-
Notifications
You must be signed in to change notification settings - Fork 14
/
chained_hash_table.rs
261 lines (217 loc) · 8.84 KB
/
chained_hash_table.rs
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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
pub const WINDOW_SIZE: usize = 32768;
const WINDOW_MASK: usize = WINDOW_SIZE - 1;
#[cfg(test)]
pub const HASH_BYTES: usize = 3;
const HASH_SHIFT: u16 = 5;
const HASH_MASK: u16 = WINDOW_MASK as u16;
/// Returns a new hash value based on the previous value and the next byte
#[inline]
fn update_hash(current_hash: u16, to_insert: u8) -> u16 {
update_hash_conf(current_hash, to_insert, HASH_SHIFT, HASH_MASK)
}
#[inline]
fn update_hash_conf(current_hash: u16, to_insert: u8, shift: u16, mask: u16) -> u16 {
((current_hash << shift) ^ (to_insert as u16)) & mask
}
fn init_array(arr: &mut [u16; WINDOW_SIZE]) {
for (n, mut b) in arr.iter_mut().enumerate() {
*b = n as u16;
}
}
pub struct ChainedHashTable {
// Current running hash value of the last 3 bytes
current_hash: u16,
// Starts of hash chains (in prev)
head: Box<[u16; WINDOW_SIZE]>,
// link to previous occurence of this hash value
prev: Box<[u16; WINDOW_SIZE]>,
// Used for testing
// Didn't find an easy way for it not to exist when debug_assertions are disabled.
pub count: u64,
}
impl ChainedHashTable {
pub fn new() -> ChainedHashTable {
let mut c = ChainedHashTable {
current_hash: 0,
head: Box::new([0; WINDOW_SIZE]),
prev: Box::new([0; WINDOW_SIZE]),
count: 0,
};
init_array(&mut c.head);
init_array(&mut c.prev);
c
}
pub fn from_starting_values(v1: u8, v2: u8) -> ChainedHashTable {
let mut t = ChainedHashTable::new();
t.current_hash = update_hash(t.current_hash, v1);
t.current_hash = update_hash(t.current_hash, v2);
t
}
/// Resets the hash value and hash chains
pub fn reset(&mut self) {
*self = ChainedHashTable::from_starting_values(55, 77);
}
pub fn add_initial_hash_values(&mut self, v1: u8, v2: u8) {
self.current_hash = update_hash(self.current_hash, v1);
self.current_hash = update_hash(self.current_hash, v2);
}
// Insert a byte into the hash table
pub fn add_hash_value(&mut self, position: usize, value: u8) {
// Check that all bytes are input in order and at the correct positions.
debug_assert_eq!(position & WINDOW_MASK, self.count as usize & WINDOW_MASK);
debug_assert!(position < WINDOW_SIZE * 2,
"Position is larger than 2 * window size! {}",
position);
// Storing the hash in a temporary variable here makes the compiler avoid the
// bounds checks in this function.
let new_hash = update_hash(self.current_hash, value);
self.prev[position & WINDOW_MASK] = self.head[new_hash as usize];
// Ignoring any bits over 16 here is deliberate, as we only concern ourselves about
// where in the buffer (which is 64k bytes) we are referring to.
self.head[new_hash as usize] = position as u16;
// Update the stored hash value with the new hash.
self.current_hash = new_hash;
if cfg!(debug_assertions) {
self.count += 1;
}
}
// Get the head of the hash chain for the current hash value
#[cfg(test)]
#[inline]
pub fn current_head(&self) -> u16 {
self.head[self.current_hash as usize]
}
#[cfg(test)]
#[inline]
pub fn current_hash(&self) -> u16 {
self.current_hash
}
#[inline]
pub fn get_prev(&self, bytes: usize) -> u16 {
self.prev[bytes & WINDOW_MASK]
}
fn slide_value(b: u16, pos: u16, bytes: u16) -> u16 {
if b >= bytes { b - bytes } else { pos }
}
fn slide_table(table: &mut [u16], bytes: u16) {
for (n, b) in table.iter_mut().enumerate() {
*b = ChainedHashTable::slide_value(*b, n as u16, bytes);
}
}
pub fn slide(&mut self, bytes: usize) {
if cfg!(debug_assertions) {
if bytes != WINDOW_SIZE {
// This should only happen in tests in this file.
self.count = 0;
} else {
assert_eq!(self.count & WINDOW_MASK as u64, 0);
}
}
ChainedHashTable::slide_table(&mut self.head[..], bytes as u16);
ChainedHashTable::slide_table(&mut self.prev[..], bytes as u16);
}
// #[cfg(test)]
pub fn _get_head_arr(&self) -> &[u16] {
&self.head[..]
}
// #[cfg(test)]
pub fn _get_prev_arr(&self) -> &[u16] {
&self.prev[..]
}
}
#[cfg(test)]
pub fn filled_hash_table(data: &[u8]) -> ChainedHashTable {
let mut hash_table = ChainedHashTable::from_starting_values(data[0], data[1]);
for (n, b) in data[2..].iter().enumerate() {
hash_table.add_hash_value(n, *b);
}
hash_table
}
#[cfg(test)]
mod test {
use super::filled_hash_table;
#[test]
fn chained_hash() {
use std::str;
let test_string = "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do \
eiusmod tempor. rum. incididunt ut labore et dolore magna aliqua. Ut \
enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi \
ut aliquip ex ea commodo consequat. rum. Duis aute irure dolor in \
reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla \
pariatur. Excepteur sint occaecat cupidatat non proident, sunt in \
culpa qui officia deserunt mollit anim id est laborum.";
let test_data = test_string.as_bytes();
let current_bytes = &test_data[test_data.len() - super::HASH_BYTES..test_data.len()];
let num_iters = test_string
.matches(str::from_utf8(current_bytes).unwrap())
.count();
let hash_table = filled_hash_table(test_data);
// Test that the positions in the chain are valid
let mut prev_value = hash_table.get_prev(hash_table.current_head() as usize) as usize;
let mut count = 0;
let mut current = hash_table.current_head() as usize;
while current != prev_value {
count += 1;
current = prev_value;
prev_value = hash_table.get_prev(prev_value) as usize;
}
// There should be at least as many occurences of the hash of the checked bytes as the
// numbers of occurences of the checked bytes themselves. As the hashes are not large enough
// to store 8 * 3 = 24 bits, there could be more with different input data.
assert!(count >= num_iters);
}
#[test]
fn table_unique() {
let mut test_data = Vec::new();
test_data.extend((0u8..255));
test_data.extend((255u8..0));
let hash_table = filled_hash_table(&test_data);
let prev_pos = hash_table.get_prev(hash_table.current_head() as usize);
// Since all sequences in the input are unique, there shouldn't be any previous values
assert_eq!(prev_pos, hash_table.current_hash());
}
#[test]
fn table_slide() {
use std::fs::File;
use std::io::Read;
use std::str;
let window_size = super::WINDOW_SIZE;
let window_size16 = super::WINDOW_SIZE as u16;
let mut input = Vec::new();
let mut f = File::open("tests/pg11.txt").unwrap();
f.read_to_end(&mut input).unwrap();
let mut hash_table = filled_hash_table(&input[..window_size + 2]);
for (n, b) in input[2..window_size + 2].iter().enumerate() {
hash_table.add_hash_value(n + window_size, *b);
}
hash_table.slide(window_size);
{
let max_head = hash_table.head.iter().max().unwrap();
// After sliding there should be no hashes referring to values
// higher than the window size
assert!(*max_head < window_size16);
assert!(*max_head > 0);
let pos = hash_table.get_prev(hash_table.current_head() as usize);
// There should be a previous occurence since we inserted the data 3 times
assert!(pos < window_size16);
assert!(pos > 0);
}
for (n, b) in input[2..(window_size / 2)].iter().enumerate() {
hash_table.add_hash_value(n + window_size, *b);
}
// There should hashes referring to values in the upper part of the input window
// at this point
let max_prev = hash_table.prev.iter().max().unwrap();
assert!(*max_prev > window_size16);
let mut pos = hash_table.current_head();
// There should be a previous occurence since we inserted the data 3 times
assert!(pos > window_size16);
let end_byte = input[(window_size / 2) - 1 - 2];
let mut iterations = 0;
while pos > window_size16 && iterations < 5000 {
assert_eq!(input[pos as usize & window_size - 1], end_byte);
pos = hash_table.get_prev(pos as usize);
iterations += 1;
}
}
}