Implement our own small-integer optimization (#7584)

## Summary

This is a follow-up to #7469 that attempts to achieve similar gains, but
without introducing malachite. Instead, this PR removes the `BigInt`
type altogether, instead opting for a simple enum that allows us to
store small integers directly and only allocate for values greater than
`i64`:

```rust
/// A Python integer literal. Represents both small (fits in an `i64`) and large integers.
#[derive(Clone, PartialEq, Eq, Hash)]
pub struct Int(Number);

#[derive(Debug, Clone, PartialEq, Eq, Hash)]
pub enum Number {
    /// A "small" number that can be represented as an `i64`.
    Small(i64),
    /// A "large" number that cannot be represented as an `i64`.
    Big(Box<str>),
}

impl std::fmt::Display for Number {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            Number::Small(value) => write!(f, "{value}"),
            Number::Big(value) => write!(f, "{value}"),
        }
    }
}
```

We typically don't care about numbers greater than `isize` -- our only
uses are comparisons against small constants (like `1`, `2`, `3`, etc.),
so there's no real loss of information, except in one or two rules where
we're now a little more conservative (with the worst-case being that we
don't flag, e.g., an `itertools.pairwise` that uses an extremely large
value for the slice start constant). For simplicity, a few diagnostics
now show a dedicated message when they see integers that are out of the
supported range (e.g., `outdated-version-block`).

An additional benefit here is that we get to remove a few dependencies,
especially `num-bigint`.

## Test Plan

`cargo test`
This commit is contained in:
Charlie Marsh 2023-09-25 11:13:21 -04:00 committed by GitHub
parent 65aebf127a
commit 93b5d8a0fb
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
40 changed files with 707 additions and 385 deletions

View file

@ -17,7 +17,6 @@ hexf-parse = "0.2.1"
is-macro.workspace = true
itertools = { workspace = true }
lexical-parse-float = { version = "0.8.0", features = ["format"] }
num-traits = { workspace = true }
unic-ucd-category = "0.9"
[dev-dependencies]

View file

@ -1,7 +1,7 @@
use crate::Case;
use num_traits::{Float, Zero};
use std::f64;
use crate::Case;
pub fn parse_str(literal: &str) -> Option<f64> {
parse_inner(literal.trim().as_bytes())
}
@ -244,46 +244,6 @@ pub fn from_hex(s: &str) -> Option<f64> {
}
}
pub fn to_hex(value: f64) -> String {
let (mantissa, exponent, sign) = value.integer_decode();
let sign_fmt = if sign < 0 { "-" } else { "" };
match value {
value if value.is_zero() => format!("{sign_fmt}0x0.0p+0"),
value if value.is_infinite() => format!("{sign_fmt}inf"),
value if value.is_nan() => "nan".to_owned(),
_ => {
const BITS: i16 = 52;
const FRACT_MASK: u64 = 0xf_ffff_ffff_ffff;
format!(
"{}{:#x}.{:013x}p{:+}",
sign_fmt,
mantissa >> BITS,
mantissa & FRACT_MASK,
exponent + BITS
)
}
}
}
#[test]
#[allow(clippy::float_cmp)]
fn test_to_hex() {
use rand::Rng;
for _ in 0..20000 {
let bytes = rand::thread_rng().gen::<[u64; 1]>();
let f = f64::from_bits(bytes[0]);
if !f.is_finite() {
continue;
}
let hex = to_hex(f);
// println!("{} -> {}", f, hex);
let roundtrip = hexf_parse::parse_hexf64(&hex, false).unwrap();
// println!(" -> {}", roundtrip);
assert_eq!(f, roundtrip, "{f} {hex} {roundtrip}");
}
}
#[test]
fn test_remove_trailing_zeros() {
assert!(remove_trailing_zeros(String::from("100")) == *"1");