[e16e8f2] | 1 | /* |
---|
| 2 | * vsscanf.c |
---|
| 3 | * |
---|
| 4 | * vsscanf(), from which the rest of the scanf() |
---|
| 5 | * family is built |
---|
| 6 | */ |
---|
| 7 | |
---|
| 8 | #include <ctype.h> |
---|
| 9 | #include <stdarg.h> |
---|
| 10 | #include <stddef.h> |
---|
| 11 | #include <inttypes.h> |
---|
| 12 | #include <string.h> |
---|
| 13 | #include <limits.h> |
---|
| 14 | #include <stdio.h> |
---|
| 15 | #include <sys/bitops.h> |
---|
| 16 | |
---|
| 17 | #ifndef LONG_BIT |
---|
| 18 | #define LONG_BIT (CHAR_BIT*sizeof(long)) |
---|
| 19 | #endif |
---|
| 20 | |
---|
| 21 | enum flags { |
---|
| 22 | FL_SPLAT = 0x01, /* Drop the value, do not assign */ |
---|
| 23 | FL_WIDTH = 0x02, /* Field width specified */ |
---|
| 24 | FL_MINUS = 0x04, /* Negative number */ |
---|
| 25 | }; |
---|
| 26 | |
---|
| 27 | enum ranks { |
---|
| 28 | rank_char = -2, |
---|
| 29 | rank_short = -1, |
---|
| 30 | rank_int = 0, |
---|
| 31 | rank_long = 1, |
---|
| 32 | rank_longlong = 2, |
---|
| 33 | rank_ptr = INT_MAX /* Special value used for pointers */ |
---|
| 34 | }; |
---|
| 35 | |
---|
| 36 | #define MIN_RANK rank_char |
---|
| 37 | #define MAX_RANK rank_longlong |
---|
| 38 | |
---|
| 39 | #define INTMAX_RANK rank_longlong |
---|
| 40 | #define SIZE_T_RANK rank_long |
---|
| 41 | #define PTRDIFF_T_RANK rank_long |
---|
| 42 | |
---|
| 43 | enum bail { |
---|
| 44 | bail_none = 0, /* No error condition */ |
---|
| 45 | bail_eof, /* Hit EOF */ |
---|
| 46 | bail_err /* Conversion mismatch */ |
---|
| 47 | }; |
---|
| 48 | |
---|
| 49 | int vsscanf(const char *buffer, const char *format, va_list ap) |
---|
| 50 | { |
---|
| 51 | const char *p = format; |
---|
| 52 | char ch; |
---|
| 53 | const char *q = buffer; |
---|
| 54 | const char *qq; |
---|
| 55 | uintmax_t val = 0; |
---|
| 56 | int rank = rank_int; /* Default rank */ |
---|
| 57 | unsigned int width = UINT_MAX; |
---|
| 58 | int base; |
---|
| 59 | enum flags flags = 0; |
---|
| 60 | enum { |
---|
| 61 | st_normal, /* Ground state */ |
---|
| 62 | st_flags, /* Special flags */ |
---|
| 63 | st_width, /* Field width */ |
---|
| 64 | st_modifiers, /* Length or conversion modifiers */ |
---|
| 65 | st_match_init, /* Initial state of %[ sequence */ |
---|
| 66 | st_match, /* Main state of %[ sequence */ |
---|
| 67 | st_match_range, /* After - in a %[ sequence */ |
---|
| 68 | } state = st_normal; |
---|
| 69 | char *sarg = NULL; /* %s %c or %[ string argument */ |
---|
| 70 | enum bail bail = bail_none; |
---|
| 71 | int converted = 0; /* Successful conversions */ |
---|
| 72 | unsigned long matchmap[((1 << CHAR_BIT) + (LONG_BIT - 1)) / LONG_BIT]; |
---|
| 73 | int matchinv = 0; /* Is match map inverted? */ |
---|
| 74 | unsigned char range_start = 0; |
---|
| 75 | |
---|
| 76 | while ((ch = *p++) && !bail) { |
---|
| 77 | switch (state) { |
---|
| 78 | case st_normal: |
---|
| 79 | if (ch == '%') { |
---|
| 80 | state = st_flags; |
---|
| 81 | flags = 0; |
---|
| 82 | rank = rank_int; |
---|
| 83 | width = UINT_MAX; |
---|
| 84 | } else if (isspace((unsigned char)ch)) { |
---|
| 85 | q = skipspace(q); |
---|
| 86 | } else { |
---|
| 87 | if (*q == ch) |
---|
| 88 | q++; |
---|
| 89 | else |
---|
| 90 | bail = bail_err; /* Match failure */ |
---|
| 91 | } |
---|
| 92 | break; |
---|
| 93 | |
---|
| 94 | case st_flags: |
---|
| 95 | switch (ch) { |
---|
| 96 | case '*': |
---|
| 97 | flags |= FL_SPLAT; |
---|
| 98 | break; |
---|
| 99 | case '0' ... '9': |
---|
| 100 | width = (ch - '0'); |
---|
| 101 | state = st_width; |
---|
| 102 | flags |= FL_WIDTH; |
---|
| 103 | break; |
---|
| 104 | default: |
---|
| 105 | state = st_modifiers; |
---|
| 106 | p--; /* Process this character again */ |
---|
| 107 | break; |
---|
| 108 | } |
---|
| 109 | break; |
---|
| 110 | |
---|
| 111 | case st_width: |
---|
| 112 | if (ch >= '0' && ch <= '9') { |
---|
| 113 | width = width * 10 + (ch - '0'); |
---|
| 114 | } else { |
---|
| 115 | state = st_modifiers; |
---|
| 116 | p--; /* Process this character again */ |
---|
| 117 | } |
---|
| 118 | break; |
---|
| 119 | |
---|
| 120 | case st_modifiers: |
---|
| 121 | switch (ch) { |
---|
| 122 | /* Length modifiers - nonterminal sequences */ |
---|
| 123 | case 'h': |
---|
| 124 | rank--; /* Shorter rank */ |
---|
| 125 | break; |
---|
| 126 | case 'l': |
---|
| 127 | rank++; /* Longer rank */ |
---|
| 128 | break; |
---|
| 129 | case 'j': |
---|
| 130 | rank = INTMAX_RANK; |
---|
| 131 | break; |
---|
| 132 | case 'z': |
---|
| 133 | rank = SIZE_T_RANK; |
---|
| 134 | break; |
---|
| 135 | case 't': |
---|
| 136 | rank = PTRDIFF_T_RANK; |
---|
| 137 | break; |
---|
| 138 | case 'L': |
---|
| 139 | case 'q': |
---|
| 140 | rank = rank_longlong; /* long double/long long */ |
---|
| 141 | break; |
---|
| 142 | |
---|
| 143 | default: |
---|
| 144 | /* Output modifiers - terminal sequences */ |
---|
| 145 | state = st_normal; /* Next state will be normal */ |
---|
| 146 | if (rank < MIN_RANK) /* Canonicalize rank */ |
---|
| 147 | rank = MIN_RANK; |
---|
| 148 | else if (rank > MAX_RANK) |
---|
| 149 | rank = MAX_RANK; |
---|
| 150 | |
---|
| 151 | switch (ch) { |
---|
| 152 | case 'P': /* Upper case pointer */ |
---|
| 153 | case 'p': /* Pointer */ |
---|
| 154 | #if 0 /* Enable this to allow null pointers by name */ |
---|
| 155 | q = skipspace(q); |
---|
| 156 | if (!isdigit((unsigned char)*q)) { |
---|
| 157 | static const char *const nullnames[] = |
---|
| 158 | { "null", "nul", "nil", "(null)", "(nul)", "(nil)", |
---|
| 159 | 0 }; |
---|
| 160 | const char *const *np; |
---|
| 161 | |
---|
| 162 | /* Check to see if it's a null pointer by name */ |
---|
| 163 | for (np = nullnames; *np; np++) { |
---|
| 164 | if (!strncasecmp(q, *np, strlen(*np))) { |
---|
| 165 | val = (uintmax_t) ((void *)NULL); |
---|
| 166 | goto set_integer; |
---|
| 167 | } |
---|
| 168 | } |
---|
| 169 | /* Failure */ |
---|
| 170 | bail = bail_err; |
---|
| 171 | break; |
---|
| 172 | } |
---|
| 173 | /* else */ |
---|
| 174 | #endif |
---|
| 175 | rank = rank_ptr; |
---|
| 176 | base = 0; |
---|
| 177 | goto scan_int; |
---|
| 178 | |
---|
| 179 | case 'i': /* Base-independent integer */ |
---|
| 180 | base = 0; |
---|
| 181 | goto scan_int; |
---|
| 182 | |
---|
| 183 | case 'd': /* Decimal integer */ |
---|
| 184 | base = 10; |
---|
| 185 | goto scan_int; |
---|
| 186 | |
---|
| 187 | case 'o': /* Octal integer */ |
---|
| 188 | base = 8; |
---|
| 189 | goto scan_int; |
---|
| 190 | |
---|
| 191 | case 'u': /* Unsigned decimal integer */ |
---|
| 192 | base = 10; |
---|
| 193 | goto scan_int; |
---|
| 194 | |
---|
| 195 | case 'x': /* Hexadecimal integer */ |
---|
| 196 | case 'X': |
---|
| 197 | base = 16; |
---|
| 198 | goto scan_int; |
---|
| 199 | |
---|
| 200 | case 'n': /* Number of characters consumed */ |
---|
| 201 | val = (q - buffer); |
---|
| 202 | goto set_integer; |
---|
| 203 | |
---|
| 204 | scan_int: |
---|
| 205 | q = skipspace(q); |
---|
| 206 | if (!*q) { |
---|
| 207 | bail = bail_eof; |
---|
| 208 | break; |
---|
| 209 | } |
---|
| 210 | val = strntoumax(q, (char **)&qq, base, width); |
---|
| 211 | if (qq == q) { |
---|
| 212 | bail = bail_err; |
---|
| 213 | break; |
---|
| 214 | } |
---|
| 215 | q = qq; |
---|
| 216 | converted++; |
---|
| 217 | /* fall through */ |
---|
| 218 | |
---|
| 219 | set_integer: |
---|
| 220 | if (!(flags & FL_SPLAT)) { |
---|
| 221 | switch (rank) { |
---|
| 222 | case rank_char: |
---|
| 223 | *va_arg(ap, unsigned char *) = (unsigned char)val; |
---|
| 224 | break; |
---|
| 225 | case rank_short: |
---|
| 226 | *va_arg(ap, unsigned short *) = (unsigned short)val; |
---|
| 227 | break; |
---|
| 228 | case rank_int: |
---|
| 229 | *va_arg(ap, unsigned int *) = (unsigned int)val; |
---|
| 230 | break; |
---|
| 231 | case rank_long: |
---|
| 232 | *va_arg(ap, unsigned long *) = (unsigned long)val; |
---|
| 233 | break; |
---|
| 234 | case rank_longlong: |
---|
| 235 | *va_arg(ap, unsigned long long *) = |
---|
| 236 | (unsigned long long)val; |
---|
| 237 | break; |
---|
| 238 | case rank_ptr: |
---|
| 239 | *va_arg(ap, void **) = (void *)(uintptr_t) val; |
---|
| 240 | break; |
---|
| 241 | } |
---|
| 242 | } |
---|
| 243 | break; |
---|
| 244 | |
---|
| 245 | case 'c': /* Character */ |
---|
| 246 | width = (flags & FL_WIDTH) ? width : 1; /* Default width == 1 */ |
---|
| 247 | sarg = va_arg(ap, char *); |
---|
| 248 | while (width--) { |
---|
| 249 | if (!*q) { |
---|
| 250 | bail = bail_eof; |
---|
| 251 | break; |
---|
| 252 | } |
---|
| 253 | *sarg++ = *q++; |
---|
| 254 | } |
---|
| 255 | if (!bail) |
---|
| 256 | converted++; |
---|
| 257 | break; |
---|
| 258 | |
---|
| 259 | case 's': /* String */ |
---|
| 260 | { |
---|
| 261 | char *sp; |
---|
| 262 | sp = sarg = va_arg(ap, char *); |
---|
| 263 | while (width-- && *q && !isspace((unsigned char)*q)) { |
---|
| 264 | *sp++ = *q++; |
---|
| 265 | } |
---|
| 266 | if (sarg != sp) { |
---|
| 267 | *sp = '\0'; /* Terminate output */ |
---|
| 268 | converted++; |
---|
| 269 | } else { |
---|
| 270 | bail = bail_eof; |
---|
| 271 | } |
---|
| 272 | } |
---|
| 273 | break; |
---|
| 274 | |
---|
| 275 | case '[': /* Character range */ |
---|
| 276 | sarg = va_arg(ap, char *); |
---|
| 277 | state = st_match_init; |
---|
| 278 | matchinv = 0; |
---|
| 279 | memset(matchmap, 0, sizeof matchmap); |
---|
| 280 | break; |
---|
| 281 | |
---|
| 282 | case '%': /* %% sequence */ |
---|
| 283 | if (*q == '%') |
---|
| 284 | q++; |
---|
| 285 | else |
---|
| 286 | bail = bail_err; |
---|
| 287 | break; |
---|
| 288 | |
---|
| 289 | default: /* Anything else */ |
---|
| 290 | bail = bail_err; /* Unknown sequence */ |
---|
| 291 | break; |
---|
| 292 | } |
---|
| 293 | } |
---|
| 294 | break; |
---|
| 295 | |
---|
| 296 | case st_match_init: /* Initial state for %[ match */ |
---|
| 297 | if (ch == '^' && !matchinv) { |
---|
| 298 | matchinv = 1; |
---|
| 299 | } else { |
---|
| 300 | range_start = (unsigned char)ch; |
---|
| 301 | set_bit((unsigned char)ch, matchmap); |
---|
| 302 | state = st_match; |
---|
| 303 | } |
---|
| 304 | break; |
---|
| 305 | |
---|
| 306 | case st_match: /* Main state for %[ match */ |
---|
| 307 | if (ch == ']') { |
---|
| 308 | goto match_run; |
---|
| 309 | } else if (ch == '-') { |
---|
| 310 | state = st_match_range; |
---|
| 311 | } else { |
---|
| 312 | range_start = (unsigned char)ch; |
---|
| 313 | set_bit((unsigned char)ch, matchmap); |
---|
| 314 | } |
---|
| 315 | break; |
---|
| 316 | |
---|
| 317 | case st_match_range: /* %[ match after - */ |
---|
| 318 | if (ch == ']') { |
---|
| 319 | set_bit((unsigned char)'-', matchmap); /* - was last character */ |
---|
| 320 | goto match_run; |
---|
| 321 | } else { |
---|
| 322 | int i; |
---|
| 323 | for (i = range_start; i <= (unsigned char)ch; i++) |
---|
| 324 | set_bit(i, matchmap); |
---|
| 325 | state = st_match; |
---|
| 326 | } |
---|
| 327 | break; |
---|
| 328 | |
---|
| 329 | match_run: /* Match expression finished */ |
---|
| 330 | qq = q; |
---|
| 331 | while (width && *q |
---|
| 332 | && test_bit((unsigned char)*q, matchmap) ^ matchinv) { |
---|
| 333 | *sarg++ = *q++; |
---|
| 334 | } |
---|
| 335 | if (q != qq) { |
---|
| 336 | *sarg = '\0'; |
---|
| 337 | converted++; |
---|
| 338 | } else { |
---|
| 339 | bail = *q ? bail_err : bail_eof; |
---|
| 340 | } |
---|
| 341 | break; |
---|
| 342 | } |
---|
| 343 | } |
---|
| 344 | |
---|
| 345 | if (bail == bail_eof && !converted) |
---|
| 346 | converted = -1; /* Return EOF (-1) */ |
---|
| 347 | |
---|
| 348 | return converted; |
---|
| 349 | } |
---|