@tanstack/db-ivm 0.1.11 → 0.1.12
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/dist/cjs/operators/topKWithFractionalIndex.cjs +8 -2
- package/dist/cjs/operators/topKWithFractionalIndex.cjs.map +1 -1
- package/dist/esm/operators/topKWithFractionalIndex.js +8 -2
- package/dist/esm/operators/topKWithFractionalIndex.js.map +1 -1
- package/package.json +1 -1
- package/src/operators/topKWithFractionalIndex.ts +17 -4
|
@@ -30,10 +30,16 @@ class TopKArray {
|
|
|
30
30
|
}) {
|
|
31
31
|
const oldOffset = this.#topKStart;
|
|
32
32
|
const oldLimit = this.#topKEnd - this.#topKStart;
|
|
33
|
-
const oldRange = [
|
|
33
|
+
const oldRange = [
|
|
34
|
+
this.#topKStart,
|
|
35
|
+
this.#topKEnd === Infinity ? this.#topKStart + this.size : this.#topKEnd
|
|
36
|
+
];
|
|
34
37
|
this.#topKStart = offset ?? oldOffset;
|
|
35
38
|
this.#topKEnd = this.#topKStart + (limit ?? oldLimit);
|
|
36
|
-
const newRange = [
|
|
39
|
+
const newRange = [
|
|
40
|
+
this.#topKStart,
|
|
41
|
+
this.#topKEnd === Infinity ? Math.max(this.#topKStart + this.size, oldRange[1]) : this.#topKEnd
|
|
42
|
+
];
|
|
37
43
|
const { onlyInA, onlyInB } = utils.diffHalfOpen(oldRange, newRange);
|
|
38
44
|
const moveIns = [];
|
|
39
45
|
onlyInB.forEach((index) => {
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"topKWithFractionalIndex.cjs","sources":["../../../src/operators/topKWithFractionalIndex.ts"],"sourcesContent":["import { generateKeyBetween } from \"fractional-indexing\"\nimport { DifferenceStreamWriter, UnaryOperator } from \"../graph.js\"\nimport { StreamBuilder } from \"../d2.js\"\nimport { MultiSet } from \"../multiset.js\"\nimport {\n binarySearch,\n diffHalfOpen,\n globalObjectIdGenerator,\n} from \"../utils.js\"\nimport type { HRange } from \"../utils.js\"\nimport type { DifferenceStreamReader } from \"../graph.js\"\nimport type { IStreamBuilder, PipedOperator } from \"../types.js\"\n\nexport interface TopKWithFractionalIndexOptions {\n limit?: number\n offset?: number\n setSizeCallback?: (getSize: () => number) => void\n setWindowFn?: (\n windowFn: (options: { offset?: number; limit?: number }) => void\n ) => void\n}\n\nexport type TopKChanges<V> = {\n /** Indicates which element moves into the topK (if any) */\n moveIn: IndexedValue<V> | null\n /** Indicates which element moves out of the topK (if any) */\n moveOut: IndexedValue<V> | null\n}\n\nexport type TopKMoveChanges<V> = {\n /** Flag that marks whether there were any changes to the topK */\n changes: boolean\n /** Indicates which elements move into the topK (if any) */\n moveIns: Array<IndexedValue<V>>\n /** Indicates which elements move out of the topK (if any) */\n moveOuts: Array<IndexedValue<V>>\n}\n\n/**\n * A topK data structure that supports insertions and deletions\n * and returns changes to the topK.\n */\nexport interface TopK<V> {\n size: number\n insert: (value: V) => TopKChanges<V>\n delete: (value: V) => TopKChanges<V>\n}\n\n/**\n * Implementation of a topK data structure.\n * Uses a sorted array internally to store the values and keeps a topK window over that array.\n * Inserts and deletes are O(n) operations because worst case an element is inserted/deleted\n * at the start of the array which causes all the elements to shift to the right/left.\n */\nclass TopKArray<V> implements TopK<V> {\n #sortedValues: Array<IndexedValue<V>> = []\n #comparator: (a: V, b: V) => number\n #topKStart: number\n #topKEnd: number\n\n constructor(\n offset: number,\n limit: number,\n comparator: (a: V, b: V) => number\n ) {\n this.#topKStart = offset\n this.#topKEnd = offset + limit\n this.#comparator = comparator\n }\n\n get size(): number {\n const offset = this.#topKStart\n const limit = this.#topKEnd - this.#topKStart\n const available = this.#sortedValues.length - offset\n return Math.max(0, Math.min(limit, available))\n }\n\n /**\n * Moves the topK window\n */\n move({\n offset,\n limit,\n }: {\n offset?: number\n limit?: number\n }): TopKMoveChanges<V> {\n const oldOffset = this.#topKStart\n const oldLimit = this.#topKEnd - this.#topKStart\n const oldRange: HRange = [this.#topKStart, this.#topKEnd]\n\n this.#topKStart = offset ?? oldOffset\n this.#topKEnd = this.#topKStart + (limit ?? oldLimit)\n\n const newRange: HRange = [this.#topKStart, this.#topKEnd]\n const { onlyInA, onlyInB } = diffHalfOpen(oldRange, newRange)\n\n const moveIns: Array<IndexedValue<V>> = []\n onlyInB.forEach((index) => {\n const value = this.#sortedValues[index]\n if (value) {\n moveIns.push(value)\n }\n })\n\n const moveOuts: Array<IndexedValue<V>> = []\n onlyInA.forEach((index) => {\n const value = this.#sortedValues[index]\n if (value) {\n moveOuts.push(value)\n }\n })\n\n // It could be that there are changes (i.e. moveIns or moveOuts)\n // but that the collection is lazy so we don't have the data yet that needs to move in/out\n // so `moveIns` and `moveOuts` will be empty but `changes` will be true\n // this will tell the caller that it needs to run the graph to load more data\n return { moveIns, moveOuts, changes: onlyInA.length + onlyInB.length > 0 }\n }\n\n insert(value: V): TopKChanges<V> {\n const result: TopKChanges<V> = { moveIn: null, moveOut: null }\n\n // Lookup insert position\n const index = this.#findIndex(value)\n // Generate fractional index based on the fractional indices of the elements before and after it\n const indexBefore =\n index === 0 ? null : getIndex(this.#sortedValues[index - 1]!)\n const indexAfter =\n index === this.#sortedValues.length\n ? null\n : getIndex(this.#sortedValues[index]!)\n const fractionalIndex = generateKeyBetween(indexBefore, indexAfter)\n\n // Insert the value at the correct position\n const val = indexedValue(value, fractionalIndex)\n // Splice is O(n) where n = all elements in the collection (i.e. n >= k) !\n this.#sortedValues.splice(index, 0, val)\n\n // Check if the topK changed\n if (index < this.#topKEnd) {\n // The inserted element is either before the top K or within the top K\n // If it is before the top K then it moves the element that was right before the topK into the topK\n // If it is within the top K then the inserted element moves into the top K\n // In both cases the last element of the old top K now moves out of the top K\n const moveInIndex = Math.max(index, this.#topKStart)\n if (moveInIndex < this.#sortedValues.length) {\n // We actually have a topK\n // because in some cases there may not be enough elements in the array to reach the start of the topK\n // e.g. [1, 2, 3] with K = 2 and offset = 3 does not have a topK\n result.moveIn = this.#sortedValues[moveInIndex]!\n\n // We need to remove the element that falls out of the top K\n // The element that falls out of the top K has shifted one to the right\n // because of the element we inserted, so we find it at index topKEnd\n if (this.#topKEnd < this.#sortedValues.length) {\n result.moveOut = this.#sortedValues[this.#topKEnd]!\n }\n }\n }\n\n return result\n }\n\n /**\n * Deletes a value that may or may not be in the topK.\n * IMPORTANT: this assumes that the value is present in the collection\n * if it's not the case it will remove the element\n * that is on the position where the provided `value` would be.\n */\n delete(value: V): TopKChanges<V> {\n const result: TopKChanges<V> = { moveIn: null, moveOut: null }\n\n // Lookup delete position\n const index = this.#findIndex(value)\n // Remove the value at that position\n const [removedElem] = this.#sortedValues.splice(index, 1)\n\n // Check if the topK changed\n if (index < this.#topKEnd) {\n // The removed element is either before the top K or within the top K\n // If it is before the top K then the first element of the topK moves out of the topK\n // If it is within the top K then the removed element moves out of the topK\n result.moveOut = removedElem!\n if (index < this.#topKStart) {\n // The removed element is before the topK\n // so actually, the first element of the topK moves out of the topK\n // and not the element that we removed\n // The first element of the topK is now at index topKStart - 1\n // since we removed an element before the topK\n const moveOutIndex = this.#topKStart - 1\n if (moveOutIndex < this.#sortedValues.length) {\n result.moveOut = this.#sortedValues[moveOutIndex]!\n } else {\n // No value is moving out of the topK\n // because there are no elements in the topK\n result.moveOut = null\n }\n }\n\n // Since we removed an element that was before or in the topK\n // the first element after the topK moved one position to the left\n // and thus falls into the topK now\n const moveInIndex = this.#topKEnd - 1\n if (moveInIndex < this.#sortedValues.length) {\n result.moveIn = this.#sortedValues[moveInIndex]!\n }\n }\n\n return result\n }\n\n // TODO: see if there is a way to refactor the code for insert and delete in the topK above\n // because they are very similar, one is shifting the topK window to the left and the other is shifting it to the right\n // so i have the feeling there is a common pattern here and we can implement both cases using that pattern\n\n #findIndex(value: V): number {\n return binarySearch(this.#sortedValues, indexedValue(value, ``), (a, b) =>\n this.#comparator(getValue(a), getValue(b))\n )\n }\n}\n\n/**\n * Operator for fractional indexed topK operations\n * This operator maintains fractional indices for sorted elements\n * and only updates indices when elements move position\n */\nexport class TopKWithFractionalIndexOperator<K, T> extends UnaryOperator<\n [K, T],\n [K, IndexedValue<T>]\n> {\n #index: Map<K, number> = new Map() // maps keys to their multiplicity\n\n /**\n * topK data structure that supports insertions and deletions\n * and returns changes to the topK.\n */\n #topK: TopK<TaggedValue<K, T>>\n\n constructor(\n id: number,\n inputA: DifferenceStreamReader<[K, T]>,\n output: DifferenceStreamWriter<[K, IndexedValue<T>]>,\n comparator: (a: T, b: T) => number,\n options: TopKWithFractionalIndexOptions\n ) {\n super(id, inputA, output)\n const limit = options.limit ?? Infinity\n const offset = options.offset ?? 0\n const compareTaggedValues = (\n a: TaggedValue<K, T>,\n b: TaggedValue<K, T>\n ) => {\n // First compare on the value\n const valueComparison = comparator(getVal(a), getVal(b))\n if (valueComparison !== 0) {\n return valueComparison\n }\n // If the values are equal, compare on the tag (object identity)\n const tieBreakerA = getTag(a)\n const tieBreakerB = getTag(b)\n return tieBreakerA - tieBreakerB\n }\n this.#topK = this.createTopK(offset, limit, compareTaggedValues)\n options.setSizeCallback?.(() => this.#topK.size)\n options.setWindowFn?.(this.moveTopK.bind(this))\n }\n\n protected createTopK(\n offset: number,\n limit: number,\n comparator: (a: TaggedValue<K, T>, b: TaggedValue<K, T>) => number\n ): TopK<TaggedValue<K, T>> {\n return new TopKArray(offset, limit, comparator)\n }\n\n /**\n * Moves the topK window based on the provided offset and limit.\n * Any changes to the topK are sent to the output.\n */\n moveTopK({ offset, limit }: { offset?: number; limit?: number }) {\n if (!(this.#topK instanceof TopKArray)) {\n throw new Error(\n `Cannot move B+-tree implementation of TopK with fractional index`\n )\n }\n\n const result: Array<[[K, IndexedValue<T>], number]> = []\n\n const diff = this.#topK.move({ offset, limit })\n\n diff.moveIns.forEach((moveIn) => this.handleMoveIn(moveIn, result))\n diff.moveOuts.forEach((moveOut) => this.handleMoveOut(moveOut, result))\n\n if (diff.changes) {\n // There are changes to the topK\n // it could be that moveIns and moveOuts are empty\n // because the collection is lazy, so we will run the graph again to load the data\n this.output.sendData(new MultiSet(result))\n }\n }\n\n run(): void {\n const result: Array<[[K, IndexedValue<T>], number]> = []\n for (const message of this.inputMessages()) {\n for (const [item, multiplicity] of message.getInner()) {\n const [key, value] = item\n this.processElement(key, value, multiplicity, result)\n }\n }\n\n if (result.length > 0) {\n this.output.sendData(new MultiSet(result))\n }\n }\n\n processElement(\n key: K,\n value: T,\n multiplicity: number,\n result: Array<[[K, IndexedValue<T>], number]>\n ): void {\n const { oldMultiplicity, newMultiplicity } = this.addKey(key, multiplicity)\n\n let res: TopKChanges<TaggedValue<K, T>> = {\n moveIn: null,\n moveOut: null,\n }\n if (oldMultiplicity <= 0 && newMultiplicity > 0) {\n // The value was invisible but should now be visible\n // Need to insert it into the array of sorted values\n const taggedValue = tagValue(key, value)\n res = this.#topK.insert(taggedValue)\n } else if (oldMultiplicity > 0 && newMultiplicity <= 0) {\n // The value was visible but should now be invisible\n // Need to remove it from the array of sorted values\n const taggedValue = tagValue(key, value)\n res = this.#topK.delete(taggedValue)\n } else {\n // The value was invisible and it remains invisible\n // or it was visible and remains visible\n // so it doesn't affect the topK\n }\n\n this.handleMoveIn(res.moveIn, result)\n this.handleMoveOut(res.moveOut, result)\n\n return\n }\n\n private handleMoveIn(\n moveIn: IndexedValue<TaggedValue<K, T>> | null,\n result: Array<[[K, IndexedValue<T>], number]>\n ) {\n if (moveIn) {\n const index = getIndex(moveIn)\n const taggedValue = getValue(moveIn)\n const k = getKey(taggedValue)\n const val = getVal(taggedValue)\n result.push([[k, [val, index]], 1])\n }\n }\n\n private handleMoveOut(\n moveOut: IndexedValue<TaggedValue<K, T>> | null,\n result: Array<[[K, IndexedValue<T>], number]>\n ) {\n if (moveOut) {\n const index = getIndex(moveOut)\n const taggedValue = getValue(moveOut)\n const k = getKey(taggedValue)\n const val = getVal(taggedValue)\n result.push([[k, [val, index]], -1])\n }\n }\n\n private getMultiplicity(key: K): number {\n return this.#index.get(key) ?? 0\n }\n\n private addKey(\n key: K,\n multiplicity: number\n ): { oldMultiplicity: number; newMultiplicity: number } {\n const oldMultiplicity = this.getMultiplicity(key)\n const newMultiplicity = oldMultiplicity + multiplicity\n if (newMultiplicity === 0) {\n this.#index.delete(key)\n } else {\n this.#index.set(key, newMultiplicity)\n }\n return { oldMultiplicity, newMultiplicity }\n }\n}\n\n/**\n * Limits the number of results based on a comparator, with optional offset.\n * Uses fractional indexing to minimize the number of changes when elements move positions.\n * Each element is assigned a fractional index that is lexicographically sortable.\n * When elements move, only the indices of the moved elements are updated, not all elements.\n *\n * @param comparator - A function that compares two elements\n * @param options - An optional object containing limit and offset properties\n * @returns A piped operator that orders the elements and limits the number of results\n */\nexport function topKWithFractionalIndex<KType, T>(\n comparator: (a: T, b: T) => number,\n options?: TopKWithFractionalIndexOptions\n): PipedOperator<[KType, T], [KType, IndexedValue<T>]> {\n const opts = options || {}\n\n return (\n stream: IStreamBuilder<[KType, T]>\n ): IStreamBuilder<[KType, IndexedValue<T>]> => {\n const output = new StreamBuilder<[KType, IndexedValue<T>]>(\n stream.graph,\n new DifferenceStreamWriter<[KType, IndexedValue<T>]>()\n )\n const operator = new TopKWithFractionalIndexOperator<KType, T>(\n stream.graph.getNextOperatorId(),\n stream.connectReader(),\n output.writer,\n comparator,\n opts\n )\n stream.graph.addOperator(operator)\n return output\n }\n}\n\n// Abstraction for fractionally indexed values\nexport type FractionalIndex = string\nexport type IndexedValue<V> = [V, FractionalIndex]\n\nexport function indexedValue<V>(\n value: V,\n index: FractionalIndex\n): IndexedValue<V> {\n return [value, index]\n}\n\nexport function getValue<V>(indexedVal: IndexedValue<V>): V {\n return indexedVal[0]\n}\n\nexport function getIndex<V>(indexedVal: IndexedValue<V>): FractionalIndex {\n return indexedVal[1]\n}\n\nexport type Tag = number\nexport type TaggedValue<K, V> = [K, V, Tag]\n\nfunction tagValue<K, V>(key: K, value: V): TaggedValue<K, V> {\n return [key, value, globalObjectIdGenerator.getId(key)]\n}\n\nfunction getKey<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): K {\n return tieBreakerTaggedValue[0]\n}\n\nfunction getVal<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): V {\n return tieBreakerTaggedValue[1]\n}\n\nfunction getTag<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): Tag {\n return tieBreakerTaggedValue[2]\n}\n"],"names":["diffHalfOpen","generateKeyBetween","binarySearch","UnaryOperator","MultiSet","StreamBuilder","DifferenceStreamWriter","globalObjectIdGenerator"],"mappings":";;;;;;;AAsDA,MAAM,UAAgC;AAAA,EACpC,gBAAwC,CAAA;AAAA,EACxC;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YACE,QACA,OACA,YACA;AACA,SAAK,aAAa;AAClB,SAAK,WAAW,SAAS;AACzB,SAAK,cAAc;AAAA,EACrB;AAAA,EAEA,IAAI,OAAe;AACjB,UAAM,SAAS,KAAK;AACpB,UAAM,QAAQ,KAAK,WAAW,KAAK;AACnC,UAAM,YAAY,KAAK,cAAc,SAAS;AAC9C,WAAO,KAAK,IAAI,GAAG,KAAK,IAAI,OAAO,SAAS,CAAC;AAAA,EAC/C;AAAA;AAAA;AAAA;AAAA,EAKA,KAAK;AAAA,IACH;AAAA,IACA;AAAA,EAAA,GAIqB;AACrB,UAAM,YAAY,KAAK;AACvB,UAAM,WAAW,KAAK,WAAW,KAAK;AACtC,UAAM,WAAmB,CAAC,KAAK,YAAY,KAAK,QAAQ;AAExD,SAAK,aAAa,UAAU;AAC5B,SAAK,WAAW,KAAK,cAAc,SAAS;AAE5C,UAAM,WAAmB,CAAC,KAAK,YAAY,KAAK,QAAQ;AACxD,UAAM,EAAE,SAAS,QAAA,IAAYA,MAAAA,aAAa,UAAU,QAAQ;AAE5D,UAAM,UAAkC,CAAA;AACxC,YAAQ,QAAQ,CAAC,UAAU;AACzB,YAAM,QAAQ,KAAK,cAAc,KAAK;AACtC,UAAI,OAAO;AACT,gBAAQ,KAAK,KAAK;AAAA,MACpB;AAAA,IACF,CAAC;AAED,UAAM,WAAmC,CAAA;AACzC,YAAQ,QAAQ,CAAC,UAAU;AACzB,YAAM,QAAQ,KAAK,cAAc,KAAK;AACtC,UAAI,OAAO;AACT,iBAAS,KAAK,KAAK;AAAA,MACrB;AAAA,IACF,CAAC;AAMD,WAAO,EAAE,SAAS,UAAU,SAAS,QAAQ,SAAS,QAAQ,SAAS,EAAA;AAAA,EACzE;AAAA,EAEA,OAAO,OAA0B;AAC/B,UAAM,SAAyB,EAAE,QAAQ,MAAM,SAAS,KAAA;AAGxD,UAAM,QAAQ,KAAK,WAAW,KAAK;AAEnC,UAAM,cACJ,UAAU,IAAI,OAAO,SAAS,KAAK,cAAc,QAAQ,CAAC,CAAE;AAC9D,UAAM,aACJ,UAAU,KAAK,cAAc,SACzB,OACA,SAAS,KAAK,cAAc,KAAK,CAAE;AACzC,UAAM,kBAAkBC,mBAAAA,mBAAmB,aAAa,UAAU;AAGlE,UAAM,MAAM,aAAa,OAAO,eAAe;AAE/C,SAAK,cAAc,OAAO,OAAO,GAAG,GAAG;AAGvC,QAAI,QAAQ,KAAK,UAAU;AAKzB,YAAM,cAAc,KAAK,IAAI,OAAO,KAAK,UAAU;AACnD,UAAI,cAAc,KAAK,cAAc,QAAQ;AAI3C,eAAO,SAAS,KAAK,cAAc,WAAW;AAK9C,YAAI,KAAK,WAAW,KAAK,cAAc,QAAQ;AAC7C,iBAAO,UAAU,KAAK,cAAc,KAAK,QAAQ;AAAA,QACnD;AAAA,MACF;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,OAAO,OAA0B;AAC/B,UAAM,SAAyB,EAAE,QAAQ,MAAM,SAAS,KAAA;AAGxD,UAAM,QAAQ,KAAK,WAAW,KAAK;AAEnC,UAAM,CAAC,WAAW,IAAI,KAAK,cAAc,OAAO,OAAO,CAAC;AAGxD,QAAI,QAAQ,KAAK,UAAU;AAIzB,aAAO,UAAU;AACjB,UAAI,QAAQ,KAAK,YAAY;AAM3B,cAAM,eAAe,KAAK,aAAa;AACvC,YAAI,eAAe,KAAK,cAAc,QAAQ;AAC5C,iBAAO,UAAU,KAAK,cAAc,YAAY;AAAA,QAClD,OAAO;AAGL,iBAAO,UAAU;AAAA,QACnB;AAAA,MACF;AAKA,YAAM,cAAc,KAAK,WAAW;AACpC,UAAI,cAAc,KAAK,cAAc,QAAQ;AAC3C,eAAO,SAAS,KAAK,cAAc,WAAW;AAAA,MAChD;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA,EAMA,WAAW,OAAkB;AAC3B,WAAOC,MAAAA;AAAAA,MAAa,KAAK;AAAA,MAAe,aAAa,OAAO,EAAE;AAAA,MAAG,CAAC,GAAG,MACnE,KAAK,YAAY,SAAS,CAAC,GAAG,SAAS,CAAC,CAAC;AAAA,IAAA;AAAA,EAE7C;AACF;AAOO,MAAM,wCAA8CC,MAAAA,cAGzD;AAAA,EACA,6BAA6B,IAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAM7B;AAAA,EAEA,YACE,IACA,QACA,QACA,YACA,SACA;AACA,UAAM,IAAI,QAAQ,MAAM;AACxB,UAAM,QAAQ,QAAQ,SAAS;AAC/B,UAAM,SAAS,QAAQ,UAAU;AACjC,UAAM,sBAAsB,CAC1B,GACA,MACG;AAEH,YAAM,kBAAkB,WAAW,OAAO,CAAC,GAAG,OAAO,CAAC,CAAC;AACvD,UAAI,oBAAoB,GAAG;AACzB,eAAO;AAAA,MACT;AAEA,YAAM,cAAc,OAAO,CAAC;AAC5B,YAAM,cAAc,OAAO,CAAC;AAC5B,aAAO,cAAc;AAAA,IACvB;AACA,SAAK,QAAQ,KAAK,WAAW,QAAQ,OAAO,mBAAmB;AAC/D,YAAQ,kBAAkB,MAAM,KAAK,MAAM,IAAI;AAC/C,YAAQ,cAAc,KAAK,SAAS,KAAK,IAAI,CAAC;AAAA,EAChD;AAAA,EAEU,WACR,QACA,OACA,YACyB;AACzB,WAAO,IAAI,UAAU,QAAQ,OAAO,UAAU;AAAA,EAChD;AAAA;AAAA;AAAA;AAAA;AAAA,EAMA,SAAS,EAAE,QAAQ,SAA8C;AAC/D,QAAI,EAAE,KAAK,iBAAiB,YAAY;AACtC,YAAM,IAAI;AAAA,QACR;AAAA,MAAA;AAAA,IAEJ;AAEA,UAAM,SAAgD,CAAA;AAEtD,UAAM,OAAO,KAAK,MAAM,KAAK,EAAE,QAAQ,OAAO;AAE9C,SAAK,QAAQ,QAAQ,CAAC,WAAW,KAAK,aAAa,QAAQ,MAAM,CAAC;AAClE,SAAK,SAAS,QAAQ,CAAC,YAAY,KAAK,cAAc,SAAS,MAAM,CAAC;AAEtE,QAAI,KAAK,SAAS;AAIhB,WAAK,OAAO,SAAS,IAAIC,SAAAA,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,MAAY;AACV,UAAM,SAAgD,CAAA;AACtD,eAAW,WAAW,KAAK,iBAAiB;AAC1C,iBAAW,CAAC,MAAM,YAAY,KAAK,QAAQ,YAAY;AACrD,cAAM,CAAC,KAAK,KAAK,IAAI;AACrB,aAAK,eAAe,KAAK,OAAO,cAAc,MAAM;AAAA,MACtD;AAAA,IACF;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAIA,SAAAA,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,eACE,KACA,OACA,cACA,QACM;AACN,UAAM,EAAE,iBAAiB,gBAAA,IAAoB,KAAK,OAAO,KAAK,YAAY;AAE1E,QAAI,MAAsC;AAAA,MACxC,QAAQ;AAAA,MACR,SAAS;AAAA,IAAA;AAEX,QAAI,mBAAmB,KAAK,kBAAkB,GAAG;AAG/C,YAAM,cAAc,SAAS,KAAK,KAAK;AACvC,YAAM,KAAK,MAAM,OAAO,WAAW;AAAA,IACrC,WAAW,kBAAkB,KAAK,mBAAmB,GAAG;AAGtD,YAAM,cAAc,SAAS,KAAK,KAAK;AACvC,YAAM,KAAK,MAAM,OAAO,WAAW;AAAA,IACrC,MAAO;AAMP,SAAK,aAAa,IAAI,QAAQ,MAAM;AACpC,SAAK,cAAc,IAAI,SAAS,MAAM;AAEtC;AAAA,EACF;AAAA,EAEQ,aACN,QACA,QACA;AACA,QAAI,QAAQ;AACV,YAAM,QAAQ,SAAS,MAAM;AAC7B,YAAM,cAAc,SAAS,MAAM;AACnC,YAAM,IAAI,OAAO,WAAW;AAC5B,YAAM,MAAM,OAAO,WAAW;AAC9B,aAAO,KAAK,CAAC,CAAC,GAAG,CAAC,KAAK,KAAK,CAAC,GAAG,CAAC,CAAC;AAAA,IACpC;AAAA,EACF;AAAA,EAEQ,cACN,SACA,QACA;AACA,QAAI,SAAS;AACX,YAAM,QAAQ,SAAS,OAAO;AAC9B,YAAM,cAAc,SAAS,OAAO;AACpC,YAAM,IAAI,OAAO,WAAW;AAC5B,YAAM,MAAM,OAAO,WAAW;AAC9B,aAAO,KAAK,CAAC,CAAC,GAAG,CAAC,KAAK,KAAK,CAAC,GAAG,EAAE,CAAC;AAAA,IACrC;AAAA,EACF;AAAA,EAEQ,gBAAgB,KAAgB;AACtC,WAAO,KAAK,OAAO,IAAI,GAAG,KAAK;AAAA,EACjC;AAAA,EAEQ,OACN,KACA,cACsD;AACtD,UAAM,kBAAkB,KAAK,gBAAgB,GAAG;AAChD,UAAM,kBAAkB,kBAAkB;AAC1C,QAAI,oBAAoB,GAAG;AACzB,WAAK,OAAO,OAAO,GAAG;AAAA,IACxB,OAAO;AACL,WAAK,OAAO,IAAI,KAAK,eAAe;AAAA,IACtC;AACA,WAAO,EAAE,iBAAiB,gBAAA;AAAA,EAC5B;AACF;AAYO,SAAS,wBACd,YACA,SACqD;AACrD,QAAM,OAAO,WAAW,CAAA;AAExB,SAAO,CACL,WAC6C;AAC7C,UAAM,SAAS,IAAIC,GAAAA;AAAAA,MACjB,OAAO;AAAA,MACP,IAAIC,MAAAA,uBAAA;AAAA,IAAiD;AAEvD,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,OAAO;AAAA,MACP;AAAA,MACA;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;AAMO,SAAS,aACd,OACA,OACiB;AACjB,SAAO,CAAC,OAAO,KAAK;AACtB;AAEO,SAAS,SAAY,YAAgC;AAC1D,SAAO,WAAW,CAAC;AACrB;AAEO,SAAS,SAAY,YAA8C;AACxE,SAAO,WAAW,CAAC;AACrB;AAKA,SAAS,SAAe,KAAQ,OAA6B;AAC3D,SAAO,CAAC,KAAK,OAAOC,MAAAA,wBAAwB,MAAM,GAAG,CAAC;AACxD;AAEA,SAAS,OAAa,uBAA6C;AACjE,SAAO,sBAAsB,CAAC;AAChC;AAEA,SAAS,OAAa,uBAA6C;AACjE,SAAO,sBAAsB,CAAC;AAChC;AAEA,SAAS,OAAa,uBAA+C;AACnE,SAAO,sBAAsB,CAAC;AAChC;;;;;;"}
|
|
1
|
+
{"version":3,"file":"topKWithFractionalIndex.cjs","sources":["../../../src/operators/topKWithFractionalIndex.ts"],"sourcesContent":["import { generateKeyBetween } from \"fractional-indexing\"\nimport { DifferenceStreamWriter, UnaryOperator } from \"../graph.js\"\nimport { StreamBuilder } from \"../d2.js\"\nimport { MultiSet } from \"../multiset.js\"\nimport {\n binarySearch,\n diffHalfOpen,\n globalObjectIdGenerator,\n} from \"../utils.js\"\nimport type { HRange } from \"../utils.js\"\nimport type { DifferenceStreamReader } from \"../graph.js\"\nimport type { IStreamBuilder, PipedOperator } from \"../types.js\"\n\nexport interface TopKWithFractionalIndexOptions {\n limit?: number\n offset?: number\n setSizeCallback?: (getSize: () => number) => void\n setWindowFn?: (\n windowFn: (options: { offset?: number; limit?: number }) => void\n ) => void\n}\n\nexport type TopKChanges<V> = {\n /** Indicates which element moves into the topK (if any) */\n moveIn: IndexedValue<V> | null\n /** Indicates which element moves out of the topK (if any) */\n moveOut: IndexedValue<V> | null\n}\n\nexport type TopKMoveChanges<V> = {\n /** Flag that marks whether there were any changes to the topK */\n changes: boolean\n /** Indicates which elements move into the topK (if any) */\n moveIns: Array<IndexedValue<V>>\n /** Indicates which elements move out of the topK (if any) */\n moveOuts: Array<IndexedValue<V>>\n}\n\n/**\n * A topK data structure that supports insertions and deletions\n * and returns changes to the topK.\n */\nexport interface TopK<V> {\n size: number\n insert: (value: V) => TopKChanges<V>\n delete: (value: V) => TopKChanges<V>\n}\n\n/**\n * Implementation of a topK data structure.\n * Uses a sorted array internally to store the values and keeps a topK window over that array.\n * Inserts and deletes are O(n) operations because worst case an element is inserted/deleted\n * at the start of the array which causes all the elements to shift to the right/left.\n */\nclass TopKArray<V> implements TopK<V> {\n #sortedValues: Array<IndexedValue<V>> = []\n #comparator: (a: V, b: V) => number\n #topKStart: number\n #topKEnd: number\n\n constructor(\n offset: number,\n limit: number,\n comparator: (a: V, b: V) => number\n ) {\n this.#topKStart = offset\n this.#topKEnd = offset + limit\n this.#comparator = comparator\n }\n\n get size(): number {\n const offset = this.#topKStart\n const limit = this.#topKEnd - this.#topKStart\n const available = this.#sortedValues.length - offset\n return Math.max(0, Math.min(limit, available))\n }\n\n /**\n * Moves the topK window\n */\n move({\n offset,\n limit,\n }: {\n offset?: number\n limit?: number\n }): TopKMoveChanges<V> {\n const oldOffset = this.#topKStart\n const oldLimit = this.#topKEnd - this.#topKStart\n\n // `this.#topKEnd` can be `Infinity` if it has no limit\n // but `diffHalfOpen` expects a finite range\n // so we restrict it to the size of the topK if topKEnd is infinite\n const oldRange: HRange = [\n this.#topKStart,\n this.#topKEnd === Infinity ? this.#topKStart + this.size : this.#topKEnd,\n ]\n\n this.#topKStart = offset ?? oldOffset\n this.#topKEnd = this.#topKStart + (limit ?? oldLimit) // can be `Infinity` if limit is `Infinity`\n\n // Also handle `Infinity` in the newRange\n const newRange: HRange = [\n this.#topKStart,\n this.#topKEnd === Infinity\n ? Math.max(this.#topKStart + this.size, oldRange[1]) // since the new limit is Infinity we need to take everything (so we need to take the biggest (finite) topKEnd)\n : this.#topKEnd,\n ]\n const { onlyInA, onlyInB } = diffHalfOpen(oldRange, newRange)\n\n const moveIns: Array<IndexedValue<V>> = []\n onlyInB.forEach((index) => {\n const value = this.#sortedValues[index]\n if (value) {\n moveIns.push(value)\n }\n })\n\n const moveOuts: Array<IndexedValue<V>> = []\n onlyInA.forEach((index) => {\n const value = this.#sortedValues[index]\n if (value) {\n moveOuts.push(value)\n }\n })\n\n // It could be that there are changes (i.e. moveIns or moveOuts)\n // but that the collection is lazy so we don't have the data yet that needs to move in/out\n // so `moveIns` and `moveOuts` will be empty but `changes` will be true\n // this will tell the caller that it needs to run the graph to load more data\n return { moveIns, moveOuts, changes: onlyInA.length + onlyInB.length > 0 }\n }\n\n insert(value: V): TopKChanges<V> {\n const result: TopKChanges<V> = { moveIn: null, moveOut: null }\n\n // Lookup insert position\n const index = this.#findIndex(value)\n // Generate fractional index based on the fractional indices of the elements before and after it\n const indexBefore =\n index === 0 ? null : getIndex(this.#sortedValues[index - 1]!)\n const indexAfter =\n index === this.#sortedValues.length\n ? null\n : getIndex(this.#sortedValues[index]!)\n const fractionalIndex = generateKeyBetween(indexBefore, indexAfter)\n\n // Insert the value at the correct position\n const val = indexedValue(value, fractionalIndex)\n // Splice is O(n) where n = all elements in the collection (i.e. n >= k) !\n this.#sortedValues.splice(index, 0, val)\n\n // Check if the topK changed\n if (index < this.#topKEnd) {\n // The inserted element is either before the top K or within the top K\n // If it is before the top K then it moves the element that was right before the topK into the topK\n // If it is within the top K then the inserted element moves into the top K\n // In both cases the last element of the old top K now moves out of the top K\n const moveInIndex = Math.max(index, this.#topKStart)\n if (moveInIndex < this.#sortedValues.length) {\n // We actually have a topK\n // because in some cases there may not be enough elements in the array to reach the start of the topK\n // e.g. [1, 2, 3] with K = 2 and offset = 3 does not have a topK\n result.moveIn = this.#sortedValues[moveInIndex]!\n\n // We need to remove the element that falls out of the top K\n // The element that falls out of the top K has shifted one to the right\n // because of the element we inserted, so we find it at index topKEnd\n if (this.#topKEnd < this.#sortedValues.length) {\n result.moveOut = this.#sortedValues[this.#topKEnd]!\n }\n }\n }\n\n return result\n }\n\n /**\n * Deletes a value that may or may not be in the topK.\n * IMPORTANT: this assumes that the value is present in the collection\n * if it's not the case it will remove the element\n * that is on the position where the provided `value` would be.\n */\n delete(value: V): TopKChanges<V> {\n const result: TopKChanges<V> = { moveIn: null, moveOut: null }\n\n // Lookup delete position\n const index = this.#findIndex(value)\n // Remove the value at that position\n const [removedElem] = this.#sortedValues.splice(index, 1)\n\n // Check if the topK changed\n if (index < this.#topKEnd) {\n // The removed element is either before the top K or within the top K\n // If it is before the top K then the first element of the topK moves out of the topK\n // If it is within the top K then the removed element moves out of the topK\n result.moveOut = removedElem!\n if (index < this.#topKStart) {\n // The removed element is before the topK\n // so actually, the first element of the topK moves out of the topK\n // and not the element that we removed\n // The first element of the topK is now at index topKStart - 1\n // since we removed an element before the topK\n const moveOutIndex = this.#topKStart - 1\n if (moveOutIndex < this.#sortedValues.length) {\n result.moveOut = this.#sortedValues[moveOutIndex]!\n } else {\n // No value is moving out of the topK\n // because there are no elements in the topK\n result.moveOut = null\n }\n }\n\n // Since we removed an element that was before or in the topK\n // the first element after the topK moved one position to the left\n // and thus falls into the topK now\n const moveInIndex = this.#topKEnd - 1\n if (moveInIndex < this.#sortedValues.length) {\n result.moveIn = this.#sortedValues[moveInIndex]!\n }\n }\n\n return result\n }\n\n // TODO: see if there is a way to refactor the code for insert and delete in the topK above\n // because they are very similar, one is shifting the topK window to the left and the other is shifting it to the right\n // so i have the feeling there is a common pattern here and we can implement both cases using that pattern\n\n #findIndex(value: V): number {\n return binarySearch(this.#sortedValues, indexedValue(value, ``), (a, b) =>\n this.#comparator(getValue(a), getValue(b))\n )\n }\n}\n\n/**\n * Operator for fractional indexed topK operations\n * This operator maintains fractional indices for sorted elements\n * and only updates indices when elements move position\n */\nexport class TopKWithFractionalIndexOperator<K, T> extends UnaryOperator<\n [K, T],\n [K, IndexedValue<T>]\n> {\n #index: Map<K, number> = new Map() // maps keys to their multiplicity\n\n /**\n * topK data structure that supports insertions and deletions\n * and returns changes to the topK.\n */\n #topK: TopK<TaggedValue<K, T>>\n\n constructor(\n id: number,\n inputA: DifferenceStreamReader<[K, T]>,\n output: DifferenceStreamWriter<[K, IndexedValue<T>]>,\n comparator: (a: T, b: T) => number,\n options: TopKWithFractionalIndexOptions\n ) {\n super(id, inputA, output)\n const limit = options.limit ?? Infinity\n const offset = options.offset ?? 0\n const compareTaggedValues = (\n a: TaggedValue<K, T>,\n b: TaggedValue<K, T>\n ) => {\n // First compare on the value\n const valueComparison = comparator(getVal(a), getVal(b))\n if (valueComparison !== 0) {\n return valueComparison\n }\n // If the values are equal, compare on the tag (object identity)\n const tieBreakerA = getTag(a)\n const tieBreakerB = getTag(b)\n return tieBreakerA - tieBreakerB\n }\n this.#topK = this.createTopK(offset, limit, compareTaggedValues)\n options.setSizeCallback?.(() => this.#topK.size)\n options.setWindowFn?.(this.moveTopK.bind(this))\n }\n\n protected createTopK(\n offset: number,\n limit: number,\n comparator: (a: TaggedValue<K, T>, b: TaggedValue<K, T>) => number\n ): TopK<TaggedValue<K, T>> {\n return new TopKArray(offset, limit, comparator)\n }\n\n /**\n * Moves the topK window based on the provided offset and limit.\n * Any changes to the topK are sent to the output.\n */\n moveTopK({ offset, limit }: { offset?: number; limit?: number }) {\n if (!(this.#topK instanceof TopKArray)) {\n throw new Error(\n `Cannot move B+-tree implementation of TopK with fractional index`\n )\n }\n\n const result: Array<[[K, IndexedValue<T>], number]> = []\n\n const diff = this.#topK.move({ offset, limit })\n\n diff.moveIns.forEach((moveIn) => this.handleMoveIn(moveIn, result))\n diff.moveOuts.forEach((moveOut) => this.handleMoveOut(moveOut, result))\n\n if (diff.changes) {\n // There are changes to the topK\n // it could be that moveIns and moveOuts are empty\n // because the collection is lazy, so we will run the graph again to load the data\n this.output.sendData(new MultiSet(result))\n }\n }\n\n run(): void {\n const result: Array<[[K, IndexedValue<T>], number]> = []\n for (const message of this.inputMessages()) {\n for (const [item, multiplicity] of message.getInner()) {\n const [key, value] = item\n this.processElement(key, value, multiplicity, result)\n }\n }\n\n if (result.length > 0) {\n this.output.sendData(new MultiSet(result))\n }\n }\n\n processElement(\n key: K,\n value: T,\n multiplicity: number,\n result: Array<[[K, IndexedValue<T>], number]>\n ): void {\n const { oldMultiplicity, newMultiplicity } = this.addKey(key, multiplicity)\n\n let res: TopKChanges<TaggedValue<K, T>> = {\n moveIn: null,\n moveOut: null,\n }\n if (oldMultiplicity <= 0 && newMultiplicity > 0) {\n // The value was invisible but should now be visible\n // Need to insert it into the array of sorted values\n const taggedValue = tagValue(key, value)\n res = this.#topK.insert(taggedValue)\n } else if (oldMultiplicity > 0 && newMultiplicity <= 0) {\n // The value was visible but should now be invisible\n // Need to remove it from the array of sorted values\n const taggedValue = tagValue(key, value)\n res = this.#topK.delete(taggedValue)\n } else {\n // The value was invisible and it remains invisible\n // or it was visible and remains visible\n // so it doesn't affect the topK\n }\n\n this.handleMoveIn(res.moveIn, result)\n this.handleMoveOut(res.moveOut, result)\n\n return\n }\n\n private handleMoveIn(\n moveIn: IndexedValue<TaggedValue<K, T>> | null,\n result: Array<[[K, IndexedValue<T>], number]>\n ) {\n if (moveIn) {\n const index = getIndex(moveIn)\n const taggedValue = getValue(moveIn)\n const k = getKey(taggedValue)\n const val = getVal(taggedValue)\n result.push([[k, [val, index]], 1])\n }\n }\n\n private handleMoveOut(\n moveOut: IndexedValue<TaggedValue<K, T>> | null,\n result: Array<[[K, IndexedValue<T>], number]>\n ) {\n if (moveOut) {\n const index = getIndex(moveOut)\n const taggedValue = getValue(moveOut)\n const k = getKey(taggedValue)\n const val = getVal(taggedValue)\n result.push([[k, [val, index]], -1])\n }\n }\n\n private getMultiplicity(key: K): number {\n return this.#index.get(key) ?? 0\n }\n\n private addKey(\n key: K,\n multiplicity: number\n ): { oldMultiplicity: number; newMultiplicity: number } {\n const oldMultiplicity = this.getMultiplicity(key)\n const newMultiplicity = oldMultiplicity + multiplicity\n if (newMultiplicity === 0) {\n this.#index.delete(key)\n } else {\n this.#index.set(key, newMultiplicity)\n }\n return { oldMultiplicity, newMultiplicity }\n }\n}\n\n/**\n * Limits the number of results based on a comparator, with optional offset.\n * Uses fractional indexing to minimize the number of changes when elements move positions.\n * Each element is assigned a fractional index that is lexicographically sortable.\n * When elements move, only the indices of the moved elements are updated, not all elements.\n *\n * @param comparator - A function that compares two elements\n * @param options - An optional object containing limit and offset properties\n * @returns A piped operator that orders the elements and limits the number of results\n */\nexport function topKWithFractionalIndex<KType, T>(\n comparator: (a: T, b: T) => number,\n options?: TopKWithFractionalIndexOptions\n): PipedOperator<[KType, T], [KType, IndexedValue<T>]> {\n const opts = options || {}\n\n return (\n stream: IStreamBuilder<[KType, T]>\n ): IStreamBuilder<[KType, IndexedValue<T>]> => {\n const output = new StreamBuilder<[KType, IndexedValue<T>]>(\n stream.graph,\n new DifferenceStreamWriter<[KType, IndexedValue<T>]>()\n )\n const operator = new TopKWithFractionalIndexOperator<KType, T>(\n stream.graph.getNextOperatorId(),\n stream.connectReader(),\n output.writer,\n comparator,\n opts\n )\n stream.graph.addOperator(operator)\n return output\n }\n}\n\n// Abstraction for fractionally indexed values\nexport type FractionalIndex = string\nexport type IndexedValue<V> = [V, FractionalIndex]\n\nexport function indexedValue<V>(\n value: V,\n index: FractionalIndex\n): IndexedValue<V> {\n return [value, index]\n}\n\nexport function getValue<V>(indexedVal: IndexedValue<V>): V {\n return indexedVal[0]\n}\n\nexport function getIndex<V>(indexedVal: IndexedValue<V>): FractionalIndex {\n return indexedVal[1]\n}\n\nexport type Tag = number\nexport type TaggedValue<K, V> = [K, V, Tag]\n\nfunction tagValue<K, V>(key: K, value: V): TaggedValue<K, V> {\n return [key, value, globalObjectIdGenerator.getId(key)]\n}\n\nfunction getKey<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): K {\n return tieBreakerTaggedValue[0]\n}\n\nfunction getVal<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): V {\n return tieBreakerTaggedValue[1]\n}\n\nfunction getTag<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): Tag {\n return tieBreakerTaggedValue[2]\n}\n"],"names":["diffHalfOpen","generateKeyBetween","binarySearch","UnaryOperator","MultiSet","StreamBuilder","DifferenceStreamWriter","globalObjectIdGenerator"],"mappings":";;;;;;;AAsDA,MAAM,UAAgC;AAAA,EACpC,gBAAwC,CAAA;AAAA,EACxC;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YACE,QACA,OACA,YACA;AACA,SAAK,aAAa;AAClB,SAAK,WAAW,SAAS;AACzB,SAAK,cAAc;AAAA,EACrB;AAAA,EAEA,IAAI,OAAe;AACjB,UAAM,SAAS,KAAK;AACpB,UAAM,QAAQ,KAAK,WAAW,KAAK;AACnC,UAAM,YAAY,KAAK,cAAc,SAAS;AAC9C,WAAO,KAAK,IAAI,GAAG,KAAK,IAAI,OAAO,SAAS,CAAC;AAAA,EAC/C;AAAA;AAAA;AAAA;AAAA,EAKA,KAAK;AAAA,IACH;AAAA,IACA;AAAA,EAAA,GAIqB;AACrB,UAAM,YAAY,KAAK;AACvB,UAAM,WAAW,KAAK,WAAW,KAAK;AAKtC,UAAM,WAAmB;AAAA,MACvB,KAAK;AAAA,MACL,KAAK,aAAa,WAAW,KAAK,aAAa,KAAK,OAAO,KAAK;AAAA,IAAA;AAGlE,SAAK,aAAa,UAAU;AAC5B,SAAK,WAAW,KAAK,cAAc,SAAS;AAG5C,UAAM,WAAmB;AAAA,MACvB,KAAK;AAAA,MACL,KAAK,aAAa,WACd,KAAK,IAAI,KAAK,aAAa,KAAK,MAAM,SAAS,CAAC,CAAC,IACjD,KAAK;AAAA,IAAA;AAEX,UAAM,EAAE,SAAS,QAAA,IAAYA,MAAAA,aAAa,UAAU,QAAQ;AAE5D,UAAM,UAAkC,CAAA;AACxC,YAAQ,QAAQ,CAAC,UAAU;AACzB,YAAM,QAAQ,KAAK,cAAc,KAAK;AACtC,UAAI,OAAO;AACT,gBAAQ,KAAK,KAAK;AAAA,MACpB;AAAA,IACF,CAAC;AAED,UAAM,WAAmC,CAAA;AACzC,YAAQ,QAAQ,CAAC,UAAU;AACzB,YAAM,QAAQ,KAAK,cAAc,KAAK;AACtC,UAAI,OAAO;AACT,iBAAS,KAAK,KAAK;AAAA,MACrB;AAAA,IACF,CAAC;AAMD,WAAO,EAAE,SAAS,UAAU,SAAS,QAAQ,SAAS,QAAQ,SAAS,EAAA;AAAA,EACzE;AAAA,EAEA,OAAO,OAA0B;AAC/B,UAAM,SAAyB,EAAE,QAAQ,MAAM,SAAS,KAAA;AAGxD,UAAM,QAAQ,KAAK,WAAW,KAAK;AAEnC,UAAM,cACJ,UAAU,IAAI,OAAO,SAAS,KAAK,cAAc,QAAQ,CAAC,CAAE;AAC9D,UAAM,aACJ,UAAU,KAAK,cAAc,SACzB,OACA,SAAS,KAAK,cAAc,KAAK,CAAE;AACzC,UAAM,kBAAkBC,mBAAAA,mBAAmB,aAAa,UAAU;AAGlE,UAAM,MAAM,aAAa,OAAO,eAAe;AAE/C,SAAK,cAAc,OAAO,OAAO,GAAG,GAAG;AAGvC,QAAI,QAAQ,KAAK,UAAU;AAKzB,YAAM,cAAc,KAAK,IAAI,OAAO,KAAK,UAAU;AACnD,UAAI,cAAc,KAAK,cAAc,QAAQ;AAI3C,eAAO,SAAS,KAAK,cAAc,WAAW;AAK9C,YAAI,KAAK,WAAW,KAAK,cAAc,QAAQ;AAC7C,iBAAO,UAAU,KAAK,cAAc,KAAK,QAAQ;AAAA,QACnD;AAAA,MACF;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,OAAO,OAA0B;AAC/B,UAAM,SAAyB,EAAE,QAAQ,MAAM,SAAS,KAAA;AAGxD,UAAM,QAAQ,KAAK,WAAW,KAAK;AAEnC,UAAM,CAAC,WAAW,IAAI,KAAK,cAAc,OAAO,OAAO,CAAC;AAGxD,QAAI,QAAQ,KAAK,UAAU;AAIzB,aAAO,UAAU;AACjB,UAAI,QAAQ,KAAK,YAAY;AAM3B,cAAM,eAAe,KAAK,aAAa;AACvC,YAAI,eAAe,KAAK,cAAc,QAAQ;AAC5C,iBAAO,UAAU,KAAK,cAAc,YAAY;AAAA,QAClD,OAAO;AAGL,iBAAO,UAAU;AAAA,QACnB;AAAA,MACF;AAKA,YAAM,cAAc,KAAK,WAAW;AACpC,UAAI,cAAc,KAAK,cAAc,QAAQ;AAC3C,eAAO,SAAS,KAAK,cAAc,WAAW;AAAA,MAChD;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA,EAMA,WAAW,OAAkB;AAC3B,WAAOC,MAAAA;AAAAA,MAAa,KAAK;AAAA,MAAe,aAAa,OAAO,EAAE;AAAA,MAAG,CAAC,GAAG,MACnE,KAAK,YAAY,SAAS,CAAC,GAAG,SAAS,CAAC,CAAC;AAAA,IAAA;AAAA,EAE7C;AACF;AAOO,MAAM,wCAA8CC,MAAAA,cAGzD;AAAA,EACA,6BAA6B,IAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAM7B;AAAA,EAEA,YACE,IACA,QACA,QACA,YACA,SACA;AACA,UAAM,IAAI,QAAQ,MAAM;AACxB,UAAM,QAAQ,QAAQ,SAAS;AAC/B,UAAM,SAAS,QAAQ,UAAU;AACjC,UAAM,sBAAsB,CAC1B,GACA,MACG;AAEH,YAAM,kBAAkB,WAAW,OAAO,CAAC,GAAG,OAAO,CAAC,CAAC;AACvD,UAAI,oBAAoB,GAAG;AACzB,eAAO;AAAA,MACT;AAEA,YAAM,cAAc,OAAO,CAAC;AAC5B,YAAM,cAAc,OAAO,CAAC;AAC5B,aAAO,cAAc;AAAA,IACvB;AACA,SAAK,QAAQ,KAAK,WAAW,QAAQ,OAAO,mBAAmB;AAC/D,YAAQ,kBAAkB,MAAM,KAAK,MAAM,IAAI;AAC/C,YAAQ,cAAc,KAAK,SAAS,KAAK,IAAI,CAAC;AAAA,EAChD;AAAA,EAEU,WACR,QACA,OACA,YACyB;AACzB,WAAO,IAAI,UAAU,QAAQ,OAAO,UAAU;AAAA,EAChD;AAAA;AAAA;AAAA;AAAA;AAAA,EAMA,SAAS,EAAE,QAAQ,SAA8C;AAC/D,QAAI,EAAE,KAAK,iBAAiB,YAAY;AACtC,YAAM,IAAI;AAAA,QACR;AAAA,MAAA;AAAA,IAEJ;AAEA,UAAM,SAAgD,CAAA;AAEtD,UAAM,OAAO,KAAK,MAAM,KAAK,EAAE,QAAQ,OAAO;AAE9C,SAAK,QAAQ,QAAQ,CAAC,WAAW,KAAK,aAAa,QAAQ,MAAM,CAAC;AAClE,SAAK,SAAS,QAAQ,CAAC,YAAY,KAAK,cAAc,SAAS,MAAM,CAAC;AAEtE,QAAI,KAAK,SAAS;AAIhB,WAAK,OAAO,SAAS,IAAIC,SAAAA,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,MAAY;AACV,UAAM,SAAgD,CAAA;AACtD,eAAW,WAAW,KAAK,iBAAiB;AAC1C,iBAAW,CAAC,MAAM,YAAY,KAAK,QAAQ,YAAY;AACrD,cAAM,CAAC,KAAK,KAAK,IAAI;AACrB,aAAK,eAAe,KAAK,OAAO,cAAc,MAAM;AAAA,MACtD;AAAA,IACF;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAIA,SAAAA,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,eACE,KACA,OACA,cACA,QACM;AACN,UAAM,EAAE,iBAAiB,gBAAA,IAAoB,KAAK,OAAO,KAAK,YAAY;AAE1E,QAAI,MAAsC;AAAA,MACxC,QAAQ;AAAA,MACR,SAAS;AAAA,IAAA;AAEX,QAAI,mBAAmB,KAAK,kBAAkB,GAAG;AAG/C,YAAM,cAAc,SAAS,KAAK,KAAK;AACvC,YAAM,KAAK,MAAM,OAAO,WAAW;AAAA,IACrC,WAAW,kBAAkB,KAAK,mBAAmB,GAAG;AAGtD,YAAM,cAAc,SAAS,KAAK,KAAK;AACvC,YAAM,KAAK,MAAM,OAAO,WAAW;AAAA,IACrC,MAAO;AAMP,SAAK,aAAa,IAAI,QAAQ,MAAM;AACpC,SAAK,cAAc,IAAI,SAAS,MAAM;AAEtC;AAAA,EACF;AAAA,EAEQ,aACN,QACA,QACA;AACA,QAAI,QAAQ;AACV,YAAM,QAAQ,SAAS,MAAM;AAC7B,YAAM,cAAc,SAAS,MAAM;AACnC,YAAM,IAAI,OAAO,WAAW;AAC5B,YAAM,MAAM,OAAO,WAAW;AAC9B,aAAO,KAAK,CAAC,CAAC,GAAG,CAAC,KAAK,KAAK,CAAC,GAAG,CAAC,CAAC;AAAA,IACpC;AAAA,EACF;AAAA,EAEQ,cACN,SACA,QACA;AACA,QAAI,SAAS;AACX,YAAM,QAAQ,SAAS,OAAO;AAC9B,YAAM,cAAc,SAAS,OAAO;AACpC,YAAM,IAAI,OAAO,WAAW;AAC5B,YAAM,MAAM,OAAO,WAAW;AAC9B,aAAO,KAAK,CAAC,CAAC,GAAG,CAAC,KAAK,KAAK,CAAC,GAAG,EAAE,CAAC;AAAA,IACrC;AAAA,EACF;AAAA,EAEQ,gBAAgB,KAAgB;AACtC,WAAO,KAAK,OAAO,IAAI,GAAG,KAAK;AAAA,EACjC;AAAA,EAEQ,OACN,KACA,cACsD;AACtD,UAAM,kBAAkB,KAAK,gBAAgB,GAAG;AAChD,UAAM,kBAAkB,kBAAkB;AAC1C,QAAI,oBAAoB,GAAG;AACzB,WAAK,OAAO,OAAO,GAAG;AAAA,IACxB,OAAO;AACL,WAAK,OAAO,IAAI,KAAK,eAAe;AAAA,IACtC;AACA,WAAO,EAAE,iBAAiB,gBAAA;AAAA,EAC5B;AACF;AAYO,SAAS,wBACd,YACA,SACqD;AACrD,QAAM,OAAO,WAAW,CAAA;AAExB,SAAO,CACL,WAC6C;AAC7C,UAAM,SAAS,IAAIC,GAAAA;AAAAA,MACjB,OAAO;AAAA,MACP,IAAIC,MAAAA,uBAAA;AAAA,IAAiD;AAEvD,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,OAAO;AAAA,MACP;AAAA,MACA;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;AAMO,SAAS,aACd,OACA,OACiB;AACjB,SAAO,CAAC,OAAO,KAAK;AACtB;AAEO,SAAS,SAAY,YAAgC;AAC1D,SAAO,WAAW,CAAC;AACrB;AAEO,SAAS,SAAY,YAA8C;AACxE,SAAO,WAAW,CAAC;AACrB;AAKA,SAAS,SAAe,KAAQ,OAA6B;AAC3D,SAAO,CAAC,KAAK,OAAOC,MAAAA,wBAAwB,MAAM,GAAG,CAAC;AACxD;AAEA,SAAS,OAAa,uBAA6C;AACjE,SAAO,sBAAsB,CAAC;AAChC;AAEA,SAAS,OAAa,uBAA6C;AACjE,SAAO,sBAAsB,CAAC;AAChC;AAEA,SAAS,OAAa,uBAA+C;AACnE,SAAO,sBAAsB,CAAC;AAChC;;;;;;"}
|
|
@@ -28,10 +28,16 @@ class TopKArray {
|
|
|
28
28
|
}) {
|
|
29
29
|
const oldOffset = this.#topKStart;
|
|
30
30
|
const oldLimit = this.#topKEnd - this.#topKStart;
|
|
31
|
-
const oldRange = [
|
|
31
|
+
const oldRange = [
|
|
32
|
+
this.#topKStart,
|
|
33
|
+
this.#topKEnd === Infinity ? this.#topKStart + this.size : this.#topKEnd
|
|
34
|
+
];
|
|
32
35
|
this.#topKStart = offset ?? oldOffset;
|
|
33
36
|
this.#topKEnd = this.#topKStart + (limit ?? oldLimit);
|
|
34
|
-
const newRange = [
|
|
37
|
+
const newRange = [
|
|
38
|
+
this.#topKStart,
|
|
39
|
+
this.#topKEnd === Infinity ? Math.max(this.#topKStart + this.size, oldRange[1]) : this.#topKEnd
|
|
40
|
+
];
|
|
35
41
|
const { onlyInA, onlyInB } = diffHalfOpen(oldRange, newRange);
|
|
36
42
|
const moveIns = [];
|
|
37
43
|
onlyInB.forEach((index) => {
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"topKWithFractionalIndex.js","sources":["../../../src/operators/topKWithFractionalIndex.ts"],"sourcesContent":["import { generateKeyBetween } from \"fractional-indexing\"\nimport { DifferenceStreamWriter, UnaryOperator } from \"../graph.js\"\nimport { StreamBuilder } from \"../d2.js\"\nimport { MultiSet } from \"../multiset.js\"\nimport {\n binarySearch,\n diffHalfOpen,\n globalObjectIdGenerator,\n} from \"../utils.js\"\nimport type { HRange } from \"../utils.js\"\nimport type { DifferenceStreamReader } from \"../graph.js\"\nimport type { IStreamBuilder, PipedOperator } from \"../types.js\"\n\nexport interface TopKWithFractionalIndexOptions {\n limit?: number\n offset?: number\n setSizeCallback?: (getSize: () => number) => void\n setWindowFn?: (\n windowFn: (options: { offset?: number; limit?: number }) => void\n ) => void\n}\n\nexport type TopKChanges<V> = {\n /** Indicates which element moves into the topK (if any) */\n moveIn: IndexedValue<V> | null\n /** Indicates which element moves out of the topK (if any) */\n moveOut: IndexedValue<V> | null\n}\n\nexport type TopKMoveChanges<V> = {\n /** Flag that marks whether there were any changes to the topK */\n changes: boolean\n /** Indicates which elements move into the topK (if any) */\n moveIns: Array<IndexedValue<V>>\n /** Indicates which elements move out of the topK (if any) */\n moveOuts: Array<IndexedValue<V>>\n}\n\n/**\n * A topK data structure that supports insertions and deletions\n * and returns changes to the topK.\n */\nexport interface TopK<V> {\n size: number\n insert: (value: V) => TopKChanges<V>\n delete: (value: V) => TopKChanges<V>\n}\n\n/**\n * Implementation of a topK data structure.\n * Uses a sorted array internally to store the values and keeps a topK window over that array.\n * Inserts and deletes are O(n) operations because worst case an element is inserted/deleted\n * at the start of the array which causes all the elements to shift to the right/left.\n */\nclass TopKArray<V> implements TopK<V> {\n #sortedValues: Array<IndexedValue<V>> = []\n #comparator: (a: V, b: V) => number\n #topKStart: number\n #topKEnd: number\n\n constructor(\n offset: number,\n limit: number,\n comparator: (a: V, b: V) => number\n ) {\n this.#topKStart = offset\n this.#topKEnd = offset + limit\n this.#comparator = comparator\n }\n\n get size(): number {\n const offset = this.#topKStart\n const limit = this.#topKEnd - this.#topKStart\n const available = this.#sortedValues.length - offset\n return Math.max(0, Math.min(limit, available))\n }\n\n /**\n * Moves the topK window\n */\n move({\n offset,\n limit,\n }: {\n offset?: number\n limit?: number\n }): TopKMoveChanges<V> {\n const oldOffset = this.#topKStart\n const oldLimit = this.#topKEnd - this.#topKStart\n const oldRange: HRange = [this.#topKStart, this.#topKEnd]\n\n this.#topKStart = offset ?? oldOffset\n this.#topKEnd = this.#topKStart + (limit ?? oldLimit)\n\n const newRange: HRange = [this.#topKStart, this.#topKEnd]\n const { onlyInA, onlyInB } = diffHalfOpen(oldRange, newRange)\n\n const moveIns: Array<IndexedValue<V>> = []\n onlyInB.forEach((index) => {\n const value = this.#sortedValues[index]\n if (value) {\n moveIns.push(value)\n }\n })\n\n const moveOuts: Array<IndexedValue<V>> = []\n onlyInA.forEach((index) => {\n const value = this.#sortedValues[index]\n if (value) {\n moveOuts.push(value)\n }\n })\n\n // It could be that there are changes (i.e. moveIns or moveOuts)\n // but that the collection is lazy so we don't have the data yet that needs to move in/out\n // so `moveIns` and `moveOuts` will be empty but `changes` will be true\n // this will tell the caller that it needs to run the graph to load more data\n return { moveIns, moveOuts, changes: onlyInA.length + onlyInB.length > 0 }\n }\n\n insert(value: V): TopKChanges<V> {\n const result: TopKChanges<V> = { moveIn: null, moveOut: null }\n\n // Lookup insert position\n const index = this.#findIndex(value)\n // Generate fractional index based on the fractional indices of the elements before and after it\n const indexBefore =\n index === 0 ? null : getIndex(this.#sortedValues[index - 1]!)\n const indexAfter =\n index === this.#sortedValues.length\n ? null\n : getIndex(this.#sortedValues[index]!)\n const fractionalIndex = generateKeyBetween(indexBefore, indexAfter)\n\n // Insert the value at the correct position\n const val = indexedValue(value, fractionalIndex)\n // Splice is O(n) where n = all elements in the collection (i.e. n >= k) !\n this.#sortedValues.splice(index, 0, val)\n\n // Check if the topK changed\n if (index < this.#topKEnd) {\n // The inserted element is either before the top K or within the top K\n // If it is before the top K then it moves the element that was right before the topK into the topK\n // If it is within the top K then the inserted element moves into the top K\n // In both cases the last element of the old top K now moves out of the top K\n const moveInIndex = Math.max(index, this.#topKStart)\n if (moveInIndex < this.#sortedValues.length) {\n // We actually have a topK\n // because in some cases there may not be enough elements in the array to reach the start of the topK\n // e.g. [1, 2, 3] with K = 2 and offset = 3 does not have a topK\n result.moveIn = this.#sortedValues[moveInIndex]!\n\n // We need to remove the element that falls out of the top K\n // The element that falls out of the top K has shifted one to the right\n // because of the element we inserted, so we find it at index topKEnd\n if (this.#topKEnd < this.#sortedValues.length) {\n result.moveOut = this.#sortedValues[this.#topKEnd]!\n }\n }\n }\n\n return result\n }\n\n /**\n * Deletes a value that may or may not be in the topK.\n * IMPORTANT: this assumes that the value is present in the collection\n * if it's not the case it will remove the element\n * that is on the position where the provided `value` would be.\n */\n delete(value: V): TopKChanges<V> {\n const result: TopKChanges<V> = { moveIn: null, moveOut: null }\n\n // Lookup delete position\n const index = this.#findIndex(value)\n // Remove the value at that position\n const [removedElem] = this.#sortedValues.splice(index, 1)\n\n // Check if the topK changed\n if (index < this.#topKEnd) {\n // The removed element is either before the top K or within the top K\n // If it is before the top K then the first element of the topK moves out of the topK\n // If it is within the top K then the removed element moves out of the topK\n result.moveOut = removedElem!\n if (index < this.#topKStart) {\n // The removed element is before the topK\n // so actually, the first element of the topK moves out of the topK\n // and not the element that we removed\n // The first element of the topK is now at index topKStart - 1\n // since we removed an element before the topK\n const moveOutIndex = this.#topKStart - 1\n if (moveOutIndex < this.#sortedValues.length) {\n result.moveOut = this.#sortedValues[moveOutIndex]!\n } else {\n // No value is moving out of the topK\n // because there are no elements in the topK\n result.moveOut = null\n }\n }\n\n // Since we removed an element that was before or in the topK\n // the first element after the topK moved one position to the left\n // and thus falls into the topK now\n const moveInIndex = this.#topKEnd - 1\n if (moveInIndex < this.#sortedValues.length) {\n result.moveIn = this.#sortedValues[moveInIndex]!\n }\n }\n\n return result\n }\n\n // TODO: see if there is a way to refactor the code for insert and delete in the topK above\n // because they are very similar, one is shifting the topK window to the left and the other is shifting it to the right\n // so i have the feeling there is a common pattern here and we can implement both cases using that pattern\n\n #findIndex(value: V): number {\n return binarySearch(this.#sortedValues, indexedValue(value, ``), (a, b) =>\n this.#comparator(getValue(a), getValue(b))\n )\n }\n}\n\n/**\n * Operator for fractional indexed topK operations\n * This operator maintains fractional indices for sorted elements\n * and only updates indices when elements move position\n */\nexport class TopKWithFractionalIndexOperator<K, T> extends UnaryOperator<\n [K, T],\n [K, IndexedValue<T>]\n> {\n #index: Map<K, number> = new Map() // maps keys to their multiplicity\n\n /**\n * topK data structure that supports insertions and deletions\n * and returns changes to the topK.\n */\n #topK: TopK<TaggedValue<K, T>>\n\n constructor(\n id: number,\n inputA: DifferenceStreamReader<[K, T]>,\n output: DifferenceStreamWriter<[K, IndexedValue<T>]>,\n comparator: (a: T, b: T) => number,\n options: TopKWithFractionalIndexOptions\n ) {\n super(id, inputA, output)\n const limit = options.limit ?? Infinity\n const offset = options.offset ?? 0\n const compareTaggedValues = (\n a: TaggedValue<K, T>,\n b: TaggedValue<K, T>\n ) => {\n // First compare on the value\n const valueComparison = comparator(getVal(a), getVal(b))\n if (valueComparison !== 0) {\n return valueComparison\n }\n // If the values are equal, compare on the tag (object identity)\n const tieBreakerA = getTag(a)\n const tieBreakerB = getTag(b)\n return tieBreakerA - tieBreakerB\n }\n this.#topK = this.createTopK(offset, limit, compareTaggedValues)\n options.setSizeCallback?.(() => this.#topK.size)\n options.setWindowFn?.(this.moveTopK.bind(this))\n }\n\n protected createTopK(\n offset: number,\n limit: number,\n comparator: (a: TaggedValue<K, T>, b: TaggedValue<K, T>) => number\n ): TopK<TaggedValue<K, T>> {\n return new TopKArray(offset, limit, comparator)\n }\n\n /**\n * Moves the topK window based on the provided offset and limit.\n * Any changes to the topK are sent to the output.\n */\n moveTopK({ offset, limit }: { offset?: number; limit?: number }) {\n if (!(this.#topK instanceof TopKArray)) {\n throw new Error(\n `Cannot move B+-tree implementation of TopK with fractional index`\n )\n }\n\n const result: Array<[[K, IndexedValue<T>], number]> = []\n\n const diff = this.#topK.move({ offset, limit })\n\n diff.moveIns.forEach((moveIn) => this.handleMoveIn(moveIn, result))\n diff.moveOuts.forEach((moveOut) => this.handleMoveOut(moveOut, result))\n\n if (diff.changes) {\n // There are changes to the topK\n // it could be that moveIns and moveOuts are empty\n // because the collection is lazy, so we will run the graph again to load the data\n this.output.sendData(new MultiSet(result))\n }\n }\n\n run(): void {\n const result: Array<[[K, IndexedValue<T>], number]> = []\n for (const message of this.inputMessages()) {\n for (const [item, multiplicity] of message.getInner()) {\n const [key, value] = item\n this.processElement(key, value, multiplicity, result)\n }\n }\n\n if (result.length > 0) {\n this.output.sendData(new MultiSet(result))\n }\n }\n\n processElement(\n key: K,\n value: T,\n multiplicity: number,\n result: Array<[[K, IndexedValue<T>], number]>\n ): void {\n const { oldMultiplicity, newMultiplicity } = this.addKey(key, multiplicity)\n\n let res: TopKChanges<TaggedValue<K, T>> = {\n moveIn: null,\n moveOut: null,\n }\n if (oldMultiplicity <= 0 && newMultiplicity > 0) {\n // The value was invisible but should now be visible\n // Need to insert it into the array of sorted values\n const taggedValue = tagValue(key, value)\n res = this.#topK.insert(taggedValue)\n } else if (oldMultiplicity > 0 && newMultiplicity <= 0) {\n // The value was visible but should now be invisible\n // Need to remove it from the array of sorted values\n const taggedValue = tagValue(key, value)\n res = this.#topK.delete(taggedValue)\n } else {\n // The value was invisible and it remains invisible\n // or it was visible and remains visible\n // so it doesn't affect the topK\n }\n\n this.handleMoveIn(res.moveIn, result)\n this.handleMoveOut(res.moveOut, result)\n\n return\n }\n\n private handleMoveIn(\n moveIn: IndexedValue<TaggedValue<K, T>> | null,\n result: Array<[[K, IndexedValue<T>], number]>\n ) {\n if (moveIn) {\n const index = getIndex(moveIn)\n const taggedValue = getValue(moveIn)\n const k = getKey(taggedValue)\n const val = getVal(taggedValue)\n result.push([[k, [val, index]], 1])\n }\n }\n\n private handleMoveOut(\n moveOut: IndexedValue<TaggedValue<K, T>> | null,\n result: Array<[[K, IndexedValue<T>], number]>\n ) {\n if (moveOut) {\n const index = getIndex(moveOut)\n const taggedValue = getValue(moveOut)\n const k = getKey(taggedValue)\n const val = getVal(taggedValue)\n result.push([[k, [val, index]], -1])\n }\n }\n\n private getMultiplicity(key: K): number {\n return this.#index.get(key) ?? 0\n }\n\n private addKey(\n key: K,\n multiplicity: number\n ): { oldMultiplicity: number; newMultiplicity: number } {\n const oldMultiplicity = this.getMultiplicity(key)\n const newMultiplicity = oldMultiplicity + multiplicity\n if (newMultiplicity === 0) {\n this.#index.delete(key)\n } else {\n this.#index.set(key, newMultiplicity)\n }\n return { oldMultiplicity, newMultiplicity }\n }\n}\n\n/**\n * Limits the number of results based on a comparator, with optional offset.\n * Uses fractional indexing to minimize the number of changes when elements move positions.\n * Each element is assigned a fractional index that is lexicographically sortable.\n * When elements move, only the indices of the moved elements are updated, not all elements.\n *\n * @param comparator - A function that compares two elements\n * @param options - An optional object containing limit and offset properties\n * @returns A piped operator that orders the elements and limits the number of results\n */\nexport function topKWithFractionalIndex<KType, T>(\n comparator: (a: T, b: T) => number,\n options?: TopKWithFractionalIndexOptions\n): PipedOperator<[KType, T], [KType, IndexedValue<T>]> {\n const opts = options || {}\n\n return (\n stream: IStreamBuilder<[KType, T]>\n ): IStreamBuilder<[KType, IndexedValue<T>]> => {\n const output = new StreamBuilder<[KType, IndexedValue<T>]>(\n stream.graph,\n new DifferenceStreamWriter<[KType, IndexedValue<T>]>()\n )\n const operator = new TopKWithFractionalIndexOperator<KType, T>(\n stream.graph.getNextOperatorId(),\n stream.connectReader(),\n output.writer,\n comparator,\n opts\n )\n stream.graph.addOperator(operator)\n return output\n }\n}\n\n// Abstraction for fractionally indexed values\nexport type FractionalIndex = string\nexport type IndexedValue<V> = [V, FractionalIndex]\n\nexport function indexedValue<V>(\n value: V,\n index: FractionalIndex\n): IndexedValue<V> {\n return [value, index]\n}\n\nexport function getValue<V>(indexedVal: IndexedValue<V>): V {\n return indexedVal[0]\n}\n\nexport function getIndex<V>(indexedVal: IndexedValue<V>): FractionalIndex {\n return indexedVal[1]\n}\n\nexport type Tag = number\nexport type TaggedValue<K, V> = [K, V, Tag]\n\nfunction tagValue<K, V>(key: K, value: V): TaggedValue<K, V> {\n return [key, value, globalObjectIdGenerator.getId(key)]\n}\n\nfunction getKey<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): K {\n return tieBreakerTaggedValue[0]\n}\n\nfunction getVal<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): V {\n return tieBreakerTaggedValue[1]\n}\n\nfunction getTag<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): Tag {\n return tieBreakerTaggedValue[2]\n}\n"],"names":[],"mappings":";;;;;AAsDA,MAAM,UAAgC;AAAA,EACpC,gBAAwC,CAAA;AAAA,EACxC;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YACE,QACA,OACA,YACA;AACA,SAAK,aAAa;AAClB,SAAK,WAAW,SAAS;AACzB,SAAK,cAAc;AAAA,EACrB;AAAA,EAEA,IAAI,OAAe;AACjB,UAAM,SAAS,KAAK;AACpB,UAAM,QAAQ,KAAK,WAAW,KAAK;AACnC,UAAM,YAAY,KAAK,cAAc,SAAS;AAC9C,WAAO,KAAK,IAAI,GAAG,KAAK,IAAI,OAAO,SAAS,CAAC;AAAA,EAC/C;AAAA;AAAA;AAAA;AAAA,EAKA,KAAK;AAAA,IACH;AAAA,IACA;AAAA,EAAA,GAIqB;AACrB,UAAM,YAAY,KAAK;AACvB,UAAM,WAAW,KAAK,WAAW,KAAK;AACtC,UAAM,WAAmB,CAAC,KAAK,YAAY,KAAK,QAAQ;AAExD,SAAK,aAAa,UAAU;AAC5B,SAAK,WAAW,KAAK,cAAc,SAAS;AAE5C,UAAM,WAAmB,CAAC,KAAK,YAAY,KAAK,QAAQ;AACxD,UAAM,EAAE,SAAS,QAAA,IAAY,aAAa,UAAU,QAAQ;AAE5D,UAAM,UAAkC,CAAA;AACxC,YAAQ,QAAQ,CAAC,UAAU;AACzB,YAAM,QAAQ,KAAK,cAAc,KAAK;AACtC,UAAI,OAAO;AACT,gBAAQ,KAAK,KAAK;AAAA,MACpB;AAAA,IACF,CAAC;AAED,UAAM,WAAmC,CAAA;AACzC,YAAQ,QAAQ,CAAC,UAAU;AACzB,YAAM,QAAQ,KAAK,cAAc,KAAK;AACtC,UAAI,OAAO;AACT,iBAAS,KAAK,KAAK;AAAA,MACrB;AAAA,IACF,CAAC;AAMD,WAAO,EAAE,SAAS,UAAU,SAAS,QAAQ,SAAS,QAAQ,SAAS,EAAA;AAAA,EACzE;AAAA,EAEA,OAAO,OAA0B;AAC/B,UAAM,SAAyB,EAAE,QAAQ,MAAM,SAAS,KAAA;AAGxD,UAAM,QAAQ,KAAK,WAAW,KAAK;AAEnC,UAAM,cACJ,UAAU,IAAI,OAAO,SAAS,KAAK,cAAc,QAAQ,CAAC,CAAE;AAC9D,UAAM,aACJ,UAAU,KAAK,cAAc,SACzB,OACA,SAAS,KAAK,cAAc,KAAK,CAAE;AACzC,UAAM,kBAAkB,mBAAmB,aAAa,UAAU;AAGlE,UAAM,MAAM,aAAa,OAAO,eAAe;AAE/C,SAAK,cAAc,OAAO,OAAO,GAAG,GAAG;AAGvC,QAAI,QAAQ,KAAK,UAAU;AAKzB,YAAM,cAAc,KAAK,IAAI,OAAO,KAAK,UAAU;AACnD,UAAI,cAAc,KAAK,cAAc,QAAQ;AAI3C,eAAO,SAAS,KAAK,cAAc,WAAW;AAK9C,YAAI,KAAK,WAAW,KAAK,cAAc,QAAQ;AAC7C,iBAAO,UAAU,KAAK,cAAc,KAAK,QAAQ;AAAA,QACnD;AAAA,MACF;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,OAAO,OAA0B;AAC/B,UAAM,SAAyB,EAAE,QAAQ,MAAM,SAAS,KAAA;AAGxD,UAAM,QAAQ,KAAK,WAAW,KAAK;AAEnC,UAAM,CAAC,WAAW,IAAI,KAAK,cAAc,OAAO,OAAO,CAAC;AAGxD,QAAI,QAAQ,KAAK,UAAU;AAIzB,aAAO,UAAU;AACjB,UAAI,QAAQ,KAAK,YAAY;AAM3B,cAAM,eAAe,KAAK,aAAa;AACvC,YAAI,eAAe,KAAK,cAAc,QAAQ;AAC5C,iBAAO,UAAU,KAAK,cAAc,YAAY;AAAA,QAClD,OAAO;AAGL,iBAAO,UAAU;AAAA,QACnB;AAAA,MACF;AAKA,YAAM,cAAc,KAAK,WAAW;AACpC,UAAI,cAAc,KAAK,cAAc,QAAQ;AAC3C,eAAO,SAAS,KAAK,cAAc,WAAW;AAAA,MAChD;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA,EAMA,WAAW,OAAkB;AAC3B,WAAO;AAAA,MAAa,KAAK;AAAA,MAAe,aAAa,OAAO,EAAE;AAAA,MAAG,CAAC,GAAG,MACnE,KAAK,YAAY,SAAS,CAAC,GAAG,SAAS,CAAC,CAAC;AAAA,IAAA;AAAA,EAE7C;AACF;AAOO,MAAM,wCAA8C,cAGzD;AAAA,EACA,6BAA6B,IAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAM7B;AAAA,EAEA,YACE,IACA,QACA,QACA,YACA,SACA;AACA,UAAM,IAAI,QAAQ,MAAM;AACxB,UAAM,QAAQ,QAAQ,SAAS;AAC/B,UAAM,SAAS,QAAQ,UAAU;AACjC,UAAM,sBAAsB,CAC1B,GACA,MACG;AAEH,YAAM,kBAAkB,WAAW,OAAO,CAAC,GAAG,OAAO,CAAC,CAAC;AACvD,UAAI,oBAAoB,GAAG;AACzB,eAAO;AAAA,MACT;AAEA,YAAM,cAAc,OAAO,CAAC;AAC5B,YAAM,cAAc,OAAO,CAAC;AAC5B,aAAO,cAAc;AAAA,IACvB;AACA,SAAK,QAAQ,KAAK,WAAW,QAAQ,OAAO,mBAAmB;AAC/D,YAAQ,kBAAkB,MAAM,KAAK,MAAM,IAAI;AAC/C,YAAQ,cAAc,KAAK,SAAS,KAAK,IAAI,CAAC;AAAA,EAChD;AAAA,EAEU,WACR,QACA,OACA,YACyB;AACzB,WAAO,IAAI,UAAU,QAAQ,OAAO,UAAU;AAAA,EAChD;AAAA;AAAA;AAAA;AAAA;AAAA,EAMA,SAAS,EAAE,QAAQ,SAA8C;AAC/D,QAAI,EAAE,KAAK,iBAAiB,YAAY;AACtC,YAAM,IAAI;AAAA,QACR;AAAA,MAAA;AAAA,IAEJ;AAEA,UAAM,SAAgD,CAAA;AAEtD,UAAM,OAAO,KAAK,MAAM,KAAK,EAAE,QAAQ,OAAO;AAE9C,SAAK,QAAQ,QAAQ,CAAC,WAAW,KAAK,aAAa,QAAQ,MAAM,CAAC;AAClE,SAAK,SAAS,QAAQ,CAAC,YAAY,KAAK,cAAc,SAAS,MAAM,CAAC;AAEtE,QAAI,KAAK,SAAS;AAIhB,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,MAAY;AACV,UAAM,SAAgD,CAAA;AACtD,eAAW,WAAW,KAAK,iBAAiB;AAC1C,iBAAW,CAAC,MAAM,YAAY,KAAK,QAAQ,YAAY;AACrD,cAAM,CAAC,KAAK,KAAK,IAAI;AACrB,aAAK,eAAe,KAAK,OAAO,cAAc,MAAM;AAAA,MACtD;AAAA,IACF;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,eACE,KACA,OACA,cACA,QACM;AACN,UAAM,EAAE,iBAAiB,gBAAA,IAAoB,KAAK,OAAO,KAAK,YAAY;AAE1E,QAAI,MAAsC;AAAA,MACxC,QAAQ;AAAA,MACR,SAAS;AAAA,IAAA;AAEX,QAAI,mBAAmB,KAAK,kBAAkB,GAAG;AAG/C,YAAM,cAAc,SAAS,KAAK,KAAK;AACvC,YAAM,KAAK,MAAM,OAAO,WAAW;AAAA,IACrC,WAAW,kBAAkB,KAAK,mBAAmB,GAAG;AAGtD,YAAM,cAAc,SAAS,KAAK,KAAK;AACvC,YAAM,KAAK,MAAM,OAAO,WAAW;AAAA,IACrC,MAAO;AAMP,SAAK,aAAa,IAAI,QAAQ,MAAM;AACpC,SAAK,cAAc,IAAI,SAAS,MAAM;AAEtC;AAAA,EACF;AAAA,EAEQ,aACN,QACA,QACA;AACA,QAAI,QAAQ;AACV,YAAM,QAAQ,SAAS,MAAM;AAC7B,YAAM,cAAc,SAAS,MAAM;AACnC,YAAM,IAAI,OAAO,WAAW;AAC5B,YAAM,MAAM,OAAO,WAAW;AAC9B,aAAO,KAAK,CAAC,CAAC,GAAG,CAAC,KAAK,KAAK,CAAC,GAAG,CAAC,CAAC;AAAA,IACpC;AAAA,EACF;AAAA,EAEQ,cACN,SACA,QACA;AACA,QAAI,SAAS;AACX,YAAM,QAAQ,SAAS,OAAO;AAC9B,YAAM,cAAc,SAAS,OAAO;AACpC,YAAM,IAAI,OAAO,WAAW;AAC5B,YAAM,MAAM,OAAO,WAAW;AAC9B,aAAO,KAAK,CAAC,CAAC,GAAG,CAAC,KAAK,KAAK,CAAC,GAAG,EAAE,CAAC;AAAA,IACrC;AAAA,EACF;AAAA,EAEQ,gBAAgB,KAAgB;AACtC,WAAO,KAAK,OAAO,IAAI,GAAG,KAAK;AAAA,EACjC;AAAA,EAEQ,OACN,KACA,cACsD;AACtD,UAAM,kBAAkB,KAAK,gBAAgB,GAAG;AAChD,UAAM,kBAAkB,kBAAkB;AAC1C,QAAI,oBAAoB,GAAG;AACzB,WAAK,OAAO,OAAO,GAAG;AAAA,IACxB,OAAO;AACL,WAAK,OAAO,IAAI,KAAK,eAAe;AAAA,IACtC;AACA,WAAO,EAAE,iBAAiB,gBAAA;AAAA,EAC5B;AACF;AAYO,SAAS,wBACd,YACA,SACqD;AACrD,QAAM,OAAO,WAAW,CAAA;AAExB,SAAO,CACL,WAC6C;AAC7C,UAAM,SAAS,IAAI;AAAA,MACjB,OAAO;AAAA,MACP,IAAI,uBAAA;AAAA,IAAiD;AAEvD,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,OAAO;AAAA,MACP;AAAA,MACA;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;AAMO,SAAS,aACd,OACA,OACiB;AACjB,SAAO,CAAC,OAAO,KAAK;AACtB;AAEO,SAAS,SAAY,YAAgC;AAC1D,SAAO,WAAW,CAAC;AACrB;AAEO,SAAS,SAAY,YAA8C;AACxE,SAAO,WAAW,CAAC;AACrB;AAKA,SAAS,SAAe,KAAQ,OAA6B;AAC3D,SAAO,CAAC,KAAK,OAAO,wBAAwB,MAAM,GAAG,CAAC;AACxD;AAEA,SAAS,OAAa,uBAA6C;AACjE,SAAO,sBAAsB,CAAC;AAChC;AAEA,SAAS,OAAa,uBAA6C;AACjE,SAAO,sBAAsB,CAAC;AAChC;AAEA,SAAS,OAAa,uBAA+C;AACnE,SAAO,sBAAsB,CAAC;AAChC;"}
|
|
1
|
+
{"version":3,"file":"topKWithFractionalIndex.js","sources":["../../../src/operators/topKWithFractionalIndex.ts"],"sourcesContent":["import { generateKeyBetween } from \"fractional-indexing\"\nimport { DifferenceStreamWriter, UnaryOperator } from \"../graph.js\"\nimport { StreamBuilder } from \"../d2.js\"\nimport { MultiSet } from \"../multiset.js\"\nimport {\n binarySearch,\n diffHalfOpen,\n globalObjectIdGenerator,\n} from \"../utils.js\"\nimport type { HRange } from \"../utils.js\"\nimport type { DifferenceStreamReader } from \"../graph.js\"\nimport type { IStreamBuilder, PipedOperator } from \"../types.js\"\n\nexport interface TopKWithFractionalIndexOptions {\n limit?: number\n offset?: number\n setSizeCallback?: (getSize: () => number) => void\n setWindowFn?: (\n windowFn: (options: { offset?: number; limit?: number }) => void\n ) => void\n}\n\nexport type TopKChanges<V> = {\n /** Indicates which element moves into the topK (if any) */\n moveIn: IndexedValue<V> | null\n /** Indicates which element moves out of the topK (if any) */\n moveOut: IndexedValue<V> | null\n}\n\nexport type TopKMoveChanges<V> = {\n /** Flag that marks whether there were any changes to the topK */\n changes: boolean\n /** Indicates which elements move into the topK (if any) */\n moveIns: Array<IndexedValue<V>>\n /** Indicates which elements move out of the topK (if any) */\n moveOuts: Array<IndexedValue<V>>\n}\n\n/**\n * A topK data structure that supports insertions and deletions\n * and returns changes to the topK.\n */\nexport interface TopK<V> {\n size: number\n insert: (value: V) => TopKChanges<V>\n delete: (value: V) => TopKChanges<V>\n}\n\n/**\n * Implementation of a topK data structure.\n * Uses a sorted array internally to store the values and keeps a topK window over that array.\n * Inserts and deletes are O(n) operations because worst case an element is inserted/deleted\n * at the start of the array which causes all the elements to shift to the right/left.\n */\nclass TopKArray<V> implements TopK<V> {\n #sortedValues: Array<IndexedValue<V>> = []\n #comparator: (a: V, b: V) => number\n #topKStart: number\n #topKEnd: number\n\n constructor(\n offset: number,\n limit: number,\n comparator: (a: V, b: V) => number\n ) {\n this.#topKStart = offset\n this.#topKEnd = offset + limit\n this.#comparator = comparator\n }\n\n get size(): number {\n const offset = this.#topKStart\n const limit = this.#topKEnd - this.#topKStart\n const available = this.#sortedValues.length - offset\n return Math.max(0, Math.min(limit, available))\n }\n\n /**\n * Moves the topK window\n */\n move({\n offset,\n limit,\n }: {\n offset?: number\n limit?: number\n }): TopKMoveChanges<V> {\n const oldOffset = this.#topKStart\n const oldLimit = this.#topKEnd - this.#topKStart\n\n // `this.#topKEnd` can be `Infinity` if it has no limit\n // but `diffHalfOpen` expects a finite range\n // so we restrict it to the size of the topK if topKEnd is infinite\n const oldRange: HRange = [\n this.#topKStart,\n this.#topKEnd === Infinity ? this.#topKStart + this.size : this.#topKEnd,\n ]\n\n this.#topKStart = offset ?? oldOffset\n this.#topKEnd = this.#topKStart + (limit ?? oldLimit) // can be `Infinity` if limit is `Infinity`\n\n // Also handle `Infinity` in the newRange\n const newRange: HRange = [\n this.#topKStart,\n this.#topKEnd === Infinity\n ? Math.max(this.#topKStart + this.size, oldRange[1]) // since the new limit is Infinity we need to take everything (so we need to take the biggest (finite) topKEnd)\n : this.#topKEnd,\n ]\n const { onlyInA, onlyInB } = diffHalfOpen(oldRange, newRange)\n\n const moveIns: Array<IndexedValue<V>> = []\n onlyInB.forEach((index) => {\n const value = this.#sortedValues[index]\n if (value) {\n moveIns.push(value)\n }\n })\n\n const moveOuts: Array<IndexedValue<V>> = []\n onlyInA.forEach((index) => {\n const value = this.#sortedValues[index]\n if (value) {\n moveOuts.push(value)\n }\n })\n\n // It could be that there are changes (i.e. moveIns or moveOuts)\n // but that the collection is lazy so we don't have the data yet that needs to move in/out\n // so `moveIns` and `moveOuts` will be empty but `changes` will be true\n // this will tell the caller that it needs to run the graph to load more data\n return { moveIns, moveOuts, changes: onlyInA.length + onlyInB.length > 0 }\n }\n\n insert(value: V): TopKChanges<V> {\n const result: TopKChanges<V> = { moveIn: null, moveOut: null }\n\n // Lookup insert position\n const index = this.#findIndex(value)\n // Generate fractional index based on the fractional indices of the elements before and after it\n const indexBefore =\n index === 0 ? null : getIndex(this.#sortedValues[index - 1]!)\n const indexAfter =\n index === this.#sortedValues.length\n ? null\n : getIndex(this.#sortedValues[index]!)\n const fractionalIndex = generateKeyBetween(indexBefore, indexAfter)\n\n // Insert the value at the correct position\n const val = indexedValue(value, fractionalIndex)\n // Splice is O(n) where n = all elements in the collection (i.e. n >= k) !\n this.#sortedValues.splice(index, 0, val)\n\n // Check if the topK changed\n if (index < this.#topKEnd) {\n // The inserted element is either before the top K or within the top K\n // If it is before the top K then it moves the element that was right before the topK into the topK\n // If it is within the top K then the inserted element moves into the top K\n // In both cases the last element of the old top K now moves out of the top K\n const moveInIndex = Math.max(index, this.#topKStart)\n if (moveInIndex < this.#sortedValues.length) {\n // We actually have a topK\n // because in some cases there may not be enough elements in the array to reach the start of the topK\n // e.g. [1, 2, 3] with K = 2 and offset = 3 does not have a topK\n result.moveIn = this.#sortedValues[moveInIndex]!\n\n // We need to remove the element that falls out of the top K\n // The element that falls out of the top K has shifted one to the right\n // because of the element we inserted, so we find it at index topKEnd\n if (this.#topKEnd < this.#sortedValues.length) {\n result.moveOut = this.#sortedValues[this.#topKEnd]!\n }\n }\n }\n\n return result\n }\n\n /**\n * Deletes a value that may or may not be in the topK.\n * IMPORTANT: this assumes that the value is present in the collection\n * if it's not the case it will remove the element\n * that is on the position where the provided `value` would be.\n */\n delete(value: V): TopKChanges<V> {\n const result: TopKChanges<V> = { moveIn: null, moveOut: null }\n\n // Lookup delete position\n const index = this.#findIndex(value)\n // Remove the value at that position\n const [removedElem] = this.#sortedValues.splice(index, 1)\n\n // Check if the topK changed\n if (index < this.#topKEnd) {\n // The removed element is either before the top K or within the top K\n // If it is before the top K then the first element of the topK moves out of the topK\n // If it is within the top K then the removed element moves out of the topK\n result.moveOut = removedElem!\n if (index < this.#topKStart) {\n // The removed element is before the topK\n // so actually, the first element of the topK moves out of the topK\n // and not the element that we removed\n // The first element of the topK is now at index topKStart - 1\n // since we removed an element before the topK\n const moveOutIndex = this.#topKStart - 1\n if (moveOutIndex < this.#sortedValues.length) {\n result.moveOut = this.#sortedValues[moveOutIndex]!\n } else {\n // No value is moving out of the topK\n // because there are no elements in the topK\n result.moveOut = null\n }\n }\n\n // Since we removed an element that was before or in the topK\n // the first element after the topK moved one position to the left\n // and thus falls into the topK now\n const moveInIndex = this.#topKEnd - 1\n if (moveInIndex < this.#sortedValues.length) {\n result.moveIn = this.#sortedValues[moveInIndex]!\n }\n }\n\n return result\n }\n\n // TODO: see if there is a way to refactor the code for insert and delete in the topK above\n // because they are very similar, one is shifting the topK window to the left and the other is shifting it to the right\n // so i have the feeling there is a common pattern here and we can implement both cases using that pattern\n\n #findIndex(value: V): number {\n return binarySearch(this.#sortedValues, indexedValue(value, ``), (a, b) =>\n this.#comparator(getValue(a), getValue(b))\n )\n }\n}\n\n/**\n * Operator for fractional indexed topK operations\n * This operator maintains fractional indices for sorted elements\n * and only updates indices when elements move position\n */\nexport class TopKWithFractionalIndexOperator<K, T> extends UnaryOperator<\n [K, T],\n [K, IndexedValue<T>]\n> {\n #index: Map<K, number> = new Map() // maps keys to their multiplicity\n\n /**\n * topK data structure that supports insertions and deletions\n * and returns changes to the topK.\n */\n #topK: TopK<TaggedValue<K, T>>\n\n constructor(\n id: number,\n inputA: DifferenceStreamReader<[K, T]>,\n output: DifferenceStreamWriter<[K, IndexedValue<T>]>,\n comparator: (a: T, b: T) => number,\n options: TopKWithFractionalIndexOptions\n ) {\n super(id, inputA, output)\n const limit = options.limit ?? Infinity\n const offset = options.offset ?? 0\n const compareTaggedValues = (\n a: TaggedValue<K, T>,\n b: TaggedValue<K, T>\n ) => {\n // First compare on the value\n const valueComparison = comparator(getVal(a), getVal(b))\n if (valueComparison !== 0) {\n return valueComparison\n }\n // If the values are equal, compare on the tag (object identity)\n const tieBreakerA = getTag(a)\n const tieBreakerB = getTag(b)\n return tieBreakerA - tieBreakerB\n }\n this.#topK = this.createTopK(offset, limit, compareTaggedValues)\n options.setSizeCallback?.(() => this.#topK.size)\n options.setWindowFn?.(this.moveTopK.bind(this))\n }\n\n protected createTopK(\n offset: number,\n limit: number,\n comparator: (a: TaggedValue<K, T>, b: TaggedValue<K, T>) => number\n ): TopK<TaggedValue<K, T>> {\n return new TopKArray(offset, limit, comparator)\n }\n\n /**\n * Moves the topK window based on the provided offset and limit.\n * Any changes to the topK are sent to the output.\n */\n moveTopK({ offset, limit }: { offset?: number; limit?: number }) {\n if (!(this.#topK instanceof TopKArray)) {\n throw new Error(\n `Cannot move B+-tree implementation of TopK with fractional index`\n )\n }\n\n const result: Array<[[K, IndexedValue<T>], number]> = []\n\n const diff = this.#topK.move({ offset, limit })\n\n diff.moveIns.forEach((moveIn) => this.handleMoveIn(moveIn, result))\n diff.moveOuts.forEach((moveOut) => this.handleMoveOut(moveOut, result))\n\n if (diff.changes) {\n // There are changes to the topK\n // it could be that moveIns and moveOuts are empty\n // because the collection is lazy, so we will run the graph again to load the data\n this.output.sendData(new MultiSet(result))\n }\n }\n\n run(): void {\n const result: Array<[[K, IndexedValue<T>], number]> = []\n for (const message of this.inputMessages()) {\n for (const [item, multiplicity] of message.getInner()) {\n const [key, value] = item\n this.processElement(key, value, multiplicity, result)\n }\n }\n\n if (result.length > 0) {\n this.output.sendData(new MultiSet(result))\n }\n }\n\n processElement(\n key: K,\n value: T,\n multiplicity: number,\n result: Array<[[K, IndexedValue<T>], number]>\n ): void {\n const { oldMultiplicity, newMultiplicity } = this.addKey(key, multiplicity)\n\n let res: TopKChanges<TaggedValue<K, T>> = {\n moveIn: null,\n moveOut: null,\n }\n if (oldMultiplicity <= 0 && newMultiplicity > 0) {\n // The value was invisible but should now be visible\n // Need to insert it into the array of sorted values\n const taggedValue = tagValue(key, value)\n res = this.#topK.insert(taggedValue)\n } else if (oldMultiplicity > 0 && newMultiplicity <= 0) {\n // The value was visible but should now be invisible\n // Need to remove it from the array of sorted values\n const taggedValue = tagValue(key, value)\n res = this.#topK.delete(taggedValue)\n } else {\n // The value was invisible and it remains invisible\n // or it was visible and remains visible\n // so it doesn't affect the topK\n }\n\n this.handleMoveIn(res.moveIn, result)\n this.handleMoveOut(res.moveOut, result)\n\n return\n }\n\n private handleMoveIn(\n moveIn: IndexedValue<TaggedValue<K, T>> | null,\n result: Array<[[K, IndexedValue<T>], number]>\n ) {\n if (moveIn) {\n const index = getIndex(moveIn)\n const taggedValue = getValue(moveIn)\n const k = getKey(taggedValue)\n const val = getVal(taggedValue)\n result.push([[k, [val, index]], 1])\n }\n }\n\n private handleMoveOut(\n moveOut: IndexedValue<TaggedValue<K, T>> | null,\n result: Array<[[K, IndexedValue<T>], number]>\n ) {\n if (moveOut) {\n const index = getIndex(moveOut)\n const taggedValue = getValue(moveOut)\n const k = getKey(taggedValue)\n const val = getVal(taggedValue)\n result.push([[k, [val, index]], -1])\n }\n }\n\n private getMultiplicity(key: K): number {\n return this.#index.get(key) ?? 0\n }\n\n private addKey(\n key: K,\n multiplicity: number\n ): { oldMultiplicity: number; newMultiplicity: number } {\n const oldMultiplicity = this.getMultiplicity(key)\n const newMultiplicity = oldMultiplicity + multiplicity\n if (newMultiplicity === 0) {\n this.#index.delete(key)\n } else {\n this.#index.set(key, newMultiplicity)\n }\n return { oldMultiplicity, newMultiplicity }\n }\n}\n\n/**\n * Limits the number of results based on a comparator, with optional offset.\n * Uses fractional indexing to minimize the number of changes when elements move positions.\n * Each element is assigned a fractional index that is lexicographically sortable.\n * When elements move, only the indices of the moved elements are updated, not all elements.\n *\n * @param comparator - A function that compares two elements\n * @param options - An optional object containing limit and offset properties\n * @returns A piped operator that orders the elements and limits the number of results\n */\nexport function topKWithFractionalIndex<KType, T>(\n comparator: (a: T, b: T) => number,\n options?: TopKWithFractionalIndexOptions\n): PipedOperator<[KType, T], [KType, IndexedValue<T>]> {\n const opts = options || {}\n\n return (\n stream: IStreamBuilder<[KType, T]>\n ): IStreamBuilder<[KType, IndexedValue<T>]> => {\n const output = new StreamBuilder<[KType, IndexedValue<T>]>(\n stream.graph,\n new DifferenceStreamWriter<[KType, IndexedValue<T>]>()\n )\n const operator = new TopKWithFractionalIndexOperator<KType, T>(\n stream.graph.getNextOperatorId(),\n stream.connectReader(),\n output.writer,\n comparator,\n opts\n )\n stream.graph.addOperator(operator)\n return output\n }\n}\n\n// Abstraction for fractionally indexed values\nexport type FractionalIndex = string\nexport type IndexedValue<V> = [V, FractionalIndex]\n\nexport function indexedValue<V>(\n value: V,\n index: FractionalIndex\n): IndexedValue<V> {\n return [value, index]\n}\n\nexport function getValue<V>(indexedVal: IndexedValue<V>): V {\n return indexedVal[0]\n}\n\nexport function getIndex<V>(indexedVal: IndexedValue<V>): FractionalIndex {\n return indexedVal[1]\n}\n\nexport type Tag = number\nexport type TaggedValue<K, V> = [K, V, Tag]\n\nfunction tagValue<K, V>(key: K, value: V): TaggedValue<K, V> {\n return [key, value, globalObjectIdGenerator.getId(key)]\n}\n\nfunction getKey<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): K {\n return tieBreakerTaggedValue[0]\n}\n\nfunction getVal<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): V {\n return tieBreakerTaggedValue[1]\n}\n\nfunction getTag<K, V>(tieBreakerTaggedValue: TaggedValue<K, V>): Tag {\n return tieBreakerTaggedValue[2]\n}\n"],"names":[],"mappings":";;;;;AAsDA,MAAM,UAAgC;AAAA,EACpC,gBAAwC,CAAA;AAAA,EACxC;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YACE,QACA,OACA,YACA;AACA,SAAK,aAAa;AAClB,SAAK,WAAW,SAAS;AACzB,SAAK,cAAc;AAAA,EACrB;AAAA,EAEA,IAAI,OAAe;AACjB,UAAM,SAAS,KAAK;AACpB,UAAM,QAAQ,KAAK,WAAW,KAAK;AACnC,UAAM,YAAY,KAAK,cAAc,SAAS;AAC9C,WAAO,KAAK,IAAI,GAAG,KAAK,IAAI,OAAO,SAAS,CAAC;AAAA,EAC/C;AAAA;AAAA;AAAA;AAAA,EAKA,KAAK;AAAA,IACH;AAAA,IACA;AAAA,EAAA,GAIqB;AACrB,UAAM,YAAY,KAAK;AACvB,UAAM,WAAW,KAAK,WAAW,KAAK;AAKtC,UAAM,WAAmB;AAAA,MACvB,KAAK;AAAA,MACL,KAAK,aAAa,WAAW,KAAK,aAAa,KAAK,OAAO,KAAK;AAAA,IAAA;AAGlE,SAAK,aAAa,UAAU;AAC5B,SAAK,WAAW,KAAK,cAAc,SAAS;AAG5C,UAAM,WAAmB;AAAA,MACvB,KAAK;AAAA,MACL,KAAK,aAAa,WACd,KAAK,IAAI,KAAK,aAAa,KAAK,MAAM,SAAS,CAAC,CAAC,IACjD,KAAK;AAAA,IAAA;AAEX,UAAM,EAAE,SAAS,QAAA,IAAY,aAAa,UAAU,QAAQ;AAE5D,UAAM,UAAkC,CAAA;AACxC,YAAQ,QAAQ,CAAC,UAAU;AACzB,YAAM,QAAQ,KAAK,cAAc,KAAK;AACtC,UAAI,OAAO;AACT,gBAAQ,KAAK,KAAK;AAAA,MACpB;AAAA,IACF,CAAC;AAED,UAAM,WAAmC,CAAA;AACzC,YAAQ,QAAQ,CAAC,UAAU;AACzB,YAAM,QAAQ,KAAK,cAAc,KAAK;AACtC,UAAI,OAAO;AACT,iBAAS,KAAK,KAAK;AAAA,MACrB;AAAA,IACF,CAAC;AAMD,WAAO,EAAE,SAAS,UAAU,SAAS,QAAQ,SAAS,QAAQ,SAAS,EAAA;AAAA,EACzE;AAAA,EAEA,OAAO,OAA0B;AAC/B,UAAM,SAAyB,EAAE,QAAQ,MAAM,SAAS,KAAA;AAGxD,UAAM,QAAQ,KAAK,WAAW,KAAK;AAEnC,UAAM,cACJ,UAAU,IAAI,OAAO,SAAS,KAAK,cAAc,QAAQ,CAAC,CAAE;AAC9D,UAAM,aACJ,UAAU,KAAK,cAAc,SACzB,OACA,SAAS,KAAK,cAAc,KAAK,CAAE;AACzC,UAAM,kBAAkB,mBAAmB,aAAa,UAAU;AAGlE,UAAM,MAAM,aAAa,OAAO,eAAe;AAE/C,SAAK,cAAc,OAAO,OAAO,GAAG,GAAG;AAGvC,QAAI,QAAQ,KAAK,UAAU;AAKzB,YAAM,cAAc,KAAK,IAAI,OAAO,KAAK,UAAU;AACnD,UAAI,cAAc,KAAK,cAAc,QAAQ;AAI3C,eAAO,SAAS,KAAK,cAAc,WAAW;AAK9C,YAAI,KAAK,WAAW,KAAK,cAAc,QAAQ;AAC7C,iBAAO,UAAU,KAAK,cAAc,KAAK,QAAQ;AAAA,QACnD;AAAA,MACF;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,OAAO,OAA0B;AAC/B,UAAM,SAAyB,EAAE,QAAQ,MAAM,SAAS,KAAA;AAGxD,UAAM,QAAQ,KAAK,WAAW,KAAK;AAEnC,UAAM,CAAC,WAAW,IAAI,KAAK,cAAc,OAAO,OAAO,CAAC;AAGxD,QAAI,QAAQ,KAAK,UAAU;AAIzB,aAAO,UAAU;AACjB,UAAI,QAAQ,KAAK,YAAY;AAM3B,cAAM,eAAe,KAAK,aAAa;AACvC,YAAI,eAAe,KAAK,cAAc,QAAQ;AAC5C,iBAAO,UAAU,KAAK,cAAc,YAAY;AAAA,QAClD,OAAO;AAGL,iBAAO,UAAU;AAAA,QACnB;AAAA,MACF;AAKA,YAAM,cAAc,KAAK,WAAW;AACpC,UAAI,cAAc,KAAK,cAAc,QAAQ;AAC3C,eAAO,SAAS,KAAK,cAAc,WAAW;AAAA,MAChD;AAAA,IACF;AAEA,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA,EAMA,WAAW,OAAkB;AAC3B,WAAO;AAAA,MAAa,KAAK;AAAA,MAAe,aAAa,OAAO,EAAE;AAAA,MAAG,CAAC,GAAG,MACnE,KAAK,YAAY,SAAS,CAAC,GAAG,SAAS,CAAC,CAAC;AAAA,IAAA;AAAA,EAE7C;AACF;AAOO,MAAM,wCAA8C,cAGzD;AAAA,EACA,6BAA6B,IAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAM7B;AAAA,EAEA,YACE,IACA,QACA,QACA,YACA,SACA;AACA,UAAM,IAAI,QAAQ,MAAM;AACxB,UAAM,QAAQ,QAAQ,SAAS;AAC/B,UAAM,SAAS,QAAQ,UAAU;AACjC,UAAM,sBAAsB,CAC1B,GACA,MACG;AAEH,YAAM,kBAAkB,WAAW,OAAO,CAAC,GAAG,OAAO,CAAC,CAAC;AACvD,UAAI,oBAAoB,GAAG;AACzB,eAAO;AAAA,MACT;AAEA,YAAM,cAAc,OAAO,CAAC;AAC5B,YAAM,cAAc,OAAO,CAAC;AAC5B,aAAO,cAAc;AAAA,IACvB;AACA,SAAK,QAAQ,KAAK,WAAW,QAAQ,OAAO,mBAAmB;AAC/D,YAAQ,kBAAkB,MAAM,KAAK,MAAM,IAAI;AAC/C,YAAQ,cAAc,KAAK,SAAS,KAAK,IAAI,CAAC;AAAA,EAChD;AAAA,EAEU,WACR,QACA,OACA,YACyB;AACzB,WAAO,IAAI,UAAU,QAAQ,OAAO,UAAU;AAAA,EAChD;AAAA;AAAA;AAAA;AAAA;AAAA,EAMA,SAAS,EAAE,QAAQ,SAA8C;AAC/D,QAAI,EAAE,KAAK,iBAAiB,YAAY;AACtC,YAAM,IAAI;AAAA,QACR;AAAA,MAAA;AAAA,IAEJ;AAEA,UAAM,SAAgD,CAAA;AAEtD,UAAM,OAAO,KAAK,MAAM,KAAK,EAAE,QAAQ,OAAO;AAE9C,SAAK,QAAQ,QAAQ,CAAC,WAAW,KAAK,aAAa,QAAQ,MAAM,CAAC;AAClE,SAAK,SAAS,QAAQ,CAAC,YAAY,KAAK,cAAc,SAAS,MAAM,CAAC;AAEtE,QAAI,KAAK,SAAS;AAIhB,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,MAAY;AACV,UAAM,SAAgD,CAAA;AACtD,eAAW,WAAW,KAAK,iBAAiB;AAC1C,iBAAW,CAAC,MAAM,YAAY,KAAK,QAAQ,YAAY;AACrD,cAAM,CAAC,KAAK,KAAK,IAAI;AACrB,aAAK,eAAe,KAAK,OAAO,cAAc,MAAM;AAAA,MACtD;AAAA,IACF;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,eACE,KACA,OACA,cACA,QACM;AACN,UAAM,EAAE,iBAAiB,gBAAA,IAAoB,KAAK,OAAO,KAAK,YAAY;AAE1E,QAAI,MAAsC;AAAA,MACxC,QAAQ;AAAA,MACR,SAAS;AAAA,IAAA;AAEX,QAAI,mBAAmB,KAAK,kBAAkB,GAAG;AAG/C,YAAM,cAAc,SAAS,KAAK,KAAK;AACvC,YAAM,KAAK,MAAM,OAAO,WAAW;AAAA,IACrC,WAAW,kBAAkB,KAAK,mBAAmB,GAAG;AAGtD,YAAM,cAAc,SAAS,KAAK,KAAK;AACvC,YAAM,KAAK,MAAM,OAAO,WAAW;AAAA,IACrC,MAAO;AAMP,SAAK,aAAa,IAAI,QAAQ,MAAM;AACpC,SAAK,cAAc,IAAI,SAAS,MAAM;AAEtC;AAAA,EACF;AAAA,EAEQ,aACN,QACA,QACA;AACA,QAAI,QAAQ;AACV,YAAM,QAAQ,SAAS,MAAM;AAC7B,YAAM,cAAc,SAAS,MAAM;AACnC,YAAM,IAAI,OAAO,WAAW;AAC5B,YAAM,MAAM,OAAO,WAAW;AAC9B,aAAO,KAAK,CAAC,CAAC,GAAG,CAAC,KAAK,KAAK,CAAC,GAAG,CAAC,CAAC;AAAA,IACpC;AAAA,EACF;AAAA,EAEQ,cACN,SACA,QACA;AACA,QAAI,SAAS;AACX,YAAM,QAAQ,SAAS,OAAO;AAC9B,YAAM,cAAc,SAAS,OAAO;AACpC,YAAM,IAAI,OAAO,WAAW;AAC5B,YAAM,MAAM,OAAO,WAAW;AAC9B,aAAO,KAAK,CAAC,CAAC,GAAG,CAAC,KAAK,KAAK,CAAC,GAAG,EAAE,CAAC;AAAA,IACrC;AAAA,EACF;AAAA,EAEQ,gBAAgB,KAAgB;AACtC,WAAO,KAAK,OAAO,IAAI,GAAG,KAAK;AAAA,EACjC;AAAA,EAEQ,OACN,KACA,cACsD;AACtD,UAAM,kBAAkB,KAAK,gBAAgB,GAAG;AAChD,UAAM,kBAAkB,kBAAkB;AAC1C,QAAI,oBAAoB,GAAG;AACzB,WAAK,OAAO,OAAO,GAAG;AAAA,IACxB,OAAO;AACL,WAAK,OAAO,IAAI,KAAK,eAAe;AAAA,IACtC;AACA,WAAO,EAAE,iBAAiB,gBAAA;AAAA,EAC5B;AACF;AAYO,SAAS,wBACd,YACA,SACqD;AACrD,QAAM,OAAO,WAAW,CAAA;AAExB,SAAO,CACL,WAC6C;AAC7C,UAAM,SAAS,IAAI;AAAA,MACjB,OAAO;AAAA,MACP,IAAI,uBAAA;AAAA,IAAiD;AAEvD,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,OAAO;AAAA,MACP;AAAA,MACA;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;AAMO,SAAS,aACd,OACA,OACiB;AACjB,SAAO,CAAC,OAAO,KAAK;AACtB;AAEO,SAAS,SAAY,YAAgC;AAC1D,SAAO,WAAW,CAAC;AACrB;AAEO,SAAS,SAAY,YAA8C;AACxE,SAAO,WAAW,CAAC;AACrB;AAKA,SAAS,SAAe,KAAQ,OAA6B;AAC3D,SAAO,CAAC,KAAK,OAAO,wBAAwB,MAAM,GAAG,CAAC;AACxD;AAEA,SAAS,OAAa,uBAA6C;AACjE,SAAO,sBAAsB,CAAC;AAChC;AAEA,SAAS,OAAa,uBAA6C;AACjE,SAAO,sBAAsB,CAAC;AAChC;AAEA,SAAS,OAAa,uBAA+C;AACnE,SAAO,sBAAsB,CAAC;AAChC;"}
|
package/package.json
CHANGED
|
@@ -87,12 +87,25 @@ class TopKArray<V> implements TopK<V> {
|
|
|
87
87
|
}): TopKMoveChanges<V> {
|
|
88
88
|
const oldOffset = this.#topKStart
|
|
89
89
|
const oldLimit = this.#topKEnd - this.#topKStart
|
|
90
|
-
const oldRange: HRange = [this.#topKStart, this.#topKEnd]
|
|
91
90
|
|
|
92
|
-
this.#
|
|
93
|
-
|
|
91
|
+
// `this.#topKEnd` can be `Infinity` if it has no limit
|
|
92
|
+
// but `diffHalfOpen` expects a finite range
|
|
93
|
+
// so we restrict it to the size of the topK if topKEnd is infinite
|
|
94
|
+
const oldRange: HRange = [
|
|
95
|
+
this.#topKStart,
|
|
96
|
+
this.#topKEnd === Infinity ? this.#topKStart + this.size : this.#topKEnd,
|
|
97
|
+
]
|
|
94
98
|
|
|
95
|
-
|
|
99
|
+
this.#topKStart = offset ?? oldOffset
|
|
100
|
+
this.#topKEnd = this.#topKStart + (limit ?? oldLimit) // can be `Infinity` if limit is `Infinity`
|
|
101
|
+
|
|
102
|
+
// Also handle `Infinity` in the newRange
|
|
103
|
+
const newRange: HRange = [
|
|
104
|
+
this.#topKStart,
|
|
105
|
+
this.#topKEnd === Infinity
|
|
106
|
+
? Math.max(this.#topKStart + this.size, oldRange[1]) // since the new limit is Infinity we need to take everything (so we need to take the biggest (finite) topKEnd)
|
|
107
|
+
: this.#topKEnd,
|
|
108
|
+
]
|
|
96
109
|
const { onlyInA, onlyInB } = diffHalfOpen(oldRange, newRange)
|
|
97
110
|
|
|
98
111
|
const moveIns: Array<IndexedValue<V>> = []
|