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
  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 in_set(const wchar_t *set, int c)
  60 {
  61         int j;
  62         const wchar_t *p = set;
  63         if (*p == '-') {
  64                 if (c=='-') return 1;
  65                 p++;
  66         } else if (*p == ']') {
  67                 if (c==']') return 1;
  68                 p++;
  69         }
  70         for (; *p && *p != ']'; p++) {
  71                 if (*p=='-' && p[1] && p[1] != ']')
  72                         for (j=p++[-1]; j<*p; j++)
  73                                 if (c==j) return 1;
  74                 if (c==*p) return 1;
  75         }
  76         return 0;
  77 }
  78
  79 #if 1
  80 #undef getwc
  81 #define getwc(f) \
  82         ((f)->rpos < (f)->rend && *(f)->rpos < 128 ? *(f)->rpos++ : (getwc)(f))
  83
  84 #undef ungetwc
  85 #define ungetwc(c,f) \
  86         ((f)->rend && (c)<128U ? *--(f)->rpos : ungetwc((c),(f)))
  87 #endif
  88
  89 int vfwscanf(FILE *restrict f, const wchar_t *restrict fmt, va_list ap)
  90 {
  91         int width;
  92         int size;
  93         int alloc;
  94         const wchar_t *p;
  95         int c, t;
  96         char *s;
  97         wchar_t *wcs;
  98         void *dest=NULL;
  99         int invert;
 100         int matches=0;
 101         off_t pos = 0, cnt;
 102         static const char size_pfx[][3] = { "hh", "h", "", "l", "L", "ll" };
 103         char tmp[3*sizeof(int)+10];
 104
 105         FLOCK(f);
 106
 107         for (p=fmt; *p; p++) {
 108
 109                 if (iswspace(*p)) {
 110                         while (iswspace(p[1])) p++;
 111                         while (iswspace((c=getwc(f)))) pos++;
 112                         ungetwc(c, f);
 113                         continue;
 114                 }
 115                 if (*p != '%' || p[1] == '%') {
 116                         p += *p=='%';
 117                         c = getwc(f);
 118                         if (c!=*p) {
 119                                 ungetwc(c, f);
 120                                 if (c<0) goto input_fail;
 121                                 goto match_fail;
 122                         }
 123                         pos++;
 124                         continue;
 125                 }
 126
 127                 p++;
 128                 if (*p=='*') {
 129                         dest = 0; p++;
 130                 } else if (iswdigit(*p) && p[1]=='$') {
 131                         dest = arg_n(ap, *p-'0'); p+=2;
 132                 } else {
 133                         dest = va_arg(ap, void *);
 134                 }
 135
 136                 for (width=0; iswdigit(*p); p++) {
 137                         width = 10*width + *p - '0';
 138                 }
 139
 140                 if (*p=='m') {
 141                         alloc = 1;
 142                         p++;
 143                 } else {
 144                         alloc = 0;
 145                 }
 146
 147                 size = SIZE_def;
 148                 switch (*p++) {
 149                 case 'h':
 150                         if (*p == 'h') p++, size = SIZE_hh;
 151                         else size = SIZE_h;
 152                         break;
 153                 case 'l':
 154                         if (*p == 'l') p++, size = SIZE_ll;
 155                         else size = SIZE_l;
 156                         break;
 157                 case 'j':
 158                         size = SIZE_ll;
 159                         break;
 160                 case 'z':
 161                 case 't':
 162                         size = SIZE_l;
 163                         break;
 164                 case 'L':
 165                         size = SIZE_L;
 166                         break;
 167                 case 'd': case 'i': case 'o': case 'u': case 'x':
 168                 case 'a': case 'e': case 'f': case 'g':
 169                 case 'A': case 'E': case 'F': case 'G': case 'X':
 170                 case 's': case 'c': case '[':
 171                 case 'S': case 'C':
 172                 case 'p': case 'n':
 173                         p--;
 174                         break;
 175                 default:
 176                         goto fmt_fail;
 177                 }
 178
 179                 t = *p;
 180
 181                 /* Transform ls,lc -> S,C */
 182                 if (size==SIZE_l && (t&15)==3) t&=~32;
 183
 184                 if (t != 'n') {
 185                         if (t != '[' && (t|32) != 'c')
 186                                 while (iswspace((c=getwc(f)))) pos++;
 187                         else
 188                                 c=getwc(f);
 189                         if (c < 0) goto input_fail;
 190                         ungetwc(c, f);
 191                 }
 192
 193                 switch (t) {
 194                 case 'n':
 195                         store_int(dest, size, pos);
 196                         /* do not increment match count, etc! */
 197                         continue;
 198
 199                 case 'c':
 200                         if (width < 1) width = 1;
 201                         s = dest;
 202                         for (; width && (c=getwc(f)) >= 0; width--) {
 203                                 int l = wctomb(s?s:tmp, c);
 204                                 if (l<0) goto input_fail;
 205                                 if (s) s+=l;
 206                                 pos++;
 207                         }
 208                         if (width) goto match_fail;
 209                         break;
 210
 211                 case 'C':
 212                         if (width < 1) width = 1;
 213                         wcs = dest;
 214                         for (; width && (c=getwc(f)) >= 0; width--)
 215                                 pos++, wcs && (*wcs++ = c);
 216                         if (width) goto match_fail;
 217                         break;
 218
 219                 case 's':
 220                         if (width < 1) width = -1;
 221                         s = dest;
 222                         while (width && !iswspace(c=getwc(f)) && c!=EOF) {
 223                                 int l = wctomb(s?s:tmp, c);
 224                                 if (l<0) goto input_fail;
 225                                 if (s) s+=l;
 226                                 pos++;
 227                                 width-=(width>0);
 228                         }
 229                         if (width) ungetwc(c, f);
 230                         if (s) *s = 0;
 231                         break;
 232
 233                 case 'S':
 234                         wcs = dest;
 235                         if (width < 1) width = -1;
 236                         while (width && !iswspace(c=getwc(f)) && c!=EOF)
 237                                 width-=(width>0), pos++, *wcs++ = c;
 238                         if (width) ungetwc(c, f);
 239                         if (wcs) *wcs = 0;
 240                         break;
 241
 242                 case '[':
 243                         s = (size == SIZE_def) ? dest : 0;
 244                         wcs = (size == SIZE_l) ? dest : 0;
 245
 246                         if (*++p == '^') p++, invert = 1;
 247                         else invert = 0;
 248
 249                         int gotmatch = 0;
 250
 251                         if (width < 1) width = -1;
 252
 253                         while (width) {
 254                                 if ((c=getwc(f))<0) break;
 255                                 if (in_set(p, c) == invert)
 256                                         break;
 257                                 if (wcs) {
 258                                         *wcs++ = c;
 259                                 } else if (size != SIZE_l) {
 260                                         int l = wctomb(s?s:tmp, c);
 261                                         if (l<0) goto input_fail;
 262                                         if (s) s+=l;
 263                                 }
 264                                 pos++;
 265                                 width-=(width>0);
 266                                 gotmatch=1;
 267                         }
 268                         if (width) ungetwc(c, f);
 269
 270                         if (!gotmatch) goto match_fail;
 271
 272                         if (*p==']') p++;
 273                         while (*p!=']') {
 274                                 if (!*p) goto fmt_fail;
 275                                 p++;
 276                         }
 277
 278                         if (wcs) *wcs++ = 0;
 279                         if (s) *s++ = 0;
 280                         break;
 281
 282                 case 'd': case 'i': case 'o': case 'u': case 'x':
 283                 case 'a': case 'e': case 'f': case 'g':
 284                 case 'A': case 'E': case 'F': case 'G': case 'X':
 285                 case 'p':
 286                         if (width < 1) width = 0;
 287                         snprintf(tmp, sizeof tmp, "%.*s%.0d%s%c%%lln",
 288                                 1+!dest, "%*", width, size_pfx[size+2], t);
 289                         cnt = 0;
 290                         if (fscanf(f, tmp, dest?dest:&cnt, &cnt) == -1)
 291                                 goto input_fail;
 292                         else if (!cnt)
 293                                 goto match_fail;
 294                         pos += cnt;
 295                         break;
 296                 default:
 297                         goto fmt_fail;
 298                 }
 299
 300                 if (dest) matches++;
 301         }
 302         if (0) {
 303 fmt_fail:
 304 input_fail:
 305                 if (!matches) matches--;
 306         }
 307 match_fail:
 308         FUNLOCK(f);
 309         return matches;
 310 }