crypt32: Remove an unnecessary test for the extended key usage extension in CA certif...
[wine.git] / dlls / msvcrt / undname.c
blobcb079050bda46e7a4f26f876f71a97ad2c5b083b
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 <stdlib.h>
28 #include "msvcrt.h"
30 #include "wine/debug.h"
32 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt);
34 /* TODO:
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
61 * {<A>}: ""
62 * {<B>}: "const "
63 * {<C>}: "volatile "
64 * {<D>}: "const volatile "
66 * in arguments:
67 * P<M>x {<M>}x*
68 * Q<M>x {<M>}x* const
69 * A<M>x {<M>}x&
70 * in data fields:
71 * same as for arguments and also the following
72 * ?<M>x {<M>}x
76 struct array
78 unsigned start; /* first valid reference in array */
79 unsigned num; /* total number of used elts */
80 unsigned max;
81 unsigned alloc;
82 char** elts;
85 /* Structure holding a parsed symbol */
86 struct 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 */
103 struct datatype_t
105 const char* left;
106 const char* right;
109 /******************************************************************
110 * und_alloc
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, unsigned int len)
118 void* ptr;
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*);
133 else
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;
148 return ptr;
149 #undef BLOCK_SIZE
150 #undef AVAIL_SIZE
153 /******************************************************************
154 * und_free
155 * Frees all the blocks in the list of large blocks allocated by
156 * und_alloc.
158 static void und_free_all(struct parsed_symbol* sym)
160 void* next;
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 /******************************************************************
172 * str_array_init
173 * Initialises an array of strings
175 static void str_array_init(struct array* a)
177 a->start = a->num = a->max = a->alloc = 0;
178 a->elts = NULL;
181 /******************************************************************
182 * str_array_push
183 * Adding a new string to an array
185 static BOOL str_array_push(struct parsed_symbol* sym, const char* ptr, int len,
186 struct array* a)
188 char** new;
190 assert(ptr);
191 assert(a);
193 if (!a->alloc)
195 new = und_alloc(sym, (a->alloc = 32) * sizeof(a->elts[0]));
196 if (!new) return FALSE;
197 a->elts = new;
199 else if (a->max >= a->alloc)
201 new = und_alloc(sym, (a->alloc * 2) * sizeof(a->elts[0]));
202 if (!new) return FALSE;
203 memcpy(new, a->elts, a->alloc * sizeof(a->elts[0]));
204 a->alloc *= 2;
205 a->elts = new;
207 if (len == -1) len = strlen(ptr);
208 a->elts[a->num] = und_alloc(sym, len + 1);
209 assert(a->elts[a->num]);
210 memcpy(a->elts[a->num], ptr, len);
211 a->elts[a->num][len] = '\0';
212 if (++a->num >= a->max) a->max = a->num;
214 int i;
215 char c;
217 for (i = a->max - 1; i >= 0; i--)
219 c = '>';
220 if (i < a->start) c = '-';
221 else if (i >= a->num) c = '}';
222 TRACE("%p\t%d%c %s\n", a, i, c, a->elts[i]);
226 return TRUE;
229 /******************************************************************
230 * str_array_get_ref
231 * Extracts a reference from an existing array (doing proper type
232 * checking)
234 static char* str_array_get_ref(struct array* cref, unsigned idx)
236 assert(cref);
237 if (cref->start + idx >= cref->max)
239 WARN("Out of bounds: %p %d + %d >= %d\n",
240 cref, cref->start, idx, cref->max);
241 return NULL;
243 TRACE("Returning %p[%d] => %s\n",
244 cref, idx, cref->elts[cref->start + idx]);
245 return cref->elts[cref->start + idx];
248 /******************************************************************
249 * str_printf
250 * Helper for printf type of command (only %s and %c are implemented)
251 * while dynamically allocating the buffer
253 static char* str_printf(struct parsed_symbol* sym, const char* format, ...)
255 va_list args;
256 unsigned int len = 1, i, sz;
257 char* tmp;
258 char* p;
259 char* t;
261 va_start(args, format);
262 for (i = 0; format[i]; i++)
264 if (format[i] == '%')
266 switch (format[++i])
268 case 's': t = va_arg(args, char*); if (t) len += strlen(t); break;
269 case 'c': (void)va_arg(args, int); len++; break;
270 default: i--; /* fall thru */
271 case '%': len++; break;
274 else len++;
276 va_end(args);
277 if (!(tmp = und_alloc(sym, len))) return NULL;
278 va_start(args, format);
279 for (p = tmp, i = 0; format[i]; i++)
281 if (format[i] == '%')
283 switch (format[++i])
285 case 's':
286 t = va_arg(args, char*);
287 if (t)
289 sz = strlen(t);
290 memcpy(p, t, sz);
291 p += sz;
293 break;
294 case 'c':
295 *p++ = (char)va_arg(args, int);
296 break;
297 default: i--; /* fall thru */
298 case '%': *p++ = '%'; break;
301 else *p++ = format[i];
303 va_end(args);
304 *p = '\0';
305 return tmp;
308 /* forward declaration */
309 static BOOL demangle_datatype(struct parsed_symbol* sym, struct datatype_t* ct,
310 struct array* pmt, BOOL in_args);
312 static const char* get_number(struct parsed_symbol* sym)
314 char* ptr;
315 BOOL sgn = FALSE;
317 if (*sym->current == '?')
319 sgn = TRUE;
320 sym->current++;
322 if (*sym->current >= '0' && *sym->current <= '8')
324 ptr = und_alloc(sym, 3);
325 if (sgn) ptr[0] = '-';
326 ptr[sgn ? 1 : 0] = *sym->current + 1;
327 ptr[sgn ? 2 : 1] = '\0';
328 sym->current++;
330 else if (*sym->current == '9')
332 ptr = und_alloc(sym, 4);
333 if (sgn) ptr[0] = '-';
334 ptr[sgn ? 1 : 0] = '1';
335 ptr[sgn ? 2 : 1] = '0';
336 ptr[sgn ? 3 : 2] = '\0';
337 sym->current++;
339 else if (*sym->current >= 'A' && *sym->current <= 'P')
341 int ret = 0;
343 while (*sym->current >= 'A' && *sym->current <= 'P')
345 ret *= 16;
346 ret += *sym->current++ - 'A';
348 if (*sym->current != '@') return NULL;
350 ptr = und_alloc(sym, 17);
351 sprintf(ptr, "%s%d", sgn ? "-" : "", ret);
352 sym->current++;
354 else return NULL;
355 return ptr;
358 /******************************************************************
359 * get_args
360 * Parses a list of function/method arguments, creates a string corresponding
361 * to the arguments' list.
363 static char* get_args(struct parsed_symbol* sym, struct array* pmt_ref, BOOL z_term,
364 char open_char, char close_char)
367 struct datatype_t ct;
368 struct array arg_collect;
369 char* args_str = NULL;
370 unsigned int i;
372 str_array_init(&arg_collect);
374 /* Now come the function arguments */
375 while (*sym->current)
377 /* Decode each data type and append it to the argument list */
378 if (*sym->current == '@')
380 sym->current++;
381 break;
383 if (!demangle_datatype(sym, &ct, pmt_ref, TRUE))
384 return NULL;
385 /* 'void' terminates an argument list in a function */
386 if (z_term && !strcmp(ct.left, "void")) break;
387 if (!str_array_push(sym, str_printf(sym, "%s%s", ct.left, ct.right), -1,
388 &arg_collect))
389 return NULL;
390 if (!strcmp(ct.left, "...")) break;
392 /* Functions are always terminated by 'Z'. If we made it this far and
393 * don't find it, we have incorrectly identified a data type.
395 if (z_term && *sym->current++ != 'Z') return NULL;
397 if (arg_collect.num == 0 ||
398 (arg_collect.num == 1 && !strcmp(arg_collect.elts[0], "void")))
399 return str_printf(sym, "%cvoid%c", open_char, close_char);
400 for (i = 1; i < arg_collect.num; i++)
402 args_str = str_printf(sym, "%s,%s", args_str, arg_collect.elts[i]);
405 if (close_char == '>' && args_str && args_str[strlen(args_str) - 1] == '>')
406 args_str = str_printf(sym, "%c%s%s %c",
407 open_char, arg_collect.elts[0], args_str, close_char);
408 else
409 args_str = str_printf(sym, "%c%s%s%c",
410 open_char, arg_collect.elts[0], args_str, close_char);
412 return args_str;
415 /******************************************************************
416 * get_modifier
417 * Parses the type modifier. Always returns a static string
419 static BOOL get_modifier(char ch, const char** ret)
421 switch (ch)
423 case 'A': *ret = NULL; break;
424 case 'B': *ret = "const"; break;
425 case 'C': *ret = "volatile"; break;
426 case 'D': *ret = "const volatile"; break;
427 default: return FALSE;
429 return TRUE;
432 static BOOL get_modified_type(struct datatype_t *ct, struct parsed_symbol* sym,
433 struct array *pmt_ref, char modif, BOOL in_args)
435 const char* modifier;
436 const char* str_modif;
438 switch (modif)
440 case 'A': str_modif = " &"; break;
441 case 'B': str_modif = " & volatile"; break;
442 case 'P': str_modif = " *"; break;
443 case 'Q': str_modif = " * const"; break;
444 case 'R': str_modif = " * volatile"; break;
445 case 'S': str_modif = " * const volatile"; break;
446 case '?': str_modif = ""; break;
447 default: return FALSE;
450 if (get_modifier(*sym->current++, &modifier))
452 unsigned mark = sym->stack.num;
453 struct datatype_t sub_ct;
455 /* multidimensional arrays */
456 if (*sym->current == 'Y')
458 const char* n1;
459 int num;
461 sym->current++;
462 if (!(n1 = get_number(sym))) return FALSE;
463 num = atoi(n1);
465 if (str_modif[0] == ' ' && !modifier)
466 str_modif++;
468 if (modifier)
470 str_modif = str_printf(sym, " (%s%s)", modifier, str_modif);
471 modifier = NULL;
473 else
474 str_modif = str_printf(sym, " (%s)", str_modif);
476 while (num--)
477 str_modif = str_printf(sym, "%s[%s]", str_modif, get_number(sym));
480 /* Recurse to get the referred-to type */
481 if (!demangle_datatype(sym, &sub_ct, pmt_ref, FALSE))
482 return FALSE;
483 if (modifier)
484 ct->left = str_printf(sym, "%s %s%s", sub_ct.left, modifier, str_modif );
485 else
487 /* don't insert a space between duplicate '*' */
488 if (!in_args && str_modif[0] && str_modif[1] == '*' && sub_ct.left[strlen(sub_ct.left)-1] == '*')
489 str_modif++;
490 ct->left = str_printf(sym, "%s%s", sub_ct.left, str_modif );
492 ct->right = sub_ct.right;
493 sym->stack.num = mark;
495 return TRUE;
498 /******************************************************************
499 * get_literal_string
500 * Gets the literal name from the current position in the mangled
501 * symbol to the first '@' character. It pushes the parsed name to
502 * the symbol names stack and returns a pointer to it or NULL in
503 * case of an error.
505 static char* get_literal_string(struct parsed_symbol* sym)
507 const char *ptr = sym->current;
509 do {
510 if (!((*sym->current >= 'A' && *sym->current <= 'Z') ||
511 (*sym->current >= 'a' && *sym->current <= 'z') ||
512 (*sym->current >= '0' && *sym->current <= '9') ||
513 *sym->current == '_' || *sym->current == '$')) {
514 TRACE("Failed at '%c' in %s\n", *sym->current, ptr);
515 return NULL;
517 } while (*++sym->current != '@');
518 sym->current++;
519 if (!str_array_push(sym, ptr, sym->current - 1 - ptr, &sym->names))
520 return NULL;
522 return str_array_get_ref(&sym->names, sym->names.num - sym->names.start - 1);
525 /******************************************************************
526 * get_template_name
527 * Parses a name with a template argument list and returns it as
528 * a string.
529 * In a template argument list the back reference to the names
530 * table is separately created. '0' points to the class component
531 * name with the template arguments. We use the same stack array
532 * to hold the names but save/restore the stack state before/after
533 * parsing the template argument list.
535 static char* get_template_name(struct parsed_symbol* sym)
537 char *name, *args;
538 unsigned num_mark = sym->names.num;
539 unsigned start_mark = sym->names.start;
540 unsigned stack_mark = sym->stack.num;
541 struct array array_pmt;
543 sym->names.start = sym->names.num;
544 if (!(name = get_literal_string(sym)))
545 return FALSE;
546 str_array_init(&array_pmt);
547 args = get_args(sym, &array_pmt, FALSE, '<', '>');
548 if (args != NULL)
549 name = str_printf(sym, "%s%s", name, args);
550 sym->names.num = num_mark;
551 sym->names.start = start_mark;
552 sym->stack.num = stack_mark;
553 return name;
556 /******************************************************************
557 * get_class
558 * Parses class as a list of parent-classes, terminated by '@' and stores the
559 * result in 'a' array. Each parent-classes, as well as the inner element
560 * (either field/method name or class name), are represented in the mangled
561 * name by a literal name ([a-zA-Z0-9_]+ terminated by '@') or a back reference
562 * ([0-9]) or a name with template arguments ('?$' literal name followed by the
563 * template argument list). The class name components appear in the reverse
564 * order in the mangled name, e.g aaa@bbb@ccc@@ will be demangled to
565 * ccc::bbb::aaa
566 * For each of these class name components a string will be allocated in the
567 * array.
569 static BOOL get_class(struct parsed_symbol* sym)
571 const char* name = NULL;
573 while (*sym->current != '@')
575 switch (*sym->current)
577 case '\0': return FALSE;
579 case '0': case '1': case '2': case '3':
580 case '4': case '5': case '6': case '7':
581 case '8': case '9':
582 name = str_array_get_ref(&sym->names, *sym->current++ - '0');
583 break;
584 case '?':
585 if (*++sym->current == '$')
587 sym->current++;
588 if ((name = get_template_name(sym)) &&
589 !str_array_push(sym, name, -1, &sym->names))
590 return FALSE;
592 break;
593 default:
594 name = get_literal_string(sym);
595 break;
597 if (!name || !str_array_push(sym, name, -1, &sym->stack))
598 return FALSE;
600 sym->current++;
601 return TRUE;
604 /******************************************************************
605 * get_class_string
606 * From an array collected by get_class in sym->stack, constructs the
607 * corresponding (allocated) string
609 static char* get_class_string(struct parsed_symbol* sym, int start)
611 int i;
612 unsigned int len, sz;
613 char* ret;
614 struct array *a = &sym->stack;
616 for (len = 0, i = start; i < a->num; i++)
618 assert(a->elts[i]);
619 len += 2 + strlen(a->elts[i]);
621 if (!(ret = und_alloc(sym, len - 1))) return NULL;
622 for (len = 0, i = a->num - 1; i >= start; i--)
624 sz = strlen(a->elts[i]);
625 memcpy(ret + len, a->elts[i], sz);
626 len += sz;
627 if (i > start)
629 ret[len++] = ':';
630 ret[len++] = ':';
633 ret[len] = '\0';
634 return ret;
637 /******************************************************************
638 * get_class_name
639 * Wrapper around get_class and get_class_string.
641 static char* get_class_name(struct parsed_symbol* sym)
643 unsigned mark = sym->stack.num;
644 char* s = NULL;
646 if (get_class(sym))
647 s = get_class_string(sym, mark);
648 sym->stack.num = mark;
649 return s;
652 /******************************************************************
653 * get_calling_convention
654 * Returns a static string corresponding to the calling convention described
655 * by char 'ch'. Sets export to TRUE iff the calling convention is exported.
657 static BOOL get_calling_convention(char ch, const char** call_conv,
658 const char** exported, unsigned flags)
660 *call_conv = *exported = NULL;
662 if (!(flags & (UNDNAME_NO_MS_KEYWORDS | UNDNAME_NO_ALLOCATION_LANGUAGE)))
664 if (flags & UNDNAME_NO_LEADING_UNDERSCORES)
666 if (((ch - 'A') % 2) == 1) *exported = "dll_export ";
667 switch (ch)
669 case 'A': case 'B': *call_conv = "cdecl"; break;
670 case 'C': case 'D': *call_conv = "pascal"; break;
671 case 'E': case 'F': *call_conv = "thiscall"; break;
672 case 'G': case 'H': *call_conv = "stdcall"; break;
673 case 'I': case 'J': *call_conv = "fastcall"; break;
674 case 'K': case 'L': break;
675 case 'M': *call_conv = "clrcall"; break;
676 default: ERR("Unknown calling convention %c\n", ch); return FALSE;
679 else
681 if (((ch - 'A') % 2) == 1) *exported = "__dll_export ";
682 switch (ch)
684 case 'A': case 'B': *call_conv = "__cdecl"; break;
685 case 'C': case 'D': *call_conv = "__pascal"; break;
686 case 'E': case 'F': *call_conv = "__thiscall"; break;
687 case 'G': case 'H': *call_conv = "__stdcall"; break;
688 case 'I': case 'J': *call_conv = "__fastcall"; break;
689 case 'K': case 'L': break;
690 case 'M': *call_conv = "__clrcall"; break;
691 default: ERR("Unknown calling convention %c\n", ch); return FALSE;
695 return TRUE;
698 /*******************************************************************
699 * get_simple_type
700 * Return a string containing an allocated string for a simple data type
702 static const char* get_simple_type(char c)
704 const char* type_string;
706 switch (c)
708 case 'C': type_string = "signed char"; break;
709 case 'D': type_string = "char"; break;
710 case 'E': type_string = "unsigned char"; break;
711 case 'F': type_string = "short"; break;
712 case 'G': type_string = "unsigned short"; break;
713 case 'H': type_string = "int"; break;
714 case 'I': type_string = "unsigned int"; break;
715 case 'J': type_string = "long"; break;
716 case 'K': type_string = "unsigned long"; break;
717 case 'M': type_string = "float"; break;
718 case 'N': type_string = "double"; break;
719 case 'O': type_string = "long double"; break;
720 case 'X': type_string = "void"; break;
721 case 'Z': type_string = "..."; break;
722 default: type_string = NULL; break;
724 return type_string;
727 /*******************************************************************
728 * get_extended_type
729 * Return a string containing an allocated string for a simple data type
731 static const char* get_extended_type(char c)
733 const char* type_string;
735 switch (c)
737 case 'D': type_string = "__int8"; break;
738 case 'E': type_string = "unsigned __int8"; break;
739 case 'F': type_string = "__int16"; break;
740 case 'G': type_string = "unsigned __int16"; break;
741 case 'H': type_string = "__int32"; break;
742 case 'I': type_string = "unsigned __int32"; break;
743 case 'J': type_string = "__int64"; break;
744 case 'K': type_string = "unsigned __int64"; break;
745 case 'L': type_string = "__int128"; break;
746 case 'M': type_string = "unsigned __int128"; break;
747 case 'N': type_string = "bool"; break;
748 case 'W': type_string = "wchar_t"; break;
749 default: type_string = NULL; break;
751 return type_string;
754 /*******************************************************************
755 * demangle_datatype
757 * Attempt to demangle a C++ data type, which may be datatype.
758 * a datatype type is made up of a number of simple types. e.g:
759 * char** = (pointer to (pointer to (char)))
761 static BOOL demangle_datatype(struct parsed_symbol* sym, struct datatype_t* ct,
762 struct array* pmt_ref, BOOL in_args)
764 char dt;
765 BOOL add_pmt = TRUE;
767 assert(ct);
768 ct->left = ct->right = NULL;
770 switch (dt = *sym->current++)
772 case '_':
773 /* MS type: __int8,__int16 etc */
774 ct->left = get_extended_type(*sym->current++);
775 break;
776 case 'C': case 'D': case 'E': case 'F': case 'G':
777 case 'H': case 'I': case 'J': case 'K': case 'M':
778 case 'N': case 'O': case 'X': case 'Z':
779 /* Simple data types */
780 ct->left = get_simple_type(dt);
781 add_pmt = FALSE;
782 break;
783 case 'T': /* union */
784 case 'U': /* struct */
785 case 'V': /* class */
786 case 'Y': /* cointerface */
787 /* Class/struct/union/cointerface */
789 const char* struct_name = NULL;
790 const char* type_name = NULL;
792 if (!(struct_name = get_class_name(sym)))
793 goto done;
794 if (!(sym->flags & UNDNAME_NO_COMPLEX_TYPE))
796 switch (dt)
798 case 'T': type_name = "union "; break;
799 case 'U': type_name = "struct "; break;
800 case 'V': type_name = "class "; break;
801 case 'Y': type_name = "cointerface "; break;
804 ct->left = str_printf(sym, "%s%s", type_name, struct_name);
806 break;
807 case '?':
808 /* not all the time is seems */
809 if (in_args)
811 const char* ptr;
812 if (!(ptr = get_number(sym))) goto done;
813 ct->left = str_printf(sym, "`template-parameter-%s'", ptr);
815 else
817 if (!get_modified_type(ct, sym, pmt_ref, '?', in_args)) goto done;
819 break;
820 case 'A': /* reference */
821 case 'B': /* volatile reference */
822 if (!get_modified_type(ct, sym, pmt_ref, dt, in_args)) goto done;
823 break;
824 case 'Q': /* const pointer */
825 case 'R': /* volatile pointer */
826 case 'S': /* const volatile pointer */
827 if (!get_modified_type(ct, sym, pmt_ref, in_args ? dt : 'P', in_args)) goto done;
828 break;
829 case 'P': /* Pointer */
830 if (isdigit(*sym->current))
832 /* FIXME: P6 = Function pointer, others who knows.. */
833 if (*sym->current++ == '6')
835 char* args = NULL;
836 const char* call_conv;
837 const char* exported;
838 struct datatype_t sub_ct;
839 unsigned mark = sym->stack.num;
841 if (!get_calling_convention(*sym->current++,
842 &call_conv, &exported,
843 sym->flags & ~UNDNAME_NO_ALLOCATION_LANGUAGE) ||
844 !demangle_datatype(sym, &sub_ct, pmt_ref, FALSE))
845 goto done;
847 args = get_args(sym, pmt_ref, TRUE, '(', ')');
848 if (!args) goto done;
849 sym->stack.num = mark;
851 ct->left = str_printf(sym, "%s%s (%s*",
852 sub_ct.left, sub_ct.right, call_conv);
853 ct->right = str_printf(sym, ")%s", args);
855 else goto done;
857 else if (!get_modified_type(ct, sym, pmt_ref, 'P', in_args)) goto done;
858 break;
859 case 'W':
860 if (*sym->current == '4')
862 char* enum_name;
863 sym->current++;
864 if (!(enum_name = get_class_name(sym)))
865 goto done;
866 if (sym->flags & UNDNAME_NO_COMPLEX_TYPE)
867 ct->left = enum_name;
868 else
869 ct->left = str_printf(sym, "enum %s", enum_name);
871 else goto done;
872 break;
873 case '0': case '1': case '2': case '3': case '4':
874 case '5': case '6': case '7': case '8': case '9':
875 /* Referring back to previously parsed type */
876 /* left and right are pushed as two separate strings */
877 ct->left = str_array_get_ref(pmt_ref, (dt - '0') * 2);
878 ct->right = str_array_get_ref(pmt_ref, (dt - '0') * 2 + 1);
879 if (!ct->left) goto done;
880 add_pmt = FALSE;
881 break;
882 case '$':
883 switch (*sym->current++)
885 case '0':
886 if (!(ct->left = get_number(sym))) goto done;
887 break;
888 case 'D':
890 const char* ptr;
891 if (!(ptr = get_number(sym))) goto done;
892 ct->left = str_printf(sym, "`template-parameter%s'", ptr);
894 break;
895 case 'F':
897 const char* p1;
898 const char* p2;
899 if (!(p1 = get_number(sym))) goto done;
900 if (!(p2 = get_number(sym))) goto done;
901 ct->left = str_printf(sym, "{%s,%s}", p1, p2);
903 break;
904 case 'G':
906 const char* p1;
907 const char* p2;
908 const char* p3;
909 if (!(p1 = get_number(sym))) goto done;
910 if (!(p2 = get_number(sym))) goto done;
911 if (!(p3 = get_number(sym))) goto done;
912 ct->left = str_printf(sym, "{%s,%s,%s}", p1, p2, p3);
914 break;
915 case 'Q':
917 const char* ptr;
918 if (!(ptr = get_number(sym))) goto done;
919 ct->left = str_printf(sym, "`non-type-template-parameter%s'", ptr);
921 break;
922 case '$':
923 if (*sym->current == 'C')
925 const char* ptr;
927 sym->current++;
928 if (!get_modifier(*sym->current++, &ptr)) goto done;
929 if (!demangle_datatype(sym, ct, pmt_ref, in_args)) goto done;
930 ct->left = str_printf(sym, "%s %s", ct->left, ptr);
932 break;
934 break;
935 default :
936 ERR("Unknown type %c\n", dt);
937 break;
939 if (add_pmt && pmt_ref && in_args)
941 /* left and right are pushed as two separate strings */
942 if (!str_array_push(sym, ct->left ? ct->left : "", -1, pmt_ref) ||
943 !str_array_push(sym, ct->right ? ct->right : "", -1, pmt_ref))
944 return FALSE;
946 done:
948 return ct->left != NULL;
951 /******************************************************************
952 * handle_data
953 * Does the final parsing and handling for a variable or a field in
954 * a class.
956 static BOOL handle_data(struct parsed_symbol* sym)
958 const char* access = NULL;
959 const char* member_type = NULL;
960 const char* modifier = NULL;
961 struct datatype_t ct;
962 char* name = NULL;
963 BOOL ret = FALSE;
965 /* 0 private static
966 * 1 protected static
967 * 2 public static
968 * 3 private non-static
969 * 4 protected non-static
970 * 5 public non-static
971 * 6 ?? static
972 * 7 ?? static
975 if (!(sym->flags & UNDNAME_NO_ACCESS_SPECIFIERS))
977 /* we only print the access for static members */
978 switch (*sym->current)
980 case '0': access = "private: "; break;
981 case '1': access = "protected: "; break;
982 case '2': access = "public: "; break;
986 if (!(sym->flags & UNDNAME_NO_MEMBER_TYPE))
988 if (*sym->current >= '0' && *sym->current <= '2')
989 member_type = "static ";
992 name = get_class_string(sym, 0);
994 switch (*sym->current++)
996 case '0': case '1': case '2':
997 case '3': case '4': case '5':
999 unsigned mark = sym->stack.num;
1000 struct array pmt;
1002 str_array_init(&pmt);
1004 if (!demangle_datatype(sym, &ct, &pmt, FALSE)) goto done;
1005 if (!get_modifier(*sym->current++, &modifier)) goto done;
1006 sym->stack.num = mark;
1008 break;
1009 case '6' : /* compiler generated static */
1010 case '7' : /* compiler generated static */
1011 ct.left = ct.right = NULL;
1012 if (!get_modifier(*sym->current++, &modifier)) goto done;
1013 if (*sym->current != '@')
1015 char* cls = NULL;
1017 if (!(cls = get_class_name(sym)))
1018 goto done;
1019 ct.right = str_printf(sym, "{for `%s'}", cls);
1021 break;
1022 case '8':
1023 case '9':
1024 modifier = ct.left = ct.right = NULL;
1025 break;
1026 default: goto done;
1028 if (sym->flags & UNDNAME_NAME_ONLY) ct.left = ct.right = modifier = NULL;
1030 sym->result = str_printf(sym, "%s%s%s%s%s%s%s%s", access,
1031 member_type, ct.left,
1032 modifier && ct.left ? " " : NULL, modifier,
1033 modifier || ct.left ? " " : NULL, name, ct.right);
1034 ret = TRUE;
1035 done:
1036 return ret;
1039 /******************************************************************
1040 * handle_method
1041 * Does the final parsing and handling for a function or a method in
1042 * a class.
1044 static BOOL handle_method(struct parsed_symbol* sym, BOOL cast_op)
1046 char accmem;
1047 const char* access = NULL;
1048 const char* member_type = NULL;
1049 struct datatype_t ct_ret;
1050 const char* call_conv;
1051 const char* modifier = NULL;
1052 const char* exported;
1053 const char* args_str = NULL;
1054 const char* name = NULL;
1055 BOOL ret = FALSE;
1056 unsigned mark;
1057 struct array array_pmt;
1059 /* FIXME: why 2 possible letters for each option?
1060 * 'A' private:
1061 * 'B' private:
1062 * 'C' private: static
1063 * 'D' private: static
1064 * 'E' private: virtual
1065 * 'F' private: virtual
1066 * 'G' private: thunk
1067 * 'H' private: thunk
1068 * 'I' protected:
1069 * 'J' protected:
1070 * 'K' protected: static
1071 * 'L' protected: static
1072 * 'M' protected: virtual
1073 * 'N' protected: virtual
1074 * 'O' protected: thunk
1075 * 'P' protected: thunk
1076 * 'Q' public:
1077 * 'R' public:
1078 * 'S' public: static
1079 * 'T' public: static
1080 * 'U' public: virtual
1081 * 'V' public: virtual
1082 * 'W' public: thunk
1083 * 'X' public: thunk
1084 * 'Y'
1085 * 'Z'
1087 accmem = *sym->current++;
1088 if (accmem < 'A' || accmem > 'Z') goto done;
1090 if (!(sym->flags & UNDNAME_NO_ACCESS_SPECIFIERS))
1092 switch ((accmem - 'A') / 8)
1094 case 0: access = "private: "; break;
1095 case 1: access = "protected: "; break;
1096 case 2: access = "public: "; break;
1099 if (!(sym->flags & UNDNAME_NO_MEMBER_TYPE))
1101 if (accmem <= 'X')
1103 switch ((accmem - 'A') % 8)
1105 case 2: case 3: member_type = "static "; break;
1106 case 4: case 5: member_type = "virtual "; break;
1107 case 6: case 7:
1108 access = str_printf(sym, "[thunk]:%s", access);
1109 member_type = "virtual ";
1110 break;
1115 name = get_class_string(sym, 0);
1117 if ((accmem - 'A') % 8 == 6 || (accmem - '8') % 8 == 7) /* a thunk */
1118 name = str_printf(sym, "%s`adjustor{%s}' ", name, get_number(sym));
1120 if (accmem <= 'X')
1122 if (((accmem - 'A') % 8) != 2 && ((accmem - 'A') % 8) != 3)
1124 /* Implicit 'this' pointer */
1125 /* If there is an implicit this pointer, const modifier follows */
1126 if (!get_modifier(*sym->current, &modifier)) goto done;
1127 sym->current++;
1131 if (!get_calling_convention(*sym->current++, &call_conv, &exported,
1132 sym->flags))
1133 goto done;
1135 str_array_init(&array_pmt);
1137 /* Return type, or @ if 'void' */
1138 if (*sym->current == '@')
1140 ct_ret.left = "void";
1141 ct_ret.right = NULL;
1142 sym->current++;
1144 else
1146 if (!demangle_datatype(sym, &ct_ret, &array_pmt, FALSE))
1147 goto done;
1149 if (sym->flags & UNDNAME_NO_FUNCTION_RETURNS)
1150 ct_ret.left = ct_ret.right = NULL;
1151 if (cast_op)
1153 name = str_printf(sym, "%s%s%s", name, ct_ret.left, ct_ret.right);
1154 ct_ret.left = ct_ret.right = NULL;
1157 mark = sym->stack.num;
1158 if (!(args_str = get_args(sym, &array_pmt, TRUE, '(', ')'))) goto done;
1159 if (sym->flags & UNDNAME_NAME_ONLY) args_str = modifier = NULL;
1160 sym->stack.num = mark;
1162 /* Note: '()' after 'Z' means 'throws', but we don't care here
1163 * Yet!!! FIXME
1165 sym->result = str_printf(sym, "%s%s%s%s%s%s%s%s%s%s%s%s",
1166 access, member_type, ct_ret.left,
1167 (ct_ret.left && !ct_ret.right) ? " " : NULL,
1168 call_conv, call_conv ? " " : NULL, exported,
1169 name, args_str, modifier,
1170 modifier ? " " : NULL, ct_ret.right);
1171 ret = TRUE;
1172 done:
1173 return ret;
1176 /******************************************************************
1177 * handle_template
1178 * Does the final parsing and handling for a name with templates
1180 static BOOL handle_template(struct parsed_symbol* sym)
1182 const char* name;
1183 const char* args;
1185 assert(*sym->current++ == '$');
1186 if (!(name = get_literal_string(sym))) return FALSE;
1187 if (!(args = get_args(sym, NULL, FALSE, '<', '>'))) return FALSE;
1188 sym->result = str_printf(sym, "%s%s", name, args);
1189 return TRUE;
1192 /*******************************************************************
1193 * symbol_demangle
1194 * Demangle a C++ linker symbol
1196 static BOOL symbol_demangle(struct parsed_symbol* sym)
1198 BOOL ret = FALSE;
1199 unsigned do_after = 0;
1200 static CHAR dashed_null[] = "--null--";
1202 /* FIXME seems wrong as name, as it demangles a simple data type */
1203 if (sym->flags & UNDNAME_NO_ARGUMENTS)
1205 struct datatype_t ct;
1207 if (demangle_datatype(sym, &ct, NULL, FALSE))
1209 sym->result = str_printf(sym, "%s%s", ct.left, ct.right);
1210 ret = TRUE;
1212 goto done;
1215 /* MS mangled names always begin with '?' */
1216 if (*sym->current != '?') return FALSE;
1217 str_array_init(&sym->names);
1218 str_array_init(&sym->stack);
1219 sym->current++;
1221 /* Then function name or operator code */
1222 if (*sym->current == '?' && sym->current[1] != '$')
1224 const char* function_name = NULL;
1226 /* C++ operator code (one character, or two if the first is '_') */
1227 switch (*++sym->current)
1229 case '0': do_after = 1; break;
1230 case '1': do_after = 2; break;
1231 case '2': function_name = "operator new"; break;
1232 case '3': function_name = "operator delete"; break;
1233 case '4': function_name = "operator="; break;
1234 case '5': function_name = "operator>>"; break;
1235 case '6': function_name = "operator<<"; break;
1236 case '7': function_name = "operator!"; break;
1237 case '8': function_name = "operator=="; break;
1238 case '9': function_name = "operator!="; break;
1239 case 'A': function_name = "operator[]"; break;
1240 case 'B': function_name = "operator "; do_after = 3; break;
1241 case 'C': function_name = "operator->"; break;
1242 case 'D': function_name = "operator*"; break;
1243 case 'E': function_name = "operator++"; break;
1244 case 'F': function_name = "operator--"; break;
1245 case 'G': function_name = "operator-"; break;
1246 case 'H': function_name = "operator+"; break;
1247 case 'I': function_name = "operator&"; break;
1248 case 'J': function_name = "operator->*"; break;
1249 case 'K': function_name = "operator/"; break;
1250 case 'L': function_name = "operator%"; break;
1251 case 'M': function_name = "operator<"; break;
1252 case 'N': function_name = "operator<="; break;
1253 case 'O': function_name = "operator>"; break;
1254 case 'P': function_name = "operator>="; break;
1255 case 'Q': function_name = "operator,"; break;
1256 case 'R': function_name = "operator()"; break;
1257 case 'S': function_name = "operator~"; break;
1258 case 'T': function_name = "operator^"; break;
1259 case 'U': function_name = "operator|"; break;
1260 case 'V': function_name = "operator&&"; break;
1261 case 'W': function_name = "operator||"; break;
1262 case 'X': function_name = "operator*="; break;
1263 case 'Y': function_name = "operator+="; break;
1264 case 'Z': function_name = "operator-="; break;
1265 case '_':
1266 switch (*++sym->current)
1268 case '0': function_name = "operator/="; break;
1269 case '1': function_name = "operator%="; break;
1270 case '2': function_name = "operator>>="; break;
1271 case '3': function_name = "operator<<="; break;
1272 case '4': function_name = "operator&="; break;
1273 case '5': function_name = "operator|="; break;
1274 case '6': function_name = "operator^="; break;
1275 case '7': function_name = "`vftable'"; break;
1276 case '8': function_name = "`vbtable'"; break;
1277 case '9': function_name = "`vcall'"; break;
1278 case 'A': function_name = "`typeof'"; break;
1279 case 'B': function_name = "`local static guard'"; break;
1280 case 'C': function_name = "`string'"; do_after = 4; break;
1281 case 'D': function_name = "`vbase destructor'"; break;
1282 case 'E': function_name = "`vector deleting destructor'"; break;
1283 case 'F': function_name = "`default constructor closure'"; break;
1284 case 'G': function_name = "`scalar deleting destructor'"; break;
1285 case 'H': function_name = "`vector constructor iterator'"; break;
1286 case 'I': function_name = "`vector destructor iterator'"; break;
1287 case 'J': function_name = "`vector vbase constructor iterator'"; break;
1288 case 'K': function_name = "`virtual displacement map'"; break;
1289 case 'L': function_name = "`eh vector constructor iterator'"; break;
1290 case 'M': function_name = "`eh vector destructor iterator'"; break;
1291 case 'N': function_name = "`eh vector vbase constructor iterator'"; break;
1292 case 'O': function_name = "`copy constructor closure'"; break;
1293 case 'R':
1294 sym->flags |= UNDNAME_NO_FUNCTION_RETURNS;
1295 switch (*++sym->current)
1297 case '0':
1299 struct datatype_t ct;
1300 struct array pmt;
1302 sym->current++;
1303 str_array_init(&pmt);
1304 demangle_datatype(sym, &ct, &pmt, FALSE);
1305 function_name = str_printf(sym, "%s%s `RTTI Type Descriptor'",
1306 ct.left, ct.right);
1307 sym->current--;
1309 break;
1310 case '1':
1312 const char* n1, *n2, *n3, *n4;
1313 sym->current++;
1314 n1 = get_number(sym);
1315 n2 = get_number(sym);
1316 n3 = get_number(sym);
1317 n4 = get_number(sym);
1318 sym->current--;
1319 function_name = str_printf(sym, "`RTTI Base Class Descriptor at (%s,%s,%s,%s)'",
1320 n1, n2, n3, n4);
1322 break;
1323 case '2': function_name = "`RTTI Base Class Array'"; break;
1324 case '3': function_name = "`RTTI Class Hierarchy Descriptor'"; break;
1325 case '4': function_name = "`RTTI Complete Object Locator'"; break;
1326 default:
1327 ERR("Unknown RTTI operator: _R%c\n", *sym->current);
1328 break;
1330 break;
1331 case 'S': function_name = "`local vftable'"; break;
1332 case 'T': function_name = "`local vftable constructor closure'"; break;
1333 case 'U': function_name = "operator new[]"; break;
1334 case 'V': function_name = "operator delete[]"; break;
1335 case 'X': function_name = "`placement delete closure'"; break;
1336 case 'Y': function_name = "`placement delete[] closure'"; break;
1337 default:
1338 ERR("Unknown operator: _%c\n", *sym->current);
1339 return FALSE;
1341 break;
1342 default:
1343 /* FIXME: Other operators */
1344 ERR("Unknown operator: %c\n", *sym->current);
1345 return FALSE;
1347 sym->current++;
1348 switch (do_after)
1350 case 1: case 2:
1351 if (!str_array_push(sym, dashed_null, -1, &sym->stack))
1352 return FALSE;
1353 break;
1354 case 4:
1355 sym->result = (char*)function_name;
1356 ret = TRUE;
1357 goto done;
1358 default:
1359 if (!str_array_push(sym, function_name, -1, &sym->stack))
1360 return FALSE;
1361 break;
1364 else if (*sym->current == '$')
1366 /* Strange construct, it's a name with a template argument list
1367 and that's all. */
1368 sym->current++;
1369 ret = (sym->result = get_template_name(sym)) != NULL;
1370 goto done;
1372 else if (*sym->current == '?' && sym->current[1] == '$')
1373 do_after = 5;
1375 /* Either a class name, or '@' if the symbol is not a class member */
1376 switch (*sym->current)
1378 case '@': sym->current++; break;
1379 case '$': break;
1380 default:
1381 /* Class the function is associated with, terminated by '@@' */
1382 if (!get_class(sym)) goto done;
1383 break;
1386 switch (do_after)
1388 case 0: default: break;
1389 case 1: case 2:
1390 /* it's time to set the member name for ctor & dtor */
1391 if (sym->stack.num <= 1) goto done;
1392 if (do_after == 1)
1393 sym->stack.elts[0] = sym->stack.elts[1];
1394 else
1395 sym->stack.elts[0] = str_printf(sym, "~%s", sym->stack.elts[1]);
1396 /* ctors and dtors don't have return type */
1397 sym->flags |= UNDNAME_NO_FUNCTION_RETURNS;
1398 break;
1399 case 3:
1400 sym->flags &= ~UNDNAME_NO_FUNCTION_RETURNS;
1401 break;
1402 case 5:
1403 sym->names.start = 1;
1404 break;
1407 /* Function/Data type and access level */
1408 if (*sym->current >= '0' && *sym->current <= '9')
1409 ret = handle_data(sym);
1410 else if (*sym->current >= 'A' && *sym->current <= 'Z')
1411 ret = handle_method(sym, do_after == 3);
1412 else if (*sym->current == '$')
1413 ret = handle_template(sym);
1414 else ret = FALSE;
1415 done:
1416 if (ret) assert(sym->result);
1417 else WARN("Failed at %s\n", sym->current);
1419 return ret;
1422 /*********************************************************************
1423 * __unDNameEx (MSVCRT.@)
1425 * Demangle a C++ identifier.
1427 * PARAMS
1428 * buffer [O] If not NULL, the place to put the demangled string
1429 * mangled [I] Mangled name of the function
1430 * buflen [I] Length of buffer
1431 * memget [I] Function to allocate memory with
1432 * memfree [I] Function to free memory with
1433 * unknown [?] Unknown, possibly a call back
1434 * flags [I] Flags determining demangled format
1436 * RETURNS
1437 * Success: A string pointing to the unmangled name, allocated with memget.
1438 * Failure: NULL.
1440 char* CDECL __unDNameEx(char* buffer, const char* mangled, int buflen,
1441 malloc_func_t memget, free_func_t memfree,
1442 void* unknown, unsigned short int flags)
1444 struct parsed_symbol sym;
1445 const char* result;
1447 TRACE("(%p,%s,%d,%p,%p,%p,%x)\n",
1448 buffer, mangled, buflen, memget, memfree, unknown, flags);
1450 /* The flags details is not documented by MS. However, it looks exactly
1451 * like the UNDNAME_ manifest constants from imagehlp.h and dbghelp.h
1452 * So, we copied those (on top of the file)
1454 memset(&sym, 0, sizeof(struct parsed_symbol));
1455 if (flags & UNDNAME_NAME_ONLY)
1456 flags |= UNDNAME_NO_FUNCTION_RETURNS | UNDNAME_NO_ACCESS_SPECIFIERS |
1457 UNDNAME_NO_MEMBER_TYPE | UNDNAME_NO_ALLOCATION_LANGUAGE |
1458 UNDNAME_NO_COMPLEX_TYPE;
1460 sym.flags = flags;
1461 sym.mem_alloc_ptr = memget;
1462 sym.mem_free_ptr = memfree;
1463 sym.current = mangled;
1465 result = symbol_demangle(&sym) ? sym.result : mangled;
1466 if (buffer && buflen)
1468 lstrcpynA( buffer, result, buflen);
1470 else
1472 buffer = memget(strlen(result) + 1);
1473 if (buffer) strcpy(buffer, result);
1476 und_free_all(&sym);
1478 return buffer;
1482 /*********************************************************************
1483 * __unDName (MSVCRT.@)
1485 char* CDECL __unDName(char* buffer, const char* mangled, int buflen,
1486 malloc_func_t memget, free_func_t memfree,
1487 unsigned short int flags)
1489 return __unDNameEx(buffer, mangled, buflen, memget, memfree, NULL, flags);