| 1 | import { assert, describe, it } from "@effect/vitest" |
| 2 | import { Chunk, Equal, Graph, Hash, HashMap, HashSet, Trie } from "effect" |
| 3 | |
| 4 | const range = (length: number): ReadonlyArray<number> => Array.from({ length }, (_, index) => index) |
| 5 | |
| 6 | const grid = <A>(size: number, make: (i: number, j: number) => A): ReadonlyArray<A> => { |
| 7 | const values = new Array<A>(size * size) |
| 8 | let index = 0 |
| 9 | for (let i = 0; i < size; i++) { |
| 10 | for (let j = 0; j < size; j++) { |
| 11 | values[index++] = make(i, j) |
| 12 | } |
| 13 | } |
| 14 | return values |
| 15 | } |
| 16 | |
| 17 | const unorderedPairs = <A>(size: number, make: (i: number, j: number) => A): ReadonlyArray<A> => { |
| 18 | const values = new Array<A>((size * (size - 1)) / 2) |
| 19 | let index = 0 |
| 20 | for (let i = 0; i < size; i++) { |
| 21 | for (let j = i + 1; j < size; j++) { |
| 22 | values[index++] = make(i, j) |
| 23 | } |
| 24 | } |
| 25 | return values |
| 26 | } |
| 27 | |
| 28 | const permutations = <A>(values: ReadonlyArray<A>): ReadonlyArray<ReadonlyArray<A>> => { |
| 29 | const output: Array<ReadonlyArray<A>> = [] |
| 30 | const visit = (prefix: ReadonlyArray<A>, remaining: ReadonlyArray<A>): void => { |
| 31 | if (remaining.length === 0) { |
| 32 | output.push(prefix) |
| 33 | return |
| 34 | } |
| 35 | for (let i = 0; i < remaining.length; i++) { |
| 36 | visit([...prefix, remaining[i]], [...remaining.slice(0, i), ...remaining.slice(i + 1)]) |
| 37 | } |
| 38 | } |
| 39 | visit([], values) |
| 40 | return output |
| 41 | } |
| 42 | |
| 43 | const assertNoHashCollisions = (values: Iterable<unknown>): void => { |
| 44 | const hashes = new Set<number>() |
| 45 | let count = 0 |
| 46 | for (const value of values) { |
| 47 | hashes.add(Hash.hash(value)) |
| 48 | count++ |
| 49 | } |
| 50 | assert.strictEqual(hashes.size, count) |
| 51 | } |
| 52 | |
| 53 | const twoNodeGraph = (type: Graph.Kind, first: number, second: number) => |
| 54 | Graph.make(type)<number, number>((mutable) => { |
| 55 | Graph.addNode(mutable, first) |
| 56 | Graph.addNode(mutable, second) |
| 57 | Graph.addEdge(mutable, 0, 1, 0) |
| 58 | }) |
| 59 | |
| 60 | describe("Hash", () => { |
| 61 | describe("collision resistance", () => { |
| 62 | it("does not collide for a 100 by 100 grid of arrays", () => { |
| 63 | assertNoHashCollisions(grid(100, (i, j) => [i, j])) |
| 64 | }) |
| 65 | |
| 66 | it("does not collide for a 100 by 100 grid of records", () => { |
| 67 | assertNoHashCollisions(grid(100, (i, j) => ({ i, j }))) |
| 68 | }) |
| 69 | |
| 70 | it("does not collide for a 100 by 100 grid of combined hashes", () => { |
| 71 | assertNoHashCollisions(grid(100, (i, j) => Hash.combine(Hash.hash(i), Hash.hash(j)))) |
| 72 | }) |
| 73 | |
| 74 | it("does not collide for a 100 by 100 grid of HashMap entries", () => { |
| 75 | assertNoHashCollisions(grid(100, (i, j) => HashMap.make([i, j]))) |
| 76 | }) |
| 77 | |
| 78 | it("does not collide for a 100 by 100 grid of Trie entries", () => { |
| 79 | assertNoHashCollisions(grid(100, (i, j) => Trie.make([String(i), j]))) |
| 80 | }) |
| 81 | |
| 82 | it("does not collide for a 100 by 100 grid of Chunks", () => { |
| 83 | assertNoHashCollisions(grid(100, (i, j) => Chunk.make(i, j))) |
| 84 | }) |
| 85 | |
| 86 | it("does not collide for native Sets of pairs below 150", () => { |
| 87 | assertNoHashCollisions(unorderedPairs(150, (i, j) => new Set([i, j]))) |
| 88 | }) |
| 89 | |
| 90 | it("does not collide for HashSets of pairs below 150", () => { |
| 91 | assertNoHashCollisions(unorderedPairs(150, (i, j) => HashSet.make(i, j))) |
| 92 | }) |
| 93 | |
| 94 | it("does not give distinct HashSets one constant hash", () => { |
| 95 | assert.notStrictEqual(Hash.hash(HashSet.make(1)), Hash.hash(HashSet.make(2))) |
| 96 | }) |
| 97 | |
| 98 | it("does not collide for every permutation of seven integers", () => { |
| 99 | assertNoHashCollisions(permutations(range(7))) |
| 100 | }) |
| 101 | |
| 102 | it("does not collide for nested arrays and records", () => { |
| 103 | const values = [] |
| 104 | for (let i = 0; i < 10; i++) { |
| 105 | for (let j = 0; j < 10; j++) { |
| 106 | for (let k = 0; k < 10; k++) { |
| 107 | values.push({ a: [i, j], b: { c: k } }) |
| 108 | } |
| 109 | } |
| 110 | } |
| 111 | assertNoHashCollisions(values) |
| 112 | }) |
| 113 | |
| 114 | it("does not collide for a 100 by 100 grid of directed two-node graphs", () => { |
| 115 | assertNoHashCollisions(grid(100, (i, j) => twoNodeGraph("directed", i, j))) |
| 116 | }) |
| 117 | |
| 118 | it("does not collide for a 100 by 100 grid of undirected two-node graphs", () => { |
| 119 | assertNoHashCollisions(grid(100, (i, j) => twoNodeGraph("undirected", i, j))) |
| 120 | }) |
| 121 | |
| 122 | it("keeps array order and length in the hash", () => { |
| 123 | assert.notStrictEqual(Hash.hash([1, 2]), Hash.hash([2, 1])) |
| 124 | assert.notStrictEqual(Hash.hash([1, 2, 3]), Hash.hash([])) |
| 125 | }) |
| 126 | |
| 127 | it("does not hash null, undefined, booleans and symbols like their names", () => { |
| 128 | for (const value of [null, undefined, true, false, Symbol.for("a")]) { |
| 129 | assert.notStrictEqual(Hash.hash(value), Hash.hash(String(value))) |
| 130 | } |
| 131 | }) |
| 132 | |
| 133 | it("does not collide for 10k multiples of 0.1", () => { |
| 134 | assertNoHashCollisions(range(10_000).map((i) => i * 0.1)) |
| 135 | }) |
| 136 | |
| 137 | it("does not collide for 10k multiples of 0.001", () => { |
| 138 | assertNoHashCollisions(range(10_000).map((i) => i * 0.001)) |
| 139 | }) |
| 140 | }) |
| 141 | |
| 142 | describe("floats", () => { |
| 143 | it("hashes 0 and -0 the same", () => { |
| 144 | assert.strictEqual(Hash.hash(0), Hash.hash(-0)) |
| 145 | }) |
| 146 | |
| 147 | it("hashes every NaN the same", () => { |
| 148 | const bytes = new Uint32Array(2) |
| 149 | const view = new Float64Array(bytes.buffer) |
| 150 | bytes[0] = 1 |
| 151 | bytes[1] = 0x7ff80000 |
| 152 | const quietPayload = view[0] |
| 153 | bytes[0] = 0 |
| 154 | bytes[1] = 0xfff00001 |
| 155 | const signalingNegative = view[0] |
| 156 | assert.isNaN(quietPayload) |
| 157 | assert.isNaN(signalingNegative) |
| 158 | assert.strictEqual(Hash.hash(NaN), Hash.hash(0 / 0)) |
| 159 | assert.strictEqual(Hash.hash(NaN), Hash.hash(quietPayload)) |
| 160 | assert.strictEqual(Hash.hash(NaN), Hash.hash(signalingNegative)) |
| 161 | assert.notStrictEqual(Hash.hash(Infinity), Hash.hash(-Infinity)) |
| 162 | }) |
| 163 | }) |
| 164 | |
| 165 | describe("hash law", () => { |
| 166 | let seed = 1 |
| 167 | const random = (n: number): number => { |
| 168 | seed = (Math.imul(seed, 1103515245) + 12345) >>> 0 |
| 169 | return (seed >>> 16) % n |
| 170 | } |
| 171 | const randomKey = (): unknown => [random(60), `k${random(60)}`, [random(4), random(4)]][random(3)] |
| 172 | const shuffle = <A>(values: ReadonlyArray<A>): Array<A> => { |
| 173 | const output = [...values] |
| 174 | for (let i = output.length - 1; i > 0; i--) { |
| 175 | const j = random(i + 1) |
| 176 | ;[output[i], output[j]] = [output[j], output[i]] |
| 177 | } |
| 178 | return output |
| 179 | } |
| 180 | |
| 181 | it("hashes HashMaps and HashSets independently of insertion order", () => { |
| 182 | for (let run = 0; run < 200; run++) { |
| 183 | const keys = Array.from(new Set(range(1 + random(80)).map(() => random(1000)))) |
| 184 | const entries = keys.map((key) => [key, random(3)] as const) |
| 185 | const reordered = shuffle(entries) |
| 186 | assert.strictEqual(Hash.hash(HashMap.fromIterable(entries)), Hash.hash(HashMap.fromIterable(reordered))) |
| 187 | assert.strictEqual(Hash.hash(HashSet.fromIterable(keys)), Hash.hash(HashSet.fromIterable(shuffle(keys)))) |
| 188 | } |
| 189 | }) |
| 190 | |
| 191 | it("hashes a HashMap like a fresh build after persistent and in-place edits", () => { |
| 192 | for (let run = 0; run < 100; run++) { |
| 193 | let map = HashMap.empty<unknown, number>() |
| 194 | for (let step = 0; step < 40; step++) { |
| 195 | const operation = random(3) |
| 196 | if (operation === 0) { |
| 197 | map = HashMap.mutate(map, (mutable) => { |
| 198 | for (let i = random(20); i >= 0; i--) { |
| 199 | if (random(3) === 0) HashMap.remove(mutable, randomKey()) |
| 200 | else HashMap.set(mutable, randomKey(), random(3)) |
| 201 | if (random(4) === 0) mutable[Hash.symbol]() |
| 202 | } |
| 203 | }) |
| 204 | } else if (operation === 1) { |
| 205 | map = HashMap.remove(map, randomKey()) |
| 206 | } else { |
| 207 | map = HashMap.set(map, randomKey(), random(3)) |
| 208 | } |
| 209 | const rebuilt = HashMap.fromIterable(shuffle(Array.from(map))) |
| 210 | assert.isTrue(Equal.equals(map, rebuilt)) |
| 211 | assert.strictEqual(Hash.hash(map), Hash.hash(rebuilt)) |
| 212 | } |
| 213 | } |
| 214 | }) |
| 215 | |
| 216 | it("invalidates CollisionNode hashes after in-place edits", () => { |
| 217 | const key = (id: number) => ({ |
| 218 | id, |
| 219 | [Hash.symbol](): number { |
| 220 | return 0 |
| 221 | } |
| 222 | }) |
| 223 | const first = key(1) |
| 224 | const second = key(2) |
| 225 | const third = key(3) |
| 226 | |
| 227 | HashMap.mutate(HashMap.empty<typeof first, number>(), (mutable) => { |
| 228 | HashMap.set(mutable, first, 1) |
| 229 | HashMap.set(mutable, second, 2) |
| 230 | mutable[Hash.symbol]() |
| 231 | |
| 232 | HashMap.set(mutable, third, 3) |
| 233 | assert.strictEqual(mutable[Hash.symbol](), Hash.hash(HashMap.fromIterable(mutable))) |
| 234 | |
| 235 | mutable[Hash.symbol]() |
| 236 | HashMap.set(mutable, second, 20) |
| 237 | assert.strictEqual(mutable[Hash.symbol](), Hash.hash(HashMap.fromIterable(mutable))) |
| 238 | |
| 239 | mutable[Hash.symbol]() |
| 240 | HashMap.remove(mutable, first) |
| 241 | assert.strictEqual(mutable[Hash.symbol](), Hash.hash(HashMap.fromIterable(mutable))) |
| 242 | }) |
| 243 | }) |
| 244 | }) |
| 245 | |
| 246 | describe("cyclic values", () => { |
| 247 | it("does not cache a hash computed from a circular sentinel", () => { |
| 248 | interface Node { |
| 249 | next?: Node |
| 250 | } |
| 251 | const a1: Node = {} |
| 252 | const b1: Node = {} |
| 253 | a1.next = b1 |
| 254 | b1.next = a1 |
| 255 | |
| 256 | const a2: Node = {} |
| 257 | const b2: Node = {} |
| 258 | a2.next = b2 |
| 259 | b2.next = a2 |
| 260 | |
| 261 | Hash.hash(b2) |
| 262 | assert.strictEqual(Hash.hash(a1), Hash.hash(a2)) |
| 263 | }) |
| 264 | |
| 265 | it("does not cache a HashMap subtree hash computed from a circular sentinel", () => { |
| 266 | const make = () => { |
| 267 | const holder: { map?: unknown; tag: number } = { tag: 1 } |
| 268 | const map = HashMap.make(["inner", { holder }]) |
| 269 | holder.map = map |
| 270 | return { holder, map } |
| 271 | } |
| 272 | const first = make() |
| 273 | const second = make() |
| 274 | Hash.hash(first.map) |
| 275 | assert.strictEqual(Hash.hash(first.holder), Hash.hash(second.holder)) |
| 276 | }) |
| 277 | |
| 278 | it("clears visited state when a custom hasher throws", () => { |
| 279 | const value = { |
| 280 | [Hash.symbol](): number { |
| 281 | throw new Error("boom") |
| 282 | } |
| 283 | } |
| 284 | |
| 285 | assert.throws(() => Hash.hash(value), /boom/) |
| 286 | assert.throws(() => Hash.hash(value), /boom/) |
| 287 | }) |
| 288 | }) |
| 289 | }) |