cosmopolitan/libc/stdio/dtoa.c
Justine Tunney 1c387727fe Add long double printf formatting support
You can now easily print numbers that have 80 bits of precision. This
change also imports many of the test vectors from the gdtoa codebase.
2022-06-14 02:18:03 -07:00

534 lines
15 KiB
C

/*-*- mode:c;indent-tabs-mode:nil;c-basic-offset:2;tab-width:8;coding:utf-8 -*-│
│vi: set net ft=c ts=2 sts=2 sw=2 fenc=utf-8 :vi│
╞══════════════════════════════════════════════════════════════════════════════╡
│ Copyright (C) 2022 Justine Alexandra Roberts Tunney │
│ Copyright (C) 1997, 1999, 2001 Lucent Technologies │
│ All Rights Reserved │
│ │
│ Permission to use, copy, modify, and distribute this software and │
│ its documentation for any purpose and without fee is hereby │
│ granted, provided that the above copyright notice appear in all │
│ copies and that both that the copyright notice and this │
│ permission notice and warranty disclaimer appear in supporting │
│ documentation, and that the name of Lucent or any of its entities │
│ not be used in advertising or publicity pertaining to │
│ distribution of the software without specific, written prior │
│ permission. │
│ │
│ LUCENT DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, │
│ INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS. │
│ IN NO EVENT SHALL LUCENT OR ANY OF ITS ENTITIES BE LIABLE FOR ANY │
│ SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES │
│ WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER │
│ IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, │
│ ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF │
│ THIS SOFTWARE. │
╚─────────────────────────────────────────────────────────────────────────────*/
#include "libc/assert.h"
#include "libc/fmt/fmt.h"
#include "libc/fmt/fmt.internal.h"
#include "libc/fmt/internal.h"
#include "libc/macros.internal.h"
#include "libc/nexgen32e/bsr.h"
#include "third_party/gdtoa/gdtoa.h"
/**
* @fileoverview Floating-Point Formatting
*
* This implements most of ANSI C's printf floating-point directives.
* omitting L, with %.0g and %.0G giving the shortest decimal string
* that rounds to the number being converted, and with negative
* precisions allowed for %f.
*/
struct FPBits {
uint32_t bits[4];
const FPI *fpi;
int sign;
int ex; // exponent
int kind;
};
union U {
double d;
uint64_t q;
long double ld;
unsigned int ui[4];
unsigned short us[5];
};
static const FPI kFpiDbl = {
.nbits = 53,
.emin = 1 - 1023 - 53 + 1,
.emax = 2046 - 1023 - 53 + 1,
.rounding = FPI_Round_near,
.sudden_underflow = 0,
};
static const FPI kFpiLdbl = {
.nbits = 64,
.emin = 1 - 16383 - 64 + 1,
.emax = 32766 - 16383 - 64 + 1,
.rounding = FPI_Round_near,
.sudden_underflow = 0,
};
static const char kSpecialFloats[2][2][4] = {
{"INF", "inf"},
{"NAN", "nan"},
};
static void dfpbits(union U *u, struct FPBits *b) {
int ex, i;
uint32_t *bits;
b->fpi = &kFpiDbl;
b->sign = u->ui[1] & 0x80000000L;
bits = b->bits;
bits[1] = u->ui[1] & 0xfffff;
bits[0] = u->ui[0];
if ((ex = (u->ui[1] & 0x7ff00000L) >> 20) != 0) {
if (ex == 0x7ff) {
// Infinity or NaN
i = bits[0] | bits[1] ? STRTOG_NaN : STRTOG_Infinite;
} else {
i = STRTOG_Normal;
bits[1] |= 0x100000;
}
} else if (bits[0] | bits[1]) {
i = STRTOG_Denormal;
ex = 1;
} else {
i = STRTOG_Zero;
}
b->kind = i;
b->ex = ex - (0x3ff + 52);
}
static void xfpbits(union U *u, struct FPBits *b) {
uint32_t *bits;
int ex, i;
b->fpi = &kFpiLdbl;
b->sign = u->us[4] & 0x8000;
bits = b->bits;
bits[1] = ((unsigned)u->us[3] << 16) | u->us[2];
bits[0] = ((unsigned)u->us[1] << 16) | u->us[0];
if ((ex = u->us[4] & 0x7fff) != 0) {
i = STRTOG_Normal;
if (ex == 0x7fff) // Infinity or NaN
i = bits[0] | bits[1] ? STRTOG_NaN : STRTOG_Infinite;
} else if (bits[0] | bits[1]) {
i = STRTOG_Denormal;
ex = 1;
} else {
i = STRTOG_Zero;
}
b->kind = i;
b->ex = ex - (0x3fff + 63);
}
// returns number of hex digits minus 1, or 0 for zero
static int fpiprec(struct FPBits *b) {
FPI *fpi;
int i, j, k, m;
uint32_t *bits;
if (b->kind == STRTOG_Zero) return (b->ex = 0);
fpi = b->fpi;
bits = b->bits;
for (k = (fpi->nbits - 1) >> 2; k > 0; --k) {
if ((bits[k >> 3] >> 4 * (k & 7)) & 0xf) {
m = k >> 3;
for (i = 0; i <= m; ++i)
if (bits[i]) {
if (i > 0) {
k -= 8 * i;
b->ex += 32 * i;
for (j = i; j <= m; ++j) {
bits[j - i] = bits[j];
}
}
break;
}
for (i = 0; i < 28 && !((bits[0] >> i) & 0xf); i += 4) donothing;
if (i) {
b->ex += i;
m = k >> 3;
k -= (i >> 2);
for (j = 0;; ++j) {
bits[j] >>= i;
if (j == m) break;
bits[j] |= bits[j + 1] << (32 - i);
}
}
break;
}
}
return k;
}
// round to prec hex digits after the "."
// prec1 = incoming precision (after ".")
static int bround(struct FPBits *b, int prec, int prec1) {
uint32_t *bits, t;
int i, inc, j, k, m, n;
m = prec1 - prec;
bits = b->bits;
inc = 0;
k = m - 1;
if ((t = bits[k >> 3] >> (j = (k & 7) * 4)) & 8) {
if (t & 7) goto inc1;
if (j && bits[k >> 3] << (32 - j)) goto inc1;
while (k >= 8) {
k -= 8;
if (bits[k >> 3]) {
inc1:
inc = 1;
goto haveinc;
}
}
}
haveinc:
b->ex += m * 4;
i = m >> 3;
k = prec1 >> 3;
j = i;
if ((n = 4 * (m & 7)))
for (;; ++j) {
bits[j - i] = bits[j] >> n;
if (j == k) break;
bits[j - i] |= bits[j + 1] << (32 - n);
}
else
for (;; ++j) {
bits[j - i] = bits[j];
if (j == k) break;
}
k = prec >> 3;
if (inc) {
for (j = 0; !(++bits[j] & 0xffffffff); ++j) donothing;
if (j > k) {
onebit:
bits[0] = 1;
b->ex += 4 * prec;
return 1;
}
if ((j = prec & 7) < 7 && bits[k] >> (j + 1) * 4) goto onebit;
}
for (i = 0; !(bits[i >> 3] & (0xf << 4 * (i & 7))); ++i) donothing;
if (i) {
b->ex += 4 * i;
prec -= i;
j = i >> 3;
i &= 7;
i *= 4;
for (m = j;; ++m) {
bits[m - j] = bits[m] >> i;
if (m == k) break;
bits[m - j] |= bits[m + 1] << (32 - i);
}
}
return prec;
}
int __fmt_dtoa(int (*out)(const char *, void *, size_t), void *arg, int d,
int flags, int prec, int sign, int width, bool longdouble,
char qchar, unsigned char signbit, const char *alphabet,
va_list va) {
double x;
union U u;
struct FPBits fpb;
char *s, *q, *se, *s0, special[8];
int c, k, i1, ui, bw, bex, sgn, prec1, decpt;
x = 0;
switch (d) {
case 'F':
case 'f':
if (!(flags & FLAGS_PRECISION)) prec = 6;
if (!longdouble) {
x = va_arg(va, double);
s = s0 = dtoa(x, 3, prec, &decpt, &fpb.sign, &se);
if (decpt == 9999) {
if (s && s[0] == 'N') {
fpb.kind = STRTOG_NaN;
} else {
fpb.kind = STRTOG_Infinite;
}
}
} else {
u.ld = va_arg(va, long double);
xfpbits(&u, &fpb);
s = s0 =
gdtoa(fpb.fpi, fpb.ex, fpb.bits, &fpb.kind, 3, prec, &decpt, &se);
}
if (decpt == 9999) {
Format9999:
if (s0) freedtoa(s0);
bzero(special, sizeof(special));
s = q = special;
if (fpb.sign) {
*q++ = '-';
} else if (flags & FLAGS_PLUS) {
*q++ = '+';
} else if (flags & FLAGS_SPACE) {
*q++ = ' ';
}
memcpy(q, kSpecialFloats[fpb.kind == STRTOG_NaN][d >= 'a'], 4);
flags &= ~(FLAGS_PRECISION | FLAGS_PLUS | FLAGS_HASH | FLAGS_SPACE);
prec = 0;
return __fmt_stoa(out, arg, s, flags, prec, width, signbit, qchar);
}
FormatReal:
if (fpb.sign /* && (x || sign) */) sign = '-';
if (prec > 0) width -= prec;
if (width > 0) {
if (sign) --width;
if (decpt <= 0) {
--width;
if (prec > 0) --width;
} else {
if (s == se) decpt = 1;
width -= decpt;
if (prec > 0 || (flags & FLAGS_HASH)) --width;
}
}
if (width > 0 && !(flags & FLAGS_LEFT)) {
if ((flags & FLAGS_ZEROPAD)) {
if (sign) __FMT_PUT(sign);
sign = 0;
do __FMT_PUT('0');
while (--width > 0);
} else
do __FMT_PUT(' ');
while (--width > 0);
}
if (sign) __FMT_PUT(sign);
if (decpt <= 0) {
__FMT_PUT('0');
if (prec > 0 || (flags & FLAGS_HASH)) __FMT_PUT('.');
while (decpt < 0) {
__FMT_PUT('0');
prec--;
decpt++;
}
} else {
do {
if ((c = *s)) {
s++;
} else {
c = '0';
}
__FMT_PUT(c);
} while (--decpt > 0);
if (prec > 0 || (flags & FLAGS_HASH)) __FMT_PUT('.');
}
while (--prec >= 0) {
if ((c = *s)) {
s++;
} else {
c = '0';
}
__FMT_PUT(c);
}
while (--width >= 0) __FMT_PUT(' ');
if (s0) freedtoa(s0);
break;
case 'G':
case 'g':
if (!(flags & FLAGS_PRECISION)) prec = 6;
if (prec < 0) prec = 0;
if (!longdouble) {
x = va_arg(va, double);
s = s0 = dtoa(x, prec ? 2 : 0, prec, &decpt, &fpb.sign, &se);
if (decpt == 9999) {
if (s && s[0] == 'N') {
fpb.kind = STRTOG_NaN;
} else {
fpb.kind = STRTOG_Infinite;
}
}
} else {
u.ld = va_arg(va, long double);
xfpbits(&u, &fpb);
s = s0 = gdtoa(fpb.fpi, fpb.ex, fpb.bits, &fpb.kind, prec ? 2 : 0, prec,
&decpt, &se);
}
if (decpt == 9999) goto Format9999;
c = se - s;
prec1 = prec;
if (!prec) {
prec = c;
prec1 = c + (s[1] || (flags & FLAGS_HASH) ? 5 : 4);
// %.0g gives 10 rather than 1e1
}
if (decpt > -4 && decpt <= prec1) {
if ((flags & FLAGS_HASH))
prec -= decpt;
else
prec = c - decpt;
if (prec < 0) prec = 0;
goto FormatReal;
}
d -= 2;
if (!(flags & FLAGS_HASH) && prec > c) prec = c;
--prec;
goto FormatExpo;
case 'e':
case 'E':
if (!(flags & FLAGS_PRECISION)) prec = 6;
if (prec < 0) prec = 0;
if (!longdouble) {
x = va_arg(va, double);
s = s0 = dtoa(x, prec ? 2 : 0, prec + 1, &decpt, &fpb.sign, &se);
if (decpt == 9999) {
if (s && s[0] == 'N') {
fpb.kind = STRTOG_NaN;
} else {
fpb.kind = STRTOG_Infinite;
}
}
} else {
u.ld = va_arg(va, long double);
xfpbits(&u, &fpb);
s = s0 = gdtoa(fpb.fpi, fpb.ex, fpb.bits, &fpb.kind, prec ? 2 : 0, prec,
&decpt, &se);
}
if (decpt == 9999) goto Format9999;
FormatExpo:
if (fpb.sign /* && (x || sign) */) sign = '-';
if ((width -= prec + 5) > 0) {
if (sign) --width;
if (prec || (flags & FLAGS_HASH)) --width;
}
if ((c = --decpt) < 0) c = -c;
while (c >= 100) {
--width;
c /= 10;
}
if (width > 0 && !(flags & FLAGS_LEFT)) {
if ((flags & FLAGS_ZEROPAD)) {
if (sign) __FMT_PUT(sign);
sign = 0;
do __FMT_PUT('0');
while (--width > 0);
} else
do __FMT_PUT(' ');
while (--width > 0);
}
if (sign) __FMT_PUT(sign);
__FMT_PUT(*s++);
if (prec || (flags & FLAGS_HASH)) __FMT_PUT('.');
while (--prec >= 0) {
if ((c = *s))
s++;
else
c = '0';
__FMT_PUT(c);
}
__FMT_PUT(d);
if (decpt < 0) {
__FMT_PUT('-');
decpt = -decpt;
} else
__FMT_PUT('+');
for (c = 2, k = 10; 10 * k <= decpt; c++, k *= 10) donothing;
for (;;) {
i1 = decpt / k;
__FMT_PUT(i1 + '0');
if (--c <= 0) break;
decpt -= i1 * k;
decpt *= 10;
}
while (--width >= 0) __FMT_PUT(' ');
freedtoa(s0);
break;
case 'A':
alphabet = "0123456789ABCDEFPX";
goto FormatBinary;
case 'a':
alphabet = "0123456789abcdefpx";
FormatBinary:
if (longdouble) {
u.ld = va_arg(va, long double);
xfpbits(&u, &fpb);
} else {
u.d = va_arg(va, double);
dfpbits(&u, &fpb);
}
if (fpb.kind == STRTOG_Infinite || fpb.kind == STRTOG_NaN) {
s0 = 0;
goto Format9999;
}
prec1 = fpiprec(&fpb);
if ((flags & FLAGS_PRECISION) && prec < prec1) {
prec1 = bround(&fpb, prec, prec1);
}
bw = 1;
bex = fpb.ex + 4 * prec1;
if (bex) {
if ((i1 = bex) < 0) i1 = -i1;
while (i1 >= 10) {
++bw;
i1 /= 10;
}
}
if (fpb.sign /* && (sign || fpb.kind != STRTOG_Zero) */) {
sign = '-';
}
if ((width -= bw + 5) > 0) {
if (sign) --width;
if (prec1 || (flags & FLAGS_HASH)) --width;
}
if ((width -= prec1) > 0 && !(flags & FLAGS_LEFT) &&
!(flags & FLAGS_ZEROPAD)) {
do __FMT_PUT(' ');
while (--width > 0);
}
if (sign) __FMT_PUT(sign);
__FMT_PUT('0');
__FMT_PUT(alphabet[17]);
if ((flags & FLAGS_ZEROPAD) && width > 0 && !(flags & FLAGS_LEFT)) {
do __FMT_PUT('0');
while (--width > 0);
}
i1 = prec1 & 7;
k = prec1 >> 3;
__FMT_PUT(alphabet[(fpb.bits[k] >> 4 * i1) & 0xf]);
if (prec1 > 0 || (flags & FLAGS_HASH)) __FMT_PUT('.');
if (prec1 > 0) {
prec -= prec1;
while (prec1 > 0) {
if (--i1 < 0) {
if (--k < 0) break;
i1 = 7;
}
__FMT_PUT(alphabet[(fpb.bits[k] >> 4 * i1) & 0xf]);
--prec1;
}
if ((flags & FLAGS_HASH) && prec > 0) {
do __FMT_PUT(0);
while (--prec > 0);
}
}
__FMT_PUT(alphabet[16]);
if (bex < 0) {
__FMT_PUT('-');
bex = -bex;
} else
__FMT_PUT('+');
for (c = 1; 10 * c <= bex; c *= 10) donothing;
for (;;) {
i1 = bex / c;
__FMT_PUT('0' + i1);
if (!--bw) break;
bex -= i1 * c;
bex *= 10;
}
while (--width >= 0) __FMT_PUT(' ');
break;
default:
unreachable;
}
return 0;
}