1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
|
//! Variable length integer encoding.
//!
//! This uses a different encoding from the leb128 encoding used for binary DRAT and CLRAT. It uses
//! the same amount of bytes for each input, but is faster to encode and decode.
//!
//! Numbers are encoded like this:
//! ```text
//! 1xxxxxxx for up to 7 bits
//! 01xxxxxx xxxxxxxx for up to 14 bits
//! 001xxxxx xxxxxxxx xxxxxxxx for up to 21 bits
//! ...
//! ```
//!
//! The x-bits store the number from LSB to MSB. This scheme allows faster encoding and decoding, as
//! the bits are kept consecutive and the length can be determined from the first or first two
//! bytes.
//!
//! The current implementations are not optimal but fast enough to not be the bottleneck when
//! checking proofs.
use std::{
convert::{TryFrom, TryInto},
io::{self, BufRead, Write},
};
/// Write an encoded 64 bit number.
pub fn write_u64(target: &mut impl Write, mut value: u64) -> Result<(), io::Error> {
let bits = (64 - value.leading_zeros()) as u32;
let blocks = (bits * (64 / 7)) / 64;
if value < (1 << (8 * 7)) {
value = ((value << 1) | 1) << blocks;
let bytes = unsafe { std::mem::transmute::<u64, [u8; 8]>(value.to_le()) };
target.write_all(&bytes[..(blocks + 1) as usize])
} else {
let lo_data = ((value << 1) | 1) << blocks;
let lo_bytes = unsafe { std::mem::transmute::<u64, [u8; 8]>(lo_data.to_le()) };
let hi_data = value >> (64 - (blocks + 1));
let hi_bytes = unsafe { std::mem::transmute::<u64, [u8; 8]>(hi_data.to_le()) };
target.write_all(&lo_bytes)?;
target.write_all(&hi_bytes[..(blocks as usize) + 1 - 8])
}
}
/// Read an encoded 64 bit number, if at least 16 bytes lookahead are available.
fn read_u64_fast(bytes: &[u8; 16]) -> (u64, usize) {
let lo_data = u64::from_le(unsafe {
std::mem::transmute::<[u8; 8], u64>(*<&[u8; 8]>::try_from(&bytes[..8]).unwrap())
});
let len = (lo_data | (1 << 9)).trailing_zeros() + 1;
if len <= 8 {
let result = lo_data & (!0u64 >> (64 - 8 * len));
let result = result >> len;
(result, len as usize)
} else {
let hi_data = u64::from_le(unsafe {
std::mem::transmute::<[u8; 8], u64>(*<&[u8; 8]>::try_from(&bytes[8..]).unwrap())
});
let hi_data = hi_data & (!0u64 >> (64 - 8 * (len - 8)));
let result = (lo_data >> len) | (hi_data << (64 - len));
(result as u64, len as usize)
}
}
/// Read an encoded 64 bit number from a buffered reader.
///
/// This uses [`read_u64_fast`] if the remaining buffer is larger than 16 bytes and falls back to a
/// slower implementation otherwise.
pub fn read_u64(source: &mut impl BufRead) -> Result<u64, io::Error> {
let buf = source.fill_buf()?;
if buf.len() >= 16 {
let next_bytes: &[u8; 16] = (&buf[..16]).try_into().unwrap();
let (value, advance) = read_u64_fast(next_bytes);
source.consume(advance);
Ok(value)
} else {
let mut scan = 1 << 9;
let mut byte: [u8; 1] = [0];
let mut begin = 1;
source.read_exact(&mut byte[..])?;
let mut result = byte[0] as u64;
scan |= byte[0] as u64;
if byte[0] == 0 {
begin += 1;
source.read_exact(&mut byte[..])?;
result |= (byte[0] as u64) << 8;
scan |= (byte[0] as u64) << 8;
}
let len = scan.trailing_zeros() + 1;
result >>= len;
for i in begin..len {
source.read_exact(&mut byte[..])?;
result |= (byte[0] as u64) << (8 * i - len);
}
Ok(result as u64)
}
}
#[cfg(test)]
mod tests {
use super::*;
use proptest::prelude::*;
proptest! {
#[test]
fn roundtrip (
numbers in prop::collection::vec(prop::num::u64::ANY, 0..10_000)
) {
let mut buf = vec![];
for &num in numbers.iter() {
write_u64(&mut buf, num)?;
}
let mut read = std::io::BufReader::with_capacity(128, &buf[..]);
let mut out = vec![];
while let Ok(num) = read_u64(&mut read) {
out.push(num)
}
prop_assert_eq!(numbers, out);
}
}
}
|
