dlls/msvcrt/undname.c

   1 /*
   2  *  Demangle VC++ symbols into C function prototypes
   3  *
   4  *  Copyright 2000 Jon Griffiths
   5  *            2004 Eric Pouech
   6  *
   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.
  11  *
  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.
  16  *
  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
  20  */
  21
  22 #include "config.h"
  23 #include "wine/port.h"
  24
  25 #include <assert.h>
  26 #include <stdio.h>
  27 #include "msvcrt.h"
  28
  29 #include "wine/debug.h"
  30
  31 WINE_DEFAULT_DEBUG_CHANNEL(msvcrt);
  32
  33 /* TODO:
  34  * - document a bit (grammar + functions)
  35  * - back-port this new code into tools/winedump/msmangle.c
  36  */
  37
  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)
  56
  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 "
  64  *
  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
  72  *
  73  */
  74
  75 #define MAX_ARRAY_ELTS  32
  76 struct array
  77 {
  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];
  82 };
  83
  84 /* Structure holding a parsed symbol */
  85 struct parsed_symbol
  86 {
  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 */
  90
  91     const char*         current;        /* pointer in input (mangled) string */
  92     char*               result;         /* demangled string */
  93
  94     struct array        names;          /* array of names for back reference */
  95     struct array        stack;          /* stack of parsed strings */
  96
  97     void*               alloc_list;     /* linked list of allocated blocks */
  98     unsigned            avail_in_first; /* number of available bytes in head block */
  99 };
 100
 101 /* Type for parsing mangled types */
 102 struct datatype_t
 103 {
 104     const char*         left;
 105     const char*         right;
 106 };
 107
 108 /******************************************************************
 109  *              und_alloc
 110  *
 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).
 114  */
 115 static void*    und_alloc(struct parsed_symbol* sym, size_t len)
 116 {
 117     void*       ptr;
 118
 119 #define BLOCK_SIZE      1024
 120 #define AVAIL_SIZE      (1024 - sizeof(void*))
 121
 122     if (len > AVAIL_SIZE)
 123     {
 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*);
 131     }
 132     else
 133     {
 134         if (len > sym->avail_in_first)
 135         {
 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;
 142         }
 143         /* grab memory from head block */
 144         ptr = (char*)sym->alloc_list + BLOCK_SIZE - sym->avail_in_first;
 145         sym->avail_in_first -= len;
 146     }
 147     return ptr;
 148 #undef BLOCK_SIZE
 149 #undef AVAIL_SIZE
 150 }
 151
 152 /******************************************************************
 153  *              und_free
 154  * Frees all the blocks in the list of large blocks allocated by
 155  * und_alloc.
 156  */
 157 static void und_free_all(struct parsed_symbol* sym)
 158 {
 159     void*       next;
 160
 161     while (sym->alloc_list)
 162     {
 163         next = *(void**)sym->alloc_list;
 164         if(sym->mem_free_ptr) sym->mem_free_ptr(sym->alloc_list);
 165         sym->alloc_list = next;
 166     }
 167     sym->avail_in_first = 0;
 168 }
 169
 170 /******************************************************************
 171  *              str_array_init
 172  * Initialises an array of strings
 173  */
 174 static void str_array_init(struct array* a)
 175 {
 176     a->start = a->num = a->max = 0;
 177 }
 178
 179 /******************************************************************
 180  *              str_array_push
 181  * Adding a new string to an array
 182  */
 183 static void str_array_push(struct parsed_symbol* sym, const char* ptr, int len,
 184                            struct array* a)
 185 {
 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;
 195     {
 196         int i;
 197         char c;
 198
 199         for (i = a->max - 1; i >= 0; i--)
 200         {
 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]);
 205         }
 206     }
 207 }
 208
 209 /******************************************************************
 210  *              str_array_get_ref
 211  * Extracts a reference from an existing array (doing proper type
 212  * checking)
 213  */
 214 static char* str_array_get_ref(struct array* cref, unsigned idx)
 215 {
 216     assert(cref);
 217     if (cref->start + idx >= cref->max)
 218     {
 219         WARN("Out of bounds: %p %d + %d >= %d\n",
 220               cref, cref->start, idx, cref->max);
 221         return NULL;
 222     }
 223     TRACE("Returning %p[%d] => %s\n",
 224           cref, idx, cref->elts[cref->start + idx]);
 225     return cref->elts[cref->start + idx];
 226 }
 227
 228 /******************************************************************
 229  *              str_printf
 230  * Helper for printf type of command (only %s and %c are implemented)
 231  * while dynamically allocating the buffer
 232  */
 233 static char* str_printf(struct parsed_symbol* sym, const char* format, ...)
 234 {
 235     va_list     args;
 236     size_t      len = 1, i, sz;
 237     char*       tmp;
 238     char*       p;
 239     char*       t;
 240
 241     va_start(args, format);
 242     for (i = 0; format[i]; i++)
 243     {
 244         if (format[i] == '%')
 245         {
 246             switch (format[++i])
 247             {
 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;
 252             }
 253         }
 254         else len++;
 255     }
 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++)
 260     {
 261         if (format[i] == '%')
 262         {
 263             switch (format[++i])
 264             {
 265             case 's':
 266                 t = va_arg(args, char*);
 267                 if (t)
 268                 {
 269                     sz = strlen(t);
 270                     memcpy(p, t, sz);
 271                     p += sz;
 272                 }
 273                 break;
 274             case 'c':
 275                 *p++ = (char)va_arg(args, int);
 276                 break;
 277             default: i--; /* fall thru */
 278             case '%': *p++ = '%'; break;
 279             }
 280         }
 281         else *p++ = format[i];
 282     }
 283     va_end(args);
 284     *p = '\0';
 285     return tmp;
 286 }
 287
 288 /* forward declaration */
 289 static BOOL demangle_datatype(struct parsed_symbol* sym, struct datatype_t* ct,
 290                               struct array* pmt, BOOL in_args);
 291
 292 /******************************************************************
 293  *              get_args
 294  * Parses a list of function/method arguments, creates a string corresponding
 295  * to the arguments' list.
 296  */
 297 static char* get_args(struct parsed_symbol* sym, struct array* pmt_ref, BOOL z_term,
 298                       char open_char, char close_char)
 299
 300 {
 301     struct datatype_t   ct;
 302     struct array        arg_collect;
 303     char*               args_str = NULL;
 304     int                 i;
 305
 306     str_array_init(&arg_collect);
 307
 308     /* Now come the function arguments */
 309     while (*sym->current)
 310     {
 311         /* Decode each data type and append it to the argument list */
 312         if (*sym->current == '@')
 313         {
 314             sym->current++;
 315             break;
 316         }
 317         if (!demangle_datatype(sym, &ct, pmt_ref, TRUE))
 318             return NULL;
 319         /* 'void' terminates an argument list */
 320         if (!strcmp(ct.left, "void"))
 321         {
 322             if (!z_term && *sym->current == '@') sym->current++;
 323             break;
 324         }
 325         str_array_push(sym, str_printf(sym, "%s%s", ct.left, ct.right), -1,
 326                        &arg_collect);
 327         if (!strcmp(ct.left, "...")) break;
 328     }
 329     /* Functions are always terminated by 'Z'. If we made it this far and
 330      * don't find it, we have incorrectly identified a data type.
 331      */
 332     if (z_term && *sym->current++ != 'Z') return NULL;
 333
 334     if (arg_collect.num == 0 ||
 335         (arg_collect.num == 1 && !strcmp(arg_collect.elts[0], "void")))
 336         return str_printf(sym, "%cvoid%c", open_char, close_char);
 337     for (i = 1; i < arg_collect.num; i++)
 338     {
 339         args_str = str_printf(sym, "%s,%s", args_str, arg_collect.elts[i]);
 340     }
 341
 342     if (close_char == '>' && args_str && args_str[strlen(args_str) - 1] == '>')
 343         args_str = str_printf(sym, "%c%s%s %c",
 344                               open_char, arg_collect.elts[0], args_str, close_char);
 345     else
 346         args_str = str_printf(sym, "%c%s%s%c",
 347                               open_char, arg_collect.elts[0], args_str, close_char);
 348
 349     return args_str;
 350 }
 351
 352 /******************************************************************
 353  *              get_modifier
 354  * Parses the type modifier. Always returns a static string
 355  */
 356 static BOOL get_modifier(char ch, const char** ret)
 357 {
 358     switch (ch)
 359     {
 360     case 'A': *ret = NULL; break;
 361     case 'B': *ret = "const"; break;
 362     case 'C': *ret = "volatile"; break;
 363     case 'D': *ret = "const volatile"; break;
 364     default: return FALSE;
 365     }
 366     return TRUE;
 367 }
 368
 369 static BOOL get_modified_type(struct datatype_t *ct, struct parsed_symbol* sym,
 370                               struct array *pmt_ref, char modif)
 371 {
 372     const char* modifier;
 373     const char* str_modif;
 374
 375     switch (modif)
 376     {
 377     case 'A': str_modif = " &"; break;
 378     case 'B': str_modif = " & volatile"; break;
 379     case 'P': str_modif = " *"; break;
 380     case 'Q': str_modif = " * const"; break;
 381     case 'R': str_modif = " * volatile"; break;
 382     case 'S': str_modif = " * const volatile"; break;
 383     case '?': str_modif = ""; break;
 384     default: return FALSE;
 385     }
 386
 387     if (get_modifier(*sym->current++, &modifier))
 388     {
 389         unsigned            mark = sym->stack.num;
 390         struct datatype_t   sub_ct;
 391
 392         /* Recurse to get the referred-to type */
 393         if (!demangle_datatype(sym, &sub_ct, pmt_ref, FALSE))
 394             return FALSE;
 395         if (modifier)
 396             ct->left = str_printf(sym, "%s %s%s", sub_ct.left, modifier, str_modif );
 397         else
 398         {
 399             /* don't insert a space between duplicate '*' */
 400             if (str_modif[0] && str_modif[1] == '*' && sub_ct.left[strlen(sub_ct.left)-1] == '*')
 401                 str_modif++;
 402             ct->left = str_printf(sym, "%s%s", sub_ct.left, str_modif );
 403         }
 404         ct->right = sub_ct.right;
 405         sym->stack.num = mark;
 406     }
 407     return TRUE;
 408 }
 409
 410 /******************************************************************
 411  *             get_literal_string
 412  * Gets the literal name from the current position in the mangled
 413  * symbol to the first '@' character. It pushes the parsed name to
 414  * the symbol names stack and returns a pointer to it or NULL in
 415  * case of an error.
 416  */
 417 static char* get_literal_string(struct parsed_symbol* sym)
 418 {
 419     const char *ptr = sym->current;
 420
 421     do {
 422         if (!((*sym->current >= 'A' && *sym->current <= 'Z') ||
 423               (*sym->current >= 'a' && *sym->current <= 'z') ||
 424               (*sym->current >= '0' && *sym->current <= '9') ||
 425               *sym->current == '_' || *sym->current == '$')) {
 426             TRACE("Failed at '%c' in %s\n", *sym->current, ptr);
 427             return NULL;
 428         }
 429     } while (*++sym->current != '@');
 430     sym->current++;
 431     str_array_push(sym, ptr, sym->current - 1 - ptr, &sym->names);
 432
 433     return str_array_get_ref(&sym->names, sym->names.num - sym->names.start - 1);
 434 }
 435
 436 /******************************************************************
 437  *              get_template_name
 438  * Parses a name with a template argument list and returns it as
 439  * a string.
 440  * In a template argument list the back reference to the names
 441  * table is separately created. '0' points to the class component
 442  * name with the template arguments.  We use the same stack array
 443  * to hold the names but save/restore the stack state before/after
 444  * parsing the template argument list.
 445  */
 446 static char* get_template_name(struct parsed_symbol* sym)
 447 {
 448     char *name, *args;
 449     unsigned num_mark = sym->names.num;
 450     unsigned start_mark = sym->names.start;
 451     unsigned stack_mark = sym->stack.num;
 452     struct array array_pmt;
 453
 454     sym->names.start = sym->names.num;
 455     if (!(name = get_literal_string(sym)))
 456         return FALSE;
 457     str_array_init(&array_pmt);
 458     args = get_args(sym, &array_pmt, FALSE, '<', '>');
 459     if (args != NULL)
 460         name = str_printf(sym, "%s%s", name, args);
 461     sym->names.num = num_mark;
 462     sym->names.start = start_mark;
 463     sym->stack.num = stack_mark;
 464     return name;
 465 }
 466
 467 /******************************************************************
 468  *              get_class
 469  * Parses class as a list of parent-classes, terminated by '@' and stores the
 470  * result in 'a' array. Each parent-classes, as well as the inner element
 471  * (either field/method name or class name), are represented in the mangled
 472  * name by a literal name ([a-zA-Z0-9_]+ terminated by '@') or a back reference
 473  * ([0-9]) or a name with template arguments ('?$' literal name followed by the
 474  * template argument list). The class name components appear in the reverse
 475  * order in the mangled name, e.g aaa@bbb@ccc@@ will be demangled to
 476  * ccc::bbb::aaa
 477  * For each of this class name componets a string will be allocated in the
 478  * array.
 479  */
 480 static BOOL get_class(struct parsed_symbol* sym)
 481 {
 482     const char* name = NULL;
 483
 484     while (*sym->current != '@')
 485     {
 486         switch (*sym->current)
 487         {
 488         case '\0': return FALSE;
 489
 490         case '0': case '1': case '2': case '3':
 491         case '4': case '5': case '6': case '7':
 492         case '8': case '9':
 493             name = str_array_get_ref(&sym->names, *sym->current++ - '0');
 494             break;
 495         case '?':
 496             if (*++sym->current == '$')
 497             {
 498                 sym->current++;
 499                 name = get_template_name(sym);
 500                 str_array_push(sym, name, -1, &sym->names);
 501             }
 502             break;
 503         default:
 504             name = get_literal_string(sym);
 505             break;
 506         }
 507         if (!name)
 508             return FALSE;
 509         str_array_push(sym, name, -1, &sym->stack);
 510     }
 511     sym->current++;
 512     return TRUE;
 513 }
 514
 515 /******************************************************************
 516  *              get_class_string
 517  * From an array collected by get_class in sym->stack, constructs the
 518  * corresponding (allocated) string
 519  */
 520 static char* get_class_string(struct parsed_symbol* sym, int start)
 521 {
 522     int         i;
 523     size_t      len, sz;
 524     char*       ret;
 525     struct array *a = &sym->stack;
 526
 527     for (len = 0, i = start; i < a->num; i++)
 528     {
 529         assert(a->elts[i]);
 530         len += 2 + strlen(a->elts[i]);
 531     }
 532     if (!(ret = und_alloc(sym, len - 1))) return NULL;
 533     for (len = 0, i = a->num - 1; i >= start; i--)
 534     {
 535         sz = strlen(a->elts[i]);
 536         memcpy(ret + len, a->elts[i], sz);
 537         len += sz;
 538         if (i > start)
 539         {
 540             ret[len++] = ':';
 541             ret[len++] = ':';
 542         }
 543     }
 544     ret[len] = '\0';
 545     return ret;
 546 }
 547
 548 /******************************************************************
 549  *            get_class_name
 550  * Wrapper around get_class and get_class_string.
 551  */
 552 static char* get_class_name(struct parsed_symbol* sym)
 553 {
 554     unsigned    mark = sym->stack.num;
 555     char*       s = NULL;
 556
 557     if (get_class(sym))
 558         s = get_class_string(sym, mark);
 559     sym->stack.num = mark;
 560     return s;
 561 }
 562
 563 /******************************************************************
 564  *              get_calling_convention
 565  * Returns a static string corresponding to the calling convention described
 566  * by char 'ch'. Sets export to TRUE iff the calling convention is exported.
 567  */
 568 static BOOL get_calling_convention(char ch, const char** call_conv,
 569                                    const char** exported, unsigned flags)
 570 {
 571     *call_conv = *exported = NULL;
 572
 573     if (!(flags & (UNDNAME_NO_MS_KEYWORDS | UNDNAME_NO_ALLOCATION_LANGUAGE)))
 574     {
 575         if (flags & UNDNAME_NO_LEADING_UNDERSCORES)
 576         {
 577             if (((ch - 'A') % 2) == 1) *exported = "dll_export ";
 578             switch (ch)
 579             {
 580             case 'A': case 'B': *call_conv = "cdecl"; break;
 581             case 'C': case 'D': *call_conv = "pascal"; break;
 582             case 'E': case 'F': *call_conv = "thiscall"; break;
 583             case 'G': case 'H': *call_conv = "stdcall"; break;
 584             case 'I': case 'J': *call_conv = "fastcall"; break;
 585             case 'K': break;
 586             default: ERR("Unknown calling convention %c\n", ch); return FALSE;
 587             }
 588         }
 589         else
 590         {
 591             if (((ch - 'A') % 2) == 1) *exported = "__dll_export ";
 592             switch (ch)
 593             {
 594             case 'A': case 'B': *call_conv = "__cdecl"; break;
 595             case 'C': case 'D': *call_conv = "__pascal"; break;
 596             case 'E': case 'F': *call_conv = "__thiscall"; break;
 597             case 'G': case 'H': *call_conv = "__stdcall"; break;
 598             case 'I': case 'J': *call_conv = "__fastcall"; break;
 599             case 'K': break;
 600             default: ERR("Unknown calling convention %c\n", ch); return FALSE;
 601             }
 602         }
 603     }
 604     return TRUE;
 605 }
 606
 607 /*******************************************************************
 608  *         get_simple_type
 609  * Return a string containing an allocated string for a simple data type
 610  */
 611 static const char* get_simple_type(char c)
 612 {
 613     const char* type_string;
 614
 615     switch (c)
 616     {
 617     case 'C': type_string = "signed char"; break;
 618     case 'D': type_string = "char"; break;
 619     case 'E': type_string = "unsigned char"; break;
 620     case 'F': type_string = "short"; break;
 621     case 'G': type_string = "unsigned short"; break;
 622     case 'H': type_string = "int"; break;
 623     case 'I': type_string = "unsigned int"; break;
 624     case 'J': type_string = "long"; break;
 625     case 'K': type_string = "unsigned long"; break;
 626     case 'M': type_string = "float"; break;
 627     case 'N': type_string = "double"; break;
 628     case 'O': type_string = "long double"; break;
 629     case 'X': type_string = "void"; break;
 630     case 'Z': type_string = "..."; break;
 631     default:  type_string = NULL; break;
 632     }
 633     return type_string;
 634 }
 635
 636 /*******************************************************************
 637  *         get_extented_type
 638  * Return a string containing an allocated string for a simple data type
 639  */
 640 static const char* get_extended_type(char c)
 641 {
 642     const char* type_string;
 643
 644     switch (c)
 645     {
 646     case 'D': type_string = "__int8"; break;
 647     case 'E': type_string = "unsigned __int8"; break;
 648     case 'F': type_string = "__int16"; break;
 649     case 'G': type_string = "unsigned __int16"; break;
 650     case 'H': type_string = "__int32"; break;
 651     case 'I': type_string = "unsigned __int32"; break;
 652     case 'J': type_string = "__int64"; break;
 653     case 'K': type_string = "unsigned __int64"; break;
 654     case 'L': type_string = "__int128"; break;
 655     case 'M': type_string = "unsigned __int128"; break;
 656     case 'N': type_string = "bool"; break;
 657     case 'W': type_string = "wchar_t"; break;
 658     default:  type_string = NULL; break;
 659     }
 660     return type_string;
 661 }
 662
 663 /*******************************************************************
 664  *         demangle_datatype
 665  *
 666  * Attempt to demangle a C++ data type, which may be datatype.
 667  * a datatype type is made up of a number of simple types. e.g:
 668  * char** = (pointer to (pointer to (char)))
 669  */
 670 static BOOL demangle_datatype(struct parsed_symbol* sym, struct datatype_t* ct,
 671                               struct array* pmt_ref, BOOL in_args)
 672 {
 673     char                dt;
 674     BOOL                add_pmt = TRUE;
 675     int                 num_args=0;
 676
 677     assert(ct);
 678     ct->left = ct->right = NULL;
 679
 680     switch (dt = *sym->current++)
 681     {
 682     case '_':
 683         /* MS type: __int8,__int16 etc */
 684         ct->left = get_extended_type(*sym->current++);
 685         break;
 686     case 'C': case 'D': case 'E': case 'F': case 'G':
 687     case 'H': case 'I': case 'J': case 'K': case 'M':
 688     case 'N': case 'O': case 'X': case 'Z':
 689         /* Simple data types */
 690         ct->left = get_simple_type(dt);
 691         add_pmt = FALSE;
 692         break;
 693     case 'T': /* union */
 694     case 'U': /* struct */
 695     case 'V': /* class */
 696         /* Class/struct/union */
 697         {
 698             const char* struct_name = NULL;
 699             const char* type_name = NULL;
 700
 701             if (!(struct_name = get_class_name(sym)))
 702                 goto done;
 703             if (!(sym->flags & UNDNAME_NO_COMPLEX_TYPE))
 704             {
 705                 switch (dt)
 706                 {
 707                 case 'T': type_name = "union ";  break;
 708                 case 'U': type_name = "struct "; break;
 709                 case 'V': type_name = "class ";  break;
 710                 }
 711             }
 712             ct->left = str_printf(sym, "%s%s", type_name, struct_name);
 713         }
 714         break;
 715     case '?':
 716         /* not all the time is seems */
 717         if (!get_modified_type(ct, sym, pmt_ref, '?')) goto done;
 718         break;
 719     case 'A': /* reference */
 720     case 'B': /* volatile reference */
 721         if (!get_modified_type(ct, sym, pmt_ref, dt)) goto done;
 722         break;
 723     case 'Q': /* const pointer */
 724     case 'R': /* volatile pointer */
 725     case 'S': /* const volatile pointer */
 726         if (!get_modified_type(ct, sym, pmt_ref, in_args ? dt : 'P')) goto done;
 727         break;
 728     case 'P': /* Pointer */
 729         if (isdigit(*sym->current))
 730         {
 731             /* FIXME: P6 = Function pointer, others who knows.. */
 732             if (*sym->current++ == '6')
 733             {
 734                 char*                   args = NULL;
 735                 const char*             call_conv;
 736                 const char*             exported;
 737                 struct datatype_t       sub_ct;
 738                 unsigned                mark = sym->stack.num;
 739
 740                 if (!get_calling_convention(*sym->current++,
 741                                             &call_conv, &exported,
 742                                             sym->flags & ~UNDNAME_NO_ALLOCATION_LANGUAGE) ||
 743                     !demangle_datatype(sym, &sub_ct, pmt_ref, FALSE))
 744                     goto done;
 745
 746                 args = get_args(sym, pmt_ref, TRUE, '(', ')');
 747                 if (!args) goto done;
 748                 sym->stack.num = mark;
 749
 750                 ct->left  = str_printf(sym, "%s%s (%s*",
 751                                        sub_ct.left, sub_ct.right, call_conv);
 752                 ct->right = str_printf(sym, ")%s", args);
 753             }
 754             else goto done;
 755         }
 756         else if (!get_modified_type(ct, sym, pmt_ref, 'P')) goto done;
 757         break;
 758     case 'W':
 759         if (*sym->current == '4')
 760         {
 761             char*               enum_name;
 762             sym->current++;
 763             if (!(enum_name = get_class_name(sym)))
 764                 goto done;
 765             if (sym->flags & UNDNAME_NO_COMPLEX_TYPE)
 766                 ct->left = enum_name;
 767             else
 768                 ct->left = str_printf(sym, "enum %s", enum_name);
 769         }
 770         else goto done;
 771         break;
 772     case '0': case '1': case '2': case '3': case '4':
 773     case '5': case '6': case '7': case '8': case '9':
 774         /* Referring back to previously parsed type */
 775         /* left and right are pushed as two separate strings */
 776         ct->left = str_array_get_ref(pmt_ref, (dt - '0') * 2);
 777         ct->right = str_array_get_ref(pmt_ref, (dt - '0') * 2 + 1);
 778         if (!ct->left) goto done;
 779         add_pmt = FALSE;
 780         break;
 781     case '$':
 782         if (sym->current[0] != '0') goto done;
 783         if (sym->current[1] >= '0' && sym->current[1] <= '9')
 784         {
 785             char*       ptr;
 786             ptr = und_alloc(sym, 2);
 787             ptr[0] = sym->current[1] + 1;
 788             ptr[1] = 0;
 789             ct->left = ptr;
 790             sym->current += 2;
 791         }
 792         else if (sym->current[1] >= 'A' && sym->current[1] <= 'P')
 793         {
 794             while (sym->current[1] >= 'A' && sym->current[1] <= 'P')
 795             {
 796                 num_args *= 16;
 797                 num_args += sym->current[1] - 'A';
 798                 sym->current += 1;
 799             }
 800             if(sym->current[1] == '@')
 801             {
 802                 char *ptr;
 803                 ptr = und_alloc(sym, 17);
 804                 sprintf(ptr,"%d",num_args);
 805                 ct->left = ptr;
 806                 sym->current += 1;
 807             }
 808         }
 809         else goto done;
 810         break;
 811     default :
 812         ERR("Unknown type %c\n", dt);
 813         break;
 814     }
 815     if (add_pmt && pmt_ref && in_args)
 816     {
 817         /* left and right are pushed as two separate strings */
 818         str_array_push(sym, ct->left ? ct->left : "", -1, pmt_ref);
 819         str_array_push(sym, ct->right ? ct->right : "", -1, pmt_ref);
 820     }
 821 done:
 822
 823     return ct->left != NULL;
 824 }
 825
 826 /******************************************************************
 827  *              handle_data
 828  * Does the final parsing and handling for a variable or a field in
 829  * a class.
 830  */
 831 static BOOL handle_data(struct parsed_symbol* sym)
 832 {
 833     const char*         access = NULL;
 834     const char*         member_type = NULL;
 835     const char*         modifier = NULL;
 836     struct datatype_t   ct;
 837     char*               name = NULL;
 838     BOOL                ret = FALSE;
 839     char                dt;
 840
 841     /* 0 private static
 842      * 1 protected static
 843      * 2 public static
 844      * 3 private non-static
 845      * 4 protected non-static
 846      * 5 public non-static
 847      * 6 ?? static
 848      * 7 ?? static
 849      */
 850
 851     if (!(sym->flags & UNDNAME_NO_ACCESS_SPECIFIERS))
 852     {
 853         /* we only print the access for static members */
 854         switch (*sym->current)
 855         {
 856         case '0': access = "private: "; break;
 857         case '1': access = "protected: "; break;
 858         case '2': access = "public: "; break;
 859         }
 860     }
 861
 862     if (!(sym->flags & UNDNAME_NO_MEMBER_TYPE))
 863     {
 864         if (*sym->current >= '0' && *sym->current <= '2')
 865             member_type = "static ";
 866     }
 867
 868     name = get_class_string(sym, 0);
 869
 870     switch (dt = *sym->current++)
 871     {
 872     case '0': case '1': case '2':
 873     case '3': case '4': case '5':
 874         {
 875             unsigned mark = sym->stack.num;
 876             struct array pmt;
 877
 878             str_array_init(&pmt);
 879
 880             if (!demangle_datatype(sym, &ct, &pmt, FALSE)) goto done;
 881             if (!get_modifier(*sym->current++, &modifier)) goto done;
 882             sym->stack.num = mark;
 883         }
 884         break;
 885     case '6' : /* compiler generated static */
 886     case '7' : /* compiler generated static */
 887         ct.left = ct.right = NULL;
 888         if (!get_modifier(*sym->current++, &modifier)) goto done;
 889         if (*sym->current != '@')
 890         {
 891             char*       cls = NULL;
 892
 893             if (!(cls = get_class_name(sym)))
 894                 goto done;
 895             ct.right = str_printf(sym, "{for `%s'}", cls);
 896         }
 897         break;
 898     default: goto done;
 899     }
 900     if (sym->flags & UNDNAME_NAME_ONLY) ct.left = ct.right = modifier = NULL;
 901     sym->result = str_printf(sym, "%s%s%s%s%s%s%s%s", access,
 902                              member_type, ct.left,
 903                              modifier && ct.left ? " " : NULL, modifier,
 904                              modifier || ct.left ? " " : NULL, name, ct.right);
 905     ret = TRUE;
 906 done:
 907     return ret;
 908 }
 909
 910 /******************************************************************
 911  *              handle_method
 912  * Does the final parsing and handling for a function or a method in
 913  * a class.
 914  */
 915 static BOOL handle_method(struct parsed_symbol* sym, BOOL cast_op)
 916 {
 917     const char*         access = NULL;
 918     const char*         member_type = NULL;
 919     struct datatype_t   ct_ret;
 920     const char*         call_conv;
 921     const char*         modifier = NULL;
 922     const char*         exported;
 923     const char*         args_str = NULL;
 924     const char*         name = NULL;
 925     BOOL                ret = FALSE;
 926     unsigned            mark;
 927     struct array        array_pmt;
 928
 929     /* FIXME: why 2 possible letters for each option?
 930      * 'A' private:
 931      * 'B' private:
 932      * 'C' private: static
 933      * 'D' private: static
 934      * 'E' private: virtual
 935      * 'F' private: virtual
 936      * 'G' private: thunk
 937      * 'H' private: thunk
 938      * 'I' protected:
 939      * 'J' protected:
 940      * 'K' protected: static
 941      * 'L' protected: static
 942      * 'M' protected: virtual
 943      * 'N' protected: virtual
 944      * 'O' protected: thunk
 945      * 'P' protected: thunk
 946      * 'Q' public:
 947      * 'R' public:
 948      * 'S' public: static
 949      * 'T' public: static
 950      * 'U' public: virtual
 951      * 'V' public: virtual
 952      * 'W' public: thunk
 953      * 'X' public: thunk
 954      * 'Y'
 955      * 'Z'
 956      */
 957
 958     if (!(sym->flags & UNDNAME_NO_ACCESS_SPECIFIERS))
 959     {
 960         switch ((*sym->current - 'A') / 8)
 961         {
 962         case 0: access = "private: "; break;
 963         case 1: access = "protected: "; break;
 964         case 2: access = "public: "; break;
 965         }
 966     }
 967     if (!(sym->flags & UNDNAME_NO_MEMBER_TYPE))
 968     {
 969         if (*sym->current >= 'A' && *sym->current <= 'X')
 970         {
 971             switch ((*sym->current - 'A') % 8)
 972             {
 973             case 2: case 3: member_type = "static "; break;
 974             case 4: case 5: member_type = "virtual "; break;
 975             case 6: case 7: member_type = "thunk "; break;
 976             }
 977         }
 978     }
 979
 980     if (*sym->current >= 'A' && *sym->current <= 'X')
 981     {
 982         if (!((*sym->current - 'A') & 2))
 983         {
 984             /* Implicit 'this' pointer */
 985             /* If there is an implicit this pointer, const modifier follows */
 986             if (!get_modifier(*++sym->current, &modifier)) goto done;
 987         }
 988     }
 989     else if (*sym->current < 'A' || *sym->current > 'Z') goto done;
 990     sym->current++;
 991
 992     name = get_class_string(sym, 0);
 993
 994     if (!get_calling_convention(*sym->current++, &call_conv, &exported,
 995                                 sym->flags))
 996         goto done;
 997
 998     str_array_init(&array_pmt);
 999
1000     /* Return type, or @ if 'void' */
1001     if (*sym->current == '@')
1002     {
1003         ct_ret.left = "void";
1004         ct_ret.right = NULL;
1005         sym->current++;
1006     }
1007     else
1008     {
1009         if (!demangle_datatype(sym, &ct_ret, &array_pmt, FALSE))
1010             goto done;
1011     }
1012     if (sym->flags & UNDNAME_NO_FUNCTION_RETURNS)
1013         ct_ret.left = ct_ret.right = NULL;
1014     if (cast_op)
1015     {
1016         name = str_printf(sym, "%s%s%s", name, ct_ret.left, ct_ret.right);
1017         ct_ret.left = ct_ret.right = NULL;
1018     }
1019
1020     mark = sym->stack.num;
1021     if (!(args_str = get_args(sym, &array_pmt, TRUE, '(', ')'))) goto done;
1022     if (sym->flags & UNDNAME_NAME_ONLY) args_str = modifier = NULL;
1023     sym->stack.num = mark;
1024
1025     /* Note: '()' after 'Z' means 'throws', but we don't care here
1026      * Yet!!! FIXME
1027      */
1028     sym->result = str_printf(sym, "%s%s%s%s%s%s%s%s%s%s%s%s",
1029                              access, member_type, ct_ret.left,
1030                              (ct_ret.left && !ct_ret.right) ? " " : NULL,
1031                              call_conv, call_conv ? " " : NULL, exported,
1032                              name, args_str, modifier,
1033                              modifier ? " " : NULL, ct_ret.right);
1034     ret = TRUE;
1035 done:
1036     return ret;
1037 }
1038
1039 /*******************************************************************
1040  *         symbol_demangle
1041  * Demangle a C++ linker symbol
1042  */
1043 static BOOL symbol_demangle(struct parsed_symbol* sym)
1044 {
1045     BOOL                ret = FALSE;
1046     unsigned            do_after = 0;
1047     static CHAR         dashed_null[] = "--null--";
1048
1049     /* FIXME seems wrong as name, as it demangles a simple data type */
1050     if (sym->flags & UNDNAME_NO_ARGUMENTS)
1051     {
1052         struct datatype_t   ct;
1053
1054         if (demangle_datatype(sym, &ct, NULL, FALSE))
1055         {
1056             sym->result = str_printf(sym, "%s%s", ct.left, ct.right);
1057             ret = TRUE;
1058         }
1059         goto done;
1060     }
1061
1062     /* MS mangled names always begin with '?' */
1063     if (*sym->current != '?') return FALSE;
1064     str_array_init(&sym->names);
1065     str_array_init(&sym->stack);
1066     sym->current++;
1067
1068     /* Then function name or operator code */
1069     if (*sym->current == '?' && sym->current[1] != '$')
1070     {
1071         const char* function_name = NULL;
1072
1073         /* C++ operator code (one character, or two if the first is '_') */
1074         switch (*++sym->current)
1075         {
1076         case '0': do_after = 1; break;
1077         case '1': do_after = 2; break;
1078         case '2': function_name = "operator new"; break;
1079         case '3': function_name = "operator delete"; break;
1080         case '4': function_name = "operator="; break;
1081         case '5': function_name = "operator>>"; break;
1082         case '6': function_name = "operator<<"; break;
1083         case '7': function_name = "operator!"; break;
1084         case '8': function_name = "operator=="; break;
1085         case '9': function_name = "operator!="; break;
1086         case 'A': function_name = "operator[]"; break;
1087         case 'B': function_name = "operator "; do_after = 3; break;
1088         case 'C': function_name = "operator->"; break;
1089         case 'D': function_name = "operator*"; break;
1090         case 'E': function_name = "operator++"; break;
1091         case 'F': function_name = "operator--"; break;
1092         case 'G': function_name = "operator-"; break;
1093         case 'H': function_name = "operator+"; break;
1094         case 'I': function_name = "operator&"; break;
1095         case 'J': function_name = "operator->*"; break;
1096         case 'K': function_name = "operator/"; break;
1097         case 'L': function_name = "operator%"; break;
1098         case 'M': function_name = "operator<"; break;
1099         case 'N': function_name = "operator<="; break;
1100         case 'O': function_name = "operator>"; break;
1101         case 'P': function_name = "operator>="; break;
1102         case 'Q': function_name = "operator,"; break;
1103         case 'R': function_name = "operator()"; break;
1104         case 'S': function_name = "operator~"; break;
1105         case 'T': function_name = "operator^"; break;
1106         case 'U': function_name = "operator|"; break;
1107         case 'V': function_name = "operator&&"; break;
1108         case 'W': function_name = "operator||"; break;
1109         case 'X': function_name = "operator*="; break;
1110         case 'Y': function_name = "operator+="; break;
1111         case 'Z': function_name = "operator-="; break;
1112         case '_':
1113             switch (*++sym->current)
1114             {
1115             case '0': function_name = "operator/="; break;
1116             case '1': function_name = "operator%="; break;
1117             case '2': function_name = "operator>>="; break;
1118             case '3': function_name = "operator<<="; break;
1119             case '4': function_name = "operator&="; break;
1120             case '5': function_name = "operator|="; break;
1121             case '6': function_name = "operator^="; break;
1122             case '7': function_name = "`vftable'"; break;
1123             case '8': function_name = "`vbtable'"; break;
1124             case '9': function_name = "`vcall'"; break;
1125             case 'A': function_name = "`typeof'"; break;
1126             case 'B': function_name = "`local static guard'"; break;
1127             case 'C': function_name = "`string'"; do_after = 4; break;
1128             case 'D': function_name = "`vbase destructor'"; break;
1129             case 'E': function_name = "`vector deleting destructor'"; break;
1130             case 'F': function_name = "`default constructor closure'"; break;
1131             case 'G': function_name = "`scalar deleting destructor'"; break;
1132             case 'H': function_name = "`vector constructor iterator'"; break;
1133             case 'I': function_name = "`vector destructor iterator'"; break;
1134             case 'J': function_name = "`vector vbase constructor iterator'"; break;
1135             case 'K': function_name = "`virtual displacement map'"; break;
1136             case 'L': function_name = "`eh vector constructor iterator'"; break;
1137             case 'M': function_name = "`eh vector destructor iterator'"; break;
1138             case 'N': function_name = "`eh vector vbase constructor iterator'"; break;
1139             case 'O': function_name = "`copy constructor closure'"; break;
1140             case 'S': function_name = "`local vftable'"; break;
1141             case 'T': function_name = "`local vftable constructor closure'"; break;
1142             case 'U': function_name = "operator new[]"; break;
1143             case 'V': function_name = "operator delete[]"; break;
1144             case 'X': function_name = "`placement delete closure'"; break;
1145             case 'Y': function_name = "`placement delete[] closure'"; break;
1146             default:
1147                 ERR("Unknown operator: _%c\n", *sym->current);
1148                 return FALSE;
1149             }
1150             break;
1151         default:
1152             /* FIXME: Other operators */
1153             ERR("Unknown operator: %c\n", *sym->current);
1154             return FALSE;
1155         }
1156         sym->current++;
1157         switch (do_after)
1158         {
1159         case 1: case 2:
1160             sym->stack.num = sym->stack.max = 1;
1161             sym->stack.elts[0] = dashed_null;
1162             break;
1163         case 4:
1164             sym->result = (char*)function_name;
1165             ret = TRUE;
1166             goto done;
1167         default:
1168             str_array_push(sym, function_name, -1, &sym->stack);
1169             break;
1170         }
1171         sym->stack.start = 1;
1172     }
1173     else if (*sym->current == '$')
1174     {
1175         /* Strange construct, it's a name with a template argument list
1176            and that's all. */
1177         sym->current++;
1178         sym->result = get_template_name(sym);
1179         ret = TRUE;
1180         goto done;
1181     }
1182
1183     /* Either a class name, or '@' if the symbol is not a class member */
1184     if (*sym->current != '@')
1185     {
1186         /* Class the function is associated with, terminated by '@@' */
1187         if (!get_class(sym)) goto done;
1188     }
1189     else sym->current++;
1190
1191     switch (do_after)
1192     {
1193     case 0: default: break;
1194     case 1: case 2:
1195         /* it's time to set the member name for ctor & dtor */
1196         if (sym->stack.num <= 1) goto done;
1197         if (do_after == 1)
1198             sym->stack.elts[0] = sym->stack.elts[1];
1199         else
1200             sym->stack.elts[0] = str_printf(sym, "~%s", sym->stack.elts[1]);
1201         /* ctors and dtors don't have return type */
1202         sym->flags |= UNDNAME_NO_FUNCTION_RETURNS;
1203         break;
1204     case 3:
1205         sym->flags &= ~UNDNAME_NO_FUNCTION_RETURNS;
1206         break;
1207     }
1208
1209     /* Function/Data type and access level */
1210     if (*sym->current >= '0' && *sym->current <= '7')
1211         ret = handle_data(sym);
1212     else if (*sym->current >= 'A' && *sym->current <= 'Z')
1213         ret = handle_method(sym, do_after == 3);
1214     else ret = FALSE;
1215 done:
1216     if (ret) assert(sym->result);
1217     else WARN("Failed at %s\n", sym->current);
1218
1219     return ret;
1220 }
1221
1222 /*********************************************************************
1223  *              __unDNameEx (MSVCRT.@)
1224  *
1225  * Demangle a C++ identifier.
1226  *
1227  * PARAMS
1228  *  buffer   [O] If not NULL, the place to put the demangled string
1229  *  mangled  [I] Mangled name of the function
1230  *  buflen   [I] Length of buffer
1231  *  memget   [I] Function to allocate memory with
1232  *  memfree  [I] Function to free memory with
1233  *  unknown  [?] Unknown, possibly a call back
1234  *  flags    [I] Flags determining demangled format
1235  *
1236  * RETURNS
1237  *  Success: A string pointing to the unmangled name, allocated with memget.
1238  *  Failure: NULL.
1239  */
1240 char* CDECL __unDNameEx(char* buffer, const char* mangled, int buflen,
1241                         malloc_func_t memget, free_func_t memfree,
1242                         void* unknown, unsigned short int flags)
1243 {
1244     struct parsed_symbol        sym;
1245     const char*                 result;
1246
1247     TRACE("(%p,%s,%d,%p,%p,%p,%x)\n",
1248           buffer, mangled, buflen, memget, memfree, unknown, flags);
1249
1250     /* The flags details is not documented by MS. However, it looks exactly
1251      * like the UNDNAME_ manifest constants from imagehlp.h and dbghelp.h
1252      * So, we copied those (on top of the file)
1253      */
1254     memset(&sym, 0, sizeof(struct parsed_symbol));
1255     if (flags & UNDNAME_NAME_ONLY)
1256         flags |= UNDNAME_NO_FUNCTION_RETURNS | UNDNAME_NO_ACCESS_SPECIFIERS |
1257             UNDNAME_NO_MEMBER_TYPE | UNDNAME_NO_ALLOCATION_LANGUAGE |
1258             UNDNAME_NO_COMPLEX_TYPE;
1259
1260     sym.flags         = flags;
1261     sym.mem_alloc_ptr = memget;
1262     sym.mem_free_ptr  = memfree;
1263     sym.current       = mangled;
1264
1265     result = symbol_demangle(&sym) ? sym.result : mangled;
1266     if (buffer && buflen)
1267     {
1268         lstrcpynA( buffer, result, buflen);
1269     }
1270     else
1271     {
1272         buffer = memget(strlen(result) + 1);
1273         if (buffer) strcpy(buffer, result);
1274     }
1275
1276     und_free_all(&sym);
1277
1278     return buffer;
1279 }
1280
1281
1282 /*********************************************************************
1283  *              __unDName (MSVCRT.@)
1284  */
1285 char* CDECL __unDName(char* buffer, const char* mangled, int buflen,
1286                       malloc_func_t memget, free_func_t memfree,
1287                       unsigned short int flags)
1288 {
1289     return __unDNameEx(buffer, mangled, buflen, memget, memfree, NULL, flags);
1290 }