@tanstack/db-ivm 0.1.24 → 0.1.26
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/d2.d.cts +3 -3
- package/dist/cjs/graph.d.cts +2 -2
- package/dist/cjs/hashing/index.d.cts +2 -2
- package/dist/cjs/index.cjs +42 -42
- package/dist/cjs/index.d.cts +6 -6
- package/dist/cjs/indexes.cjs +6 -3
- package/dist/cjs/indexes.cjs.map +1 -1
- package/dist/cjs/indexes.d.cts +38 -8
- package/dist/cjs/multiset.cjs +12 -4
- package/dist/cjs/multiset.cjs.map +1 -1
- package/dist/cjs/multiset.d.cts +9 -4
- package/dist/cjs/operators/concat.d.cts +2 -2
- package/dist/cjs/operators/consolidate.d.cts +2 -2
- package/dist/cjs/operators/count.cjs.map +1 -1
- package/dist/cjs/operators/count.d.cts +11 -5
- package/dist/cjs/operators/debug.d.cts +2 -2
- package/dist/cjs/operators/distinct.d.cts +2 -2
- package/dist/cjs/operators/filter.d.cts +3 -3
- package/dist/cjs/operators/filterBy.cjs.map +1 -1
- package/dist/cjs/operators/filterBy.d.cts +2 -2
- package/dist/cjs/operators/groupBy.cjs.map +1 -1
- package/dist/cjs/operators/groupBy.d.cts +5 -2
- package/dist/cjs/operators/groupedOrderBy.d.cts +8 -2
- package/dist/cjs/operators/groupedTopKWithFractionalIndex.d.cts +34 -7
- package/dist/cjs/operators/index.d.cts +21 -21
- package/dist/cjs/operators/join.cjs.map +1 -1
- package/dist/cjs/operators/join.d.cts +49 -10
- package/dist/cjs/operators/keying.d.cts +5 -2
- package/dist/cjs/operators/map.d.cts +3 -3
- package/dist/cjs/operators/negate.d.cts +3 -3
- package/dist/cjs/operators/orderBy.d.cts +20 -5
- package/dist/cjs/operators/output.d.cts +3 -3
- package/dist/cjs/operators/pipe.d.cts +1 -1
- package/dist/cjs/operators/reduce.cjs.map +1 -1
- package/dist/cjs/operators/reduce.d.cts +29 -5
- package/dist/cjs/operators/tap.d.cts +3 -3
- package/dist/cjs/operators/topK.cjs.map +1 -1
- package/dist/cjs/operators/topK.d.cts +5 -2
- package/dist/cjs/operators/topKArray.d.cts +11 -2
- package/dist/cjs/operators/topKState.d.cts +44 -8
- package/dist/cjs/operators/topKWithFractionalIndex.d.cts +41 -8
- package/dist/cjs/operators/topKWithFractionalIndexBTree.d.cts +20 -5
- package/dist/cjs/types.d.cts +6 -3
- package/dist/cjs/utils.cjs +9 -7
- package/dist/cjs/utils.cjs.map +1 -1
- package/dist/cjs/utils.d.cts +7 -4
- package/dist/esm/index.js +15 -15
- package/dist/esm/indexes.js +6 -3
- package/dist/esm/indexes.js.map +1 -1
- package/dist/esm/multiset.d.ts +1 -2
- package/dist/esm/multiset.js +12 -4
- package/dist/esm/multiset.js.map +1 -1
- package/dist/esm/operators/count.d.ts +1 -1
- package/dist/esm/operators/count.js.map +1 -1
- package/dist/esm/operators/filterBy.d.ts +1 -1
- package/dist/esm/operators/filterBy.js.map +1 -1
- package/dist/esm/operators/groupBy.js.map +1 -1
- package/dist/esm/operators/join.d.ts +6 -6
- package/dist/esm/operators/join.js.map +1 -1
- package/dist/esm/operators/reduce.d.ts +1 -1
- package/dist/esm/operators/reduce.js.map +1 -1
- package/dist/esm/operators/topK.d.ts +2 -2
- package/dist/esm/operators/topK.js.map +1 -1
- package/dist/esm/operators/topKArray.js +1 -1
- package/dist/esm/utils.d.ts +3 -3
- package/dist/esm/utils.js +9 -7
- package/dist/esm/utils.js.map +1 -1
- package/package.json +2 -2
- package/src/indexes.ts +8 -3
- package/src/multiset.ts +13 -7
- package/src/operators/count.ts +2 -2
- package/src/operators/filterBy.ts +1 -1
- package/src/operators/groupBy.ts +1 -2
- package/src/operators/join.ts +6 -6
- package/src/operators/reduce.ts +2 -2
- package/src/operators/topK.ts +4 -4
- package/src/utils.ts +14 -10
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"groupBy.cjs","sources":["../../../src/operators/groupBy.ts"],"sourcesContent":["import { serializeValue } from '../utils.js'\nimport { map } from './map.js'\nimport { reduce } from './reduce.js'\nimport type { IStreamBuilder, KeyValue } from '../types.js'\n\ntype GroupKey = Record<string, unknown>\n\ntype BasicAggregateFunction<T, R, V = unknown> = {\n preMap: (data: T) => V\n reduce: (values: Array<[V, number]>) => V\n postMap?: (result: V) => R\n}\n\ntype PipedAggregateFunction<T, R> = {\n pipe: (stream: IStreamBuilder<T>) => IStreamBuilder<KeyValue<string, R>>\n}\n\ntype AggregateFunction<T, R, V = unknown> =\n | BasicAggregateFunction<T, R, V>\n | PipedAggregateFunction<T, R>\n\ntype ExtractAggregateReturnType<T, A> =\n A extends AggregateFunction<T, infer R, any> ? R : never\n\ntype AggregatesReturnType<T, A> = {\n [K in keyof A]: ExtractAggregateReturnType<T, A[K]>\n}\n\nfunction isPipedAggregateFunction<T, R>(\n aggregate: AggregateFunction<T, R>,\n): aggregate is PipedAggregateFunction<T, R> {\n return `pipe` in aggregate\n}\n\n/**\n * Groups data by key and applies multiple aggregate operations\n * @param keyExtractor Function to extract grouping key from data\n * @param aggregates Object mapping aggregate names to aggregate functions\n */\nexport function groupBy<\n T,\n K extends GroupKey,\n A extends Record<string, AggregateFunction<T, any, any>>,\n>(keyExtractor: (data: T) => K, aggregates: A = {} as A) {\n type ResultType = K & AggregatesReturnType<T, A>\n\n const basicAggregates = Object.fromEntries(\n Object.entries(aggregates).filter(\n ([_, aggregate]) => !isPipedAggregateFunction(aggregate),\n ),\n ) as Record<string, BasicAggregateFunction<T, any, any>>\n\n // @ts-expect-error - TODO: we don't use this yet, but we will\n // eslint-disable-next-line @typescript-eslint/no-unused-vars\n const pipedAggregates = Object.fromEntries(\n Object.entries(aggregates).filter(([_, aggregate]) =>\n isPipedAggregateFunction(aggregate),\n ),\n ) as Record<string, PipedAggregateFunction<T, any>>\n\n return (\n stream: IStreamBuilder<T>,\n ): IStreamBuilder<KeyValue<string, ResultType>> => {\n // Special key to store the original key object\n const KEY_SENTINEL = Symbol(`original_group_key`)\n\n // First map to extract keys and pre-aggregate values\n const withKeysAndValues = stream.pipe(\n map((data) => {\n const key = keyExtractor(data)\n const keyString = serializeValue(key)\n\n // Create values object with pre-aggregated values\n const values: Record<string | symbol, unknown> = {}\n\n // Store the original key object\n values[KEY_SENTINEL] = key\n\n // Add pre-aggregated values\n for (const [name, aggregate] of Object.entries(basicAggregates)) {\n values[name] = aggregate.preMap(data)\n }\n\n return [keyString, values] as KeyValue<\n string,\n Record<string | symbol, unknown>\n >\n }),\n )\n\n // Then reduce to compute aggregates\n const reduced = withKeysAndValues.pipe(\n reduce((values) => {\n // Calculate total multiplicity to check if the group should exist\n let totalMultiplicity = 0\n for (const [_, multiplicity] of values) {\n totalMultiplicity += multiplicity\n }\n\n // If total multiplicity is 0 or negative, the group should be removed completely\n if (totalMultiplicity <= 0) {\n return []\n }\n\n const result: Record<PropertyKey, unknown> = {}\n\n // Get the original key from first value in group\n const originalKey = values[0]?.[0]?.[KEY_SENTINEL]\n result[KEY_SENTINEL] = originalKey\n\n // Apply each aggregate function\n for (const [name, aggregate] of Object.entries(basicAggregates)) {\n const preValues = values.map(\n ([v, m]) => [v[name], m] as [any, number],\n )\n result[name] = aggregate.reduce(preValues)\n }\n\n return [[result, 1]]\n }),\n )\n\n // Finally map to extract the key and include all values\n return reduced.pipe(\n map(([keyString, values]) => {\n // Extract the original key\n const key = values[KEY_SENTINEL] as K\n\n // Create intermediate result with key values and aggregate results\n const result: Record<string, unknown> = {}\n\n // Add key properties to result\n Object.assign(result, key)\n\n // Apply postMap if provided\n for (const [name, aggregate] of Object.entries(basicAggregates)) {\n if (aggregate.postMap) {\n result[name] = aggregate.postMap(values[name])\n } else {\n result[name] = values[name]\n }\n }\n\n // Return with the string key instead of the object\n return [keyString, result] as KeyValue<string, ResultType>\n }),\n )\n }\n}\n\n/**\n * Creates a sum aggregate function\n */\nexport function sum<T>(\n valueExtractor: (value: T) => number = (v) => v as unknown as number,\n): AggregateFunction<T, number, number> {\n return {\n preMap: (data: T) => valueExtractor(data),\n reduce: (values: Array<[number, number]>) => {\n let total = 0\n for (const [value, multiplicity] of values) {\n total += value * multiplicity\n }\n return total\n },\n }\n}\n\n/**\n * Creates a count aggregate function\n */\nexport function count<T>(\n valueExtractor: (value: T) => any = (v) => v,\n): AggregateFunction<T, number, number> {\n return {\n // Count only not-null values (the `== null` comparison gives true for both null and undefined)\n preMap: (data: T) => (valueExtractor(data) == null ? 0 : 1),\n reduce: (values: Array<[number, number]>) => {\n let totalCount = 0\n for (const [nullMultiplier, multiplicity] of values) {\n totalCount += nullMultiplier * multiplicity\n }\n return totalCount\n },\n }\n}\n\n/**\n * Creates an average aggregate function\n */\nexport function avg<T>(\n valueExtractor: (value: T) => number = (v) => v as unknown as number,\n): AggregateFunction<T, number, { sum: number; count: number }> {\n return {\n preMap: (data: T) => ({\n sum: valueExtractor(data),\n count: 0,\n }),\n reduce: (values: Array<[{ sum: number; count: number }, number]>) => {\n let totalSum = 0\n let totalCount = 0\n for (const [value, multiplicity] of values) {\n totalSum += value.sum * multiplicity\n totalCount += multiplicity\n }\n return {\n sum: totalSum,\n count: totalCount,\n }\n },\n postMap: (result: { sum: number; count: number }) => {\n return result.sum / result.count\n },\n }\n}\n\ntype CanMinMax = number | Date | bigint | string\n\n/**\n * Creates a min aggregate function that computes the minimum value in a group\n * @param valueExtractor Function to extract a comparable value from each data entry\n */\nexport function min<T extends CanMinMax>(): AggregateFunction<\n T,\n T | undefined,\n T | undefined\n>\nexport function min<T, V extends CanMinMax>(\n valueExtractor: (value: T) => V,\n): AggregateFunction<T, V | undefined, V | undefined>\nexport function min<T, V extends CanMinMax>(\n valueExtractor?: (value: T) => V,\n): AggregateFunction<T, V | undefined, V | undefined> {\n const extractor = valueExtractor ?? ((v: T) => v as unknown as V)\n return {\n preMap: (data: T) => extractor(data),\n reduce: (values) => {\n let minValue: V | undefined\n for (const [value, _multiplicity] of values) {\n if (\n value !== undefined &&\n (minValue === undefined || value < minValue)\n ) {\n minValue = value\n }\n }\n return minValue\n },\n }\n}\n\n/**\n * Creates a max aggregate function that computes the maximum value in a group\n * @param valueExtractor Function to extract a comparable value from each data entry\n */\nexport function max<T extends CanMinMax>(): AggregateFunction<\n T,\n T | undefined,\n T | undefined\n>\nexport function max<T, V extends CanMinMax>(\n valueExtractor: (value: T) => V,\n): AggregateFunction<T, V | undefined, V | undefined>\nexport function max<T, V extends CanMinMax>(\n valueExtractor?: (value: T) => V,\n): AggregateFunction<T, V | undefined, V | undefined> {\n const extractor = valueExtractor ?? ((v: T) => v as unknown as V)\n return {\n preMap: (data: T) => extractor(data),\n reduce: (values) => {\n let maxValue: V | undefined\n for (const [value, _multiplicity] of values) {\n if (\n value !== undefined &&\n (maxValue === undefined || value > maxValue)\n ) {\n maxValue = value\n }\n }\n return maxValue\n },\n }\n}\n\n/**\n * Creates a median aggregate function that computes the middle value in a sorted group\n * If there's an even number of values, returns the average of the two middle values\n * @param valueExtractor Function to extract a numeric value from each data entry\n */\nexport function median<T>(\n valueExtractor: (value: T) => number = (v) => v as unknown as number,\n): AggregateFunction<T, number, Array<number>> {\n return {\n preMap: (data: T) => [valueExtractor(data)],\n reduce: (values: Array<[Array<number>, number]>) => {\n // Flatten all values, taking multiplicity into account\n const allValues: Array<number> = []\n for (const [valueArray, multiplicity] of values) {\n for (const value of valueArray) {\n // Add each value multiple times based on multiplicity\n for (let i = 0; i < multiplicity; i++) {\n allValues.push(value)\n }\n }\n }\n\n // Return empty array if no values\n if (allValues.length === 0) {\n return []\n }\n\n // Sort values\n allValues.sort((a, b) => a - b)\n\n return allValues\n },\n postMap: (result: Array<number>) => {\n if (result.length === 0) return 0\n\n const mid = Math.floor(result.length / 2)\n\n // If even number of values, average the two middle values\n if (result.length % 2 === 0) {\n return (result[mid - 1]! + result[mid]!) / 2\n }\n\n // If odd number of values, return the middle value\n return result[mid]!\n },\n }\n}\n\n/**\n * Creates a mode aggregate function that computes the most frequent value in a group\n * If multiple values have the same highest frequency, returns the first one encountered\n * @param valueExtractor Function to extract a value from each data entry\n */\nexport function mode<T>(\n valueExtractor: (value: T) => number = (v) => v as unknown as number,\n): AggregateFunction<T, number, Map<number, number>> {\n return {\n preMap: (data: T) => {\n const value = valueExtractor(data)\n const frequencyMap = new Map<number, number>()\n frequencyMap.set(value, 1)\n return frequencyMap\n },\n reduce: (values: Array<[Map<number, number>, number]>) => {\n // Combine all frequency maps\n const combinedMap = new Map<number, number>()\n\n for (const [frequencyMap, multiplicity] of values) {\n for (const [value, frequencyCount] of frequencyMap.entries()) {\n const currentCount = combinedMap.get(value) || 0\n combinedMap.set(value, currentCount + frequencyCount * multiplicity)\n }\n }\n\n return combinedMap\n },\n postMap: (result: Map<number, number>) => {\n if (result.size === 0) return 0\n\n let modeValue = 0\n let maxFrequency = 0\n\n for (const [value, frequency] of result.entries()) {\n if (frequency > maxFrequency) {\n maxFrequency = frequency\n modeValue = value\n }\n }\n\n return modeValue\n },\n }\n}\n\nexport const groupByOperators = {\n sum,\n count,\n avg,\n min,\n max,\n median,\n mode,\n}\n"],"names":["map","serializeValue","reduce"],"mappings":";;;;;AA4BA,SAAS,yBACP,WAC2C;AAC3C,SAAO,UAAU;AACnB;AAOO,SAAS,QAId,cAA8B,aAAgB,IAAS;AAGvD,QAAM,kBAAkB,OAAO;AAAA,IAC7B,OAAO,QAAQ,UAAU,EAAE;AAAA,MACzB,CAAC,CAAC,GAAG,SAAS,MAAM,CAAC,yBAAyB,SAAS;AAAA,IAAA;AAAA,EACzD;AAKsB,SAAO;AAAA,IAC7B,OAAO,QAAQ,UAAU,EAAE;AAAA,MAAO,CAAC,CAAC,GAAG,SAAS,MAC9C,yBAAyB,SAAS;AAAA,IAAA;AAAA,EACpC;AAGF,SAAO,CACL,WACiD;AAEjD,UAAM,sCAAsB,oBAAoB;AAGhD,UAAM,oBAAoB,OAAO;AAAA,MAC/BA,IAAAA,IAAI,CAAC,SAAS;AACZ,cAAM,MAAM,aAAa,IAAI;AAC7B,cAAM,YAAYC,MAAAA,eAAe,GAAG;AAGpC,cAAM,SAA2C,CAAA;AAGjD,eAAO,YAAY,IAAI;AAGvB,mBAAW,CAAC,MAAM,SAAS,KAAK,OAAO,QAAQ,eAAe,GAAG;AAC/D,iBAAO,IAAI,IAAI,UAAU,OAAO,IAAI;AAAA,QACtC;AAEA,eAAO,CAAC,WAAW,MAAM;AAAA,MAI3B,CAAC;AAAA,IAAA;AAIH,UAAM,UAAU,kBAAkB;AAAA,MAChCC,OAAAA,OAAO,CAAC,WAAW;AAEjB,YAAI,oBAAoB;AACxB,mBAAW,CAAC,GAAG,YAAY,KAAK,QAAQ;AACtC,+BAAqB;AAAA,QACvB;AAGA,YAAI,qBAAqB,GAAG;AAC1B,iBAAO,CAAA;AAAA,QACT;AAEA,cAAM,SAAuC,CAAA;AAG7C,cAAM,cAAc,OAAO,CAAC,IAAI,CAAC,IAAI,YAAY;AACjD,eAAO,YAAY,IAAI;AAGvB,mBAAW,CAAC,MAAM,SAAS,KAAK,OAAO,QAAQ,eAAe,GAAG;AAC/D,gBAAM,YAAY,OAAO;AAAA,YACvB,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,IAAI,GAAG,CAAC;AAAA,UAAA;AAEzB,iBAAO,IAAI,IAAI,UAAU,OAAO,SAAS;AAAA,QAC3C;AAEA,eAAO,CAAC,CAAC,QAAQ,CAAC,CAAC;AAAA,MACrB,CAAC;AAAA,IAAA;AAIH,WAAO,QAAQ;AAAA,MACbF,IAAAA,IAAI,CAAC,CAAC,WAAW,MAAM,MAAM;AAE3B,cAAM,MAAM,OAAO,YAAY;AAG/B,cAAM,SAAkC,CAAA;AAGxC,eAAO,OAAO,QAAQ,GAAG;AAGzB,mBAAW,CAAC,MAAM,SAAS,KAAK,OAAO,QAAQ,eAAe,GAAG;AAC/D,cAAI,UAAU,SAAS;AACrB,mBAAO,IAAI,IAAI,UAAU,QAAQ,OAAO,IAAI,CAAC;AAAA,UAC/C,OAAO;AACL,mBAAO,IAAI,IAAI,OAAO,IAAI;AAAA,UAC5B;AAAA,QACF;AAGA,eAAO,CAAC,WAAW,MAAM;AAAA,MAC3B,CAAC;AAAA,IAAA;AAAA,EAEL;AACF;AAKO,SAAS,IACd,iBAAuC,CAAC,MAAM,GACR;AACtC,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,eAAe,IAAI;AAAA,IACxC,QAAQ,CAAC,WAAoC;AAC3C,UAAI,QAAQ;AACZ,iBAAW,CAAC,OAAO,YAAY,KAAK,QAAQ;AAC1C,iBAAS,QAAQ;AAAA,MACnB;AACA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAKO,SAAS,MACd,iBAAoC,CAAC,MAAM,GACL;AACtC,SAAO;AAAA;AAAA,IAEL,QAAQ,CAAC,SAAa,eAAe,IAAI,KAAK,OAAO,IAAI;AAAA,IACzD,QAAQ,CAAC,WAAoC;AAC3C,UAAI,aAAa;AACjB,iBAAW,CAAC,gBAAgB,YAAY,KAAK,QAAQ;AACnD,sBAAc,iBAAiB;AAAA,MACjC;AACA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAKO,SAAS,IACd,iBAAuC,CAAC,MAAM,GACgB;AAC9D,SAAO;AAAA,IACL,QAAQ,CAAC,UAAa;AAAA,MACpB,KAAK,eAAe,IAAI;AAAA,MACxB,OAAO;AAAA,IAAA;AAAA,IAET,QAAQ,CAAC,WAA4D;AACnE,UAAI,WAAW;AACf,UAAI,aAAa;AACjB,iBAAW,CAAC,OAAO,YAAY,KAAK,QAAQ;AAC1C,oBAAY,MAAM,MAAM;AACxB,sBAAc;AAAA,MAChB;AACA,aAAO;AAAA,QACL,KAAK;AAAA,QACL,OAAO;AAAA,MAAA;AAAA,IAEX;AAAA,IACA,SAAS,CAAC,WAA2C;AACnD,aAAO,OAAO,MAAM,OAAO;AAAA,IAC7B;AAAA,EAAA;AAEJ;AAgBO,SAAS,IACd,gBACoD;AACpD,QAAM,YAAY,mBAAmB,CAAC,MAAS;AAC/C,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,UAAU,IAAI;AAAA,IACnC,QAAQ,CAAC,WAAW;AAClB,UAAI;AACJ,iBAAW,CAAC,OAAO,aAAa,KAAK,QAAQ;AAC3C,YACE,UAAU,WACT,aAAa,UAAa,QAAQ,WACnC;AACA,qBAAW;AAAA,QACb;AAAA,MACF;AACA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAcO,SAAS,IACd,gBACoD;AACpD,QAAM,YAAY,mBAAmB,CAAC,MAAS;AAC/C,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,UAAU,IAAI;AAAA,IACnC,QAAQ,CAAC,WAAW;AAClB,UAAI;AACJ,iBAAW,CAAC,OAAO,aAAa,KAAK,QAAQ;AAC3C,YACE,UAAU,WACT,aAAa,UAAa,QAAQ,WACnC;AACA,qBAAW;AAAA,QACb;AAAA,MACF;AACA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAOO,SAAS,OACd,iBAAuC,CAAC,MAAM,GACD;AAC7C,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,CAAC,eAAe,IAAI,CAAC;AAAA,IAC1C,QAAQ,CAAC,WAA2C;AAElD,YAAM,YAA2B,CAAA;AACjC,iBAAW,CAAC,YAAY,YAAY,KAAK,QAAQ;AAC/C,mBAAW,SAAS,YAAY;AAE9B,mBAAS,IAAI,GAAG,IAAI,cAAc,KAAK;AACrC,sBAAU,KAAK,KAAK;AAAA,UACtB;AAAA,QACF;AAAA,MACF;AAGA,UAAI,UAAU,WAAW,GAAG;AAC1B,eAAO,CAAA;AAAA,MACT;AAGA,gBAAU,KAAK,CAAC,GAAG,MAAM,IAAI,CAAC;AAE9B,aAAO;AAAA,IACT;AAAA,IACA,SAAS,CAAC,WAA0B;AAClC,UAAI,OAAO,WAAW,EAAG,QAAO;AAEhC,YAAM,MAAM,KAAK,MAAM,OAAO,SAAS,CAAC;AAGxC,UAAI,OAAO,SAAS,MAAM,GAAG;AAC3B,gBAAQ,OAAO,MAAM,CAAC,IAAK,OAAO,GAAG,KAAM;AAAA,MAC7C;AAGA,aAAO,OAAO,GAAG;AAAA,IACnB;AAAA,EAAA;AAEJ;AAOO,SAAS,KACd,iBAAuC,CAAC,MAAM,GACK;AACnD,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY;AACnB,YAAM,QAAQ,eAAe,IAAI;AACjC,YAAM,mCAAmB,IAAA;AACzB,mBAAa,IAAI,OAAO,CAAC;AACzB,aAAO;AAAA,IACT;AAAA,IACA,QAAQ,CAAC,WAAiD;AAExD,YAAM,kCAAkB,IAAA;AAExB,iBAAW,CAAC,cAAc,YAAY,KAAK,QAAQ;AACjD,mBAAW,CAAC,OAAO,cAAc,KAAK,aAAa,WAAW;AAC5D,gBAAM,eAAe,YAAY,IAAI,KAAK,KAAK;AAC/C,sBAAY,IAAI,OAAO,eAAe,iBAAiB,YAAY;AAAA,QACrE;AAAA,MACF;AAEA,aAAO;AAAA,IACT;AAAA,IACA,SAAS,CAAC,WAAgC;AACxC,UAAI,OAAO,SAAS,EAAG,QAAO;AAE9B,UAAI,YAAY;AAChB,UAAI,eAAe;AAEnB,iBAAW,CAAC,OAAO,SAAS,KAAK,OAAO,WAAW;AACjD,YAAI,YAAY,cAAc;AAC5B,yBAAe;AACf,sBAAY;AAAA,QACd;AAAA,MACF;AAEA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAEO,MAAM,mBAAmB;AAAA,EAC9B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;;;;;;;;;;"}
|
|
1
|
+
{"version":3,"file":"groupBy.cjs","sources":["../../../src/operators/groupBy.ts"],"sourcesContent":["import { serializeValue } from '../utils.js'\nimport { map } from './map.js'\nimport { reduce } from './reduce.js'\nimport type { IStreamBuilder, KeyValue } from '../types.js'\n\ntype GroupKey = Record<string, unknown>\n\ntype BasicAggregateFunction<T, R, V = unknown> = {\n preMap: (data: T) => V\n reduce: (values: Array<[V, number]>) => V\n postMap?: (result: V) => R\n}\n\ntype PipedAggregateFunction<T, R> = {\n pipe: (stream: IStreamBuilder<T>) => IStreamBuilder<KeyValue<string, R>>\n}\n\ntype AggregateFunction<T, R, V = unknown> =\n BasicAggregateFunction<T, R, V> | PipedAggregateFunction<T, R>\n\ntype ExtractAggregateReturnType<T, A> =\n A extends AggregateFunction<T, infer R, any> ? R : never\n\ntype AggregatesReturnType<T, A> = {\n [K in keyof A]: ExtractAggregateReturnType<T, A[K]>\n}\n\nfunction isPipedAggregateFunction<T, R>(\n aggregate: AggregateFunction<T, R>,\n): aggregate is PipedAggregateFunction<T, R> {\n return `pipe` in aggregate\n}\n\n/**\n * Groups data by key and applies multiple aggregate operations\n * @param keyExtractor Function to extract grouping key from data\n * @param aggregates Object mapping aggregate names to aggregate functions\n */\nexport function groupBy<\n T,\n K extends GroupKey,\n A extends Record<string, AggregateFunction<T, any, any>>,\n>(keyExtractor: (data: T) => K, aggregates: A = {} as A) {\n type ResultType = K & AggregatesReturnType<T, A>\n\n const basicAggregates = Object.fromEntries(\n Object.entries(aggregates).filter(\n ([_, aggregate]) => !isPipedAggregateFunction(aggregate),\n ),\n ) as Record<string, BasicAggregateFunction<T, any, any>>\n\n // @ts-expect-error - TODO: we don't use this yet, but we will\n // eslint-disable-next-line @typescript-eslint/no-unused-vars\n const pipedAggregates = Object.fromEntries(\n Object.entries(aggregates).filter(([_, aggregate]) =>\n isPipedAggregateFunction(aggregate),\n ),\n ) as Record<string, PipedAggregateFunction<T, any>>\n\n return (\n stream: IStreamBuilder<T>,\n ): IStreamBuilder<KeyValue<string, ResultType>> => {\n // Special key to store the original key object\n const KEY_SENTINEL = Symbol(`original_group_key`)\n\n // First map to extract keys and pre-aggregate values\n const withKeysAndValues = stream.pipe(\n map((data) => {\n const key = keyExtractor(data)\n const keyString = serializeValue(key)\n\n // Create values object with pre-aggregated values\n const values: Record<string | symbol, unknown> = {}\n\n // Store the original key object\n values[KEY_SENTINEL] = key\n\n // Add pre-aggregated values\n for (const [name, aggregate] of Object.entries(basicAggregates)) {\n values[name] = aggregate.preMap(data)\n }\n\n return [keyString, values] as KeyValue<\n string,\n Record<string | symbol, unknown>\n >\n }),\n )\n\n // Then reduce to compute aggregates\n const reduced = withKeysAndValues.pipe(\n reduce((values) => {\n // Calculate total multiplicity to check if the group should exist\n let totalMultiplicity = 0\n for (const [_, multiplicity] of values) {\n totalMultiplicity += multiplicity\n }\n\n // If total multiplicity is 0 or negative, the group should be removed completely\n if (totalMultiplicity <= 0) {\n return []\n }\n\n const result: Record<PropertyKey, unknown> = {}\n\n // Get the original key from first value in group\n const originalKey = values[0]?.[0]?.[KEY_SENTINEL]\n result[KEY_SENTINEL] = originalKey\n\n // Apply each aggregate function\n for (const [name, aggregate] of Object.entries(basicAggregates)) {\n const preValues = values.map(\n ([v, m]) => [v[name], m] as [any, number],\n )\n result[name] = aggregate.reduce(preValues)\n }\n\n return [[result, 1]]\n }),\n )\n\n // Finally map to extract the key and include all values\n return reduced.pipe(\n map(([keyString, values]) => {\n // Extract the original key\n const key = values[KEY_SENTINEL] as K\n\n // Create intermediate result with key values and aggregate results\n const result: Record<string, unknown> = {}\n\n // Add key properties to result\n Object.assign(result, key)\n\n // Apply postMap if provided\n for (const [name, aggregate] of Object.entries(basicAggregates)) {\n if (aggregate.postMap) {\n result[name] = aggregate.postMap(values[name])\n } else {\n result[name] = values[name]\n }\n }\n\n // Return with the string key instead of the object\n return [keyString, result] as KeyValue<string, ResultType>\n }),\n )\n }\n}\n\n/**\n * Creates a sum aggregate function\n */\nexport function sum<T>(\n valueExtractor: (value: T) => number = (v) => v as unknown as number,\n): AggregateFunction<T, number, number> {\n return {\n preMap: (data: T) => valueExtractor(data),\n reduce: (values: Array<[number, number]>) => {\n let total = 0\n for (const [value, multiplicity] of values) {\n total += value * multiplicity\n }\n return total\n },\n }\n}\n\n/**\n * Creates a count aggregate function\n */\nexport function count<T>(\n valueExtractor: (value: T) => any = (v) => v,\n): AggregateFunction<T, number, number> {\n return {\n // Count only not-null values (the `== null` comparison gives true for both null and undefined)\n preMap: (data: T) => (valueExtractor(data) == null ? 0 : 1),\n reduce: (values: Array<[number, number]>) => {\n let totalCount = 0\n for (const [nullMultiplier, multiplicity] of values) {\n totalCount += nullMultiplier * multiplicity\n }\n return totalCount\n },\n }\n}\n\n/**\n * Creates an average aggregate function\n */\nexport function avg<T>(\n valueExtractor: (value: T) => number = (v) => v as unknown as number,\n): AggregateFunction<T, number, { sum: number; count: number }> {\n return {\n preMap: (data: T) => ({\n sum: valueExtractor(data),\n count: 0,\n }),\n reduce: (values: Array<[{ sum: number; count: number }, number]>) => {\n let totalSum = 0\n let totalCount = 0\n for (const [value, multiplicity] of values) {\n totalSum += value.sum * multiplicity\n totalCount += multiplicity\n }\n return {\n sum: totalSum,\n count: totalCount,\n }\n },\n postMap: (result: { sum: number; count: number }) => {\n return result.sum / result.count\n },\n }\n}\n\ntype CanMinMax = number | Date | bigint | string\n\n/**\n * Creates a min aggregate function that computes the minimum value in a group\n * @param valueExtractor Function to extract a comparable value from each data entry\n */\nexport function min<T extends CanMinMax>(): AggregateFunction<\n T,\n T | undefined,\n T | undefined\n>\nexport function min<T, V extends CanMinMax>(\n valueExtractor: (value: T) => V,\n): AggregateFunction<T, V | undefined, V | undefined>\nexport function min<T, V extends CanMinMax>(\n valueExtractor?: (value: T) => V,\n): AggregateFunction<T, V | undefined, V | undefined> {\n const extractor = valueExtractor ?? ((v: T) => v as unknown as V)\n return {\n preMap: (data: T) => extractor(data),\n reduce: (values) => {\n let minValue: V | undefined\n for (const [value, _multiplicity] of values) {\n if (\n value !== undefined &&\n (minValue === undefined || value < minValue)\n ) {\n minValue = value\n }\n }\n return minValue\n },\n }\n}\n\n/**\n * Creates a max aggregate function that computes the maximum value in a group\n * @param valueExtractor Function to extract a comparable value from each data entry\n */\nexport function max<T extends CanMinMax>(): AggregateFunction<\n T,\n T | undefined,\n T | undefined\n>\nexport function max<T, V extends CanMinMax>(\n valueExtractor: (value: T) => V,\n): AggregateFunction<T, V | undefined, V | undefined>\nexport function max<T, V extends CanMinMax>(\n valueExtractor?: (value: T) => V,\n): AggregateFunction<T, V | undefined, V | undefined> {\n const extractor = valueExtractor ?? ((v: T) => v as unknown as V)\n return {\n preMap: (data: T) => extractor(data),\n reduce: (values) => {\n let maxValue: V | undefined\n for (const [value, _multiplicity] of values) {\n if (\n value !== undefined &&\n (maxValue === undefined || value > maxValue)\n ) {\n maxValue = value\n }\n }\n return maxValue\n },\n }\n}\n\n/**\n * Creates a median aggregate function that computes the middle value in a sorted group\n * If there's an even number of values, returns the average of the two middle values\n * @param valueExtractor Function to extract a numeric value from each data entry\n */\nexport function median<T>(\n valueExtractor: (value: T) => number = (v) => v as unknown as number,\n): AggregateFunction<T, number, Array<number>> {\n return {\n preMap: (data: T) => [valueExtractor(data)],\n reduce: (values: Array<[Array<number>, number]>) => {\n // Flatten all values, taking multiplicity into account\n const allValues: Array<number> = []\n for (const [valueArray, multiplicity] of values) {\n for (const value of valueArray) {\n // Add each value multiple times based on multiplicity\n for (let i = 0; i < multiplicity; i++) {\n allValues.push(value)\n }\n }\n }\n\n // Return empty array if no values\n if (allValues.length === 0) {\n return []\n }\n\n // Sort values\n allValues.sort((a, b) => a - b)\n\n return allValues\n },\n postMap: (result: Array<number>) => {\n if (result.length === 0) return 0\n\n const mid = Math.floor(result.length / 2)\n\n // If even number of values, average the two middle values\n if (result.length % 2 === 0) {\n return (result[mid - 1]! + result[mid]!) / 2\n }\n\n // If odd number of values, return the middle value\n return result[mid]!\n },\n }\n}\n\n/**\n * Creates a mode aggregate function that computes the most frequent value in a group\n * If multiple values have the same highest frequency, returns the first one encountered\n * @param valueExtractor Function to extract a value from each data entry\n */\nexport function mode<T>(\n valueExtractor: (value: T) => number = (v) => v as unknown as number,\n): AggregateFunction<T, number, Map<number, number>> {\n return {\n preMap: (data: T) => {\n const value = valueExtractor(data)\n const frequencyMap = new Map<number, number>()\n frequencyMap.set(value, 1)\n return frequencyMap\n },\n reduce: (values: Array<[Map<number, number>, number]>) => {\n // Combine all frequency maps\n const combinedMap = new Map<number, number>()\n\n for (const [frequencyMap, multiplicity] of values) {\n for (const [value, frequencyCount] of frequencyMap.entries()) {\n const currentCount = combinedMap.get(value) || 0\n combinedMap.set(value, currentCount + frequencyCount * multiplicity)\n }\n }\n\n return combinedMap\n },\n postMap: (result: Map<number, number>) => {\n if (result.size === 0) return 0\n\n let modeValue = 0\n let maxFrequency = 0\n\n for (const [value, frequency] of result.entries()) {\n if (frequency > maxFrequency) {\n maxFrequency = frequency\n modeValue = value\n }\n }\n\n return modeValue\n },\n }\n}\n\nexport const groupByOperators = {\n sum,\n count,\n avg,\n min,\n max,\n median,\n mode,\n}\n"],"names":["map","serializeValue","reduce"],"mappings":";;;;;AA2BA,SAAS,yBACP,WAC2C;AAC3C,SAAO,UAAU;AACnB;AAOO,SAAS,QAId,cAA8B,aAAgB,IAAS;AAGvD,QAAM,kBAAkB,OAAO;AAAA,IAC7B,OAAO,QAAQ,UAAU,EAAE;AAAA,MACzB,CAAC,CAAC,GAAG,SAAS,MAAM,CAAC,yBAAyB,SAAS;AAAA,IAAA;AAAA,EACzD;AAKsB,SAAO;AAAA,IAC7B,OAAO,QAAQ,UAAU,EAAE;AAAA,MAAO,CAAC,CAAC,GAAG,SAAS,MAC9C,yBAAyB,SAAS;AAAA,IAAA;AAAA,EACpC;AAGF,SAAO,CACL,WACiD;AAEjD,UAAM,sCAAsB,oBAAoB;AAGhD,UAAM,oBAAoB,OAAO;AAAA,MAC/BA,IAAAA,IAAI,CAAC,SAAS;AACZ,cAAM,MAAM,aAAa,IAAI;AAC7B,cAAM,YAAYC,MAAAA,eAAe,GAAG;AAGpC,cAAM,SAA2C,CAAA;AAGjD,eAAO,YAAY,IAAI;AAGvB,mBAAW,CAAC,MAAM,SAAS,KAAK,OAAO,QAAQ,eAAe,GAAG;AAC/D,iBAAO,IAAI,IAAI,UAAU,OAAO,IAAI;AAAA,QACtC;AAEA,eAAO,CAAC,WAAW,MAAM;AAAA,MAI3B,CAAC;AAAA,IAAA;AAIH,UAAM,UAAU,kBAAkB;AAAA,MAChCC,OAAAA,OAAO,CAAC,WAAW;AAEjB,YAAI,oBAAoB;AACxB,mBAAW,CAAC,GAAG,YAAY,KAAK,QAAQ;AACtC,+BAAqB;AAAA,QACvB;AAGA,YAAI,qBAAqB,GAAG;AAC1B,iBAAO,CAAA;AAAA,QACT;AAEA,cAAM,SAAuC,CAAA;AAG7C,cAAM,cAAc,OAAO,CAAC,IAAI,CAAC,IAAI,YAAY;AACjD,eAAO,YAAY,IAAI;AAGvB,mBAAW,CAAC,MAAM,SAAS,KAAK,OAAO,QAAQ,eAAe,GAAG;AAC/D,gBAAM,YAAY,OAAO;AAAA,YACvB,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,IAAI,GAAG,CAAC;AAAA,UAAA;AAEzB,iBAAO,IAAI,IAAI,UAAU,OAAO,SAAS;AAAA,QAC3C;AAEA,eAAO,CAAC,CAAC,QAAQ,CAAC,CAAC;AAAA,MACrB,CAAC;AAAA,IAAA;AAIH,WAAO,QAAQ;AAAA,MACbF,IAAAA,IAAI,CAAC,CAAC,WAAW,MAAM,MAAM;AAE3B,cAAM,MAAM,OAAO,YAAY;AAG/B,cAAM,SAAkC,CAAA;AAGxC,eAAO,OAAO,QAAQ,GAAG;AAGzB,mBAAW,CAAC,MAAM,SAAS,KAAK,OAAO,QAAQ,eAAe,GAAG;AAC/D,cAAI,UAAU,SAAS;AACrB,mBAAO,IAAI,IAAI,UAAU,QAAQ,OAAO,IAAI,CAAC;AAAA,UAC/C,OAAO;AACL,mBAAO,IAAI,IAAI,OAAO,IAAI;AAAA,UAC5B;AAAA,QACF;AAGA,eAAO,CAAC,WAAW,MAAM;AAAA,MAC3B,CAAC;AAAA,IAAA;AAAA,EAEL;AACF;AAKO,SAAS,IACd,iBAAuC,CAAC,MAAM,GACR;AACtC,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,eAAe,IAAI;AAAA,IACxC,QAAQ,CAAC,WAAoC;AAC3C,UAAI,QAAQ;AACZ,iBAAW,CAAC,OAAO,YAAY,KAAK,QAAQ;AAC1C,iBAAS,QAAQ;AAAA,MACnB;AACA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAKO,SAAS,MACd,iBAAoC,CAAC,MAAM,GACL;AACtC,SAAO;AAAA;AAAA,IAEL,QAAQ,CAAC,SAAa,eAAe,IAAI,KAAK,OAAO,IAAI;AAAA,IACzD,QAAQ,CAAC,WAAoC;AAC3C,UAAI,aAAa;AACjB,iBAAW,CAAC,gBAAgB,YAAY,KAAK,QAAQ;AACnD,sBAAc,iBAAiB;AAAA,MACjC;AACA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAKO,SAAS,IACd,iBAAuC,CAAC,MAAM,GACgB;AAC9D,SAAO;AAAA,IACL,QAAQ,CAAC,UAAa;AAAA,MACpB,KAAK,eAAe,IAAI;AAAA,MACxB,OAAO;AAAA,IAAA;AAAA,IAET,QAAQ,CAAC,WAA4D;AACnE,UAAI,WAAW;AACf,UAAI,aAAa;AACjB,iBAAW,CAAC,OAAO,YAAY,KAAK,QAAQ;AAC1C,oBAAY,MAAM,MAAM;AACxB,sBAAc;AAAA,MAChB;AACA,aAAO;AAAA,QACL,KAAK;AAAA,QACL,OAAO;AAAA,MAAA;AAAA,IAEX;AAAA,IACA,SAAS,CAAC,WAA2C;AACnD,aAAO,OAAO,MAAM,OAAO;AAAA,IAC7B;AAAA,EAAA;AAEJ;AAgBO,SAAS,IACd,gBACoD;AACpD,QAAM,YAAY,mBAAmB,CAAC,MAAS;AAC/C,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,UAAU,IAAI;AAAA,IACnC,QAAQ,CAAC,WAAW;AAClB,UAAI;AACJ,iBAAW,CAAC,OAAO,aAAa,KAAK,QAAQ;AAC3C,YACE,UAAU,WACT,aAAa,UAAa,QAAQ,WACnC;AACA,qBAAW;AAAA,QACb;AAAA,MACF;AACA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAcO,SAAS,IACd,gBACoD;AACpD,QAAM,YAAY,mBAAmB,CAAC,MAAS;AAC/C,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,UAAU,IAAI;AAAA,IACnC,QAAQ,CAAC,WAAW;AAClB,UAAI;AACJ,iBAAW,CAAC,OAAO,aAAa,KAAK,QAAQ;AAC3C,YACE,UAAU,WACT,aAAa,UAAa,QAAQ,WACnC;AACA,qBAAW;AAAA,QACb;AAAA,MACF;AACA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAOO,SAAS,OACd,iBAAuC,CAAC,MAAM,GACD;AAC7C,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,CAAC,eAAe,IAAI,CAAC;AAAA,IAC1C,QAAQ,CAAC,WAA2C;AAElD,YAAM,YAA2B,CAAA;AACjC,iBAAW,CAAC,YAAY,YAAY,KAAK,QAAQ;AAC/C,mBAAW,SAAS,YAAY;AAE9B,mBAAS,IAAI,GAAG,IAAI,cAAc,KAAK;AACrC,sBAAU,KAAK,KAAK;AAAA,UACtB;AAAA,QACF;AAAA,MACF;AAGA,UAAI,UAAU,WAAW,GAAG;AAC1B,eAAO,CAAA;AAAA,MACT;AAGA,gBAAU,KAAK,CAAC,GAAG,MAAM,IAAI,CAAC;AAE9B,aAAO;AAAA,IACT;AAAA,IACA,SAAS,CAAC,WAA0B;AAClC,UAAI,OAAO,WAAW,EAAG,QAAO;AAEhC,YAAM,MAAM,KAAK,MAAM,OAAO,SAAS,CAAC;AAGxC,UAAI,OAAO,SAAS,MAAM,GAAG;AAC3B,gBAAQ,OAAO,MAAM,CAAC,IAAK,OAAO,GAAG,KAAM;AAAA,MAC7C;AAGA,aAAO,OAAO,GAAG;AAAA,IACnB;AAAA,EAAA;AAEJ;AAOO,SAAS,KACd,iBAAuC,CAAC,MAAM,GACK;AACnD,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY;AACnB,YAAM,QAAQ,eAAe,IAAI;AACjC,YAAM,mCAAmB,IAAA;AACzB,mBAAa,IAAI,OAAO,CAAC;AACzB,aAAO;AAAA,IACT;AAAA,IACA,QAAQ,CAAC,WAAiD;AAExD,YAAM,kCAAkB,IAAA;AAExB,iBAAW,CAAC,cAAc,YAAY,KAAK,QAAQ;AACjD,mBAAW,CAAC,OAAO,cAAc,KAAK,aAAa,WAAW;AAC5D,gBAAM,eAAe,YAAY,IAAI,KAAK,KAAK;AAC/C,sBAAY,IAAI,OAAO,eAAe,iBAAiB,YAAY;AAAA,QACrE;AAAA,MACF;AAEA,aAAO;AAAA,IACT;AAAA,IACA,SAAS,CAAC,WAAgC;AACxC,UAAI,OAAO,SAAS,EAAG,QAAO;AAE9B,UAAI,YAAY;AAChB,UAAI,eAAe;AAEnB,iBAAW,CAAC,OAAO,SAAS,KAAK,OAAO,WAAW;AACjD,YAAI,YAAY,cAAc;AAC5B,yBAAe;AACf,sBAAY;AAAA,QACd;AAAA,MACF;AAEA,aAAO;AAAA,IACT;AAAA,EAAA;AAEJ;AAEO,MAAM,mBAAmB;AAAA,EAC9B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;;;;;;;;;;"}
|
|
@@ -1,8 +1,11 @@
|
|
|
1
|
-
import { IStreamBuilder, KeyValue } from
|
|
1
|
+
import { IStreamBuilder, KeyValue } from "../types.cjs";
|
|
2
2
|
type GroupKey = Record<string, unknown>;
|
|
3
3
|
type BasicAggregateFunction<T, R, V = unknown> = {
|
|
4
4
|
preMap: (data: T) => V;
|
|
5
|
-
reduce: (values: Array<[
|
|
5
|
+
reduce: (values: Array<[
|
|
6
|
+
V,
|
|
7
|
+
number
|
|
8
|
+
]>) => V;
|
|
6
9
|
postMap?: (result: V) => R;
|
|
7
10
|
};
|
|
8
11
|
type PipedAggregateFunction<T, R> = {
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { IStreamBuilder, KeyValue } from
|
|
1
|
+
import { IStreamBuilder, KeyValue } from "../types.cjs";
|
|
2
2
|
export interface GroupedOrderByOptions<Ve> {
|
|
3
3
|
comparator?: (a: Ve, b: Ve) => number;
|
|
4
4
|
limit?: number;
|
|
@@ -28,4 +28,10 @@ export interface GroupedOrderByWithFractionalIndexOptions<Ve, KeyType = unknown,
|
|
|
28
28
|
* @param options - Configuration including groupKeyFn, comparator, limit, and offset
|
|
29
29
|
* @returns A piped operator that orders the elements per group and limits the number of results per group
|
|
30
30
|
*/
|
|
31
|
-
export declare function groupedOrderByWithFractionalIndex<T extends KeyValue<unknown, unknown>, Ve = unknown>(valueExtractor: (value: T extends KeyValue<unknown, infer V> ? V : never) => Ve, options: GroupedOrderByWithFractionalIndexOptions<Ve, T extends KeyValue<infer K, unknown> ? K : never, T extends KeyValue<unknown, infer V> ? V : never>): (stream: IStreamBuilder<T>) => IStreamBuilder<[
|
|
31
|
+
export declare function groupedOrderByWithFractionalIndex<T extends KeyValue<unknown, unknown>, Ve = unknown>(valueExtractor: (value: T extends KeyValue<unknown, infer V> ? V : never) => Ve, options: GroupedOrderByWithFractionalIndexOptions<Ve, T extends KeyValue<infer K, unknown> ? K : never, T extends KeyValue<unknown, infer V> ? V : never>): (stream: IStreamBuilder<T>) => IStreamBuilder<[
|
|
32
|
+
T extends KeyValue<infer K, unknown> ? K : never,
|
|
33
|
+
[
|
|
34
|
+
T extends KeyValue<unknown, infer V> ? V : never,
|
|
35
|
+
string
|
|
36
|
+
]
|
|
37
|
+
]>;
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { DifferenceStreamWriter, UnaryOperator, DifferenceStreamReader } from
|
|
2
|
-
import { IndexedValue, TopK } from
|
|
3
|
-
import { PipedOperator } from
|
|
1
|
+
import { DifferenceStreamWriter, UnaryOperator, DifferenceStreamReader } from "../graph.cjs";
|
|
2
|
+
import { IndexedValue, TopK } from "./topKArray.cjs";
|
|
3
|
+
import { PipedOperator } from "../types.cjs";
|
|
4
4
|
export interface GroupedTopKWithFractionalIndexOptions<K, T> {
|
|
5
5
|
limit?: number;
|
|
6
6
|
offset?: number;
|
|
@@ -24,14 +24,35 @@ export interface GroupedTopKWithFractionalIndexOptions<K, T> {
|
|
|
24
24
|
* Elements are grouped by the groupKeyFn, and each group maintains
|
|
25
25
|
* its own sorted collection with independent limit/offset.
|
|
26
26
|
*/
|
|
27
|
-
export declare class GroupedTopKWithFractionalIndexOperator<K extends string | number, T> extends UnaryOperator<[
|
|
27
|
+
export declare class GroupedTopKWithFractionalIndexOperator<K extends string | number, T> extends UnaryOperator<[
|
|
28
|
+
K,
|
|
29
|
+
T
|
|
30
|
+
], [
|
|
31
|
+
K,
|
|
32
|
+
IndexedValue<T>
|
|
33
|
+
]> {
|
|
28
34
|
#private;
|
|
29
|
-
constructor(id: number, inputA: DifferenceStreamReader<[
|
|
35
|
+
constructor(id: number, inputA: DifferenceStreamReader<[
|
|
36
|
+
K,
|
|
37
|
+
T
|
|
38
|
+
]>, output: DifferenceStreamWriter<[
|
|
39
|
+
K,
|
|
40
|
+
IndexedValue<T>
|
|
41
|
+
]>, comparator: (a: T, b: T) => number, options: GroupedTopKWithFractionalIndexOptions<K, T>);
|
|
30
42
|
/**
|
|
31
43
|
* Creates a new TopK data structure for a group.
|
|
32
44
|
* Can be overridden in subclasses to use different implementations (e.g., B+ tree).
|
|
33
45
|
*/
|
|
34
|
-
protected createTopK(offset: number, limit: number, comparator: (a: [
|
|
46
|
+
protected createTopK(offset: number, limit: number, comparator: (a: [
|
|
47
|
+
K,
|
|
48
|
+
T
|
|
49
|
+
], b: [
|
|
50
|
+
K,
|
|
51
|
+
T
|
|
52
|
+
]) => number): TopK<[
|
|
53
|
+
K,
|
|
54
|
+
T
|
|
55
|
+
]>;
|
|
35
56
|
run(): void;
|
|
36
57
|
}
|
|
37
58
|
/**
|
|
@@ -47,4 +68,10 @@ export declare class GroupedTopKWithFractionalIndexOperator<K extends string | n
|
|
|
47
68
|
* @param options - Configuration including groupKeyFn, limit, and offset
|
|
48
69
|
* @returns A piped operator that orders elements per group and limits results per group
|
|
49
70
|
*/
|
|
50
|
-
export declare function groupedTopKWithFractionalIndex<K extends string | number, T>(comparator: (a: T, b: T) => number, options: GroupedTopKWithFractionalIndexOptions<K, T>): PipedOperator<[
|
|
71
|
+
export declare function groupedTopKWithFractionalIndex<K extends string | number, T>(comparator: (a: T, b: T) => number, options: GroupedTopKWithFractionalIndexOptions<K, T>): PipedOperator<[
|
|
72
|
+
K,
|
|
73
|
+
T
|
|
74
|
+
], [
|
|
75
|
+
K,
|
|
76
|
+
IndexedValue<T>
|
|
77
|
+
]>;
|
|
@@ -1,21 +1,21 @@
|
|
|
1
|
-
export * from
|
|
2
|
-
export * from
|
|
3
|
-
export * from
|
|
4
|
-
export * from
|
|
5
|
-
export * from
|
|
6
|
-
export * from
|
|
7
|
-
export * from
|
|
8
|
-
export * from
|
|
9
|
-
export * from
|
|
10
|
-
export * from
|
|
11
|
-
export * from
|
|
12
|
-
export * from
|
|
13
|
-
export * from
|
|
14
|
-
export * from
|
|
15
|
-
export * from
|
|
16
|
-
export * from
|
|
17
|
-
export * from
|
|
18
|
-
export * from
|
|
19
|
-
export * from
|
|
20
|
-
export * from
|
|
21
|
-
export { groupBy, groupByOperators } from
|
|
1
|
+
export * from "./pipe.cjs";
|
|
2
|
+
export * from "./map.cjs";
|
|
3
|
+
export * from "./tap.cjs";
|
|
4
|
+
export * from "./filter.cjs";
|
|
5
|
+
export * from "./negate.cjs";
|
|
6
|
+
export * from "./concat.cjs";
|
|
7
|
+
export * from "./debug.cjs";
|
|
8
|
+
export * from "./output.cjs";
|
|
9
|
+
export * from "./consolidate.cjs";
|
|
10
|
+
export * from "./join.cjs";
|
|
11
|
+
export * from "./reduce.cjs";
|
|
12
|
+
export * from "./count.cjs";
|
|
13
|
+
export * from "./distinct.cjs";
|
|
14
|
+
export * from "./keying.cjs";
|
|
15
|
+
export * from "./topK.cjs";
|
|
16
|
+
export * from "./topKWithFractionalIndex.cjs";
|
|
17
|
+
export * from "./groupedTopKWithFractionalIndex.cjs";
|
|
18
|
+
export * from "./orderBy.cjs";
|
|
19
|
+
export * from "./groupedOrderBy.cjs";
|
|
20
|
+
export * from "./filterBy.cjs";
|
|
21
|
+
export { groupBy, groupByOperators } from "./groupBy.cjs";
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"join.cjs","sources":["../../../src/operators/join.ts"],"sourcesContent":["/**\n * # Direct Join Algorithms for Incremental View Maintenance\n *\n * High-performance join operations implementing all join types (inner, left, right, full, anti)\n * with minimal state and optimized performance.\n *\n * ## Algorithm\n *\n * For each tick, the algorithm processes incoming changes (deltas) and emits join results:\n *\n * 1. **Build deltas**: Create delta indexes from input messages using `Index.fromMultiSet()`\n * 2. **Inner results**: Emit `ΔA⋈B_old + A_old⋈ΔB + ΔA⋈ΔB` (matched pairs)\n * 3. **Outer results**: For unmatched rows, emit null-extended tuples:\n * - New unmatched rows from deltas (when opposite side empty)\n * - Presence transitions: when key goes `0→>0` (retract nulls) or `>0→0` (emit nulls)\n * 4. **Update state**: Append deltas to indexes (consolidated multiplicity tracking automatic)\n *\n * **Consolidated multiplicity tracking** enables O(1) presence checks instead of scanning index buckets.\n *\n * ## State\n *\n * **Indexes** store the actual data:\n * - `indexA: Index<K, V1>` - all left-side rows accumulated over time\n * - `indexB: Index<K, V2>` - all right-side rows accumulated over time\n *\n * **Consolidated multiplicity tracking** (built into Index):\n * - Each Index maintains sum of multiplicities per key internally\n * - Provides O(1) presence checks: `index.hasPresence(key)` and `index.getConsolidatedMultiplicity(key)`\n * - Avoids scanning entire index buckets just to check if key has any rows\n *\n * ## Join Types\n *\n * - **Inner**: Standard delta terms only\n * - **Outer**: Inner results + null-extended unmatched rows with transition handling\n * - **Anti**: Unmatched rows only (no inner results)\n *\n * ## Key Optimizations\n *\n * - **No temp copying**: Uses `(A⊎ΔA)⋈ΔB = A⋈ΔB ⊎ ΔA⋈ΔB` distributive property\n * - **Early-out checks**: Skip phases when no deltas present\n * - **Zero-entry pruning**: Keep maps compact, O(distinct keys) memory\n * - **Final presence logic**: Avoid emit→retract churn within same tick\n *\n * ## Correctness\n *\n * - **Ordering**: Pre-append snapshots for emissions, post-emit state updates\n * - **Presence**: Key matched iff mass ≠ 0, transitions trigger null handling\n * - **Bag semantics**: Proper multiplicity handling including negatives\n */\n\nimport { BinaryOperator, DifferenceStreamWriter } from '../graph.js'\nimport { StreamBuilder } from '../d2.js'\nimport { MultiSet } from '../multiset.js'\nimport { Index } from '../indexes.js'\nimport type { DifferenceStreamReader } from '../graph.js'\nimport type { IStreamBuilder, KeyValue, PipedOperator } from '../types.js'\n\n/**\n * Type of join to perform\n */\nexport type JoinType = `inner` | `left` | `right` | `full` | `anti`\n\n/**\n * Operator that joins two input streams using direct join algorithms\n */\nexport class JoinOperator<K, V1, V2> extends BinaryOperator<\n [K, V1] | [K, V2] | [K, [V1, V2]] | [K, [V1 | null, V2 | null]]\n> {\n #indexA = new Index<K, V1>()\n #indexB = new Index<K, V2>()\n #mode: JoinType\n\n constructor(\n id: number,\n inputA: DifferenceStreamReader<[K, V1]>,\n inputB: DifferenceStreamReader<[K, V2]>,\n output: DifferenceStreamWriter<any>,\n mode: JoinType = `inner`,\n ) {\n super(id, inputA, inputB, output)\n this.#mode = mode\n }\n\n run(): void {\n // Build deltas from input messages\n const deltaA = Index.fromMultiSets<K, V1>(\n this.inputAMessages() as Array<MultiSet<[K, V1]>>,\n )\n const deltaB = Index.fromMultiSets<K, V2>(\n this.inputBMessages() as Array<MultiSet<[K, V2]>>,\n )\n\n // Early-out if nothing changed\n if (deltaA.size === 0 && deltaB.size === 0) return\n\n const results = new MultiSet<any>()\n\n // Emit inner results (all modes except anti)\n if (this.#mode !== `anti`) {\n this.emitInnerResults(deltaA, deltaB, results)\n }\n\n // Emit left outer/anti results\n if (\n this.#mode === `left` ||\n this.#mode === `full` ||\n this.#mode === `anti`\n ) {\n this.emitLeftOuterResults(deltaA, deltaB, results)\n }\n\n // Emit right outer results\n if (this.#mode === `right` || this.#mode === `full`) {\n this.emitRightOuterResults(deltaA, deltaB, results)\n }\n\n // Update state and send results\n // IMPORTANT: All emissions use pre-append snapshots of indexA/indexB.\n // Now append ALL deltas to indices - this happens unconditionally for every key,\n // regardless of whether presence flipped. Consolidated multiplicity tracking is automatic.\n this.#indexA.append(deltaA)\n this.#indexB.append(deltaB)\n\n // Send results\n if (results.getInner().length > 0) {\n this.output.sendData(results)\n }\n }\n\n private emitInnerResults(\n deltaA: Index<K, V1>,\n deltaB: Index<K, V2>,\n results: MultiSet<any>,\n ): void {\n // Emit the three standard delta terms: ΔA⋈B_old, A_old⋈ΔB, ΔA⋈ΔB\n if (deltaA.size > 0) results.extend(deltaA.join(this.#indexB))\n if (deltaB.size > 0) results.extend(this.#indexA.join(deltaB))\n if (deltaA.size > 0 && deltaB.size > 0) results.extend(deltaA.join(deltaB))\n }\n\n private emitLeftOuterResults(\n deltaA: Index<K, V1>,\n deltaB: Index<K, V2>,\n results: MultiSet<any>,\n ): void {\n // Emit unmatched left rows from deltaA\n if (deltaA.size > 0) {\n for (const [key, valueIterator] of deltaA.entriesIterators()) {\n const currentMultiplicityB =\n this.#indexB.getConsolidatedMultiplicity(key)\n const deltaMultiplicityB = deltaB.getConsolidatedMultiplicity(key)\n const finalMultiplicityB = currentMultiplicityB + deltaMultiplicityB\n\n if (finalMultiplicityB === 0) {\n for (const [value, multiplicity] of valueIterator) {\n if (multiplicity !== 0) {\n results.add([key, [value, null]], multiplicity)\n }\n }\n }\n }\n }\n\n // Handle presence transitions from right side changes\n if (deltaB.size > 0) {\n for (const key of deltaB.getPresenceKeys()) {\n const before = this.#indexB.getConsolidatedMultiplicity(key)\n const deltaMult = deltaB.getConsolidatedMultiplicity(key)\n if (deltaMult === 0) continue\n const after = before + deltaMult\n\n // Skip transition handling if presence doesn't flip (both zero or both non-zero)\n // Note: Index updates happen later regardless - we're only skipping null-extension emissions here\n if ((before === 0) === (after === 0)) continue\n\n // Determine the type of transition:\n // - 0 → non-zero: Right becomes non-empty, left rows transition from unmatched to matched\n // → RETRACT previously emitted null-extended rows (emit with negative multiplicity)\n // - non-zero → 0: Right becomes empty, left rows transition from matched to unmatched\n // → EMIT new null-extended rows (emit with positive multiplicity)\n const transitioningToMatched = before === 0\n\n for (const [value, multiplicity] of this.#indexA.getIterator(key)) {\n if (multiplicity !== 0) {\n results.add(\n [key, [value, null]],\n transitioningToMatched ? -multiplicity : +multiplicity,\n )\n }\n }\n }\n }\n }\n\n private emitRightOuterResults(\n deltaA: Index<K, V1>,\n deltaB: Index<K, V2>,\n results: MultiSet<any>,\n ): void {\n // Emit unmatched right rows from deltaB\n if (deltaB.size > 0) {\n for (const [key, valueIterator] of deltaB.entriesIterators()) {\n const currentMultiplicityA =\n this.#indexA.getConsolidatedMultiplicity(key)\n const deltaMultiplicityA = deltaA.getConsolidatedMultiplicity(key)\n const finalMultiplicityA = currentMultiplicityA + deltaMultiplicityA\n\n if (finalMultiplicityA === 0) {\n for (const [value, multiplicity] of valueIterator) {\n if (multiplicity !== 0) {\n results.add([key, [null, value]], multiplicity)\n }\n }\n }\n }\n }\n\n // Handle presence transitions from left side changes\n if (deltaA.size > 0) {\n for (const key of deltaA.getPresenceKeys()) {\n const before = this.#indexA.getConsolidatedMultiplicity(key)\n const deltaMult = deltaA.getConsolidatedMultiplicity(key)\n if (deltaMult === 0) continue\n const after = before + deltaMult\n\n // Skip transition handling if presence doesn't flip (both zero or both non-zero)\n // Note: Index updates happen later regardless - we're only skipping null-extension emissions here\n if ((before === 0) === (after === 0)) continue\n\n // Determine the type of transition:\n // - 0 → non-zero: Left becomes non-empty, right rows transition from unmatched to matched\n // → RETRACT previously emitted null-extended rows (emit with negative multiplicity)\n // - non-zero → 0: Left becomes empty, right rows transition from matched to unmatched\n // → EMIT new null-extended rows (emit with positive multiplicity)\n const transitioningToMatched = before === 0\n\n for (const [value, multiplicity] of this.#indexB.getIterator(key)) {\n if (multiplicity !== 0) {\n results.add(\n [key, [null, value]],\n transitioningToMatched ? -multiplicity : +multiplicity,\n )\n }\n }\n }\n }\n }\n}\n\n/**\n * Joins two input streams\n * @param other - The other stream to join with\n * @param type - The type of join to perform\n */\nexport function join<\n K,\n V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never,\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n type: JoinType = `inner`,\n): PipedOperator<T, KeyValue<K, [V1 | null, V2 | null]>> {\n return (\n stream: IStreamBuilder<T>,\n ): IStreamBuilder<KeyValue<K, [V1 | null, V2 | null]>> => {\n if (stream.graph !== other.graph) {\n throw new Error(`Cannot join streams from different graphs`)\n }\n const output = new StreamBuilder<KeyValue<K, [V1 | null, V2 | null]>>(\n stream.graph,\n new DifferenceStreamWriter<KeyValue<K, [V1 | null, V2 | null]>>(),\n )\n const operator = new JoinOperator<K, V1, V2>(\n stream.graph.getNextOperatorId(),\n stream.connectReader() as DifferenceStreamReader<KeyValue<K, V1>>,\n other.connectReader(),\n output.writer,\n type,\n )\n stream.graph.addOperator(operator)\n return output\n }\n}\n\n/**\n * Joins two input streams (inner join)\n * @param other - The other stream to join with\n */\nexport function innerJoin<\n K,\n V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never,\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1, V2]>> {\n return join(other, `inner`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1, V2]>\n >\n}\n\n/**\n * Joins two input streams (anti join)\n * @param other - The other stream to join with\n */\nexport function antiJoin<\n K,\n V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never,\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1, null]>> {\n return join(other, `anti`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1, null]>\n >\n}\n\n/**\n * Joins two input streams (left join)\n * @param other - The other stream to join with\n */\nexport function leftJoin<\n K,\n V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never,\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1, V2 | null]>> {\n return join(other, `left`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1, V2 | null]>\n >\n}\n\n/**\n * Joins two input streams (right join)\n * @param other - The other stream to join with\n */\nexport function rightJoin<\n K,\n V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never,\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1 | null, V2]>> {\n return join(other, `right`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1 | null, V2]>\n >\n}\n\n/**\n * Joins two input streams (full join)\n * @param other - The other stream to join with\n */\nexport function fullJoin<\n K,\n V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never,\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1 | null, V2 | null]>> {\n return join(other, `full`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1 | null, V2 | null]>\n >\n}\n"],"names":["BinaryOperator","Index","MultiSet","StreamBuilder","DifferenceStreamWriter"],"mappings":";;;;;;AAiEO,MAAM,qBAAgCA,MAAAA,eAE3C;AAAA,EACA,UAAU,IAAIC,QAAAA,MAAA;AAAA,EACd,UAAU,IAAIA,QAAAA,MAAA;AAAA,EACd;AAAA,EAEA,YACE,IACA,QACA,QACA,QACA,OAAiB,SACjB;AACA,UAAM,IAAI,QAAQ,QAAQ,MAAM;AAChC,SAAK,QAAQ;AAAA,EACf;AAAA,EAEA,MAAY;AAEV,UAAM,SAASA,QAAAA,MAAM;AAAA,MACnB,KAAK,eAAA;AAAA,IAAe;AAEtB,UAAM,SAASA,QAAAA,MAAM;AAAA,MACnB,KAAK,eAAA;AAAA,IAAe;AAItB,QAAI,OAAO,SAAS,KAAK,OAAO,SAAS,EAAG;AAE5C,UAAM,UAAU,IAAIC,kBAAA;AAGpB,QAAI,KAAK,UAAU,QAAQ;AACzB,WAAK,iBAAiB,QAAQ,QAAQ,OAAO;AAAA,IAC/C;AAGA,QACE,KAAK,UAAU,UACf,KAAK,UAAU,UACf,KAAK,UAAU,QACf;AACA,WAAK,qBAAqB,QAAQ,QAAQ,OAAO;AAAA,IACnD;AAGA,QAAI,KAAK,UAAU,WAAW,KAAK,UAAU,QAAQ;AACnD,WAAK,sBAAsB,QAAQ,QAAQ,OAAO;AAAA,IACpD;AAMA,SAAK,QAAQ,OAAO,MAAM;AAC1B,SAAK,QAAQ,OAAO,MAAM;AAG1B,QAAI,QAAQ,WAAW,SAAS,GAAG;AACjC,WAAK,OAAO,SAAS,OAAO;AAAA,IAC9B;AAAA,EACF;AAAA,EAEQ,iBACN,QACA,QACA,SACM;AAEN,QAAI,OAAO,OAAO,EAAG,SAAQ,OAAO,OAAO,KAAK,KAAK,OAAO,CAAC;AAC7D,QAAI,OAAO,OAAO,EAAG,SAAQ,OAAO,KAAK,QAAQ,KAAK,MAAM,CAAC;AAC7D,QAAI,OAAO,OAAO,KAAK,OAAO,OAAO,EAAG,SAAQ,OAAO,OAAO,KAAK,MAAM,CAAC;AAAA,EAC5E;AAAA,EAEQ,qBACN,QACA,QACA,SACM;AAEN,QAAI,OAAO,OAAO,GAAG;AACnB,iBAAW,CAAC,KAAK,aAAa,KAAK,OAAO,oBAAoB;AAC5D,cAAM,uBACJ,KAAK,QAAQ,4BAA4B,GAAG;AAC9C,cAAM,qBAAqB,OAAO,4BAA4B,GAAG;AACjE,cAAM,qBAAqB,uBAAuB;AAElD,YAAI,uBAAuB,GAAG;AAC5B,qBAAW,CAAC,OAAO,YAAY,KAAK,eAAe;AACjD,gBAAI,iBAAiB,GAAG;AACtB,sBAAQ,IAAI,CAAC,KAAK,CAAC,OAAO,IAAI,CAAC,GAAG,YAAY;AAAA,YAChD;AAAA,UACF;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAGA,QAAI,OAAO,OAAO,GAAG;AACnB,iBAAW,OAAO,OAAO,mBAAmB;AAC1C,cAAM,SAAS,KAAK,QAAQ,4BAA4B,GAAG;AAC3D,cAAM,YAAY,OAAO,4BAA4B,GAAG;AACxD,YAAI,cAAc,EAAG;AACrB,cAAM,QAAQ,SAAS;AAIvB,YAAK,WAAW,OAAQ,UAAU,GAAI;AAOtC,cAAM,yBAAyB,WAAW;AAE1C,mBAAW,CAAC,OAAO,YAAY,KAAK,KAAK,QAAQ,YAAY,GAAG,GAAG;AACjE,cAAI,iBAAiB,GAAG;AACtB,oBAAQ;AAAA,cACN,CAAC,KAAK,CAAC,OAAO,IAAI,CAAC;AAAA,cACnB,yBAAyB,CAAC,eAAe,CAAC;AAAA,YAAA;AAAA,UAE9C;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA,EAEQ,sBACN,QACA,QACA,SACM;AAEN,QAAI,OAAO,OAAO,GAAG;AACnB,iBAAW,CAAC,KAAK,aAAa,KAAK,OAAO,oBAAoB;AAC5D,cAAM,uBACJ,KAAK,QAAQ,4BAA4B,GAAG;AAC9C,cAAM,qBAAqB,OAAO,4BAA4B,GAAG;AACjE,cAAM,qBAAqB,uBAAuB;AAElD,YAAI,uBAAuB,GAAG;AAC5B,qBAAW,CAAC,OAAO,YAAY,KAAK,eAAe;AACjD,gBAAI,iBAAiB,GAAG;AACtB,sBAAQ,IAAI,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC,GAAG,YAAY;AAAA,YAChD;AAAA,UACF;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAGA,QAAI,OAAO,OAAO,GAAG;AACnB,iBAAW,OAAO,OAAO,mBAAmB;AAC1C,cAAM,SAAS,KAAK,QAAQ,4BAA4B,GAAG;AAC3D,cAAM,YAAY,OAAO,4BAA4B,GAAG;AACxD,YAAI,cAAc,EAAG;AACrB,cAAM,QAAQ,SAAS;AAIvB,YAAK,WAAW,OAAQ,UAAU,GAAI;AAOtC,cAAM,yBAAyB,WAAW;AAE1C,mBAAW,CAAC,OAAO,YAAY,KAAK,KAAK,QAAQ,YAAY,GAAG,GAAG;AACjE,cAAI,iBAAiB,GAAG;AACtB,oBAAQ;AAAA,cACN,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC;AAAA,cACnB,yBAAyB,CAAC,eAAe,CAAC;AAAA,YAAA;AAAA,UAE9C;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAAA,EACF;AACF;AAOO,SAAS,KAMd,OACA,OAAiB,SACsC;AACvD,SAAO,CACL,WACwD;AACxD,QAAI,OAAO,UAAU,MAAM,OAAO;AAChC,YAAM,IAAI,MAAM,2CAA2C;AAAA,IAC7D;AACA,UAAM,SAAS,IAAIC,GAAAA;AAAAA,MACjB,OAAO;AAAA,MACP,IAAIC,MAAAA,uBAAA;AAAA,IAA4D;AAElE,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,MAAM,cAAA;AAAA,MACN,OAAO;AAAA,MACP;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;AAMO,SAAS,UAMd,OACyC;AACzC,SAAO,KAAK,OAAO,OAAO;AAI5B;AAMO,SAAS,SAMd,OAC2C;AAC3C,SAAO,KAAK,OAAO,MAAM;AAI3B;AAMO,SAAS,SAMd,OACgD;AAChD,SAAO,KAAK,OAAO,MAAM;AAI3B;AAMO,SAAS,UAMd,OACgD;AAChD,SAAO,KAAK,OAAO,OAAO;AAI5B;AAMO,SAAS,SAMd,OACuD;AACvD,SAAO,KAAK,OAAO,MAAM;AAI3B;;;;;;;;"}
|
|
1
|
+
{"version":3,"file":"join.cjs","sources":["../../../src/operators/join.ts"],"sourcesContent":["/**\n * # Direct Join Algorithms for Incremental View Maintenance\n *\n * High-performance join operations implementing all join types (inner, left, right, full, anti)\n * with minimal state and optimized performance.\n *\n * ## Algorithm\n *\n * For each tick, the algorithm processes incoming changes (deltas) and emits join results:\n *\n * 1. **Build deltas**: Create delta indexes from input messages using `Index.fromMultiSet()`\n * 2. **Inner results**: Emit `ΔA⋈B_old + A_old⋈ΔB + ΔA⋈ΔB` (matched pairs)\n * 3. **Outer results**: For unmatched rows, emit null-extended tuples:\n * - New unmatched rows from deltas (when opposite side empty)\n * - Presence transitions: when key goes `0→>0` (retract nulls) or `>0→0` (emit nulls)\n * 4. **Update state**: Append deltas to indexes (consolidated multiplicity tracking automatic)\n *\n * **Consolidated multiplicity tracking** enables O(1) presence checks instead of scanning index buckets.\n *\n * ## State\n *\n * **Indexes** store the actual data:\n * - `indexA: Index<K, V1>` - all left-side rows accumulated over time\n * - `indexB: Index<K, V2>` - all right-side rows accumulated over time\n *\n * **Consolidated multiplicity tracking** (built into Index):\n * - Each Index maintains sum of multiplicities per key internally\n * - Provides O(1) presence checks: `index.hasPresence(key)` and `index.getConsolidatedMultiplicity(key)`\n * - Avoids scanning entire index buckets just to check if key has any rows\n *\n * ## Join Types\n *\n * - **Inner**: Standard delta terms only\n * - **Outer**: Inner results + null-extended unmatched rows with transition handling\n * - **Anti**: Unmatched rows only (no inner results)\n *\n * ## Key Optimizations\n *\n * - **No temp copying**: Uses `(A⊎ΔA)⋈ΔB = A⋈ΔB ⊎ ΔA⋈ΔB` distributive property\n * - **Early-out checks**: Skip phases when no deltas present\n * - **Zero-entry pruning**: Keep maps compact, O(distinct keys) memory\n * - **Final presence logic**: Avoid emit→retract churn within same tick\n *\n * ## Correctness\n *\n * - **Ordering**: Pre-append snapshots for emissions, post-emit state updates\n * - **Presence**: Key matched iff mass ≠ 0, transitions trigger null handling\n * - **Bag semantics**: Proper multiplicity handling including negatives\n */\n\nimport { BinaryOperator, DifferenceStreamWriter } from '../graph.js'\nimport { StreamBuilder } from '../d2.js'\nimport { MultiSet } from '../multiset.js'\nimport { Index } from '../indexes.js'\nimport type { DifferenceStreamReader } from '../graph.js'\nimport type { IStreamBuilder, KeyValue, PipedOperator } from '../types.js'\n\n/**\n * Type of join to perform\n */\nexport type JoinType = `inner` | `left` | `right` | `full` | `anti`\n\n/**\n * Operator that joins two input streams using direct join algorithms\n */\nexport class JoinOperator<K, V1, V2> extends BinaryOperator<\n [K, V1] | [K, V2] | [K, [V1, V2]] | [K, [V1 | null, V2 | null]]\n> {\n #indexA = new Index<K, V1>()\n #indexB = new Index<K, V2>()\n #mode: JoinType\n\n constructor(\n id: number,\n inputA: DifferenceStreamReader<[K, V1]>,\n inputB: DifferenceStreamReader<[K, V2]>,\n output: DifferenceStreamWriter<any>,\n mode: JoinType = `inner`,\n ) {\n super(id, inputA, inputB, output)\n this.#mode = mode\n }\n\n run(): void {\n // Build deltas from input messages\n const deltaA = Index.fromMultiSets<K, V1>(\n this.inputAMessages() as Array<MultiSet<[K, V1]>>,\n )\n const deltaB = Index.fromMultiSets<K, V2>(\n this.inputBMessages() as Array<MultiSet<[K, V2]>>,\n )\n\n // Early-out if nothing changed\n if (deltaA.size === 0 && deltaB.size === 0) return\n\n const results = new MultiSet<any>()\n\n // Emit inner results (all modes except anti)\n if (this.#mode !== `anti`) {\n this.emitInnerResults(deltaA, deltaB, results)\n }\n\n // Emit left outer/anti results\n if (\n this.#mode === `left` ||\n this.#mode === `full` ||\n this.#mode === `anti`\n ) {\n this.emitLeftOuterResults(deltaA, deltaB, results)\n }\n\n // Emit right outer results\n if (this.#mode === `right` || this.#mode === `full`) {\n this.emitRightOuterResults(deltaA, deltaB, results)\n }\n\n // Update state and send results\n // IMPORTANT: All emissions use pre-append snapshots of indexA/indexB.\n // Now append ALL deltas to indices - this happens unconditionally for every key,\n // regardless of whether presence flipped. Consolidated multiplicity tracking is automatic.\n this.#indexA.append(deltaA)\n this.#indexB.append(deltaB)\n\n // Send results\n if (results.getInner().length > 0) {\n this.output.sendData(results)\n }\n }\n\n private emitInnerResults(\n deltaA: Index<K, V1>,\n deltaB: Index<K, V2>,\n results: MultiSet<any>,\n ): void {\n // Emit the three standard delta terms: ΔA⋈B_old, A_old⋈ΔB, ΔA⋈ΔB\n if (deltaA.size > 0) results.extend(deltaA.join(this.#indexB))\n if (deltaB.size > 0) results.extend(this.#indexA.join(deltaB))\n if (deltaA.size > 0 && deltaB.size > 0) results.extend(deltaA.join(deltaB))\n }\n\n private emitLeftOuterResults(\n deltaA: Index<K, V1>,\n deltaB: Index<K, V2>,\n results: MultiSet<any>,\n ): void {\n // Emit unmatched left rows from deltaA\n if (deltaA.size > 0) {\n for (const [key, valueIterator] of deltaA.entriesIterators()) {\n const currentMultiplicityB =\n this.#indexB.getConsolidatedMultiplicity(key)\n const deltaMultiplicityB = deltaB.getConsolidatedMultiplicity(key)\n const finalMultiplicityB = currentMultiplicityB + deltaMultiplicityB\n\n if (finalMultiplicityB === 0) {\n for (const [value, multiplicity] of valueIterator) {\n if (multiplicity !== 0) {\n results.add([key, [value, null]], multiplicity)\n }\n }\n }\n }\n }\n\n // Handle presence transitions from right side changes\n if (deltaB.size > 0) {\n for (const key of deltaB.getPresenceKeys()) {\n const before = this.#indexB.getConsolidatedMultiplicity(key)\n const deltaMult = deltaB.getConsolidatedMultiplicity(key)\n if (deltaMult === 0) continue\n const after = before + deltaMult\n\n // Skip transition handling if presence doesn't flip (both zero or both non-zero)\n // Note: Index updates happen later regardless - we're only skipping null-extension emissions here\n if ((before === 0) === (after === 0)) continue\n\n // Determine the type of transition:\n // - 0 → non-zero: Right becomes non-empty, left rows transition from unmatched to matched\n // → RETRACT previously emitted null-extended rows (emit with negative multiplicity)\n // - non-zero → 0: Right becomes empty, left rows transition from matched to unmatched\n // → EMIT new null-extended rows (emit with positive multiplicity)\n const transitioningToMatched = before === 0\n\n for (const [value, multiplicity] of this.#indexA.getIterator(key)) {\n if (multiplicity !== 0) {\n results.add(\n [key, [value, null]],\n transitioningToMatched ? -multiplicity : +multiplicity,\n )\n }\n }\n }\n }\n }\n\n private emitRightOuterResults(\n deltaA: Index<K, V1>,\n deltaB: Index<K, V2>,\n results: MultiSet<any>,\n ): void {\n // Emit unmatched right rows from deltaB\n if (deltaB.size > 0) {\n for (const [key, valueIterator] of deltaB.entriesIterators()) {\n const currentMultiplicityA =\n this.#indexA.getConsolidatedMultiplicity(key)\n const deltaMultiplicityA = deltaA.getConsolidatedMultiplicity(key)\n const finalMultiplicityA = currentMultiplicityA + deltaMultiplicityA\n\n if (finalMultiplicityA === 0) {\n for (const [value, multiplicity] of valueIterator) {\n if (multiplicity !== 0) {\n results.add([key, [null, value]], multiplicity)\n }\n }\n }\n }\n }\n\n // Handle presence transitions from left side changes\n if (deltaA.size > 0) {\n for (const key of deltaA.getPresenceKeys()) {\n const before = this.#indexA.getConsolidatedMultiplicity(key)\n const deltaMult = deltaA.getConsolidatedMultiplicity(key)\n if (deltaMult === 0) continue\n const after = before + deltaMult\n\n // Skip transition handling if presence doesn't flip (both zero or both non-zero)\n // Note: Index updates happen later regardless - we're only skipping null-extension emissions here\n if ((before === 0) === (after === 0)) continue\n\n // Determine the type of transition:\n // - 0 → non-zero: Left becomes non-empty, right rows transition from unmatched to matched\n // → RETRACT previously emitted null-extended rows (emit with negative multiplicity)\n // - non-zero → 0: Left becomes empty, right rows transition from matched to unmatched\n // → EMIT new null-extended rows (emit with positive multiplicity)\n const transitioningToMatched = before === 0\n\n for (const [value, multiplicity] of this.#indexB.getIterator(key)) {\n if (multiplicity !== 0) {\n results.add(\n [key, [null, value]],\n transitioningToMatched ? -multiplicity : +multiplicity,\n )\n }\n }\n }\n }\n }\n}\n\n/**\n * Joins two input streams\n * @param other - The other stream to join with\n * @param type - The type of join to perform\n */\nexport function join<\n K,\n V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never),\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n type: JoinType = `inner`,\n): PipedOperator<T, KeyValue<K, [V1 | null, V2 | null]>> {\n return (\n stream: IStreamBuilder<T>,\n ): IStreamBuilder<KeyValue<K, [V1 | null, V2 | null]>> => {\n if (stream.graph !== other.graph) {\n throw new Error(`Cannot join streams from different graphs`)\n }\n const output = new StreamBuilder<KeyValue<K, [V1 | null, V2 | null]>>(\n stream.graph,\n new DifferenceStreamWriter<KeyValue<K, [V1 | null, V2 | null]>>(),\n )\n const operator = new JoinOperator<K, V1, V2>(\n stream.graph.getNextOperatorId(),\n stream.connectReader() as DifferenceStreamReader<KeyValue<K, V1>>,\n other.connectReader(),\n output.writer,\n type,\n )\n stream.graph.addOperator(operator)\n return output\n }\n}\n\n/**\n * Joins two input streams (inner join)\n * @param other - The other stream to join with\n */\nexport function innerJoin<\n K,\n V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never),\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1, V2]>> {\n return join(other, `inner`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1, V2]>\n >\n}\n\n/**\n * Joins two input streams (anti join)\n * @param other - The other stream to join with\n */\nexport function antiJoin<\n K,\n V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never),\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1, null]>> {\n return join(other, `anti`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1, null]>\n >\n}\n\n/**\n * Joins two input streams (left join)\n * @param other - The other stream to join with\n */\nexport function leftJoin<\n K,\n V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never),\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1, V2 | null]>> {\n return join(other, `left`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1, V2 | null]>\n >\n}\n\n/**\n * Joins two input streams (right join)\n * @param other - The other stream to join with\n */\nexport function rightJoin<\n K,\n V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never),\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1 | null, V2]>> {\n return join(other, `right`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1 | null, V2]>\n >\n}\n\n/**\n * Joins two input streams (full join)\n * @param other - The other stream to join with\n */\nexport function fullJoin<\n K,\n V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never),\n V2,\n T,\n>(\n other: IStreamBuilder<KeyValue<K, V2>>,\n): PipedOperator<T, KeyValue<K, [V1 | null, V2 | null]>> {\n return join(other, `full`) as unknown as PipedOperator<\n T,\n KeyValue<K, [V1 | null, V2 | null]>\n >\n}\n"],"names":["BinaryOperator","Index","MultiSet","StreamBuilder","DifferenceStreamWriter"],"mappings":";;;;;;AAiEO,MAAM,qBAAgCA,MAAAA,eAE3C;AAAA,EACA,UAAU,IAAIC,QAAAA,MAAA;AAAA,EACd,UAAU,IAAIA,QAAAA,MAAA;AAAA,EACd;AAAA,EAEA,YACE,IACA,QACA,QACA,QACA,OAAiB,SACjB;AACA,UAAM,IAAI,QAAQ,QAAQ,MAAM;AAChC,SAAK,QAAQ;AAAA,EACf;AAAA,EAEA,MAAY;AAEV,UAAM,SAASA,QAAAA,MAAM;AAAA,MACnB,KAAK,eAAA;AAAA,IAAe;AAEtB,UAAM,SAASA,QAAAA,MAAM;AAAA,MACnB,KAAK,eAAA;AAAA,IAAe;AAItB,QAAI,OAAO,SAAS,KAAK,OAAO,SAAS,EAAG;AAE5C,UAAM,UAAU,IAAIC,kBAAA;AAGpB,QAAI,KAAK,UAAU,QAAQ;AACzB,WAAK,iBAAiB,QAAQ,QAAQ,OAAO;AAAA,IAC/C;AAGA,QACE,KAAK,UAAU,UACf,KAAK,UAAU,UACf,KAAK,UAAU,QACf;AACA,WAAK,qBAAqB,QAAQ,QAAQ,OAAO;AAAA,IACnD;AAGA,QAAI,KAAK,UAAU,WAAW,KAAK,UAAU,QAAQ;AACnD,WAAK,sBAAsB,QAAQ,QAAQ,OAAO;AAAA,IACpD;AAMA,SAAK,QAAQ,OAAO,MAAM;AAC1B,SAAK,QAAQ,OAAO,MAAM;AAG1B,QAAI,QAAQ,WAAW,SAAS,GAAG;AACjC,WAAK,OAAO,SAAS,OAAO;AAAA,IAC9B;AAAA,EACF;AAAA,EAEQ,iBACN,QACA,QACA,SACM;AAEN,QAAI,OAAO,OAAO,EAAG,SAAQ,OAAO,OAAO,KAAK,KAAK,OAAO,CAAC;AAC7D,QAAI,OAAO,OAAO,EAAG,SAAQ,OAAO,KAAK,QAAQ,KAAK,MAAM,CAAC;AAC7D,QAAI,OAAO,OAAO,KAAK,OAAO,OAAO,EAAG,SAAQ,OAAO,OAAO,KAAK,MAAM,CAAC;AAAA,EAC5E;AAAA,EAEQ,qBACN,QACA,QACA,SACM;AAEN,QAAI,OAAO,OAAO,GAAG;AACnB,iBAAW,CAAC,KAAK,aAAa,KAAK,OAAO,oBAAoB;AAC5D,cAAM,uBACJ,KAAK,QAAQ,4BAA4B,GAAG;AAC9C,cAAM,qBAAqB,OAAO,4BAA4B,GAAG;AACjE,cAAM,qBAAqB,uBAAuB;AAElD,YAAI,uBAAuB,GAAG;AAC5B,qBAAW,CAAC,OAAO,YAAY,KAAK,eAAe;AACjD,gBAAI,iBAAiB,GAAG;AACtB,sBAAQ,IAAI,CAAC,KAAK,CAAC,OAAO,IAAI,CAAC,GAAG,YAAY;AAAA,YAChD;AAAA,UACF;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAGA,QAAI,OAAO,OAAO,GAAG;AACnB,iBAAW,OAAO,OAAO,mBAAmB;AAC1C,cAAM,SAAS,KAAK,QAAQ,4BAA4B,GAAG;AAC3D,cAAM,YAAY,OAAO,4BAA4B,GAAG;AACxD,YAAI,cAAc,EAAG;AACrB,cAAM,QAAQ,SAAS;AAIvB,YAAK,WAAW,OAAQ,UAAU,GAAI;AAOtC,cAAM,yBAAyB,WAAW;AAE1C,mBAAW,CAAC,OAAO,YAAY,KAAK,KAAK,QAAQ,YAAY,GAAG,GAAG;AACjE,cAAI,iBAAiB,GAAG;AACtB,oBAAQ;AAAA,cACN,CAAC,KAAK,CAAC,OAAO,IAAI,CAAC;AAAA,cACnB,yBAAyB,CAAC,eAAe,CAAC;AAAA,YAAA;AAAA,UAE9C;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAAA,EACF;AAAA,EAEQ,sBACN,QACA,QACA,SACM;AAEN,QAAI,OAAO,OAAO,GAAG;AACnB,iBAAW,CAAC,KAAK,aAAa,KAAK,OAAO,oBAAoB;AAC5D,cAAM,uBACJ,KAAK,QAAQ,4BAA4B,GAAG;AAC9C,cAAM,qBAAqB,OAAO,4BAA4B,GAAG;AACjE,cAAM,qBAAqB,uBAAuB;AAElD,YAAI,uBAAuB,GAAG;AAC5B,qBAAW,CAAC,OAAO,YAAY,KAAK,eAAe;AACjD,gBAAI,iBAAiB,GAAG;AACtB,sBAAQ,IAAI,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC,GAAG,YAAY;AAAA,YAChD;AAAA,UACF;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAGA,QAAI,OAAO,OAAO,GAAG;AACnB,iBAAW,OAAO,OAAO,mBAAmB;AAC1C,cAAM,SAAS,KAAK,QAAQ,4BAA4B,GAAG;AAC3D,cAAM,YAAY,OAAO,4BAA4B,GAAG;AACxD,YAAI,cAAc,EAAG;AACrB,cAAM,QAAQ,SAAS;AAIvB,YAAK,WAAW,OAAQ,UAAU,GAAI;AAOtC,cAAM,yBAAyB,WAAW;AAE1C,mBAAW,CAAC,OAAO,YAAY,KAAK,KAAK,QAAQ,YAAY,GAAG,GAAG;AACjE,cAAI,iBAAiB,GAAG;AACtB,oBAAQ;AAAA,cACN,CAAC,KAAK,CAAC,MAAM,KAAK,CAAC;AAAA,cACnB,yBAAyB,CAAC,eAAe,CAAC;AAAA,YAAA;AAAA,UAE9C;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAAA,EACF;AACF;AAOO,SAAS,KAMd,OACA,OAAiB,SACsC;AACvD,SAAO,CACL,WACwD;AACxD,QAAI,OAAO,UAAU,MAAM,OAAO;AAChC,YAAM,IAAI,MAAM,2CAA2C;AAAA,IAC7D;AACA,UAAM,SAAS,IAAIC,GAAAA;AAAAA,MACjB,OAAO;AAAA,MACP,IAAIC,MAAAA,uBAAA;AAAA,IAA4D;AAElE,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,MAAM,cAAA;AAAA,MACN,OAAO;AAAA,MACP;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;AAMO,SAAS,UAMd,OACyC;AACzC,SAAO,KAAK,OAAO,OAAO;AAI5B;AAMO,SAAS,SAMd,OAC2C;AAC3C,SAAO,KAAK,OAAO,MAAM;AAI3B;AAMO,SAAS,SAMd,OACgD;AAChD,SAAO,KAAK,OAAO,MAAM;AAI3B;AAMO,SAAS,UAMd,OACgD;AAChD,SAAO,KAAK,OAAO,OAAO;AAI5B;AAMO,SAAS,SAMd,OACuD;AACvD,SAAO,KAAK,OAAO,MAAM;AAI3B;;;;;;;;"}
|
|
@@ -1,5 +1,5 @@
|
|
|
1
|
-
import { BinaryOperator, DifferenceStreamWriter, DifferenceStreamReader } from
|
|
2
|
-
import { IStreamBuilder, KeyValue, PipedOperator } from
|
|
1
|
+
import { BinaryOperator, DifferenceStreamWriter, DifferenceStreamReader } from "../graph.cjs";
|
|
2
|
+
import { IStreamBuilder, KeyValue, PipedOperator } from "../types.cjs";
|
|
3
3
|
/**
|
|
4
4
|
* Type of join to perform
|
|
5
5
|
*/
|
|
@@ -10,9 +10,30 @@ export type JoinType = `inner` | `left` | `right` | `full` | `anti`;
|
|
|
10
10
|
export declare class JoinOperator<K, V1, V2> extends BinaryOperator<[
|
|
11
11
|
K,
|
|
12
12
|
V1
|
|
13
|
-
] | [
|
|
13
|
+
] | [
|
|
14
|
+
K,
|
|
15
|
+
V2
|
|
16
|
+
] | [
|
|
17
|
+
K,
|
|
18
|
+
[
|
|
19
|
+
V1,
|
|
20
|
+
V2
|
|
21
|
+
]
|
|
22
|
+
] | [
|
|
23
|
+
K,
|
|
24
|
+
[
|
|
25
|
+
V1 | null,
|
|
26
|
+
V2 | null
|
|
27
|
+
]
|
|
28
|
+
]> {
|
|
14
29
|
#private;
|
|
15
|
-
constructor(id: number, inputA: DifferenceStreamReader<[
|
|
30
|
+
constructor(id: number, inputA: DifferenceStreamReader<[
|
|
31
|
+
K,
|
|
32
|
+
V1
|
|
33
|
+
]>, inputB: DifferenceStreamReader<[
|
|
34
|
+
K,
|
|
35
|
+
V2
|
|
36
|
+
]>, output: DifferenceStreamWriter<any>, mode?: JoinType);
|
|
16
37
|
run(): void;
|
|
17
38
|
private emitInnerResults;
|
|
18
39
|
private emitLeftOuterResults;
|
|
@@ -23,29 +44,47 @@ export declare class JoinOperator<K, V1, V2> extends BinaryOperator<[
|
|
|
23
44
|
* @param other - The other stream to join with
|
|
24
45
|
* @param type - The type of join to perform
|
|
25
46
|
*/
|
|
26
|
-
export declare function join<K, V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never, V2, T>(other: IStreamBuilder<KeyValue<K, V2>>, type?: JoinType): PipedOperator<T, KeyValue<K, [
|
|
47
|
+
export declare function join<K, V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never), V2, T>(other: IStreamBuilder<KeyValue<K, V2>>, type?: JoinType): PipedOperator<T, KeyValue<K, [
|
|
48
|
+
V1 | null,
|
|
49
|
+
V2 | null
|
|
50
|
+
]>>;
|
|
27
51
|
/**
|
|
28
52
|
* Joins two input streams (inner join)
|
|
29
53
|
* @param other - The other stream to join with
|
|
30
54
|
*/
|
|
31
|
-
export declare function innerJoin<K, V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never, V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
55
|
+
export declare function innerJoin<K, V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never), V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
56
|
+
V1,
|
|
57
|
+
V2
|
|
58
|
+
]>>;
|
|
32
59
|
/**
|
|
33
60
|
* Joins two input streams (anti join)
|
|
34
61
|
* @param other - The other stream to join with
|
|
35
62
|
*/
|
|
36
|
-
export declare function antiJoin<K, V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never, V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
63
|
+
export declare function antiJoin<K, V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never), V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
64
|
+
V1,
|
|
65
|
+
null
|
|
66
|
+
]>>;
|
|
37
67
|
/**
|
|
38
68
|
* Joins two input streams (left join)
|
|
39
69
|
* @param other - The other stream to join with
|
|
40
70
|
*/
|
|
41
|
-
export declare function leftJoin<K, V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never, V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
71
|
+
export declare function leftJoin<K, V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never), V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
72
|
+
V1,
|
|
73
|
+
V2 | null
|
|
74
|
+
]>>;
|
|
42
75
|
/**
|
|
43
76
|
* Joins two input streams (right join)
|
|
44
77
|
* @param other - The other stream to join with
|
|
45
78
|
*/
|
|
46
|
-
export declare function rightJoin<K, V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never, V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
79
|
+
export declare function rightJoin<K, V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never), V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
80
|
+
V1 | null,
|
|
81
|
+
V2
|
|
82
|
+
]>>;
|
|
47
83
|
/**
|
|
48
84
|
* Joins two input streams (full join)
|
|
49
85
|
* @param other - The other stream to join with
|
|
50
86
|
*/
|
|
51
|
-
export declare function fullJoin<K, V1 extends T extends KeyValue<infer _KT, infer VT> ? VT : never, V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
87
|
+
export declare function fullJoin<K, V1 extends (T extends KeyValue<infer _KT, infer VT> ? VT : never), V2, T>(other: IStreamBuilder<KeyValue<K, V2>>): PipedOperator<T, KeyValue<K, [
|
|
88
|
+
V1 | null,
|
|
89
|
+
V2 | null
|
|
90
|
+
]>>;
|
|
@@ -1,8 +1,11 @@
|
|
|
1
|
-
import { PipedOperator } from
|
|
1
|
+
import { PipedOperator } from "../types.cjs";
|
|
2
2
|
/**
|
|
3
3
|
* Type for a keyed value
|
|
4
4
|
*/
|
|
5
|
-
export type Keyed<K, V> = [
|
|
5
|
+
export type Keyed<K, V> = [
|
|
6
|
+
K,
|
|
7
|
+
V
|
|
8
|
+
];
|
|
6
9
|
/**
|
|
7
10
|
* Takes an unkeyed input and returns a keyed result
|
|
8
11
|
* @param keyFn - Function to generate the key for each value
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { DifferenceStreamWriter, LinearUnaryOperator, DifferenceStreamReader } from
|
|
2
|
-
import { PipedOperator } from
|
|
3
|
-
import { MultiSet } from
|
|
1
|
+
import { DifferenceStreamWriter, LinearUnaryOperator, DifferenceStreamReader } from "../graph.cjs";
|
|
2
|
+
import { PipedOperator } from "../types.cjs";
|
|
3
|
+
import { MultiSet } from "../multiset.cjs";
|
|
4
4
|
/**
|
|
5
5
|
* Operator that applies a function to each element in the input stream
|
|
6
6
|
*/
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { LinearUnaryOperator } from
|
|
2
|
-
import { PipedOperator } from
|
|
3
|
-
import { MultiSet } from
|
|
1
|
+
import { LinearUnaryOperator } from "../graph.cjs";
|
|
2
|
+
import { PipedOperator } from "../types.cjs";
|
|
3
|
+
import { MultiSet } from "../multiset.cjs";
|
|
4
4
|
/**
|
|
5
5
|
* Operator that negates the multiplicities in the input stream
|
|
6
6
|
*/
|
|
@@ -1,5 +1,5 @@
|
|
|
1
|
-
import { topKWithFractionalIndex } from
|
|
2
|
-
import { IStreamBuilder, KeyValue } from
|
|
1
|
+
import { topKWithFractionalIndex } from "./topKWithFractionalIndex.cjs";
|
|
2
|
+
import { IStreamBuilder, KeyValue } from "../types.cjs";
|
|
3
3
|
export interface OrderByOptions<Ve> {
|
|
4
4
|
comparator?: (a: Ve, b: Ve) => number;
|
|
5
5
|
limit?: number;
|
|
@@ -30,8 +30,17 @@ export declare function orderBy<T extends KeyValue<unknown, unknown>, Ve = unkno
|
|
|
30
30
|
* @param options - An optional object containing comparator, limit and offset properties
|
|
31
31
|
* @returns A piped operator that orders the elements and limits the number of results
|
|
32
32
|
*/
|
|
33
|
-
export declare function orderByWithIndex<T extends KeyValue<unknown, unknown>, Ve = unknown>(valueExtractor: (value: T extends KeyValue<unknown, infer V> ? V : never) => Ve, options?: OrderByOptions<Ve>): (stream: IStreamBuilder<T>) => IStreamBuilder<KeyValue<T extends KeyValue<infer K, unknown> ? K : never, [
|
|
34
|
-
|
|
33
|
+
export declare function orderByWithIndex<T extends KeyValue<unknown, unknown>, Ve = unknown>(valueExtractor: (value: T extends KeyValue<unknown, infer V> ? V : never) => Ve, options?: OrderByOptions<Ve>): (stream: IStreamBuilder<T>) => IStreamBuilder<KeyValue<T extends KeyValue<infer K, unknown> ? K : never, [
|
|
34
|
+
T extends KeyValue<unknown, infer V> ? V : never,
|
|
35
|
+
number
|
|
36
|
+
]>>;
|
|
37
|
+
export declare function orderByWithFractionalIndexBase<T extends KeyValue<unknown, unknown>, Ve = unknown>(topKFunction: typeof topKWithFractionalIndex, valueExtractor: (value: T extends KeyValue<unknown, infer V> ? V : never) => Ve, options?: OrderByWithFractionalIndexOptions<Ve>): (stream: IStreamBuilder<T>) => IStreamBuilder<[
|
|
38
|
+
T extends KeyValue<infer K, unknown> ? K : never,
|
|
39
|
+
[
|
|
40
|
+
T extends KeyValue<unknown, infer V> ? V : never,
|
|
41
|
+
string
|
|
42
|
+
]
|
|
43
|
+
]>;
|
|
35
44
|
/**
|
|
36
45
|
* Orders the elements and limits the number of results, with optional offset and
|
|
37
46
|
* annotates the value with a fractional index.
|
|
@@ -41,5 +50,11 @@ export declare function orderByWithFractionalIndexBase<T extends KeyValue<unknow
|
|
|
41
50
|
* @param options - An optional object containing comparator, limit and offset properties
|
|
42
51
|
* @returns A piped operator that orders the elements and limits the number of results
|
|
43
52
|
*/
|
|
44
|
-
export declare function orderByWithFractionalIndex<T extends KeyValue<unknown, unknown>, Ve = unknown>(valueExtractor: (value: T extends KeyValue<unknown, infer V> ? V : never) => Ve, options?: OrderByWithFractionalIndexOptions<Ve>): (stream: IStreamBuilder<T>) => IStreamBuilder<[
|
|
53
|
+
export declare function orderByWithFractionalIndex<T extends KeyValue<unknown, unknown>, Ve = unknown>(valueExtractor: (value: T extends KeyValue<unknown, infer V> ? V : never) => Ve, options?: OrderByWithFractionalIndexOptions<Ve>): (stream: IStreamBuilder<T>) => IStreamBuilder<[
|
|
54
|
+
T extends KeyValue<infer K, unknown> ? K : never,
|
|
55
|
+
[
|
|
56
|
+
T extends KeyValue<unknown, infer V> ? V : never,
|
|
57
|
+
string
|
|
58
|
+
]
|
|
59
|
+
]>;
|
|
45
60
|
export {};
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { DifferenceStreamWriter, UnaryOperator, DifferenceStreamReader } from
|
|
2
|
-
import { PipedOperator } from
|
|
3
|
-
import { MultiSet } from
|
|
1
|
+
import { DifferenceStreamWriter, UnaryOperator, DifferenceStreamReader } from "../graph.cjs";
|
|
2
|
+
import { PipedOperator } from "../types.cjs";
|
|
3
|
+
import { MultiSet } from "../multiset.cjs";
|
|
4
4
|
/**
|
|
5
5
|
* Operator that outputs the messages in the stream
|
|
6
6
|
*/
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import { PipedOperator } from
|
|
1
|
+
import { PipedOperator } from "../types.cjs";
|
|
2
2
|
export declare function pipe<T, O>(o1: PipedOperator<T, O>): PipedOperator<T, O>;
|
|
3
3
|
export declare function pipe<T, T2, O>(o1: PipedOperator<T, T2>, o2: PipedOperator<T2, O>): PipedOperator<T, O>;
|
|
4
4
|
export declare function pipe<T, T2, T3, O>(o1: PipedOperator<T, T2>, o2: PipedOperator<T2, T3>, o3: PipedOperator<T3, O>): PipedOperator<T, O>;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"reduce.cjs","sources":["../../../src/operators/reduce.ts"],"sourcesContent":["import { DifferenceStreamWriter, UnaryOperator } from '../graph.js'\nimport { StreamBuilder } from '../d2.js'\nimport { MultiSet } from '../multiset.js'\nimport { Index } from '../indexes.js'\nimport type { DifferenceStreamReader } from '../graph.js'\nimport type { IStreamBuilder, KeyValue } from '../types.js'\n\n/**\n * Base operator for reduction operations (version-free)\n */\nexport class ReduceOperator<K, V1, V2> extends UnaryOperator<[K, V1], [K, V2]> {\n #index = new Index<K, V1>()\n #indexOut = new Index<K, V2>()\n #f: (values: Array<[V1, number]>) => Array<[V2, number]>\n\n constructor(\n id: number,\n inputA: DifferenceStreamReader<[K, V1]>,\n output: DifferenceStreamWriter<[K, V2]>,\n f: (values: Array<[V1, number]>) => Array<[V2, number]>,\n ) {\n super(id, inputA, output)\n this.#f = f\n }\n\n run(): void {\n // Collect all input messages and update the index\n const keysTodo = new Set<K>()\n for (const message of this.inputMessages()) {\n for (const [item, multiplicity] of message.getInner()) {\n const [key, value] = item\n this.#index.addValue(key, [value, multiplicity])\n keysTodo.add(key)\n }\n }\n\n // For each key, compute the reduction and delta\n const result: Array<[[K, V2], number]> = []\n for (const key of keysTodo) {\n const curr = this.#index.get(key)\n const currOut = this.#indexOut.get(key)\n const out = this.#f(curr)\n\n // Create maps for current and previous outputs using values directly as keys\n const newOutputMap = new Map<V2, number>()\n const oldOutputMap = new Map<V2, number>()\n\n // Process new output\n for (const [value, multiplicity] of out) {\n const existing = newOutputMap.get(value) ?? 0\n newOutputMap.set(value, existing + multiplicity)\n }\n\n // Process previous output\n for (const [value, multiplicity] of currOut) {\n const existing = oldOutputMap.get(value) ?? 0\n oldOutputMap.set(value, existing + multiplicity)\n }\n\n // First, emit removals for old values that are no longer present\n for (const [value, multiplicity] of oldOutputMap) {\n if (!newOutputMap.has(value)) {\n // Remove the old value entirely\n result.push([[key, value], -multiplicity])\n this.#indexOut.addValue(key, [value, -multiplicity])\n }\n }\n\n // Then, emit additions for new values that are not present in old\n for (const [value, multiplicity] of newOutputMap) {\n if (!oldOutputMap.has(value)) {\n // Add the new value only if it has non-zero multiplicity\n if (multiplicity !== 0) {\n result.push([[key, value], multiplicity])\n this.#indexOut.addValue(key, [value, multiplicity])\n }\n }\n }\n\n // Finally, emit multiplicity changes for values that were present and are still present\n for (const [value, newMultiplicity] of newOutputMap) {\n const oldMultiplicity = oldOutputMap.get(value)\n if (oldMultiplicity !== undefined) {\n const delta = newMultiplicity - oldMultiplicity\n // Only emit actual changes, i.e. non-zero deltas\n if (delta !== 0) {\n result.push([[key, value], delta])\n this.#indexOut.addValue(key, [value, delta])\n }\n }\n }\n }\n\n if (result.length > 0) {\n this.output.sendData(new MultiSet(result))\n }\n }\n}\n\n/**\n * Reduces the elements in the stream by key (version-free)\n */\nexport function reduce<\n KType extends T extends KeyValue<infer K, infer _V> ? K : never,\n V1Type extends T extends KeyValue<KType, infer V> ? V : never,\n R,\n T,\n>(f: (values: Array<[V1Type, number]>) => Array<[R, number]>) {\n return (stream: IStreamBuilder<T>): IStreamBuilder<KeyValue<KType, R>> => {\n const output = new StreamBuilder<KeyValue<KType, R>>(\n stream.graph,\n new DifferenceStreamWriter<KeyValue<KType, R>>(),\n )\n const operator = new ReduceOperator<KType, V1Type, R>(\n stream.graph.getNextOperatorId(),\n stream.connectReader() as DifferenceStreamReader<KeyValue<KType, V1Type>>,\n output.writer,\n f,\n )\n stream.graph.addOperator(operator)\n return output\n }\n}\n"],"names":["UnaryOperator","Index","MultiSet","StreamBuilder","DifferenceStreamWriter"],"mappings":";;;;;;AAUO,MAAM,uBAAkCA,MAAAA,cAAgC;AAAA,EAC7E,SAAS,IAAIC,QAAAA,MAAA;AAAA,EACb,YAAY,IAAIA,QAAAA,MAAA;AAAA,EAChB;AAAA,EAEA,YACE,IACA,QACA,QACA,GACA;AACA,UAAM,IAAI,QAAQ,MAAM;AACxB,SAAK,KAAK;AAAA,EACZ;AAAA,EAEA,MAAY;AAEV,UAAM,+BAAe,IAAA;AACrB,eAAW,WAAW,KAAK,iBAAiB;AAC1C,iBAAW,CAAC,MAAM,YAAY,KAAK,QAAQ,YAAY;AACrD,cAAM,CAAC,KAAK,KAAK,IAAI;AACrB,aAAK,OAAO,SAAS,KAAK,CAAC,OAAO,YAAY,CAAC;AAC/C,iBAAS,IAAI,GAAG;AAAA,MAClB;AAAA,IACF;AAGA,UAAM,SAAmC,CAAA;AACzC,eAAW,OAAO,UAAU;AAC1B,YAAM,OAAO,KAAK,OAAO,IAAI,GAAG;AAChC,YAAM,UAAU,KAAK,UAAU,IAAI,GAAG;AACtC,YAAM,MAAM,KAAK,GAAG,IAAI;AAGxB,YAAM,mCAAmB,IAAA;AACzB,YAAM,mCAAmB,IAAA;AAGzB,iBAAW,CAAC,OAAO,YAAY,KAAK,KAAK;AACvC,cAAM,WAAW,aAAa,IAAI,KAAK,KAAK;AAC5C,qBAAa,IAAI,OAAO,WAAW,YAAY;AAAA,MACjD;AAGA,iBAAW,CAAC,OAAO,YAAY,KAAK,SAAS;AAC3C,cAAM,WAAW,aAAa,IAAI,KAAK,KAAK;AAC5C,qBAAa,IAAI,OAAO,WAAW,YAAY;AAAA,MACjD;AAGA,iBAAW,CAAC,OAAO,YAAY,KAAK,cAAc;AAChD,YAAI,CAAC,aAAa,IAAI,KAAK,GAAG;AAE5B,iBAAO,KAAK,CAAC,CAAC,KAAK,KAAK,GAAG,CAAC,YAAY,CAAC;AACzC,eAAK,UAAU,SAAS,KAAK,CAAC,OAAO,CAAC,YAAY,CAAC;AAAA,QACrD;AAAA,MACF;AAGA,iBAAW,CAAC,OAAO,YAAY,KAAK,cAAc;AAChD,YAAI,CAAC,aAAa,IAAI,KAAK,GAAG;AAE5B,cAAI,iBAAiB,GAAG;AACtB,mBAAO,KAAK,CAAC,CAAC,KAAK,KAAK,GAAG,YAAY,CAAC;AACxC,iBAAK,UAAU,SAAS,KAAK,CAAC,OAAO,YAAY,CAAC;AAAA,UACpD;AAAA,QACF;AAAA,MACF;AAGA,iBAAW,CAAC,OAAO,eAAe,KAAK,cAAc;AACnD,cAAM,kBAAkB,aAAa,IAAI,KAAK;AAC9C,YAAI,oBAAoB,QAAW;AACjC,gBAAM,QAAQ,kBAAkB;AAEhC,cAAI,UAAU,GAAG;AACf,mBAAO,KAAK,CAAC,CAAC,KAAK,KAAK,GAAG,KAAK,CAAC;AACjC,iBAAK,UAAU,SAAS,KAAK,CAAC,OAAO,KAAK,CAAC;AAAA,UAC7C;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAIC,SAAAA,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AACF;AAKO,SAAS,OAKd,GAA4D;AAC5D,SAAO,CAAC,WAAkE;AACxE,UAAM,SAAS,IAAIC,GAAAA;AAAAA,MACjB,OAAO;AAAA,MACP,IAAIC,MAAAA,uBAAA;AAAA,IAA2C;AAEjD,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,OAAO;AAAA,MACP;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;;;"}
|
|
1
|
+
{"version":3,"file":"reduce.cjs","sources":["../../../src/operators/reduce.ts"],"sourcesContent":["import { DifferenceStreamWriter, UnaryOperator } from '../graph.js'\nimport { StreamBuilder } from '../d2.js'\nimport { MultiSet } from '../multiset.js'\nimport { Index } from '../indexes.js'\nimport type { DifferenceStreamReader } from '../graph.js'\nimport type { IStreamBuilder, KeyValue } from '../types.js'\n\n/**\n * Base operator for reduction operations (version-free)\n */\nexport class ReduceOperator<K, V1, V2> extends UnaryOperator<[K, V1], [K, V2]> {\n #index = new Index<K, V1>()\n #indexOut = new Index<K, V2>()\n #f: (values: Array<[V1, number]>) => Array<[V2, number]>\n\n constructor(\n id: number,\n inputA: DifferenceStreamReader<[K, V1]>,\n output: DifferenceStreamWriter<[K, V2]>,\n f: (values: Array<[V1, number]>) => Array<[V2, number]>,\n ) {\n super(id, inputA, output)\n this.#f = f\n }\n\n run(): void {\n // Collect all input messages and update the index\n const keysTodo = new Set<K>()\n for (const message of this.inputMessages()) {\n for (const [item, multiplicity] of message.getInner()) {\n const [key, value] = item\n this.#index.addValue(key, [value, multiplicity])\n keysTodo.add(key)\n }\n }\n\n // For each key, compute the reduction and delta\n const result: Array<[[K, V2], number]> = []\n for (const key of keysTodo) {\n const curr = this.#index.get(key)\n const currOut = this.#indexOut.get(key)\n const out = this.#f(curr)\n\n // Create maps for current and previous outputs using values directly as keys\n const newOutputMap = new Map<V2, number>()\n const oldOutputMap = new Map<V2, number>()\n\n // Process new output\n for (const [value, multiplicity] of out) {\n const existing = newOutputMap.get(value) ?? 0\n newOutputMap.set(value, existing + multiplicity)\n }\n\n // Process previous output\n for (const [value, multiplicity] of currOut) {\n const existing = oldOutputMap.get(value) ?? 0\n oldOutputMap.set(value, existing + multiplicity)\n }\n\n // First, emit removals for old values that are no longer present\n for (const [value, multiplicity] of oldOutputMap) {\n if (!newOutputMap.has(value)) {\n // Remove the old value entirely\n result.push([[key, value], -multiplicity])\n this.#indexOut.addValue(key, [value, -multiplicity])\n }\n }\n\n // Then, emit additions for new values that are not present in old\n for (const [value, multiplicity] of newOutputMap) {\n if (!oldOutputMap.has(value)) {\n // Add the new value only if it has non-zero multiplicity\n if (multiplicity !== 0) {\n result.push([[key, value], multiplicity])\n this.#indexOut.addValue(key, [value, multiplicity])\n }\n }\n }\n\n // Finally, emit multiplicity changes for values that were present and are still present\n for (const [value, newMultiplicity] of newOutputMap) {\n const oldMultiplicity = oldOutputMap.get(value)\n if (oldMultiplicity !== undefined) {\n const delta = newMultiplicity - oldMultiplicity\n // Only emit actual changes, i.e. non-zero deltas\n if (delta !== 0) {\n result.push([[key, value], delta])\n this.#indexOut.addValue(key, [value, delta])\n }\n }\n }\n }\n\n if (result.length > 0) {\n this.output.sendData(new MultiSet(result))\n }\n }\n}\n\n/**\n * Reduces the elements in the stream by key (version-free)\n */\nexport function reduce<\n KType extends (T extends KeyValue<infer K, infer _V> ? K : never),\n V1Type extends (T extends KeyValue<KType, infer V> ? V : never),\n R,\n T,\n>(f: (values: Array<[V1Type, number]>) => Array<[R, number]>) {\n return (stream: IStreamBuilder<T>): IStreamBuilder<KeyValue<KType, R>> => {\n const output = new StreamBuilder<KeyValue<KType, R>>(\n stream.graph,\n new DifferenceStreamWriter<KeyValue<KType, R>>(),\n )\n const operator = new ReduceOperator<KType, V1Type, R>(\n stream.graph.getNextOperatorId(),\n stream.connectReader() as DifferenceStreamReader<KeyValue<KType, V1Type>>,\n output.writer,\n f,\n )\n stream.graph.addOperator(operator)\n return output\n }\n}\n"],"names":["UnaryOperator","Index","MultiSet","StreamBuilder","DifferenceStreamWriter"],"mappings":";;;;;;AAUO,MAAM,uBAAkCA,MAAAA,cAAgC;AAAA,EAC7E,SAAS,IAAIC,QAAAA,MAAA;AAAA,EACb,YAAY,IAAIA,QAAAA,MAAA;AAAA,EAChB;AAAA,EAEA,YACE,IACA,QACA,QACA,GACA;AACA,UAAM,IAAI,QAAQ,MAAM;AACxB,SAAK,KAAK;AAAA,EACZ;AAAA,EAEA,MAAY;AAEV,UAAM,+BAAe,IAAA;AACrB,eAAW,WAAW,KAAK,iBAAiB;AAC1C,iBAAW,CAAC,MAAM,YAAY,KAAK,QAAQ,YAAY;AACrD,cAAM,CAAC,KAAK,KAAK,IAAI;AACrB,aAAK,OAAO,SAAS,KAAK,CAAC,OAAO,YAAY,CAAC;AAC/C,iBAAS,IAAI,GAAG;AAAA,MAClB;AAAA,IACF;AAGA,UAAM,SAAmC,CAAA;AACzC,eAAW,OAAO,UAAU;AAC1B,YAAM,OAAO,KAAK,OAAO,IAAI,GAAG;AAChC,YAAM,UAAU,KAAK,UAAU,IAAI,GAAG;AACtC,YAAM,MAAM,KAAK,GAAG,IAAI;AAGxB,YAAM,mCAAmB,IAAA;AACzB,YAAM,mCAAmB,IAAA;AAGzB,iBAAW,CAAC,OAAO,YAAY,KAAK,KAAK;AACvC,cAAM,WAAW,aAAa,IAAI,KAAK,KAAK;AAC5C,qBAAa,IAAI,OAAO,WAAW,YAAY;AAAA,MACjD;AAGA,iBAAW,CAAC,OAAO,YAAY,KAAK,SAAS;AAC3C,cAAM,WAAW,aAAa,IAAI,KAAK,KAAK;AAC5C,qBAAa,IAAI,OAAO,WAAW,YAAY;AAAA,MACjD;AAGA,iBAAW,CAAC,OAAO,YAAY,KAAK,cAAc;AAChD,YAAI,CAAC,aAAa,IAAI,KAAK,GAAG;AAE5B,iBAAO,KAAK,CAAC,CAAC,KAAK,KAAK,GAAG,CAAC,YAAY,CAAC;AACzC,eAAK,UAAU,SAAS,KAAK,CAAC,OAAO,CAAC,YAAY,CAAC;AAAA,QACrD;AAAA,MACF;AAGA,iBAAW,CAAC,OAAO,YAAY,KAAK,cAAc;AAChD,YAAI,CAAC,aAAa,IAAI,KAAK,GAAG;AAE5B,cAAI,iBAAiB,GAAG;AACtB,mBAAO,KAAK,CAAC,CAAC,KAAK,KAAK,GAAG,YAAY,CAAC;AACxC,iBAAK,UAAU,SAAS,KAAK,CAAC,OAAO,YAAY,CAAC;AAAA,UACpD;AAAA,QACF;AAAA,MACF;AAGA,iBAAW,CAAC,OAAO,eAAe,KAAK,cAAc;AACnD,cAAM,kBAAkB,aAAa,IAAI,KAAK;AAC9C,YAAI,oBAAoB,QAAW;AACjC,gBAAM,QAAQ,kBAAkB;AAEhC,cAAI,UAAU,GAAG;AACf,mBAAO,KAAK,CAAC,CAAC,KAAK,KAAK,GAAG,KAAK,CAAC;AACjC,iBAAK,UAAU,SAAS,KAAK,CAAC,OAAO,KAAK,CAAC;AAAA,UAC7C;AAAA,QACF;AAAA,MACF;AAAA,IACF;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAIC,SAAAA,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AACF;AAKO,SAAS,OAKd,GAA4D;AAC5D,SAAO,CAAC,WAAkE;AACxE,UAAM,SAAS,IAAIC,GAAAA;AAAAA,MACjB,OAAO;AAAA,MACP,IAAIC,MAAAA,uBAAA;AAAA,IAA2C;AAEjD,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,OAAO;AAAA,MACP;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;;;"}
|
|
@@ -1,14 +1,38 @@
|
|
|
1
|
-
import { DifferenceStreamWriter, UnaryOperator, DifferenceStreamReader } from
|
|
2
|
-
import { IStreamBuilder, KeyValue } from
|
|
1
|
+
import { DifferenceStreamWriter, UnaryOperator, DifferenceStreamReader } from "../graph.cjs";
|
|
2
|
+
import { IStreamBuilder, KeyValue } from "../types.cjs";
|
|
3
3
|
/**
|
|
4
4
|
* Base operator for reduction operations (version-free)
|
|
5
5
|
*/
|
|
6
|
-
export declare class ReduceOperator<K, V1, V2> extends UnaryOperator<[
|
|
6
|
+
export declare class ReduceOperator<K, V1, V2> extends UnaryOperator<[
|
|
7
|
+
K,
|
|
8
|
+
V1
|
|
9
|
+
], [
|
|
10
|
+
K,
|
|
11
|
+
V2
|
|
12
|
+
]> {
|
|
7
13
|
#private;
|
|
8
|
-
constructor(id: number, inputA: DifferenceStreamReader<[
|
|
14
|
+
constructor(id: number, inputA: DifferenceStreamReader<[
|
|
15
|
+
K,
|
|
16
|
+
V1
|
|
17
|
+
]>, output: DifferenceStreamWriter<[
|
|
18
|
+
K,
|
|
19
|
+
V2
|
|
20
|
+
]>, f: (values: Array<[
|
|
21
|
+
V1,
|
|
22
|
+
number
|
|
23
|
+
]>) => Array<[
|
|
24
|
+
V2,
|
|
25
|
+
number
|
|
26
|
+
]>);
|
|
9
27
|
run(): void;
|
|
10
28
|
}
|
|
11
29
|
/**
|
|
12
30
|
* Reduces the elements in the stream by key (version-free)
|
|
13
31
|
*/
|
|
14
|
-
export declare function reduce<KType extends T extends KeyValue<infer K, infer _V> ? K : never, V1Type extends T extends KeyValue<KType, infer V> ? V : never, R, T>(f: (values: Array<[
|
|
32
|
+
export declare function reduce<KType extends (T extends KeyValue<infer K, infer _V> ? K : never), V1Type extends (T extends KeyValue<KType, infer V> ? V : never), R, T>(f: (values: Array<[
|
|
33
|
+
V1Type,
|
|
34
|
+
number
|
|
35
|
+
]>) => Array<[
|
|
36
|
+
R,
|
|
37
|
+
number
|
|
38
|
+
]>): (stream: IStreamBuilder<T>) => IStreamBuilder<KeyValue<KType, R>>;
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { DifferenceStreamWriter, LinearUnaryOperator, DifferenceStreamReader } from
|
|
2
|
-
import { PipedOperator } from
|
|
3
|
-
import { MultiSet } from
|
|
1
|
+
import { DifferenceStreamWriter, LinearUnaryOperator, DifferenceStreamReader } from "../graph.cjs";
|
|
2
|
+
import { PipedOperator } from "../types.cjs";
|
|
3
|
+
import { MultiSet } from "../multiset.cjs";
|
|
4
4
|
/**
|
|
5
5
|
* Operator that applies a function to each multi-set in the input stream
|
|
6
6
|
*/
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"topK.cjs","sources":["../../../src/operators/topK.ts"],"sourcesContent":["import { MultiSet } from '../multiset.js'\nimport { reduce } from './reduce.js'\nimport type { IStreamBuilder, KeyValue, PipedOperator } from '../types'\n\ninterface TopKOptions {\n limit?: number\n offset?: number\n}\n\n/**\n * Limits the number of results based on a comparator, with optional offset.\n * This works on a keyed stream, where the key is the first element of the tuple\n * The ordering is withing a key group, i.e. elements are sorted within a key group\n * and the limit + offset is applied to that sorted group.\n * To order the entire stream, key by the same value for all elements such as null.\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 limits the number of results\n */\nexport function topK<\n KType extends T extends KeyValue<infer K, infer _V> ? K : never,\n V1Type extends T extends KeyValue<KType, infer V> ? V : never,\n T,\n>(\n comparator: (a: V1Type, b: V1Type) => number,\n options?: TopKOptions,\n): PipedOperator<T, T> {\n const limit = options?.limit ?? Infinity\n const offset = options?.offset ?? 0\n\n return (stream: IStreamBuilder<T>): IStreamBuilder<T> => {\n const reduced = stream.pipe(\n reduce((values) => {\n // `values` is a list of tuples, first element is the value, second is the multiplicity\n const consolidated = new MultiSet(values).consolidate()\n const sortedValues = consolidated\n .getInner()\n .sort((a, b) => comparator(a[0] as V1Type, b[0] as V1Type))\n return sortedValues.slice(offset, offset + limit)\n }),\n )\n return reduced as IStreamBuilder<T>\n }\n}\n\n/**\n * Limits the number of results based on a comparator, with optional offset.\n * This works on a keyed stream, where the key is the first element of the tuple\n * The ordering is withing a key group, i.e. elements are sorted within a key group\n * and the limit + offset is applied to that sorted group.\n * To order the entire stream, key by the same value for all elements such as null.\n * Adds the index of the element to the result as [key, [value, index]]\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 topKWithIndex<\n KType extends T extends KeyValue<infer K, infer _V> ? K : never,\n V1Type extends T extends KeyValue<KType, infer V> ? V : never,\n T,\n>(\n comparator: (a: V1Type, b: V1Type) => number,\n options?: TopKOptions,\n): PipedOperator<T, KeyValue<KType, [V1Type, number]>> {\n const limit = options?.limit ?? Infinity\n const offset = options?.offset ?? 0\n\n return (\n stream: IStreamBuilder<T>,\n ): IStreamBuilder<KeyValue<KType, [V1Type, number]>> => {\n const reduced = stream.pipe(\n reduce<KType, V1Type, [V1Type, number], T>((values) => {\n // `values` is a list of tuples, first element is the value, second is the multiplicity\n const consolidated = new MultiSet(values).consolidate()\n let i = offset\n const sortedValues = consolidated\n .getInner()\n .sort((a, b) => comparator(a[0], b[0]))\n .slice(offset, offset + limit)\n .map(([value, multiplicity]): [[V1Type, number], number] => [\n [value, i++],\n multiplicity,\n ])\n return sortedValues\n }),\n )\n return reduced\n }\n}\n"],"names":["reduce","MultiSet"],"mappings":";;;;AAoBO,SAAS,KAKd,YACA,SACqB;AACrB,QAAM,QAAQ,SAAS,SAAS;AAChC,QAAM,SAAS,SAAS,UAAU;AAElC,SAAO,CAAC,WAAiD;AACvD,UAAM,UAAU,OAAO;AAAA,MACrBA,OAAAA,OAAO,CAAC,WAAW;AAEjB,cAAM,eAAe,IAAIC,SAAAA,SAAS,MAAM,EAAE,YAAA;AAC1C,cAAM,eAAe,aAClB,SAAA,EACA,KAAK,CAAC,GAAG,MAAM,WAAW,EAAE,CAAC,GAAa,EAAE,CAAC,CAAW,CAAC;AAC5D,eAAO,aAAa,MAAM,QAAQ,SAAS,KAAK;AAAA,MAClD,CAAC;AAAA,IAAA;AAEH,WAAO;AAAA,EACT;AACF;AAcO,SAAS,cAKd,YACA,SACqD;AACrD,QAAM,QAAQ,SAAS,SAAS;AAChC,QAAM,SAAS,SAAS,UAAU;AAElC,SAAO,CACL,WACsD;AACtD,UAAM,UAAU,OAAO;AAAA,MACrBD,OAAAA,OAA2C,CAAC,WAAW;AAErD,cAAM,eAAe,IAAIC,SAAAA,SAAS,MAAM,EAAE,YAAA;AAC1C,YAAI,IAAI;AACR,cAAM,eAAe,aAClB,SAAA,EACA,KAAK,CAAC,GAAG,MAAM,WAAW,EAAE,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,EACrC,MAAM,QAAQ,SAAS,KAAK,EAC5B,IAAI,CAAC,CAAC,OAAO,YAAY,MAAkC;AAAA,UAC1D,CAAC,OAAO,GAAG;AAAA,UACX;AAAA,QAAA,CACD;AACH,eAAO;AAAA,MACT,CAAC;AAAA,IAAA;AAEH,WAAO;AAAA,EACT;AACF;;;"}
|
|
1
|
+
{"version":3,"file":"topK.cjs","sources":["../../../src/operators/topK.ts"],"sourcesContent":["import { MultiSet } from '../multiset.js'\nimport { reduce } from './reduce.js'\nimport type { IStreamBuilder, KeyValue, PipedOperator } from '../types'\n\ninterface TopKOptions {\n limit?: number\n offset?: number\n}\n\n/**\n * Limits the number of results based on a comparator, with optional offset.\n * This works on a keyed stream, where the key is the first element of the tuple\n * The ordering is withing a key group, i.e. elements are sorted within a key group\n * and the limit + offset is applied to that sorted group.\n * To order the entire stream, key by the same value for all elements such as null.\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 limits the number of results\n */\nexport function topK<\n KType extends (T extends KeyValue<infer K, infer _V> ? K : never),\n V1Type extends (T extends KeyValue<KType, infer V> ? V : never),\n T,\n>(\n comparator: (a: V1Type, b: V1Type) => number,\n options?: TopKOptions,\n): PipedOperator<T, T> {\n const limit = options?.limit ?? Infinity\n const offset = options?.offset ?? 0\n\n return (stream: IStreamBuilder<T>): IStreamBuilder<T> => {\n const reduced = stream.pipe(\n reduce((values) => {\n // `values` is a list of tuples, first element is the value, second is the multiplicity\n const consolidated = new MultiSet(values).consolidate()\n const sortedValues = consolidated\n .getInner()\n .sort((a, b) => comparator(a[0] as V1Type, b[0] as V1Type))\n return sortedValues.slice(offset, offset + limit)\n }),\n )\n return reduced as IStreamBuilder<T>\n }\n}\n\n/**\n * Limits the number of results based on a comparator, with optional offset.\n * This works on a keyed stream, where the key is the first element of the tuple\n * The ordering is withing a key group, i.e. elements are sorted within a key group\n * and the limit + offset is applied to that sorted group.\n * To order the entire stream, key by the same value for all elements such as null.\n * Adds the index of the element to the result as [key, [value, index]]\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 topKWithIndex<\n KType extends (T extends KeyValue<infer K, infer _V> ? K : never),\n V1Type extends (T extends KeyValue<KType, infer V> ? V : never),\n T,\n>(\n comparator: (a: V1Type, b: V1Type) => number,\n options?: TopKOptions,\n): PipedOperator<T, KeyValue<KType, [V1Type, number]>> {\n const limit = options?.limit ?? Infinity\n const offset = options?.offset ?? 0\n\n return (\n stream: IStreamBuilder<T>,\n ): IStreamBuilder<KeyValue<KType, [V1Type, number]>> => {\n const reduced = stream.pipe(\n reduce<KType, V1Type, [V1Type, number], T>((values) => {\n // `values` is a list of tuples, first element is the value, second is the multiplicity\n const consolidated = new MultiSet(values).consolidate()\n let i = offset\n const sortedValues = consolidated\n .getInner()\n .sort((a, b) => comparator(a[0], b[0]))\n .slice(offset, offset + limit)\n .map(([value, multiplicity]): [[V1Type, number], number] => [\n [value, i++],\n multiplicity,\n ])\n return sortedValues\n }),\n )\n return reduced\n }\n}\n"],"names":["reduce","MultiSet"],"mappings":";;;;AAoBO,SAAS,KAKd,YACA,SACqB;AACrB,QAAM,QAAQ,SAAS,SAAS;AAChC,QAAM,SAAS,SAAS,UAAU;AAElC,SAAO,CAAC,WAAiD;AACvD,UAAM,UAAU,OAAO;AAAA,MACrBA,OAAAA,OAAO,CAAC,WAAW;AAEjB,cAAM,eAAe,IAAIC,SAAAA,SAAS,MAAM,EAAE,YAAA;AAC1C,cAAM,eAAe,aAClB,SAAA,EACA,KAAK,CAAC,GAAG,MAAM,WAAW,EAAE,CAAC,GAAa,EAAE,CAAC,CAAW,CAAC;AAC5D,eAAO,aAAa,MAAM,QAAQ,SAAS,KAAK;AAAA,MAClD,CAAC;AAAA,IAAA;AAEH,WAAO;AAAA,EACT;AACF;AAcO,SAAS,cAKd,YACA,SACqD;AACrD,QAAM,QAAQ,SAAS,SAAS;AAChC,QAAM,SAAS,SAAS,UAAU;AAElC,SAAO,CACL,WACsD;AACtD,UAAM,UAAU,OAAO;AAAA,MACrBD,OAAAA,OAA2C,CAAC,WAAW;AAErD,cAAM,eAAe,IAAIC,SAAAA,SAAS,MAAM,EAAE,YAAA;AAC1C,YAAI,IAAI;AACR,cAAM,eAAe,aAClB,SAAA,EACA,KAAK,CAAC,GAAG,MAAM,WAAW,EAAE,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,EACrC,MAAM,QAAQ,SAAS,KAAK,EAC5B,IAAI,CAAC,CAAC,OAAO,YAAY,MAAkC;AAAA,UAC1D,CAAC,OAAO,GAAG;AAAA,UACX;AAAA,QAAA,CACD;AACH,eAAO;AAAA,MACT,CAAC;AAAA,IAAA;AAEH,WAAO;AAAA,EACT;AACF;;;"}
|