tools/protoparse.h

   1
   2 struct parsed_proto;
   3
   4 struct parsed_type {
   5         char *name;
   6         unsigned int is_array:1;
   7         unsigned int is_ptr:1;
   8         unsigned int is_struct:1; // split for args
   9 };
  10
  11 struct parsed_proto_arg {
  12         char *reg;
  13         struct parsed_type type;
  14         struct parsed_proto *fptr;
  15         void *datap;
  16 };
  17
  18 struct parsed_proto {
  19         char name[256];
  20         union {
  21                 struct parsed_type ret_type;
  22                 struct parsed_type type;
  23         };
  24         struct parsed_proto_arg arg[16];
  25         int argc;
  26         int argc_stack;
  27         int argc_reg;
  28         unsigned int is_func:1;
  29         unsigned int is_stdcall:1;
  30         unsigned int is_fastcall:1;
  31         unsigned int is_vararg:1;
  32         unsigned int is_fptr:1;
  33         unsigned int is_noreturn:1;
  34         unsigned int is_unresolved:1;
  35         unsigned int has_structarg:1;
  36 };
  37
  38 static const char *hdrfn;
  39 static int hdrfline = 0;
  40
  41 static void pp_copy_arg(struct parsed_proto_arg *d,
  42         const struct parsed_proto_arg *s);
  43
  44 static int b_pp_c_handler(char *proto, const char *fname);
  45
  46 static int do_protostrs(FILE *fhdr, const char *fname)
  47 {
  48         const char *finc_name;
  49         const char *hdrfn_saved;
  50         char protostr[256];
  51         char path[256];
  52         char fname_inc[256];
  53         FILE *finc;
  54         int line = 0;
  55         int ret;
  56         char *p;
  57
  58         hdrfn_saved = hdrfn;
  59         hdrfn = fname;
  60
  61         while (fgets(protostr, sizeof(protostr), fhdr))
  62         {
  63                 line++;
  64                 if (strncmp(protostr, "//#include ", 11) == 0) {
  65                         finc_name = protostr + 11;
  66                         p = strpbrk(finc_name, "\r\n ");
  67                         if (p != NULL)
  68                                 *p = 0;
  69
  70                         path[0] = 0;
  71                         p = strrchr(hdrfn_saved, '/');
  72                         if (p) {
  73                                 memcpy(path, hdrfn_saved,
  74                                         p - hdrfn_saved + 1);
  75                                 path[p - hdrfn_saved + 1] = 0;
  76                         }
  77                         snprintf(fname_inc, sizeof(fname_inc), "%s%s",
  78                                 path, finc_name);
  79                         finc = fopen(fname_inc, "r");
  80                         if (finc == NULL) {
  81                                 printf("%s:%d: can't open '%s'\n",
  82                                         fname_inc, line, finc_name);
  83                                 continue;
  84                         }
  85                         ret = do_protostrs(finc, finc_name);
  86                         fclose(finc);
  87                         if (ret < 0)
  88                                 break;
  89                         continue;
  90                 }
  91                 if (strncmp(sskip(protostr), "//", 2) == 0)
  92                         continue;
  93
  94                 p = protostr + strlen(protostr);
  95                 for (p--; p >= protostr && my_isblank(*p); --p)
  96                         *p = 0;
  97                 if (p < protostr)
  98                         continue;
  99
 100                 hdrfline = line;
 101
 102                 ret = b_pp_c_handler(protostr, hdrfn);
 103                 if (ret < 0)
 104                         break;
 105         }
 106
 107         hdrfn = hdrfn_saved;
 108
 109         if (feof(fhdr))
 110                 return 0;
 111
 112         return -1;
 113 }
 114
 115 static int get_regparm(char *dst, size_t dlen, char *p)
 116 {
 117         int i, o;
 118
 119         if (*p != '<')
 120                 return 0;
 121
 122         for (o = 0, i = 1; o < dlen; i++) {
 123                 if (p[i] == 0)
 124                         return 0;
 125                 if (p[i] == '>')
 126                         break;
 127                 dst[o++] = p[i];
 128         }
 129         dst[o] = 0;
 130         return i + 1;
 131 }
 132
 133 // hmh..
 134 static const char *known_type_mod[] = {
 135         "const",
 136         "signed",
 137         "unsigned",
 138         "struct",
 139         "enum",
 140         "CONST",
 141 };
 142
 143 static const char *known_ptr_types[] = {
 144         "FARPROC",
 145         "HACCEL",
 146         "HANDLE",
 147         "HBITMAP",
 148         "HCURSOR",
 149         "HDC",
 150         "HFONT",
 151         "HGDIOBJ",
 152         "HGLOBAL",
 153         "HICON",
 154         "HINSTANCE",
 155         //"HIMC", // DWORD
 156         "HMODULE",
 157         "HPALETTE",
 158         "HRGN",
 159         "HRSRC",
 160         "HKEY",
 161         "HMENU",
 162         "HWND",
 163         "PCRITICAL_SECTION",
 164         "PDWORD",
 165         "PHKEY",
 166         "PLONG",
 167         "PMEMORY_BASIC_INFORMATION",
 168         "PUINT",
 169         "PVOID",
 170         "PCVOID",
 171         "DLGPROC",
 172         "TIMERPROC",
 173         "WNDENUMPROC",
 174         "va_list",
 175         "__VALIST",
 176 };
 177
 178 static const char *ignored_keywords[] = {
 179         "extern",
 180         "WINBASEAPI",
 181         "WINUSERAPI",
 182         "WINGDIAPI",
 183         "WINADVAPI",
 184 };
 185
 186 // returns ptr to char after type ends
 187 static int typecmp(const char *n, const char *t)
 188 {
 189         for (; *t != 0; n++, t++) {
 190                 while (n[0] == ' ' && (n[1] == ' ' || n[1] == '*'))
 191                         n++;
 192                 while (t[0] == ' ' && (t[1] == ' ' || t[1] == '*'))
 193                         t++;
 194                 if (*n != *t)
 195                         return *n - *t;
 196         }
 197
 198         return 0;
 199 }
 200
 201 static const char *skip_type_mod(const char *n)
 202 {
 203         int len;
 204         int i;
 205
 206         for (i = 0; i < ARRAY_SIZE(known_type_mod); i++) {
 207                 len = strlen(known_type_mod[i]);
 208                 if (strncmp(n, known_type_mod[i], len) != 0)
 209                         continue;
 210                 if (!my_isblank(n[len]))
 211                         continue;
 212
 213                 n += len;
 214                 while (my_isblank(*n))
 215                         n++;
 216                 i = 0;
 217         }
 218
 219         return n;
 220 }
 221
 222 static int check_type(const char *name, struct parsed_type *type)
 223 {
 224         const char *n, *n1;
 225         int ret = -1;
 226         int i;
 227
 228         n = skip_type_mod(name);
 229
 230         for (i = 0; i < ARRAY_SIZE(known_ptr_types); i++) {
 231                 if (typecmp(n, known_ptr_types[i]))
 232                         continue;
 233
 234                 type->is_ptr = 1;
 235                 break;
 236         }
 237
 238         if (n[0] == 'L' && n[1] == 'P' && strncmp(n, "LPARAM", 6))
 239                 type->is_ptr = 1;
 240
 241         // assume single word
 242         while (!my_isblank(*n) && !my_issep(*n))
 243                 n++;
 244
 245         while (1) {
 246                 n1 = n;
 247                 while (my_isblank(*n))
 248                         n++;
 249                 if (*n == '*') {
 250                         type->is_ptr = 1;
 251                         n++;
 252                         continue;
 253                 }
 254                 break;
 255         }
 256
 257         ret = n1 - name;
 258         type->name = strndup(name, ret);
 259         if (IS(type->name, "VOID"))
 260                 memcpy(type->name, "void", 4);
 261
 262         return ret;
 263 }
 264
 265 /* args are always expanded to 32bit */
 266 static const char *map_reg(const char *reg)
 267 {
 268         const char *regs_f[] = { "eax", "ebx", "ecx", "edx", "esi", "edi" };
 269         const char *regs_w[] = { "ax",  "bx",  "cx",  "dx",  "si",  "di" };
 270         const char *regs_b[] = { "al",  "bl",  "cl",  "dl" };
 271         int i;
 272
 273         for (i = 0; i < ARRAY_SIZE(regs_w); i++)
 274                 if (IS(reg, regs_w[i]))
 275                         return regs_f[i];
 276
 277         for (i = 0; i < ARRAY_SIZE(regs_b); i++)
 278                 if (IS(reg, regs_b[i]))
 279                         return regs_f[i];
 280
 281         return reg;
 282 }
 283
 284 static int check_struct_arg(struct parsed_proto_arg *arg)
 285 {
 286         if (IS(arg->type.name, "POINT"))
 287                 return 2 - 1;
 288
 289         return 0;
 290 }
 291
 292 static int parse_protostr(char *protostr, struct parsed_proto *pp)
 293 {
 294         struct parsed_proto_arg *arg;
 295         char regparm[16];
 296         char buf[256];
 297         char cconv[32];
 298         int xarg = 0;
 299         char *p, *p1;
 300         int i, l;
 301         int ret;
 302
 303         p = sskip(protostr);
 304         if (p[0] == '/' && p[1] == '/') {
 305                 printf("%s:%d: commented out?\n", hdrfn, hdrfline);
 306                 p = sskip(p + 2);
 307         }
 308
 309         // strip unneeded stuff
 310         for (p1 = p; p1[0] != 0 && p1[1] != 0; p1++) {
 311                 if ((p1[0] == '/' && p1[1] == '*')
 312                  || (p1[0] == '*' && p1[1] == '/'))
 313                         p1[0] = p1[1] = ' ';
 314         }
 315
 316         if (!strncmp(p, "DECLSPEC_NORETURN ", 18)) {
 317                 pp->is_noreturn = 1;
 318                 p = sskip(p + 18);
 319         }
 320
 321         for (i = 0; i < ARRAY_SIZE(ignored_keywords); i++) {
 322                 l = strlen(ignored_keywords[i]);
 323                 if (!strncmp(p, ignored_keywords[i], l) && my_isblank(p[l]))
 324                         p = sskip(p + l + 1);
 325         }
 326
 327         ret = check_type(p, &pp->ret_type);
 328         if (ret <= 0) {
 329                 printf("%s:%d:%zd: unhandled return in '%s'\n",
 330                         hdrfn, hdrfline, (p - protostr) + 1, protostr);
 331                 return -1;
 332         }
 333         p = sskip(p + ret);
 334
 335         if (!strncmp(p, "noreturn ", 9)) {
 336                 pp->is_noreturn = 1;
 337                 p = sskip(p + 9);
 338         }
 339
 340         if (!strchr(p, ')')) {
 341                 p = next_idt(buf, sizeof(buf), p);
 342                 p = sskip(p);
 343                 if (buf[0] == 0) {
 344                         printf("%s:%d:%zd: var name missing\n",
 345                                 hdrfn, hdrfline, (p - protostr) + 1);
 346                         return -1;
 347                 }
 348                 strcpy(pp->name, buf);
 349
 350                 p1 = strchr(p, ']');
 351                 if (p1 != NULL) {
 352                         p = p1 + 1;
 353                         pp->ret_type.is_array = 1;
 354                 }
 355                 return p - protostr;
 356         }
 357
 358         pp->is_func = 1;
 359
 360         if (*p == '(') {
 361                 pp->is_fptr = 1;
 362                 p = sskip(p + 1);
 363         }
 364
 365         p = next_word(cconv, sizeof(cconv), p);
 366         p = sskip(p);
 367         if (cconv[0] == 0) {
 368                 printf("%s:%d:%zd: cconv missing\n",
 369                         hdrfn, hdrfline, (p - protostr) + 1);
 370                 return -1;
 371         }
 372         if      (IS(cconv, "__cdecl"))
 373                 pp->is_stdcall = 0;
 374         else if (IS(cconv, "__stdcall"))
 375                 pp->is_stdcall = 1;
 376         else if (IS(cconv, "__fastcall")) {
 377                 pp->is_fastcall = 1;
 378                 pp->is_stdcall = 1; // sort of..
 379         }
 380         else if (IS(cconv, "__thiscall"))
 381                 pp->is_stdcall = 1;
 382         else if (IS(cconv, "__userpurge"))
 383                 pp->is_stdcall = 1; // IDA
 384         else if (IS(cconv, "__usercall"))
 385                 pp->is_stdcall = 0; // IDA
 386         else if (IS(cconv, "WINAPI"))
 387                 pp->is_stdcall = 1;
 388         else {
 389                 printf("%s:%d:%zd: unhandled cconv: '%s'\n",
 390                         hdrfn, hdrfline, (p - protostr) + 1, cconv);
 391                 return -1;
 392         }
 393
 394         if (pp->is_fptr) {
 395                 if (*p != '*') {
 396                         printf("%s:%d:%zd: '*' expected\n",
 397                                 hdrfn, hdrfline, (p - protostr) + 1);
 398                         return -1;
 399                 }
 400                 p++;
 401                 // XXX: skipping extra asterisks, for now
 402                 while (*p == '*')
 403                         p++;
 404                 p = sskip(p);
 405         }
 406
 407         p = next_idt(buf, sizeof(buf), p);
 408         p = sskip(p);
 409         if (buf[0] == 0) {
 410                 //printf("%s:%d:%zd: func name missing\n",
 411                 //      hdrfn, hdrfline, (p - protostr) + 1);
 412                 //return -1;
 413         }
 414         strcpy(pp->name, buf);
 415
 416         ret = get_regparm(regparm, sizeof(regparm), p);
 417         if (ret > 0) {
 418                 if (!IS(regparm, "eax") && !IS(regparm, "ax")
 419                  && !IS(regparm, "al") && !IS(regparm, "edx:eax"))
 420                 {
 421                         printf("%s:%d:%zd: bad regparm: %s\n",
 422                                 hdrfn, hdrfline, (p - protostr) + 1, regparm);
 423                         return -1;
 424                 }
 425                 p += ret;
 426                 p = sskip(p);
 427         }
 428
 429         if (pp->is_fptr) {
 430                 if (*p == '[') {
 431                         // not really ret_type is array, but ohwell
 432                         pp->ret_type.is_array = 1;
 433                         p = strchr(p + 1, ']');
 434                         if (p == NULL) {
 435                                 printf("%s:%d:%zd: ']' expected\n",
 436                                  hdrfn, hdrfline, (p - protostr) + 1);
 437                                 return -1;
 438                         }
 439                         p = sskip(p + 1);
 440                 }
 441                 if (*p != ')') {
 442                         printf("%s:%d:%zd: ')' expected\n",
 443                                 hdrfn, hdrfline, (p - protostr) + 1);
 444                         return -1;
 445                 }
 446                 p = sskip(p + 1);
 447         }
 448
 449         if (*p != '(') {
 450                 printf("%s:%d:%zd: '(' expected, got '%c'\n",
 451                                 hdrfn, hdrfline, (p - protostr) + 1, *p);
 452                 return -1;
 453         }
 454         p++;
 455
 456         // check for x(void)
 457         p = sskip(p);
 458         if ((!strncmp(p, "void", 4) || !strncmp(p, "VOID", 4))
 459            && *sskip(p + 4) == ')')
 460                 p += 4;
 461
 462         while (1) {
 463                 p = sskip(p);
 464                 if (*p == ')') {
 465                         p++;
 466                         break;
 467                 }
 468                 if (xarg > 0) {
 469                         if (*p != ',') {
 470                                 printf("%s:%d:%zd: ',' expected\n",
 471                                  hdrfn, hdrfline, (p - protostr) + 1);
 472                                 return -1;
 473                         }
 474                         p = sskip(p + 1);
 475                 }
 476
 477                 if (!strncmp(p, "...", 3)) {
 478                         pp->is_vararg = 1;
 479                         p = sskip(p + 3);
 480                         if (*p == ')') {
 481                                 p++;
 482                                 break;
 483                         }
 484                         printf("%s:%d:%zd: ')' expected\n",
 485                                 hdrfn, hdrfline, (p - protostr) + 1);
 486                         return -1;
 487                 }
 488
 489                 arg = &pp->arg[xarg];
 490                 xarg++;
 491
 492                 p1 = p;
 493                 ret = check_type(p, &arg->type);
 494                 if (ret <= 0) {
 495                         printf("%s:%d:%zd: unhandled type for arg%d\n",
 496                                 hdrfn, hdrfline, (p - protostr) + 1, xarg);
 497                         return -1;
 498                 }
 499                 p = sskip(p + ret);
 500
 501                 if (*p == '(') {
 502                         // func ptr
 503                         arg->fptr = calloc(1, sizeof(*arg->fptr));
 504                         ret = parse_protostr(p1, arg->fptr);
 505                         if (ret < 0) {
 506                                 printf("%s:%d:%zd: funcarg parse failed\n",
 507                                         hdrfn, hdrfline, p1 - protostr);
 508                                 return -1;
 509                         }
 510                         // we'll treat it as void * for non-calls
 511                         arg->type.name = strdup("void *");
 512                         arg->type.is_ptr = 1;
 513
 514                         p = p1 + ret;
 515                 }
 516
 517                 p = next_idt(buf, sizeof(buf), p);
 518                 p = sskip(p);
 519 #if 0
 520                 if (buf[0] == 0) {
 521                         printf("%s:%d:%zd: idt missing for arg%d\n",
 522                                 hdrfn, hdrfline, (p - protostr) + 1, xarg);
 523                         return -1;
 524                 }
 525 #endif
 526                 arg->reg = NULL;
 527
 528                 ret = get_regparm(regparm, sizeof(regparm), p);
 529                 if (ret > 0) {
 530                         p += ret;
 531                         p = sskip(p);
 532
 533                         arg->reg = strdup(map_reg(regparm));
 534                 }
 535
 536                 ret = check_struct_arg(arg);
 537                 if (ret > 0) {
 538                         pp->has_structarg = 1;
 539                         arg->type.is_struct = 1;
 540                         free(arg->type.name);
 541                         arg->type.name = strdup("int");
 542                         for (l = 0; l < ret; l++) {
 543                                 pp_copy_arg(&pp->arg[xarg], arg);
 544                                 xarg++;
 545                         }
 546                 }
 547         }
 548
 549         if (xarg > 0 && (IS(cconv, "__fastcall") || IS(cconv, "__thiscall"))) {
 550                 if (pp->arg[0].reg != NULL) {
 551                         printf("%s:%d: %s with arg1 spec %s?\n",
 552                                 hdrfn, hdrfline, cconv, pp->arg[0].reg);
 553                 }
 554                 pp->arg[0].reg = strdup("ecx");
 555         }
 556
 557         if (xarg > 1 && IS(cconv, "__fastcall")) {
 558                 if (pp->arg[1].reg != NULL) {
 559                         printf("%s:%d: %s with arg2 spec %s?\n",
 560                                 hdrfn, hdrfline, cconv, pp->arg[1].reg);
 561                 }
 562                 pp->arg[1].reg = strdup("edx");
 563         }
 564
 565         pp->argc = xarg;
 566
 567         for (i = 0; i < pp->argc; i++) {
 568                 if (pp->arg[i].reg == NULL)
 569                         pp->argc_stack++;
 570                 else
 571                         pp->argc_reg++;
 572         }
 573
 574         if (pp->argc == 1 && pp->arg[0].reg != NULL
 575             && IS(pp->arg[0].reg, "ecx"))
 576         {
 577                 pp->is_fastcall = 1;
 578         }
 579         else if (pp->argc_reg == 2
 580           && pp->arg[0].reg != NULL && IS(pp->arg[0].reg, "ecx")
 581           && pp->arg[1].reg != NULL && IS(pp->arg[1].reg, "edx"))
 582         {
 583                 pp->is_fastcall = 1;
 584         }
 585
 586         if (pp->is_vararg && (pp->is_stdcall || pp->is_fastcall)) {
 587                 printf("%s:%d: vararg %s?\n", hdrfn, hdrfline, cconv);
 588                 return -1;
 589         }
 590
 591         return p - protostr;
 592 }
 593
 594 static int pp_name_cmp(const void *p1, const void *p2)
 595 {
 596         const struct parsed_proto *pp1 = p1, *pp2 = p2;
 597         return strcmp(pp1->name, pp2->name);
 598 }
 599
 600 static struct parsed_proto *pp_cache;
 601 static int pp_cache_size;
 602 static int pp_cache_alloc;
 603
 604 static int b_pp_c_handler(char *proto, const char *fname)
 605 {
 606         int ret;
 607
 608         if (pp_cache_size >= pp_cache_alloc) {
 609                 pp_cache_alloc = pp_cache_alloc * 2 + 64;
 610                 pp_cache = realloc(pp_cache, pp_cache_alloc
 611                                 * sizeof(pp_cache[0]));
 612                 my_assert_not(pp_cache, NULL);
 613                 memset(pp_cache + pp_cache_size, 0,
 614                         (pp_cache_alloc - pp_cache_size)
 615                          * sizeof(pp_cache[0]));
 616         }
 617
 618         ret = parse_protostr(proto, &pp_cache[pp_cache_size]);
 619         if (ret < 0)
 620                 return -1;
 621
 622         pp_cache_size++;
 623         return 0;
 624 }
 625
 626 static void build_pp_cache(FILE *fhdr)
 627 {
 628         long pos;
 629         int ret;
 630
 631         pos = ftell(fhdr);
 632         rewind(fhdr);
 633
 634         ret = do_protostrs(fhdr, hdrfn);
 635         if (ret < 0)
 636                 exit(1);
 637
 638         qsort(pp_cache, pp_cache_size, sizeof(pp_cache[0]), pp_name_cmp);
 639         fseek(fhdr, pos, SEEK_SET);
 640 }
 641
 642 static const struct parsed_proto *proto_parse(FILE *fhdr, const char *sym,
 643         int quiet)
 644 {
 645         const struct parsed_proto *pp_ret;
 646         struct parsed_proto pp_search;
 647
 648         if (pp_cache == NULL)
 649                 build_pp_cache(fhdr);
 650
 651         if (sym[0] == '_') // && strncmp(fname, "stdc", 4) == 0)
 652                 sym++;
 653
 654         strcpy(pp_search.name, sym);
 655         pp_ret = bsearch(&pp_search, pp_cache, pp_cache_size,
 656                         sizeof(pp_cache[0]), pp_name_cmp);
 657         if (pp_ret == NULL && !quiet)
 658                 printf("%s: sym '%s' is missing\n", hdrfn, sym);
 659
 660         return pp_ret;
 661 }
 662
 663 static void pp_copy_arg(struct parsed_proto_arg *d,
 664         const struct parsed_proto_arg *s)
 665 {
 666         memcpy(d, s, sizeof(*d));
 667
 668         if (s->reg != NULL) {
 669                 d->reg = strdup(s->reg);
 670                 my_assert_not(d->reg, NULL);
 671         }
 672         if (s->type.name != NULL) {
 673                 d->type.name = strdup(s->type.name);
 674                 my_assert_not(d->type.name, NULL);
 675         }
 676         if (s->fptr != NULL) {
 677                 d->fptr = malloc(sizeof(*d->fptr));
 678                 my_assert_not(d->fptr, NULL);
 679                 memcpy(d->fptr, s->fptr, sizeof(*d->fptr));
 680         }
 681 }
 682
 683 struct parsed_proto *proto_clone(const struct parsed_proto *pp_c)
 684 {
 685         struct parsed_proto *pp;
 686         int i;
 687
 688         pp = malloc(sizeof(*pp));
 689         my_assert_not(pp, NULL);
 690         memcpy(pp, pp_c, sizeof(*pp)); // lazy..
 691
 692         // do the actual deep copy..
 693         for (i = 0; i < pp_c->argc; i++)
 694                 pp_copy_arg(&pp->arg[i], &pp_c->arg[i]);
 695         if (pp_c->ret_type.name != NULL)
 696                 pp->ret_type.name = strdup(pp_c->ret_type.name);
 697
 698         return pp;
 699 }
 700
 701 static inline void proto_release(struct parsed_proto *pp)
 702 {
 703         int i;
 704
 705         for (i = 0; i < pp->argc; i++) {
 706                 if (pp->arg[i].reg != NULL)
 707                         free(pp->arg[i].reg);
 708                 if (pp->arg[i].type.name != NULL)
 709                         free(pp->arg[i].type.name);
 710                 if (pp->arg[i].fptr != NULL)
 711                         free(pp->arg[i].fptr);
 712         }
 713         if (pp->ret_type.name != NULL)
 714                 free(pp->ret_type.name);
 715         free(pp);
 716 }