nsz Git - cparser/blob - format_check.c

   1 #include <wctype.h>
   2
   3 #include "ast_t.h"
   4 #include "diagnostic.h"
   5 #include "format_check.h"
   6 #include "types.h"
   7
   8
   9 typedef enum format_flag_t {
  10         FMT_FLAG_NONE  = 0,
  11         FMT_FLAG_HASH  = 1U << 0,
  12         FMT_FLAG_ZERO  = 1U << 1,
  13         FMT_FLAG_MINUS = 1U << 2,
  14         FMT_FLAG_SPACE = 1U << 3,
  15         FMT_FLAG_PLUS  = 1U << 4,
  16         FMT_FLAG_TICK  = 1U << 5
  17 } format_flag_t;
  18
  19 typedef unsigned format_flags_t;
  20
  21 typedef enum format_length_modifier_t {
  22         FMT_MOD_NONE,
  23         FMT_MOD_L,
  24         FMT_MOD_hh,
  25         FMT_MOD_h,
  26         FMT_MOD_l,
  27         FMT_MOD_ll,
  28         FMT_MOD_j,
  29         FMT_MOD_t,
  30         FMT_MOD_z,
  31         FMT_MOD_q
  32 } format_length_modifier_t;
  33
  34 static void warn_invalid_length_modifier(const source_position_t pos,
  35                                          const format_length_modifier_t mod,
  36                                          const char conversion)
  37 {
  38         static const char* const names[] = {
  39                 [FMT_MOD_NONE] = "",
  40                 [FMT_MOD_L]    = "L",
  41                 [FMT_MOD_hh]   = "hh",
  42                 [FMT_MOD_h]    = "h",
  43                 [FMT_MOD_l]    = "l",
  44                 [FMT_MOD_ll]   = "ll",
  45                 [FMT_MOD_j]    = "j",
  46                 [FMT_MOD_t]    = "t",
  47                 [FMT_MOD_z]    = "z",
  48                 [FMT_MOD_q]    = "q"
  49         };
  50         assert(mod < sizeof(names) / sizeof(*names));
  51
  52         warningf(pos,
  53                 "invalid length modifier '%s' for conversion specifier '%%%c'",
  54                 names[mod], conversion
  55         );
  56 }
  57
  58 static void check_format_arguments(const call_argument_t *const fmt_arg, const call_argument_t* arg)
  59 {
  60         const expression_t *fmt_expr = fmt_arg->expression;
  61         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
  62                 fmt_expr = fmt_expr->unary.value;
  63         }
  64
  65         if (fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
  66                 return;
  67
  68         const source_position_t    pos     = fmt_expr->base.source_position;
  69         const wide_string_t *const wstring = &fmt_expr->wide_string.value;
  70         const wchar_rep_t *fmt = wstring->begin;
  71         for (; *fmt != '\0'; ++fmt) {
  72                 if (*fmt != '%')
  73                         continue;
  74                 ++fmt;
  75
  76                 if (*fmt == '%')
  77                         continue;
  78
  79                 format_flags_t fmt_flags = FMT_FLAG_NONE;
  80                 if (*fmt == '0') {
  81                         ++fmt;
  82                         fmt_flags |= FMT_FLAG_ZERO;
  83                 }
  84
  85                 /* argument selector or minimum field width */
  86                 if (iswdigit(*fmt)) {
  87                         do {
  88                                 ++fmt;
  89                         } while (iswdigit(*fmt));
  90
  91                         /* digit string was ... */
  92                         if (*fmt == '$') {
  93                                 /* ... argument selector */
  94                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
  95                                 /* TODO implement */
  96                                 return;
  97                         }
  98                         /* ... minimum field width */
  99                 } else {
 100                         /* flags */
 101                         for (;;) {
 102                                 format_flags_t flag;
 103                                 switch (*fmt) {
 104                                         case '#':  flag = FMT_FLAG_HASH;  break;
 105                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 106                                         case '-':  flag = FMT_FLAG_MINUS; break;
 107                                         case '\'': flag = FMT_FLAG_TICK;  break;
 108
 109                                         case ' ':
 110                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 111                                                         warningf(pos, "' ' is overridden by prior '+' in conversion specification");
 112                                                 }
 113                                                 flag = FMT_FLAG_SPACE;
 114                                                 break;
 115
 116                                         case '+':
 117                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 118                                                         warningf(pos, "'+' overrides prior ' ' in conversion specification");
 119                                                 }
 120                                                 flag = FMT_FLAG_PLUS;
 121                                                 break;
 122
 123                                         default: goto break_fmt_flags;
 124                                 }
 125                                 if (fmt_flags & flag) {
 126                                         warningf(pos, "repeated flag '%c' in conversion specification", (char)*fmt);
 127                                 }
 128                                 fmt_flags |= flag;
 129                                 ++fmt;
 130                         }
 131 break_fmt_flags:
 132
 133                         /* minimum field width */
 134                         if (*fmt == '*') {
 135                                 if (arg == NULL) {
 136                                         warningf(pos, "missing argument for '*' field width in conversion specification");
 137                                         return;
 138                                 }
 139                                 const type_t *const arg_type = arg->expression->base.datatype;
 140                                 if (arg_type != type_int) {
 141                                         warningf(pos, "argument for '*' field width in conversion specification is not an 'int', but an '%T'", arg_type);
 142                                 }
 143                                 arg = arg->next;
 144                         } else {
 145                                 while (iswdigit(*fmt)) {
 146                                         ++fmt;
 147                                 }
 148                         }
 149                 }
 150
 151                 /* precision */
 152                 if (*fmt == '.') {
 153                         ++fmt;
 154                         if (*fmt == '*') {
 155                                 if (arg == NULL) {
 156                                         warningf(pos, "missing argument for '*' precision in conversion specification");
 157                                         return;
 158                                 }
 159                                 const type_t *const arg_type = arg->expression->base.datatype;
 160                                 if (arg_type != type_int) {
 161                                         warningf(pos, "argument for '*' precision in conversion specification is not an 'int', but an '%T'", arg_type);
 162                                 }
 163                                 arg = arg->next;
 164                         } else {
 165                                 /* digit string may be omitted */
 166                                 while (iswdigit(*fmt)) {
 167                                         ++fmt;
 168                                 }
 169                         }
 170                 }
 171
 172                 /* length modifier */
 173                 format_length_modifier_t fmt_mod;
 174                 switch (*fmt) {
 175                         case 'h':
 176                                 ++fmt;
 177                                 if (*fmt == 'h') {
 178                                         ++fmt;
 179                                         fmt_mod = FMT_MOD_hh;
 180                                 } else {
 181                                         fmt_mod = FMT_MOD_h;
 182                                 }
 183                                 break;
 184
 185                         case 'l':
 186                                 ++fmt;
 187                                 if (*fmt == 'l') {
 188                                         ++fmt;
 189                                         fmt_mod = FMT_MOD_ll;
 190                                 } else {
 191                                         fmt_mod = FMT_MOD_l;
 192                                 }
 193                                 break;
 194
 195                         case 'L': ++fmt; fmt_mod = FMT_MOD_L;    break;
 196                         case 'j': ++fmt; fmt_mod = FMT_MOD_j;    break;
 197                         case 't': ++fmt; fmt_mod = FMT_MOD_t;    break;
 198                         case 'z': ++fmt; fmt_mod = FMT_MOD_z;    break;
 199                         case 'q': ++fmt; fmt_mod = FMT_MOD_q;    break;
 200                         default:         fmt_mod = FMT_MOD_NONE; break;
 201                 }
 202
 203                 if (*fmt == '\0') {
 204                         warningf(pos, "dangling %% in format string");
 205                         break;
 206                 }
 207
 208                 const type_t   *expected_type;
 209                 format_flags_t  allowed_flags;
 210                 switch (*fmt) {
 211                         case 'd':
 212                         case 'i':
 213                                 switch (fmt_mod) {
 214                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 215                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 216                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 217                                         case FMT_MOD_l:    expected_type = type_long;      break;
 218                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 219                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 220                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 221                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 222
 223                                         default:
 224                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 225                                                 break;
 226                                 }
 227                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 228                                 break;
 229
 230                         case 'o':
 231                         case 'X':
 232                         case 'x':
 233                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 234                                 goto eval_fmt_mod_unsigned;
 235                                 break;
 236
 237                         case 'u':
 238                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 239 eval_fmt_mod_unsigned:
 240                                 switch (fmt_mod) {
 241                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 242                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 243                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 244                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 245                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 246                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 247                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 248                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 249
 250                                         default:
 251                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 252                                                 break;
 253                                 }
 254                                 break;
 255
 256                         case 'A':
 257                         case 'a':
 258                         case 'E':
 259                         case 'e':
 260                         case 'F':
 261                         case 'f':
 262                         case 'G':
 263                         case 'g':
 264                                 switch (fmt_mod) {
 265                                         case FMT_MOD_l:    /* l modifier is ignored */
 266                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 267                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 268
 269                                         default:
 270                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 271                                                 break;
 272                                 }
 273                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 274                                 break;
 275
 276                         case 'C':
 277                                 if (fmt_mod != FMT_MOD_NONE) {
 278                                         warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 279                                 }
 280                                 expected_type = type_wchar_t;
 281                                 allowed_flags = FMT_FLAG_NONE;
 282                                 break;
 283
 284                         case 'c':
 285                                 expected_type = type_int;
 286                                 switch (fmt_mod) {
 287                                         case FMT_MOD_NONE: expected_type = type_int;    break; /* TODO promoted char */
 288                                         case FMT_MOD_l:    expected_type = type_wint_t; break;
 289
 290                                         default:
 291                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 292                                                 break;
 293                                 }
 294                                 allowed_flags = FMT_FLAG_NONE;
 295                                 break;
 296
 297                         case 'S':
 298                                 if (fmt_mod != FMT_MOD_NONE) {
 299                                         warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 300                                 }
 301                                 expected_type = type_wchar_t_ptr;
 302                                 allowed_flags = FMT_FLAG_NONE;
 303                                 break;
 304
 305                         case 's':
 306                                 switch (fmt_mod) {
 307                                         case FMT_MOD_NONE: expected_type = type_string;      break;
 308                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 309
 310                                         default:
 311                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 312                                                 break;
 313                                 }
 314                                 allowed_flags = FMT_FLAG_NONE;
 315                                 break;
 316
 317                         case 'p':
 318                                 if (fmt_mod != FMT_MOD_NONE) {
 319                                         warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 320                                 }
 321                                 expected_type = type_void_ptr;
 322                                 allowed_flags = FMT_FLAG_NONE;
 323                                 break;
 324
 325                         case 'n':
 326                                 switch (fmt_mod) {
 327                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 328                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 329                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 330                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 331                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 332                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 333                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 334                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 335
 336                                         default:
 337                                                 warn_invalid_length_modifier(pos, fmt_mod, *fmt);
 338                                                 break;
 339                                 }
 340                                 allowed_flags = FMT_FLAG_NONE;
 341                                 break;
 342
 343                         default:
 344                                 warningf(pos, "encountered unknown conversion specifier '%%%C'", (wint_t)*fmt);
 345                                 arg = arg->next;
 346                                 continue;
 347                 }
 348
 349                 if ((fmt_flags & ~allowed_flags) != 0) {
 350                         /* TODO better warning message text */
 351                         warningf(pos, "invalid format flags in conversion specification");
 352                 }
 353
 354                 if (arg == NULL) {
 355                         warningf(pos, "too few arguments for format string");
 356                         return;
 357                 }
 358
 359                 const type_t *const arg_type = arg->expression->base.datatype;
 360                 if (arg_type != expected_type) {
 361                         warningf(pos, "argument type '%T' does not match conversion specifier '%%%c'\n", arg_type, (char)*fmt);
 362                 }
 363
 364                 arg = arg->next;
 365         }
 366         if (fmt + 1 != wstring->begin + wstring->size) {
 367                 warningf(pos, "format string contains NUL");
 368         }
 369         if (arg != NULL) {
 370                 warningf(pos, "too many arguments for format string");
 371         }
 372 }
 373
 374 void check_format(const call_expression_t *const call)
 375 {
 376         const expression_t *const func_expr = call->function;
 377         if (func_expr->kind != EXPR_REFERENCE)
 378                 return;
 379
 380         const char            *const name = func_expr->reference.symbol->string;
 381         const call_argument_t *      arg  = call->arguments;
 382         if (strcmp(name, "wprintf") == 0) { /* TODO gammlig */
 383                 check_format_arguments(arg, arg->next);
 384         } else if (strcmp(name, "swprintf") == 0) {
 385                 arg = arg->next->next; /* skip destination buffer and size */
 386                 check_format_arguments(arg, arg->next);
 387         }
 388 }