1 #![doc(html_root_url="https://docs.rs/phf_generator/0.7")]
2 use phf_shared::{PhfHash, HashKey};
3 use rand::{SeedableRng, Rng};
4 use rand::distributions::Standard;
5 use rand::rngs::SmallRng;
6
7 const DEFAULT_LAMBDA: usize = 5;
8
9 const FIXED_SEED: u64 = 1234567890;
10
11 pub struct HashState {
12 pub key: HashKey,
13 pub disps: Vec<(u32, u32)>,
14 pub map: Vec<usize>,
15 }
16
generate_hash<H: PhfHash>(entries: &[H]) -> HashState17 pub fn generate_hash<H: PhfHash>(entries: &[H]) -> HashState {
18 SmallRng::seed_from_u64(FIXED_SEED)
19 .sample_iter(Standard)
20 .find_map(|key| try_generate_hash(entries, key))
21 .expect("failed to solve PHF")
22 }
23
try_generate_hash<H: PhfHash>(entries: &[H], key: HashKey) -> Option<HashState>24 fn try_generate_hash<H: PhfHash>(entries: &[H], key: HashKey) -> Option<HashState> {
25 struct Bucket {
26 idx: usize,
27 keys: Vec<usize>,
28 }
29
30 let hashes: Vec<_> = entries.iter().map(|entry| phf_shared::hash(entry, &key)).collect();
31
32 let buckets_len = (hashes.len() + DEFAULT_LAMBDA - 1) / DEFAULT_LAMBDA;
33 let mut buckets = (0..buckets_len)
34 .map(|i| {
35 Bucket {
36 idx: i,
37 keys: vec![],
38 }
39 })
40 .collect::<Vec<_>>();
41
42 for (i, hash) in hashes.iter().enumerate() {
43 buckets[(hash.g % (buckets_len as u32)) as usize].keys.push(i);
44 }
45
46 // Sort descending
47 buckets.sort_by(|a, b| a.keys.len().cmp(&b.keys.len()).reverse());
48
49 let table_len = hashes.len();
50 let mut map = vec![None; table_len];
51 let mut disps = vec![(0u32, 0u32); buckets_len];
52
53 // store whether an element from the bucket being placed is
54 // located at a certain position, to allow for efficient overlap
55 // checks. It works by storing the generation in each cell and
56 // each new placement-attempt is a new generation, so you can tell
57 // if this is legitimately full by checking that the generations
58 // are equal. (A u64 is far too large to overflow in a reasonable
59 // time for current hardware.)
60 let mut try_map = vec![0u64; table_len];
61 let mut generation = 0u64;
62
63 // the actual values corresponding to the markers above, as
64 // (index, key) pairs, for adding to the main map once we've
65 // chosen the right disps.
66 let mut values_to_add = vec![];
67
68 'buckets: for bucket in &buckets {
69 for d1 in 0..(table_len as u32) {
70 'disps: for d2 in 0..(table_len as u32) {
71 values_to_add.clear();
72 generation += 1;
73
74 for &key in &bucket.keys {
75 let idx = (phf_shared::displace(hashes[key].f1, hashes[key].f2, d1, d2) %
76 (table_len as u32)) as usize;
77 if map[idx].is_some() || try_map[idx] == generation {
78 continue 'disps;
79 }
80 try_map[idx] = generation;
81 values_to_add.push((idx, key));
82 }
83
84 // We've picked a good set of disps
85 disps[bucket.idx] = (d1, d2);
86 for &(idx, key) in &values_to_add {
87 map[idx] = Some(key);
88 }
89 continue 'buckets;
90 }
91 }
92
93 // Unable to find displacements for a bucket
94 return None;
95 }
96
97 Some(HashState {
98 key,
99 disps,
100 map: map.into_iter().map(|i| i.unwrap()).collect(),
101 })
102 }
103