nsz Git - musl/blob - src/stdio/vfscanf.c

   1 #include <stdio.h>
   2 #include <stdlib.h>
   3 #include <stdarg.h>
   4 #include <ctype.h>
   5 #include <wchar.h>
   6 #include <wctype.h>
   7 #include <limits.h>
   8 #include <string.h>
   9 #include <errno.h>
  10 #include <math.h>
  11 #include <float.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
  92         FLOCK(f);
  93
  94         for (p=(const unsigned char *)fmt; *p; p++) {
  95
  96                 if (isspace(*p)) {
  97                         while (isspace(p[1])) p++;
  98                         shlim(f, 0);
  99                         while (isspace(shgetc(f)));
 100                         shunget(f);
 101                         pos += shcnt(f);
 102                         continue;
 103                 }
 104                 if (*p != '%' || p[1] == '%') {
 105                         p += *p=='%';
 106                         shlim(f, 0);
 107                         c = shgetc(f);
 108                         if (c!=*p) {
 109                                 shunget(f);
 110                                 if (c<0) goto input_fail;
 111                                 goto match_fail;
 112                         }
 113                         pos++;
 114                         continue;
 115                 }
 116
 117                 p++;
 118                 if (*p=='*') {
 119                         dest = 0; p++;
 120                 } else if (isdigit(*p) && p[1]=='$') {
 121                         dest = arg_n(ap, *p-'0'); p+=2;
 122                 } else {
 123                         dest = va_arg(ap, void *);
 124                 }
 125
 126                 for (width=0; isdigit(*p); p++) {
 127                         width = 10*width + *p - '0';
 128                 }
 129
 130                 if (*p=='m') {
 131                         alloc = 1;
 132                         p++;
 133                 } else {
 134                         alloc = 0;
 135                 }
 136
 137                 size = SIZE_def;
 138                 switch (*p++) {
 139                 case 'h':
 140                         if (*p == 'h') p++, size = SIZE_hh;
 141                         else size = SIZE_h;
 142                         break;
 143                 case 'l':
 144                         if (*p == 'l') p++, size = SIZE_ll;
 145                         else size = SIZE_l;
 146                         break;
 147                 case 'j':
 148                         size = SIZE_ll;
 149                         break;
 150                 case 'z':
 151                 case 't':
 152                         size = SIZE_l;
 153                         break;
 154                 case 'L':
 155                         size = SIZE_L;
 156                         break;
 157                 case 'd': case 'i': case 'o': case 'u': case 'x':
 158                 case 'a': case 'e': case 'f': case 'g':
 159                 case 'A': case 'E': case 'F': case 'G': case 'X':
 160                 case 's': case 'c': case '[':
 161                 case 'S': case 'C':
 162                 case 'p': case 'n':
 163                         p--;
 164                         break;
 165                 default:
 166                         goto fmt_fail;
 167                 }
 168
 169                 t = *p;
 170
 171                 switch (t) {
 172                 case 'C':
 173                 case 'c':
 174                         if (width < 1) width = 1;
 175                 case 's':
 176                         if (size == SIZE_l) t &= ~0x20;
 177                 case 'd': case 'i': case 'o': case 'u': case 'x':
 178                 case 'a': case 'e': case 'f': case 'g':
 179                 case 'A': case 'E': case 'F': case 'G': case 'X':
 180                 case '[': case 'S':
 181                 case 'p': case 'n':
 182                         if (width < 1) width = 0;
 183                         break;
 184                 default:
 185                         goto fmt_fail;
 186                 }
 187
 188                 shlim(f, width);
 189
 190                 if (t != 'n') {
 191                         if (shgetc(f) < 0) goto input_fail;
 192                         shunget(f);
 193                 }
 194
 195                 switch (t) {
 196                 case 'n':
 197                         store_int(dest, size, pos);
 198                         /* do not increment match count, etc! */
 199                         continue;
 200                 case 'C':
 201                         wcs = dest;
 202                         st = (mbstate_t){ 0 };
 203                         while ((c=shgetc(f)) >= 0) {
 204                                 if (readwc(c, &wcs, &st) < 0)
 205                                         goto input_fail;
 206                         }
 207                         if (!mbsinit(&st)) goto input_fail;
 208                         if (shcnt(f) != width) goto match_fail;
 209                         break;
 210                 case 'c':
 211                         if (dest) {
 212                                 s = dest;
 213                                 while ((c=shgetc(f)) >= 0) *s++ = c;
 214                         } else {
 215                                 while (shgetc(f)>=0);
 216                         }
 217                         if (shcnt(f) < width) goto match_fail;
 218                         break;
 219                 case '[':
 220                         s = dest;
 221                         wcs = dest;
 222
 223                         if (*++p == '^') p++, invert = 1;
 224                         else invert = 0;
 225
 226                         unsigned char scanset[257];
 227                         memset(scanset, invert, sizeof scanset);
 228
 229                         scanset[0] = 0;
 230                         if (*p == '-') p++, scanset[1+'-'] = 1-invert;
 231                         else if (*p == ']') p++, scanset[1+']'] = 1-invert;
 232                         for (; *p != ']'; p++) {
 233                                 if (!*p) goto fmt_fail;
 234                                 if (*p=='-' && p[1] && p[1] != ']')
 235                                         for (c=p++[-1]; c<*p; c++)
 236                                                 scanset[1+c] = 1-invert;
 237                                 scanset[1+*p] = 1-invert;
 238                         }
 239
 240                         if (size == SIZE_l) {
 241                                 st = (mbstate_t){0};
 242                                 while (scanset[(c=shgetc(f))+1]) {
 243                                         if (readwc(c, &wcs, &st) < 0)
 244                                                 goto input_fail;
 245                                 }
 246                                 if (!mbsinit(&st)) goto input_fail;
 247                                 s = 0;
 248                         } else if (s) {
 249                                 while (scanset[(c=shgetc(f))+1])
 250                                         *s++ = c;
 251                                 wcs = 0;
 252                         } else {
 253                                 while (scanset[(c=shgetc(f))+1]);
 254                         }
 255                         shunget(f);
 256                         if (!shcnt(f)) goto match_fail;
 257                         if (s) *s = 0;
 258                         if (wcs) *wcs = 0;
 259                         break;
 260                 default:
 261                         shlim(f, 0);
 262                         while (isspace(shgetc(f)));
 263                         shunget(f);
 264                         pos += shcnt(f);
 265                         shlim(f, width);
 266                         if (shgetc(f) < 0) goto input_fail;
 267                         shunget(f);
 268                 }
 269
 270                 switch (t) {
 271                 case 'p':
 272                 case 'X':
 273                 case 'x':
 274                         base = 16;
 275                         goto int_common;
 276                 case 'o':
 277                         base = 8;
 278                         goto int_common;
 279                 case 'd':
 280                 case 'u':
 281                         base = 10;
 282                         goto int_common;
 283                 case 'i':
 284                         base = 0;
 285                 int_common:
 286                         x = __intscan(f, base, 0, ULLONG_MAX);
 287                         if (!shcnt(f)) goto match_fail;
 288                         if (t=='p' && dest) *(void **)dest = (void *)(uintptr_t)x;
 289                         else store_int(dest, size, x);
 290                         break;
 291                 case 'a': case 'A':
 292                 case 'e': case 'E':
 293                 case 'f': case 'F':
 294                 case 'g': case 'G':
 295                         y = __floatscan(f, size, 0);
 296                         if (!shcnt(f)) goto match_fail;
 297                         if (dest) switch (size) {
 298                         case SIZE_def:
 299                                 *(float *)dest = y;
 300                                 break;
 301                         case SIZE_l:
 302                                 *(double *)dest = y;
 303                                 break;
 304                         case SIZE_L:
 305                                 *(long double *)dest = y;
 306                                 break;
 307                         }
 308                         break;
 309                 case 'S':
 310                         wcs = dest;
 311                         st = (mbstate_t){ 0 };
 312                         while (!isspace(c=shgetc(f)) && c!=EOF) {
 313                                 if (readwc(c, &wcs, &st) < 0)
 314                                         goto input_fail;
 315                         }
 316                         shunget(f);
 317                         if (!mbsinit(&st)) goto input_fail;
 318                         if (dest) *wcs++ = 0;
 319                         break;
 320                 case 's':
 321                         if (dest) {
 322                                 s = dest;
 323                                 while (!isspace(c=shgetc(f)) && c!=EOF)
 324                                         *s++ = c;
 325                                 *s = 0;
 326                         } else {
 327                                 while (!isspace(c=shgetc(f)) && c!=EOF);
 328                         }
 329                         shunget(f);
 330                         break;
 331                 }
 332
 333                 pos += shcnt(f);
 334                 if (dest) matches++;
 335         }
 336         if (0) {
 337 fmt_fail:
 338 input_fail:
 339                 if (!matches) matches--;
 340         }
 341 match_fail:
 342         FUNLOCK(f);
 343         return matches;
 344 }