nsz Git - musl/blob - 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 static int readwc(int c, wchar_t **wcs, mbstate_t *st)
  60 {
  61         char ch = c;
  62         wchar_t wc;
  63         switch (mbrtowc(&wc, &ch, 1, st)) {
  64         case -1:
  65                 return -1;
  66         case -2:
  67                 break;
  68         default:
  69                 if (*wcs) *(*wcs)++ = wc;
  70         }
  71         return 0;
  72 }
  73
  74 int vfscanf(FILE *restrict f, const char *restrict fmt, va_list ap)
  75 {
  76         int width;
  77         int size;
  78         int alloc;
  79         int base;
  80         const unsigned char *p;
  81         int c, t;
  82         char *s;
  83         wchar_t *wcs;
  84         mbstate_t st;
  85         void *dest=NULL;
  86         int invert;
  87         int matches=0;
  88         unsigned long long x;
  89         long double y;
  90         off_t pos = 0;
  91         unsigned char scanset[257];
  92
  93         FLOCK(f);
  94
  95         for (p=(const unsigned char *)fmt; *p; p++) {
  96
  97                 if (isspace(*p)) {
  98                         while (isspace(p[1])) p++;
  99                         shlim(f, 0);
 100                         while (isspace(shgetc(f)));
 101                         shunget(f);
 102                         pos += shcnt(f);
 103                         continue;
 104                 }
 105                 if (*p != '%' || p[1] == '%') {
 106                         p += *p=='%';
 107                         shlim(f, 0);
 108                         c = shgetc(f);
 109                         if (c!=*p) {
 110                                 shunget(f);
 111                                 if (c<0) goto input_fail;
 112                                 goto match_fail;
 113                         }
 114                         pos++;
 115                         continue;
 116                 }
 117
 118                 p++;
 119                 if (*p=='*') {
 120                         dest = 0; p++;
 121                 } else if (isdigit(*p) && p[1]=='$') {
 122                         dest = arg_n(ap, *p-'0'); p+=2;
 123                 } else {
 124                         dest = va_arg(ap, void *);
 125                 }
 126
 127                 for (width=0; isdigit(*p); p++) {
 128                         width = 10*width + *p - '0';
 129                 }
 130
 131                 if (*p=='m') {
 132                         alloc = 1;
 133                         p++;
 134                 } else {
 135                         alloc = 0;
 136                 }
 137
 138                 size = SIZE_def;
 139                 switch (*p++) {
 140                 case 'h':
 141                         if (*p == 'h') p++, size = SIZE_hh;
 142                         else size = SIZE_h;
 143                         break;
 144                 case 'l':
 145                         if (*p == 'l') p++, size = SIZE_ll;
 146                         else size = SIZE_l;
 147                         break;
 148                 case 'j':
 149                         size = SIZE_ll;
 150                         break;
 151                 case 'z':
 152                 case 't':
 153                         size = SIZE_l;
 154                         break;
 155                 case 'L':
 156                         size = SIZE_L;
 157                         break;
 158                 case 'd': case 'i': case 'o': case 'u': case 'x':
 159                 case 'a': case 'e': case 'f': case 'g':
 160                 case 'A': case 'E': case 'F': case 'G': case 'X':
 161                 case 's': case 'c': case '[':
 162                 case 'S': case 'C':
 163                 case 'p': case 'n':
 164                         p--;
 165                         break;
 166                 default:
 167                         goto fmt_fail;
 168                 }
 169
 170                 t = *p;
 171
 172                 switch (t) {
 173                 case 'C':
 174                         if (width < 1) width = 1;
 175                 case 'S':
 176                         t |= 32;
 177                         size = SIZE_l;
 178                         break;
 179                 case 'c':
 180                         if (width < 1) width = 1;
 181                 case 'd': case 'i': case 'o': case 'u': case 'x':
 182                 case 'a': case 'e': case 'f': case 'g':
 183                 case 'A': case 'E': case 'F': case 'G': case 'X':
 184                 case '[': case 's':
 185                 case 'p': case 'n':
 186                         break;
 187                 default:
 188                         goto fmt_fail;
 189                 }
 190
 191                 if (t == 'n') {
 192                         store_int(dest, size, pos);
 193                         /* do not increment match count, etc! */
 194                         continue;
 195                 }
 196
 197                 if (t != '[' && (t|32) != 'c') {
 198                         shlim(f, 0);
 199                         while (isspace(shgetc(f)));
 200                         shunget(f);
 201                         pos += shcnt(f);
 202                 }
 203
 204                 shlim(f, width);
 205                 if (shgetc(f) < 0) goto input_fail;
 206                 shunget(f);
 207
 208                 switch (t) {
 209                 case 's':
 210                 case 'c':
 211                 case '[':
 212                         if (t == 'c' || t == 's') {
 213                                 memset(scanset, -1, sizeof scanset);
 214                                 scanset[0] = 0;
 215                                 if (t == 's') {
 216                                         scanset[1+'\t'] = 0;
 217                                         scanset[1+'\n'] = 0;
 218                                         scanset[1+'\v'] = 0;
 219                                         scanset[1+'\f'] = 0;
 220                                         scanset[1+'\r'] = 0;
 221                                         scanset[1+' '] = 0;
 222                                 }
 223                         } else {
 224                                 if (*++p == '^') p++, invert = 1;
 225                                 else invert = 0;
 226                                 memset(scanset, invert, sizeof scanset);
 227                                 scanset[0] = 0;
 228                                 if (*p == '-') p++, scanset[1+'-'] = 1-invert;
 229                                 else if (*p == ']') p++, scanset[1+']'] = 1-invert;
 230                                 for (; *p != ']'; p++) {
 231                                         if (!*p) goto fmt_fail;
 232                                         if (*p=='-' && p[1] && p[1] != ']')
 233                                                 for (c=p++[-1]; c<*p; c++)
 234                                                         scanset[1+c] = 1-invert;
 235                                         scanset[1+*p] = 1-invert;
 236                                 }
 237                         }
 238                         wcs = 0;
 239                         s = 0;
 240                         if (size == SIZE_l) {
 241                                 wcs = dest;
 242                                 st = (mbstate_t){0};
 243                                 while (scanset[(c=shgetc(f))+1]) {
 244                                         if (readwc(c, &wcs, &st) < 0)
 245                                                 goto input_fail;
 246                                 }
 247                                 if (!mbsinit(&st)) goto input_fail;
 248                         } else if ((s = dest)) {
 249                                 while (scanset[(c=shgetc(f))+1])
 250                                         *s++ = c;
 251                         } else {
 252                                 while (scanset[(c=shgetc(f))+1]);
 253                         }
 254                         shunget(f);
 255                         if (!shcnt(f)) goto match_fail;
 256                         if (t == 'c' && shcnt(f) != width) goto match_fail;
 257                         if (wcs) *wcs = 0;
 258                         if (s) *s = 0;
 259                         break;
 260                 case 'p':
 261                 case 'X':
 262                 case 'x':
 263                         base = 16;
 264                         goto int_common;
 265                 case 'o':
 266                         base = 8;
 267                         goto int_common;
 268                 case 'd':
 269                 case 'u':
 270                         base = 10;
 271                         goto int_common;
 272                 case 'i':
 273                         base = 0;
 274                 int_common:
 275                         x = __intscan(f, base, 0, ULLONG_MAX);
 276                         if (!shcnt(f)) goto match_fail;
 277                         if (t=='p' && dest) *(void **)dest = (void *)(uintptr_t)x;
 278                         else store_int(dest, size, x);
 279                         break;
 280                 case 'a': case 'A':
 281                 case 'e': case 'E':
 282                 case 'f': case 'F':
 283                 case 'g': case 'G':
 284                         y = __floatscan(f, size, 0);
 285                         if (!shcnt(f)) goto match_fail;
 286                         if (dest) switch (size) {
 287                         case SIZE_def:
 288                                 *(float *)dest = y;
 289                                 break;
 290                         case SIZE_l:
 291                                 *(double *)dest = y;
 292                                 break;
 293                         case SIZE_L:
 294                                 *(long double *)dest = y;
 295                                 break;
 296                         }
 297                         break;
 298                 }
 299
 300                 pos += shcnt(f);
 301                 if (dest) matches++;
 302         }
 303         if (0) {
 304 fmt_fail:
 305 input_fail:
 306                 if (!matches) matches--;
 307         }
 308 match_fail:
 309         FUNLOCK(f);
 310         return matches;
 311 }