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
76 #define MAX_ARRAY_ELTS 32
79 unsigned start; /* first valid reference in array */
80 unsigned num; /* total number of used elts */
82 char* elts[MAX_ARRAY_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 /******************************************************************
112 * Internal allocator. Uses a simple linked list of large blocks
113 * where we use a poor-man allocator. It's fast, and since all
114 * allocation is pool, memory management is easy (esp. freeing).
116 static void* und_alloc(struct parsed_symbol* sym, size_t len)
120 #define BLOCK_SIZE 1024
121 #define AVAIL_SIZE (1024 - sizeof(void*))
123 if (len > AVAIL_SIZE)
125 /* allocate a specific block */
126 ptr = sym->mem_alloc_ptr(sizeof(void*) + len);
127 if (!ptr) return NULL;
128 *(void**)ptr = sym->alloc_list;
129 sym->alloc_list = ptr;
130 sym->avail_in_first = 0;
131 ptr = (char*)sym->alloc_list + sizeof(void*);
135 if (len > sym->avail_in_first)
137 /* add a new block */
138 ptr = sym->mem_alloc_ptr(BLOCK_SIZE);
139 if (!ptr) return NULL;
140 *(void**)ptr = sym->alloc_list;
141 sym->alloc_list = ptr;
142 sym->avail_in_first = AVAIL_SIZE;
144 /* grab memory from head block */
145 ptr = (char*)sym->alloc_list + BLOCK_SIZE - sym->avail_in_first;
146 sym->avail_in_first -= len;
153 /******************************************************************
155 * Frees all the blocks in the list of large blocks allocated by
158 static void und_free_all(struct parsed_symbol* sym)
162 while (sym->alloc_list)
164 next = *(void**)sym->alloc_list;
165 if(sym->mem_free_ptr) sym->mem_free_ptr(sym->alloc_list);
166 sym->alloc_list = next;
168 sym->avail_in_first = 0;
171 /******************************************************************
173 * Initialises an array of strings
175 static void str_array_init(struct array* a)
177 a->start = a->num = a->max = 0;
180 /******************************************************************
182 * Adding a new string to an array
184 static void str_array_push(struct parsed_symbol* sym, const char* ptr, int len,
189 assert(a->num < MAX_ARRAY_ELTS);
190 if (len == -1) len = strlen(ptr);
191 a->elts[a->num] = und_alloc(sym, len + 1);
192 assert(a->elts[a->num]);
193 memcpy(a->elts[a->num], ptr, len);
194 a->elts[a->num][len] = '\0';
195 if (++a->num >= a->max) a->max = a->num;
200 for (i = a->max - 1; i >= 0; i--)
203 if (i < a->start) c = '-';
204 else if (i >= a->num) c = '}';
205 TRACE("%p\t%d%c %s\n", a, i, c, a->elts[i]);
210 /******************************************************************
212 * Extracts a reference from an existing array (doing proper type
215 static char* str_array_get_ref(struct array* cref, unsigned idx)
218 if (cref->start + idx >= cref->max)
220 WARN("Out of bounds: %p %d + %d >= %d\n",
221 cref, cref->start, idx, cref->max);
224 TRACE("Returning %p[%d] => %s\n",
225 cref, idx, cref->elts[cref->start + idx]);
226 return cref->elts[cref->start + idx];
229 /******************************************************************
231 * Helper for printf type of command (only %s and %c are implemented)
232 * while dynamically allocating the buffer
234 static char* str_printf(struct parsed_symbol* sym, const char* format, ...)
237 size_t len = 1, i, sz;
242 va_start(args, format);
243 for (i = 0; format[i]; i++)
245 if (format[i] == '%')
249 case 's': t = va_arg(args, char*); if (t) len += strlen(t); break;
250 case 'c': (void)va_arg(args, int); len++; break;
251 default: i--; /* fall thru */
252 case '%': len++; break;
258 if (!(tmp = und_alloc(sym, len))) return NULL;
259 va_start(args, format);
260 for (p = tmp, i = 0; format[i]; i++)
262 if (format[i] == '%')
267 t = va_arg(args, char*);
276 *p++ = (char)va_arg(args, int);
278 default: i--; /* fall thru */
279 case '%': *p++ = '%'; break;
282 else *p++ = format[i];
289 /* forward declaration */
290 static BOOL demangle_datatype(struct parsed_symbol* sym, struct datatype_t* ct,
291 struct array* pmt, BOOL in_args);
293 static const char* get_number(struct parsed_symbol* sym)
298 if (*sym->current == '?')
303 if (*sym->current >= '0' && *sym->current <= '8')
305 ptr = und_alloc(sym, 3);
306 if (sgn) ptr[0] = '-';
307 ptr[sgn ? 1 : 0] = *sym->current + 1;
308 ptr[sgn ? 2 : 1] = '\0';
311 else if (*sym->current == '9')
313 ptr = und_alloc(sym, 4);
314 if (sgn) ptr[0] = '-';
315 ptr[sgn ? 1 : 0] = '1';
316 ptr[sgn ? 2 : 1] = '0';
317 ptr[sgn ? 3 : 2] = '\0';
320 else if (*sym->current >= 'A' && *sym->current <= 'P')
324 while (*sym->current >= 'A' && *sym->current <= 'P')
327 ret += *sym->current++ - 'A';
329 if (*sym->current != '@') return NULL;
331 ptr = und_alloc(sym, 17);
332 sprintf(ptr, "%s%ld", sgn ? "-" : "", ret);
339 /******************************************************************
341 * Parses a list of function/method arguments, creates a string corresponding
342 * to the arguments' list.
344 static char* get_args(struct parsed_symbol* sym, struct array* pmt_ref, BOOL z_term,
345 char open_char, char close_char)
348 struct datatype_t ct;
349 struct array arg_collect;
350 char* args_str = NULL;
353 str_array_init(&arg_collect);
355 /* Now come the function arguments */
356 while (*sym->current)
358 /* Decode each data type and append it to the argument list */
359 if (*sym->current == '@')
364 if (!demangle_datatype(sym, &ct, pmt_ref, TRUE))
366 /* 'void' terminates an argument list in a function */
367 if (z_term && !strcmp(ct.left, "void")) break;
368 str_array_push(sym, str_printf(sym, "%s%s", ct.left, ct.right), -1,
370 if (!strcmp(ct.left, "...")) break;
372 /* Functions are always terminated by 'Z'. If we made it this far and
373 * don't find it, we have incorrectly identified a data type.
375 if (z_term && *sym->current++ != 'Z') return NULL;
377 if (arg_collect.num == 0 ||
378 (arg_collect.num == 1 && !strcmp(arg_collect.elts[0], "void")))
379 return str_printf(sym, "%cvoid%c", open_char, close_char);
380 for (i = 1; i < arg_collect.num; i++)
382 args_str = str_printf(sym, "%s,%s", args_str, arg_collect.elts[i]);
385 if (close_char == '>' && args_str && args_str[strlen(args_str) - 1] == '>')
386 args_str = str_printf(sym, "%c%s%s %c",
387 open_char, arg_collect.elts[0], args_str, close_char);
389 args_str = str_printf(sym, "%c%s%s%c",
390 open_char, arg_collect.elts[0], args_str, close_char);
395 /******************************************************************
397 * Parses the type modifier. Always returns a static string
399 static BOOL get_modifier(char ch, const char** ret)
403 case 'A': *ret = NULL; break;
404 case 'B': *ret = "const"; break;
405 case 'C': *ret = "volatile"; break;
406 case 'D': *ret = "const volatile"; break;
407 default: return FALSE;
412 static BOOL get_modified_type(struct datatype_t *ct, struct parsed_symbol* sym,
413 struct array *pmt_ref, char modif, BOOL in_args)
415 const char* modifier;
416 const char* str_modif;
420 case 'A': str_modif = " &"; break;
421 case 'B': str_modif = " & volatile"; break;
422 case 'P': str_modif = " *"; break;
423 case 'Q': str_modif = " * const"; break;
424 case 'R': str_modif = " * volatile"; break;
425 case 'S': str_modif = " * const volatile"; break;
426 case '?': str_modif = ""; break;
427 default: return FALSE;
430 if (get_modifier(*sym->current++, &modifier))
432 unsigned mark = sym->stack.num;
433 struct datatype_t sub_ct;
435 /* multidimensional arrays */
436 if (*sym->current == 'Y')
442 if (!(n1 = get_number(sym))) return FALSE;
445 if (str_modif[0] == ' ' && !modifier)
450 str_modif = str_printf(sym, " (%s%s)", modifier, str_modif);
454 str_modif = str_printf(sym, " (%s)", str_modif);
457 str_modif = str_printf(sym, "%s[%s]", str_modif, get_number(sym));
460 /* Recurse to get the referred-to type */
461 if (!demangle_datatype(sym, &sub_ct, pmt_ref, FALSE))
464 ct->left = str_printf(sym, "%s %s%s", sub_ct.left, modifier, str_modif );
467 /* don't insert a space between duplicate '*' */
468 if (!in_args && str_modif[0] && str_modif[1] == '*' && sub_ct.left[strlen(sub_ct.left)-1] == '*')
470 ct->left = str_printf(sym, "%s%s", sub_ct.left, str_modif );
472 ct->right = sub_ct.right;
473 sym->stack.num = mark;
478 /******************************************************************
480 * Gets the literal name from the current position in the mangled
481 * symbol to the first '@' character. It pushes the parsed name to
482 * the symbol names stack and returns a pointer to it or NULL in
485 static char* get_literal_string(struct parsed_symbol* sym)
487 const char *ptr = sym->current;
490 if (!((*sym->current >= 'A' && *sym->current <= 'Z') ||
491 (*sym->current >= 'a' && *sym->current <= 'z') ||
492 (*sym->current >= '0' && *sym->current <= '9') ||
493 *sym->current == '_' || *sym->current == '$')) {
494 TRACE("Failed at '%c' in %s\n", *sym->current, ptr);
497 } while (*++sym->current != '@');
499 str_array_push(sym, ptr, sym->current - 1 - ptr, &sym->names);
501 return str_array_get_ref(&sym->names, sym->names.num - sym->names.start - 1);
504 /******************************************************************
506 * Parses a name with a template argument list and returns it as
508 * In a template argument list the back reference to the names
509 * table is separately created. '0' points to the class component
510 * name with the template arguments. We use the same stack array
511 * to hold the names but save/restore the stack state before/after
512 * parsing the template argument list.
514 static char* get_template_name(struct parsed_symbol* sym)
517 unsigned num_mark = sym->names.num;
518 unsigned start_mark = sym->names.start;
519 unsigned stack_mark = sym->stack.num;
520 struct array array_pmt;
522 sym->names.start = sym->names.num;
523 if (!(name = get_literal_string(sym)))
525 str_array_init(&array_pmt);
526 args = get_args(sym, &array_pmt, FALSE, '<', '>');
528 name = str_printf(sym, "%s%s", name, args);
529 sym->names.num = num_mark;
530 sym->names.start = start_mark;
531 sym->stack.num = stack_mark;
535 /******************************************************************
537 * Parses class as a list of parent-classes, terminated by '@' and stores the
538 * result in 'a' array. Each parent-classes, as well as the inner element
539 * (either field/method name or class name), are represented in the mangled
540 * name by a literal name ([a-zA-Z0-9_]+ terminated by '@') or a back reference
541 * ([0-9]) or a name with template arguments ('?$' literal name followed by the
542 * template argument list). The class name components appear in the reverse
543 * order in the mangled name, e.g aaa@bbb@ccc@@ will be demangled to
545 * For each of these class name components a string will be allocated in the
548 static BOOL get_class(struct parsed_symbol* sym)
550 const char* name = NULL;
552 while (*sym->current != '@')
554 switch (*sym->current)
556 case '\0': return FALSE;
558 case '0': case '1': case '2': case '3':
559 case '4': case '5': case '6': case '7':
561 name = str_array_get_ref(&sym->names, *sym->current++ - '0');
564 if (*++sym->current == '$')
567 if ((name = get_template_name(sym)))
568 str_array_push(sym, name, -1, &sym->names);
572 name = get_literal_string(sym);
577 str_array_push(sym, name, -1, &sym->stack);
583 /******************************************************************
585 * From an array collected by get_class in sym->stack, constructs the
586 * corresponding (allocated) string
588 static char* get_class_string(struct parsed_symbol* sym, int start)
593 struct array *a = &sym->stack;
595 for (len = 0, i = start; i < a->num; i++)
598 len += 2 + strlen(a->elts[i]);
600 if (!(ret = und_alloc(sym, len - 1))) return NULL;
601 for (len = 0, i = a->num - 1; i >= start; i--)
603 sz = strlen(a->elts[i]);
604 memcpy(ret + len, a->elts[i], sz);
616 /******************************************************************
618 * Wrapper around get_class and get_class_string.
620 static char* get_class_name(struct parsed_symbol* sym)
622 unsigned mark = sym->stack.num;
626 s = get_class_string(sym, mark);
627 sym->stack.num = mark;
631 /******************************************************************
632 * get_calling_convention
633 * Returns a static string corresponding to the calling convention described
634 * by char 'ch'. Sets export to TRUE iff the calling convention is exported.
636 static BOOL get_calling_convention(char ch, const char** call_conv,
637 const char** exported, unsigned flags)
639 *call_conv = *exported = NULL;
641 if (!(flags & (UNDNAME_NO_MS_KEYWORDS | UNDNAME_NO_ALLOCATION_LANGUAGE)))
643 if (flags & UNDNAME_NO_LEADING_UNDERSCORES)
645 if (((ch - 'A') % 2) == 1) *exported = "dll_export ";
648 case 'A': case 'B': *call_conv = "cdecl"; break;
649 case 'C': case 'D': *call_conv = "pascal"; break;
650 case 'E': case 'F': *call_conv = "thiscall"; break;
651 case 'G': case 'H': *call_conv = "stdcall"; break;
652 case 'I': case 'J': *call_conv = "fastcall"; break;
653 case 'K': case 'L': break;
654 case 'M': *call_conv = "clrcall"; break;
655 default: ERR("Unknown calling convention %c\n", ch); return FALSE;
660 if (((ch - 'A') % 2) == 1) *exported = "__dll_export ";
663 case 'A': case 'B': *call_conv = "__cdecl"; break;
664 case 'C': case 'D': *call_conv = "__pascal"; break;
665 case 'E': case 'F': *call_conv = "__thiscall"; break;
666 case 'G': case 'H': *call_conv = "__stdcall"; break;
667 case 'I': case 'J': *call_conv = "__fastcall"; break;
668 case 'K': case 'L': break;
669 case 'M': *call_conv = "__clrcall"; break;
670 default: ERR("Unknown calling convention %c\n", ch); return FALSE;
677 /*******************************************************************
679 * Return a string containing an allocated string for a simple data type
681 static const char* get_simple_type(char c)
683 const char* type_string;
687 case 'C': type_string = "signed char"; break;
688 case 'D': type_string = "char"; break;
689 case 'E': type_string = "unsigned char"; break;
690 case 'F': type_string = "short"; break;
691 case 'G': type_string = "unsigned short"; break;
692 case 'H': type_string = "int"; break;
693 case 'I': type_string = "unsigned int"; break;
694 case 'J': type_string = "long"; break;
695 case 'K': type_string = "unsigned long"; break;
696 case 'M': type_string = "float"; break;
697 case 'N': type_string = "double"; break;
698 case 'O': type_string = "long double"; break;
699 case 'X': type_string = "void"; break;
700 case 'Z': type_string = "..."; break;
701 default: type_string = NULL; break;
706 /*******************************************************************
708 * Return a string containing an allocated string for a simple data type
710 static const char* get_extended_type(char c)
712 const char* type_string;
716 case 'D': type_string = "__int8"; break;
717 case 'E': type_string = "unsigned __int8"; break;
718 case 'F': type_string = "__int16"; break;
719 case 'G': type_string = "unsigned __int16"; break;
720 case 'H': type_string = "__int32"; break;
721 case 'I': type_string = "unsigned __int32"; break;
722 case 'J': type_string = "__int64"; break;
723 case 'K': type_string = "unsigned __int64"; break;
724 case 'L': type_string = "__int128"; break;
725 case 'M': type_string = "unsigned __int128"; break;
726 case 'N': type_string = "bool"; break;
727 case 'W': type_string = "wchar_t"; break;
728 default: type_string = NULL; break;
733 /*******************************************************************
736 * Attempt to demangle a C++ data type, which may be datatype.
737 * a datatype type is made up of a number of simple types. e.g:
738 * char** = (pointer to (pointer to (char)))
740 static BOOL demangle_datatype(struct parsed_symbol* sym, struct datatype_t* ct,
741 struct array* pmt_ref, BOOL in_args)
747 ct->left = ct->right = NULL;
749 switch (dt = *sym->current++)
752 /* MS type: __int8,__int16 etc */
753 ct->left = get_extended_type(*sym->current++);
755 case 'C': case 'D': case 'E': case 'F': case 'G':
756 case 'H': case 'I': case 'J': case 'K': case 'M':
757 case 'N': case 'O': case 'X': case 'Z':
758 /* Simple data types */
759 ct->left = get_simple_type(dt);
762 case 'T': /* union */
763 case 'U': /* struct */
764 case 'V': /* class */
765 case 'Y': /* cointerface */
766 /* Class/struct/union/cointerface */
768 const char* struct_name = NULL;
769 const char* type_name = NULL;
771 if (!(struct_name = get_class_name(sym)))
773 if (!(sym->flags & UNDNAME_NO_COMPLEX_TYPE))
777 case 'T': type_name = "union "; break;
778 case 'U': type_name = "struct "; break;
779 case 'V': type_name = "class "; break;
780 case 'Y': type_name = "cointerface "; break;
783 ct->left = str_printf(sym, "%s%s", type_name, struct_name);
787 /* not all the time is seems */
791 if (!(ptr = get_number(sym))) goto done;
792 ct->left = str_printf(sym, "`template-parameter-%s'", ptr);
796 if (!get_modified_type(ct, sym, pmt_ref, '?', in_args)) goto done;
799 case 'A': /* reference */
800 case 'B': /* volatile reference */
801 if (!get_modified_type(ct, sym, pmt_ref, dt, in_args)) goto done;
803 case 'Q': /* const pointer */
804 case 'R': /* volatile pointer */
805 case 'S': /* const volatile pointer */
806 if (!get_modified_type(ct, sym, pmt_ref, in_args ? dt : 'P', in_args)) goto done;
808 case 'P': /* Pointer */
809 if (isdigit(*sym->current))
811 /* FIXME: P6 = Function pointer, others who knows.. */
812 if (*sym->current++ == '6')
815 const char* call_conv;
816 const char* exported;
817 struct datatype_t sub_ct;
818 unsigned mark = sym->stack.num;
820 if (!get_calling_convention(*sym->current++,
821 &call_conv, &exported,
822 sym->flags & ~UNDNAME_NO_ALLOCATION_LANGUAGE) ||
823 !demangle_datatype(sym, &sub_ct, pmt_ref, FALSE))
826 args = get_args(sym, pmt_ref, TRUE, '(', ')');
827 if (!args) goto done;
828 sym->stack.num = mark;
830 ct->left = str_printf(sym, "%s%s (%s*",
831 sub_ct.left, sub_ct.right, call_conv);
832 ct->right = str_printf(sym, ")%s", args);
836 else if (!get_modified_type(ct, sym, pmt_ref, 'P', in_args)) goto done;
839 if (*sym->current == '4')
843 if (!(enum_name = get_class_name(sym)))
845 if (sym->flags & UNDNAME_NO_COMPLEX_TYPE)
846 ct->left = enum_name;
848 ct->left = str_printf(sym, "enum %s", enum_name);
852 case '0': case '1': case '2': case '3': case '4':
853 case '5': case '6': case '7': case '8': case '9':
854 /* Referring back to previously parsed type */
855 /* left and right are pushed as two separate strings */
856 ct->left = str_array_get_ref(pmt_ref, (dt - '0') * 2);
857 ct->right = str_array_get_ref(pmt_ref, (dt - '0') * 2 + 1);
858 if (!ct->left) goto done;
862 switch (*sym->current++)
865 if (!(ct->left = get_number(sym))) goto done;
870 if (!(ptr = get_number(sym))) goto done;
871 ct->left = str_printf(sym, "`template-parameter%s'", ptr);
878 if (!(p1 = get_number(sym))) goto done;
879 if (!(p2 = get_number(sym))) goto done;
880 ct->left = str_printf(sym, "{%s,%s}", p1, p2);
888 if (!(p1 = get_number(sym))) goto done;
889 if (!(p2 = get_number(sym))) goto done;
890 if (!(p3 = get_number(sym))) goto done;
891 ct->left = str_printf(sym, "{%s,%s,%s}", p1, p2, p3);
897 if (!(ptr = get_number(sym))) goto done;
898 ct->left = str_printf(sym, "`non-type-template-parameter%s'", ptr);
902 if (*sym->current == 'C')
907 if (!get_modifier(*sym->current++, &ptr)) goto done;
908 if (!demangle_datatype(sym, ct, pmt_ref, in_args)) goto done;
909 ct->left = str_printf(sym, "%s %s", ct->left, ptr);
915 ERR("Unknown type %c\n", dt);
918 if (add_pmt && pmt_ref && in_args)
920 /* left and right are pushed as two separate strings */
921 str_array_push(sym, ct->left ? ct->left : "", -1, pmt_ref);
922 str_array_push(sym, ct->right ? ct->right : "", -1, pmt_ref);
926 return ct->left != NULL;
929 /******************************************************************
931 * Does the final parsing and handling for a variable or a field in
934 static BOOL handle_data(struct parsed_symbol* sym)
936 const char* access = NULL;
937 const char* member_type = NULL;
938 const char* modifier = NULL;
939 struct datatype_t ct;
946 * 3 private non-static
947 * 4 protected non-static
948 * 5 public non-static
953 if (!(sym->flags & UNDNAME_NO_ACCESS_SPECIFIERS))
955 /* we only print the access for static members */
956 switch (*sym->current)
958 case '0': access = "private: "; break;
959 case '1': access = "protected: "; break;
960 case '2': access = "public: "; break;
964 if (!(sym->flags & UNDNAME_NO_MEMBER_TYPE))
966 if (*sym->current >= '0' && *sym->current <= '2')
967 member_type = "static ";
970 name = get_class_string(sym, 0);
972 switch (*sym->current++)
974 case '0': case '1': case '2':
975 case '3': case '4': case '5':
977 unsigned mark = sym->stack.num;
980 str_array_init(&pmt);
982 if (!demangle_datatype(sym, &ct, &pmt, FALSE)) goto done;
983 if (!get_modifier(*sym->current++, &modifier)) goto done;
984 sym->stack.num = mark;
987 case '6' : /* compiler generated static */
988 case '7' : /* compiler generated static */
989 ct.left = ct.right = NULL;
990 if (!get_modifier(*sym->current++, &modifier)) goto done;
991 if (*sym->current != '@')
995 if (!(cls = get_class_name(sym)))
997 ct.right = str_printf(sym, "{for `%s'}", cls);
1002 modifier = ct.left = ct.right = NULL;
1006 if (sym->flags & UNDNAME_NAME_ONLY) ct.left = ct.right = modifier = NULL;
1008 sym->result = str_printf(sym, "%s%s%s%s%s%s%s%s", access,
1009 member_type, ct.left,
1010 modifier && ct.left ? " " : NULL, modifier,
1011 modifier || ct.left ? " " : NULL, name, ct.right);
1017 /******************************************************************
1019 * Does the final parsing and handling for a function or a method in
1022 static BOOL handle_method(struct parsed_symbol* sym, BOOL cast_op)
1025 const char* access = NULL;
1026 const char* member_type = NULL;
1027 struct datatype_t ct_ret;
1028 const char* call_conv;
1029 const char* modifier = NULL;
1030 const char* exported;
1031 const char* args_str = NULL;
1032 const char* name = NULL;
1035 struct array array_pmt;
1037 /* FIXME: why 2 possible letters for each option?
1040 * 'C' private: static
1041 * 'D' private: static
1042 * 'E' private: virtual
1043 * 'F' private: virtual
1044 * 'G' private: thunk
1045 * 'H' private: thunk
1048 * 'K' protected: static
1049 * 'L' protected: static
1050 * 'M' protected: virtual
1051 * 'N' protected: virtual
1052 * 'O' protected: thunk
1053 * 'P' protected: thunk
1056 * 'S' public: static
1057 * 'T' public: static
1058 * 'U' public: virtual
1059 * 'V' public: virtual
1065 accmem = *sym->current++;
1066 if (accmem < 'A' || accmem > 'Z') goto done;
1068 if (!(sym->flags & UNDNAME_NO_ACCESS_SPECIFIERS))
1070 switch ((accmem - 'A') / 8)
1072 case 0: access = "private: "; break;
1073 case 1: access = "protected: "; break;
1074 case 2: access = "public: "; break;
1077 if (!(sym->flags & UNDNAME_NO_MEMBER_TYPE))
1081 switch ((accmem - 'A') % 8)
1083 case 2: case 3: member_type = "static "; break;
1084 case 4: case 5: member_type = "virtual "; break;
1086 access = str_printf(sym, "[thunk]:%s", access);
1087 member_type = "virtual ";
1093 name = get_class_string(sym, 0);
1095 if ((accmem - 'A') % 8 == 6 || (accmem - '8') % 8 == 7) /* a thunk */
1096 name = str_printf(sym, "%s`adjustor{%s}' ", name, get_number(sym));
1100 if (((accmem - 'A') % 8) != 2 && ((accmem - 'A') % 8) != 3)
1102 /* Implicit 'this' pointer */
1103 /* If there is an implicit this pointer, const modifier follows */
1104 if (!get_modifier(*sym->current, &modifier)) goto done;
1109 if (!get_calling_convention(*sym->current++, &call_conv, &exported,
1113 str_array_init(&array_pmt);
1115 /* Return type, or @ if 'void' */
1116 if (*sym->current == '@')
1118 ct_ret.left = "void";
1119 ct_ret.right = NULL;
1124 if (!demangle_datatype(sym, &ct_ret, &array_pmt, FALSE))
1127 if (sym->flags & UNDNAME_NO_FUNCTION_RETURNS)
1128 ct_ret.left = ct_ret.right = NULL;
1131 name = str_printf(sym, "%s%s%s", name, ct_ret.left, ct_ret.right);
1132 ct_ret.left = ct_ret.right = NULL;
1135 mark = sym->stack.num;
1136 if (!(args_str = get_args(sym, &array_pmt, TRUE, '(', ')'))) goto done;
1137 if (sym->flags & UNDNAME_NAME_ONLY) args_str = modifier = NULL;
1138 sym->stack.num = mark;
1140 /* Note: '()' after 'Z' means 'throws', but we don't care here
1143 sym->result = str_printf(sym, "%s%s%s%s%s%s%s%s%s%s%s%s",
1144 access, member_type, ct_ret.left,
1145 (ct_ret.left && !ct_ret.right) ? " " : NULL,
1146 call_conv, call_conv ? " " : NULL, exported,
1147 name, args_str, modifier,
1148 modifier ? " " : NULL, ct_ret.right);
1154 /******************************************************************
1156 * Does the final parsing and handling for a name with templates
1158 static BOOL handle_template(struct parsed_symbol* sym)
1163 assert(*sym->current++ == '$');
1164 if (!(name = get_literal_string(sym))) return FALSE;
1165 if (!(args = get_args(sym, NULL, FALSE, '<', '>'))) return FALSE;
1166 sym->result = str_printf(sym, "%s%s", name, args);
1170 /*******************************************************************
1172 * Demangle a C++ linker symbol
1174 static BOOL symbol_demangle(struct parsed_symbol* sym)
1177 unsigned do_after = 0;
1178 static CHAR dashed_null[] = "--null--";
1180 /* FIXME seems wrong as name, as it demangles a simple data type */
1181 if (sym->flags & UNDNAME_NO_ARGUMENTS)
1183 struct datatype_t ct;
1185 if (demangle_datatype(sym, &ct, NULL, FALSE))
1187 sym->result = str_printf(sym, "%s%s", ct.left, ct.right);
1193 /* MS mangled names always begin with '?' */
1194 if (*sym->current != '?') return FALSE;
1195 str_array_init(&sym->names);
1196 str_array_init(&sym->stack);
1199 /* Then function name or operator code */
1200 if (*sym->current == '?' && sym->current[1] != '$')
1202 const char* function_name = NULL;
1204 /* C++ operator code (one character, or two if the first is '_') */
1205 switch (*++sym->current)
1207 case '0': do_after = 1; break;
1208 case '1': do_after = 2; break;
1209 case '2': function_name = "operator new"; break;
1210 case '3': function_name = "operator delete"; break;
1211 case '4': function_name = "operator="; break;
1212 case '5': function_name = "operator>>"; break;
1213 case '6': function_name = "operator<<"; break;
1214 case '7': function_name = "operator!"; break;
1215 case '8': function_name = "operator=="; break;
1216 case '9': function_name = "operator!="; break;
1217 case 'A': function_name = "operator[]"; break;
1218 case 'B': function_name = "operator "; do_after = 3; break;
1219 case 'C': function_name = "operator->"; break;
1220 case 'D': function_name = "operator*"; break;
1221 case 'E': function_name = "operator++"; break;
1222 case 'F': function_name = "operator--"; break;
1223 case 'G': function_name = "operator-"; break;
1224 case 'H': function_name = "operator+"; break;
1225 case 'I': function_name = "operator&"; break;
1226 case 'J': function_name = "operator->*"; break;
1227 case 'K': function_name = "operator/"; break;
1228 case 'L': function_name = "operator%"; break;
1229 case 'M': function_name = "operator<"; break;
1230 case 'N': function_name = "operator<="; break;
1231 case 'O': function_name = "operator>"; break;
1232 case 'P': function_name = "operator>="; break;
1233 case 'Q': function_name = "operator,"; break;
1234 case 'R': function_name = "operator()"; break;
1235 case 'S': function_name = "operator~"; break;
1236 case 'T': function_name = "operator^"; break;
1237 case 'U': function_name = "operator|"; break;
1238 case 'V': function_name = "operator&&"; break;
1239 case 'W': function_name = "operator||"; break;
1240 case 'X': function_name = "operator*="; break;
1241 case 'Y': function_name = "operator+="; break;
1242 case 'Z': function_name = "operator-="; break;
1244 switch (*++sym->current)
1246 case '0': function_name = "operator/="; break;
1247 case '1': function_name = "operator%="; break;
1248 case '2': function_name = "operator>>="; break;
1249 case '3': function_name = "operator<<="; break;
1250 case '4': function_name = "operator&="; break;
1251 case '5': function_name = "operator|="; break;
1252 case '6': function_name = "operator^="; break;
1253 case '7': function_name = "`vftable'"; break;
1254 case '8': function_name = "`vbtable'"; break;
1255 case '9': function_name = "`vcall'"; break;
1256 case 'A': function_name = "`typeof'"; break;
1257 case 'B': function_name = "`local static guard'"; break;
1258 case 'C': function_name = "`string'"; do_after = 4; break;
1259 case 'D': function_name = "`vbase destructor'"; break;
1260 case 'E': function_name = "`vector deleting destructor'"; break;
1261 case 'F': function_name = "`default constructor closure'"; break;
1262 case 'G': function_name = "`scalar deleting destructor'"; break;
1263 case 'H': function_name = "`vector constructor iterator'"; break;
1264 case 'I': function_name = "`vector destructor iterator'"; break;
1265 case 'J': function_name = "`vector vbase constructor iterator'"; break;
1266 case 'K': function_name = "`virtual displacement map'"; break;
1267 case 'L': function_name = "`eh vector constructor iterator'"; break;
1268 case 'M': function_name = "`eh vector destructor iterator'"; break;
1269 case 'N': function_name = "`eh vector vbase constructor iterator'"; break;
1270 case 'O': function_name = "`copy constructor closure'"; break;
1272 sym->flags |= UNDNAME_NO_FUNCTION_RETURNS;
1273 switch (*++sym->current)
1277 struct datatype_t ct;
1281 str_array_init(&pmt);
1282 demangle_datatype(sym, &ct, &pmt, FALSE);
1283 function_name = str_printf(sym, "%s%s `RTTI Type Descriptor'",
1290 const char* n1, *n2, *n3, *n4;
1292 n1 = get_number(sym);
1293 n2 = get_number(sym);
1294 n3 = get_number(sym);
1295 n4 = get_number(sym);
1297 function_name = str_printf(sym, "`RTTI Base Class Descriptor at (%s,%s,%s,%s)'",
1301 case '2': function_name = "`RTTI Base Class Array'"; break;
1302 case '3': function_name = "`RTTI Class Hierarchy Descriptor'"; break;
1303 case '4': function_name = "`RTTI Complete Object Locator'"; break;
1305 ERR("Unknown RTTI operator: _R%c\n", *sym->current);
1309 case 'S': function_name = "`local vftable'"; break;
1310 case 'T': function_name = "`local vftable constructor closure'"; break;
1311 case 'U': function_name = "operator new[]"; break;
1312 case 'V': function_name = "operator delete[]"; break;
1313 case 'X': function_name = "`placement delete closure'"; break;
1314 case 'Y': function_name = "`placement delete[] closure'"; break;
1316 ERR("Unknown operator: _%c\n", *sym->current);
1321 /* FIXME: Other operators */
1322 ERR("Unknown operator: %c\n", *sym->current);
1329 sym->stack.num = sym->stack.max = 1;
1330 sym->stack.elts[0] = dashed_null;
1333 sym->result = (char*)function_name;
1337 str_array_push(sym, function_name, -1, &sym->stack);
1341 else if (*sym->current == '$')
1343 /* Strange construct, it's a name with a template argument list
1346 ret = (sym->result = get_template_name(sym)) != NULL;
1349 else if (*sym->current == '?' && sym->current[1] == '$')
1352 /* Either a class name, or '@' if the symbol is not a class member */
1353 switch (*sym->current)
1355 case '@': sym->current++; break;
1358 /* Class the function is associated with, terminated by '@@' */
1359 if (!get_class(sym)) goto done;
1365 case 0: default: break;
1367 /* it's time to set the member name for ctor & dtor */
1368 if (sym->stack.num <= 1) goto done;
1370 sym->stack.elts[0] = sym->stack.elts[1];
1372 sym->stack.elts[0] = str_printf(sym, "~%s", sym->stack.elts[1]);
1373 /* ctors and dtors don't have return type */
1374 sym->flags |= UNDNAME_NO_FUNCTION_RETURNS;
1377 sym->flags &= ~UNDNAME_NO_FUNCTION_RETURNS;
1380 sym->names.start = 1;
1384 /* Function/Data type and access level */
1385 if (*sym->current >= '0' && *sym->current <= '9')
1386 ret = handle_data(sym);
1387 else if (*sym->current >= 'A' && *sym->current <= 'Z')
1388 ret = handle_method(sym, do_after == 3);
1389 else if (*sym->current == '$')
1390 ret = handle_template(sym);
1393 if (ret) assert(sym->result);
1394 else WARN("Failed at %s\n", sym->current);
1399 /*********************************************************************
1400 * __unDNameEx (MSVCRT.@)
1402 * Demangle a C++ identifier.
1405 * buffer [O] If not NULL, the place to put the demangled string
1406 * mangled [I] Mangled name of the function
1407 * buflen [I] Length of buffer
1408 * memget [I] Function to allocate memory with
1409 * memfree [I] Function to free memory with
1410 * unknown [?] Unknown, possibly a call back
1411 * flags [I] Flags determining demangled format
1414 * Success: A string pointing to the unmangled name, allocated with memget.
1417 char* CDECL __unDNameEx(char* buffer, const char* mangled, int buflen,
1418 malloc_func_t memget, free_func_t memfree,
1419 void* unknown, unsigned short int flags)
1421 struct parsed_symbol sym;
1424 TRACE("(%p,%s,%d,%p,%p,%p,%x)\n",
1425 buffer, mangled, buflen, memget, memfree, unknown, flags);
1427 /* The flags details is not documented by MS. However, it looks exactly
1428 * like the UNDNAME_ manifest constants from imagehlp.h and dbghelp.h
1429 * So, we copied those (on top of the file)
1431 memset(&sym, 0, sizeof(struct parsed_symbol));
1432 if (flags & UNDNAME_NAME_ONLY)
1433 flags |= UNDNAME_NO_FUNCTION_RETURNS | UNDNAME_NO_ACCESS_SPECIFIERS |
1434 UNDNAME_NO_MEMBER_TYPE | UNDNAME_NO_ALLOCATION_LANGUAGE |
1435 UNDNAME_NO_COMPLEX_TYPE;
1438 sym.mem_alloc_ptr = memget;
1439 sym.mem_free_ptr = memfree;
1440 sym.current = mangled;
1442 result = symbol_demangle(&sym) ? sym.result : mangled;
1443 if (buffer && buflen)
1445 lstrcpynA( buffer, result, buflen);
1449 buffer = memget(strlen(result) + 1);
1450 if (buffer) strcpy(buffer, result);
1459 /*********************************************************************
1460 * __unDName (MSVCRT.@)
1462 char* CDECL __unDName(char* buffer, const char* mangled, int buflen,
1463 malloc_func_t memget, free_func_t memfree,
1464 unsigned short int flags)
1466 return __unDNameEx(buffer, mangled, buflen, memget, memfree, NULL, flags);