src/stdio/vfscanf.c

   1 #include <stdlib.h>
   2 #include <stdarg.h>
   3 #include <ctype.h>
   4 #include <wchar.h>
   5 #include <wctype.h>
   6 #include <limits.h>
   7 #include <string.h>
   8 #include <stdint.h>
   9
  10 #include "stdio_impl.h"
  11 #include "shgetc.h"
  12 #include "intscan.h"
  13 #include "floatscan.h"
  14
  15 #define SIZE_hh -2
  16 #define SIZE_h  -1
  17 #define SIZE_def 0
  18 #define SIZE_l   1
  19 #define SIZE_L   2
  20 #define SIZE_ll  3
  21
  22 static void store_int(void *dest, int size, unsigned long long i)
  23 {
  24         if (!dest) return;
  25         switch (size) {
  26         case SIZE_hh:
  27                 *(char *)dest = i;
  28                 break;
  29         case SIZE_h:
  30                 *(short *)dest = i;
  31                 break;
  32         case SIZE_def:
  33                 *(int *)dest = i;
  34                 break;
  35         case SIZE_l:
  36                 *(long *)dest = i;
  37                 break;
  38         case SIZE_ll:
  39                 *(long long *)dest = i;
  40                 break;
  41         }
  42 }
  43
  44 static void *arg_n(va_list ap, unsigned int n)
  45 {
  46         void *p;
  47         unsigned int i;
  48         va_list ap2;
  49         va_copy(ap2, ap);
  50         for (i=n; i>1; i--) va_arg(ap2, void *);
  51         p = va_arg(ap2, void *);
  52         va_end(ap2);
  53         return p;
  54 }
  55
  56 int vfscanf(FILE *restrict f, const char *restrict fmt, va_list ap)
  57 {
  58         int width;
  59         int size;
  60         int alloc;
  61         int base;
  62         const unsigned char *p;
  63         int c, t;
  64         char *s;
  65         wchar_t *wcs;
  66         mbstate_t st;
  67         void *dest=NULL;
  68         int invert;
  69         int matches=0;
  70         unsigned long long x;
  71         long double y;
  72         off_t pos = 0;
  73         unsigned char scanset[257];
  74         size_t i, k;
  75         wchar_t wc;
  76
  77         FLOCK(f);
  78
  79         if (!f->rpos) __toread(f);
  80         if (!f->rpos) goto input_fail;
  81
  82         for (p=(const unsigned char *)fmt; *p; p++) {
  83
  84                 alloc = 0;
  85
  86                 if (isspace(*p)) {
  87                         while (isspace(p[1])) p++;
  88                         shlim(f, 0);
  89                         while (isspace(shgetc(f)));
  90                         shunget(f);
  91                         pos += shcnt(f);
  92                         continue;
  93                 }
  94                 if (*p != '%' || p[1] == '%') {
  95                         shlim(f, 0);
  96                         if (*p == '%') {
  97                                 p++;
  98                                 while (isspace((c=shgetc(f))));
  99                         } else {
 100                                 c = shgetc(f);
 101                         }
 102                         if (c!=*p) {
 103                                 shunget(f);
 104                                 if (c<0) goto input_fail;
 105                                 goto match_fail;
 106                         }
 107                         pos += shcnt(f);
 108                         continue;
 109                 }
 110
 111                 p++;
 112                 if (*p=='*') {
 113                         dest = 0; p++;
 114                 } else if (isdigit(*p) && p[1]=='$') {
 115                         dest = arg_n(ap, *p-'0'); p+=2;
 116                 } else {
 117                         dest = va_arg(ap, void *);
 118                 }
 119
 120                 for (width=0; isdigit(*p); p++) {
 121                         width = 10*width + *p - '0';
 122                 }
 123
 124                 if (*p=='m') {
 125                         wcs = 0;
 126                         s = 0;
 127                         alloc = !!dest;
 128                         p++;
 129                 } else {
 130                         alloc = 0;
 131                 }
 132
 133                 size = SIZE_def;
 134                 switch (*p++) {
 135                 case 'h':
 136                         if (*p == 'h') p++, size = SIZE_hh;
 137                         else size = SIZE_h;
 138                         break;
 139                 case 'l':
 140                         if (*p == 'l') p++, size = SIZE_ll;
 141                         else size = SIZE_l;
 142                         break;
 143                 case 'j':
 144                         size = SIZE_ll;
 145                         break;
 146                 case 'z':
 147                 case 't':
 148                         size = SIZE_l;
 149                         break;
 150                 case 'L':
 151                         size = SIZE_L;
 152                         break;
 153                 case 'd': case 'i': case 'o': case 'u': case 'x':
 154                 case 'a': case 'e': case 'f': case 'g':
 155                 case 'A': case 'E': case 'F': case 'G': case 'X':
 156                 case 's': case 'c': case '[':
 157                 case 'S': case 'C':
 158                 case 'p': case 'n':
 159                         p--;
 160                         break;
 161                 default:
 162                         goto fmt_fail;
 163                 }
 164
 165                 t = *p;
 166
 167                 /* C or S */
 168                 if ((t&0x2f) == 3) {
 169                         t |= 32;
 170                         size = SIZE_l;
 171                 }
 172
 173                 switch (t) {
 174                 case 'c':
 175                         if (width < 1) width = 1;
 176                 case '[':
 177                         break;
 178                 case 'n':
 179                         store_int(dest, size, pos);
 180                         /* do not increment match count, etc! */
 181                         continue;
 182                 default:
 183                         shlim(f, 0);
 184                         while (isspace(shgetc(f)));
 185                         shunget(f);
 186                         pos += shcnt(f);
 187                 }
 188
 189                 shlim(f, width);
 190                 if (shgetc(f) < 0) goto input_fail;
 191                 shunget(f);
 192
 193                 switch (t) {
 194                 case 's':
 195                 case 'c':
 196                 case '[':
 197                         if (t == 'c' || t == 's') {
 198                                 memset(scanset, -1, sizeof scanset);
 199                                 scanset[0] = 0;
 200                                 if (t == 's') {
 201                                         scanset[1+'\t'] = 0;
 202                                         scanset[1+'\n'] = 0;
 203                                         scanset[1+'\v'] = 0;
 204                                         scanset[1+'\f'] = 0;
 205                                         scanset[1+'\r'] = 0;
 206                                         scanset[1+' '] = 0;
 207                                 }
 208                         } else {
 209                                 if (*++p == '^') p++, invert = 1;
 210                                 else invert = 0;
 211                                 memset(scanset, invert, sizeof scanset);
 212                                 scanset[0] = 0;
 213                                 if (*p == '-') p++, scanset[1+'-'] = 1-invert;
 214                                 else if (*p == ']') p++, scanset[1+']'] = 1-invert;
 215                                 for (; *p != ']'; p++) {
 216                                         if (!*p) goto fmt_fail;
 217                                         if (*p=='-' && p[1] && p[1] != ']')
 218                                                 for (c=p++[-1]; c<*p; c++)
 219                                                         scanset[1+c] = 1-invert;
 220                                         scanset[1+*p] = 1-invert;
 221                                 }
 222                         }
 223                         wcs = 0;
 224                         s = 0;
 225                         i = 0;
 226                         k = t=='c' ? width+1U : 31;
 227                         if (size == SIZE_l) {
 228                                 if (alloc) {
 229                                         wcs = malloc(k*sizeof(wchar_t));
 230                                         if (!wcs) goto alloc_fail;
 231                                 } else {
 232                                         wcs = dest;
 233                                 }
 234                                 st = (mbstate_t){0};
 235                                 while (scanset[(c=shgetc(f))+1]) {
 236                                         switch (mbrtowc(&wc, &(char){c}, 1, &st)) {
 237                                         case -1:
 238                                                 goto input_fail;
 239                                         case -2:
 240                                                 continue;
 241                                         }
 242                                         if (wcs) wcs[i++] = wc;
 243                                         if (alloc && i==k) {
 244                                                 k+=k+1;
 245                                                 wchar_t *tmp = realloc(wcs, k*sizeof(wchar_t));
 246                                                 if (!tmp) goto alloc_fail;
 247                                                 wcs = tmp;
 248                                         }
 249                                 }
 250                                 if (!mbsinit(&st)) goto input_fail;
 251                         } else if (alloc) {
 252                                 s = malloc(k);
 253                                 if (!s) goto alloc_fail;
 254                                 while (scanset[(c=shgetc(f))+1]) {
 255                                         s[i++] = c;
 256                                         if (i==k) {
 257                                                 k+=k+1;
 258                                                 char *tmp = realloc(s, k);
 259                                                 if (!tmp) goto alloc_fail;
 260                                                 s = tmp;
 261                                         }
 262                                 }
 263                         } else if ((s = dest)) {
 264                                 while (scanset[(c=shgetc(f))+1])
 265                                         s[i++] = c;
 266                         } else {
 267                                 while (scanset[(c=shgetc(f))+1]);
 268                         }
 269                         shunget(f);
 270                         if (!shcnt(f)) goto match_fail;
 271                         if (t == 'c' && shcnt(f) != width) goto match_fail;
 272                         if (alloc) {
 273                                 if (size == SIZE_l) *(wchar_t **)dest = wcs;
 274                                 else *(char **)dest = s;
 275                         }
 276                         if (t != 'c') {
 277                                 if (wcs) wcs[i] = 0;
 278                                 if (s) s[i] = 0;
 279                         }
 280                         break;
 281                 case 'p':
 282                 case 'X':
 283                 case 'x':
 284                         base = 16;
 285                         goto int_common;
 286                 case 'o':
 287                         base = 8;
 288                         goto int_common;
 289                 case 'd':
 290                 case 'u':
 291                         base = 10;
 292                         goto int_common;
 293                 case 'i':
 294                         base = 0;
 295                 int_common:
 296                         x = __intscan(f, base, 0, ULLONG_MAX);
 297                         if (!shcnt(f)) goto match_fail;
 298                         if (t=='p' && dest) *(void **)dest = (void *)(uintptr_t)x;
 299                         else store_int(dest, size, x);
 300                         break;
 301                 case 'a': case 'A':
 302                 case 'e': case 'E':
 303                 case 'f': case 'F':
 304                 case 'g': case 'G':
 305                         y = __floatscan(f, size, 0);
 306                         if (!shcnt(f)) goto match_fail;
 307                         if (dest) switch (size) {
 308                         case SIZE_def:
 309                                 *(float *)dest = y;
 310                                 break;
 311                         case SIZE_l:
 312                                 *(double *)dest = y;
 313                                 break;
 314                         case SIZE_L:
 315                                 *(long double *)dest = y;
 316                                 break;
 317                         }
 318                         break;
 319                 }
 320
 321                 pos += shcnt(f);
 322                 if (dest) matches++;
 323         }
 324         if (0) {
 325 fmt_fail:
 326 alloc_fail:
 327 input_fail:
 328                 if (!matches) matches--;
 329 match_fail:
 330                 if (alloc) {
 331                         free(s);
 332                         free(wcs);
 333                 }
 334         }
 335         FUNLOCK(f);
 336         return matches;
 337 }
 338
 339 weak_alias(vfscanf,__isoc99_vfscanf);