nsz Git - cparser/blob - format_check.c

   1 /*
   2  * This file is part of cparser.
   3  * Copyright (C) 2007-2008 Matthias Braun <matze@braunis.de>
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public License
   7  * as published by the Free Software Foundation; either version 2
   8  * of the License, or (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  18  * 02111-1307, USA.
  19  */
  20 #include <ctype.h>
  21 #include <wctype.h>
  22
  23 #include "format_check.h"
  24 #include "symbol_t.h"
  25 #include "ast_t.h"
  26 #include "diagnostic.h"
  27 #include "types.h"
  28 #include "type_t.h"
  29 #include "warning.h"
  30 #include "lang_features.h"
  31
  32 typedef enum format_flag_t {
  33         FMT_FLAG_NONE  = 0,
  34         FMT_FLAG_HASH  = 1U << 0,
  35         FMT_FLAG_ZERO  = 1U << 1,
  36         FMT_FLAG_MINUS = 1U << 2,
  37         FMT_FLAG_SPACE = 1U << 3,
  38         FMT_FLAG_PLUS  = 1U << 4,
  39         FMT_FLAG_TICK  = 1U << 5
  40 } format_flag_t;
  41
  42 typedef unsigned format_flags_t;
  43
  44 typedef enum format_length_modifier_t {
  45         FMT_MOD_NONE,
  46         FMT_MOD_L,
  47         FMT_MOD_hh,
  48         FMT_MOD_h,
  49         FMT_MOD_l,
  50         FMT_MOD_ll,
  51         FMT_MOD_j,
  52         FMT_MOD_t,
  53         FMT_MOD_z,
  54         FMT_MOD_q,
  55         /* only in microsoft mode */
  56         FMT_MOD_w,
  57         FMT_MOD_I,
  58         FMT_MOD_I32,
  59         FMT_MOD_I64
  60 } format_length_modifier_t;
  61
  62 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  63 {
  64         static const char* const names[] = {
  65                 [FMT_MOD_NONE] = "",
  66                 [FMT_MOD_L]    = "L",
  67                 [FMT_MOD_hh]   = "hh",
  68                 [FMT_MOD_h]    = "h",
  69                 [FMT_MOD_l]    = "l",
  70                 [FMT_MOD_ll]   = "ll",
  71                 [FMT_MOD_j]    = "j",
  72                 [FMT_MOD_t]    = "t",
  73                 [FMT_MOD_z]    = "z",
  74                 [FMT_MOD_q]    = "q",
  75                 /* only in microsoft mode */
  76                 [FMT_MOD_w]    = "w",
  77                 [FMT_MOD_I]    = "I",
  78                 [FMT_MOD_I32]  = "I32",
  79                 [FMT_MOD_I64]  = "I64"
  80         };
  81         assert(mod < sizeof(names) / sizeof(*names));
  82         return names[mod];
  83 }
  84
  85 static void warn_invalid_length_modifier(const source_position_t pos,
  86                                          const format_length_modifier_t mod,
  87                                          const wchar_rep_t conversion)
  88 {
  89         warningf(pos,
  90                 "invalid length modifier '%s' for conversion specifier '%%%c'",
  91                 get_length_modifier_name(mod), conversion
  92         );
  93 }
  94
  95 typedef struct vchar_t vchar_t;
  96 struct vchar_t {
  97         const void *string;   /**< the string */
  98         size_t     position;  /**< current position */
  99         size_t     size;      /**< size of the string */
 100
 101         /** return the first character of the string and setthe position to 0. */
 102         unsigned (*first)(vchar_t *self);
 103         /** return the next character of the string */
 104         unsigned (*next)(vchar_t *self);
 105         /** return non_zero if the given character is a digit */
 106         int (*is_digit)(unsigned vchar);
 107 };
 108
 109 static unsigned string_first(vchar_t *self) {
 110         self->position = 0;
 111         const string_t *string = self->string;
 112         return string->begin[0];
 113 }
 114
 115 static unsigned string_next(vchar_t *self) {
 116         ++self->position;
 117         const string_t *string = self->string;
 118         return string->begin[self->position];
 119 }
 120
 121 static int string_isdigit(unsigned vchar) {
 122         return isdigit(vchar);
 123 }
 124
 125 static unsigned wstring_first(vchar_t *self) {
 126         self->position = 0;
 127         const wide_string_t *wstring = self->string;
 128         return wstring->begin[0];
 129 }
 130
 131 static unsigned wstring_next(vchar_t *self) {
 132         ++self->position;
 133         const wide_string_t *wstring = self->string;
 134         return wstring->begin[self->position];
 135 }
 136
 137 static int wstring_isdigit(unsigned vchar) {
 138         return iswdigit(vchar);
 139 }
 140
 141 static bool atend(vchar_t *self) {
 142         return self->position + 1 == self->size;
 143 }
 144
 145 static void check_format_arguments(const call_argument_t *const fmt_arg, const call_argument_t* arg)
 146 {
 147         const expression_t *fmt_expr = fmt_arg->expression;
 148         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 149                 fmt_expr = fmt_expr->unary.value;
 150         }
 151
 152         vchar_t vchar;
 153         if (fmt_expr->kind == EXPR_WIDE_STRING_LITERAL) {
 154                 vchar.string   = &fmt_expr->wide_string.value;
 155                 vchar.size     = fmt_expr->wide_string.value.size;
 156                 vchar.first    = wstring_first;
 157                 vchar.next     = wstring_next;
 158                 vchar.is_digit = wstring_isdigit;
 159         } else if (fmt_expr->kind == EXPR_STRING_LITERAL) {
 160                 vchar.string   = &fmt_expr->string.value;
 161                 vchar.size     = fmt_expr->string.value.size;
 162                 vchar.first    = string_first;
 163                 vchar.next     = string_next;
 164                 vchar.is_digit = string_isdigit;
 165         } else {
 166                 return;
 167         }
 168         const source_position_t    pos     = fmt_expr->base.source_position;
 169         unsigned fmt = vchar.first(&vchar);
 170         for (; fmt != '\0'; fmt = vchar.next(&vchar)) {
 171                 if (fmt != '%')
 172                         continue;
 173                 fmt = vchar.next(&vchar);
 174
 175                 if (fmt == '%')
 176                         continue;
 177
 178                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 179                 if (fmt == '0') {
 180                         fmt = vchar.next(&vchar);
 181                         fmt_flags |= FMT_FLAG_ZERO;
 182                 }
 183
 184                 /* argument selector or minimum field width */
 185                 if (vchar.is_digit(fmt)) {
 186                         do {
 187                                 fmt = vchar.next(&vchar);
 188                         } while (vchar.is_digit(fmt));
 189
 190                         /* digit string was ... */
 191                         if (fmt == '$') {
 192                                 /* ... argument selector */
 193                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 194                                 /* TODO implement */
 195                                 return;
 196                         }
 197                         /* ... minimum field width */
 198                 } else {
 199                         /* flags */
 200                         for (;;) {
 201                                 format_flags_t flag;
 202                                 switch (fmt) {
 203                                         case '#':  flag = FMT_FLAG_HASH;  break;
 204                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 205                                         case '-':  flag = FMT_FLAG_MINUS; break;
 206                                         case '\'': flag = FMT_FLAG_TICK;  break;
 207
 208                                         case ' ':
 209                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 210                                                         warningf(pos, "' ' is overridden by prior '+' in conversion specification");
 211                                                 }
 212                                                 flag = FMT_FLAG_SPACE;
 213                                                 break;
 214
 215                                         case '+':
 216                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 217                                                         warningf(pos, "'+' overrides prior ' ' in conversion specification");
 218                                                 }
 219                                                 flag = FMT_FLAG_PLUS;
 220                                                 break;
 221
 222                                         default: goto break_fmt_flags;
 223                                 }
 224                                 if (fmt_flags & flag) {
 225                                         warningf(pos, "repeated flag '%c' in conversion specification", (char)fmt);
 226                                 }
 227                                 fmt_flags |= flag;
 228                                 fmt = vchar.next(&vchar);
 229                         }
 230 break_fmt_flags:
 231
 232                         /* minimum field width */
 233                         if (fmt == '*') {
 234                                 if (arg == NULL) {
 235                                         warningf(pos, "missing argument for '*' field width in conversion specification");
 236                                         return;
 237                                 }
 238                                 const type_t *const arg_type = arg->expression->base.type;
 239                                 if (arg_type != type_int) {
 240                                         warningf(pos, "argument for '*' field width in conversion specification is not an 'int', but an '%T'", arg_type);
 241                                 }
 242                                 arg = arg->next;
 243                         } else {
 244                                 while (vchar.is_digit(fmt)) {
 245                                         fmt = vchar.next(&vchar);
 246                                 }
 247                         }
 248                 }
 249
 250                 /* precision */
 251                 if (fmt == '.') {
 252                         fmt = vchar.next(&vchar);
 253                         if (fmt == '*') {
 254                                 if (arg == NULL) {
 255                                         warningf(pos, "missing argument for '*' precision in conversion specification");
 256                                         return;
 257                                 }
 258                                 const type_t *const arg_type = arg->expression->base.type;
 259                                 if (arg_type != type_int) {
 260                                         warningf(pos, "argument for '*' precision in conversion specification is not an 'int', but an '%T'", arg_type);
 261                                 }
 262                                 arg = arg->next;
 263                         } else {
 264                                 /* digit string may be omitted */
 265                                 while (vchar.is_digit(fmt)) {
 266                                         fmt = vchar.next(&vchar);
 267                                 }
 268                         }
 269                 }
 270
 271                 /* length modifier */
 272                 format_length_modifier_t fmt_mod;
 273                 switch (fmt) {
 274                         case 'h':
 275                                 fmt = vchar.next(&vchar);
 276                                 if (fmt == 'h') {
 277                                         fmt = vchar.next(&vchar);
 278                                         fmt_mod = FMT_MOD_hh;
 279                                 } else {
 280                                         fmt_mod = FMT_MOD_h;
 281                                 }
 282                                 break;
 283
 284                         case 'l':
 285                                 fmt = vchar.next(&vchar);
 286                                 if (fmt == 'l') {
 287                                         fmt = vchar.next(&vchar);
 288                                         fmt_mod = FMT_MOD_ll;
 289                                 } else {
 290                                         fmt_mod = FMT_MOD_l;
 291                                 }
 292                                 break;
 293
 294                         case 'L': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_L;    break;
 295                         case 'j': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_j;    break;
 296                         case 't': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_t;    break;
 297                         case 'z': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_z;    break;
 298                         case 'q': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_q;    break;
 299                         /* microsoft mode */
 300                         case 'w':
 301                                 if (c_mode & _MS) {
 302                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_w;
 303                                 } else {
 304                                         fmt_mod = FMT_MOD_NONE;
 305                                 }
 306                                 break;
 307                         case 'I':
 308                                 if (c_mode & _MS) {
 309                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_I;
 310                                         if (fmt == '3') {
 311                                                 fmt = vchar.next(&vchar);
 312                                                 if (fmt == '2') {
 313                                                         fmt = vchar.next(&vchar);
 314                                                         fmt_mod = FMT_MOD_I32;
 315                                                 } else {
 316                                                         /* rewind */
 317                                                         --vchar.position;
 318                                                 }
 319                                         } else if (fmt == '6') {
 320                                                 fmt = vchar.next(&vchar);
 321                                                 if (fmt == '4') {
 322                                                         fmt = vchar.next(&vchar);
 323                                                         fmt_mod = FMT_MOD_I64;
 324                                                 } else {
 325                                                         /* rewind */
 326                                                         --vchar.position;
 327                                                 }
 328                                         }
 329                                 } else {
 330                                         fmt_mod = FMT_MOD_NONE;
 331                                 }
 332                                 break;
 333                         default:
 334                                 fmt_mod = FMT_MOD_NONE;
 335                                 break;
 336                 }
 337
 338                 if (fmt == '\0') {
 339                         warningf(pos, "dangling %% in format string");
 340                         break;
 341                 }
 342
 343                 const type_t      *expected_type;
 344                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 345                 format_flags_t     allowed_flags;
 346                 switch (fmt) {
 347                         case 'd':
 348                         case 'i':
 349                                 switch (fmt_mod) {
 350                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 351                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 352                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 353                                         case FMT_MOD_l:    expected_type = type_long;      break;
 354                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 355                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 356                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 357                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 358                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t; break;
 359                                         case FMT_MOD_I32:  expected_type = type_int32;     break;
 360                                         case FMT_MOD_I64:  expected_type = type_int64;     break;
 361
 362                                         default:
 363                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 364                                                 goto next_arg;
 365                                 }
 366                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 367                                 break;
 368
 369                         case 'o':
 370                         case 'X':
 371                         case 'x':
 372                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 373                                 goto eval_fmt_mod_unsigned;
 374                                 break;
 375
 376                         case 'u':
 377                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 378 eval_fmt_mod_unsigned:
 379                                 switch (fmt_mod) {
 380                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 381                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 382                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 383                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 384                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 385                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 386                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 387                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 388                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 389                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 390                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 391
 392                                         default:
 393                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 394                                                 goto next_arg;
 395                                 }
 396                                 break;
 397
 398                         case 'A':
 399                         case 'a':
 400                         case 'E':
 401                         case 'e':
 402                         case 'F':
 403                         case 'f':
 404                         case 'G':
 405                         case 'g':
 406                                 switch (fmt_mod) {
 407                                         case FMT_MOD_l:    /* l modifier is ignored */
 408                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 409                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 410
 411                                         default:
 412                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 413                                                 goto next_arg;
 414                                 }
 415                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 416                                 break;
 417
 418                         case 'C':
 419                                 if (fmt_mod != FMT_MOD_NONE) {
 420                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 421                                         goto next_arg;
 422                                 }
 423                                 expected_type = type_wchar_t;
 424                                 allowed_flags = FMT_FLAG_NONE;
 425                                 break;
 426
 427                         case 'c':
 428                                 expected_type = type_int;
 429                                 switch (fmt_mod) {
 430                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 431                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 432                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 433
 434                                         default:
 435                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 436                                                 goto next_arg;
 437                                 }
 438                                 allowed_flags = FMT_FLAG_NONE;
 439                                 break;
 440
 441                         case 'S':
 442                                 if (fmt_mod != FMT_MOD_NONE) {
 443                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 444                                         goto next_arg;
 445                                 }
 446                                 expected_type = type_wchar_t_ptr;
 447                                 expected_qual = TYPE_QUALIFIER_CONST;
 448                                 allowed_flags = FMT_FLAG_NONE;
 449                                 break;
 450
 451                         case 's':
 452                                 switch (fmt_mod) {
 453                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 454                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 455                                         case FMT_MOD_w:    expected_type = type_wchar_t_ptr; break;
 456
 457                                         default:
 458                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 459                                                 goto next_arg;
 460                                 }
 461                                 expected_qual = TYPE_QUALIFIER_CONST;
 462                                 allowed_flags = FMT_FLAG_NONE;
 463                                 break;
 464
 465                         case 'p':
 466                                 if (fmt_mod != FMT_MOD_NONE) {
 467                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 468                                         goto next_arg;
 469                                 }
 470                                 expected_type = type_void_ptr;
 471                                 allowed_flags = FMT_FLAG_NONE;
 472                                 break;
 473
 474                         case 'n':
 475                                 switch (fmt_mod) {
 476                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 477                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 478                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 479                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 480                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 481                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 482                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 483                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 484
 485                                         default:
 486                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 487                                                 goto next_arg;
 488                                 }
 489                                 allowed_flags = FMT_FLAG_NONE;
 490                                 break;
 491
 492                         default:
 493                                 warningf(pos, "encountered unknown conversion specifier '%%%C'", (wint_t)fmt);
 494                                 goto next_arg;
 495                 }
 496
 497                 if ((fmt_flags & ~allowed_flags) != 0) {
 498                         /* TODO better warning message text */
 499                         warningf(pos, "invalid format flags in conversion specification");
 500                 }
 501
 502                 if (arg == NULL) {
 503                         warningf(pos, "too few arguments for format string");
 504                         return;
 505                 }
 506
 507                 {       /* create a scope here to prevent warning about the jump to next_arg */
 508                         type_t *const arg_type = arg->expression->base.type;
 509                         if (is_type_pointer(expected_type)) {
 510                                 type_t *const arg_skip = skip_typeref(arg_type);
 511                                 if (is_type_pointer(arg_skip)) {
 512                                         type_t *const exp_to = skip_typeref(expected_type->pointer.points_to);
 513                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 514                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 515                                                 get_unqualified_type(arg_to) == exp_to) {
 516                                                 goto next_arg;
 517                                         }
 518                                 }
 519                         } else {
 520                                 if (get_unqualified_type(skip_typeref(arg_type)) == expected_type) {
 521                                         goto next_arg;
 522                                 }
 523                         }
 524                         if (is_type_valid(arg_type)) {
 525                                 warningf(pos,
 526                                         "argument type '%T' does not match conversion specifier '%%%s%c'",
 527                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt);
 528                         }
 529                 }
 530 next_arg:
 531                 arg = arg->next;
 532         }
 533         if (!atend(&vchar)) {
 534                 warningf(pos, "format string contains NUL");
 535         }
 536         if (arg != NULL) {
 537                 warningf(pos, "too many arguments for format string");
 538         }
 539 }
 540
 541 void check_format(const call_expression_t *const call)
 542 {
 543         if (!warning.check_format)
 544                 return;
 545
 546         const expression_t *const func_expr = call->function;
 547         if (func_expr->kind != EXPR_REFERENCE)
 548                 return;
 549
 550         const char            *const name = func_expr->reference.symbol->string;
 551         const call_argument_t *      arg  = call->arguments;
 552         if (strcmp(name, "wprintf") == 0) { /* TODO gammlig */
 553                 check_format_arguments(arg, arg->next);
 554         } else if (strcmp(name, "printf") == 0) {
 555                 check_format_arguments(arg, arg->next);
 556         } else if (strcmp(name, "swprintf") == 0) {
 557                 arg = arg->next->next; /* skip destination buffer and size */
 558                 check_format_arguments(arg, arg->next);
 559         } else if (strcmp(name, "sprintf") == 0) {
 560                 arg = arg->next->next; /* skip destination buffer and size */
 561                 check_format_arguments(arg, arg->next);
 562         }
 563 }