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 <wctype.h>
  21
  22 #include "format_check.h"
  23 #include "symbol_t.h"
  24 #include "ast_t.h"
  25 #include "diagnostic.h"
  26 #include "types.h"
  27 #include "type_t.h"
  28 #include "warning.h"
  29
  30
  31 typedef enum format_flag_t {
  32         FMT_FLAG_NONE  = 0,
  33         FMT_FLAG_HASH  = 1U << 0,
  34         FMT_FLAG_ZERO  = 1U << 1,
  35         FMT_FLAG_MINUS = 1U << 2,
  36         FMT_FLAG_SPACE = 1U << 3,
  37         FMT_FLAG_PLUS  = 1U << 4,
  38         FMT_FLAG_TICK  = 1U << 5
  39 } format_flag_t;
  40
  41 typedef unsigned format_flags_t;
  42
  43 typedef enum format_length_modifier_t {
  44         FMT_MOD_NONE,
  45         FMT_MOD_L,
  46         FMT_MOD_hh,
  47         FMT_MOD_h,
  48         FMT_MOD_l,
  49         FMT_MOD_ll,
  50         FMT_MOD_j,
  51         FMT_MOD_t,
  52         FMT_MOD_z,
  53         FMT_MOD_q
  54 } format_length_modifier_t;
  55
  56 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  57 {
  58         static const char* const names[] = {
  59                 [FMT_MOD_NONE] = "",
  60                 [FMT_MOD_L]    = "L",
  61                 [FMT_MOD_hh]   = "hh",
  62                 [FMT_MOD_h]    = "h",
  63                 [FMT_MOD_l]    = "l",
  64                 [FMT_MOD_ll]   = "ll",
  65                 [FMT_MOD_j]    = "j",
  66                 [FMT_MOD_t]    = "t",
  67                 [FMT_MOD_z]    = "z",
  68                 [FMT_MOD_q]    = "q"
  69         };
  70         assert(mod < sizeof(names) / sizeof(*names));
  71         return names[mod];
  72 }
  73
  74 static void warn_invalid_length_modifier(const source_position_t pos,
  75                                          const format_length_modifier_t mod,
  76                                          const wchar_rep_t conversion)
  77 {
  78         warningf(pos,
  79                 "invalid length modifier '%s' for conversion specifier '%%%c'",
  80                 get_length_modifier_name(mod), conversion
  81         );
  82 }
  83
  84 static void check_format_arguments(const call_argument_t *const fmt_arg, const call_argument_t* arg)
  85 {
  86         const expression_t *fmt_expr = fmt_arg->expression;
  87         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
  88                 fmt_expr = fmt_expr->unary.value;
  89         }
  90
  91         if (fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
  92                 return;
  93
  94         const source_position_t    pos     = fmt_expr->base.source_position;
  95         const wide_string_t *const wstring = &fmt_expr->wide_string.value;
  96         const wchar_rep_t *fmt = wstring->begin;
  97         for (; *fmt != '\0'; ++fmt) {
  98                 if (*fmt != '%')
  99                         continue;
 100                 ++fmt;
 101
 102                 if (*fmt == '%')
 103                         continue;
 104
 105                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 106                 if (*fmt == '0') {
 107                         ++fmt;
 108                         fmt_flags |= FMT_FLAG_ZERO;
 109                 }
 110
 111                 /* argument selector or minimum field width */
 112                 if (iswdigit(*fmt)) {
 113                         do {
 114                                 ++fmt;
 115                         } while (iswdigit(*fmt));
 116
 117                         /* digit string was ... */
 118                         if (*fmt == '$') {
 119                                 /* ... argument selector */
 120                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 121                                 /* TODO implement */
 122                                 return;
 123                         }
 124                         /* ... minimum field width */
 125                 } else {
 126                         /* flags */
 127                         for (;;) {
 128                                 format_flags_t flag;
 129                                 switch (*fmt) {
 130                                         case '#':  flag = FMT_FLAG_HASH;  break;
 131                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 132                                         case '-':  flag = FMT_FLAG_MINUS; break;
 133                                         case '\'': flag = FMT_FLAG_TICK;  break;
 134
 135                                         case ' ':
 136                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 137                                                         warningf(pos, "' ' is overridden by prior '+' in conversion specification");
 138                                                 }
 139                                                 flag = FMT_FLAG_SPACE;
 140                                                 break;
 141
 142                                         case '+':
 143                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 144                                                         warningf(pos, "'+' overrides prior ' ' in conversion specification");
 145                                                 }
 146                                                 flag = FMT_FLAG_PLUS;
 147                                                 break;
 148
 149                                         default: goto break_fmt_flags;
 150                                 }
 151                                 if (fmt_flags & flag) {
 152                                         warningf(pos, "repeated flag '%c' in conversion specification", (char)*fmt);
 153                                 }
 154                                 fmt_flags |= flag;
 155                                 ++fmt;
 156                         }
 157 break_fmt_flags:
 158
 159                         /* minimum field width */
 160                         if (*fmt == '*') {
 161                                 if (arg == NULL) {
 162                                         warningf(pos, "missing argument for '*' field width in conversion specification");
 163                                         return;
 164                                 }
 165                                 const type_t *const arg_type = arg->expression->base.type;
 166                                 if (arg_type != type_int) {
 167                                         warningf(pos, "argument for '*' field width in conversion specification is not an 'int', but an '%T'", arg_type);
 168                                 }
 169                                 arg = arg->next;
 170                         } else {
 171                                 while (iswdigit(*fmt)) {
 172                                         ++fmt;
 173                                 }
 174                         }
 175                 }
 176
 177                 /* precision */
 178                 if (*fmt == '.') {
 179                         ++fmt;
 180                         if (*fmt == '*') {
 181                                 if (arg == NULL) {
 182                                         warningf(pos, "missing argument for '*' precision in conversion specification");
 183                                         return;
 184                                 }
 185                                 const type_t *const arg_type = arg->expression->base.type;
 186                                 if (arg_type != type_int) {
 187                                         warningf(pos, "argument for '*' precision in conversion specification is not an 'int', but an '%T'", arg_type);
 188                                 }
 189                                 arg = arg->next;
 190                         } else {
 191                                 /* digit string may be omitted */
 192                                 while (iswdigit(*fmt)) {
 193                                         ++fmt;
 194                                 }
 195                         }
 196                 }
 197
 198                 /* length modifier */
 199                 format_length_modifier_t fmt_mod;
 200                 switch (*fmt) {
 201                         case 'h':
 202                                 ++fmt;
 203                                 if (*fmt == 'h') {
 204                                         ++fmt;
 205                                         fmt_mod = FMT_MOD_hh;
 206                                 } else {
 207                                         fmt_mod = FMT_MOD_h;
 208                                 }
 209                                 break;
 210
 211                         case 'l':
 212                                 ++fmt;
 213                                 if (*fmt == 'l') {
 214                                         ++fmt;
 215                                         fmt_mod = FMT_MOD_ll;
 216                                 } else {
 217                                         fmt_mod = FMT_MOD_l;
 218                                 }
 219                                 break;
 220
 221                         case 'L': ++fmt; fmt_mod = FMT_MOD_L;    break;
 222                         case 'j': ++fmt; fmt_mod = FMT_MOD_j;    break;
 223                         case 't': ++fmt; fmt_mod = FMT_MOD_t;    break;
 224                         case 'z': ++fmt; fmt_mod = FMT_MOD_z;    break;
 225                         case 'q': ++fmt; fmt_mod = FMT_MOD_q;    break;
 226                         default:         fmt_mod = FMT_MOD_NONE; break;
 227                 }
 228
 229                 if (*fmt == '\0') {
 230                         warningf(pos, "dangling %% in format string");
 231                         break;
 232                 }
 233
 234                 const type_t      *expected_type;
 235                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 236                 format_flags_t     allowed_flags;
 237                 switch (*fmt) {
 238                         case 'd':
 239                         case 'i':
 240                                 switch (fmt_mod) {
 241                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 242                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 243                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 244                                         case FMT_MOD_l:    expected_type = type_long;      break;
 245                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 246                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 247                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 248                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 249
 250                                         default:
 251                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 252                                                 goto next_arg;
 253                                 }
 254                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 255                                 break;
 256
 257                         case 'o':
 258                         case 'X':
 259                         case 'x':
 260                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 261                                 goto eval_fmt_mod_unsigned;
 262                                 break;
 263
 264                         case 'u':
 265                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 266 eval_fmt_mod_unsigned:
 267                                 switch (fmt_mod) {
 268                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 269                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 270                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 271                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 272                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 273                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 274                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 275                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 276
 277                                         default:
 278                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 279                                                 goto next_arg;
 280                                 }
 281                                 break;
 282
 283                         case 'A':
 284                         case 'a':
 285                         case 'E':
 286                         case 'e':
 287                         case 'F':
 288                         case 'f':
 289                         case 'G':
 290                         case 'g':
 291                                 switch (fmt_mod) {
 292                                         case FMT_MOD_l:    /* l modifier is ignored */
 293                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 294                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 295
 296                                         default:
 297                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 298                                                 goto next_arg;
 299                                 }
 300                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 301                                 break;
 302
 303                         case 'C':
 304                                 if (fmt_mod != FMT_MOD_NONE) {
 305                                         warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 306                                         goto next_arg;
 307                                 }
 308                                 expected_type = type_wchar_t;
 309                                 allowed_flags = FMT_FLAG_NONE;
 310                                 break;
 311
 312                         case 'c':
 313                                 expected_type = type_int;
 314                                 switch (fmt_mod) {
 315                                         case FMT_MOD_NONE: expected_type = type_int;    break; /* TODO promoted char */
 316                                         case FMT_MOD_l:    expected_type = type_wint_t; break;
 317
 318                                         default:
 319                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 320                                                 goto next_arg;
 321                                 }
 322                                 allowed_flags = FMT_FLAG_NONE;
 323                                 break;
 324
 325                         case 'S':
 326                                 if (fmt_mod != FMT_MOD_NONE) {
 327                                         warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 328                                         goto next_arg;
 329                                 }
 330                                 expected_type = type_wchar_t_ptr;
 331                                 expected_qual = TYPE_QUALIFIER_CONST;
 332                                 allowed_flags = FMT_FLAG_NONE;
 333                                 break;
 334
 335                         case 's':
 336                                 switch (fmt_mod) {
 337                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 338                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 339
 340                                         default:
 341                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 342                                                 goto next_arg;
 343                                 }
 344                                 expected_qual = TYPE_QUALIFIER_CONST;
 345                                 allowed_flags = FMT_FLAG_NONE;
 346                                 break;
 347
 348                         case 'p':
 349                                 if (fmt_mod != FMT_MOD_NONE) {
 350                                         warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 351                                         goto next_arg;
 352                                 }
 353                                 expected_type = type_void_ptr;
 354                                 allowed_flags = FMT_FLAG_NONE;
 355                                 break;
 356
 357                         case 'n':
 358                                 switch (fmt_mod) {
 359                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 360                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 361                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 362                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 363                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 364                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 365                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 366                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 367
 368                                         default:
 369                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 370                                                 goto next_arg;
 371                                 }
 372                                 allowed_flags = FMT_FLAG_NONE;
 373                                 break;
 374
 375                         default:
 376                                 warningf(pos, "encountered unknown conversion specifier '%%%C'", (wint_t)*fmt);
 377                                 goto next_arg;
 378                 }
 379
 380                 if ((fmt_flags & ~allowed_flags) != 0) {
 381                         /* TODO better warning message text */
 382                         warningf(pos, "invalid format flags in conversion specification");
 383                 }
 384
 385                 if (arg == NULL) {
 386                         warningf(pos, "too few arguments for format string");
 387                         return;
 388                 }
 389
 390                 {       /* create a scope here to prevent warning about the jump to next_arg */
 391                         type_t *const arg_type = arg->expression->base.type;
 392                         if (is_type_pointer(expected_type)) {
 393                                 type_t *const arg_skip = skip_typeref(arg_type);
 394                                 if (is_type_pointer(arg_skip)) {
 395                                         type_t *const exp_to = skip_typeref(expected_type->pointer.points_to);
 396                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 397                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 398                                                 get_unqualified_type(arg_to) == exp_to) {
 399                                                 goto next_arg;
 400                                         }
 401                                 }
 402                         } else {
 403                                 if (get_unqualified_type(skip_typeref(arg_type)) == expected_type) {
 404                                         goto next_arg;
 405                                 }
 406                         }
 407                         if (is_type_valid(arg_type)) {
 408                                 warningf(pos,
 409                                         "argument type '%T' does not match conversion specifier '%%%s%c'",
 410                                         arg_type, get_length_modifier_name(fmt_mod), (char)*fmt);
 411                         }
 412                 }
 413 next_arg:
 414                 arg = arg->next;
 415         }
 416         if (fmt + 1 != wstring->begin + wstring->size) {
 417                 warningf(pos, "format string contains NUL");
 418         }
 419         if (arg != NULL) {
 420                 warningf(pos, "too many arguments for format string");
 421         }
 422 }
 423
 424 void check_format(const call_expression_t *const call)
 425 {
 426         if (!warning.check_format)
 427                 return;
 428
 429         const expression_t *const func_expr = call->function;
 430         if (func_expr->kind != EXPR_REFERENCE)
 431                 return;
 432
 433         const char            *const name = func_expr->reference.symbol->string;
 434         const call_argument_t *      arg  = call->arguments;
 435         if (strcmp(name, "wprintf") == 0) { /* TODO gammlig */
 436                 check_format_arguments(arg, arg->next);
 437         } else if (strcmp(name, "swprintf") == 0) {
 438                 arg = arg->next->next; /* skip destination buffer and size */
 439                 check_format_arguments(arg, arg->next);
 440         }
 441 }