@tanstack/db-ivm 0.1.2 → 0.1.3

package/dist/esm/operators/groupBy.d.ts

@@ -26,7 +26,7 @@ export declare function sum<T>(valueExtractor?: (value: T) => number): Aggregate
 /**
  * Creates a count aggregate function
  */
-export declare function count<T>(): AggregateFunction<T, number, number>;
+export declare function count<T>(valueExtractor?: (value: T) => any): AggregateFunction<T, number, number>;
 /**
  * Creates an average aggregate function
  */

package/dist/esm/operators/groupBy.js

@@ -79,13 +79,14 @@ function sum(valueExtractor = (v) => v) {
     }
   };
 }
-function count() {
+function count(valueExtractor = (v) => v) {
   return {
-    preMap: () => 1,
+    // Count only not-null values (the `== null` comparison gives true for both null and undefined)
+    preMap: (data) => valueExtractor(data) == null ? 0 : 1,
     reduce: (values) => {
       let totalCount = 0;
-      for (const [_, multiplicity] of values) {
-        totalCount += multiplicity;
+      for (const [nullMultiplier, multiplicity] of values) {
+        totalCount += nullMultiplier * multiplicity;
       }
       return totalCount;
     }

package/dist/esm/operators/groupBy.js.map

@@ -1 +1 @@
-{"version":3,"file":"groupBy.js","sources":["../../../src/operators/groupBy.ts"],"sourcesContent":["import { 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 = `__original_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 = JSON.stringify(key)\n\n        // Create values object with pre-aggregated values\n        const values: Record<string, 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<string, Record<string, unknown>>\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<string, 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>(): AggregateFunction<T, number, number> {\n  return {\n    preMap: () => 1,\n    reduce: (values: Array<[number, number]>) => {\n      let totalCount = 0\n      for (const [_, multiplicity] of values) {\n        totalCount += 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\n/**\n * Creates a min aggregate function that computes the minimum value in a group\n * @param valueExtractor Function to extract a numeric value from each data entry\n */\nexport function min<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 minValue = Number.POSITIVE_INFINITY\n      for (const [value, _multiplicity] of values) {\n        if (value < minValue) {\n          minValue = value\n        }\n      }\n      return minValue === Number.POSITIVE_INFINITY ? 0 : 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 numeric value from each data entry\n */\nexport function max<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 maxValue = Number.NEGATIVE_INFINITY\n      for (const [value, _multiplicity] of values) {\n        if (value > maxValue) {\n          maxValue = value\n        }\n      }\n      return maxValue === Number.NEGATIVE_INFINITY ? 0 : 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":[],"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,eAAe;AAGrB,UAAM,oBAAoB,OAAO;AAAA,MAC/B,IAAI,CAAC,SAAS;AACZ,cAAM,MAAM,aAAa,IAAI;AAC7B,cAAM,YAAY,KAAK,UAAU,GAAG;AAGpC,cAAM,SAAkC,CAAA;AAGxC,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,MAC3B,CAAC;AAAA,IAAA;AAIH,UAAM,UAAU,kBAAkB;AAAA,MAChC,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,SAAkC,CAAA;AAGxC,cAAM,eAAc,kBAAO,CAAC,MAAR,mBAAY,OAAZ,mBAAiB;AACrC,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,MACb,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,QAAiD;AAC/D,SAAO;AAAA,IACL,QAAQ,MAAM;AAAA,IACd,QAAQ,CAAC,WAAoC;AAC3C,UAAI,aAAa;AACjB,iBAAW,CAAC,GAAG,YAAY,KAAK,QAAQ;AACtC,sBAAc;AAAA,MAChB;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;AAMO,SAAS,IACd,iBAAuC,CAAC,MAAM,GACR;AACtC,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,eAAe,IAAI;AAAA,IACxC,QAAQ,CAAC,WAAoC;AAC3C,UAAI,WAAW,OAAO;AACtB,iBAAW,CAAC,OAAO,aAAa,KAAK,QAAQ;AAC3C,YAAI,QAAQ,UAAU;AACpB,qBAAW;AAAA,QACb;AAAA,MACF;AACA,aAAO,aAAa,OAAO,oBAAoB,IAAI;AAAA,IACrD;AAAA,EAAA;AAEJ;AAMO,SAAS,IACd,iBAAuC,CAAC,MAAM,GACR;AACtC,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,eAAe,IAAI;AAAA,IACxC,QAAQ,CAAC,WAAoC;AAC3C,UAAI,WAAW,OAAO;AACtB,iBAAW,CAAC,OAAO,aAAa,KAAK,QAAQ;AAC3C,YAAI,QAAQ,UAAU;AACpB,qBAAW;AAAA,QACb;AAAA,MACF;AACA,aAAO,aAAa,OAAO,oBAAoB,IAAI;AAAA,IACrD;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;"}
+{"version":3,"file":"groupBy.js","sources":["../../../src/operators/groupBy.ts"],"sourcesContent":["import { 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 = `__original_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 = JSON.stringify(key)\n\n        // Create values object with pre-aggregated values\n        const values: Record<string, 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<string, Record<string, unknown>>\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<string, 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\n/**\n * Creates a min aggregate function that computes the minimum value in a group\n * @param valueExtractor Function to extract a numeric value from each data entry\n */\nexport function min<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 minValue = Number.POSITIVE_INFINITY\n      for (const [value, _multiplicity] of values) {\n        if (value < minValue) {\n          minValue = value\n        }\n      }\n      return minValue === Number.POSITIVE_INFINITY ? 0 : 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 numeric value from each data entry\n */\nexport function max<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 maxValue = Number.NEGATIVE_INFINITY\n      for (const [value, _multiplicity] of values) {\n        if (value > maxValue) {\n          maxValue = value\n        }\n      }\n      return maxValue === Number.NEGATIVE_INFINITY ? 0 : 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":[],"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,eAAe;AAGrB,UAAM,oBAAoB,OAAO;AAAA,MAC/B,IAAI,CAAC,SAAS;AACZ,cAAM,MAAM,aAAa,IAAI;AAC7B,cAAM,YAAY,KAAK,UAAU,GAAG;AAGpC,cAAM,SAAkC,CAAA;AAGxC,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,MAC3B,CAAC;AAAA,IAAA;AAIH,UAAM,UAAU,kBAAkB;AAAA,MAChC,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,SAAkC,CAAA;AAGxC,cAAM,eAAc,kBAAO,CAAC,MAAR,mBAAY,OAAZ,mBAAiB;AACrC,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,MACb,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;AAMO,SAAS,IACd,iBAAuC,CAAC,MAAM,GACR;AACtC,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,eAAe,IAAI;AAAA,IACxC,QAAQ,CAAC,WAAoC;AAC3C,UAAI,WAAW,OAAO;AACtB,iBAAW,CAAC,OAAO,aAAa,KAAK,QAAQ;AAC3C,YAAI,QAAQ,UAAU;AACpB,qBAAW;AAAA,QACb;AAAA,MACF;AACA,aAAO,aAAa,OAAO,oBAAoB,IAAI;AAAA,IACrD;AAAA,EAAA;AAEJ;AAMO,SAAS,IACd,iBAAuC,CAAC,MAAM,GACR;AACtC,SAAO;AAAA,IACL,QAAQ,CAAC,SAAY,eAAe,IAAI;AAAA,IACxC,QAAQ,CAAC,WAAoC;AAC3C,UAAI,WAAW,OAAO;AACtB,iBAAW,CAAC,OAAO,aAAa,KAAK,QAAQ;AAC3C,YAAI,QAAQ,UAAU;AACpB,qBAAW;AAAA,QACb;AAAA,MACF;AACA,aAAO,aAAa,OAAO,oBAAoB,IAAI;AAAA,IACrD;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;"}

package/dist/esm/utils.d.ts

@@ -1,3 +1,8 @@
+/**
+ * Simple assertion function for runtime checks.
+ * Throws an error if the condition is false.
+ */
+export declare function assert(condition: unknown, message?: string): asserts condition;
 /**
  * A map that returns a default value for keys that are not present.
  */
@@ -11,10 +16,6 @@ export declare class DefaultMap<K, V> extends Map<K, V> {
     update(key: K, updater: (value: V) => V): V;
 }
 export declare function chunkedArrayPush(array: Array<unknown>, other: Array<unknown>): void;
-/**
- * A hash method that caches the hash of a value in a week map
- */
-export declare function hash(data: any): string;
 export declare function binarySearch<T>(array: Array<T>, value: T, comparator: (a: T, b: T) => number): number;
 /**
  * Utility for generating unique IDs for objects and values.
@@ -38,3 +39,5 @@ export declare class ObjectIdGenerator {
  * Global instance for cases where a shared object ID space is needed.
  */
 export declare const globalObjectIdGenerator: ObjectIdGenerator;
+export declare function concatIterable<T>(...iterables: Array<Iterable<T>>): Iterable<T>;
+export declare function mapIterable<T, U>(it: Iterable<T>, fn: (t: T) => U): Iterable<U>;

package/dist/esm/utils.js

@@ -1,4 +1,3 @@
-import murmurhash from "murmurhash-js";
 class DefaultMap extends Map {
   constructor(defaultValue, entries) {
     super(entries);
@@ -6,7 +5,7 @@ class DefaultMap extends Map {
   }
   get(key) {
     if (!this.has(key)) {
-      this.set(key, this.defaultValue());
+      return this.defaultValue();
     }
     return super.get(key);
   }
@@ -31,36 +30,6 @@ function chunkedArrayPush(array, other) {
     }
   }
 }
-const hashCache = /* @__PURE__ */ new WeakMap();
-function hashReplacer(_key, value) {
-  if (typeof value === `bigint`) {
-    return String(value);
-  } else if (typeof value === `symbol`) {
-    return String(value);
-  } else if (typeof value === `function`) {
-    return String(value);
-  } else if (value === void 0) {
-    return `undefined`;
-  } else if (value instanceof Map) {
-    return `Map(${JSON.stringify(Array.from(value.entries()), hashReplacer)})`;
-  } else if (value instanceof Set) {
-    return `Set(${JSON.stringify(Array.from(value.values()), hashReplacer)})`;
-  }
-  return value;
-}
-function hash(data) {
-  if (data === null || data === void 0 || typeof data !== `object` && typeof data !== `function`) {
-    const serialized2 = JSON.stringify(data, hashReplacer);
-    return murmurhash.murmur3(serialized2).toString(16);
-  }
-  if (hashCache.has(data)) {
-    return hashCache.get(data);
-  }
-  const serialized = JSON.stringify(data, hashReplacer);
-  const hashValue = murmurhash.murmur3(serialized).toString(16);
-  hashCache.set(data, hashValue);
-  return hashValue;
-}
 function binarySearch(array, value, comparator) {
   let low = 0;
   let high = array.length;
@@ -114,12 +83,23 @@ class ObjectIdGenerator {
   }
 }
 const globalObjectIdGenerator = new ObjectIdGenerator();
+function* concatIterable(...iterables) {
+  for (const iterable of iterables) {
+    yield* iterable;
+  }
+}
+function* mapIterable(it, fn) {
+  for (const t of it) {
+    yield fn(t);
+  }
+}
 export {
   DefaultMap,
   ObjectIdGenerator,
   binarySearch,
   chunkedArrayPush,
+  concatIterable,
   globalObjectIdGenerator,
-  hash
+  mapIterable
 };
 //# sourceMappingURL=utils.js.map

package/dist/esm/utils.js.map

@@ -1 +1 @@
-{"version":3,"file":"utils.js","sources":["../../src/utils.ts"],"sourcesContent":["import murmurhash from \"murmurhash-js\"\n\n/**\n * A map that returns a default value for keys that are not present.\n */\nexport class DefaultMap<K, V> extends Map<K, V> {\n  constructor(\n    private defaultValue: () => V,\n    entries?: Iterable<[K, V]>\n  ) {\n    super(entries)\n  }\n\n  get(key: K): V {\n    if (!this.has(key)) {\n      this.set(key, this.defaultValue())\n    }\n    return super.get(key)!\n  }\n\n  /**\n   * Update the value for a key using a function.\n   */\n  update(key: K, updater: (value: V) => V): V {\n    const value = this.get(key)\n    const newValue = updater(value)\n    this.set(key, newValue)\n    return newValue\n  }\n}\n\n// JS engines have various limits on how many args can be passed to a function\n// with a spread operator, so we need to split the operation into chunks\n// 32767 is the max for Chrome 14, all others are higher\n// TODO: investigate the performance of this and other approaches\nconst chunkSize = 30000\nexport function chunkedArrayPush(array: Array<unknown>, other: Array<unknown>) {\n  if (other.length <= chunkSize) {\n    array.push(...other)\n  } else {\n    for (let i = 0; i < other.length; i += chunkSize) {\n      const chunk = other.slice(i, i + chunkSize)\n      array.push(...chunk)\n    }\n  }\n}\n\nconst hashCache = new WeakMap()\n\n/**\n * Replacer function for JSON.stringify that converts unsupported types to strings\n */\nfunction hashReplacer(_key: string, value: any): any {\n  if (typeof value === `bigint`) {\n    return String(value)\n  } else if (typeof value === `symbol`) {\n    return String(value)\n  } else if (typeof value === `function`) {\n    return String(value)\n  } else if (value === undefined) {\n    return `undefined`\n  } else if (value instanceof Map) {\n    return `Map(${JSON.stringify(Array.from(value.entries()), hashReplacer)})`\n  } else if (value instanceof Set) {\n    return `Set(${JSON.stringify(Array.from(value.values()), hashReplacer)})`\n  }\n  return value\n}\n\n/**\n * A hash method that caches the hash of a value in a week map\n */\nexport function hash(data: any): string {\n  if (\n    data === null ||\n    data === undefined ||\n    (typeof data !== `object` && typeof data !== `function`)\n  ) {\n    // Can't be cached in the weak map because it's not an object\n    const serialized = JSON.stringify(data, hashReplacer)\n    return murmurhash.murmur3(serialized).toString(16)\n  }\n\n  if (hashCache.has(data)) {\n    return hashCache.get(data)\n  }\n\n  const serialized = JSON.stringify(data, hashReplacer)\n  const hashValue = murmurhash.murmur3(serialized).toString(16)\n  hashCache.set(data, hashValue)\n  return hashValue\n}\n\nexport function binarySearch<T>(\n  array: Array<T>,\n  value: T,\n  comparator: (a: T, b: T) => number\n): number {\n  let low = 0\n  let high = array.length\n  while (low < high) {\n    const mid = Math.floor((low + high) / 2)\n    const comparison = comparator(array[mid]!, value)\n    if (comparison < 0) {\n      low = mid + 1\n    } else if (comparison > 0) {\n      high = mid\n    } else {\n      return mid\n    }\n  }\n  return low\n}\n\n/**\n * Utility for generating unique IDs for objects and values.\n * Uses WeakMap for object reference tracking and consistent hashing for primitives.\n */\nexport class ObjectIdGenerator {\n  private objectIds = new WeakMap<object, number>()\n  private nextId = 0\n\n  /**\n   * Get a unique identifier for any value.\n   * - Objects: Uses WeakMap for reference-based identity\n   * - Primitives: Uses consistent string-based hashing\n   */\n  getId(value: any): number {\n    // For primitives, use a simple hash of their string representation\n    if (typeof value !== `object` || value === null) {\n      const str = String(value)\n      let hashValue = 0\n      for (let i = 0; i < str.length; i++) {\n        const char = str.charCodeAt(i)\n        hashValue = (hashValue << 5) - hashValue + char\n        hashValue = hashValue & hashValue // Convert to 32-bit integer\n      }\n      return hashValue\n    }\n\n    // For objects, use WeakMap to assign unique IDs\n    if (!this.objectIds.has(value)) {\n      this.objectIds.set(value, this.nextId++)\n    }\n    return this.objectIds.get(value)!\n  }\n\n  /**\n   * Get a string representation of the ID for use in composite keys.\n   */\n  getStringId(value: any): string {\n    if (value === null) return `null`\n    if (value === undefined) return `undefined`\n    if (typeof value !== `object`) return `str_${String(value)}`\n\n    return `obj_${this.getId(value)}`\n  }\n}\n\n/**\n * Global instance for cases where a shared object ID space is needed.\n */\nexport const globalObjectIdGenerator = new ObjectIdGenerator()\n"],"names":["serialized"],"mappings":";AAKO,MAAM,mBAAyB,IAAU;AAAA,EAC9C,YACU,cACR,SACA;AACA,UAAM,OAAO;AAHL,SAAA,eAAA;AAAA,EAIV;AAAA,EAEA,IAAI,KAAW;AACb,QAAI,CAAC,KAAK,IAAI,GAAG,GAAG;AAClB,WAAK,IAAI,KAAK,KAAK,aAAA,CAAc;AAAA,IACnC;AACA,WAAO,MAAM,IAAI,GAAG;AAAA,EACtB;AAAA;AAAA;AAAA;AAAA,EAKA,OAAO,KAAQ,SAA6B;AAC1C,UAAM,QAAQ,KAAK,IAAI,GAAG;AAC1B,UAAM,WAAW,QAAQ,KAAK;AAC9B,SAAK,IAAI,KAAK,QAAQ;AACtB,WAAO;AAAA,EACT;AACF;AAMA,MAAM,YAAY;AACX,SAAS,iBAAiB,OAAuB,OAAuB;AAC7E,MAAI,MAAM,UAAU,WAAW;AAC7B,UAAM,KAAK,GAAG,KAAK;AAAA,EACrB,OAAO;AACL,aAAS,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK,WAAW;AAChD,YAAM,QAAQ,MAAM,MAAM,GAAG,IAAI,SAAS;AAC1C,YAAM,KAAK,GAAG,KAAK;AAAA,IACrB;AAAA,EACF;AACF;AAEA,MAAM,gCAAgB,QAAA;AAKtB,SAAS,aAAa,MAAc,OAAiB;AACnD,MAAI,OAAO,UAAU,UAAU;AAC7B,WAAO,OAAO,KAAK;AAAA,EACrB,WAAW,OAAO,UAAU,UAAU;AACpC,WAAO,OAAO,KAAK;AAAA,EACrB,WAAW,OAAO,UAAU,YAAY;AACtC,WAAO,OAAO,KAAK;AAAA,EACrB,WAAW,UAAU,QAAW;AAC9B,WAAO;AAAA,EACT,WAAW,iBAAiB,KAAK;AAC/B,WAAO,OAAO,KAAK,UAAU,MAAM,KAAK,MAAM,QAAA,CAAS,GAAG,YAAY,CAAC;AAAA,EACzE,WAAW,iBAAiB,KAAK;AAC/B,WAAO,OAAO,KAAK,UAAU,MAAM,KAAK,MAAM,OAAA,CAAQ,GAAG,YAAY,CAAC;AAAA,EACxE;AACA,SAAO;AACT;AAKO,SAAS,KAAK,MAAmB;AACtC,MACE,SAAS,QACT,SAAS,UACR,OAAO,SAAS,YAAY,OAAO,SAAS,YAC7C;AAEA,UAAMA,cAAa,KAAK,UAAU,MAAM,YAAY;AACpD,WAAO,WAAW,QAAQA,WAAU,EAAE,SAAS,EAAE;AAAA,EACnD;AAEA,MAAI,UAAU,IAAI,IAAI,GAAG;AACvB,WAAO,UAAU,IAAI,IAAI;AAAA,EAC3B;AAEA,QAAM,aAAa,KAAK,UAAU,MAAM,YAAY;AACpD,QAAM,YAAY,WAAW,QAAQ,UAAU,EAAE,SAAS,EAAE;AAC5D,YAAU,IAAI,MAAM,SAAS;AAC7B,SAAO;AACT;AAEO,SAAS,aACd,OACA,OACA,YACQ;AACR,MAAI,MAAM;AACV,MAAI,OAAO,MAAM;AACjB,SAAO,MAAM,MAAM;AACjB,UAAM,MAAM,KAAK,OAAO,MAAM,QAAQ,CAAC;AACvC,UAAM,aAAa,WAAW,MAAM,GAAG,GAAI,KAAK;AAChD,QAAI,aAAa,GAAG;AAClB,YAAM,MAAM;AAAA,IACd,WAAW,aAAa,GAAG;AACzB,aAAO;AAAA,IACT,OAAO;AACL,aAAO;AAAA,IACT;AAAA,EACF;AACA,SAAO;AACT;AAMO,MAAM,kBAAkB;AAAA,EAAxB,cAAA;AACL,SAAQ,gCAAgB,QAAA;AACxB,SAAQ,SAAS;AAAA,EAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOjB,MAAM,OAAoB;AAExB,QAAI,OAAO,UAAU,YAAY,UAAU,MAAM;AAC/C,YAAM,MAAM,OAAO,KAAK;AACxB,UAAI,YAAY;AAChB,eAAS,IAAI,GAAG,IAAI,IAAI,QAAQ,KAAK;AACnC,cAAM,OAAO,IAAI,WAAW,CAAC;AAC7B,qBAAa,aAAa,KAAK,YAAY;AAC3C,oBAAY,YAAY;AAAA,MAC1B;AACA,aAAO;AAAA,IACT;AAGA,QAAI,CAAC,KAAK,UAAU,IAAI,KAAK,GAAG;AAC9B,WAAK,UAAU,IAAI,OAAO,KAAK,QAAQ;AAAA,IACzC;AACA,WAAO,KAAK,UAAU,IAAI,KAAK;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA,EAKA,YAAY,OAAoB;AAC9B,QAAI,UAAU,KAAM,QAAO;AAC3B,QAAI,UAAU,OAAW,QAAO;AAChC,QAAI,OAAO,UAAU,iBAAiB,OAAO,OAAO,KAAK,CAAC;AAE1D,WAAO,OAAO,KAAK,MAAM,KAAK,CAAC;AAAA,EACjC;AACF;AAKO,MAAM,0BAA0B,IAAI,kBAAA;"}
+{"version":3,"file":"utils.js","sources":["../../src/utils.ts"],"sourcesContent":["/**\n * Simple assertion function for runtime checks.\n * Throws an error if the condition is false.\n */\nexport function assert(\n  condition: unknown,\n  message?: string\n): asserts condition {\n  if (!condition) {\n    throw new Error(message || `Assertion failed`)\n  }\n}\n\n/**\n * A map that returns a default value for keys that are not present.\n */\nexport class DefaultMap<K, V> extends Map<K, V> {\n  constructor(\n    private defaultValue: () => V,\n    entries?: Iterable<[K, V]>\n  ) {\n    super(entries)\n  }\n\n  get(key: K): V {\n    if (!this.has(key)) {\n      // this.set(key, this.defaultValue())\n      return this.defaultValue()\n    }\n    return super.get(key)!\n  }\n\n  /**\n   * Update the value for a key using a function.\n   */\n  update(key: K, updater: (value: V) => V): V {\n    const value = this.get(key)\n    const newValue = updater(value)\n    this.set(key, newValue)\n    return newValue\n  }\n}\n\n// JS engines have various limits on how many args can be passed to a function\n// with a spread operator, so we need to split the operation into chunks\n// 32767 is the max for Chrome 14, all others are higher\n// TODO: investigate the performance of this and other approaches\nconst chunkSize = 30000\nexport function chunkedArrayPush(array: Array<unknown>, other: Array<unknown>) {\n  if (other.length <= chunkSize) {\n    array.push(...other)\n  } else {\n    for (let i = 0; i < other.length; i += chunkSize) {\n      const chunk = other.slice(i, i + chunkSize)\n      array.push(...chunk)\n    }\n  }\n}\n\nexport function binarySearch<T>(\n  array: Array<T>,\n  value: T,\n  comparator: (a: T, b: T) => number\n): number {\n  let low = 0\n  let high = array.length\n  while (low < high) {\n    const mid = Math.floor((low + high) / 2)\n    const comparison = comparator(array[mid]!, value)\n    if (comparison < 0) {\n      low = mid + 1\n    } else if (comparison > 0) {\n      high = mid\n    } else {\n      return mid\n    }\n  }\n  return low\n}\n\n/**\n * Utility for generating unique IDs for objects and values.\n * Uses WeakMap for object reference tracking and consistent hashing for primitives.\n */\nexport class ObjectIdGenerator {\n  private objectIds = new WeakMap<object, number>()\n  private nextId = 0\n\n  /**\n   * Get a unique identifier for any value.\n   * - Objects: Uses WeakMap for reference-based identity\n   * - Primitives: Uses consistent string-based hashing\n   */\n  getId(value: any): number {\n    // For primitives, use a simple hash of their string representation\n    if (typeof value !== `object` || value === null) {\n      const str = String(value)\n      let hashValue = 0\n      for (let i = 0; i < str.length; i++) {\n        const char = str.charCodeAt(i)\n        hashValue = (hashValue << 5) - hashValue + char\n        hashValue = hashValue & hashValue // Convert to 32-bit integer\n      }\n      return hashValue\n    }\n\n    // For objects, use WeakMap to assign unique IDs\n    if (!this.objectIds.has(value)) {\n      this.objectIds.set(value, this.nextId++)\n    }\n    return this.objectIds.get(value)!\n  }\n\n  /**\n   * Get a string representation of the ID for use in composite keys.\n   */\n  getStringId(value: any): string {\n    if (value === null) return `null`\n    if (value === undefined) return `undefined`\n    if (typeof value !== `object`) return `str_${String(value)}`\n\n    return `obj_${this.getId(value)}`\n  }\n}\n\n/**\n * Global instance for cases where a shared object ID space is needed.\n */\nexport const globalObjectIdGenerator = new ObjectIdGenerator()\n\nexport function* concatIterable<T>(\n  ...iterables: Array<Iterable<T>>\n): Iterable<T> {\n  for (const iterable of iterables) {\n    yield* iterable\n  }\n}\n\nexport function* mapIterable<T, U>(\n  it: Iterable<T>,\n  fn: (t: T) => U\n): Iterable<U> {\n  for (const t of it) {\n    yield fn(t)\n  }\n}\n"],"names":[],"mappings":"AAgBO,MAAM,mBAAyB,IAAU;AAAA,EAC9C,YACU,cACR,SACA;AACA,UAAM,OAAO;AAHL,SAAA,eAAA;AAAA,EAIV;AAAA,EAEA,IAAI,KAAW;AACb,QAAI,CAAC,KAAK,IAAI,GAAG,GAAG;AAElB,aAAO,KAAK,aAAA;AAAA,IACd;AACA,WAAO,MAAM,IAAI,GAAG;AAAA,EACtB;AAAA;AAAA;AAAA;AAAA,EAKA,OAAO,KAAQ,SAA6B;AAC1C,UAAM,QAAQ,KAAK,IAAI,GAAG;AAC1B,UAAM,WAAW,QAAQ,KAAK;AAC9B,SAAK,IAAI,KAAK,QAAQ;AACtB,WAAO;AAAA,EACT;AACF;AAMA,MAAM,YAAY;AACX,SAAS,iBAAiB,OAAuB,OAAuB;AAC7E,MAAI,MAAM,UAAU,WAAW;AAC7B,UAAM,KAAK,GAAG,KAAK;AAAA,EACrB,OAAO;AACL,aAAS,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK,WAAW;AAChD,YAAM,QAAQ,MAAM,MAAM,GAAG,IAAI,SAAS;AAC1C,YAAM,KAAK,GAAG,KAAK;AAAA,IACrB;AAAA,EACF;AACF;AAEO,SAAS,aACd,OACA,OACA,YACQ;AACR,MAAI,MAAM;AACV,MAAI,OAAO,MAAM;AACjB,SAAO,MAAM,MAAM;AACjB,UAAM,MAAM,KAAK,OAAO,MAAM,QAAQ,CAAC;AACvC,UAAM,aAAa,WAAW,MAAM,GAAG,GAAI,KAAK;AAChD,QAAI,aAAa,GAAG;AAClB,YAAM,MAAM;AAAA,IACd,WAAW,aAAa,GAAG;AACzB,aAAO;AAAA,IACT,OAAO;AACL,aAAO;AAAA,IACT;AAAA,EACF;AACA,SAAO;AACT;AAMO,MAAM,kBAAkB;AAAA,EAAxB,cAAA;AACL,SAAQ,gCAAgB,QAAA;AACxB,SAAQ,SAAS;AAAA,EAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOjB,MAAM,OAAoB;AAExB,QAAI,OAAO,UAAU,YAAY,UAAU,MAAM;AAC/C,YAAM,MAAM,OAAO,KAAK;AACxB,UAAI,YAAY;AAChB,eAAS,IAAI,GAAG,IAAI,IAAI,QAAQ,KAAK;AACnC,cAAM,OAAO,IAAI,WAAW,CAAC;AAC7B,qBAAa,aAAa,KAAK,YAAY;AAC3C,oBAAY,YAAY;AAAA,MAC1B;AACA,aAAO;AAAA,IACT;AAGA,QAAI,CAAC,KAAK,UAAU,IAAI,KAAK,GAAG;AAC9B,WAAK,UAAU,IAAI,OAAO,KAAK,QAAQ;AAAA,IACzC;AACA,WAAO,KAAK,UAAU,IAAI,KAAK;AAAA,EACjC;AAAA;AAAA;AAAA;AAAA,EAKA,YAAY,OAAoB;AAC9B,QAAI,UAAU,KAAM,QAAO;AAC3B,QAAI,UAAU,OAAW,QAAO;AAChC,QAAI,OAAO,UAAU,iBAAiB,OAAO,OAAO,KAAK,CAAC;AAE1D,WAAO,OAAO,KAAK,MAAM,KAAK,CAAC;AAAA,EACjC;AACF;AAKO,MAAM,0BAA0B,IAAI,kBAAA;AAEpC,UAAU,kBACZ,WACU;AACb,aAAW,YAAY,WAAW;AAChC,WAAO;AAAA,EACT;AACF;AAEO,UAAU,YACf,IACA,IACa;AACb,aAAW,KAAK,IAAI;AAClB,UAAM,GAAG,CAAC;AAAA,EACZ;AACF;"}

package/dist/esm/valueIndex.d.ts

@@ -0,0 +1,17 @@
+/**
+ * A map from a difference collection trace's keys -> (value, multiplicities) that changed.
+ * Used in operations like join and reduce where the operation needs to
+ * exploit the key-value structure of the data to run efficiently.
+ */
+export declare class ValueIndex<K, V> {
+    #private;
+    constructor();
+    toString(indent?: boolean): string;
+    get(key: K): [V, number] | undefined;
+    getMultiplicity(key: K): number;
+    entries(): MapIterator<[K, [V, number]]>;
+    has(key: K): boolean;
+    delete(key: K): void;
+    get size(): number;
+    addValue(key: K, v: [V, number]): void;
+}

package/dist/esm/valueIndex.js

@@ -0,0 +1,74 @@
+var __typeError = (msg) => {
+  throw TypeError(msg);
+};
+var __accessCheck = (obj, member, msg) => member.has(obj) || __typeError("Cannot " + msg);
+var __privateGet = (obj, member, getter) => (__accessCheck(obj, member, "read from private field"), getter ? getter.call(obj) : member.get(obj));
+var __privateAdd = (obj, member, value) => member.has(obj) ? __typeError("Cannot add the same private member more than once") : member instanceof WeakSet ? member.add(obj) : member.set(obj, value);
+var __privateSet = (obj, member, value, setter) => (__accessCheck(obj, member, "write to private field"), setter ? setter.call(obj, value) : member.set(obj, value), value);
+var __privateMethod = (obj, member, method) => (__accessCheck(obj, member, "access private method"), method);
+var _inner, _ValueIndex_instances, setMultiplicity_fn;
+import { hash } from "./hashing/hash.js";
+class ValueIndex {
+  // Maps key to the value and its multiplicity
+  constructor() {
+    __privateAdd(this, _ValueIndex_instances);
+    __privateAdd(this, _inner);
+    __privateSet(this, _inner, /* @__PURE__ */ new Map());
+  }
+  toString(indent = false) {
+    return `ValueIndex(${JSON.stringify(
+      [...__privateGet(this, _inner).entries()],
+      void 0,
+      indent ? 2 : void 0
+    )})`;
+  }
+  get(key) {
+    return __privateGet(this, _inner).get(key);
+  }
+  getMultiplicity(key) {
+    var _a;
+    return ((_a = this.get(key)) == null ? void 0 : _a[1]) ?? 0;
+  }
+  entries() {
+    return __privateGet(this, _inner).entries();
+  }
+  has(key) {
+    return __privateGet(this, _inner).has(key);
+  }
+  delete(key) {
+    __privateGet(this, _inner).delete(key);
+  }
+  get size() {
+    return __privateGet(this, _inner).size;
+  }
+  addValue(key, v) {
+    const [value, multiplicity] = v;
+    if (multiplicity === 0) {
+      return;
+    }
+    if (this.has(key)) {
+      const [currValue, currMultiplicity] = this.get(key);
+      if (hash(value) === hash(currValue)) {
+        __privateMethod(this, _ValueIndex_instances, setMultiplicity_fn).call(this, key, value, currMultiplicity + multiplicity);
+        return;
+      }
+      throw new Error(
+        `Cannot add value for key ${key} because it already exists in ValueIndex with a different value`
+      );
+    }
+    __privateGet(this, _inner).set(key, [value, multiplicity]);
+  }
+}
+_inner = new WeakMap();
+_ValueIndex_instances = new WeakSet();
+setMultiplicity_fn = function(key, value, multiplicity) {
+  if (multiplicity === 0) {
+    __privateGet(this, _inner).delete(key);
+  } else {
+    __privateGet(this, _inner).set(key, [value, multiplicity]);
+  }
+};
+export {
+  ValueIndex
+};
+//# sourceMappingURL=valueIndex.js.map

package/dist/esm/valueIndex.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"valueIndex.js","sources":["../../src/valueIndex.ts"],"sourcesContent":["import { hash } from \"./hashing/index.js\"\n\n/**\n * A map from a difference collection trace's keys -> (value, multiplicities) that changed.\n * Used in operations like join and reduce where the operation needs to\n * exploit the key-value structure of the data to run efficiently.\n */\nexport class ValueIndex<K, V> {\n  #inner: Map<K, [V, number]> // Maps key to the value and its multiplicity\n\n  constructor() {\n    this.#inner = new Map()\n  }\n\n  toString(indent = false): string {\n    return `ValueIndex(${JSON.stringify(\n      [...this.#inner.entries()],\n      undefined,\n      indent ? 2 : undefined\n    )})`\n  }\n\n  get(key: K): [V, number] | undefined {\n    return this.#inner.get(key)\n  }\n\n  getMultiplicity(key: K): number {\n    return this.get(key)?.[1] ?? 0\n  }\n\n  entries() {\n    return this.#inner.entries()\n  }\n\n  has(key: K): boolean {\n    return this.#inner.has(key)\n  }\n\n  delete(key: K): void {\n    this.#inner.delete(key)\n  }\n\n  get size(): number {\n    return this.#inner.size\n  }\n\n  addValue(key: K, v: [V, number]): void {\n    const [value, multiplicity] = v\n\n    if (multiplicity === 0) {\n      return\n    }\n\n    if (this.has(key)) {\n      const [currValue, currMultiplicity] = this.get(key)!\n      if (hash(value) === hash(currValue)) {\n        // Update the multiplicity\n        this.#setMultiplicity(key, value, currMultiplicity + multiplicity)\n        return\n      }\n      // Different value, not allowed.\n      // ValueIndex only supports one value per key.\n      throw new Error(\n        `Cannot add value for key ${key} because it already exists in ValueIndex with a different value`\n      )\n    }\n\n    this.#inner.set(key, [value, multiplicity])\n  }\n\n  #setMultiplicity(key: K, value: V, multiplicity: number): void {\n    if (multiplicity === 0) {\n      this.#inner.delete(key)\n    } else {\n      this.#inner.set(key, [value, multiplicity])\n    }\n  }\n}\n"],"names":[],"mappings":";;;;;;;;;;AAOO,MAAM,WAAiB;AAAA;AAAA,EAG5B,cAAc;AAHT;AACL;AAGE,uBAAK,4BAAa,IAAA;AAAA,EACpB;AAAA,EAEA,SAAS,SAAS,OAAe;AAC/B,WAAO,cAAc,KAAK;AAAA,MACxB,CAAC,GAAG,mBAAK,QAAO,SAAS;AAAA,MACzB;AAAA,MACA,SAAS,IAAI;AAAA,IAAA,CACd;AAAA,EACH;AAAA,EAEA,IAAI,KAAiC;AACnC,WAAO,mBAAK,QAAO,IAAI,GAAG;AAAA,EAC5B;AAAA,EAEA,gBAAgB,KAAgB;;AAC9B,aAAO,UAAK,IAAI,GAAG,MAAZ,mBAAgB,OAAM;AAAA,EAC/B;AAAA,EAEA,UAAU;AACR,WAAO,mBAAK,QAAO,QAAA;AAAA,EACrB;AAAA,EAEA,IAAI,KAAiB;AACnB,WAAO,mBAAK,QAAO,IAAI,GAAG;AAAA,EAC5B;AAAA,EAEA,OAAO,KAAc;AACnB,uBAAK,QAAO,OAAO,GAAG;AAAA,EACxB;AAAA,EAEA,IAAI,OAAe;AACjB,WAAO,mBAAK,QAAO;AAAA,EACrB;AAAA,EAEA,SAAS,KAAQ,GAAsB;AACrC,UAAM,CAAC,OAAO,YAAY,IAAI;AAE9B,QAAI,iBAAiB,GAAG;AACtB;AAAA,IACF;AAEA,QAAI,KAAK,IAAI,GAAG,GAAG;AACjB,YAAM,CAAC,WAAW,gBAAgB,IAAI,KAAK,IAAI,GAAG;AAClD,UAAI,KAAK,KAAK,MAAM,KAAK,SAAS,GAAG;AAEnC,8BAAK,2CAAL,WAAsB,KAAK,OAAO,mBAAmB;AACrD;AAAA,MACF;AAGA,YAAM,IAAI;AAAA,QACR,4BAA4B,GAAG;AAAA,MAAA;AAAA,IAEnC;AAEA,uBAAK,QAAO,IAAI,KAAK,CAAC,OAAO,YAAY,CAAC;AAAA,EAC5C;AASF;AArEE;AADK;AA+DL,qBAAA,SAAiB,KAAQ,OAAU,cAA4B;AAC7D,MAAI,iBAAiB,GAAG;AACtB,uBAAK,QAAO,OAAO,GAAG;AAAA,EACxB,OAAO;AACL,uBAAK,QAAO,IAAI,KAAK,CAAC,OAAO,YAAY,CAAC;AAAA,EAC5C;AACF;"}

package/package.json

@@ -1,15 +1,13 @@
 {
   "name": "@tanstack/db-ivm",
   "description": "Incremental View Maintenance for TanStack DB based on Differential Dataflow",
-  "version": "0.1.2",
+  "version": "0.1.3",
   "dependencies": {
     "fractional-indexing": "^3.2.0",
-    "murmurhash-js": "^1.0.0",
     "sorted-btree": "^1.8.1"
   },
   "devDependencies": {
     "@types/debug": "^4.1.12",
-    "@types/murmurhash-js": "^1.0.6",
     "@vitest/coverage-istanbul": "^3.0.9"
   },
   "exports": {

package/src/hashIndex.ts

@@ -0,0 +1,94 @@
+import { DefaultMap } from "./utils.js"
+import { hash } from "./hashing/index.js"
+import type { Hash } from "./hashing/index.js"
+
+/**
+ * A map from a difference collection trace's keys -> (value, multiplicities) that changed.
+ * Used in operations like join and reduce where the operation needs to
+ * exploit the key-value structure of the data to run efficiently.
+ */
+export class HashIndex<K, V> {
+  #inner: DefaultMap<K, DefaultMap<Hash, [V, number]>>
+
+  constructor() {
+    this.#inner = new DefaultMap<K, DefaultMap<Hash, [V, number]>>(
+      () => new DefaultMap<Hash, [V, number]>(() => [undefined as any as V, 0])
+    )
+    // #inner is as map of:
+    // {
+    //   [key]: {
+    //     [hash(value)]: [value, multiplicity]
+    //   }
+    // }
+  }
+
+  toString(indent = false): string {
+    return `HashIndex(${JSON.stringify(
+      [...this.#inner].map(([k, valueMap]) => [k, [...valueMap]]),
+      undefined,
+      indent ? 2 : undefined
+    )})`
+  }
+
+  get(key: K): Array<[V, number]> {
+    const valueMap = this.#inner.get(key)
+    return [...valueMap.values()]
+  }
+
+  getMultiplicity(key: K, value: V): number {
+    const valueMap = this.#inner.get(key)
+    const valueHash = hash(value)
+    const [, multiplicity] = valueMap.get(valueHash)
+    return multiplicity
+  }
+
+  entries() {
+    return this.#inner.entries()
+  }
+
+  *entriesIterator(): Generator<[K, [V, number]]> {
+    for (const [key, valueMap] of this.#inner.entries()) {
+      for (const [_valueHash, [value, multiplicity]] of valueMap.entries()) {
+        yield [key, [value, multiplicity]]
+      }
+    }
+  }
+
+  has(key: K): boolean {
+    return this.#inner.has(key)
+  }
+
+  delete(key: K): void {
+    this.#inner.delete(key)
+  }
+
+  get size(): number {
+    return this.#inner.size
+  }
+
+  /**
+   * Adds a value to the index and does not return anything
+   * except if the addition caused the value to be removed
+   * and the key to be left with only a single value.
+   * In that case, we return the single remaining value.
+   */
+  addValue(key: K, value: [V, number]): [V, number] | void {
+    const [val, multiplicity] = value
+    const valueMap = this.#inner.get(key)
+    const valueHash = hash(val)
+    const [, existingMultiplicity] = valueMap.get(valueHash)
+    const newMultiplicity = existingMultiplicity + multiplicity
+    if (multiplicity !== 0) {
+      if (newMultiplicity === 0) {
+        valueMap.delete(valueHash)
+        if (valueMap.size === 1) {
+          // Signal that the key only has a single remaining value
+          return valueMap.entries().next().value![1]
+        }
+      } else {
+        valueMap.set(valueHash, [val, newMultiplicity])
+      }
+    }
+    this.#inner.set(key, valueMap)
+  }
+}

package/src/hashing/hash.ts

@@ -0,0 +1,157 @@
+import { MurmurHashStream, randomHash } from "./murmur.js"
+import type { Hasher } from "./murmur.js"
+
+/*
+ * Implementation of structural hashing based on the Composites polyfill implementation:
+ * https://github.com/tc39/proposal-composites
+ */
+
+const TRUE = randomHash()
+const FALSE = randomHash()
+const NULL = randomHash()
+const UNDEFINED = randomHash()
+const KEY = randomHash()
+const FUNCTIONS = randomHash()
+const DATE_MARKER = randomHash()
+const OBJECT_MARKER = randomHash()
+const ARRAY_MARKER = randomHash()
+const MAP_MARKER = randomHash()
+const SET_MARKER = randomHash()
+
+const hashCache = new WeakMap<object, number>()
+
+export function hash(input: any): number {
+  const hasher = new MurmurHashStream()
+  updateHasher(hasher, input)
+  return hasher.digest()
+}
+
+function hashObject(input: object): number {
+  const cachedHash = hashCache.get(input)
+  if (cachedHash !== undefined) {
+    return cachedHash
+  }
+
+  let valueHash: number | undefined
+  if (input instanceof Date) {
+    valueHash = hashDate(input)
+  } else {
+    let plainObjectInput = input
+    let marker = OBJECT_MARKER
+
+    if (input instanceof Array) {
+      marker = ARRAY_MARKER
+    }
+
+    if (input instanceof Map) {
+      marker = MAP_MARKER
+      plainObjectInput = [...input.entries()]
+    }
+
+    if (input instanceof Set) {
+      marker = SET_MARKER
+      plainObjectInput = [...input.entries()]
+    }
+
+    if (
+      input instanceof Buffer ||
+      input instanceof Uint8Array ||
+      input instanceof File
+    ) {
+      // Deeply hashing these objects would be too costly
+      // but we also don't want to ignore them
+      // so we track them by reference and cache them in a weak map
+      return cachedReferenceHash(input)
+    }
+
+    valueHash = hashPlainObject(plainObjectInput, marker)
+  }
+
+  hashCache.set(input, valueHash)
+  return valueHash
+}
+
+function hashDate(input: Date): number {
+  const hasher = new MurmurHashStream()
+  hasher.update(DATE_MARKER)
+  hasher.update(input.getTime())
+  return hasher.digest()
+}
+
+function hashPlainObject(input: object, marker: number): number {
+  const hasher = new MurmurHashStream()
+
+  // Mark the type of the input
+  hasher.update(marker)
+  const keys = Object.keys(input)
+  keys.sort(keySort)
+  for (const key of keys) {
+    hasher.update(KEY)
+    hasher.update(key)
+    updateHasher(hasher, input[key as keyof typeof input])
+  }
+
+  return hasher.digest()
+}
+
+function updateHasher(hasher: Hasher, input: unknown): void {
+  if (input === null) {
+    hasher.update(NULL)
+    return
+  }
+  switch (typeof input) {
+    case `undefined`:
+      hasher.update(UNDEFINED)
+      return
+    case `boolean`:
+      hasher.update(input ? TRUE : FALSE)
+      return
+    case `number`:
+      // Normalize NaNs and -0
+      hasher.update(isNaN(input) ? NaN : input === 0 ? 0 : input)
+      return
+    case `bigint`:
+    case `string`:
+    case `symbol`:
+      hasher.update(input)
+      return
+    case `object`:
+      hasher.update(getCachedHash(input))
+      return
+    case `function`:
+      // Functions are assigned a globally unique ID
+      // and that ID is cached in the weak map
+      hasher.update(cachedReferenceHash(input))
+      return
+    default:
+      console.warn(
+        `Ignored input during hashing because it is of type ${typeof input} which is not supported`
+      )
+  }
+}
+
+function getCachedHash(input: object): number {
+  let valueHash = hashCache.get(input)
+  if (valueHash === undefined) {
+    valueHash = hashObject(input)
+  }
+  return valueHash
+}
+
+let nextRefId = 1
+function cachedReferenceHash(fn: object): number {
+  let valueHash = hashCache.get(fn)
+  if (valueHash === undefined) {
+    valueHash = nextRefId ^ FUNCTIONS
+    nextRefId++
+    hashCache.set(fn, valueHash)
+  }
+  return valueHash
+}
+
+/**
+ * Strings sorted lexicographically.
+ */
+function keySort(a: string, b: string): number {
+  return a.localeCompare(b)
+}

package/src/hashing/index.ts

@@ -0,0 +1,2 @@
+export { hash } from "./hash.js"
+export type { Hash, Hasher } from "./murmur.js"