDECL
__unDName(char* buffer
, const char* mangled
, int buflen
,
1543 malloc_func_t memget
, free_func_t memfree
,
1544 unsigned short int flags
)
1546 return __unDNameEx(buffer
, mangled
, buflen
, memget
, memfree
, NULL
, flags
);