/*
* Copyright (c) 2000-2001, 2004 Proofpoint, Inc. and its suppliers.
* All rights reserved.
* Copyright (c) 1990, 1993
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* Chris Torek.
*
* By using this file, you agree to the terms and conditions set
* forth in the LICENSE file which can be found at the top level of
* the sendmail distribution.
*/
#include <ctype.h>
#include <stdlib.h>
#include <errno.h>
#include <setjmp.h>
#include "local.h"
/* Flags used during conversion. */
/*
** The following are used in numeric conversions only:
** SIGNOK, NDIGITS, DPTOK, and EXPOK are for floating point;
** SIGNOK, NDIGITS, PFXOK, and NZDIGITS are for integral.
*/
/* Conversion types. */
/*
** SCANALRM -- handler when timeout activated for sm_io_vfscanf()
**
** Returns flow of control to where setjmp(ScanTimeOut) was set.
**
** Parameters:
** sig -- unused
**
** Returns:
** does not return
**
** Side Effects:
** returns flow of control to setjmp(ScanTimeOut).
**
** NOTE: THIS CAN BE CALLED FROM A SIGNAL HANDLER. DO NOT ADD
** ANYTHING TO THIS ROUTINE UNLESS YOU KNOW WHAT YOU ARE
** DOING.
*/
/* ARGSUSED0 */
static void
int sig;
{
}
/*
** SM_VFSCANF -- convert input into data units
**
** Parameters:
** fp -- file pointer for input data
** timeout -- time intvl allowed to complete (milliseconds)
** fmt0 -- format for finding data units
** ap -- vectors for memory location for storing data units
**
** Results:
** Success: number of data units assigned
** Failure: SM_IO_EOF
*/
int
int SM_NONVOLATILE timeout;
char const *fmt0;
{
register int c; /* character from format, or conversion */
register char *p; /* points into all kinds of strings */
register int n; /* handy integer */
/* `basefix' is used to avoid `if' tests in the integer scanner */
{ 10, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 };
if (timeout == SM_TIME_DEFAULT)
if (timeout == SM_TIME_IMMEDIATE)
{
/*
** Filling the buffer will take time and we are wanted to
** return immediately. So...
*/
return SM_IO_EOF;
}
if (timeout != SM_TIME_FOREVER)
{
if (setjmp(ScanTimeOut) != 0)
{
return SM_IO_EOF;
}
}
nassigned = 0;
nread = 0;
base = 0; /* XXX just to keep gcc happy */
for (;;)
{
c = *fmt++;
if (c == 0)
{
return nassigned;
}
if (isspace(c))
{
== 0) &&
continue;
}
if (c != '%')
goto literal;
width = 0;
flags = 0;
/*
** switch on the format. continue if done;
** break once format type is derived.
*/
switch (c)
{
case '%':
goto input_failure;
goto match_failure;
nread++;
continue;
case '*':
goto again;
case 'h':
goto again;
case 'l':
if (*fmt == 'l')
{
fmt++;
}
else
{
}
goto again;
case 'q':
goto again;
case '0': case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
goto again;
/*
** Conversions.
** Those marked `compat' are for 4.[123]BSD compatibility.
**
** (According to ANSI, E and X formats are supposed
** to the same as e and x. Sorry about that.)
*/
case 'D': /* compat */
/* FALLTHROUGH */
case 'd':
c = CT_INT;
base = 10;
break;
case 'i':
c = CT_INT;
base = 0;
break;
case 'O': /* compat */
/* FALLTHROUGH */
case 'o':
c = CT_INT;
ccfn = sm_strtoull;
base = 8;
break;
case 'u':
c = CT_INT;
ccfn = sm_strtoull;
base = 10;
break;
case 'X':
case 'x':
c = CT_INT;
ccfn = sm_strtoull;
base = 16;
break;
case 'E':
case 'G':
case 'e':
case 'f':
case 'g':
c = CT_FLOAT;
break;
case 's':
c = CT_STRING;
break;
case '[':
c = CT_CCL;
break;
case 'c':
c = CT_CHAR;
break;
case 'p': /* pointer format is like hex */
c = CT_INT;
ccfn = sm_strtoull;
base = 16;
break;
case 'n':
continue;
else
continue;
/* Disgusting backwards compatibility hacks. XXX */
case '\0': /* compat */
return SM_IO_EOF;
default: /* compat */
if (isupper(c))
c = CT_INT;
base = 10;
break;
}
/* We have a conversion that requires input. */
goto input_failure;
/*
** Consume leading white space, except for formats
** that suppress this.
*/
{
{
nread++;
goto input_failure;
}
/*
** Note that there is at least one character in
** the buffer, so conversions that do not set NOSKIP
** can no longer result in an input failure.
*/
}
/* Do the conversion. */
switch (c)
{
case CT_CHAR:
/* scan arbitrary characters (sets NOSKIP) */
if (width == 0)
width = 1;
{
for (;;)
{
{
sum += n;
width -= n;
{
if (sum == 0)
goto input_failure;
break;
}
}
else
{
break;
}
}
}
else
{
size_t r;
width);
if (r == 0)
goto input_failure;
nread += r;
nassigned++;
}
break;
case CT_CCL:
/* scan a (nonempty) character class (sets NOSKIP) */
if (width == 0)
/* take only those things in the class */
{
n = 0;
{
if (--width == 0)
break;
{
if (n == 0) /* XXX how? */
goto input_failure;
break;
}
}
if (n == 0)
goto match_failure;
}
else
{
{
if (--width == 0)
break;
{
if (p == p0)
goto input_failure;
break;
}
}
n = p - p0;
if (n == 0)
goto match_failure;
*p = 0;
nassigned++;
}
nread += n;
break;
case CT_STRING:
/* like CCL, but zero-length string OK, & no NOSKIP */
if (width == 0)
{
n = 0;
{
if (--width == 0)
break;
break;
}
nread += n;
}
else
{
{
if (--width == 0)
break;
break;
}
*p = 0;
nassigned++;
}
continue;
case CT_INT:
#else /* SM_CONF_BROKEN_SIZE_T */
/* size_t is unsigned, hence this optimisation */
width++;
#endif /* SM_CONF_BROKEN_SIZE_T */
{
/*
** Switch on the character; `goto ok'
** if we accept it as a part of number.
*/
switch (c)
{
/*
** The digit 0 is always legal, but is
** special. For %i conversions, if no
** digits (zero or nonzero) have been
** scanned (only signs), we will have
** base==0. In that case, we should set
** it to 8 and enable 0x prefixing.
** Also, if we have not scanned zero digits
** before this, do not turn off prefixing
** (someone else will turn it off if we
** have scanned any nonzero digits).
*/
case '0':
if (base == 0)
{
base = 8;
}
else
goto ok;
/* 1 through 7 always legal */
case '1': case '2': case '3':
case '4': case '5': case '6': case '7':
goto ok;
/* digits 8 and 9 ok iff decimal or hex */
case '8': case '9':
if (base <= 8)
break; /* not legal here */
goto ok;
/* letters ok iff hex */
case 'A': case 'B': case 'C':
case 'D': case 'E': case 'F':
case 'a': case 'b': case 'c':
case 'd': case 'e': case 'f':
/* no need to fix base here */
if (base <= 10)
break; /* not legal here */
goto ok;
/* sign ok only as first character */
case '+': case '-':
{
goto ok;
}
break;
/* x ok iff flag still set & 2nd char */
case 'x': case 'X':
{
goto ok;
}
break;
}
/*
** If we got here, c is not a legal character
** for a number. Stop accumulating digits.
*/
break;
ok:
/* c is legal: store it and look at the next. */
*p++ = c;
break; /* SM_IO_EOF */
}
/*
** If we had only a sign, it is no good; push
** back the sign. If the number ends in `x',
** it was [sign] '0' 'x', so push back the x
** and treat it as [sign] '0'.
*/
{
if (p > buf)
*(unsigned char *)--p);
goto match_failure;
}
c = ((unsigned char *)p)[-1];
if (c == 'x' || c == 'X')
{
--p;
}
{
*p = 0;
(void *)(long) res;
else
nassigned++;
}
break;
case CT_FLOAT:
/* scan a floating point number as if by strtod */
{
/*
** This code mimicks the integer conversion
** code, but is much simpler.
*/
switch (c)
{
case '0': case '1': case '2': case '3':
case '4': case '5': case '6': case '7':
case '8': case '9':
goto fok;
case '+': case '-':
{
goto fok;
}
break;
case '.':
{
goto fok;
}
break;
case 'e': case 'E':
/* no exponent without some digits */
{
flags =
goto fok;
}
break;
}
break;
fok:
*p++ = c;
break; /* SM_IO_EOF */
}
/*
** If no digits, might be missing exponent digits
** (just give back the exponent) or might be missing
*/
{
{
/* no digits at all */
while (p > buf)
(void) sm_io_ungetc(fp,
*(unsigned char *)--p);
goto match_failure;
}
/* just a bad exponent (e and maybe sign) */
c = *(unsigned char *) --p;
if (c != 'e' && c != 'E')
{
c); /* sign */
c = *(unsigned char *)--p;
}
}
{
double res;
*p = 0;
else
nassigned++;
}
break;
}
}
return nassigned;
}
/*
** SM_SCCL -- sequenced character comparison list
**
** Fill in the given table from the scanset at the given format
** (just after `['). Return a pointer to the character past the
** closing `]'. The table has a 1 wherever characters should be
** considered part of the scanset.
**
** Parameters:
** tab -- array flagging "active" char's to match (returned)
** fmt -- character list (within "[]")
**
** Results:
*/
static unsigned char *
register char *tab;
register unsigned char *fmt;
{
register int c, n, v;
/* first `clear' the whole table */
c = *fmt++; /* first char hat => negated scanset */
if (c == '^')
{
v = 1; /* default => accept */
c = *fmt++; /* get new first char */
}
else
v = 0; /* default => reject */
/* should probably use memset here */
for (n = 0; n < 256; n++)
tab[n] = v;
if (c == 0)
/*
** Now set the entries corresponding to the actual scanset
** to the opposite of the above.
**
** The first character may be ']' (or '-') without being special;
** the last character may be '-'.
*/
v = 1 - v;
for (;;)
{
tab[c] = v; /* take character c */
n = *fmt++; /* and examine the next */
switch (n)
{
case 0: /* format ended too soon */
return fmt - 1;
case '-':
/*
** A scanset of the form
** [01+-]
** is defined as `the digit 0, the digit 1,
** the character +, the character -', but
** the effect of a scanset such as
** [a-zA-Z0-9]
** is implementation defined. The V7 Unix
** scanf treats `a-z' as `the letters a through
** z', but treats `a-a' as `the letter a, the
** character -, and the letter a'.
**
** For compatibility, the `-' is not considerd
** to define a range if the character following
** it is either a close bracket (required by ANSI)
** or is not numerically greater than the character
** we just stored in the table (c).
*/
n = *fmt;
if (n == ']' || n < c)
{
c = '-';
break; /* resume the for(;;) */
}
fmt++;
do
{
/* fill in the range */
tab[++c] = v;
} while (c < n);
#if 1 /* XXX another disgusting compatibility hack */
/*
** Alas, the V7 Unix scanf also treats formats
** such as [a-c-e] as `the letters a through e'.
** This too is permitted by the standard....
*/
goto doswitch;
#else
c = *fmt++;
if (c == 0)
return fmt - 1;
if (c == ']')
return fmt;
break;
#endif
case ']': /* end of scanset */
return fmt;
default: /* just another character */
c = n;
break;
}
}
/* NOTREACHED */
}