This commit is contained in:
NoryiE
2025-02-16 14:12:49 +00:00
parent c6a89e5b35
commit e0aeb9b06e
2737 changed files with 5220 additions and 1039045 deletions

File diff suppressed because it is too large Load Diff

File diff suppressed because it is too large Load Diff

View File

@@ -1,314 +0,0 @@
/* eslint-env jest */
import SearchableMap from './SearchableMap'
import * as fc from 'fast-check'
describe('SearchableMap', () => {
const strings = ['bin', 'border', 'acqua', 'aqua', 'poisson', 'parachute',
'parapendio', 'acquamarina', 'summertime', 'summer', 'join', 'mediterraneo',
'perciò', 'borderline', 'bo']
const keyValues = strings.map((key, i) => [key, i])
const object = keyValues.reduce((obj, [key, value]) => ({ ...obj, [key]: value }))
const editDistance = function (a, b, mem = [[0]]) {
mem[a.length] = mem[a.length] || [a.length]
if (mem[a.length][b.length] !== undefined) { return mem[a.length][b.length] }
const d = (a[a.length - 1] === b[b.length - 1]) ? 0 : 1
const distance = (a.length === 1 && b.length === 1)
? d
: Math.min(
((a.length > 0) ? editDistance(a.slice(0, -1), b, mem) + 1 : Infinity),
((b.length > 0) ? editDistance(a, b.slice(0, -1), mem) + 1 : Infinity),
((a.length > 0 && b.length > 0) ? editDistance(a.slice(0, -1), b.slice(0, -1), mem) + d : Infinity)
)
mem[a.length][b.length] = distance
return distance
}
describe('clear', () => {
it('empties the map', () => {
const map = SearchableMap.from(keyValues)
map.clear()
expect(Array.from(map.entries())).toEqual([])
})
})
describe('delete', () => {
it('deletes the entry at the given key', () => {
const map = SearchableMap.from(keyValues)
map.delete('border')
expect(map.has('border')).toBe(false)
expect(map.has('summer')).toBe(true)
expect(map.has('borderline')).toBe(true)
expect(map.has('bo')).toBe(true)
})
it('changes the size of the map', () => {
const map = SearchableMap.from(keyValues)
const sizeBefore = map.size
map.delete('summertime')
expect(map.size).toEqual(sizeBefore - 1)
})
it('does nothing if the entry did not exist', () => {
const map = new SearchableMap()
expect(() => map.delete('something')).not.toThrow()
})
it('leaves the radix tree in the same state as before the entry was added', () => {
const map = new SearchableMap()
map.set('hello', 1)
const before = new SearchableMap(new Map(map._tree))
map.set('help', 2)
map.delete('help')
expect(map).toEqual(before)
})
})
describe('entries', () => {
it('returns an iterator of entries', () => {
const map = SearchableMap.from(keyValues)
const entries = Array.from({ [Symbol.iterator]: () => map.entries() })
expect(entries.sort()).toEqual(keyValues.sort())
})
it('returns an iterable of entries', () => {
const map = SearchableMap.from(keyValues)
const entries = Array.from(map.entries())
expect(entries.sort()).toEqual(keyValues.sort())
})
it('returns empty iterator, if the map is empty', () => {
const map = new SearchableMap()
const entries = Array.from(map.entries())
expect(entries).toEqual([])
})
})
describe('forEach', () => {
it('iterates through each entry', () => {
const entries = []
const fn = (key, value) => entries.push([key, value])
const map = SearchableMap.from(keyValues)
map.forEach(fn)
expect(entries).toEqual(Array.from(map.entries()))
})
})
describe('get', () => {
it('gets the value at key', () => {
const key = 'foo'
const value = 42
const map = SearchableMap.fromObject({ [key]: value })
expect(map.get(key)).toBe(value)
})
it('returns undefined if the key is not present', () => {
const map = new SearchableMap()
expect(map.get('not-existent')).toBe(undefined)
})
})
describe('has', () => {
it('returns true if the given key exists in the map', () => {
const map = new SearchableMap()
map.set('something', 42)
expect(map.has('something')).toBe(true)
map.set('something else', null)
expect(map.has('something else')).toBe(true)
})
it('returns false if the given key does not exist in the map', () => {
const map = SearchableMap.fromObject({ something: 42 })
expect(map.has('not-existing')).toBe(false)
expect(map.has('some')).toBe(false)
})
})
describe('keys', () => {
it('returns an iterator of keys', () => {
const map = SearchableMap.from(keyValues)
const keys = Array.from({ [Symbol.iterator]: () => map.keys() })
expect(keys.sort()).toEqual(strings.sort())
})
it('returns an iterable of keys', () => {
const map = SearchableMap.from(keyValues)
const keys = Array.from(map.keys())
expect(keys.sort()).toEqual(strings.sort())
})
it('returns empty iterator, if the map is empty', () => {
const map = new SearchableMap()
const keys = Array.from(map.keys())
expect(keys).toEqual([])
})
})
describe('set', () => {
it('sets a value at key', () => {
const map = new SearchableMap()
const key = 'foo'
const value = 42
map.set(key, value)
expect(map.get(key)).toBe(value)
})
it('overrides a value at key if it already exists', () => {
const map = SearchableMap.fromObject({ foo: 123 })
const key = 'foo'
const value = 42
map.set(key, value)
expect(map.get(key)).toBe(value)
})
it('throws error if the given key is not a string', () => {
const map = new SearchableMap()
expect(() => map.set(123, 'foo')).toThrow('key must be a string')
})
})
describe('size', () => {
it('is a property containing the size of the map', () => {
const map = SearchableMap.from(keyValues)
expect(map.size).toEqual(keyValues.length)
map.set('foo', 42)
expect(map.size).toEqual(keyValues.length + 1)
map.delete('border')
expect(map.size).toEqual(keyValues.length)
map.clear()
expect(map.size).toEqual(0)
})
})
describe('update', () => {
it('sets a value at key applying a function to the previous value', () => {
const map = new SearchableMap()
const key = 'foo'
const fn = jest.fn(x => (x || 0) + 1)
map.update(key, fn)
expect(fn).toHaveBeenCalledWith(undefined)
expect(map.get(key)).toBe(1)
map.update(key, fn)
expect(fn).toHaveBeenCalledWith(1)
expect(map.get(key)).toBe(2)
})
it('throws error if the given key is not a string', () => {
const map = new SearchableMap()
expect(() => map.update(123, () => {})).toThrow('key must be a string')
})
})
describe('values', () => {
it('returns an iterator of values', () => {
const map = SearchableMap.fromObject(object)
const values = Array.from({ [Symbol.iterator]: () => map.values() })
expect(values.sort()).toEqual(Object.values(object).sort())
})
it('returns an iterable of values', () => {
const map = SearchableMap.fromObject(object)
const values = Array.from(map.values())
expect(values.sort()).toEqual(Object.values(object).sort())
})
it('returns empty iterator, if the map is empty', () => {
const map = new SearchableMap()
const values = Array.from(map.values())
expect(values).toEqual([])
})
})
describe('atPrefix', () => {
it('returns the submap at the given prefix', () => {
const map = SearchableMap.from(keyValues)
const sum = map.atPrefix('sum')
expect(Array.from(sum.keys()).sort()).toEqual(strings.filter(string => string.startsWith('sum')).sort())
const summer = sum.atPrefix('summer')
expect(Array.from(summer.keys()).sort()).toEqual(strings.filter(string => string.startsWith('summer')).sort())
const xyz = map.atPrefix('xyz')
expect(Array.from(xyz.keys())).toEqual([])
expect(() => sum.atPrefix('xyz')).toThrow()
})
it('correctly computes the size', () => {
const map = SearchableMap.from(keyValues)
const sum = map.atPrefix('sum')
expect(sum.size).toEqual(strings.filter(string => string.startsWith('sum')).length)
})
})
describe('fuzzyGet', () => {
const terms = ['summer', 'acqua', 'aqua', 'acquire', 'poisson', 'qua']
const keyValues = terms.map((key, i) => [key, i])
const map = SearchableMap.from(keyValues)
it('returns all entries having the given maximum edit distance from the given key', () => {
[0, 1, 2, 3].forEach(distance => {
const results = map.fuzzyGet('acqua', distance)
const entries = Array.from(results)
expect(entries.map(([key, [value, dist]]) => [key, dist]).sort())
.toEqual(terms.map(term => [term, editDistance('acqua', term)]).filter(([, d]) => d <= distance).sort())
expect(entries.every(([key, [value]]) => map.get(key) === value)).toBe(true)
})
})
it('returns an empty object if no matching entries are found', () => {
expect(map.fuzzyGet('winter', 1)).toEqual(new Map())
})
it('returns entries if edit distance is longer than key', () => {
const map = SearchableMap.from([['x', 1], [' x', 2]])
expect(Array.from(map.fuzzyGet('x', 2).values())).toEqual([[1, 0], [2, 1]])
})
})
describe('with generated test data', () => {
it('adds and removes entries', () => {
const arrayOfStrings = fc.array(fc.oneof(fc.unicodeString(), fc.string()), { maxLength: 70 })
const string = fc.oneof(fc.unicodeString({ minLength: 0, maxLength: 4 }), fc.string({ minLength: 0, maxLength: 4 }))
const int = fc.integer({ min: 1, max: 4 })
fc.assert(fc.property(arrayOfStrings, string, int, (terms, prefix, maxDist) => {
const map = new SearchableMap()
const standardMap = new Map()
const uniqueTerms = [...new Set(terms)]
terms.forEach((term, i) => {
map.set(term, i)
standardMap.set(term, i)
expect(map.has(term)).toBe(true)
expect(standardMap.get(term)).toEqual(i)
})
expect(map.size).toEqual(standardMap.size)
expect(Array.from(map.entries()).sort()).toEqual(Array.from(standardMap.entries()).sort())
expect(Array.from(map.atPrefix(prefix).keys()).sort())
.toEqual(Array.from(new Set(terms)).filter(t => t.startsWith(prefix)).sort())
const fuzzy = map.fuzzyGet(terms[0], maxDist)
expect(Array.from(fuzzy, ([key, [value, dist]]) => [key, dist]).sort())
.toEqual(uniqueTerms.map(term => [term, editDistance(terms[0], term)])
.filter(([, dist]) => dist <= maxDist).sort())
terms.forEach(term => {
map.delete(term)
expect(map.has(term)).toBe(false)
expect(map.get(term)).toEqual(undefined)
})
expect(map.size).toEqual(0)
}))
})
})
})

View File

@@ -1,422 +0,0 @@
/* eslint-disable no-labels */
import { TreeIterator, ENTRIES, KEYS, VALUES, LEAF } from './TreeIterator'
import fuzzySearch, { FuzzyResults } from './fuzzySearch'
import { RadixTree, Entry, Path } from './types'
/**
* A class implementing the same interface as a standard JavaScript
* [`Map`](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map)
* with string keys, but adding support for efficiently searching entries with
* prefix or fuzzy search. This class is used internally by [[MiniSearch]] as
* the inverted index data structure. The implementation is a radix tree
* (compressed prefix tree).
*
* Since this class can be of general utility beyond _MiniSearch_, it is
* exported by the `minisearch` package and can be imported (or required) as
* `minisearch/SearchableMap`.
*
* @typeParam T The type of the values stored in the map.
*/
export default class SearchableMap<T = any> {
/**
* @internal
*/
_tree: RadixTree<T>
/**
* @internal
*/
_prefix: string
private _size: number | undefined = undefined
/**
* The constructor is normally called without arguments, creating an empty
* map. In order to create a [[SearchableMap]] from an iterable or from an
* object, check [[SearchableMap.from]] and [[SearchableMap.fromObject]].
*
* The constructor arguments are for internal use, when creating derived
* mutable views of a map at a prefix.
*/
constructor (tree: RadixTree<T> = new Map(), prefix = '') {
this._tree = tree
this._prefix = prefix
}
/**
* Creates and returns a mutable view of this [[SearchableMap]], containing only
* entries that share the given prefix.
*
* ### Usage:
*
* ```javascript
* let map = new SearchableMap()
* map.set("unicorn", 1)
* map.set("universe", 2)
* map.set("university", 3)
* map.set("unique", 4)
* map.set("hello", 5)
*
* let uni = map.atPrefix("uni")
* uni.get("unique") // => 4
* uni.get("unicorn") // => 1
* uni.get("hello") // => undefined
*
* let univer = map.atPrefix("univer")
* univer.get("unique") // => undefined
* univer.get("universe") // => 2
* univer.get("university") // => 3
* ```
*
* @param prefix The prefix
* @return A [[SearchableMap]] representing a mutable view of the original Map at the given prefix
*/
atPrefix (prefix: string): SearchableMap<T> {
if (!prefix.startsWith(this._prefix)) { throw new Error('Mismatched prefix') }
const [node, path] = trackDown(this._tree, prefix.slice(this._prefix.length))
if (node === undefined) {
const [parentNode, key] = last(path)
for (const k of parentNode!.keys()) {
if (k !== LEAF && k.startsWith(key)) {
const node = new Map()
node.set(k.slice(key.length), parentNode!.get(k)!)
return new SearchableMap(node, prefix)
}
}
}
return new SearchableMap<T>(node, prefix)
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/clear
*/
clear (): void {
this._size = undefined
this._tree.clear()
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/delete
* @param key Key to delete
*/
delete (key: string): void {
this._size = undefined
return remove(this._tree, key)
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/entries
* @return An iterator iterating through `[key, value]` entries.
*/
entries () {
return new TreeIterator(this, ENTRIES)
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/forEach
* @param fn Iteration function
*/
forEach (fn: (key: string, value: T, map: SearchableMap) => void): void {
for (const [key, value] of this) {
fn(key, value, this)
}
}
/**
* Returns a Map of all the entries that have a key within the given edit
* distance from the search key. The keys of the returned Map are the matching
* keys, while the values are two-element arrays where the first element is
* the value associated to the key, and the second is the edit distance of the
* key to the search key.
*
* ### Usage:
*
* ```javascript
* let map = new SearchableMap()
* map.set('hello', 'world')
* map.set('hell', 'yeah')
* map.set('ciao', 'mondo')
*
* // Get all entries that match the key 'hallo' with a maximum edit distance of 2
* map.fuzzyGet('hallo', 2)
* // => Map(2) { 'hello' => ['world', 1], 'hell' => ['yeah', 2] }
*
* // In the example, the "hello" key has value "world" and edit distance of 1
* // (change "e" to "a"), the key "hell" has value "yeah" and edit distance of 2
* // (change "e" to "a", delete "o")
* ```
*
* @param key The search key
* @param maxEditDistance The maximum edit distance (Levenshtein)
* @return A Map of the matching keys to their value and edit distance
*/
fuzzyGet (key: string, maxEditDistance: number): FuzzyResults<T> {
return fuzzySearch<T>(this._tree, key, maxEditDistance)
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/get
* @param key Key to get
* @return Value associated to the key, or `undefined` if the key is not
* found.
*/
get (key: string): T | undefined {
const node = lookup<T>(this._tree, key)
return node !== undefined ? node.get(LEAF) : undefined
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/has
* @param key Key
* @return True if the key is in the map, false otherwise
*/
has (key: string): boolean {
const node = lookup(this._tree, key)
return node !== undefined && node.has(LEAF)
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/keys
* @return An `Iterable` iterating through keys
*/
keys () {
return new TreeIterator(this, KEYS)
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/set
* @param key Key to set
* @param value Value to associate to the key
* @return The [[SearchableMap]] itself, to allow chaining
*/
set (key: string, value: T): SearchableMap<T> {
if (typeof key !== 'string') { throw new Error('key must be a string') }
this._size = undefined
const node = createPath(this._tree, key)
node.set(LEAF, value)
return this
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/size
*/
get size (): number {
if (this._size) { return this._size }
/** @ignore */
this._size = 0
const iter = this.entries()
while (!iter.next().done) this._size! += 1
return this._size
}
/**
* Updates the value at the given key using the provided function. The function
* is called with the current value at the key, and its return value is used as
* the new value to be set.
*
* ### Example:
*
* ```javascript
* // Increment the current value by one
* searchableMap.update('somekey', (currentValue) => currentValue == null ? 0 : currentValue + 1)
* ```
*
* If the value at the given key is or will be an object, it might not require
* re-assignment. In that case it is better to use `fetch()`, because it is
* faster.
*
* @param key The key to update
* @param fn The function used to compute the new value from the current one
* @return The [[SearchableMap]] itself, to allow chaining
*/
update (key: string, fn: (value: T | undefined) => T): SearchableMap<T> {
if (typeof key !== 'string') { throw new Error('key must be a string') }
this._size = undefined
const node = createPath(this._tree, key)
node.set(LEAF, fn(node.get(LEAF)))
return this
}
/**
* Fetches the value of the given key. If the value does not exist, calls the
* given function to create a new value, which is inserted at the given key
* and subsequently returned.
*
* ### Example:
*
* ```javascript
* const map = searchableMap.fetch('somekey', () => new Map())
* map.set('foo', 'bar')
* ```
*
* @param key The key to update
* @param defaultValue A function that creates a new value if the key does not exist
* @return The existing or new value at the given key
*/
fetch (key: string, initial: () => T): T {
if (typeof key !== 'string') { throw new Error('key must be a string') }
this._size = undefined
const node = createPath(this._tree, key)
let value = node.get(LEAF)
if (value === undefined) {
node.set(LEAF, value = initial())
}
return value
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/values
* @return An `Iterable` iterating through values.
*/
values () {
return new TreeIterator(this, VALUES)
}
/**
* @see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Map/@@iterator
*/
[Symbol.iterator] () {
return this.entries()
}
/**
* Creates a [[SearchableMap]] from an `Iterable` of entries
*
* @param entries Entries to be inserted in the [[SearchableMap]]
* @return A new [[SearchableMap]] with the given entries
*/
static from<T = any> (entries: Iterable<Entry<T>> | Entry<T>[]) {
const tree = new SearchableMap()
for (const [key, value] of entries) {
tree.set(key, value)
}
return tree
}
/**
* Creates a [[SearchableMap]] from the iterable properties of a JavaScript object
*
* @param object Object of entries for the [[SearchableMap]]
* @return A new [[SearchableMap]] with the given entries
*/
static fromObject<T = any> (object: { [key: string]: T }) {
return SearchableMap.from<T>(Object.entries(object))
}
}
const trackDown = <T = any>(tree: RadixTree<T> | undefined, key: string, path: Path<T> = []): [RadixTree<T> | undefined, Path<T>] => {
if (key.length === 0 || tree == null) { return [tree, path] }
for (const k of tree.keys()) {
if (k !== LEAF && key.startsWith(k)) {
path.push([tree, k]) // performance: update in place
return trackDown(tree.get(k)!, key.slice(k.length), path)
}
}
path.push([tree, key]) // performance: update in place
return trackDown(undefined, '', path)
}
const lookup = <T = any>(tree: RadixTree<T>, key: string): RadixTree<T> | undefined => {
if (key.length === 0 || tree == null) { return tree }
for (const k of tree.keys()) {
if (k !== LEAF && key.startsWith(k)) {
return lookup(tree.get(k)!, key.slice(k.length))
}
}
}
// Create a path in the radix tree for the given key, and returns the deepest
// node. This function is in the hot path for indexing. It avoids unnecessary
// string operations and recursion for performance.
const createPath = <T = any>(node: RadixTree<T>, key: string): RadixTree<T> => {
const keyLength = key.length
outer: for (let pos = 0; node && pos < keyLength;) {
for (const k of node.keys()) {
// Check whether this key is a candidate: the first characters must match.
if (k !== LEAF && key[pos] === k[0]) {
const len = Math.min(keyLength - pos, k.length)
// Advance offset to the point where key and k no longer match.
let offset = 1
while (offset < len && key[pos + offset] === k[offset]) ++offset
const child = node.get(k)!
if (offset === k.length) {
// The existing key is shorter than the key we need to create.
node = child
} else {
// Partial match: we need to insert an intermediate node to contain
// both the existing subtree and the new node.
const intermediate = new Map()
intermediate.set(k.slice(offset), child)
node.set(key.slice(pos, pos + offset), intermediate)
node.delete(k)
node = intermediate
}
pos += offset
continue outer
}
}
// Create a final child node to contain the final suffix of the key.
const child = new Map()
node.set(key.slice(pos), child)
return child
}
return node
}
const remove = <T = any>(tree: RadixTree<T>, key: string): void => {
const [node, path] = trackDown(tree, key)
if (node === undefined) { return }
node.delete(LEAF)
if (node.size === 0) {
cleanup(path)
} else if (node.size === 1) {
const [key, value] = node.entries().next().value
merge(path, key, value)
}
}
const cleanup = <T = any>(path: Path<T>): void => {
if (path.length === 0) { return }
const [node, key] = last(path)
node!.delete(key)
if (node!.size === 0) {
cleanup(path.slice(0, -1))
} else if (node!.size === 1) {
const [key, value] = node!.entries().next().value
if (key !== LEAF) {
merge(path.slice(0, -1), key, value)
}
}
}
const merge = <T = any>(path: Path<T>, key: string, value: RadixTree<T>): void => {
if (path.length === 0) { return }
const [node, nodeKey] = last(path)
node!.set(nodeKey + key, value)
node!.delete(nodeKey)
}
const last = <T = any>(array: T[]): T => {
return array[array.length - 1]
}

View File

@@ -1,103 +0,0 @@
import { RadixTree, Entry, LeafType } from './types'
/** @ignore */
const ENTRIES = 'ENTRIES'
/** @ignore */
const KEYS = 'KEYS'
/** @ignore */
const VALUES = 'VALUES'
/** @ignore */
const LEAF = '' as LeafType
interface Iterators<T> {
ENTRIES: Entry<T>
KEYS: string
VALUES: T
}
type Kind<T> = keyof Iterators<T>
type Result<T, K extends keyof Iterators<T>> = Iterators<T>[K]
type IteratorPath<T> = {
node: RadixTree<T>,
keys: string[]
}[]
export type IterableSet<T> = {
_tree: RadixTree<T>,
_prefix: string
}
/**
* @private
*/
class TreeIterator<T, K extends Kind<T>> implements Iterator<Result<T, K>> {
set: IterableSet<T>
_type: K
_path: IteratorPath<T>
constructor (set: IterableSet<T>, type: K) {
const node = set._tree
const keys = Array.from(node.keys())
this.set = set
this._type = type
this._path = keys.length > 0 ? [{ node, keys }] : []
}
next (): IteratorResult<Result<T, K>> {
const value = this.dive()
this.backtrack()
return value
}
dive (): IteratorResult<Result<T, K>> {
if (this._path.length === 0) { return { done: true, value: undefined } }
const { node, keys } = last(this._path)!
if (last(keys) === LEAF) { return { done: false, value: this.result() } }
const child = node.get(last(keys)!)!
this._path.push({ node: child, keys: Array.from(child.keys()) })
return this.dive()
}
backtrack (): void {
if (this._path.length === 0) { return }
const keys = last(this._path)!.keys
keys.pop()
if (keys.length > 0) { return }
this._path.pop()
this.backtrack()
}
key (): string {
return this.set._prefix + this._path
.map(({ keys }) => last(keys))
.filter(key => key !== LEAF)
.join('')
}
value (): T {
return last(this._path)!.node.get(LEAF)!
}
result (): Result<T, K> {
switch (this._type) {
case VALUES: return this.value() as Result<T, K>
case KEYS: return this.key() as Result<T, K>
default: return [this.key(), this.value()] as Result<T, K>
}
}
[Symbol.iterator] () {
return this
}
}
const last = <T>(array: T[]): T | undefined => {
return array[array.length - 1]
}
export { TreeIterator, ENTRIES, KEYS, VALUES, LEAF }

View File

@@ -1,130 +0,0 @@
/* eslint-disable no-labels */
import { LEAF } from './TreeIterator'
import { RadixTree } from './types'
export type FuzzyResult<T> = [T, number]
export type FuzzyResults<T> = Map<string, FuzzyResult<T>>
/**
* @ignore
*/
export const fuzzySearch = <T = any>(node: RadixTree<T>, query: string, maxDistance: number): FuzzyResults<T> => {
const results: FuzzyResults<T> = new Map()
if (query === undefined) return results
// Number of columns in the Levenshtein matrix.
const n = query.length + 1
// Matching terms can never be longer than N + maxDistance.
const m = n + maxDistance
// Fill first matrix row and column with numbers: 0 1 2 3 ...
const matrix = new Uint8Array(m * n).fill(maxDistance + 1)
for (let j = 0; j < n; ++j) matrix[j] = j
for (let i = 1; i < m; ++i) matrix[i * n] = i
recurse(
node,
query,
maxDistance,
results,
matrix,
1,
n,
''
)
return results
}
// Modified version of http://stevehanov.ca/blog/?id=114
// This builds a Levenshtein matrix for a given query and continuously updates
// it for nodes in the radix tree that fall within the given maximum edit
// distance. Keeping the same matrix around is beneficial especially for larger
// edit distances.
//
// k a t e <-- query
// 0 1 2 3 4
// c 1 1 2 3 4
// a 2 2 1 2 3
// t 3 3 2 1 [2] <-- edit distance
// ^
// ^ term in radix tree, rows are added and removed as needed
const recurse = <T = any>(
node: RadixTree<T>,
query: string,
maxDistance: number,
results: FuzzyResults<T>,
matrix: Uint8Array,
m: number,
n: number,
prefix: string
): void => {
const offset = m * n
key: for (const key of node.keys()) {
if (key === LEAF) {
// We've reached a leaf node. Check if the edit distance acceptable and
// store the result if it is.
const distance = matrix[offset - 1]
if (distance <= maxDistance) {
results.set(prefix, [node.get(key)!, distance])
}
} else {
// Iterate over all characters in the key. Update the Levenshtein matrix
// and check if the minimum distance in the last row is still within the
// maximum edit distance. If it is, we can recurse over all child nodes.
let i = m
for (let pos = 0; pos < key.length; ++pos, ++i) {
const char = key[pos]
const thisRowOffset = n * i
const prevRowOffset = thisRowOffset - n
// Set the first column based on the previous row, and initialize the
// minimum distance in the current row.
let minDistance = matrix[thisRowOffset]
const jmin = Math.max(0, i - maxDistance - 1)
const jmax = Math.min(n - 1, i + maxDistance)
// Iterate over remaining columns (characters in the query).
for (let j = jmin; j < jmax; ++j) {
const different = char !== query[j]
// It might make sense to only read the matrix positions used for
// deletion/insertion if the characters are different. But we want to
// avoid conditional reads for performance reasons.
const rpl = matrix[prevRowOffset + j] + +different
const del = matrix[prevRowOffset + j + 1] + 1
const ins = matrix[thisRowOffset + j] + 1
const dist = matrix[thisRowOffset + j + 1] = Math.min(rpl, del, ins)
if (dist < minDistance) minDistance = dist
}
// Because distance will never decrease, we can stop. There will be no
// matching child nodes.
if (minDistance > maxDistance) {
continue key
}
}
recurse(
node.get(key)!,
query,
maxDistance,
results,
matrix,
i,
n,
prefix + key
)
}
}
}
export default fuzzySearch

View File

@@ -1,18 +0,0 @@
export type LeafType = '' & { readonly __tag: unique symbol }
export interface RadixTree<T> extends Map<string, T | RadixTree<T>> {
// Distinguish between an empty string indicating a leaf node and a non-empty
// string indicating a subtree. Overriding these types avoids a lot of type
// assertions elsewhere in the code. It is not 100% foolproof because you can
// still pass in a blank string '' disguised as `string` and potentially get a
// leaf value.
get(key: LeafType): T | undefined
get(key: string): RadixTree<T> | undefined
set(key: LeafType, value: T): this
set(key: string, value: RadixTree<T>): this
}
export type Entry<T> = [string, T]
export type Path<T> = [RadixTree<T> | undefined, string][]

View File

@@ -1,4 +0,0 @@
import MiniSearch from './MiniSearch'
export * from './MiniSearch'
export default MiniSearch

View File

@@ -1 +0,0 @@
/* eslint-env jest */