Date: Tue, 29 Jan 2002 21:31:13 -0800 From: Bill Fenner <fenner@research.att.com> To: standards@freebsd.org Subject: scanf(3) patches for review Message-ID: <200201300531.g0U5VEh48095@stash.attlabs.att.com>
next in thread | raw e-mail | index | archive | help
--19701020
Content-Type: text/plain; charset=US-ASCII
Content-Disposition: inline
Here are some long-standing scanf(3) patches, which add the new c99
size modifiers to scanf(3). They've been languishing in my tree waiting
for me to get around to implementing %n$, but I am clearly not getting
to it so there's nothing to win by sitting on 'em.
One thing that I haven't decided yet is whether it makes sense to
rewrite the size modifier narrative to a table in the same way as
I did for printf(3). Input is solicited.
Thanks,
Bill
--19701020
Content-Type: text/plain; name="scanf.diff"; x-unix-mode=0664
Content-Disposition: attachment; filename="scanf.diff"
Index: scanf.3
===================================================================
RCS file: /home/ncvs/src/lib/libc/stdio/scanf.3,v
retrieving revision 1.13
diff -u -r1.13 scanf.3
--- scanf.3 1 Oct 2001 16:08:59 -0000 1.13
+++ scanf.3 30 Jan 2002 05:04:55 -0000
@@ -137,11 +137,20 @@
there may be a number of
.Em flag
characters, as follows:
-.Bl -tag -width indent
+.Bl -tag -width ".Cm l No (ell)"
.It Cm *
Suppresses assignment.
The conversion that follows occurs as usual, but no pointer is used;
the result of the conversion is simply discarded.
+.It Cm hh
+Indicates that the conversion will be one of
+.Cm dioux
+or
+.Cm n
+and the next pointer is a pointer to a
+.Em char
+(rather than
+.Em int ) .
.It Cm h
Indicates that the conversion will be one of
.Cm dioux
@@ -151,7 +160,7 @@
.Em short int
(rather than
.Em int ) .
-.It Cm l
+.It Cm l No (ell)
Indicates either that the conversion will be one of
.Cm dioux
or
@@ -160,21 +169,66 @@
.Em long int
(rather than
.Em int ) ,
-or that the conversion will be one of
-.Cm efg
+that the conversion will be one of
+.Cm aefg
and the next pointer is a pointer to
.Em double
(rather than
-.Em float ) .
+.Em float ) ,
+orthat the conversion will be one of
+.Cm c
+or
+.Cm s
+and the next pointer is a pointer to an array of
+.Em wchar_t
+(rather than
+.Em char ) .
+.It Cm ll No (ell ell)
+Indicates either that the conversion will be one of
+.Cm dioux
+or
+.Cm n
+and the next pointer is a pointer to a
+.Em long long int
+(rather than
+.Em int ) .
.It Cm L
Indicates that the conversion will be
-.Cm efg
+.Cm aefg
and the next pointer is a pointer to
.Em long double .
(This type is not implemented; the
.Cm L
flag is currently ignored.)
+.It Cm j
+Indicates either that the conversion will be one of
+.Cm dioux
+or
+.Cm n
+and the next pointer is a pointer to a
+.Em intmax_t
+(rather than
+.Em int ) .
+.It Cm t
+Indicates either that the conversion will be one of
+.Cm dioux
+or
+.Cm n
+and the next pointer is a pointer to a
+.Em ptrdiff_t
+(rather than
+.Em int ) .
+.It Cm z
+Indicates either that the conversion will be one of
+.Cm dioux
+or
+.Cm n
+and the next pointer is a pointer to a
+.Em size_t
+(rather than
+.Em int ) .
.It Cm q
+(deprecated.)
Indicates either that the conversion will be one of
.Cm dioux
or
@@ -182,7 +236,7 @@
and the next pointer is a pointer to a
.Em long long int
(rather than
-.Em int ) ,
+.Em int ) .
.El
.Pp
In addition to these flags,
@@ -210,10 +264,6 @@
Matches an optionally signed decimal integer;
the next pointer must be a pointer to
.Em int .
-.It Cm D
-Equivalent to
-.Cm ld ;
-this exists only for backwards compatibility.
.It Cm i
Matches an optionally signed integer;
the next pointer must be a pointer to
@@ -231,48 +281,37 @@
Matches an octal integer;
the next pointer must be a pointer to
.Em unsigned int .
-.It Cm O
-Equivalent to
-.Cm lo ;
-this exists for backwards compatibility.
.It Cm u
Matches an optionally signed decimal integer;
the next pointer must be a pointer to
.Em unsigned int .
-.It Cm x
+.It Cm x , X
Matches an optionally signed hexadecimal integer;
the next pointer must be a pointer to
.Em unsigned int .
-.It Cm X
-Equivalent to
-.Cm lx ;
-this violates the
-.St -isoC ,
-but is backwards compatible with previous
-.Ux
-systems.
-.It Cm f
+.It Cm e , E , f , F , g , G
Matches an optionally signed floating-point number;
the next pointer must be a pointer to
.Em float .
-.It Cm e
-Equivalent to
-.Cm f .
-.It Cm g
-Equivalent to
-.Cm f .
-.It Cm E
-Equivalent to
-.Cm lf ;
-this violates the
-.St -isoC ,
-but is backwards compatible with previous
-.Ux
-systems.
-.It Cm F
-Equivalent to
-.Cm lf ;
-this exists only for backwards compatibility.
+.It Cm a , A
+Matches a hexadecimal number represented in the style
+.Sm off
+.Oo \- Oc Li 0x Ar h Li \&. Ar hhhp Oo \\*[Pm] Oc Ar d .
+.Sm off
+This is an exact coversion of the mantissa+exponent internal
+floating point representation; the
+.Sm off
+.Oo \- Oc Li 0x Ar h Li \&. Ar hhh
+.Sm on
+portion represents exactly the mantissa; only denormalized
+mantissas have a zero value to the left of the hexadecimal
+point.
+The
+.Cm p
+is a literal character
+.Ql p ;
+the exponent is preceded by a positive or negative sign
+and is represented in decimal.
.It Cm s
Matches a sequence of non-white-space characters;
the next pointer must be a pointer to
@@ -283,6 +322,9 @@
character.
The input string stops at white space
or at the maximum field width, whichever occurs first.
+.It Cm S
+The same as
+.Cm ls .
.It Cm c
Matches a sequence of
.Em width
@@ -296,6 +338,9 @@
is added).
The usual skip of leading white space is suppressed.
To skip white space first, use an explicit space in the format.
+.It Cm C
+The same as
+.Cm lc .
.It Cm \&[
Matches a nonempty sequence of characters from the specified set
of accepted characters;
@@ -363,29 +408,10 @@
.Dv LC_NUMERIC ) .
.Pp
For backwards compatibility,
-other conversion characters (except
-.Ql \e0 )
-are taken as if they were
-.Ql %d
-or, if uppercase,
-.Ql %ld ,
-and a `conversion' of
+a `conversion' of
.Ql %\e0
causes an immediate return of
.Dv EOF .
-The
-.Cm F
-and
-.Cm X
-conversions will be changed in the future
-to conform to the
-.Tn ANSI
-C standard,
-after which they will act like
-.Cm f
-and
-.Cm x
-respectively.
.Sh RETURN VALUES
These
functions
@@ -419,7 +445,7 @@
and
.Fn sscanf
conform to
-.St -isoC .
+.St -isoC-99 .
.Sh HISTORY
The functions
.Fn vscanf ,
@@ -428,13 +454,21 @@
.Fn vfscanf
are new to this release.
.Sh BUGS
-The current situation with
-.Cm %F
+Earlier implementations of
+.Nm
+treated
+.Cm \&%D , \&%E , \&%F , \&%O
and
-.Cm %X
-conversions is unfortunate.
-.Pp
-All of the backwards compatibility formats will be removed in the future.
+.Cm \&%X
+as their lowercase equivalents with an
+.Cm l
+modifier. In addition,
+.Nm
+treated an unknown conversion character as
+.Cm \&%d
+or
+.Cm \&%D ,
+depending on its case. This functionality has been removed.
.Pp
Numerical strings are truncated to 512 characters; for example,
.Cm %f
@@ -444,3 +478,25 @@
.Cm %512f
and
.Cm %512d .
+.Pp
+The
+.Cm %n$
+modifiers for positional arguments are not implemented.
+.Pp
+The
+.Cm l
+modifier for
+.Cm %c
+and
+.Cm %s
+(and
+.Cm %C
+and
+.Cm %S )
+to specify wide characters and strings is not implemented.
+.Pp
+The
+.Cm \&%a
+and
+.Cm \&%A
+floating-point formats are not implemented.
Index: vfscanf.c
===================================================================
RCS file: /home/ncvs/src/lib/libc/stdio/vfscanf.c,v
retrieving revision 1.19
diff -u -r1.19 vfscanf.c
--- vfscanf.c 29 Nov 2001 03:03:55 -0000 1.19
+++ vfscanf.c 30 Jan 2002 05:16:58 -0000
@@ -45,6 +45,7 @@
#include "namespace.h"
#include <stdio.h>
#include <stdlib.h>
+#include <stddef.h>
#include <ctype.h>
#if __STDC__
#include <stdarg.h>
@@ -52,6 +53,7 @@
#include <varargs.h>
#endif
#include <string.h>
+#include <inttypes.h>
#include "un-namespace.h"
#include "collate.h"
@@ -76,7 +78,11 @@
#define SUPPRESS 0x08 /* suppress assignment */
#define POINTER 0x10 /* weird %p pointer (`fake hex') */
#define NOSKIP 0x20 /* do not skip blanks */
-#define QUAD 0x400
+#define LONGLONG 0x400 /* ll: long long (+ deprecated q: quad) */
+#define INTMAXT 0x800 /* j: intmax_t */
+#define PTRDIFFT 0x1000 /* t: ptrdiff_t */
+#define SIZET 0x2000 /* z: size_t */
+#define SHORTSHORT 0x4000 /* hh: char */
/*
* The following are used in numeric conversions only:
@@ -98,13 +104,10 @@
#define CT_CHAR 0 /* %c conversion */
#define CT_CCL 1 /* %[...] conversion */
#define CT_STRING 2 /* %s conversion */
-#define CT_INT 3 /* integer, i.e., strtoq or strtouq */
+#define CT_INT 3 /* integer, i.e., strtoimax or strtoumax */
#define CT_FLOAT 4 /* floating, i.e., strtod */
-#define u_char unsigned char
-#define u_long unsigned long
-
-static u_char *__sccl(char *, u_char *);
+static const u_char *__sccl(char *, const u_char *);
/*
* __vfscanf - MT-safe version
@@ -124,9 +127,9 @@
* __svfscanf - non-MT-safe version of __vfscanf
*/
int
-__svfscanf(FILE *fp, char const *fmt0, va_list ap)
+__svfscanf(FILE *fp, const char *fmt0, va_list ap)
{
- u_char *fmt = (u_char *)fmt0;
+ const u_char *fmt = (const u_char *)fmt0;
int c; /* character from format, or conversion */
size_t width; /* field width, or 0 */
char *p; /* points into all kinds of strings */
@@ -136,8 +139,8 @@
int nassigned; /* number of fields assigned */
int nconversions; /* number of conversions */
int nread; /* number of characters consumed from fp */
- int base; /* base argument to strtoq/strtouq */
- u_quad_t(*ccfn)(); /* conversion function (strtoq/strtouq) */
+ int base; /* base argument to strtoimax/strtoumax */
+ uintmax_t(*ccfn)(); /* conversion function (strtoimax/strtoumax) */
char ccltab[256]; /* character class table for %[...] */
char buf[BUF]; /* buffer for numeric conversions */
@@ -185,17 +188,34 @@
case '*':
flags |= SUPPRESS;
goto again;
+ case 'j':
+ flags |= INTMAXT;
+ goto again;
case 'l':
- flags |= LONG;
+ if (flags & LONG) {
+ flags &= ~LONG;
+ flags |= LONGLONG;
+ } else
+ flags |= LONG;
goto again;
case 'q':
- flags |= QUAD;
+ flags |= LONGLONG; /* not quite */
+ goto again;
+ case 't':
+ flags |= PTRDIFFT;
+ goto again;
+ case 'z':
+ flags |= SIZET;
goto again;
case 'L':
flags |= LONGDBL;
goto again;
case 'h':
- flags |= SHORT;
+ if (flags & SHORT) {
+ flags &= ~SHORT;
+ flags |= SHORTSHORT;
+ } else
+ flags |= SHORT;
goto again;
case '0': case '1': case '2': case '3': case '4':
@@ -205,61 +225,49 @@
/*
* 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 */
- flags |= LONG;
- /* FALLTHROUGH */
case 'd':
c = CT_INT;
- ccfn = (u_quad_t (*)())strtoq;
+ ccfn = (uintmax_t (*)())strtoimax;
base = 10;
break;
case 'i':
c = CT_INT;
- ccfn = (u_quad_t (*)())strtoq;
+ ccfn = (uintmax_t (*)())strtoimax;
base = 0;
break;
- case 'O': /* compat */
- flags |= LONG;
- /* FALLTHROUGH */
case 'o':
c = CT_INT;
- ccfn = strtouq;
+ ccfn = strtoumax;
base = 8;
break;
case 'u':
c = CT_INT;
- ccfn = strtouq;
+ ccfn = strtoumax;
base = 10;
break;
- case 'X': /* compat XXX */
- flags |= LONG;
- /* FALLTHROUGH */
+ case 'X':
case 'x':
flags |= PFXOK; /* enable 0x prefixing */
c = CT_INT;
- ccfn = strtouq;
+ ccfn = strtoumax;
base = 16;
break;
#ifdef FLOATING_POINT
- case 'E': /* compat XXX */
- case 'F': /* compat */
- flags |= LONG;
- /* FALLTHROUGH */
+ case 'E': case 'F': case 'G':
case 'e': case 'f': case 'g':
c = CT_FLOAT;
break;
#endif
+ case 'S':
+ flags |= LONG;
+ /* FALLTHROUGH */
case 's':
c = CT_STRING;
break;
@@ -270,6 +278,9 @@
c = CT_CCL;
break;
+ case 'C':
+ flags |= LONG;
+ /* FALLTHROUGH */
case 'c':
flags |= NOSKIP;
c = CT_CHAR;
@@ -278,7 +289,7 @@
case 'p': /* pointer format is like hex */
flags |= POINTER | PFXOK;
c = CT_INT;
- ccfn = strtouq;
+ ccfn = strtoumax;
base = 16;
break;
@@ -286,29 +297,32 @@
nconversions++;
if (flags & SUPPRESS) /* ??? */
continue;
- if (flags & SHORT)
+ if (flags & SHORTSHORT)
+ *va_arg(ap, char *) = nread;
+ else if (flags & SHORT)
*va_arg(ap, short *) = nread;
else if (flags & LONG)
*va_arg(ap, long *) = nread;
- else if (flags & QUAD)
- *va_arg(ap, quad_t *) = nread;
+ else if (flags & LONGLONG)
+ *va_arg(ap, long long *) = nread;
+ else if (flags & INTMAXT)
+ *va_arg(ap, intmax_t *) = nread;
+ else if (flags & SIZET)
+ *va_arg(ap, size_t *) = nread;
+ else if (flags & PTRDIFFT)
+ *va_arg(ap, ptrdiff_t *) = nread;
else
*va_arg(ap, int *) = nread;
continue;
+ default:
+ goto match_failure;
+
/*
- * Disgusting backwards compatibility hacks. XXX
+ * Disgusting backwards compatibility hack. XXX
*/
case '\0': /* compat */
return (EOF);
-
- default: /* compat */
- if (isupper(c))
- flags |= LONG;
- c = CT_INT;
- ccfn = (u_quad_t (*)())strtoq;
- base = 10;
- break;
}
/*
@@ -451,7 +465,7 @@
continue;
case CT_INT:
- /* scan an integer as if by strtoq/strtouq */
+ /* scan an integer as if by strtoimax/strtoumax */
#ifdef hardway
if (width == 0 || width > sizeof(buf) - 1)
width = sizeof(buf) - 1;
@@ -569,19 +583,27 @@
(void) __ungetc(c, fp);
}
if ((flags & SUPPRESS) == 0) {
- u_quad_t res;
+ uintmax_t res;
*p = 0;
res = (*ccfn)(buf, (char **)NULL, base);
if (flags & POINTER)
*va_arg(ap, void **) =
- (void *)(u_long)res;
+ (void *)(uintptr_t)res;
+ else if (flags & SHORTSHORT)
+ *va_arg(ap, char *) = res;
else if (flags & SHORT)
*va_arg(ap, short *) = res;
else if (flags & LONG)
*va_arg(ap, long *) = res;
- else if (flags & QUAD)
- *va_arg(ap, quad_t *) = res;
+ else if (flags & LONGLONG)
+ *va_arg(ap, long long *) = res;
+ else if (flags & INTMAXT)
+ *va_arg(ap, intmax_t *) = res;
+ else if (flags & PTRDIFFT)
+ *va_arg(ap, ptrdiff_t *) = res;
+ else if (flags & SIZET)
+ *va_arg(ap, size_t *) = res;
else
*va_arg(ap, int *) = res;
nassigned++;
@@ -700,10 +722,10 @@
* closing `]'. The table has a 1 wherever characters should be
* considered part of the scanset.
*/
-static u_char *
+static const u_char *
__sccl(tab, fmt)
register char *tab;
- register u_char *fmt;
+ register const u_char *fmt;
{
register int c, n, v, i;
--19701020--
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-standards" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200201300531.g0U5VEh48095>