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
  10 #include "stdio_impl.h"
  11 #include "shgetc.h"
  12 #include "intscan.h"
  13 #include "floatscan.h"
  14 #include "libc.h"
  15
  16 #define SIZE_hh -2
  17 #define SIZE_h  -1
  18 #define SIZE_def 0
  19 #define SIZE_l   1
  20 #define SIZE_L   2
  21 #define SIZE_ll  3
  22
  23 static void store_int(void *dest, int size, unsigned long long i)
  24 {
  25         if (!dest) return;
  26         switch (size) {
  27         case SIZE_hh:
  28                 *(char *)dest = i;
  29                 break;
  30         case SIZE_h:
  31                 *(short *)dest = i;
  32                 break;
  33         case SIZE_def:
  34                 *(int *)dest = i;
  35                 break;
  36         case SIZE_l:
  37                 *(long *)dest = i;
  38                 break;
  39         case SIZE_ll:
  40                 *(long long *)dest = i;
  41                 break;
  42         }
  43 }
  44
  45 static void *arg_n(va_list ap, unsigned int n)
  46 {
  47         void *p;
  48         unsigned int i;
  49         va_list ap2;
  50         va_copy(ap2, ap);
  51         for (i=n; i>1; i--) va_arg(ap2, void *);
  52         p = va_arg(ap2, void *);
  53         va_end(ap2);
  54         return p;
  55 }
  56
  57 static int in_set(const wchar_t *set, int c)
  58 {
  59         int j;
  60         const wchar_t *p = set;
  61         if (*p == '-') {
  62                 if (c=='-') return 1;
  63                 p++;
  64         } else if (*p == ']') {
  65                 if (c==']') return 1;
  66                 p++;
  67         }
  68         for (; *p && *p != ']'; p++) {
  69                 if (*p=='-' && p[1] && p[1] != ']')
  70                         for (j=p++[-1]; j<*p; j++)
  71                                 if (c==j) return 1;
  72                 if (c==*p) return 1;
  73         }
  74         return 0;
  75 }
  76
  77 #if 1
  78 #undef getwc
  79 #define getwc(f) \
  80         ((f)->rpos < (f)->rend && *(f)->rpos < 128 ? *(f)->rpos++ : (getwc)(f))
  81
  82 #undef ungetwc
  83 #define ungetwc(c,f) \
  84         ((f)->rend && (c)<128U ? *--(f)->rpos : ungetwc((c),(f)))
  85 #endif
  86
  87 int vfwscanf(FILE *restrict f, const wchar_t *restrict fmt, va_list ap)
  88 {
  89         int width;
  90         int size;
  91         int alloc;
  92         const wchar_t *p;
  93         int c, t;
  94         char *s;
  95         wchar_t *wcs;
  96         void *dest=NULL;
  97         int invert;
  98         int matches=0;
  99         off_t pos = 0, cnt;
 100         static const char size_pfx[][3] = { "hh", "h", "", "l", "L", "ll" };
 101         char tmp[3*sizeof(int)+10];
 102         const wchar_t *set;
 103         size_t i, k;
 104
 105         FLOCK(f);
 106
 107         for (p=fmt; *p; p++) {
 108
 109                 alloc = 0;
 110
 111                 if (iswspace(*p)) {
 112                         while (iswspace(p[1])) p++;
 113                         while (iswspace((c=getwc(f)))) pos++;
 114                         ungetwc(c, f);
 115                         continue;
 116                 }
 117                 if (*p != '%' || p[1] == '%') {
 118                         p += *p=='%';
 119                         c = getwc(f);
 120                         if (c!=*p) {
 121                                 ungetwc(c, f);
 122                                 if (c<0) goto input_fail;
 123                                 goto match_fail;
 124                         }
 125                         pos++;
 126                         continue;
 127                 }
 128
 129                 p++;
 130                 if (*p=='*') {
 131                         dest = 0; p++;
 132                 } else if (iswdigit(*p) && p[1]=='$') {
 133                         dest = arg_n(ap, *p-'0'); p+=2;
 134                 } else {
 135                         dest = va_arg(ap, void *);
 136                 }
 137
 138                 for (width=0; iswdigit(*p); p++) {
 139                         width = 10*width + *p - '0';
 140                 }
 141
 142                 if (*p=='m') {
 143                         wcs = 0;
 144                         s = 0;
 145                         alloc = !!dest;
 146                         p++;
 147                 } else {
 148                         alloc = 0;
 149                 }
 150
 151                 size = SIZE_def;
 152                 switch (*p++) {
 153                 case 'h':
 154                         if (*p == 'h') p++, size = SIZE_hh;
 155                         else size = SIZE_h;
 156                         break;
 157                 case 'l':
 158                         if (*p == 'l') p++, size = SIZE_ll;
 159                         else size = SIZE_l;
 160                         break;
 161                 case 'j':
 162                         size = SIZE_ll;
 163                         break;
 164                 case 'z':
 165                 case 't':
 166                         size = SIZE_l;
 167                         break;
 168                 case 'L':
 169                         size = SIZE_L;
 170                         break;
 171                 case 'd': case 'i': case 'o': case 'u': case 'x':
 172                 case 'a': case 'e': case 'f': case 'g':
 173                 case 'A': case 'E': case 'F': case 'G': case 'X':
 174                 case 's': case 'c': case '[':
 175                 case 'S': case 'C':
 176                 case 'p': case 'n':
 177                         p--;
 178                         break;
 179                 default:
 180                         goto fmt_fail;
 181                 }
 182
 183                 t = *p;
 184
 185                 /* Transform S,C -> ls,lc */
 186                 if ((t&0x2f)==3) {
 187                         size = SIZE_l;
 188                         t |= 32;
 189                 }
 190
 191                 if (t != 'n') {
 192                         if (t != '[' && (t|32) != 'c')
 193                                 while (iswspace((c=getwc(f)))) pos++;
 194                         else
 195                                 c=getwc(f);
 196                         if (c < 0) goto input_fail;
 197                         ungetwc(c, f);
 198                 }
 199
 200                 switch (t) {
 201                 case 'n':
 202                         store_int(dest, size, pos);
 203                         /* do not increment match count, etc! */
 204                         continue;
 205
 206                 case 's':
 207                 case 'c':
 208                 case '[':
 209                         if (t == 'c') {
 210                                 if (width<1) width = 1;
 211                                 invert = 1;
 212                                 set = L"";
 213                         } else if (t == 's') {
 214                                 invert = 1;
 215                                 set = (const wchar_t[]){
 216                                         ' ', '\t', '\n', '\r', 11, 12,  0x0085,
 217                                         0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005,
 218                                         0x2006, 0x2008, 0x2009, 0x200a,
 219                                         0x2028, 0x2029, 0x205f, 0x3000, 0 };
 220                         } else {
 221                                 if (*++p == '^') p++, invert = 1;
 222                                 else invert = 0;
 223                                 set = p;
 224                                 if (*p==']') p++;
 225                                 while (*p!=']') {
 226                                         if (!*p) goto fmt_fail;
 227                                         p++;
 228                                 }
 229                         }
 230
 231                         s = (size == SIZE_def) ? dest : 0;
 232                         wcs = (size == SIZE_l) ? dest : 0;
 233
 234                         int gotmatch = 0;
 235
 236                         if (width < 1) width = -1;
 237
 238                         i = 0;
 239                         if (alloc) {
 240                                 k = t=='c' ? width+1U : 31;
 241                                 if (size == SIZE_l) {
 242                                         wcs = malloc(k*sizeof(wchar_t));
 243                                         if (!wcs) goto alloc_fail;
 244                                 } else {
 245                                         s = malloc(k);
 246                                         if (!s) goto alloc_fail;
 247                                 }
 248                         }
 249                         while (width) {
 250                                 if ((c=getwc(f))<0) break;
 251                                 if (in_set(set, c) == invert)
 252                                         break;
 253                                 if (wcs) {
 254                                         wcs[i++] = c;
 255                                         if (alloc && i==k) {
 256                                                 k += k+1;
 257                                                 wchar_t *tmp = realloc(wcs, k*sizeof(wchar_t));
 258                                                 if (!tmp) goto alloc_fail;
 259                                                 wcs = tmp;
 260                                         }
 261                                 } else if (size != SIZE_l) {
 262                                         int l = wctomb(s?s+i:tmp, c);
 263                                         if (l<0) goto input_fail;
 264                                         i += l;
 265                                         if (alloc && i > k-4) {
 266                                                 k += k+1;
 267                                                 char *tmp = realloc(s, k);
 268                                                 if (!tmp) goto alloc_fail;
 269                                                 s = tmp;
 270                                         }
 271                                 }
 272                                 pos++;
 273                                 width-=(width>0);
 274                                 gotmatch=1;
 275                         }
 276                         if (width) {
 277                                 ungetwc(c, f);
 278                                 if (t == 'c' || !gotmatch) goto match_fail;
 279                         }
 280
 281                         if (alloc) {
 282                                 if (size == SIZE_l) *(wchar_t **)dest = wcs;
 283                                 else *(char **)dest = s;
 284                         }
 285                         if (t != 'c') {
 286                                 if (wcs) wcs[i] = 0;
 287                                 if (s) s[i] = 0;
 288                         }
 289                         break;
 290
 291                 case 'd': case 'i': case 'o': case 'u': case 'x':
 292                 case 'a': case 'e': case 'f': case 'g':
 293                 case 'A': case 'E': case 'F': case 'G': case 'X':
 294                 case 'p':
 295                         if (width < 1) width = 0;
 296                         snprintf(tmp, sizeof tmp, "%.*s%.0d%s%c%%lln",
 297                                 1+!dest, "%*", width, size_pfx[size+2], t);
 298                         cnt = 0;
 299                         if (fscanf(f, tmp, dest?dest:&cnt, &cnt) == -1)
 300                                 goto input_fail;
 301                         else if (!cnt)
 302                                 goto match_fail;
 303                         pos += cnt;
 304                         break;
 305                 default:
 306                         goto fmt_fail;
 307                 }
 308
 309                 if (dest) matches++;
 310         }
 311         if (0) {
 312 fmt_fail:
 313 alloc_fail:
 314 input_fail:
 315                 if (!matches) matches--;
 316 match_fail:
 317                 if (alloc) {
 318                         free(s);
 319                         free(wcs);
 320                 }
 321         }
 322         FUNLOCK(f);
 323         return matches;
 324 }
 325
 326 weak_alias(vfwscanf,__isoc99_vfwscanf);