2018-08-28 10:47:58 +00:00
|
|
|
/* $OpenBSD: fmt_scaled.c,v 1.17 2018/05/14 04:39:04 djm Exp $ */
|
2008-07-23 09:33:08 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Copyright (c) 2001, 2002, 2003 Ian F. Darwin. All rights reserved.
|
|
|
|
*
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
* modification, are permitted provided that the following conditions
|
|
|
|
* are met:
|
|
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer.
|
|
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
|
|
* documentation and/or other materials provided with the distribution.
|
|
|
|
* 3. The name of the author may not be used to endorse or promote products
|
|
|
|
* derived from this software without specific prior written permission.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
|
|
|
|
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
|
|
|
|
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
|
|
|
|
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
|
|
|
|
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
|
|
|
|
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
|
|
|
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
|
|
|
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
|
|
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
|
|
|
|
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* OPENBSD ORIGINAL: lib/libutil/fmt_scaled.c */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* fmt_scaled: Format numbers scaled for human comprehension
|
|
|
|
* scan_scaled: Scan numbers in this format.
|
|
|
|
*
|
|
|
|
* "Human-readable" output uses 4 digits max, and puts a unit suffix at
|
|
|
|
* the end. Makes output compact and easy-to-read esp. on huge disks.
|
|
|
|
* Formatting code was originally in OpenBSD "df", converted to library routine.
|
|
|
|
* Scanning code written for OpenBSD libutil.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "includes.h"
|
|
|
|
|
|
|
|
#ifndef HAVE_FMT_SCALED
|
|
|
|
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <errno.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <ctype.h>
|
|
|
|
#include <limits.h>
|
|
|
|
|
|
|
|
typedef enum {
|
|
|
|
NONE = 0, KILO = 1, MEGA = 2, GIGA = 3, TERA = 4, PETA = 5, EXA = 6
|
|
|
|
} unit_type;
|
|
|
|
|
|
|
|
/* These three arrays MUST be in sync! XXX make a struct */
|
|
|
|
static unit_type units[] = { NONE, KILO, MEGA, GIGA, TERA, PETA, EXA };
|
|
|
|
static char scale_chars[] = "BKMGTPE";
|
|
|
|
static long long scale_factors[] = {
|
|
|
|
1LL,
|
|
|
|
1024LL,
|
|
|
|
1024LL*1024,
|
|
|
|
1024LL*1024*1024,
|
|
|
|
1024LL*1024*1024*1024,
|
|
|
|
1024LL*1024*1024*1024*1024,
|
|
|
|
1024LL*1024*1024*1024*1024*1024,
|
|
|
|
};
|
|
|
|
#define SCALE_LENGTH (sizeof(units)/sizeof(units[0]))
|
|
|
|
|
|
|
|
#define MAX_DIGITS (SCALE_LENGTH * 3) /* XXX strlen(sprintf("%lld", -1)? */
|
|
|
|
|
2017-08-03 10:10:20 +00:00
|
|
|
/* Convert the given input string "scaled" into numeric in "result".
|
2008-07-23 09:33:08 +00:00
|
|
|
* Return 0 on success, -1 and errno set on error.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
scan_scaled(char *scaled, long long *result)
|
|
|
|
{
|
|
|
|
char *p = scaled;
|
|
|
|
int sign = 0;
|
|
|
|
unsigned int i, ndigits = 0, fract_digits = 0;
|
|
|
|
long long scale_fact = 1, whole = 0, fpart = 0;
|
|
|
|
|
|
|
|
/* Skip leading whitespace */
|
2017-08-03 10:10:20 +00:00
|
|
|
while (isascii((unsigned char)*p) && isspace((unsigned char)*p))
|
2008-07-23 09:33:08 +00:00
|
|
|
++p;
|
|
|
|
|
|
|
|
/* Then at most one leading + or - */
|
|
|
|
while (*p == '-' || *p == '+') {
|
|
|
|
if (*p == '-') {
|
|
|
|
if (sign) {
|
|
|
|
errno = EINVAL;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
sign = -1;
|
|
|
|
++p;
|
|
|
|
} else if (*p == '+') {
|
|
|
|
if (sign) {
|
|
|
|
errno = EINVAL;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
sign = +1;
|
|
|
|
++p;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Main loop: Scan digits, find decimal point, if present.
|
|
|
|
* We don't allow exponentials, so no scientific notation
|
|
|
|
* (but note that E for Exa might look like e to some!).
|
|
|
|
* Advance 'p' to end, to get scale factor.
|
|
|
|
*/
|
2017-08-03 10:10:20 +00:00
|
|
|
for (; isascii((unsigned char)*p) &&
|
|
|
|
(isdigit((unsigned char)*p) || *p=='.'); ++p) {
|
2008-07-23 09:33:08 +00:00
|
|
|
if (*p == '.') {
|
|
|
|
if (fract_digits > 0) { /* oops, more than one '.' */
|
|
|
|
errno = EINVAL;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
fract_digits = 1;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
i = (*p) - '0'; /* whew! finally a digit we can use */
|
|
|
|
if (fract_digits > 0) {
|
|
|
|
if (fract_digits >= MAX_DIGITS-1)
|
|
|
|
/* ignore extra fractional digits */
|
|
|
|
continue;
|
|
|
|
fract_digits++; /* for later scaling */
|
2018-05-06 12:24:45 +00:00
|
|
|
if (fpart > LLONG_MAX / 10) {
|
2017-08-03 10:10:20 +00:00
|
|
|
errno = ERANGE;
|
|
|
|
return -1;
|
|
|
|
}
|
2008-07-23 09:33:08 +00:00
|
|
|
fpart *= 10;
|
2018-05-06 12:24:45 +00:00
|
|
|
if (i > LLONG_MAX - fpart) {
|
|
|
|
errno = ERANGE;
|
|
|
|
return -1;
|
|
|
|
}
|
2008-07-23 09:33:08 +00:00
|
|
|
fpart += i;
|
|
|
|
} else { /* normal digit */
|
|
|
|
if (++ndigits >= MAX_DIGITS) {
|
|
|
|
errno = ERANGE;
|
|
|
|
return -1;
|
|
|
|
}
|
2018-05-06 12:24:45 +00:00
|
|
|
if (whole > LLONG_MAX / 10) {
|
2017-08-03 10:10:20 +00:00
|
|
|
errno = ERANGE;
|
|
|
|
return -1;
|
|
|
|
}
|
2008-07-23 09:33:08 +00:00
|
|
|
whole *= 10;
|
2018-05-06 12:24:45 +00:00
|
|
|
if (i > LLONG_MAX - whole) {
|
|
|
|
errno = ERANGE;
|
|
|
|
return -1;
|
|
|
|
}
|
2008-07-23 09:33:08 +00:00
|
|
|
whole += i;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (sign) {
|
|
|
|
whole *= sign;
|
|
|
|
fpart *= sign;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* If no scale factor given, we're done. fraction is discarded. */
|
|
|
|
if (!*p) {
|
|
|
|
*result = whole;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Validate scale factor, and scale whole and fraction by it. */
|
|
|
|
for (i = 0; i < SCALE_LENGTH; i++) {
|
|
|
|
|
2017-08-03 10:10:20 +00:00
|
|
|
/* Are we there yet? */
|
2008-07-23 09:33:08 +00:00
|
|
|
if (*p == scale_chars[i] ||
|
2017-08-03 10:10:20 +00:00
|
|
|
*p == tolower((unsigned char)scale_chars[i])) {
|
2008-07-23 09:33:08 +00:00
|
|
|
|
|
|
|
/* If it ends with alphanumerics after the scale char, bad. */
|
2017-08-03 10:10:20 +00:00
|
|
|
if (isalnum((unsigned char)*(p+1))) {
|
2008-07-23 09:33:08 +00:00
|
|
|
errno = EINVAL;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
scale_fact = scale_factors[i];
|
|
|
|
|
2018-05-06 12:24:45 +00:00
|
|
|
/* check for overflow and underflow after scaling */
|
|
|
|
if (whole > LLONG_MAX / scale_fact ||
|
|
|
|
whole < LLONG_MIN / scale_fact) {
|
2017-08-03 10:10:20 +00:00
|
|
|
errno = ERANGE;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2008-07-23 09:33:08 +00:00
|
|
|
/* scale whole part */
|
|
|
|
whole *= scale_fact;
|
|
|
|
|
2018-08-28 10:47:58 +00:00
|
|
|
/* truncate fpart so it doesn't overflow.
|
2008-07-23 09:33:08 +00:00
|
|
|
* then scale fractional part.
|
|
|
|
*/
|
|
|
|
while (fpart >= LLONG_MAX / scale_fact) {
|
|
|
|
fpart /= 10;
|
|
|
|
fract_digits--;
|
|
|
|
}
|
|
|
|
fpart *= scale_fact;
|
|
|
|
if (fract_digits > 0) {
|
|
|
|
for (i = 0; i < fract_digits -1; i++)
|
|
|
|
fpart /= 10;
|
|
|
|
}
|
|
|
|
whole += fpart;
|
|
|
|
*result = whole;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
}
|
2017-08-03 10:10:20 +00:00
|
|
|
|
|
|
|
/* Invalid unit or character */
|
|
|
|
errno = EINVAL;
|
2008-07-23 09:33:08 +00:00
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Format the given "number" into human-readable form in "result".
|
|
|
|
* Result must point to an allocated buffer of length FMT_SCALED_STRSIZE.
|
|
|
|
* Return 0 on success, -1 and errno set if error.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
fmt_scaled(long long number, char *result)
|
|
|
|
{
|
|
|
|
long long abval, fract = 0;
|
|
|
|
unsigned int i;
|
|
|
|
unit_type unit = NONE;
|
|
|
|
|
2017-08-03 10:10:20 +00:00
|
|
|
abval = llabs(number);
|
2008-07-23 09:33:08 +00:00
|
|
|
|
|
|
|
/* Not every negative long long has a positive representation.
|
|
|
|
* Also check for numbers that are just too darned big to format
|
|
|
|
*/
|
|
|
|
if (abval < 0 || abval / 1024 >= scale_factors[SCALE_LENGTH-1]) {
|
|
|
|
errno = ERANGE;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* scale whole part; get unscaled fraction */
|
|
|
|
for (i = 0; i < SCALE_LENGTH; i++) {
|
|
|
|
if (abval/1024 < scale_factors[i]) {
|
|
|
|
unit = units[i];
|
|
|
|
fract = (i == 0) ? 0 : abval % scale_factors[i];
|
|
|
|
number /= scale_factors[i];
|
|
|
|
if (i > 0)
|
|
|
|
fract /= scale_factors[i - 1];
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
fract = (10 * fract + 512) / 1024;
|
|
|
|
/* if the result would be >= 10, round main number */
|
2018-08-28 10:47:58 +00:00
|
|
|
if (fract >= 10) {
|
2008-07-23 09:33:08 +00:00
|
|
|
if (number >= 0)
|
|
|
|
number++;
|
|
|
|
else
|
|
|
|
number--;
|
|
|
|
fract = 0;
|
2018-08-28 10:47:58 +00:00
|
|
|
} else if (fract < 0) {
|
|
|
|
/* shouldn't happen */
|
|
|
|
fract = 0;
|
2008-07-23 09:33:08 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if (number == 0)
|
|
|
|
strlcpy(result, "0B", FMT_SCALED_STRSIZE);
|
|
|
|
else if (unit == NONE || number >= 100 || number <= -100) {
|
|
|
|
if (fract >= 5) {
|
|
|
|
if (number >= 0)
|
|
|
|
number++;
|
|
|
|
else
|
|
|
|
number--;
|
|
|
|
}
|
|
|
|
(void)snprintf(result, FMT_SCALED_STRSIZE, "%lld%c",
|
|
|
|
number, scale_chars[unit]);
|
|
|
|
} else
|
|
|
|
(void)snprintf(result, FMT_SCALED_STRSIZE, "%lld.%1lld%c",
|
|
|
|
number, fract, scale_chars[unit]);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef MAIN
|
|
|
|
/*
|
|
|
|
* This is the original version of the program in the man page.
|
|
|
|
* Copy-and-paste whatever you need from it.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
main(int argc, char **argv)
|
|
|
|
{
|
|
|
|
char *cinput = "1.5K", buf[FMT_SCALED_STRSIZE];
|
|
|
|
long long ninput = 10483892, result;
|
|
|
|
|
|
|
|
if (scan_scaled(cinput, &result) == 0)
|
|
|
|
printf("\"%s\" -> %lld\n", cinput, result);
|
|
|
|
else
|
|
|
|
perror(cinput);
|
|
|
|
|
|
|
|
if (fmt_scaled(ninput, buf) == 0)
|
|
|
|
printf("%lld -> \"%s\"\n", ninput, buf);
|
|
|
|
else
|
|
|
|
fprintf(stderr, "%lld invalid (%s)\n", ninput, strerror(errno));
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#endif /* HAVE_FMT_SCALED */
|