com32/lib/vsscanf.c

   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
  16 #ifndef LONG_BIT
  17 #define LONG_BIT (CHAR_BIT*sizeof(long))
  18 #endif
  19
  20 enum flags {
  21   FL_SPLAT  = 0x01,             /* Drop the value, do not assign */
  22   FL_INV    = 0x02,             /* Character-set with inverse */
  23   FL_WIDTH  = 0x04,             /* Field width specified */
  24   FL_MINUS  = 0x08,             /* 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 static inline const char *
  50 skipspace(const char *p)
  51 {
  52   while ( isspace((unsigned char)*p) ) p++;
  53   return p;
  54 }
  55
  56 #undef set_bit
  57 static inline void
  58 set_bit(unsigned long *bitmap, unsigned int bit)
  59 {
  60   bitmap[bit/LONG_BIT] |= 1UL << (bit%LONG_BIT);
  61 }
  62
  63 #undef test_bit
  64 static inline int
  65 test_bit(unsigned long *bitmap, unsigned int bit)
  66 {
  67   return (int)(bitmap[bit/LONG_BIT] >> (bit%LONG_BIT)) & 1;
  68 }
  69
  70 int vsscanf(const char *buffer, const char *format, va_list ap)
  71 {
  72   const char *p = format;
  73   char ch;
  74   const char *q = buffer;
  75   const char *qq;
  76   uintmax_t val = 0;
  77   int rank = rank_int;          /* Default rank */
  78   unsigned int width = UINT_MAX;
  79   int base;
  80   enum flags flags = 0;
  81   enum {
  82     st_normal,                  /* Ground state */
  83     st_flags,                   /* Special flags */
  84     st_width,                   /* Field width */
  85     st_modifiers,               /* Length or conversion modifiers */
  86     st_match_init,              /* Initial state of %[ sequence */
  87     st_match,                   /* Main state of %[ sequence */
  88     st_match_range,             /* After - in a %[ sequence */
  89   } state = st_normal;
  90   char *sarg = NULL;            /* %s %c or %[ string argument */
  91   enum bail bail = bail_none;
  92   int sign;
  93   int converted = 0;            /* Successful conversions */
  94   unsigned long matchmap[((1 << CHAR_BIT)+(LONG_BIT-1))/LONG_BIT];
  95   int matchinv = 0;             /* Is match map inverted? */
  96   unsigned char range_start = 0;
  97
  98   while ( (ch = *p++) && !bail ) {
  99     switch ( state ) {
 100     case st_normal:
 101       if ( ch == '%' ) {
 102         state = st_flags;
 103         flags = 0; rank = rank_int; width = UINT_MAX;
 104       } else if ( isspace((unsigned char)ch) ) {
 105         q = skipspace(q);
 106       } else {
 107         if ( *q == ch )
 108           q++;
 109         else
 110           bail = bail_err;      /* Match failure */
 111       }
 112       break;
 113
 114     case st_flags:
 115       switch ( ch ) {
 116       case '*':
 117         flags |= FL_SPLAT;
 118         break;
 119       case '0' ... '9':
 120         width = (ch-'0');
 121         state = st_width;
 122         flags |= FL_WIDTH;
 123         break;
 124       default:
 125         state = st_modifiers;
 126         p--;                    /* Process this character again */
 127         break;
 128       }
 129       break;
 130
 131     case st_width:
 132       if ( ch >= '0' && ch <= '9' ) {
 133         width = width*10+(ch-'0');
 134       } else {
 135         state = st_modifiers;
 136         p--;                    /* Process this character again */
 137       }
 138       break;
 139
 140     case st_modifiers:
 141       switch ( ch ) {
 142         /* Length modifiers - nonterminal sequences */
 143       case 'h':
 144         rank--;                 /* Shorter rank */
 145         break;
 146       case 'l':
 147         rank++;                 /* Longer rank */
 148         break;
 149       case 'j':
 150         rank = INTMAX_RANK;
 151         break;
 152       case 'z':
 153         rank = SIZE_T_RANK;
 154         break;
 155       case 't':
 156         rank = PTRDIFF_T_RANK;
 157         break;
 158       case 'L':
 159       case 'q':
 160         rank = rank_longlong;   /* long double/long long */
 161         break;
 162
 163       default:
 164         /* Output modifiers - terminal sequences */
 165         state = st_normal;      /* Next state will be normal */
 166         if ( rank < MIN_RANK )  /* Canonicalize rank */
 167           rank = MIN_RANK;
 168         else if ( rank > MAX_RANK )
 169           rank = MAX_RANK;
 170
 171         switch ( ch ) {
 172         case 'P':               /* Upper case pointer */
 173         case 'p':               /* Pointer */
 174 #if 0   /* Enable this to allow null pointers by name */
 175           q = skipspace(q);
 176           if ( !isdigit((unsigned char)*q) ) {
 177             static const char * const nullnames[] =
 178             { "null", "nul", "nil", "(null)", "(nul)", "(nil)", 0 };
 179             const char * const *np;
 180
 181             /* Check to see if it's a null pointer by name */
 182             for ( np = nullnames ; *np ; np++ ) {
 183               if ( !strncasecmp(q, *np, strlen(*np)) ) {
 184                 val = (uintmax_t)((void *)NULL);
 185                 goto set_integer;
 186               }
 187             }
 188             /* Failure */
 189             bail = bail_err;
 190             break;
 191           }
 192           /* else */
 193 #endif
 194           rank = rank_ptr;
 195           base = 0; sign = 0;
 196           goto scan_int;
 197
 198         case 'i':               /* Base-independent integer */
 199           base = 0; sign = 1;
 200           goto scan_int;
 201
 202         case 'd':               /* Decimal integer */
 203           base = 10; sign = 1;
 204           goto scan_int;
 205
 206         case 'o':               /* Octal integer */
 207           base = 8; sign = 0;
 208           goto scan_int;
 209
 210         case 'u':               /* Unsigned decimal integer */
 211           base = 10; sign = 0;
 212           goto scan_int;
 213
 214         case 'x':               /* Hexadecimal integer */
 215         case 'X':
 216           base = 16; sign = 0;
 217           goto scan_int;
 218
 219         case 'n':               /* Number of characters consumed */
 220           val = (q-buffer);
 221           goto set_integer;
 222
 223         scan_int:
 224           q = skipspace(q);
 225           if ( !*q ) {
 226             bail = bail_eof;
 227             break;
 228           }
 229           val = strntoumax(q, (char **)&qq, base, width);
 230           if ( qq == q ) {
 231             bail = bail_err;
 232             break;
 233           }
 234           q = qq;
 235           converted++;
 236           /* fall through */
 237
 238         set_integer:
 239           if ( !(flags & FL_SPLAT) ) {
 240             switch(rank) {
 241             case rank_char:
 242               *va_arg(ap, unsigned char *) = (unsigned char)val;
 243               break;
 244             case rank_short:
 245               *va_arg(ap, unsigned short *) = (unsigned short)val;
 246               break;
 247             case rank_int:
 248               *va_arg(ap, unsigned int *) = (unsigned int)val;
 249               break;
 250             case rank_long:
 251               *va_arg(ap, unsigned long *) = (unsigned long)val;
 252               break;
 253             case rank_longlong:
 254               *va_arg(ap, unsigned long long *) = (unsigned long long)val;
 255               break;
 256             case rank_ptr:
 257               *va_arg(ap, void **) = (void *)(uintptr_t)val;
 258               break;
 259             }
 260           }
 261           break;
 262
 263         case 'c':               /* Character */
 264           width = (flags & FL_WIDTH) ? width : 1; /* Default width == 1 */
 265           sarg = va_arg(ap, char *);
 266           while ( width-- ) {
 267             if ( !*q ) {
 268               bail = bail_eof;
 269               break;
 270             }
 271             *sarg++ = *q++;
 272           }
 273           if ( !bail )
 274             converted++;
 275           break;
 276
 277         case 's':               /* String */
 278           {
 279             char *sp;
 280             sp = sarg = va_arg(ap, char *);
 281             while ( width-- && *q && !isspace((unsigned char)*q) ) {
 282               *sp++ = *q++;
 283             }
 284             if ( sarg != sp ) {
 285               *sp = '\0';       /* Terminate output */
 286               converted++;
 287             } else {
 288               bail = bail_eof;
 289             }
 290           }
 291           break;
 292
 293         case '[':               /* Character range */
 294           sarg = va_arg(ap, char *);
 295           state = st_match_init;
 296           matchinv = 0;
 297           memset(matchmap, 0, sizeof matchmap);
 298           break;
 299
 300         case '%':               /* %% sequence */
 301           if ( *q == '%' )
 302             q++;
 303           else
 304             bail = bail_err;
 305           break;
 306
 307         default:                /* Anything else */
 308           bail = bail_err;      /* Unknown sequence */
 309           break;
 310         }
 311       }
 312       break;
 313
 314     case st_match_init:         /* Initial state for %[ match */
 315       if ( ch == '^' && !(flags & FL_INV) ) {
 316         matchinv = 1;
 317       } else {
 318         set_bit(matchmap, (unsigned char)ch);
 319         state = st_match;
 320       }
 321       break;
 322
 323     case st_match:              /* Main state for %[ match */
 324       if ( ch == ']' ) {
 325         goto match_run;
 326       } else if ( ch == '-' ) {
 327         range_start = (unsigned char)ch;
 328         state = st_match_range;
 329       } else {
 330         set_bit(matchmap, (unsigned char)ch);
 331       }
 332       break;
 333
 334     case st_match_range:                /* %[ match after - */
 335       if ( ch == ']' ) {
 336         set_bit(matchmap, (unsigned char)'-'); /* - was last character */
 337         goto match_run;
 338       } else {
 339         int i;
 340         for ( i = range_start ; i < (unsigned char)ch ; i++ )
 341           set_bit(matchmap, i);
 342         state = st_match;
 343       }
 344       break;
 345
 346     match_run:                  /* Match expression finished */
 347       qq = q;
 348       while ( width && *q && test_bit(matchmap, (unsigned char)*q)^matchinv ) {
 349         *sarg++ = *q++;
 350       }
 351       if ( q != qq ) {
 352         *sarg = '\0';
 353         converted++;
 354       } else {
 355         bail = *q ? bail_err : bail_eof;
 356       }
 357       break;
 358     }
 359   }
 360
 361   if ( bail == bail_eof && !converted )
 362     converted = -1;             /* Return EOF (-1) */
 363
 364   return converted;
 365 }