mirror of
https://github.com/storytold/vectorcraft.git
synced 2026-11-10 23:01:02 +00:00
dotCooCoo
02fffcd00c
A realm now has String, Number, RegExp and Date as ECMAScript 5.1 sections 15.5, 15.7, 15.9 and 15.10 define them, with Annex B's substr, getYear, setYear and toGMTString. - String: every ES5 prototype method; split, replace, match and search take strings and RegExps. Each new string's length is checked against the heap limit before it is built, and scans take fuel. - Number: toString(radix), toFixed, toExponential and toPrecision format from the number's exact decimal value. - RegExp: a backtracking matcher keeps its choice points and capture state on the heap, within the heap room left, and takes fuel for every step, so a catastrophic pattern stops at the fuel limit. Patterns nested past Options::max_depth give a SyntaxError. Flags g, i and m, lastIndex, exec, test and toString. - Date: chrono converts days to calendar dates, shifting by whole 400-year cycles past its years. Realm::set_clock and Realm::set_time_zone set the clock and local time (UTC by default). Without a hint, a Date converts to a primitive as a string. cargo xtask test262 also runs the suite's built-ins tests for the four and prints their counts.
374 lines
13 KiB
Rust
374 lines
13 KiB
Rust
//! Number to string conversion (ECMA-262 5.1 section 9.8.1), for numeric property names, and the
|
||
//! formats of `Number.prototype`'s `toString(radix)`, `toFixed`, `toExponential` and
|
||
//! `toPrecision` (section 15.7.4). The fixed, exponential and precision formats round the
|
||
//! number's exact decimal value, ties away from zero, as the specification's "larger n" says.
|
||
|
||
/// The string ToString gives for the number `x`: the shortest decimal digits that read back as
|
||
/// `x`, laid out as section 9.8.1 says.
|
||
pub fn to_string(x: f64) -> String {
|
||
if x.is_nan() {
|
||
return "NaN".into();
|
||
}
|
||
if x == 0.0 {
|
||
return "0".into();
|
||
}
|
||
if x.is_infinite() {
|
||
return if x > 0.0 { "Infinity".into() } else { "-Infinity".into() };
|
||
}
|
||
if x < 0.0 {
|
||
return format!("-{}", to_string(-x));
|
||
}
|
||
// Rust's `{:e}` writes the shortest digits that round-trip: `d.ddde±x`.
|
||
let sci = format!("{x:e}");
|
||
let (mantissa, exp) = sci.split_once('e').unwrap_or((&sci, "0"));
|
||
let digits: String = mantissa.chars().filter(char::is_ascii_digit).collect();
|
||
let k = digits.len() as i64;
|
||
let n = exp.parse::<i64>().unwrap_or(0) + 1;
|
||
if k <= n && n <= 21 {
|
||
let zeros = usize::try_from(n - k).unwrap_or(0);
|
||
format!("{digits}{}", "0".repeat(zeros))
|
||
} else if 0 < n && n <= 21 {
|
||
let (int, frac) = digits.split_at_checked(usize::try_from(n).unwrap_or(0)).unwrap_or((&digits, ""));
|
||
format!("{int}.{frac}")
|
||
} else if -6 < n && n <= 0 {
|
||
let zeros = usize::try_from(-n).unwrap_or(0);
|
||
format!("0.{}{digits}", "0".repeat(zeros))
|
||
} else {
|
||
let e = n - 1;
|
||
let sign = if e < 0 { '-' } else { '+' };
|
||
let (first, rest) = digits.split_at_checked(1).unwrap_or((&digits, ""));
|
||
if rest.is_empty() { format!("{first}e{sign}{}", e.abs()) } else { format!("{first}.{rest}e{sign}{}", e.abs()) }
|
||
}
|
||
}
|
||
|
||
/// The exact decimal digits of a positive finite `x`, without trailing zeros, and the position of
|
||
/// the decimal point: `x = 0.d1d2… × 10^point`.
|
||
fn exact_digits(x: f64) -> (Vec<u8>, i64) {
|
||
// A double's exact decimal expansion has at most 767 significant digits.
|
||
let sci = format!("{x:.800e}");
|
||
let (mantissa, exp) = sci.split_once('e').unwrap_or((&sci, "0"));
|
||
let mut digits: Vec<u8> = mantissa.bytes().filter(u8::is_ascii_digit).map(|b| b - b'0').collect();
|
||
while digits.len() > 1 && digits.last() == Some(&0) {
|
||
digits.pop();
|
||
}
|
||
(digits, exp.parse::<i64>().unwrap_or(0) + 1)
|
||
}
|
||
|
||
/// The shortest digits that read back as a positive finite `x`, and the decimal point's position.
|
||
fn shortest_digits(x: f64) -> (Vec<u8>, i64) {
|
||
let sci = format!("{x:e}");
|
||
let (mantissa, exp) = sci.split_once('e').unwrap_or((&sci, "0"));
|
||
(mantissa.bytes().filter(u8::is_ascii_digit).map(|b| b - b'0').collect(), exp.parse::<i64>().unwrap_or(0) + 1)
|
||
}
|
||
|
||
/// Rounds `digits` (a number `0.d1d2… × 10^point`) to its first `keep` digits, a half rounding
|
||
/// up. Returns exactly `keep` digits and the new point; `keep` may be 0 or negative, which leaves
|
||
/// either nothing or a single `1` one place up.
|
||
fn round_digits(digits: &[u8], point: i64, keep: i64) -> (Vec<u8>, i64) {
|
||
if keep < 0 {
|
||
return (Vec::new(), point);
|
||
}
|
||
let k = usize::try_from(keep).unwrap_or(usize::MAX);
|
||
if digits.len() <= k {
|
||
let mut out = digits.to_vec();
|
||
out.resize(k, 0);
|
||
return (out, point);
|
||
}
|
||
let mut out: Vec<u8> = digits.get(..k).unwrap_or_default().to_vec();
|
||
if digits.get(k).is_some_and(|&d| d >= 5) {
|
||
let mut i = out.len();
|
||
loop {
|
||
if i == 0 {
|
||
out.insert(0, 1);
|
||
out.truncate(k.max(1));
|
||
return (out, point + 1);
|
||
}
|
||
i -= 1;
|
||
match out.get_mut(i) {
|
||
Some(d) if *d == 9 => *d = 0,
|
||
Some(d) => {
|
||
*d += 1;
|
||
break;
|
||
}
|
||
None => break,
|
||
}
|
||
}
|
||
}
|
||
(out, point)
|
||
}
|
||
|
||
fn digits_text(digits: &[u8]) -> String {
|
||
digits.iter().map(|d| char::from(b'0' + d)).collect()
|
||
}
|
||
|
||
/// `Number.prototype.toFixed` (section 15.7.4.5) for `f` digits after the point, 0 to 20.
|
||
pub(crate) fn to_fixed(x: f64, f: usize) -> String {
|
||
if x.is_nan() {
|
||
return "NaN".into();
|
||
}
|
||
if x.abs() >= 1e21 {
|
||
return to_string(x);
|
||
}
|
||
let sign = if x < 0.0 { "-" } else { "" };
|
||
let x = x.abs();
|
||
let fi = i64::try_from(f).unwrap_or(0);
|
||
let n = if x == 0.0 {
|
||
vec![0]
|
||
} else {
|
||
let (digits, point) = exact_digits(x);
|
||
let (mut n, new_point) = round_digits(&digits, point, point + fi);
|
||
if n.is_empty() {
|
||
n.push(0);
|
||
} else if new_point > point {
|
||
// A carry past the first digit leaves one digit more than kept.
|
||
n.push(0);
|
||
n.truncate(usize::try_from(new_point + fi).unwrap_or(n.len()));
|
||
}
|
||
n
|
||
};
|
||
let mut m = digits_text(&n);
|
||
let m_trim = m.trim_start_matches('0');
|
||
m = if m_trim.is_empty() { "0".into() } else { m_trim.into() };
|
||
if f == 0 {
|
||
return format!("{sign}{m}");
|
||
}
|
||
if m.len() <= f {
|
||
m = format!("{}{m}", "0".repeat(f + 1 - m.len()));
|
||
}
|
||
let (int, frac) = m.split_at(m.len() - f);
|
||
format!("{sign}{int}.{frac}")
|
||
}
|
||
|
||
/// `Number.prototype.toExponential` (section 15.7.4.6) for `f` digits after the point (0 to 20),
|
||
/// or as many as the number needs when `f` is `None`.
|
||
pub(crate) fn to_exponential(x: f64, f: Option<usize>) -> String {
|
||
if !x.is_finite() {
|
||
return to_string(x);
|
||
}
|
||
let sign = if x < 0.0 { "-" } else { "" };
|
||
let x = x.abs();
|
||
let (digits, e) = if x == 0.0 {
|
||
(vec![0; f.unwrap_or(0) + 1], 0)
|
||
} else {
|
||
match f {
|
||
None => {
|
||
let (d, point) = shortest_digits(x);
|
||
(d, point - 1)
|
||
}
|
||
Some(f) => {
|
||
let (d, point) = exact_digits(x);
|
||
let (d, point) = round_digits(&d, point, i64::try_from(f).unwrap_or(0) + 1);
|
||
(d, point - 1)
|
||
}
|
||
}
|
||
};
|
||
let text = digits_text(&digits);
|
||
let (first, rest) = text.split_at(1.min(text.len()));
|
||
let mantissa = if rest.is_empty() { first.to_string() } else { format!("{first}.{rest}") };
|
||
let esign = if e < 0 { '-' } else { '+' };
|
||
format!("{sign}{mantissa}e{esign}{}", e.abs())
|
||
}
|
||
|
||
/// `Number.prototype.toPrecision` (section 15.7.4.7) for `p` significant digits, 1 to 21.
|
||
pub(crate) fn to_precision(x: f64, p: usize) -> String {
|
||
if !x.is_finite() {
|
||
return to_string(x);
|
||
}
|
||
let sign = if x < 0.0 { "-" } else { "" };
|
||
let x = x.abs();
|
||
let pi = i64::try_from(p).unwrap_or(1);
|
||
let (digits, e) = if x == 0.0 {
|
||
(vec![0; p], 0)
|
||
} else {
|
||
let (d, point) = exact_digits(x);
|
||
let (d, point) = round_digits(&d, point, pi);
|
||
(d, point - 1)
|
||
};
|
||
let m = digits_text(&digits);
|
||
if e < -6 || e >= pi {
|
||
let (first, rest) = m.split_at(1.min(m.len()));
|
||
let mantissa = if rest.is_empty() { first.to_string() } else { format!("{first}.{rest}") };
|
||
let esign = if e < 0 { '-' } else { '+' };
|
||
return format!("{sign}{mantissa}e{esign}{}", e.abs());
|
||
}
|
||
if e == pi - 1 {
|
||
return format!("{sign}{m}");
|
||
}
|
||
if e >= 0 {
|
||
let (int, frac) = m.split_at(usize::try_from(e + 1).unwrap_or(0).min(m.len()));
|
||
return format!("{sign}{int}.{frac}");
|
||
}
|
||
format!("{sign}0.{}{m}", "0".repeat(usize::try_from(-(e + 1)).unwrap_or(0)))
|
||
}
|
||
|
||
/// `Number.prototype.toString(radix)` (section 15.7.4.2) for a radix of 2 to 36 other than 10:
|
||
/// the integer part exactly, and fraction digits until the number would read back unchanged.
|
||
pub(crate) fn to_radix_string(x: f64, radix: u32) -> String {
|
||
if !x.is_finite() || x == 0.0 {
|
||
return to_string(x);
|
||
}
|
||
const CHARS: &[u8; 36] = b"0123456789abcdefghijklmnopqrstuvwxyz";
|
||
let r = f64::from(radix);
|
||
let sign = if x < 0.0 { "-" } else { "" };
|
||
let x = x.abs();
|
||
let mut integer = x.floor();
|
||
let mut fraction = x - integer;
|
||
// Half the distance to the next double: digits past this resolution are not needed.
|
||
let next = f64::from_bits(x.to_bits() + 1);
|
||
let mut delta = (0.5 * (next - x)).max(f64::from_bits(1));
|
||
let mut frac_digits: Vec<u8> = Vec::new();
|
||
if fraction >= delta {
|
||
loop {
|
||
fraction *= r;
|
||
delta *= r;
|
||
let digit = fraction.floor();
|
||
frac_digits.push(digit as u8);
|
||
fraction -= digit;
|
||
if (fraction > 0.5 || (fraction == 0.5 && (digit as u8) & 1 == 1)) && fraction + delta > 1.0 {
|
||
// Round up, carrying into the integer part when every digit is the largest.
|
||
loop {
|
||
match frac_digits.last_mut() {
|
||
None => {
|
||
integer += 1.0;
|
||
break;
|
||
}
|
||
Some(d) if u32::from(*d) + 1 == radix => {
|
||
frac_digits.pop();
|
||
}
|
||
Some(d) => {
|
||
*d += 1;
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
break;
|
||
}
|
||
if fraction < delta {
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
let mut int_digits: Vec<u8> = Vec::new();
|
||
// Digits below the double's precision are zeros.
|
||
while (integer / r) >= 9_007_199_254_740_992.0 {
|
||
integer /= r;
|
||
int_digits.push(0);
|
||
}
|
||
loop {
|
||
let rem = integer % r;
|
||
int_digits.push(rem as u8);
|
||
integer = (integer - rem) / r;
|
||
if integer <= 0.0 {
|
||
break;
|
||
}
|
||
}
|
||
let mut out = String::from(sign);
|
||
out.extend(int_digits.iter().rev().map(|&d| char::from(CHARS.get(usize::from(d)).copied().unwrap_or(b'0'))));
|
||
if !frac_digits.is_empty() {
|
||
out.push('.');
|
||
out.extend(frac_digits.iter().map(|&d| char::from(CHARS.get(usize::from(d)).copied().unwrap_or(b'0'))));
|
||
}
|
||
out
|
||
}
|
||
|
||
#[cfg(test)]
|
||
mod tests {
|
||
use super::{to_exponential, to_fixed, to_precision, to_radix_string, to_string};
|
||
|
||
#[test]
|
||
fn fixed_exponential_and_precision_round_the_exact_value_with_ties_away_from_zero() {
|
||
let fixed: &[(f64, usize, &str)] = &[
|
||
(0.5, 0, "1"),
|
||
(2.5, 0, "3"),
|
||
(1.45, 1, "1.4"),
|
||
(1.255, 2, "1.25"),
|
||
(1.005, 2, "1.00"),
|
||
(0.1, 20, "0.10000000000000000555"),
|
||
(123.456, 2, "123.46"),
|
||
(-1.5, 0, "-2"),
|
||
(-0.0000001, 2, "-0.00"),
|
||
(-0.0, 2, "0.00"),
|
||
(0.0, 0, "0"),
|
||
(99.995, 2, "100.00"),
|
||
(9.5, 0, "10"),
|
||
(1e21, 2, "1e+21"),
|
||
(1000000000000000128.0, 0, "1000000000000000128"),
|
||
(f64::NAN, 2, "NaN"),
|
||
];
|
||
for (x, f, want) in fixed {
|
||
assert_eq!(to_fixed(*x, *f), *want, "({x}).toFixed({f})");
|
||
}
|
||
let exp: &[(f64, Option<usize>, &str)] = &[
|
||
(123456.0, Some(2), "1.23e+5"),
|
||
(123456.0, None, "1.23456e+5"),
|
||
(0.00015, Some(1), "1.5e-4"),
|
||
(1.25, Some(1), "1.3e+0"),
|
||
(0.0, Some(2), "0.00e+0"),
|
||
(0.0, None, "0e+0"),
|
||
(-6.9e-11, Some(1), "-6.9e-11"),
|
||
(9.99, Some(1), "1.0e+1"),
|
||
(f64::INFINITY, Some(2), "Infinity"),
|
||
];
|
||
for (x, f, want) in exp {
|
||
assert_eq!(to_exponential(*x, *f), *want, "({x}).toExponential({f:?})");
|
||
}
|
||
let prec: &[(f64, usize, &str)] = &[
|
||
(123.456, 4, "123.5"),
|
||
(0.000123, 2, "0.00012"),
|
||
(0.0000001234, 2, "1.2e-7"),
|
||
(123456.0, 2, "1.2e+5"),
|
||
(123456.0, 6, "123456"),
|
||
(0.0, 3, "0.00"),
|
||
(1.0, 1, "1"),
|
||
(99.5, 2, "1.0e+2"),
|
||
(1e21, 21, "1.00000000000000000000e+21"),
|
||
];
|
||
for (x, p, want) in prec {
|
||
assert_eq!(to_precision(*x, *p), *want, "({x}).toPrecision({p})");
|
||
}
|
||
}
|
||
|
||
#[test]
|
||
fn radix_strings_give_exact_integers_and_enough_fraction_digits() {
|
||
let cases: &[(f64, u32, &str)] = &[
|
||
(255.0, 16, "ff"),
|
||
(255.0, 2, "11111111"),
|
||
(-255.0, 36, "-73"),
|
||
(0.5, 2, "0.1"),
|
||
(0.5, 16, "0.8"),
|
||
(3.75, 2, "11.11"),
|
||
(2f64.powi(60), 16, "1000000000000000"),
|
||
(1e21, 36, "5v1j4f4ds7c000"),
|
||
(0.1, 3, "0.0022002200220022002200220022002201"),
|
||
];
|
||
for (x, r, want) in cases {
|
||
assert_eq!(to_radix_string(*x, *r), *want, "({x}).toString({r})");
|
||
}
|
||
}
|
||
|
||
#[test]
|
||
fn numbers_print_as_section_9_8_1_lays_them_out() {
|
||
let cases: &[(f64, &str)] = &[
|
||
(0.0, "0"),
|
||
(-0.0, "0"),
|
||
(1.0, "1"),
|
||
(-1.5, "-1.5"),
|
||
(123.456, "123.456"),
|
||
(1e21, "1e+21"),
|
||
(1e20, "100000000000000000000"),
|
||
(0.000001, "0.000001"),
|
||
(1e-7, "1e-7"),
|
||
(1.5e-7, "1.5e-7"),
|
||
(0.1 + 0.2, "0.30000000000000004"),
|
||
(f64::NAN, "NaN"),
|
||
(f64::INFINITY, "Infinity"),
|
||
(f64::NEG_INFINITY, "-Infinity"),
|
||
(5e-324, "5e-324"),
|
||
(1.7976931348623157e308, "1.7976931348623157e+308"),
|
||
];
|
||
for (x, want) in cases {
|
||
assert_eq!(to_string(*x), *want, "{x:?}");
|
||
}
|
||
}
|
||
}
|