distillate 0.11.0 → 0.13.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +49 -3
- package/dist/bloom/index.cjs +1 -1
- package/dist/bloom/index.d.cts +1 -1
- package/dist/bloom/index.d.ts +1 -1
- package/dist/bloom/index.js +1 -1
- package/dist/countmin/index.cjs +194 -0
- package/dist/countmin/index.d.cts +178 -0
- package/dist/countmin/index.d.ts +178 -0
- package/dist/countmin/index.js +182 -0
- package/dist/cuckoo/index.cjs +278 -0
- package/dist/cuckoo/index.d.cts +164 -0
- package/dist/cuckoo/index.d.ts +164 -0
- package/dist/cuckoo/index.js +267 -0
- package/dist/hll/index.cjs +1 -1
- package/dist/hll/index.d.cts +1 -1
- package/dist/hll/index.d.ts +1 -1
- package/dist/hll/index.js +1 -1
- package/dist/index.cjs +1 -1
- package/dist/index.js +1 -1
- package/dist/scalable/index.cjs +22 -15
- package/dist/scalable/index.js +22 -15
- package/dist/{sizing-wYOW27eY.d.cts → sizing-BmKfHf_V.d.cts} +13 -1
- package/dist/{sizing-wYOW27eY.d.ts → sizing-BmKfHf_V.d.ts} +13 -1
- package/dist/{sizing-BtUrtvF_.js → sizing-D3-_7pdP.js} +9 -1
- package/dist/{sizing-BLzZk2zr.cjs → sizing-R1pwbzjC.cjs} +14 -0
- package/package.json +17 -1
|
@@ -0,0 +1,164 @@
|
|
|
1
|
+
import { t as BytesLike } from "../bytes-DCuYtUVS.js";
|
|
2
|
+
import { a as ReservedBitsError, c as UnknownHashVariantError, l as UnknownVersionError, n as ChecksumError, o as SerializationError, r as FilterJSON, s as TruncatedError, t as BadMagicError } from "../serialize--xR6vGSW.js";
|
|
3
|
+
import { t as ParamError } from "../params-DnqJBqLS.js";
|
|
4
|
+
//#region src/cuckoo/cuckoo.d.ts
|
|
5
|
+
/** Thrown when an add cannot find room; the filter is left exactly as it was. */
|
|
6
|
+
export declare class CuckooFullError extends Error {
|
|
7
|
+
/** Discriminates this error from other `Error`s. */
|
|
8
|
+
override readonly name = "CuckooFullError";
|
|
9
|
+
}
|
|
10
|
+
/** Settings for a {@link CuckooFilter}. */
|
|
11
|
+
interface CuckooParams {
|
|
12
|
+
/** Expected number of keys. */
|
|
13
|
+
n: number;
|
|
14
|
+
/** Target false-positive rate at `n` keys, e.g. `0.01` for 1%. */
|
|
15
|
+
epsilon: number;
|
|
16
|
+
/** Hash seed; defaults to `0`. */
|
|
17
|
+
seed?: number;
|
|
18
|
+
}
|
|
19
|
+
/** The optional settings {@link CuckooFilter.create} takes after `n` and `epsilon`. */
|
|
20
|
+
type CuckooOptions = Omit<CuckooParams, "n" | "epsilon">;
|
|
21
|
+
/**
|
|
22
|
+
* A cuckoo filter: a set with a tunable false-positive rate, zero false
|
|
23
|
+
* negatives, and delete. Keys are stored as short fingerprints in one of two
|
|
24
|
+
* candidate buckets (Fan et al., "Cuckoo Filter: Practically Better Than
|
|
25
|
+
* Bloom", 2014).
|
|
26
|
+
*/
|
|
27
|
+
export declare class CuckooFilter {
|
|
28
|
+
#private;
|
|
29
|
+
/**
|
|
30
|
+
* Creates a filter sized for `n` expected keys at a target false-positive rate.
|
|
31
|
+
*
|
|
32
|
+
* @param n - Expected number of keys.
|
|
33
|
+
* @param epsilon - Target false-positive rate, e.g. `0.01` for 1%.
|
|
34
|
+
* @param options - Optional seed.
|
|
35
|
+
* @returns A new, empty filter.
|
|
36
|
+
*/
|
|
37
|
+
static create(n: number, epsilon: number, options?: CuckooOptions): CuckooFilter;
|
|
38
|
+
/**
|
|
39
|
+
* Builds a filter from the given keys; duplicates are ignored. It is sized
|
|
40
|
+
* for the distinct keys at the target false-positive rate, and holds each
|
|
41
|
+
* once, so a key listed twice needs one delete.
|
|
42
|
+
*
|
|
43
|
+
* @param keys - The keys to insert.
|
|
44
|
+
* @param epsilon - Target false-positive rate, e.g. `0.01` for 1%.
|
|
45
|
+
* @param options - Optional seed.
|
|
46
|
+
* @returns A new filter containing every key.
|
|
47
|
+
*/
|
|
48
|
+
static from(keys: Iterable<BytesLike>, epsilon: number, options?: CuckooOptions): CuckooFilter;
|
|
49
|
+
/**
|
|
50
|
+
* Restores a filter from its {@link CuckooFilter.toBytes} serialization.
|
|
51
|
+
*
|
|
52
|
+
* @param bytes - The serialized filter.
|
|
53
|
+
* @returns The reconstructed filter.
|
|
54
|
+
*/
|
|
55
|
+
static fromBytes(bytes: Uint8Array): CuckooFilter;
|
|
56
|
+
/**
|
|
57
|
+
* Restores a filter from its {@link CuckooFilter.toJSON} envelope.
|
|
58
|
+
*
|
|
59
|
+
* @param value - The parsed JSON envelope.
|
|
60
|
+
* @returns The reconstructed filter.
|
|
61
|
+
*/
|
|
62
|
+
static fromJSON(value: unknown): CuckooFilter;
|
|
63
|
+
/**
|
|
64
|
+
* Constructs a filter from {@link CuckooParams}.
|
|
65
|
+
*
|
|
66
|
+
* @throws {@link ParamError} if a setting is invalid.
|
|
67
|
+
*/
|
|
68
|
+
constructor({ n, epsilon, seed }: CuckooParams);
|
|
69
|
+
/**
|
|
70
|
+
* Adds `key`. Every call stores a fingerprint, including for a key already
|
|
71
|
+
* present, so each add needs its own delete.
|
|
72
|
+
*
|
|
73
|
+
* @throws {@link CuckooFullError} if the filter has no room for it.
|
|
74
|
+
*/
|
|
75
|
+
add(key: BytesLike): void;
|
|
76
|
+
/**
|
|
77
|
+
* Removes one copy of `key`: the first slot holding its fingerprint in
|
|
78
|
+
* either candidate bucket.
|
|
79
|
+
*
|
|
80
|
+
* Only delete keys you added. A key never added can share a fingerprint and
|
|
81
|
+
* bucket with one that was, and deleting it removes that key's fingerprint
|
|
82
|
+
* instead, a false negative the filter cannot detect: it stores fingerprints,
|
|
83
|
+
* not keys, so it cannot tell the two apart.
|
|
84
|
+
*
|
|
85
|
+
* @returns `true` if a matching fingerprint was removed; `false`, with the
|
|
86
|
+
* filter unchanged, if there was none.
|
|
87
|
+
*/
|
|
88
|
+
delete(key: BytesLike): boolean;
|
|
89
|
+
/**
|
|
90
|
+
* Tests whether `key` may be in the filter.
|
|
91
|
+
*
|
|
92
|
+
* @returns `false` if `key` is definitely absent; `true` if it was added
|
|
93
|
+
* (and not deleted) or on a false positive.
|
|
94
|
+
*/
|
|
95
|
+
has(key: BytesLike): boolean;
|
|
96
|
+
/**
|
|
97
|
+
* Estimates the current false-positive rate from the load: a query compares
|
|
98
|
+
* its fingerprint against about `8 * count / capacity` occupied slots across
|
|
99
|
+
* its two buckets, each matching with probability `1 / (2^f - 1)`.
|
|
100
|
+
*
|
|
101
|
+
* @returns The estimated false-positive rate, `0` for an empty filter.
|
|
102
|
+
*/
|
|
103
|
+
rate(): number;
|
|
104
|
+
/**
|
|
105
|
+
* Serializes the filter to a DSTL type 7 frame, readable by
|
|
106
|
+
* {@link CuckooFilter.fromBytes} and by any reader of the documented format.
|
|
107
|
+
*
|
|
108
|
+
* @returns The frame bytes.
|
|
109
|
+
*/
|
|
110
|
+
toBytes(): Uint8Array;
|
|
111
|
+
/**
|
|
112
|
+
* Whether `other` holds the same settings and slots, so the two serialize to
|
|
113
|
+
* identical bytes. Which slot a key lands in depends on insertion order, so
|
|
114
|
+
* filters holding the same keys can still differ.
|
|
115
|
+
*
|
|
116
|
+
* @param other - The filter to compare against.
|
|
117
|
+
* @returns `true` if the two are indistinguishable.
|
|
118
|
+
*/
|
|
119
|
+
equals(other: CuckooFilter): boolean;
|
|
120
|
+
/**
|
|
121
|
+
* Serializes the filter as a JSON envelope around its {@link CuckooFilter.toBytes} frame.
|
|
122
|
+
*
|
|
123
|
+
* @returns The JSON-safe envelope.
|
|
124
|
+
*/
|
|
125
|
+
toJSON(): FilterJSON;
|
|
126
|
+
/** Fingerprints stored, one per add not undone by a delete. */
|
|
127
|
+
get count(): number;
|
|
128
|
+
/** Slots in the table, `4 * buckets`. */
|
|
129
|
+
get capacity(): number;
|
|
130
|
+
/** Number of 4-slot buckets. */
|
|
131
|
+
get buckets(): number;
|
|
132
|
+
/** Width of each stored fingerprint, in bits. */
|
|
133
|
+
get fingerprintBits(): number;
|
|
134
|
+
/** Total slot bits, `fingerprintBits * capacity`. */
|
|
135
|
+
get m(): number;
|
|
136
|
+
/** Slot bits per expected key, `m / n`. */
|
|
137
|
+
get bitsPerKey(): number;
|
|
138
|
+
/** Hash seed. */
|
|
139
|
+
get seed(): number;
|
|
140
|
+
/** Target false-positive rate at `n` keys. */
|
|
141
|
+
get epsilon(): number;
|
|
142
|
+
}
|
|
143
|
+
//#endregion
|
|
144
|
+
//#region src/cuckoo/sizing.d.ts
|
|
145
|
+
/** Cuckoo filter geometry: what a sizing solve determines. */
|
|
146
|
+
interface CuckooSizing {
|
|
147
|
+
/** Fingerprint width in bits. */
|
|
148
|
+
f: number;
|
|
149
|
+
/** Number of 4-slot buckets. */
|
|
150
|
+
buckets: number;
|
|
151
|
+
/** Slot bits per expected key, `4 * f * buckets / n`. */
|
|
152
|
+
bitsPerKey: number;
|
|
153
|
+
}
|
|
154
|
+
/**
|
|
155
|
+
* Cuckoo geometry for `n` keys at target FPR `epsilon`: `f` bits from the
|
|
156
|
+
* `2b / epsilon` bound with 4-slot buckets, and enough buckets for 95% load
|
|
157
|
+
* plus `sqrt(n)` slack. At exactly 95%, small tables fail to take `n` keys at
|
|
158
|
+
* about 0.6% of sizes (a few buckets collect more keys than they hold, which
|
|
159
|
+
* more kicks do not fix); the slack removed every failure in about 170k
|
|
160
|
+
* prototype runs and still leaves a million-key filter at 94.6% load.
|
|
161
|
+
*/
|
|
162
|
+
export declare function cuckooSizing(n: number, epsilon: number): CuckooSizing;
|
|
163
|
+
//#endregion
|
|
164
|
+
export { BadMagicError, ChecksumError, type CuckooOptions, type CuckooParams, type CuckooSizing, type FilterJSON, ParamError, ReservedBitsError, SerializationError, TruncatedError, UnknownHashVariantError, UnknownVersionError };
|
|
@@ -0,0 +1,267 @@
|
|
|
1
|
+
import { i as hash128KeyInto, u as reduce } from "../hasher-3Cyn7Yg-.js";
|
|
2
|
+
import { a as TruncatedError, c as assertBodyLength, d as bytesEqual, f as fromJSONEnvelope, g as writeFrame, h as toJSONEnvelope, i as SerializationError, l as assertMinBodyLength, m as readHeader, n as ChecksumError, o as UnknownHashVariantError, r as ReservedBitsError, s as UnknownVersionError, t as BadMagicError, u as assertParamsPadding } from "../serialize-CMSYFym-.js";
|
|
3
|
+
import { i as assertProbability, o as assertUint32, r as assertPositiveInt, t as ParamError } from "../params-CeajSrwv.js";
|
|
4
|
+
function cuckooSizing(n, epsilon) {
|
|
5
|
+
assertProbability(epsilon, "epsilon");
|
|
6
|
+
const f = Math.max(4, Math.ceil(Math.log2(8 / epsilon)));
|
|
7
|
+
if (f > 32) throw new ParamError(`epsilon ${String(epsilon)} needs a ${String(f)}-bit fingerprint, above the maximum ${String(32)} bits`);
|
|
8
|
+
const buckets = Math.ceil(n / 3.8) + Math.ceil(Math.sqrt(n));
|
|
9
|
+
return {
|
|
10
|
+
f,
|
|
11
|
+
buckets,
|
|
12
|
+
bitsPerKey: 4 * f * buckets / n
|
|
13
|
+
};
|
|
14
|
+
}
|
|
15
|
+
//#endregion
|
|
16
|
+
//#region src/cuckoo/cuckoo.ts
|
|
17
|
+
const TYPE = 7;
|
|
18
|
+
const PARAMS_SIZE = 32;
|
|
19
|
+
const PARAMS_FIELDS_END = 28;
|
|
20
|
+
const padded8 = (length) => Math.ceil(length / 8) * 8;
|
|
21
|
+
const wordBytes = (words) => new Uint8Array(words.buffer, words.byteOffset, words.byteLength);
|
|
22
|
+
const SLOTS = 4;
|
|
23
|
+
const MAX_KICKS = 500;
|
|
24
|
+
const UNDO_SLOT = new Float64Array(MAX_KICKS);
|
|
25
|
+
const UNDO_FP = new Uint32Array(MAX_KICKS);
|
|
26
|
+
const HASH = {
|
|
27
|
+
w0: 0,
|
|
28
|
+
w1: 0,
|
|
29
|
+
w2: 0,
|
|
30
|
+
w3: 0
|
|
31
|
+
};
|
|
32
|
+
let restoring;
|
|
33
|
+
var CuckooFullError = class extends Error {
|
|
34
|
+
name = "CuckooFullError";
|
|
35
|
+
};
|
|
36
|
+
var CuckooFilter = class CuckooFilter {
|
|
37
|
+
#n;
|
|
38
|
+
#epsilon;
|
|
39
|
+
#seed;
|
|
40
|
+
#f;
|
|
41
|
+
#buckets;
|
|
42
|
+
#mask;
|
|
43
|
+
#words;
|
|
44
|
+
#count = 0;
|
|
45
|
+
#fp = 0;
|
|
46
|
+
#i1 = 0;
|
|
47
|
+
#i2 = 0;
|
|
48
|
+
static create(n, epsilon, options = {}) {
|
|
49
|
+
return new CuckooFilter({
|
|
50
|
+
...options,
|
|
51
|
+
n,
|
|
52
|
+
epsilon
|
|
53
|
+
});
|
|
54
|
+
}
|
|
55
|
+
static from(keys, epsilon, options = {}) {
|
|
56
|
+
const seen = new Set();
|
|
57
|
+
const distinct = [];
|
|
58
|
+
for (const key of keys) {
|
|
59
|
+
hash128KeyInto(key, options.seed ?? 0, HASH);
|
|
60
|
+
const id = `${String(HASH.w0)},${String(HASH.w1)},${String(HASH.w2)},${String(HASH.w3)}`;
|
|
61
|
+
if (seen.has(id)) continue;
|
|
62
|
+
seen.add(id);
|
|
63
|
+
distinct.push(key);
|
|
64
|
+
}
|
|
65
|
+
const f = CuckooFilter.create(Math.max(1, distinct.length), epsilon, options);
|
|
66
|
+
for (const k of distinct) f.add(k);
|
|
67
|
+
return f;
|
|
68
|
+
}
|
|
69
|
+
static fromBytes(bytes) {
|
|
70
|
+
const { type, flags, body } = readHeader(bytes);
|
|
71
|
+
if (type !== TYPE) throw new SerializationError(`expected DSTL type ${String(TYPE)}, got ${String(type)}`);
|
|
72
|
+
if ((flags & 15) !== 0) throw new UnknownHashVariantError(`unsupported hash variant ${String(flags & 15)}`);
|
|
73
|
+
assertMinBodyLength(body.length, PARAMS_SIZE, "cuckoo");
|
|
74
|
+
assertParamsPadding(body, PARAMS_FIELDS_END, PARAMS_SIZE, "cuckoo");
|
|
75
|
+
const view = new DataView(body.buffer, body.byteOffset, body.byteLength);
|
|
76
|
+
const params = {
|
|
77
|
+
n: view.getUint32(0, true),
|
|
78
|
+
seed: view.getUint32(4, true),
|
|
79
|
+
epsilon: view.getFloat64(8, true)
|
|
80
|
+
};
|
|
81
|
+
let filter;
|
|
82
|
+
try {
|
|
83
|
+
const f = view.getUint32(16, true);
|
|
84
|
+
const buckets = view.getUint32(20, true);
|
|
85
|
+
if (f < 4 || f > 32) throw new SerializationError(`cuckoo: stored f ${String(f)} is outside 4..${String(32)}`);
|
|
86
|
+
if (buckets < 1) throw new SerializationError("cuckoo: stored buckets is 0");
|
|
87
|
+
const m = f * SLOTS * buckets;
|
|
88
|
+
if (m > 4294967295) throw new SerializationError(`cuckoo: stored geometry gives m ${String(m)} bits, more than 2^32 - 1`);
|
|
89
|
+
const words = Math.ceil(m / 32);
|
|
90
|
+
assertBodyLength(body.length, PARAMS_SIZE + padded8(4 * words), "cuckoo");
|
|
91
|
+
restoring = {
|
|
92
|
+
f,
|
|
93
|
+
buckets
|
|
94
|
+
};
|
|
95
|
+
try {
|
|
96
|
+
filter = new CuckooFilter(params);
|
|
97
|
+
} finally {
|
|
98
|
+
restoring = void 0;
|
|
99
|
+
}
|
|
100
|
+
} catch (err) {
|
|
101
|
+
if (err instanceof ParamError) throw new SerializationError(`cuckoo: ${err.message}`);
|
|
102
|
+
throw err;
|
|
103
|
+
}
|
|
104
|
+
const words = filter.#words;
|
|
105
|
+
for (let w = 0; w < words.length; w++) words[w] = view.getUint32(PARAMS_SIZE + 4 * w, true);
|
|
106
|
+
const tail = filter.m % 32;
|
|
107
|
+
if (tail !== 0 && (words[words.length - 1] ?? 0) >>> tail !== 0) throw new SerializationError("cuckoo: bits set past the last slot");
|
|
108
|
+
if (body.subarray(PARAMS_SIZE + 4 * words.length).some((b) => b !== 0)) throw new SerializationError("cuckoo: slot padding is not zero");
|
|
109
|
+
const count = view.getUint32(24, true);
|
|
110
|
+
let occupied = 0;
|
|
111
|
+
for (let j = 0; j < filter.capacity; j++) if (filter.#slot(j) !== 0) occupied++;
|
|
112
|
+
if (count !== occupied) throw new SerializationError(`cuckoo: count ${String(count)} does not match ${String(occupied)} occupied slots`);
|
|
113
|
+
filter.#count = count;
|
|
114
|
+
return filter;
|
|
115
|
+
}
|
|
116
|
+
static fromJSON(value) {
|
|
117
|
+
return CuckooFilter.fromBytes(fromJSONEnvelope(value));
|
|
118
|
+
}
|
|
119
|
+
constructor({ n, epsilon, seed = 0 }) {
|
|
120
|
+
assertPositiveInt(n, "n");
|
|
121
|
+
assertUint32(n, "n");
|
|
122
|
+
assertProbability(epsilon, "epsilon");
|
|
123
|
+
assertUint32(seed, "seed");
|
|
124
|
+
const { f, buckets } = restoring ?? cuckooSizing(n, epsilon);
|
|
125
|
+
assertUint32(f * SLOTS * buckets, "m");
|
|
126
|
+
this.#n = n;
|
|
127
|
+
this.#epsilon = epsilon;
|
|
128
|
+
this.#seed = seed;
|
|
129
|
+
this.#f = f;
|
|
130
|
+
this.#buckets = buckets;
|
|
131
|
+
this.#mask = 2 ** f - 1;
|
|
132
|
+
this.#words = new Uint32Array(Math.ceil(f * SLOTS * buckets / 32));
|
|
133
|
+
}
|
|
134
|
+
add(key) {
|
|
135
|
+
this.#hash(key);
|
|
136
|
+
if (this.#replace(this.#i1, 0, this.#fp) || this.#replace(this.#i2, 0, this.#fp)) {
|
|
137
|
+
this.#count++;
|
|
138
|
+
return;
|
|
139
|
+
}
|
|
140
|
+
let i = (HASH.w2 & 1) === 0 ? this.#i1 : this.#i2;
|
|
141
|
+
let x = (HASH.w3 | 1) >>> 0;
|
|
142
|
+
let fp = this.#fp;
|
|
143
|
+
for (let kick = 0; kick < MAX_KICKS; kick++) {
|
|
144
|
+
x ^= x << 13;
|
|
145
|
+
x >>>= 0;
|
|
146
|
+
x ^= x >>> 17;
|
|
147
|
+
x ^= x << 5;
|
|
148
|
+
x >>>= 0;
|
|
149
|
+
const j = i * SLOTS + (x & 3);
|
|
150
|
+
const victim = this.#slot(j);
|
|
151
|
+
UNDO_SLOT[kick] = j;
|
|
152
|
+
UNDO_FP[kick] = victim;
|
|
153
|
+
this.#setSlot(j, fp);
|
|
154
|
+
fp = victim;
|
|
155
|
+
i = this.#alt(i, fp);
|
|
156
|
+
if (this.#replace(i, 0, fp)) {
|
|
157
|
+
this.#count++;
|
|
158
|
+
return;
|
|
159
|
+
}
|
|
160
|
+
}
|
|
161
|
+
for (let kick = 499; kick >= 0; kick--) this.#setSlot(UNDO_SLOT[kick] ?? 0, UNDO_FP[kick] ?? 0);
|
|
162
|
+
throw new CuckooFullError(`cuckoo filter is full at ${String(this.#count)} of ${String(this.capacity)} slots`);
|
|
163
|
+
}
|
|
164
|
+
delete(key) {
|
|
165
|
+
this.#hash(key);
|
|
166
|
+
const removed = this.#replace(this.#i1, this.#fp, 0) || this.#i2 !== this.#i1 && this.#replace(this.#i2, this.#fp, 0);
|
|
167
|
+
if (removed) this.#count--;
|
|
168
|
+
return removed;
|
|
169
|
+
}
|
|
170
|
+
has(key) {
|
|
171
|
+
this.#hash(key);
|
|
172
|
+
return this.#find(this.#i1, this.#fp) >= 0 || this.#find(this.#i2, this.#fp) >= 0;
|
|
173
|
+
}
|
|
174
|
+
rate() {
|
|
175
|
+
const occupied = 8 * this.#count / this.capacity;
|
|
176
|
+
return 1 - (1 - 1 / this.#mask) ** occupied;
|
|
177
|
+
}
|
|
178
|
+
toBytes() {
|
|
179
|
+
const words = this.#words;
|
|
180
|
+
return writeFrame({
|
|
181
|
+
version: 5,
|
|
182
|
+
type: TYPE,
|
|
183
|
+
flags: 0
|
|
184
|
+
}, PARAMS_SIZE, padded8(4 * words.length), (_, view) => {
|
|
185
|
+
view.setUint32(0, this.#n, true);
|
|
186
|
+
view.setUint32(4, this.#seed, true);
|
|
187
|
+
view.setFloat64(8, this.#epsilon, true);
|
|
188
|
+
view.setUint32(16, this.#f, true);
|
|
189
|
+
view.setUint32(20, this.#buckets, true);
|
|
190
|
+
view.setUint32(24, this.#count, true);
|
|
191
|
+
words.forEach((word, w) => {
|
|
192
|
+
view.setUint32(PARAMS_SIZE + 4 * w, word, true);
|
|
193
|
+
});
|
|
194
|
+
});
|
|
195
|
+
}
|
|
196
|
+
equals(other) {
|
|
197
|
+
return this.#n === other.#n && this.#epsilon === other.#epsilon && this.#seed === other.#seed && this.#count === other.#count && this.#f === other.#f && this.#buckets === other.#buckets && bytesEqual(wordBytes(this.#words), wordBytes(other.#words));
|
|
198
|
+
}
|
|
199
|
+
toJSON() {
|
|
200
|
+
return toJSONEnvelope(this.toBytes());
|
|
201
|
+
}
|
|
202
|
+
#hash(key) {
|
|
203
|
+
hash128KeyInto(key, this.#seed, HASH);
|
|
204
|
+
this.#fp = HASH.w1 >>> 32 - this.#f || 1;
|
|
205
|
+
this.#i1 = reduce(HASH.w0, this.#buckets);
|
|
206
|
+
this.#i2 = this.#alt(this.#i1, this.#fp);
|
|
207
|
+
}
|
|
208
|
+
#alt(i, fp) {
|
|
209
|
+
const b = this.#buckets;
|
|
210
|
+
return ((Math.imul(fp, 1540483477) >>> 0) % b + b - i) % b;
|
|
211
|
+
}
|
|
212
|
+
#find(bucket, value) {
|
|
213
|
+
for (let j = bucket * SLOTS; j < bucket * SLOTS + SLOTS; j++) if (this.#slot(j) === value) return j;
|
|
214
|
+
return -1;
|
|
215
|
+
}
|
|
216
|
+
#replace(bucket, from, to) {
|
|
217
|
+
const j = this.#find(bucket, from);
|
|
218
|
+
if (j < 0) return false;
|
|
219
|
+
this.#setSlot(j, to);
|
|
220
|
+
return true;
|
|
221
|
+
}
|
|
222
|
+
#slot(j) {
|
|
223
|
+
const bit = j * this.#f;
|
|
224
|
+
const w = bit >>> 5;
|
|
225
|
+
const off = bit & 31;
|
|
226
|
+
let v = (this.#words[w] ?? 0) >>> off;
|
|
227
|
+
if (off + this.#f > 32) v |= (this.#words[w + 1] ?? 0) << 32 - off;
|
|
228
|
+
return (v & this.#mask) >>> 0;
|
|
229
|
+
}
|
|
230
|
+
#setSlot(j, fp) {
|
|
231
|
+
const bit = j * this.#f;
|
|
232
|
+
const w = bit >>> 5;
|
|
233
|
+
const off = bit & 31;
|
|
234
|
+
const words = this.#words;
|
|
235
|
+
words[w] = (words[w] ?? 0) & ~(this.#mask << off) | fp << off;
|
|
236
|
+
if (off + this.#f > 32) {
|
|
237
|
+
const shift = 32 - off;
|
|
238
|
+
words[w + 1] = (words[w + 1] ?? 0) & ~(this.#mask >>> shift) | fp >>> shift;
|
|
239
|
+
}
|
|
240
|
+
}
|
|
241
|
+
get count() {
|
|
242
|
+
return this.#count;
|
|
243
|
+
}
|
|
244
|
+
get capacity() {
|
|
245
|
+
return SLOTS * this.#buckets;
|
|
246
|
+
}
|
|
247
|
+
get buckets() {
|
|
248
|
+
return this.#buckets;
|
|
249
|
+
}
|
|
250
|
+
get fingerprintBits() {
|
|
251
|
+
return this.#f;
|
|
252
|
+
}
|
|
253
|
+
get m() {
|
|
254
|
+
return this.#f * this.capacity;
|
|
255
|
+
}
|
|
256
|
+
get bitsPerKey() {
|
|
257
|
+
return this.m / this.#n;
|
|
258
|
+
}
|
|
259
|
+
get seed() {
|
|
260
|
+
return this.#seed;
|
|
261
|
+
}
|
|
262
|
+
get epsilon() {
|
|
263
|
+
return this.#epsilon;
|
|
264
|
+
}
|
|
265
|
+
};
|
|
266
|
+
//#endregion
|
|
267
|
+
export { BadMagicError, ChecksumError, CuckooFilter, CuckooFullError, ParamError, ReservedBitsError, SerializationError, TruncatedError, UnknownHashVariantError, UnknownVersionError, cuckooSizing };
|
package/dist/hll/index.cjs
CHANGED
|
@@ -2,7 +2,7 @@ Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
|
|
|
2
2
|
const require_hasher = require("../hasher-MhY7aQ8F.cjs");
|
|
3
3
|
const require_serialize = require("../serialize-CHikZ4GH.cjs");
|
|
4
4
|
const require_params = require("../params-UTZbJ22c.cjs");
|
|
5
|
-
const require_sizing = require("../sizing-
|
|
5
|
+
const require_sizing = require("../sizing-R1pwbzjC.cjs");
|
|
6
6
|
//#region src/hll/estimate.ts
|
|
7
7
|
const ALPHA_INF = .5 / Math.LN2;
|
|
8
8
|
function sigma(x) {
|
package/dist/hll/index.d.cts
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import { t as BytesLike } from "../bytes-DCuYtUVS.cjs";
|
|
2
2
|
import { a as ReservedBitsError, c as UnknownHashVariantError, l as UnknownVersionError, n as ChecksumError, o as SerializationError, r as FilterJSON, s as TruncatedError, t as BadMagicError } from "../serialize--xR6vGSW.cjs";
|
|
3
3
|
import { t as ParamError } from "../params-DnqJBqLS.cjs";
|
|
4
|
-
import {
|
|
4
|
+
import { o as hllSizing, r as HllSizing } from "../sizing-BmKfHf_V.cjs";
|
|
5
5
|
//#region src/hll/hll.d.ts
|
|
6
6
|
/** Construction parameters for a {@link HyperLogLog} sketch. */
|
|
7
7
|
interface HllParams {
|
package/dist/hll/index.d.ts
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import { t as BytesLike } from "../bytes-DCuYtUVS.js";
|
|
2
2
|
import { a as ReservedBitsError, c as UnknownHashVariantError, l as UnknownVersionError, n as ChecksumError, o as SerializationError, r as FilterJSON, s as TruncatedError, t as BadMagicError } from "../serialize--xR6vGSW.js";
|
|
3
3
|
import { t as ParamError } from "../params-DnqJBqLS.js";
|
|
4
|
-
import {
|
|
4
|
+
import { o as hllSizing, r as HllSizing } from "../sizing-BmKfHf_V.js";
|
|
5
5
|
//#region src/hll/hll.d.ts
|
|
6
6
|
/** Construction parameters for a {@link HyperLogLog} sketch. */
|
|
7
7
|
interface HllParams {
|
package/dist/hll/index.js
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import { i as hash128KeyInto } from "../hasher-3Cyn7Yg-.js";
|
|
2
2
|
import { a as TruncatedError, c as assertBodyLength, f as fromJSONEnvelope, g as writeFrame, h as toJSONEnvelope, i as SerializationError, l as assertMinBodyLength, m as readHeader, n as ChecksumError, o as UnknownHashVariantError, r as ReservedBitsError, s as UnknownVersionError, t as BadMagicError, u as assertParamsPadding } from "../serialize-CMSYFym-.js";
|
|
3
3
|
import { o as assertUint32, t as ParamError } from "../params-CeajSrwv.js";
|
|
4
|
-
import {
|
|
4
|
+
import { r as hllSizing } from "../sizing-D3-_7pdP.js";
|
|
5
5
|
//#region src/hll/estimate.ts
|
|
6
6
|
const ALPHA_INF = .5 / Math.LN2;
|
|
7
7
|
function sigma(x) {
|
package/dist/index.cjs
CHANGED
package/dist/index.js
CHANGED
package/dist/scalable/index.cjs
CHANGED
|
@@ -3,7 +3,7 @@ const require_bitset = require("../bitset-QYGqxnEf.cjs");
|
|
|
3
3
|
const require_hasher = require("../hasher-MhY7aQ8F.cjs");
|
|
4
4
|
const require_serialize = require("../serialize-CHikZ4GH.cjs");
|
|
5
5
|
const require_params = require("../params-UTZbJ22c.cjs");
|
|
6
|
-
const require_sizing = require("../sizing-
|
|
6
|
+
const require_sizing = require("../sizing-R1pwbzjC.cjs");
|
|
7
7
|
//#region src/scalable/scalable.ts
|
|
8
8
|
const MAX_BITS = 4294967295;
|
|
9
9
|
const TYPE = 6;
|
|
@@ -156,8 +156,11 @@ var ScalableBloomFilter = class ScalableBloomFilter {
|
|
|
156
156
|
}
|
|
157
157
|
#adopt(stages) {
|
|
158
158
|
this.#stages.length = 0;
|
|
159
|
-
|
|
160
|
-
const
|
|
159
|
+
let k = 0;
|
|
160
|
+
for (const s of stages) {
|
|
161
|
+
this.#stages.push(s);
|
|
162
|
+
if (s.k > k) k = s.k;
|
|
163
|
+
}
|
|
161
164
|
if (k > this.#probes.length) this.#probes = new Uint32Array(k);
|
|
162
165
|
return stages.reduce((_, s) => s);
|
|
163
166
|
}
|
|
@@ -201,13 +204,6 @@ var ScalableBloomFilter = class ScalableBloomFilter {
|
|
|
201
204
|
}
|
|
202
205
|
union(other) {
|
|
203
206
|
this.#assertMergeable(other);
|
|
204
|
-
const result = new ScalableBloomFilter({
|
|
205
|
-
n: this.#n,
|
|
206
|
-
epsilon: this.#epsilon,
|
|
207
|
-
growth: this.#growth,
|
|
208
|
-
tightening: this.#tightening,
|
|
209
|
-
seed: this.#seed
|
|
210
|
-
});
|
|
211
207
|
const [longer, shorter] = this.#stages.length >= other.#stages.length ? [this.#stages, other.#stages] : [other.#stages, this.#stages];
|
|
212
208
|
const stages = longer.map((stage, i) => {
|
|
213
209
|
const bits = new require_bitset.BitSet(stage.m);
|
|
@@ -226,8 +222,13 @@ var ScalableBloomFilter = class ScalableBloomFilter {
|
|
|
226
222
|
count: Math.min(stage.capacity, stage.count + (twin?.count ?? 0))
|
|
227
223
|
};
|
|
228
224
|
});
|
|
229
|
-
|
|
230
|
-
|
|
225
|
+
return ScalableBloomFilter.#withStages({
|
|
226
|
+
n: this.#n,
|
|
227
|
+
epsilon: this.#epsilon,
|
|
228
|
+
growth: this.#growth,
|
|
229
|
+
tightening: this.#tightening,
|
|
230
|
+
seed: this.#seed
|
|
231
|
+
}, stages);
|
|
231
232
|
}
|
|
232
233
|
equals(other) {
|
|
233
234
|
if (this.#settingPairs(other).some(([, a, b]) => a !== b)) return false;
|
|
@@ -252,18 +253,24 @@ var ScalableBloomFilter = class ScalableBloomFilter {
|
|
|
252
253
|
const entries = readTable(view, stageCount);
|
|
253
254
|
const expected = entries.reduce((sum, e) => sum + padded8(Math.ceil(e.m / 8)), tableEnd);
|
|
254
255
|
require_serialize.assertBodyLength(body.length, expected, "scalable");
|
|
255
|
-
|
|
256
|
+
const stages = readStages(body, entries, tableEnd);
|
|
256
257
|
try {
|
|
257
|
-
return
|
|
258
|
+
return ScalableBloomFilter.#withStages({
|
|
258
259
|
n: view.getUint32(0, true),
|
|
259
260
|
seed: view.getUint32(4, true),
|
|
260
261
|
epsilon: view.getFloat64(8, true),
|
|
261
262
|
growth: view.getFloat64(16, true),
|
|
262
263
|
tightening: view.getFloat64(24, true)
|
|
263
|
-
});
|
|
264
|
+
}, stages);
|
|
264
265
|
} catch (err) {
|
|
265
266
|
if (err instanceof require_params.ParamError) throw new require_serialize.SerializationError(`scalable: ${err.message}`);
|
|
266
267
|
throw err;
|
|
268
|
+
}
|
|
269
|
+
}
|
|
270
|
+
static #withStages(params, stages) {
|
|
271
|
+
restoring = stages;
|
|
272
|
+
try {
|
|
273
|
+
return new ScalableBloomFilter(params);
|
|
267
274
|
} finally {
|
|
268
275
|
restoring = void 0;
|
|
269
276
|
}
|
package/dist/scalable/index.js
CHANGED
|
@@ -2,7 +2,7 @@ import { t as BitSet } from "../bitset-Cf2jwRJX.js";
|
|
|
2
2
|
import { i as hash128KeyInto, l as probeLanesInto, s as probeAt } from "../hasher-3Cyn7Yg-.js";
|
|
3
3
|
import { a as TruncatedError, c as assertBodyLength, d as bytesEqual, f as fromJSONEnvelope, g as writeFrame, h as toJSONEnvelope, i as SerializationError, l as assertMinBodyLength, m as readHeader, n as ChecksumError, o as UnknownHashVariantError, r as ReservedBitsError, s as UnknownVersionError, t as BadMagicError, u as assertParamsPadding } from "../serialize-CMSYFym-.js";
|
|
4
4
|
import { i as assertProbability, o as assertUint32, r as assertPositiveInt, t as ParamError } from "../params-CeajSrwv.js";
|
|
5
|
-
import { t as bloomSizing } from "../sizing-
|
|
5
|
+
import { t as bloomSizing } from "../sizing-D3-_7pdP.js";
|
|
6
6
|
//#region src/scalable/scalable.ts
|
|
7
7
|
const MAX_BITS = 4294967295;
|
|
8
8
|
const TYPE = 6;
|
|
@@ -155,8 +155,11 @@ var ScalableBloomFilter = class ScalableBloomFilter {
|
|
|
155
155
|
}
|
|
156
156
|
#adopt(stages) {
|
|
157
157
|
this.#stages.length = 0;
|
|
158
|
-
|
|
159
|
-
const
|
|
158
|
+
let k = 0;
|
|
159
|
+
for (const s of stages) {
|
|
160
|
+
this.#stages.push(s);
|
|
161
|
+
if (s.k > k) k = s.k;
|
|
162
|
+
}
|
|
160
163
|
if (k > this.#probes.length) this.#probes = new Uint32Array(k);
|
|
161
164
|
return stages.reduce((_, s) => s);
|
|
162
165
|
}
|
|
@@ -200,13 +203,6 @@ var ScalableBloomFilter = class ScalableBloomFilter {
|
|
|
200
203
|
}
|
|
201
204
|
union(other) {
|
|
202
205
|
this.#assertMergeable(other);
|
|
203
|
-
const result = new ScalableBloomFilter({
|
|
204
|
-
n: this.#n,
|
|
205
|
-
epsilon: this.#epsilon,
|
|
206
|
-
growth: this.#growth,
|
|
207
|
-
tightening: this.#tightening,
|
|
208
|
-
seed: this.#seed
|
|
209
|
-
});
|
|
210
206
|
const [longer, shorter] = this.#stages.length >= other.#stages.length ? [this.#stages, other.#stages] : [other.#stages, this.#stages];
|
|
211
207
|
const stages = longer.map((stage, i) => {
|
|
212
208
|
const bits = new BitSet(stage.m);
|
|
@@ -225,8 +221,13 @@ var ScalableBloomFilter = class ScalableBloomFilter {
|
|
|
225
221
|
count: Math.min(stage.capacity, stage.count + (twin?.count ?? 0))
|
|
226
222
|
};
|
|
227
223
|
});
|
|
228
|
-
|
|
229
|
-
|
|
224
|
+
return ScalableBloomFilter.#withStages({
|
|
225
|
+
n: this.#n,
|
|
226
|
+
epsilon: this.#epsilon,
|
|
227
|
+
growth: this.#growth,
|
|
228
|
+
tightening: this.#tightening,
|
|
229
|
+
seed: this.#seed
|
|
230
|
+
}, stages);
|
|
230
231
|
}
|
|
231
232
|
equals(other) {
|
|
232
233
|
if (this.#settingPairs(other).some(([, a, b]) => a !== b)) return false;
|
|
@@ -251,18 +252,24 @@ var ScalableBloomFilter = class ScalableBloomFilter {
|
|
|
251
252
|
const entries = readTable(view, stageCount);
|
|
252
253
|
const expected = entries.reduce((sum, e) => sum + padded8(Math.ceil(e.m / 8)), tableEnd);
|
|
253
254
|
assertBodyLength(body.length, expected, "scalable");
|
|
254
|
-
|
|
255
|
+
const stages = readStages(body, entries, tableEnd);
|
|
255
256
|
try {
|
|
256
|
-
return
|
|
257
|
+
return ScalableBloomFilter.#withStages({
|
|
257
258
|
n: view.getUint32(0, true),
|
|
258
259
|
seed: view.getUint32(4, true),
|
|
259
260
|
epsilon: view.getFloat64(8, true),
|
|
260
261
|
growth: view.getFloat64(16, true),
|
|
261
262
|
tightening: view.getFloat64(24, true)
|
|
262
|
-
});
|
|
263
|
+
}, stages);
|
|
263
264
|
} catch (err) {
|
|
264
265
|
if (err instanceof ParamError) throw new SerializationError(`scalable: ${err.message}`);
|
|
265
266
|
throw err;
|
|
267
|
+
}
|
|
268
|
+
}
|
|
269
|
+
static #withStages(params, stages) {
|
|
270
|
+
restoring = stages;
|
|
271
|
+
try {
|
|
272
|
+
return new ScalableBloomFilter(params);
|
|
266
273
|
} finally {
|
|
267
274
|
restoring = void 0;
|
|
268
275
|
}
|
|
@@ -18,5 +18,17 @@ interface HllSizing {
|
|
|
18
18
|
* `relativeError`. Throws {@link ParamError} when even `HLL_MAX_P` cannot.
|
|
19
19
|
*/
|
|
20
20
|
declare function hllSizing(relativeError: number): HllSizing;
|
|
21
|
+
/** Count-Min geometry: the `CountMinParams` fields a sizing solve determines. */
|
|
22
|
+
interface CountMinSizing {
|
|
23
|
+
/** Counters per row. */
|
|
24
|
+
width: number;
|
|
25
|
+
/** Number of rows, one probe each. */
|
|
26
|
+
depth: number;
|
|
27
|
+
}
|
|
28
|
+
/**
|
|
29
|
+
* Count-Min sizing: `width` columns and `depth` rows for an estimate at most
|
|
30
|
+
* `epsilon * total` above the true count, with probability `1 - delta`.
|
|
31
|
+
*/
|
|
32
|
+
declare function countMinSizing(epsilon: number, delta: number): CountMinSizing;
|
|
21
33
|
//#endregion
|
|
22
|
-
export {
|
|
34
|
+
export { countMinSizing as a, bloomSizing as i, CountMinSizing as n, hllSizing as o, HllSizing as r, BloomSizing as t };
|