nsz Git - musl/blob - src/stdio/vfwscanf.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 #include "libc.h"
  18
  19 #define SIZE_hh -2
  20 #define SIZE_h  -1
  21 #define SIZE_def 0
  22 #define SIZE_l   1
  23 #define SIZE_L   2
  24 #define SIZE_ll  3
  25
  26 static void store_int(void *dest, int size, unsigned long long i)
  27 {
  28         if (!dest) return;
  29         switch (size) {
  30         case SIZE_hh:
  31                 *(char *)dest = i;
  32                 break;
  33         case SIZE_h:
  34                 *(short *)dest = i;
  35                 break;
  36         case SIZE_def:
  37                 *(int *)dest = i;
  38                 break;
  39         case SIZE_l:
  40                 *(long *)dest = i;
  41                 break;
  42         case SIZE_ll:
  43                 *(long long *)dest = i;
  44                 break;
  45         }
  46 }
  47
  48 static void *arg_n(va_list ap, unsigned int n)
  49 {
  50         void *p;
  51         unsigned int i;
  52         va_list ap2;
  53         va_copy(ap2, ap);
  54         for (i=n; i>1; i--) va_arg(ap2, void *);
  55         p = va_arg(ap2, void *);
  56         va_end(ap2);
  57         return p;
  58 }
  59
  60 static int in_set(const wchar_t *set, int c)
  61 {
  62         int j;
  63         const wchar_t *p = set;
  64         if (*p == '-') {
  65                 if (c=='-') return 1;
  66                 p++;
  67         } else if (*p == ']') {
  68                 if (c==']') return 1;
  69                 p++;
  70         }
  71         for (; *p && *p != ']'; p++) {
  72                 if (*p=='-' && p[1] && p[1] != ']')
  73                         for (j=p++[-1]; j<*p; j++)
  74                                 if (c==j) return 1;
  75                 if (c==*p) return 1;
  76         }
  77         return 0;
  78 }
  79
  80 #if 1
  81 #undef getwc
  82 #define getwc(f) \
  83         ((f)->rpos < (f)->rend && *(f)->rpos < 128 ? *(f)->rpos++ : (getwc)(f))
  84
  85 #undef ungetwc
  86 #define ungetwc(c,f) \
  87         ((f)->rend && (c)<128U ? *--(f)->rpos : ungetwc((c),(f)))
  88 #endif
  89
  90 int vfwscanf(FILE *restrict f, const wchar_t *restrict fmt, va_list ap)
  91 {
  92         int width;
  93         int size;
  94         int alloc;
  95         const wchar_t *p;
  96         int c, t;
  97         char *s;
  98         wchar_t *wcs;
  99         void *dest=NULL;
 100         int invert;
 101         int matches=0;
 102         off_t pos = 0, cnt;
 103         static const char size_pfx[][3] = { "hh", "h", "", "l", "L", "ll" };
 104         char tmp[3*sizeof(int)+10];
 105         const wchar_t *set;
 106         size_t i, k;
 107
 108         FLOCK(f);
 109
 110         for (p=fmt; *p; p++) {
 111
 112                 alloc = 0;
 113
 114                 if (iswspace(*p)) {
 115                         while (iswspace(p[1])) p++;
 116                         while (iswspace((c=getwc(f)))) pos++;
 117                         ungetwc(c, f);
 118                         continue;
 119                 }
 120                 if (*p != '%' || p[1] == '%') {
 121                         p += *p=='%';
 122                         c = getwc(f);
 123                         if (c!=*p) {
 124                                 ungetwc(c, f);
 125                                 if (c<0) goto input_fail;
 126                                 goto match_fail;
 127                         }
 128                         pos++;
 129                         continue;
 130                 }
 131
 132                 p++;
 133                 if (*p=='*') {
 134                         dest = 0; p++;
 135                 } else if (iswdigit(*p) && p[1]=='$') {
 136                         dest = arg_n(ap, *p-'0'); p+=2;
 137                 } else {
 138                         dest = va_arg(ap, void *);
 139                 }
 140
 141                 for (width=0; iswdigit(*p); p++) {
 142                         width = 10*width + *p - '0';
 143                 }
 144
 145                 if (*p=='m') {
 146                         alloc = !!dest;
 147                         p++;
 148                 } else {
 149                         alloc = 0;
 150                 }
 151
 152                 size = SIZE_def;
 153                 switch (*p++) {
 154                 case 'h':
 155                         if (*p == 'h') p++, size = SIZE_hh;
 156                         else size = SIZE_h;
 157                         break;
 158                 case 'l':
 159                         if (*p == 'l') p++, size = SIZE_ll;
 160                         else size = SIZE_l;
 161                         break;
 162                 case 'j':
 163                         size = SIZE_ll;
 164                         break;
 165                 case 'z':
 166                 case 't':
 167                         size = SIZE_l;
 168                         break;
 169                 case 'L':
 170                         size = SIZE_L;
 171                         break;
 172                 case 'd': case 'i': case 'o': case 'u': case 'x':
 173                 case 'a': case 'e': case 'f': case 'g':
 174                 case 'A': case 'E': case 'F': case 'G': case 'X':
 175                 case 's': case 'c': case '[':
 176                 case 'S': case 'C':
 177                 case 'p': case 'n':
 178                         p--;
 179                         break;
 180                 default:
 181                         goto fmt_fail;
 182                 }
 183
 184                 t = *p;
 185
 186                 /* Transform S,C -> ls,lc */
 187                 if ((t&0x2f)==3) {
 188                         size = SIZE_l;
 189                         t |= 32;
 190                 }
 191
 192                 if (t != 'n') {
 193                         if (t != '[' && (t|32) != 'c')
 194                                 while (iswspace((c=getwc(f)))) pos++;
 195                         else
 196                                 c=getwc(f);
 197                         if (c < 0) goto input_fail;
 198                         ungetwc(c, f);
 199                 }
 200
 201                 switch (t) {
 202                 case 'n':
 203                         store_int(dest, size, pos);
 204                         /* do not increment match count, etc! */
 205                         continue;
 206
 207                 case 's':
 208                 case 'c':
 209                 case '[':
 210                         if (t == 'c') {
 211                                 if (width<1) width = 1;
 212                                 invert = 1;
 213                                 set = L"";
 214                         } else if (t == 's') {
 215                                 invert = 1;
 216                                 set = (const wchar_t[]){
 217                                         ' ', '\t', '\n', '\r', 11, 12,  0x0085,
 218                                         0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005,
 219                                         0x2006, 0x2008, 0x2009, 0x200a,
 220                                         0x2028, 0x2029, 0x205f, 0x3000, 0 };
 221                         } else {
 222                                 if (*++p == '^') p++, invert = 1;
 223                                 else invert = 0;
 224                                 set = p;
 225                                 if (*p==']') p++;
 226                                 while (*p!=']') {
 227                                         if (!*p) goto fmt_fail;
 228                                         p++;
 229                                 }
 230                         }
 231
 232                         s = (size == SIZE_def) ? dest : 0;
 233                         wcs = (size == SIZE_l) ? dest : 0;
 234
 235                         int gotmatch = 0;
 236
 237                         if (width < 1) width = -1;
 238
 239                         i = 0;
 240                         if (alloc) {
 241                                 k = t=='c' ? width+1U : 31;
 242                                 if (size == SIZE_l) {
 243                                         wcs = malloc(k*sizeof(wchar_t));
 244                                         if (!wcs) goto alloc_fail;
 245                                 } else {
 246                                         s = malloc(k);
 247                                         if (!s) goto alloc_fail;
 248                                 }
 249                         }
 250                         while (width) {
 251                                 if ((c=getwc(f))<0) break;
 252                                 if (in_set(set, c) == invert)
 253                                         break;
 254                                 if (wcs) {
 255                                         wcs[i++] = c;
 256                                         if (alloc && i==k) {
 257                                                 k += k+1;
 258                                                 wchar_t *tmp = realloc(wcs, k*sizeof(wchar_t));
 259                                                 if (!tmp) goto alloc_fail;
 260                                                 wcs = tmp;
 261                                         }
 262                                 } else if (size != SIZE_l) {
 263                                         int l = wctomb(s?s+i:tmp, c);
 264                                         if (l<0) goto input_fail;
 265                                         i += l;
 266                                         if (alloc && i > k-4) {
 267                                                 k += k+1;
 268                                                 char *tmp = realloc(s, k);
 269                                                 if (!tmp) goto alloc_fail;
 270                                                 s = tmp;
 271                                         }
 272                                 }
 273                                 pos++;
 274                                 width-=(width>0);
 275                                 gotmatch=1;
 276                         }
 277                         if (width) {
 278                                 ungetwc(c, f);
 279                                 if (t == 'c' || !gotmatch) goto match_fail;
 280                         }
 281
 282                         if (alloc) {
 283                                 if (size == SIZE_l) *(wchar_t **)dest = wcs;
 284                                 else *(char **)dest = s;
 285                         }
 286                         if (t != 'c') {
 287                                 if (wcs) wcs[i] = 0;
 288                                 if (s) s[i] = 0;
 289                         }
 290                         break;
 291
 292                 case 'd': case 'i': case 'o': case 'u': case 'x':
 293                 case 'a': case 'e': case 'f': case 'g':
 294                 case 'A': case 'E': case 'F': case 'G': case 'X':
 295                 case 'p':
 296                         if (width < 1) width = 0;
 297                         snprintf(tmp, sizeof tmp, "%.*s%.0d%s%c%%lln",
 298                                 1+!dest, "%*", width, size_pfx[size+2], t);
 299                         cnt = 0;
 300                         if (fscanf(f, tmp, dest?dest:&cnt, &cnt) == -1)
 301                                 goto input_fail;
 302                         else if (!cnt)
 303                                 goto match_fail;
 304                         pos += cnt;
 305                         break;
 306                 default:
 307                         goto fmt_fail;
 308                 }
 309
 310                 if (dest) matches++;
 311         }
 312         if (0) {
 313 fmt_fail:
 314 alloc_fail:
 315 input_fail:
 316                 if (!matches) matches--;
 317 match_fail:
 318                 if (alloc) {
 319                         free(s);
 320                         free(wcs);
 321                 }
 322         }
 323         FUNLOCK(f);
 324         return matches;
 325 }
 326
 327 weak_alias(vfwscanf,__isoc99_vfwscanf);