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
  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 } format_length_modifier_t;
  56
  57 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  58 {
  59         static const char* const names[] = {
  60                 [FMT_MOD_NONE] = "",
  61                 [FMT_MOD_L]    = "L",
  62                 [FMT_MOD_hh]   = "hh",
  63                 [FMT_MOD_h]    = "h",
  64                 [FMT_MOD_l]    = "l",
  65                 [FMT_MOD_ll]   = "ll",
  66                 [FMT_MOD_j]    = "j",
  67                 [FMT_MOD_t]    = "t",
  68                 [FMT_MOD_z]    = "z",
  69                 [FMT_MOD_q]    = "q"
  70         };
  71         assert(mod < sizeof(names) / sizeof(*names));
  72         return names[mod];
  73 }
  74
  75 static void warn_invalid_length_modifier(const source_position_t pos,
  76                                          const format_length_modifier_t mod,
  77                                          const wchar_rep_t conversion)
  78 {
  79         warningf(pos,
  80                 "invalid length modifier '%s' for conversion specifier '%%%c'",
  81                 get_length_modifier_name(mod), conversion
  82         );
  83 }
  84
  85 typedef struct vchar_t vchar_t;
  86 struct vchar_t {
  87         const void *string;   /**< the string */
  88         size_t     position;
  89         size_t     size;
  90
  91         unsigned (*first)(vchar_t *self);
  92         unsigned (*next)(vchar_t *self);
  93         int (*is_digit)(unsigned vchar);
  94 };
  95
  96 static unsigned string_first(vchar_t *self) {
  97         self->position = 0;
  98         const string_t *string = self->string;
  99         return string->begin[0];
 100 }
 101
 102 static unsigned string_next(vchar_t *self) {
 103         ++self->position;
 104         const string_t *string = self->string;
 105         return string->begin[self->position];
 106 }
 107
 108 static int string_isdigit(unsigned vchar) {
 109         return isdigit(vchar);
 110 }
 111
 112 static unsigned wstring_first(vchar_t *self) {
 113         self->position = 0;
 114         const wide_string_t *wstring = self->string;
 115         return wstring->begin[0];
 116 }
 117
 118 static unsigned wstring_next(vchar_t *self) {
 119         ++self->position;
 120         const wide_string_t *wstring = self->string;
 121         return wstring->begin[self->position];
 122 }
 123
 124 static int wstring_isdigit(unsigned vchar) {
 125         return iswdigit(vchar);
 126 }
 127
 128 static bool atend(vchar_t *self) {
 129         return self->position + 1 == self->size;
 130 }
 131
 132 static void check_format_arguments(const call_argument_t *const fmt_arg, const call_argument_t* arg)
 133 {
 134         const expression_t *fmt_expr = fmt_arg->expression;
 135         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 136                 fmt_expr = fmt_expr->unary.value;
 137         }
 138
 139         vchar_t vchar;
 140         if (fmt_expr->kind == EXPR_WIDE_STRING_LITERAL) {
 141                 vchar.string   = &fmt_expr->wide_string.value;
 142                 vchar.size     = fmt_expr->wide_string.value.size;
 143                 vchar.first    = wstring_first;
 144                 vchar.next     = wstring_next;
 145                 vchar.is_digit = wstring_isdigit;
 146         } else if (fmt_expr->kind == EXPR_STRING_LITERAL) {
 147                 vchar.string   = &fmt_expr->string.value;
 148                 vchar.size     = fmt_expr->string.value.size;
 149                 vchar.first    = string_first;
 150                 vchar.next     = string_next;
 151                 vchar.is_digit = string_isdigit;
 152         } else {
 153                 return;
 154         }
 155         const source_position_t    pos     = fmt_expr->base.source_position;
 156         unsigned fmt = vchar.first(&vchar);
 157         for (; fmt != '\0'; fmt = vchar.next(&vchar)) {
 158                 if (fmt != '%')
 159                         continue;
 160                 fmt = vchar.next(&vchar);
 161
 162                 if (fmt == '%')
 163                         continue;
 164
 165                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 166                 if (fmt == '0') {
 167                         fmt = vchar.next(&vchar);
 168                         fmt_flags |= FMT_FLAG_ZERO;
 169                 }
 170
 171                 /* argument selector or minimum field width */
 172                 if (vchar.is_digit(fmt)) {
 173                         do {
 174                                 fmt = vchar.next(&vchar);
 175                         } while (vchar.is_digit(fmt));
 176
 177                         /* digit string was ... */
 178                         if (fmt == '$') {
 179                                 /* ... argument selector */
 180                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 181                                 /* TODO implement */
 182                                 return;
 183                         }
 184                         /* ... minimum field width */
 185                 } else {
 186                         /* flags */
 187                         for (;;) {
 188                                 format_flags_t flag;
 189                                 switch (fmt) {
 190                                         case '#':  flag = FMT_FLAG_HASH;  break;
 191                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 192                                         case '-':  flag = FMT_FLAG_MINUS; break;
 193                                         case '\'': flag = FMT_FLAG_TICK;  break;
 194
 195                                         case ' ':
 196                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 197                                                         warningf(pos, "' ' is overridden by prior '+' in conversion specification");
 198                                                 }
 199                                                 flag = FMT_FLAG_SPACE;
 200                                                 break;
 201
 202                                         case '+':
 203                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 204                                                         warningf(pos, "'+' overrides prior ' ' in conversion specification");
 205                                                 }
 206                                                 flag = FMT_FLAG_PLUS;
 207                                                 break;
 208
 209                                         default: goto break_fmt_flags;
 210                                 }
 211                                 if (fmt_flags & flag) {
 212                                         warningf(pos, "repeated flag '%c' in conversion specification", (char)fmt);
 213                                 }
 214                                 fmt_flags |= flag;
 215                                 fmt = vchar.next(&vchar);
 216                         }
 217 break_fmt_flags:
 218
 219                         /* minimum field width */
 220                         if (fmt == '*') {
 221                                 if (arg == NULL) {
 222                                         warningf(pos, "missing argument for '*' field width in conversion specification");
 223                                         return;
 224                                 }
 225                                 const type_t *const arg_type = arg->expression->base.type;
 226                                 if (arg_type != type_int) {
 227                                         warningf(pos, "argument for '*' field width in conversion specification is not an 'int', but an '%T'", arg_type);
 228                                 }
 229                                 arg = arg->next;
 230                         } else {
 231                                 while (vchar.is_digit(fmt)) {
 232                                         fmt = vchar.next(&vchar);
 233                                 }
 234                         }
 235                 }
 236
 237                 /* precision */
 238                 if (fmt == '.') {
 239                         fmt = vchar.next(&vchar);
 240                         if (fmt == '*') {
 241                                 if (arg == NULL) {
 242                                         warningf(pos, "missing argument for '*' precision in conversion specification");
 243                                         return;
 244                                 }
 245                                 const type_t *const arg_type = arg->expression->base.type;
 246                                 if (arg_type != type_int) {
 247                                         warningf(pos, "argument for '*' precision in conversion specification is not an 'int', but an '%T'", arg_type);
 248                                 }
 249                                 arg = arg->next;
 250                         } else {
 251                                 /* digit string may be omitted */
 252                                 while (vchar.is_digit(fmt)) {
 253                                         fmt = vchar.next(&vchar);
 254                                 }
 255                         }
 256                 }
 257
 258                 /* length modifier */
 259                 format_length_modifier_t fmt_mod;
 260                 switch (fmt) {
 261                         case 'h':
 262                                 fmt = vchar.next(&vchar);
 263                                 if (fmt == 'h') {
 264                                         fmt = vchar.next(&vchar);
 265                                         fmt_mod = FMT_MOD_hh;
 266                                 } else {
 267                                         fmt_mod = FMT_MOD_h;
 268                                 }
 269                                 break;
 270
 271                         case 'l':
 272                                 fmt = vchar.next(&vchar);
 273                                 if (fmt == 'l') {
 274                                         fmt = vchar.next(&vchar);
 275                                         fmt_mod = FMT_MOD_ll;
 276                                 } else {
 277                                         fmt_mod = FMT_MOD_l;
 278                                 }
 279                                 break;
 280
 281                         case 'L': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_L;    break;
 282                         case 'j': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_j;    break;
 283                         case 't': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_t;    break;
 284                         case 'z': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_z;    break;
 285                         case 'q': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_q;    break;
 286                         default:                            fmt_mod = FMT_MOD_NONE; break;
 287                 }
 288
 289                 if (fmt == '\0') {
 290                         warningf(pos, "dangling %% in format string");
 291                         break;
 292                 }
 293
 294                 const type_t      *expected_type;
 295                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 296                 format_flags_t     allowed_flags;
 297                 switch (fmt) {
 298                         case 'd':
 299                         case 'i':
 300                                 switch (fmt_mod) {
 301                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 302                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 303                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 304                                         case FMT_MOD_l:    expected_type = type_long;      break;
 305                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 306                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 307                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 308                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 309
 310                                         default:
 311                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 312                                                 goto next_arg;
 313                                 }
 314                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 315                                 break;
 316
 317                         case 'o':
 318                         case 'X':
 319                         case 'x':
 320                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 321                                 goto eval_fmt_mod_unsigned;
 322                                 break;
 323
 324                         case 'u':
 325                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 326 eval_fmt_mod_unsigned:
 327                                 switch (fmt_mod) {
 328                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 329                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 330                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 331                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 332                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 333                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 334                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 335                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 336
 337                                         default:
 338                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 339                                                 goto next_arg;
 340                                 }
 341                                 break;
 342
 343                         case 'A':
 344                         case 'a':
 345                         case 'E':
 346                         case 'e':
 347                         case 'F':
 348                         case 'f':
 349                         case 'G':
 350                         case 'g':
 351                                 switch (fmt_mod) {
 352                                         case FMT_MOD_l:    /* l modifier is ignored */
 353                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 354                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 355
 356                                         default:
 357                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 358                                                 goto next_arg;
 359                                 }
 360                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 361                                 break;
 362
 363                         case 'C':
 364                                 if (fmt_mod != FMT_MOD_NONE) {
 365                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 366                                         goto next_arg;
 367                                 }
 368                                 expected_type = type_wchar_t;
 369                                 allowed_flags = FMT_FLAG_NONE;
 370                                 break;
 371
 372                         case 'c':
 373                                 expected_type = type_int;
 374                                 switch (fmt_mod) {
 375                                         case FMT_MOD_NONE: expected_type = type_int;    break; /* TODO promoted char */
 376                                         case FMT_MOD_l:    expected_type = type_wint_t; break;
 377
 378                                         default:
 379                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 380                                                 goto next_arg;
 381                                 }
 382                                 allowed_flags = FMT_FLAG_NONE;
 383                                 break;
 384
 385                         case 'S':
 386                                 if (fmt_mod != FMT_MOD_NONE) {
 387                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 388                                         goto next_arg;
 389                                 }
 390                                 expected_type = type_wchar_t_ptr;
 391                                 expected_qual = TYPE_QUALIFIER_CONST;
 392                                 allowed_flags = FMT_FLAG_NONE;
 393                                 break;
 394
 395                         case 's':
 396                                 switch (fmt_mod) {
 397                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 398                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 399
 400                                         default:
 401                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 402                                                 goto next_arg;
 403                                 }
 404                                 expected_qual = TYPE_QUALIFIER_CONST;
 405                                 allowed_flags = FMT_FLAG_NONE;
 406                                 break;
 407
 408                         case 'p':
 409                                 if (fmt_mod != FMT_MOD_NONE) {
 410                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 411                                         goto next_arg;
 412                                 }
 413                                 expected_type = type_void_ptr;
 414                                 allowed_flags = FMT_FLAG_NONE;
 415                                 break;
 416
 417                         case 'n':
 418                                 switch (fmt_mod) {
 419                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 420                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 421                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 422                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 423                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 424                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 425                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 426                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 427
 428                                         default:
 429                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 430                                                 goto next_arg;
 431                                 }
 432                                 allowed_flags = FMT_FLAG_NONE;
 433                                 break;
 434
 435                         default:
 436                                 warningf(pos, "encountered unknown conversion specifier '%%%C'", (wint_t)fmt);
 437                                 goto next_arg;
 438                 }
 439
 440                 if ((fmt_flags & ~allowed_flags) != 0) {
 441                         /* TODO better warning message text */
 442                         warningf(pos, "invalid format flags in conversion specification");
 443                 }
 444
 445                 if (arg == NULL) {
 446                         warningf(pos, "too few arguments for format string");
 447                         return;
 448                 }
 449
 450                 {       /* create a scope here to prevent warning about the jump to next_arg */
 451                         type_t *const arg_type = arg->expression->base.type;
 452                         if (is_type_pointer(expected_type)) {
 453                                 type_t *const arg_skip = skip_typeref(arg_type);
 454                                 if (is_type_pointer(arg_skip)) {
 455                                         type_t *const exp_to = skip_typeref(expected_type->pointer.points_to);
 456                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 457                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 458                                                 get_unqualified_type(arg_to) == exp_to) {
 459                                                 goto next_arg;
 460                                         }
 461                                 }
 462                         } else {
 463                                 if (get_unqualified_type(skip_typeref(arg_type)) == expected_type) {
 464                                         goto next_arg;
 465                                 }
 466                         }
 467                         if (is_type_valid(arg_type)) {
 468                                 warningf(pos,
 469                                         "argument type '%T' does not match conversion specifier '%%%s%c'",
 470                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt);
 471                         }
 472                 }
 473 next_arg:
 474                 arg = arg->next;
 475         }
 476         if (!atend(&vchar)) {
 477                 warningf(pos, "format string contains NUL");
 478         }
 479         if (arg != NULL) {
 480                 warningf(pos, "too many arguments for format string");
 481         }
 482 }
 483
 484 void check_format(const call_expression_t *const call)
 485 {
 486         if (!warning.check_format)
 487                 return;
 488
 489         const expression_t *const func_expr = call->function;
 490         if (func_expr->kind != EXPR_REFERENCE)
 491                 return;
 492
 493         const char            *const name = func_expr->reference.symbol->string;
 494         const call_argument_t *      arg  = call->arguments;
 495         if (strcmp(name, "wprintf") == 0) { /* TODO gammlig */
 496                 check_format_arguments(arg, arg->next);
 497         } else if (strcmp(name, "printf") == 0) {
 498                 check_format_arguments(arg, arg->next);
 499         } else if (strcmp(name, "swprintf") == 0) {
 500                 arg = arg->next->next; /* skip destination buffer and size */
 501                 check_format_arguments(arg, arg->next);
 502         } else if (strcmp(name, "sprintf") == 0) {
 503                 arg = arg->next->next; /* skip destination buffer and size */
 504                 check_format_arguments(arg, arg->next);
 505         }
 506 }