push 014043c4937c940c54cd1214c96e33a3b3c8cf7d
[wine/hacks.git] / dlls / msvcrt / undname.c
blobdfc2241409c51540c677285b3ddabb9b9d778368
1 /*
2 * Demangle VC++ symbols into C function prototypes
4 * Copyright 2000 Jon Griffiths
5 * 2004 Eric Pouech
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
22 #include "config.h"
23 #include "wine/port.h"
25 #include <assert.h>
26 #include <stdio.h>
27 #include "msvcrt.h"
29 #include "wine/debug.h"
31 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt);
33 /* TODO:
34 * - document a bit (grammar + functions)
35 * - back-port this new code into tools/winedump/msmangle.c
38 #define UNDNAME_COMPLETE (0x0000)
39 #define UNDNAME_NO_LEADING_UNDERSCORES (0x0001) /* Don't show __ in calling convention */
40 #define UNDNAME_NO_MS_KEYWORDS (0x0002) /* Don't show calling convention at all */
41 #define UNDNAME_NO_FUNCTION_RETURNS (0x0004) /* Don't show function/method return value */
42 #define UNDNAME_NO_ALLOCATION_MODEL (0x0008)
43 #define UNDNAME_NO_ALLOCATION_LANGUAGE (0x0010)
44 #define UNDNAME_NO_MS_THISTYPE (0x0020)
45 #define UNDNAME_NO_CV_THISTYPE (0x0040)
46 #define UNDNAME_NO_THISTYPE (0x0060)
47 #define UNDNAME_NO_ACCESS_SPECIFIERS (0x0080) /* Don't show access specifier (public/protected/private) */
48 #define UNDNAME_NO_THROW_SIGNATURES (0x0100)
49 #define UNDNAME_NO_MEMBER_TYPE (0x0200) /* Don't show static/virtual specifier */
50 #define UNDNAME_NO_RETURN_UDT_MODEL (0x0400)
51 #define UNDNAME_32_BIT_DECODE (0x0800)
52 #define UNDNAME_NAME_ONLY (0x1000) /* Only report the variable/method name */
53 #define UNDNAME_NO_ARGUMENTS (0x2000) /* Don't show method arguments */
54 #define UNDNAME_NO_SPECIAL_SYMS (0x4000)
55 #define UNDNAME_NO_COMPLEX_TYPE (0x8000)
57 /* How data types modifiers are stored:
58 * M (in the following definitions) is defined for
59 * 'A', 'B', 'C' and 'D' as follows
60 * {<A>}: ""
61 * {<B>}: "const "
62 * {<C>}: "volatile "
63 * {<D>}: "const volatile "
65 * in arguments:
66 * P<M>x {<M>}x*
67 * Q<M>x {<M>}x* const
68 * A<M>x {<M>}x&
69 * in data fields:
70 * same as for arguments and also the following
71 * ?<M>x {<M>}x
75 #define MAX_ARRAY_ELTS 32
76 struct array
78 unsigned start; /* first valid reference in array */
79 unsigned num; /* total number of used elts */
80 unsigned max;
81 char* elts[MAX_ARRAY_ELTS];
84 /* Structure holding a parsed symbol */
85 struct parsed_symbol
87 unsigned flags; /* the UNDNAME_ flags used for demangling */
88 malloc_func_t mem_alloc_ptr; /* internal allocator */
89 free_func_t mem_free_ptr; /* internal deallocator */
91 const char* current; /* pointer in input (mangled) string */
92 char* result; /* demangled string */
94 struct array names; /* array of names for back reference */
95 struct array stack; /* stack of parsed strings */
97 void* alloc_list; /* linked list of allocated blocks */
98 unsigned avail_in_first; /* number of available bytes in head block */
101 /* Type for parsing mangled types */
102 struct datatype_t
104 const char* left;
105 const char* right;
108 /******************************************************************
109 * und_alloc
111 * Internal allocator. Uses a simple linked list of large blocks
112 * where we use a poor-man allocator. It's fast, and since all
113 * allocation is pool, memory management is easy (esp. freeing).
115 static void* und_alloc(struct parsed_symbol* sym, size_t len)
117 void* ptr;
119 #define BLOCK_SIZE 1024
120 #define AVAIL_SIZE (1024 - sizeof(void*))
122 if (len > AVAIL_SIZE)
124 /* allocate a specific block */
125 ptr = sym->mem_alloc_ptr(sizeof(void*) + len);
126 if (!ptr) return NULL;
127 *(void**)ptr = sym->alloc_list;
128 sym->alloc_list = ptr;
129 sym->avail_in_first = 0;
130 ptr = (char*)sym->alloc_list + sizeof(void*);
132 else
134 if (len > sym->avail_in_first)
136 /* add a new block */
137 ptr = sym->mem_alloc_ptr(BLOCK_SIZE);
138 if (!ptr) return NULL;
139 *(void**)ptr = sym->alloc_list;
140 sym->alloc_list = ptr;
141 sym->avail_in_first = AVAIL_SIZE;
143 /* grab memory from head block */
144 ptr = (char*)sym->alloc_list + BLOCK_SIZE - sym->avail_in_first;
145 sym->avail_in_first -= len;
147 return ptr;
148 #undef BLOCK_SIZE
149 #undef AVAIL_SIZE
152 /******************************************************************
153 * und_free
154 * Frees all the blocks in the list of large blocks allocated by
155 * und_alloc.
157 static void und_free_all(struct parsed_symbol* sym)
159 void* next;
161 while (sym->alloc_list)
163 next = *(void**)sym->alloc_list;
164 if(sym->mem_free_ptr) sym->mem_free_ptr(sym->alloc_list);
165 sym->alloc_list = next;
167 sym->avail_in_first = 0;
170 /******************************************************************
171 * str_array_init
172 * Initialises an array of strings
174 static void str_array_init(struct array* a)
176 a->start = a->num = a->max = 0;
179 /******************************************************************
180 * str_array_push
181 * Adding a new string to an array
183 static void str_array_push(struct parsed_symbol* sym, const char* ptr, int len,
184 struct array* a)
186 assert(ptr);
187 assert(a);
188 assert(a->num < MAX_ARRAY_ELTS);
189 if (len == -1) len = strlen(ptr);
190 a->elts[a->num] = und_alloc(sym, len + 1);
191 assert(a->elts[a->num]);
192 memcpy(a->elts[a->num], ptr, len);
193 a->elts[a->num][len] = '\0';
194 if (++a->num >= a->max) a->max = a->num;
196 int i;
197 char c;
199 for (i = a->max - 1; i >= 0; i--)
201 c = '>';
202 if (i < a->start) c = '-';
203 else if (i >= a->num) c = '}';
204 TRACE("%p\t%d%c %s\n", a, i, c, a->elts[i]);
209 /******************************************************************
210 * str_array_get_ref
211 * Extracts a reference from an existing array (doing proper type
212 * checking)
214 static char* str_array_get_ref(struct array* cref, unsigned idx)
216 assert(cref);
217 if (cref->start + idx >= cref->max)
219 WARN("Out of bounds: %p %d + %d >= %d\n",
220 cref, cref->start, idx, cref->max);
221 return NULL;
223 TRACE("Returning %p[%d] => %s\n",
224 cref, idx, cref->elts[cref->start + idx]);
225 return cref->elts[cref->start + idx];
228 /******************************************************************
229 * str_printf
230 * Helper for printf type of command (only %s and %c are implemented)
231 * while dynamically allocating the buffer
233 static char* str_printf(struct parsed_symbol* sym, const char* format, ...)
235 va_list args;
236 size_t len = 1, i, sz;
237 char* tmp;
238 char* p;
239 char* t;
241 va_start(args, format);
242 for (i = 0; format[i]; i++)
244 if (format[i] == '%')
246 switch (format[++i])
248 case 's': t = va_arg(args, char*); if (t) len += strlen(t); break;
249 case 'c': (void)va_arg(args, int); len++; break;
250 default: i--; /* fall thru */
251 case '%': len++; break;
254 else len++;
256 va_end(args);
257 if (!(tmp = und_alloc(sym, len))) return NULL;
258 va_start(args, format);
259 for (p = tmp, i = 0; format[i]; i++)
261 if (format[i] == '%')
263 switch (format[++i])
265 case 's':
266 t = va_arg(args, char*);
267 if (t)
269 sz = strlen(t);
270 memcpy(p, t, sz);
271 p += sz;
273 break;
274 case 'c':
275 *p++ = (char)va_arg(args, int);
276 break;
277 default: i--; /* fall thru */
278 case '%': *p++ = '%'; break;
281 else *p++ = format[i];
283 va_end(args);
284 *p = '\0';
285 return tmp;
288 /* forward declaration */
289 static BOOL demangle_datatype(struct parsed_symbol* sym, struct datatype_t* ct,
290 struct array* pmt, BOOL in_args);
292 static const char* get_number(struct parsed_symbol* sym)
294 char* ptr;
295 BOOL sgn = FALSE;
297 if (*sym->current == '?')
299 sgn = TRUE;
300 sym->current++;
302 if (*sym->current >= '0' && *sym->current <= '8')
304 ptr = und_alloc(sym, 3);
305 if (sgn) ptr[0] = '-';
306 ptr[sgn ? 1 : 0] = *sym->current + 1;
307 ptr[sgn ? 2 : 1] = '\0';
308 sym->current++;
310 else if (*sym->current == '9')
312 ptr = und_alloc(sym, 4);
313 if (sgn) ptr[0] = '-';
314 ptr[sgn ? 1 : 0] = '1';
315 ptr[sgn ? 2 : 1] = '0';
316 ptr[sgn ? 3 : 2] = '\0';
317 sym->current++;
319 else if (*sym->current >= 'A' && *sym->current <= 'P')
321 long ret = 0;
323 while (*sym->current >= 'A' && *sym->current <= 'P')
325 ret *= 16;
326 ret += *sym->current++ - 'A';
328 if (*sym->current != '@') return NULL;
330 ptr = und_alloc(sym, 17);
331 sprintf(ptr, "%s%ld", sgn ? "-" : "", ret);
332 sym->current++;
334 else return NULL;
335 return ptr;
338 /******************************************************************
339 * get_args
340 * Parses a list of function/method arguments, creates a string corresponding
341 * to the arguments' list.
343 static char* get_args(struct parsed_symbol* sym, struct array* pmt_ref, BOOL z_term,
344 char open_char, char close_char)
347 struct datatype_t ct;
348 struct array arg_collect;
349 char* args_str = NULL;
350 int i;
352 str_array_init(&arg_collect);
354 /* Now come the function arguments */
355 while (*sym->current)
357 /* Decode each data type and append it to the argument list */
358 if (*sym->current == '@')
360 sym->current++;
361 break;
363 if (!demangle_datatype(sym, &ct, pmt_ref, TRUE))
364 return NULL;
365 /* 'void' terminates an argument list in a function */
366 if (z_term && !strcmp(ct.left, "void")) break;
367 str_array_push(sym, str_printf(sym, "%s%s", ct.left, ct.right), -1,
368 &arg_collect);
369 if (!strcmp(ct.left, "...")) break;
371 /* Functions are always terminated by 'Z'. If we made it this far and
372 * don't find it, we have incorrectly identified a data type.
374 if (z_term && *sym->current++ != 'Z') return NULL;
376 if (arg_collect.num == 0 ||
377 (arg_collect.num == 1 && !strcmp(arg_collect.elts[0], "void")))
378 return str_printf(sym, "%cvoid%c", open_char, close_char);
379 for (i = 1; i < arg_collect.num; i++)
381 args_str = str_printf(sym, "%s,%s", args_str, arg_collect.elts[i]);
384 if (close_char == '>' && args_str && args_str[strlen(args_str) - 1] == '>')
385 args_str = str_printf(sym, "%c%s%s %c",
386 open_char, arg_collect.elts[0], args_str, close_char);
387 else
388 args_str = str_printf(sym, "%c%s%s%c",
389 open_char, arg_collect.elts[0], args_str, close_char);
391 return args_str;
394 /******************************************************************
395 * get_modifier
396 * Parses the type modifier. Always returns a static string
398 static BOOL get_modifier(char ch, const char** ret)
400 switch (ch)
402 case 'A': *ret = NULL; break;
403 case 'B': *ret = "const"; break;
404 case 'C': *ret = "volatile"; break;
405 case 'D': *ret = "const volatile"; break;
406 default: return FALSE;
408 return TRUE;
411 static BOOL get_modified_type(struct datatype_t *ct, struct parsed_symbol* sym,
412 struct array *pmt_ref, char modif)
414 const char* modifier;
415 const char* str_modif;
417 switch (modif)
419 case 'A': str_modif = " &"; break;
420 case 'B': str_modif = " & volatile"; break;
421 case 'P': str_modif = " *"; break;
422 case 'Q': str_modif = " * const"; break;
423 case 'R': str_modif = " * volatile"; break;
424 case 'S': str_modif = " * const volatile"; break;
425 case '?': str_modif = ""; break;
426 default: return FALSE;
429 if (get_modifier(*sym->current++, &modifier))
431 unsigned mark = sym->stack.num;
432 struct datatype_t sub_ct;
434 /* Recurse to get the referred-to type */
435 if (!demangle_datatype(sym, &sub_ct, pmt_ref, FALSE))
436 return FALSE;
437 if (modifier)
438 ct->left = str_printf(sym, "%s %s%s", sub_ct.left, modifier, str_modif );
439 else
441 /* don't insert a space between duplicate '*' */
442 if (str_modif[0] && str_modif[1] == '*' && sub_ct.left[strlen(sub_ct.left)-1] == '*')
443 str_modif++;
444 ct->left = str_printf(sym, "%s%s", sub_ct.left, str_modif );
446 ct->right = sub_ct.right;
447 sym->stack.num = mark;
449 return TRUE;
452 /******************************************************************
453 * get_literal_string
454 * Gets the literal name from the current position in the mangled
455 * symbol to the first '@' character. It pushes the parsed name to
456 * the symbol names stack and returns a pointer to it or NULL in
457 * case of an error.
459 static char* get_literal_string(struct parsed_symbol* sym)
461 const char *ptr = sym->current;
463 do {
464 if (!((*sym->current >= 'A' && *sym->current <= 'Z') ||
465 (*sym->current >= 'a' && *sym->current <= 'z') ||
466 (*sym->current >= '0' && *sym->current <= '9') ||
467 *sym->current == '_' || *sym->current == '$')) {
468 TRACE("Failed at '%c' in %s\n", *sym->current, ptr);
469 return NULL;
471 } while (*++sym->current != '@');
472 sym->current++;
473 str_array_push(sym, ptr, sym->current - 1 - ptr, &sym->names);
475 return str_array_get_ref(&sym->names, sym->names.num - sym->names.start - 1);
478 /******************************************************************
479 * get_template_name
480 * Parses a name with a template argument list and returns it as
481 * a string.
482 * In a template argument list the back reference to the names
483 * table is separately created. '0' points to the class component
484 * name with the template arguments. We use the same stack array
485 * to hold the names but save/restore the stack state before/after
486 * parsing the template argument list.
488 static char* get_template_name(struct parsed_symbol* sym)
490 char *name, *args;
491 unsigned num_mark = sym->names.num;
492 unsigned start_mark = sym->names.start;
493 unsigned stack_mark = sym->stack.num;
494 struct array array_pmt;
496 sym->names.start = sym->names.num;
497 if (!(name = get_literal_string(sym)))
498 return FALSE;
499 str_array_init(&array_pmt);
500 args = get_args(sym, &array_pmt, FALSE, '<', '>');
501 if (args != NULL)
502 name = str_printf(sym, "%s%s", name, args);
503 sym->names.num = num_mark;
504 sym->names.start = start_mark;
505 sym->stack.num = stack_mark;
506 return name;
509 /******************************************************************
510 * get_class
511 * Parses class as a list of parent-classes, terminated by '@' and stores the
512 * result in 'a' array. Each parent-classes, as well as the inner element
513 * (either field/method name or class name), are represented in the mangled
514 * name by a literal name ([a-zA-Z0-9_]+ terminated by '@') or a back reference
515 * ([0-9]) or a name with template arguments ('?$' literal name followed by the
516 * template argument list). The class name components appear in the reverse
517 * order in the mangled name, e.g aaa@bbb@ccc@@ will be demangled to
518 * ccc::bbb::aaa
519 * For each of this class name componets a string will be allocated in the
520 * array.
522 static BOOL get_class(struct parsed_symbol* sym)
524 const char* name = NULL;
526 while (*sym->current != '@')
528 switch (*sym->current)
530 case '\0': return FALSE;
532 case '0': case '1': case '2': case '3':
533 case '4': case '5': case '6': case '7':
534 case '8': case '9':
535 name = str_array_get_ref(&sym->names, *sym->current++ - '0');
536 break;
537 case '?':
538 if (*++sym->current == '$')
540 sym->current++;
541 if ((name = get_template_name(sym)))
542 str_array_push(sym, name, -1, &sym->names);
544 break;
545 default:
546 name = get_literal_string(sym);
547 break;
549 if (!name)
550 return FALSE;
551 str_array_push(sym, name, -1, &sym->stack);
553 sym->current++;
554 return TRUE;
557 /******************************************************************
558 * get_class_string
559 * From an array collected by get_class in sym->stack, constructs the
560 * corresponding (allocated) string
562 static char* get_class_string(struct parsed_symbol* sym, int start)
564 int i;
565 size_t len, sz;
566 char* ret;
567 struct array *a = &sym->stack;
569 for (len = 0, i = start; i < a->num; i++)
571 assert(a->elts[i]);
572 len += 2 + strlen(a->elts[i]);
574 if (!(ret = und_alloc(sym, len - 1))) return NULL;
575 for (len = 0, i = a->num - 1; i >= start; i--)
577 sz = strlen(a->elts[i]);
578 memcpy(ret + len, a->elts[i], sz);
579 len += sz;
580 if (i > start)
582 ret[len++] = ':';
583 ret[len++] = ':';
586 ret[len] = '\0';
587 return ret;
590 /******************************************************************
591 * get_class_name
592 * Wrapper around get_class and get_class_string.
594 static char* get_class_name(struct parsed_symbol* sym)
596 unsigned mark = sym->stack.num;
597 char* s = NULL;
599 if (get_class(sym))
600 s = get_class_string(sym, mark);
601 sym->stack.num = mark;
602 return s;
605 /******************************************************************
606 * get_calling_convention
607 * Returns a static string corresponding to the calling convention described
608 * by char 'ch'. Sets export to TRUE iff the calling convention is exported.
610 static BOOL get_calling_convention(char ch, const char** call_conv,
611 const char** exported, unsigned flags)
613 *call_conv = *exported = NULL;
615 if (!(flags & (UNDNAME_NO_MS_KEYWORDS | UNDNAME_NO_ALLOCATION_LANGUAGE)))
617 if (flags & UNDNAME_NO_LEADING_UNDERSCORES)
619 if (((ch - 'A') % 2) == 1) *exported = "dll_export ";
620 switch (ch)
622 case 'A': case 'B': *call_conv = "cdecl"; break;
623 case 'C': case 'D': *call_conv = "pascal"; break;
624 case 'E': case 'F': *call_conv = "thiscall"; break;
625 case 'G': case 'H': *call_conv = "stdcall"; break;
626 case 'I': case 'J': *call_conv = "fastcall"; break;
627 case 'K': case 'L': break;
628 case 'M': *call_conv = "clrcall"; break;
629 default: ERR("Unknown calling convention %c\n", ch); return FALSE;
632 else
634 if (((ch - 'A') % 2) == 1) *exported = "__dll_export ";
635 switch (ch)
637 case 'A': case 'B': *call_conv = "__cdecl"; break;
638 case 'C': case 'D': *call_conv = "__pascal"; break;
639 case 'E': case 'F': *call_conv = "__thiscall"; break;
640 case 'G': case 'H': *call_conv = "__stdcall"; break;
641 case 'I': case 'J': *call_conv = "__fastcall"; break;
642 case 'K': case 'L': break;
643 case 'M': *call_conv = "__clrcall"; break;
644 default: ERR("Unknown calling convention %c\n", ch); return FALSE;
648 return TRUE;
651 /*******************************************************************
652 * get_simple_type
653 * Return a string containing an allocated string for a simple data type
655 static const char* get_simple_type(char c)
657 const char* type_string;
659 switch (c)
661 case 'C': type_string = "signed char"; break;
662 case 'D': type_string = "char"; break;
663 case 'E': type_string = "unsigned char"; break;
664 case 'F': type_string = "short"; break;
665 case 'G': type_string = "unsigned short"; break;
666 case 'H': type_string = "int"; break;
667 case 'I': type_string = "unsigned int"; break;
668 case 'J': type_string = "long"; break;
669 case 'K': type_string = "unsigned long"; break;
670 case 'M': type_string = "float"; break;
671 case 'N': type_string = "double"; break;
672 case 'O': type_string = "long double"; break;
673 case 'X': type_string = "void"; break;
674 case 'Z': type_string = "..."; break;
675 default: type_string = NULL; break;
677 return type_string;
680 /*******************************************************************
681 * get_extented_type
682 * Return a string containing an allocated string for a simple data type
684 static const char* get_extended_type(char c)
686 const char* type_string;
688 switch (c)
690 case 'D': type_string = "__int8"; break;
691 case 'E': type_string = "unsigned __int8"; break;
692 case 'F': type_string = "__int16"; break;
693 case 'G': type_string = "unsigned __int16"; break;
694 case 'H': type_string = "__int32"; break;
695 case 'I': type_string = "unsigned __int32"; break;
696 case 'J': type_string = "__int64"; break;
697 case 'K': type_string = "unsigned __int64"; break;
698 case 'L': type_string = "__int128"; break;
699 case 'M': type_string = "unsigned __int128"; break;
700 case 'N': type_string = "bool"; break;
701 case 'W': type_string = "wchar_t"; break;
702 default: type_string = NULL; break;
704 return type_string;
707 /*******************************************************************
708 * demangle_datatype
710 * Attempt to demangle a C++ data type, which may be datatype.
711 * a datatype type is made up of a number of simple types. e.g:
712 * char** = (pointer to (pointer to (char)))
714 static BOOL demangle_datatype(struct parsed_symbol* sym, struct datatype_t* ct,
715 struct array* pmt_ref, BOOL in_args)
717 char dt;
718 BOOL add_pmt = TRUE;
720 assert(ct);
721 ct->left = ct->right = NULL;
723 switch (dt = *sym->current++)
725 case '_':
726 /* MS type: __int8,__int16 etc */
727 ct->left = get_extended_type(*sym->current++);
728 break;
729 case 'C': case 'D': case 'E': case 'F': case 'G':
730 case 'H': case 'I': case 'J': case 'K': case 'M':
731 case 'N': case 'O': case 'X': case 'Z':
732 /* Simple data types */
733 ct->left = get_simple_type(dt);
734 add_pmt = FALSE;
735 break;
736 case 'T': /* union */
737 case 'U': /* struct */
738 case 'V': /* class */
739 case 'Y': /* cointerface */
740 /* Class/struct/union/cointerface */
742 const char* struct_name = NULL;
743 const char* type_name = NULL;
745 if (!(struct_name = get_class_name(sym)))
746 goto done;
747 if (!(sym->flags & UNDNAME_NO_COMPLEX_TYPE))
749 switch (dt)
751 case 'T': type_name = "union "; break;
752 case 'U': type_name = "struct "; break;
753 case 'V': type_name = "class "; break;
754 case 'Y': type_name = "cointerface "; break;
757 ct->left = str_printf(sym, "%s%s", type_name, struct_name);
759 break;
760 case '?':
761 /* not all the time is seems */
762 if (!get_modified_type(ct, sym, pmt_ref, '?')) goto done;
763 break;
764 case 'A': /* reference */
765 case 'B': /* volatile reference */
766 if (!get_modified_type(ct, sym, pmt_ref, dt)) goto done;
767 break;
768 case 'Q': /* const pointer */
769 case 'R': /* volatile pointer */
770 case 'S': /* const volatile pointer */
771 if (!get_modified_type(ct, sym, pmt_ref, in_args ? dt : 'P')) goto done;
772 break;
773 case 'P': /* Pointer */
774 if (isdigit(*sym->current))
776 /* FIXME: P6 = Function pointer, others who knows.. */
777 if (*sym->current++ == '6')
779 char* args = NULL;
780 const char* call_conv;
781 const char* exported;
782 struct datatype_t sub_ct;
783 unsigned mark = sym->stack.num;
785 if (!get_calling_convention(*sym->current++,
786 &call_conv, &exported,
787 sym->flags & ~UNDNAME_NO_ALLOCATION_LANGUAGE) ||
788 !demangle_datatype(sym, &sub_ct, pmt_ref, FALSE))
789 goto done;
791 args = get_args(sym, pmt_ref, TRUE, '(', ')');
792 if (!args) goto done;
793 sym->stack.num = mark;
795 ct->left = str_printf(sym, "%s%s (%s*",
796 sub_ct.left, sub_ct.right, call_conv);
797 ct->right = str_printf(sym, ")%s", args);
799 else goto done;
801 else if (!get_modified_type(ct, sym, pmt_ref, 'P')) goto done;
802 break;
803 case 'W':
804 if (*sym->current == '4')
806 char* enum_name;
807 sym->current++;
808 if (!(enum_name = get_class_name(sym)))
809 goto done;
810 if (sym->flags & UNDNAME_NO_COMPLEX_TYPE)
811 ct->left = enum_name;
812 else
813 ct->left = str_printf(sym, "enum %s", enum_name);
815 else goto done;
816 break;
817 case '0': case '1': case '2': case '3': case '4':
818 case '5': case '6': case '7': case '8': case '9':
819 /* Referring back to previously parsed type */
820 /* left and right are pushed as two separate strings */
821 ct->left = str_array_get_ref(pmt_ref, (dt - '0') * 2);
822 ct->right = str_array_get_ref(pmt_ref, (dt - '0') * 2 + 1);
823 if (!ct->left) goto done;
824 add_pmt = FALSE;
825 break;
826 case '$':
827 switch (*sym->current++)
829 case '0':
830 if (!(ct->left = get_number(sym))) goto done;
831 break;
832 case 'D':
834 const char* ptr;
835 if (!(ptr = get_number(sym))) goto done;
836 ct->left = str_printf(sym, "`template-parameter%s'", ptr);
838 break;
839 case 'F':
841 const char* p1;
842 const char* p2;
843 if (!(p1 = get_number(sym))) goto done;
844 if (!(p2 = get_number(sym))) goto done;
845 ct->left = str_printf(sym, "{%s,%s}", p1, p2);
847 break;
848 case 'G':
850 const char* p1;
851 const char* p2;
852 const char* p3;
853 if (!(p1 = get_number(sym))) goto done;
854 if (!(p2 = get_number(sym))) goto done;
855 if (!(p3 = get_number(sym))) goto done;
856 ct->left = str_printf(sym, "{%s,%s,%s}", p1, p2, p3);
858 break;
859 case 'Q':
861 const char* ptr;
862 if (!(ptr = get_number(sym))) goto done;
863 ct->left = str_printf(sym, "`non-type-template-parameter%s'", ptr);
865 break;
867 break;
868 default :
869 ERR("Unknown type %c\n", dt);
870 break;
872 if (add_pmt && pmt_ref && in_args)
874 /* left and right are pushed as two separate strings */
875 str_array_push(sym, ct->left ? ct->left : "", -1, pmt_ref);
876 str_array_push(sym, ct->right ? ct->right : "", -1, pmt_ref);
878 done:
880 return ct->left != NULL;
883 /******************************************************************
884 * handle_data
885 * Does the final parsing and handling for a variable or a field in
886 * a class.
888 static BOOL handle_data(struct parsed_symbol* sym)
890 const char* access = NULL;
891 const char* member_type = NULL;
892 const char* modifier = NULL;
893 struct datatype_t ct;
894 char* name = NULL;
895 BOOL ret = FALSE;
896 char dt;
898 /* 0 private static
899 * 1 protected static
900 * 2 public static
901 * 3 private non-static
902 * 4 protected non-static
903 * 5 public non-static
904 * 6 ?? static
905 * 7 ?? static
908 if (!(sym->flags & UNDNAME_NO_ACCESS_SPECIFIERS))
910 /* we only print the access for static members */
911 switch (*sym->current)
913 case '0': access = "private: "; break;
914 case '1': access = "protected: "; break;
915 case '2': access = "public: "; break;
919 if (!(sym->flags & UNDNAME_NO_MEMBER_TYPE))
921 if (*sym->current >= '0' && *sym->current <= '2')
922 member_type = "static ";
925 name = get_class_string(sym, 0);
927 switch (dt = *sym->current++)
929 case '0': case '1': case '2':
930 case '3': case '4': case '5':
932 unsigned mark = sym->stack.num;
933 struct array pmt;
935 str_array_init(&pmt);
937 if (!demangle_datatype(sym, &ct, &pmt, FALSE)) goto done;
938 if (!get_modifier(*sym->current++, &modifier)) goto done;
939 sym->stack.num = mark;
941 break;
942 case '6' : /* compiler generated static */
943 case '7' : /* compiler generated static */
944 ct.left = ct.right = NULL;
945 if (!get_modifier(*sym->current++, &modifier)) goto done;
946 if (*sym->current != '@')
948 char* cls = NULL;
950 if (!(cls = get_class_name(sym)))
951 goto done;
952 ct.right = str_printf(sym, "{for `%s'}", cls);
954 break;
955 case '8':
956 case '9':
957 modifier = ct.left = ct.right = NULL;
958 break;
959 default: goto done;
961 if (sym->flags & UNDNAME_NAME_ONLY) ct.left = ct.right = modifier = NULL;
963 sym->result = str_printf(sym, "%s%s%s%s%s%s%s%s", access,
964 member_type, ct.left,
965 modifier && ct.left ? " " : NULL, modifier,
966 modifier || ct.left ? " " : NULL, name, ct.right);
967 ret = TRUE;
968 done:
969 return ret;
972 /******************************************************************
973 * handle_method
974 * Does the final parsing and handling for a function or a method in
975 * a class.
977 static BOOL handle_method(struct parsed_symbol* sym, BOOL cast_op)
979 char accmem;
980 const char* access = NULL;
981 const char* member_type = NULL;
982 struct datatype_t ct_ret;
983 const char* call_conv;
984 const char* modifier = NULL;
985 const char* exported;
986 const char* args_str = NULL;
987 const char* name = NULL;
988 BOOL ret = FALSE;
989 unsigned mark;
990 struct array array_pmt;
992 /* FIXME: why 2 possible letters for each option?
993 * 'A' private:
994 * 'B' private:
995 * 'C' private: static
996 * 'D' private: static
997 * 'E' private: virtual
998 * 'F' private: virtual
999 * 'G' private: thunk
1000 * 'H' private: thunk
1001 * 'I' protected:
1002 * 'J' protected:
1003 * 'K' protected: static
1004 * 'L' protected: static
1005 * 'M' protected: virtual
1006 * 'N' protected: virtual
1007 * 'O' protected: thunk
1008 * 'P' protected: thunk
1009 * 'Q' public:
1010 * 'R' public:
1011 * 'S' public: static
1012 * 'T' public: static
1013 * 'U' public: virtual
1014 * 'V' public: virtual
1015 * 'W' public: thunk
1016 * 'X' public: thunk
1017 * 'Y'
1018 * 'Z'
1020 accmem = *sym->current++;
1021 if (accmem < 'A' || accmem > 'Z') goto done;
1023 if (!(sym->flags & UNDNAME_NO_ACCESS_SPECIFIERS))
1025 switch ((accmem - 'A') / 8)
1027 case 0: access = "private: "; break;
1028 case 1: access = "protected: "; break;
1029 case 2: access = "public: "; break;
1032 if (!(sym->flags & UNDNAME_NO_MEMBER_TYPE))
1034 if (accmem <= 'X')
1036 switch ((accmem - 'A') % 8)
1038 case 2: case 3: member_type = "static "; break;
1039 case 4: case 5: member_type = "virtual "; break;
1040 case 6: case 7:
1041 access = str_printf(sym, "[thunk]:%s", access);
1042 member_type = "virtual ";
1043 break;
1048 name = get_class_string(sym, 0);
1050 if ((accmem - 'A') % 8 == 6 || (accmem - '8') % 8 == 7) /* a thunk */
1051 name = str_printf(sym, "%s`adjustor{%s}' ", name, get_number(sym));
1053 if (accmem <= 'X')
1055 if (((accmem - 'A') % 8) != 2 && ((accmem - 'A') % 8) != 3)
1057 /* Implicit 'this' pointer */
1058 /* If there is an implicit this pointer, const modifier follows */
1059 if (!get_modifier(*sym->current, &modifier)) goto done;
1060 sym->current++;
1064 if (!get_calling_convention(*sym->current++, &call_conv, &exported,
1065 sym->flags))
1066 goto done;
1068 str_array_init(&array_pmt);
1070 /* Return type, or @ if 'void' */
1071 if (*sym->current == '@')
1073 ct_ret.left = "void";
1074 ct_ret.right = NULL;
1075 sym->current++;
1077 else
1079 if (!demangle_datatype(sym, &ct_ret, &array_pmt, FALSE))
1080 goto done;
1082 if (sym->flags & UNDNAME_NO_FUNCTION_RETURNS)
1083 ct_ret.left = ct_ret.right = NULL;
1084 if (cast_op)
1086 name = str_printf(sym, "%s%s%s", name, ct_ret.left, ct_ret.right);
1087 ct_ret.left = ct_ret.right = NULL;
1090 mark = sym->stack.num;
1091 if (!(args_str = get_args(sym, &array_pmt, TRUE, '(', ')'))) goto done;
1092 if (sym->flags & UNDNAME_NAME_ONLY) args_str = modifier = NULL;
1093 sym->stack.num = mark;
1095 /* Note: '()' after 'Z' means 'throws', but we don't care here
1096 * Yet!!! FIXME
1098 sym->result = str_printf(sym, "%s%s%s%s%s%s%s%s%s%s%s%s",
1099 access, member_type, ct_ret.left,
1100 (ct_ret.left && !ct_ret.right) ? " " : NULL,
1101 call_conv, call_conv ? " " : NULL, exported,
1102 name, args_str, modifier,
1103 modifier ? " " : NULL, ct_ret.right);
1104 ret = TRUE;
1105 done:
1106 return ret;
1109 /******************************************************************
1110 * handle_template
1111 * Does the final parsing and handling for a name with templates
1113 static BOOL handle_template(struct parsed_symbol* sym)
1115 const char* name;
1116 const char* args;
1118 assert(*sym->current++ == '$');
1119 if (!(name = get_literal_string(sym))) return FALSE;
1120 if (!(args = get_args(sym, NULL, FALSE, '<', '>'))) return FALSE;
1121 sym->result = str_printf(sym, "%s%s", name, args);
1122 return TRUE;
1125 /*******************************************************************
1126 * symbol_demangle
1127 * Demangle a C++ linker symbol
1129 static BOOL symbol_demangle(struct parsed_symbol* sym)
1131 BOOL ret = FALSE;
1132 unsigned do_after = 0;
1133 static CHAR dashed_null[] = "--null--";
1135 /* FIXME seems wrong as name, as it demangles a simple data type */
1136 if (sym->flags & UNDNAME_NO_ARGUMENTS)
1138 struct datatype_t ct;
1140 if (demangle_datatype(sym, &ct, NULL, FALSE))
1142 sym->result = str_printf(sym, "%s%s", ct.left, ct.right);
1143 ret = TRUE;
1145 goto done;
1148 /* MS mangled names always begin with '?' */
1149 if (*sym->current != '?') return FALSE;
1150 str_array_init(&sym->names);
1151 str_array_init(&sym->stack);
1152 sym->current++;
1154 /* Then function name or operator code */
1155 if (*sym->current == '?' && sym->current[1] != '$')
1157 const char* function_name = NULL;
1159 /* C++ operator code (one character, or two if the first is '_') */
1160 switch (*++sym->current)
1162 case '0': do_after = 1; break;
1163 case '1': do_after = 2; break;
1164 case '2': function_name = "operator new"; break;
1165 case '3': function_name = "operator delete"; break;
1166 case '4': function_name = "operator="; break;
1167 case '5': function_name = "operator>>"; break;
1168 case '6': function_name = "operator<<"; break;
1169 case '7': function_name = "operator!"; break;
1170 case '8': function_name = "operator=="; break;
1171 case '9': function_name = "operator!="; break;
1172 case 'A': function_name = "operator[]"; break;
1173 case 'B': function_name = "operator "; do_after = 3; break;
1174 case 'C': function_name = "operator->"; break;
1175 case 'D': function_name = "operator*"; break;
1176 case 'E': function_name = "operator++"; break;
1177 case 'F': function_name = "operator--"; break;
1178 case 'G': function_name = "operator-"; break;
1179 case 'H': function_name = "operator+"; break;
1180 case 'I': function_name = "operator&"; break;
1181 case 'J': function_name = "operator->*"; break;
1182 case 'K': function_name = "operator/"; break;
1183 case 'L': function_name = "operator%"; break;
1184 case 'M': function_name = "operator<"; break;
1185 case 'N': function_name = "operator<="; break;
1186 case 'O': function_name = "operator>"; break;
1187 case 'P': function_name = "operator>="; break;
1188 case 'Q': function_name = "operator,"; break;
1189 case 'R': function_name = "operator()"; break;
1190 case 'S': function_name = "operator~"; break;
1191 case 'T': function_name = "operator^"; break;
1192 case 'U': function_name = "operator|"; break;
1193 case 'V': function_name = "operator&&"; break;
1194 case 'W': function_name = "operator||"; break;
1195 case 'X': function_name = "operator*="; break;
1196 case 'Y': function_name = "operator+="; break;
1197 case 'Z': function_name = "operator-="; break;
1198 case '_':
1199 switch (*++sym->current)
1201 case '0': function_name = "operator/="; break;
1202 case '1': function_name = "operator%="; break;
1203 case '2': function_name = "operator>>="; break;
1204 case '3': function_name = "operator<<="; break;
1205 case '4': function_name = "operator&="; break;
1206 case '5': function_name = "operator|="; break;
1207 case '6': function_name = "operator^="; break;
1208 case '7': function_name = "`vftable'"; break;
1209 case '8': function_name = "`vbtable'"; break;
1210 case '9': function_name = "`vcall'"; break;
1211 case 'A': function_name = "`typeof'"; break;
1212 case 'B': function_name = "`local static guard'"; break;
1213 case 'C': function_name = "`string'"; do_after = 4; break;
1214 case 'D': function_name = "`vbase destructor'"; break;
1215 case 'E': function_name = "`vector deleting destructor'"; break;
1216 case 'F': function_name = "`default constructor closure'"; break;
1217 case 'G': function_name = "`scalar deleting destructor'"; break;
1218 case 'H': function_name = "`vector constructor iterator'"; break;
1219 case 'I': function_name = "`vector destructor iterator'"; break;
1220 case 'J': function_name = "`vector vbase constructor iterator'"; break;
1221 case 'K': function_name = "`virtual displacement map'"; break;
1222 case 'L': function_name = "`eh vector constructor iterator'"; break;
1223 case 'M': function_name = "`eh vector destructor iterator'"; break;
1224 case 'N': function_name = "`eh vector vbase constructor iterator'"; break;
1225 case 'O': function_name = "`copy constructor closure'"; break;
1226 case 'R':
1227 sym->flags |= UNDNAME_NO_FUNCTION_RETURNS;
1228 switch (*++sym->current)
1230 case '0':
1232 struct datatype_t ct;
1233 struct array pmt;
1235 sym->current++;
1236 str_array_init(&pmt);
1237 demangle_datatype(sym, &ct, &pmt, FALSE);
1238 function_name = str_printf(sym, "%s%s `RTTI Type Descriptor'",
1239 ct.left, ct.right);
1240 sym->current--;
1242 break;
1243 case '1':
1245 const char* n1, *n2, *n3, *n4;
1246 sym->current++;
1247 n1 = get_number(sym);
1248 n2 = get_number(sym);
1249 n3 = get_number(sym);
1250 n4 = get_number(sym);
1251 sym->current--;
1252 function_name = str_printf(sym, "`RTTI Base Class Descriptor at (%s,%s,%s,%s)'",
1253 n1, n2, n3, n4);
1255 break;
1256 case '2': function_name = "`RTTI Base Class Array'"; break;
1257 case '3': function_name = "`RTTI Class Hierarchy Descriptor'"; break;
1258 case '4': function_name = "`RTTI Complete Object Locator'"; break;
1259 default:
1260 ERR("Unknown RTTI operator: _R%c\n", *sym->current);
1261 break;
1263 break;
1264 case 'S': function_name = "`local vftable'"; break;
1265 case 'T': function_name = "`local vftable constructor closure'"; break;
1266 case 'U': function_name = "operator new[]"; break;
1267 case 'V': function_name = "operator delete[]"; break;
1268 case 'X': function_name = "`placement delete closure'"; break;
1269 case 'Y': function_name = "`placement delete[] closure'"; break;
1270 default:
1271 ERR("Unknown operator: _%c\n", *sym->current);
1272 return FALSE;
1274 break;
1275 default:
1276 /* FIXME: Other operators */
1277 ERR("Unknown operator: %c\n", *sym->current);
1278 return FALSE;
1280 sym->current++;
1281 switch (do_after)
1283 case 1: case 2:
1284 sym->stack.num = sym->stack.max = 1;
1285 sym->stack.elts[0] = dashed_null;
1286 break;
1287 case 4:
1288 sym->result = (char*)function_name;
1289 ret = TRUE;
1290 goto done;
1291 default:
1292 str_array_push(sym, function_name, -1, &sym->stack);
1293 break;
1296 else if (*sym->current == '$')
1298 /* Strange construct, it's a name with a template argument list
1299 and that's all. */
1300 sym->current++;
1301 ret = (sym->result = get_template_name(sym)) != NULL;
1302 goto done;
1304 else if (*sym->current == '?' && sym->current[1] == '$')
1305 do_after = 5;
1307 /* Either a class name, or '@' if the symbol is not a class member */
1308 switch (*sym->current)
1310 case '@': sym->current++; break;
1311 case '$': break;
1312 default:
1313 /* Class the function is associated with, terminated by '@@' */
1314 if (!get_class(sym)) goto done;
1315 break;
1318 switch (do_after)
1320 case 0: default: break;
1321 case 1: case 2:
1322 /* it's time to set the member name for ctor & dtor */
1323 if (sym->stack.num <= 1) goto done;
1324 if (do_after == 1)
1325 sym->stack.elts[0] = sym->stack.elts[1];
1326 else
1327 sym->stack.elts[0] = str_printf(sym, "~%s", sym->stack.elts[1]);
1328 /* ctors and dtors don't have return type */
1329 sym->flags |= UNDNAME_NO_FUNCTION_RETURNS;
1330 break;
1331 case 3:
1332 sym->flags &= ~UNDNAME_NO_FUNCTION_RETURNS;
1333 break;
1334 case 5:
1335 sym->names.start = 1;
1336 break;
1339 /* Function/Data type and access level */
1340 if (*sym->current >= '0' && *sym->current <= '9')
1341 ret = handle_data(sym);
1342 else if (*sym->current >= 'A' && *sym->current <= 'Z')
1343 ret = handle_method(sym, do_after == 3);
1344 else if (*sym->current == '$')
1345 ret = handle_template(sym);
1346 else ret = FALSE;
1347 done:
1348 if (ret) assert(sym->result);
1349 else WARN("Failed at %s\n", sym->current);
1351 return ret;
1354 /*********************************************************************
1355 * __unDNameEx (MSVCRT.@)
1357 * Demangle a C++ identifier.
1359 * PARAMS
1360 * buffer [O] If not NULL, the place to put the demangled string
1361 * mangled [I] Mangled name of the function
1362 * buflen [I] Length of buffer
1363 * memget [I] Function to allocate memory with
1364 * memfree [I] Function to free memory with
1365 * unknown [?] Unknown, possibly a call back
1366 * flags [I] Flags determining demangled format
1368 * RETURNS
1369 * Success: A string pointing to the unmangled name, allocated with memget.
1370 * Failure: NULL.
1372 char* CDECL __unDNameEx(char* buffer, const char* mangled, int buflen,
1373 malloc_func_t memget, free_func_t memfree,
1374 void* unknown, unsigned short int flags)
1376 struct parsed_symbol sym;
1377 const char* result;
1379 TRACE("(%p,%s,%d,%p,%p,%p,%x)\n",
1380 buffer, mangled, buflen, memget, memfree, unknown, flags);
1382 /* The flags details is not documented by MS. However, it looks exactly
1383 * like the UNDNAME_ manifest constants from imagehlp.h and dbghelp.h
1384 * So, we copied those (on top of the file)
1386 memset(&sym, 0, sizeof(struct parsed_symbol));
1387 if (flags & UNDNAME_NAME_ONLY)
1388 flags |= UNDNAME_NO_FUNCTION_RETURNS | UNDNAME_NO_ACCESS_SPECIFIERS |
1389 UNDNAME_NO_MEMBER_TYPE | UNDNAME_NO_ALLOCATION_LANGUAGE |
1390 UNDNAME_NO_COMPLEX_TYPE;
1392 sym.flags = flags;
1393 sym.mem_alloc_ptr = memget;
1394 sym.mem_free_ptr = memfree;
1395 sym.current = mangled;
1397 result = symbol_demangle(&sym) ? sym.result : mangled;
1398 if (buffer && buflen)
1400 lstrcpynA( buffer, result, buflen);
1402 else
1404 buffer = memget(strlen(result) + 1);
1405 if (buffer) strcpy(buffer, result);
1408 und_free_all(&sym);
1410 return buffer;
1414 /*********************************************************************
1415 * __unDName (MSVCRT.@)
1417 char* CDECL __unDName(char* buffer, const char* mangled, int buflen,
1418 malloc_func_t memget, free_func_t memfree,
1419 unsigned short int flags)
1421 return __unDNameEx(buffer, mangled, buflen, memget, memfree, NULL, flags);