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