272 lines
10 KiB
Rust
272 lines
10 KiB
Rust
/// Note: window size 32 or 64, is faster because we can
|
|
/// speedup modulo operations, but always computes hash 0
|
|
/// for constant data streams .. 0,0,0,0,0,0
|
|
/// so we use a modified the chunk boundary test too not
|
|
/// use hash value 0 to detect a boundary.
|
|
const CA_CHUNKER_WINDOW_SIZE: usize = 64;
|
|
|
|
/// Sliding window chunker (Buzhash)
|
|
///
|
|
/// This is a rewrite of *casync* chunker (cachunker.h) in rust.
|
|
///
|
|
/// Hashing by cyclic polynomial (also called Buzhash) has the benefit
|
|
/// of avoiding multiplications, using barrel shifts instead. For more
|
|
/// information please take a look at the [Rolling
|
|
/// Hash](https://en.wikipedia.org/wiki/Rolling_hash) article from
|
|
/// Wikipedia.
|
|
|
|
pub struct Chunker {
|
|
h: u32,
|
|
window_size: usize,
|
|
chunk_size: usize,
|
|
|
|
chunk_size_min: usize,
|
|
chunk_size_max: usize,
|
|
_chunk_size_avg: usize,
|
|
|
|
_discriminator: u32,
|
|
|
|
break_test_mask: u32,
|
|
break_test_minimum: u32,
|
|
|
|
window: [u8; CA_CHUNKER_WINDOW_SIZE],
|
|
}
|
|
|
|
const BUZHASH_TABLE: [u32; 256] = [
|
|
0x458be752, 0xc10748cc, 0xfbbcdbb8, 0x6ded5b68, 0xb10a82b5, 0x20d75648, 0xdfc5665f, 0xa8428801,
|
|
0x7ebf5191, 0x841135c7, 0x65cc53b3, 0x280a597c, 0x16f60255, 0xc78cbc3e, 0x294415f5, 0xb938d494,
|
|
0xec85c4e6, 0xb7d33edc, 0xe549b544, 0xfdeda5aa, 0x882bf287, 0x3116737c, 0x05569956, 0xe8cc1f68,
|
|
0x0806ac5e, 0x22a14443, 0x15297e10, 0x50d090e7, 0x4ba60f6f, 0xefd9f1a7, 0x5c5c885c, 0x82482f93,
|
|
0x9bfd7c64, 0x0b3e7276, 0xf2688e77, 0x8fad8abc, 0xb0509568, 0xf1ada29f, 0xa53efdfe, 0xcb2b1d00,
|
|
0xf2a9e986, 0x6463432b, 0x95094051, 0x5a223ad2, 0x9be8401b, 0x61e579cb, 0x1a556a14, 0x5840fdc2,
|
|
0x9261ddf6, 0xcde002bb, 0x52432bb0, 0xbf17373e, 0x7b7c222f, 0x2955ed16, 0x9f10ca59, 0xe840c4c9,
|
|
0xccabd806, 0x14543f34, 0x1462417a, 0x0d4a1f9c, 0x087ed925, 0xd7f8f24c, 0x7338c425, 0xcf86c8f5,
|
|
0xb19165cd, 0x9891c393, 0x325384ac, 0x0308459d, 0x86141d7e, 0xc922116a, 0xe2ffa6b6, 0x53f52aed,
|
|
0x2cd86197, 0xf5b9f498, 0xbf319c8f, 0xe0411fae, 0x977eb18c, 0xd8770976, 0x9833466a, 0xc674df7f,
|
|
0x8c297d45, 0x8ca48d26, 0xc49ed8e2, 0x7344f874, 0x556f79c7, 0x6b25eaed, 0xa03e2b42, 0xf68f66a4,
|
|
0x8e8b09a2, 0xf2e0e62a, 0x0d3a9806, 0x9729e493, 0x8c72b0fc, 0x160b94f6, 0x450e4d3d, 0x7a320e85,
|
|
0xbef8f0e1, 0x21d73653, 0x4e3d977a, 0x1e7b3929, 0x1cc6c719, 0xbe478d53, 0x8d752809, 0xe6d8c2c6,
|
|
0x275f0892, 0xc8acc273, 0x4cc21580, 0xecc4a617, 0xf5f7be70, 0xe795248a, 0x375a2fe9, 0x425570b6,
|
|
0x8898dcf8, 0xdc2d97c4, 0x0106114b, 0x364dc22f, 0x1e0cad1f, 0xbe63803c, 0x5f69fac2, 0x4d5afa6f,
|
|
0x1bc0dfb5, 0xfb273589, 0x0ea47f7b, 0x3c1c2b50, 0x21b2a932, 0x6b1223fd, 0x2fe706a8, 0xf9bd6ce2,
|
|
0xa268e64e, 0xe987f486, 0x3eacf563, 0x1ca2018c, 0x65e18228, 0x2207360a, 0x57cf1715, 0x34c37d2b,
|
|
0x1f8f3cde, 0x93b657cf, 0x31a019fd, 0xe69eb729, 0x8bca7b9b, 0x4c9d5bed, 0x277ebeaf, 0xe0d8f8ae,
|
|
0xd150821c, 0x31381871, 0xafc3f1b0, 0x927db328, 0xe95effac, 0x305a47bd, 0x426ba35b, 0x1233af3f,
|
|
0x686a5b83, 0x50e072e5, 0xd9d3bb2a, 0x8befc475, 0x487f0de6, 0xc88dff89, 0xbd664d5e, 0x971b5d18,
|
|
0x63b14847, 0xd7d3c1ce, 0x7f583cf3, 0x72cbcb09, 0xc0d0a81c, 0x7fa3429b, 0xe9158a1b, 0x225ea19a,
|
|
0xd8ca9ea3, 0xc763b282, 0xbb0c6341, 0x020b8293, 0xd4cd299d, 0x58cfa7f8, 0x91b4ee53, 0x37e4d140,
|
|
0x95ec764c, 0x30f76b06, 0x5ee68d24, 0x679c8661, 0xa41979c2, 0xf2b61284, 0x4fac1475, 0x0adb49f9,
|
|
0x19727a23, 0x15a7e374, 0xc43a18d5, 0x3fb1aa73, 0x342fc615, 0x924c0793, 0xbee2d7f0, 0x8a279de9,
|
|
0x4aa2d70c, 0xe24dd37f, 0xbe862c0b, 0x177c22c2, 0x5388e5ee, 0xcd8a7510, 0xf901b4fd, 0xdbc13dbc,
|
|
0x6c0bae5b, 0x64efe8c7, 0x48b02079, 0x80331a49, 0xca3d8ae6, 0xf3546190, 0xfed7108b, 0xc49b941b,
|
|
0x32baf4a9, 0xeb833a4a, 0x88a3f1a5, 0x3a91ce0a, 0x3cc27da1, 0x7112e684, 0x4a3096b1, 0x3794574c,
|
|
0xa3c8b6f3, 0x1d213941, 0x6e0a2e00, 0x233479f1, 0x0f4cd82f, 0x6093edd2, 0x5d7d209e, 0x464fe319,
|
|
0xd4dcac9e, 0x0db845cb, 0xfb5e4bc3, 0xe0256ce1, 0x09fb4ed1, 0x0914be1e, 0xa5bdb2c3, 0xc6eb57bb,
|
|
0x30320350, 0x3f397e91, 0xa67791bc, 0x86bc0e2c, 0xefa0a7e2, 0xe9ff7543, 0xe733612c, 0xd185897b,
|
|
0x329e5388, 0x91dd236b, 0x2ecb0d93, 0xf4d82a3d, 0x35b5c03f, 0xe4e606f0, 0x05b21843, 0x37b45964,
|
|
0x5eff22f4, 0x6027f4cc, 0x77178b3c, 0xae507131, 0x7bf7cabc, 0xf9c18d66, 0x593ade65, 0xd95ddf11,
|
|
];
|
|
|
|
impl Chunker {
|
|
/// Create a new Chunker instance, which produces and average
|
|
/// chunk size of `chunk_size_avg` (need to be a power of two). We
|
|
/// allow variation from `chunk_size_avg/4` up to a maximum of
|
|
/// `chunk_size_avg*4`.
|
|
pub fn new(chunk_size_avg: usize) -> Self {
|
|
// The chunk cut discriminator. In order to get an average
|
|
// chunk size of avg, we cut whenever for a hash value "h" at
|
|
// byte "i" given the descriminator "d(avg)": h(i) mod d(avg)
|
|
// == d(avg) - 1. Note that the discriminator calculated like
|
|
// this only yields correct results as long as the minimal
|
|
// chunk size is picked as avg/4, and the maximum chunk size
|
|
// as avg*4. If they are picked differently the result might
|
|
// be skewed into either direction.
|
|
let avg = chunk_size_avg as f64;
|
|
let discriminator = (avg / (-1.42888852e-7 * avg + 1.33237515)) as u32;
|
|
|
|
if chunk_size_avg.count_ones() != 1 {
|
|
panic!("got unexpected chunk size - not a power of two.");
|
|
}
|
|
|
|
let break_test_mask = (chunk_size_avg * 2 - 1) as u32;
|
|
let break_test_minimum = break_test_mask - 2;
|
|
|
|
Self {
|
|
h: 0,
|
|
window_size: 0,
|
|
chunk_size: 0,
|
|
chunk_size_min: chunk_size_avg >> 2,
|
|
chunk_size_max: chunk_size_avg << 2,
|
|
_chunk_size_avg: chunk_size_avg,
|
|
_discriminator: discriminator,
|
|
break_test_mask,
|
|
break_test_minimum,
|
|
window: [0u8; CA_CHUNKER_WINDOW_SIZE],
|
|
}
|
|
}
|
|
|
|
/// Scans the specified data for a chunk border. Returns 0 if none
|
|
/// was found (and the function should be called with more data
|
|
/// later on), or another value indicating the position of a
|
|
/// border.
|
|
pub fn scan(&mut self, data: &[u8]) -> usize {
|
|
let window_len = self.window.len();
|
|
let data_len = data.len();
|
|
|
|
let mut pos = 0;
|
|
|
|
if self.window_size < window_len {
|
|
let need = window_len - self.window_size;
|
|
let copy_len = if need < data_len { need } else { data_len };
|
|
|
|
for _i in 0..copy_len {
|
|
let byte = data[pos];
|
|
self.window[self.window_size] = byte;
|
|
self.h = self.h.rotate_left(1) ^ BUZHASH_TABLE[byte as usize];
|
|
pos += 1;
|
|
self.window_size += 1;
|
|
}
|
|
|
|
self.chunk_size += copy_len;
|
|
|
|
// return if window is still not full
|
|
if self.window_size < window_len {
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
//let mut idx = self.chunk_size % CA_CHUNKER_WINDOW_SIZE;
|
|
let mut idx = self.chunk_size & 0x3f;
|
|
|
|
while pos < data_len {
|
|
// roll window
|
|
let enter = data[pos];
|
|
let leave = self.window[idx];
|
|
self.h = self.h.rotate_left(1) ^
|
|
//BUZHASH_TABLE[leave as usize].rotate_left(CA_CHUNKER_WINDOW_SIZE as u32) ^
|
|
BUZHASH_TABLE[leave as usize] ^
|
|
BUZHASH_TABLE[enter as usize];
|
|
|
|
self.chunk_size += 1;
|
|
pos += 1;
|
|
|
|
self.window[idx] = enter;
|
|
|
|
if self.shall_break() {
|
|
self.h = 0;
|
|
self.chunk_size = 0;
|
|
self.window_size = 0;
|
|
return pos;
|
|
}
|
|
|
|
//idx = self.chunk_size % CA_CHUNKER_WINDOW_SIZE;
|
|
idx = self.chunk_size & 0x3f;
|
|
//idx += 1; if idx >= CA_CHUNKER_WINDOW_SIZE { idx = 0 };
|
|
}
|
|
|
|
0
|
|
}
|
|
|
|
// fast implementation avoiding modulo
|
|
// #[inline(always)]
|
|
fn shall_break(&self) -> bool {
|
|
if self.chunk_size >= self.chunk_size_max {
|
|
return true;
|
|
}
|
|
|
|
if self.chunk_size < self.chunk_size_min {
|
|
return false;
|
|
}
|
|
|
|
//(self.h & 0x1ffff) <= 2 //THIS IS SLOW!!!
|
|
|
|
//(self.h & self.break_test_mask) <= 2 // Bad on 0 streams
|
|
|
|
(self.h & self.break_test_mask) >= self.break_test_minimum
|
|
}
|
|
|
|
// This is the original implementation from casync
|
|
/*
|
|
#[inline(always)]
|
|
fn shall_break_orig(&self) -> bool {
|
|
|
|
if self.chunk_size >= self.chunk_size_max { return true; }
|
|
|
|
if self.chunk_size < self.chunk_size_min { return false; }
|
|
|
|
(self.h % self.discriminator) == (self.discriminator - 1)
|
|
}
|
|
*/
|
|
}
|
|
|
|
#[test]
|
|
fn test_chunker1() {
|
|
let mut buffer = Vec::new();
|
|
|
|
for i in 0..(256 * 1024) {
|
|
for j in 0..4 {
|
|
let byte = ((i >> (j << 3)) & 0xff) as u8;
|
|
buffer.push(byte);
|
|
}
|
|
}
|
|
let mut chunker = Chunker::new(64 * 1024);
|
|
|
|
let mut pos = 0;
|
|
let mut last = 0;
|
|
|
|
let mut chunks1: Vec<(usize, usize)> = vec![];
|
|
let mut chunks2: Vec<(usize, usize)> = vec![];
|
|
|
|
// test1: feed single bytes
|
|
while pos < buffer.len() {
|
|
let k = chunker.scan(&buffer[pos..pos + 1]);
|
|
pos += 1;
|
|
if k != 0 {
|
|
let prev = last;
|
|
last = pos;
|
|
chunks1.push((prev, pos - prev));
|
|
}
|
|
}
|
|
chunks1.push((last, buffer.len() - last));
|
|
|
|
let mut chunker = Chunker::new(64 * 1024);
|
|
|
|
let mut pos = 0;
|
|
|
|
// test2: feed with whole buffer
|
|
while pos < buffer.len() {
|
|
let k = chunker.scan(&buffer[pos..]);
|
|
if k != 0 {
|
|
chunks2.push((pos, k));
|
|
pos += k;
|
|
} else {
|
|
break;
|
|
}
|
|
}
|
|
|
|
chunks2.push((pos, buffer.len() - pos));
|
|
|
|
if chunks1 != chunks2 {
|
|
let mut size1 = 0;
|
|
for (_offset, len) in &chunks1 {
|
|
size1 += len;
|
|
}
|
|
println!("Chunks1:{}\n{:?}\n", size1, chunks1);
|
|
|
|
let mut size2 = 0;
|
|
for (_offset, len) in &chunks2 {
|
|
size2 += len;
|
|
}
|
|
println!("Chunks2:{}\n{:?}\n", size2, chunks2);
|
|
|
|
if size1 != 256 * 4 * 1024 {
|
|
panic!("wrong size for chunks1");
|
|
}
|
|
if size2 != 256 * 4 * 1024 {
|
|
panic!("wrong size for chunks2");
|
|
}
|
|
|
|
panic!("got different chunks");
|
|
}
|
|
}
|