@tanstack/db-ivm 0.1.20 → 0.1.24

package/dist/esm/hashing/hash.js.map

@@ -1 +1 @@
-{"version":3,"file":"hash.js","sources":["../../../src/hashing/hash.ts"],"sourcesContent":["import { MurmurHashStream, getSymbolIdentity, randomHash } from './murmur.js'\nimport type { Hasher } from './murmur.js'\n\n/*\n * Implementation of structural hashing based on the Composites polyfill implementation:\n * https://github.com/tc39/proposal-composites\n */\n\nconst TRUE = randomHash()\nconst FALSE = randomHash()\nconst NULL = randomHash()\nconst UNDEFINED = randomHash()\nconst KEY = randomHash()\nconst FUNCTIONS = randomHash()\nconst DATE_MARKER = randomHash()\nconst OBJECT_MARKER = randomHash()\nconst ARRAY_MARKER = randomHash()\nconst MAP_MARKER = randomHash()\nconst SET_MARKER = randomHash()\nconst UINT8ARRAY_MARKER = randomHash()\nconst TEMPORAL_MARKER = randomHash()\n// Bound structural recursion and value visits. Shared acyclic subtrees are\n// cached; cycles are rejected rather than given context-dependent hashes.\nconst MAX_STRUCTURAL_HASH_WORK = 1_000_000\nconst MAX_STRUCTURAL_HASH_DEPTH = 768\n\nconst temporalTypes = new Set([\n  `Temporal.Duration`,\n  `Temporal.Instant`,\n  `Temporal.PlainDate`,\n  `Temporal.PlainDateTime`,\n  `Temporal.PlainMonthDay`,\n  `Temporal.PlainTime`,\n  `Temporal.PlainYearMonth`,\n  `Temporal.ZonedDateTime`,\n])\n\ninterface TemporalLike {\n  [Symbol.toStringTag]: string\n  toString: () => string\n}\n\nfunction isTemporal(input: object): input is TemporalLike {\n  const tag = (input as Record<symbol, unknown>)[Symbol.toStringTag]\n  return typeof tag === `string` && temporalTypes.has(tag)\n}\n\n// Maximum byte length for Uint8Arrays to hash by content instead of reference\n// Arrays smaller than this will be hashed by content, allowing proper equality comparisons\n// for small arrays like ULIDs (16 bytes) while still avoiding performance costs for large arrays\nconst UINT8ARRAY_CONTENT_HASH_THRESHOLD = 128\n\nconst hashCache = new WeakMap<object, number>()\n\n/** @internal Register a mutable handle before it enters a structural value. */\nexport function registerOpaqueHash(value: object): void {\n  cachedReferenceHash(value)\n}\n\ntype HashContext = {\n  activeObjects: Set<object>\n  work: number\n  pendingHashes: Map<object, number>\n}\n\nexport function hash(input: any): number {\n  const hasher = new MurmurHashStream()\n  updateHasher(hasher, input)\n  return hasher.digest()\n}\n\nfunction hashObject(input: object, context: HashContext): number {\n  if (context.activeObjects.size >= MAX_STRUCTURAL_HASH_DEPTH) {\n    throw new RangeError(\n      `Value is too complex to hash safely: structural depth`,\n    )\n  }\n\n  context.activeObjects.add(input)\n\n  let valueHash: number | undefined\n  try {\n    if (input instanceof Date) {\n      valueHash = hashDate(input)\n    } else if (isBinaryValue(input)) {\n      valueHash = hashUint8Array(input)\n    } else if (isTemporal(input)) {\n      valueHash = hashTemporal(input)\n    } else {\n      let plainObjectInput = input\n      let marker = OBJECT_MARKER\n\n      if (input instanceof Array) {\n        marker = ARRAY_MARKER\n      }\n\n      if (input instanceof Map) {\n        marker = MAP_MARKER\n        plainObjectInput = [...input.entries()]\n      }\n\n      if (input instanceof Set) {\n        marker = SET_MARKER\n        plainObjectInput = [...input.entries()]\n      }\n\n      valueHash = hashPlainObject(plainObjectInput, marker, context)\n    }\n  } finally {\n    context.activeObjects.delete(input)\n  }\n\n  context.pendingHashes.set(input, valueHash)\n  return valueHash\n}\n\nfunction hashDate(input: Date): number {\n  const hasher = new MurmurHashStream()\n  hasher.update(DATE_MARKER)\n  hasher.update(input.getTime())\n  return hasher.digest()\n}\n\nfunction hashUint8Array(input: Uint8Array): number {\n  const hasher = new MurmurHashStream()\n  hasher.update(UINT8ARRAY_MARKER)\n  // Hash the byte length first to differentiate arrays of different sizes\n  hasher.update(input.byteLength)\n  // Hash each byte in the array\n  for (let i = 0; i < input.byteLength; i++) {\n    hasher.writeByte(input[i]!)\n  }\n  return hasher.digest()\n}\n\nfunction hashTemporal(input: TemporalLike): number {\n  const hasher = new MurmurHashStream()\n  hasher.update(TEMPORAL_MARKER)\n  hasher.update(input[Symbol.toStringTag])\n  hasher.update(input.toString())\n  return hasher.digest()\n}\n\nfunction hashPlainObject(\n  input: object,\n  marker: number,\n  context: HashContext,\n): number {\n  const hasher = new MurmurHashStream()\n\n  // Mark the type of the input\n  hasher.update(marker)\n  const keys = Object.keys(input)\n  keys.sort(keySort)\n  for (const key of keys) {\n    hasher.update(KEY)\n    hasher.update(key)\n    updateHasher(hasher, input[key as keyof typeof input], context)\n  }\n  const symbolKeys = Object.getOwnPropertySymbols(input)\n    .filter((key) => Object.prototype.propertyIsEnumerable.call(input, key))\n    .sort((left, right) => getSymbolIdentity(left) - getSymbolIdentity(right))\n  for (const key of symbolKeys) {\n    hasher.update(KEY)\n    hasher.update(key)\n    updateHasher(hasher, input[key as keyof typeof input], context)\n  }\n\n  return hasher.digest()\n}\n\nfunction updateHasher(\n  hasher: Hasher,\n  input: unknown,\n  context?: HashContext,\n): void {\n  if (context && ++context.work > MAX_STRUCTURAL_HASH_WORK) {\n    throw new RangeError(`Value is too complex to hash safely: structural work`)\n  }\n  if (input === null) {\n    hasher.update(NULL)\n    return\n  }\n  switch (typeof input) {\n    case `undefined`:\n      hasher.update(UNDEFINED)\n      return\n    case `boolean`:\n      hasher.update(input ? TRUE : FALSE)\n      return\n    case `number`:\n      // Normalize NaNs and -0\n      hasher.update(isNaN(input) ? NaN : input === 0 ? 0 : input)\n      return\n    case `bigint`:\n    case `string`:\n    case `symbol`:\n      hasher.update(input)\n      return\n    case `object`:\n      hasher.update(getCachedHash(input, context))\n      return\n    case `function`:\n      // Functions are assigned a globally unique ID\n      // and that ID is cached in the weak map\n      hasher.update(cachedReferenceHash(input))\n      return\n    default:\n      console.warn(\n        `Ignored input during hashing because it is of type ${typeof input} which is not supported`,\n      )\n  }\n}\n\nfunction getCachedHash(input: object, context?: HashContext): number {\n  if (!context) {\n    const cached = hashCache.get(input)\n    if (cached !== undefined) return cached\n    if (isReferenceHashedObject(input)) return cachedReferenceHash(input)\n\n    // Only an uncached structural root needs graph traversal state. Commit its\n    // cache entries after success so a failed traversal cannot poison retries.\n    context = {\n      activeObjects: new Set(),\n      work: 0,\n      pendingHashes: new Map(),\n    }\n    const result = hashObject(input, context)\n    for (const [object, valueHash] of context.pendingHashes) {\n      hashCache.set(object, valueHash)\n    }\n    return result\n  }\n\n  if (context.activeObjects.has(input)) {\n    throw new TypeError(`Cannot hash cyclic structural values`)\n  }\n\n  // Opaque leaves cannot contain structural back-references. Resolve them\n  // before entering structural recursion, even when they have user properties.\n  if (isReferenceHashedObject(input)) return cachedReferenceHash(input)\n\n  const valueHash = hashCache.get(input) ?? context.pendingHashes.get(input)\n  if (valueHash !== undefined) return valueHash\n\n  return hashObject(input, context)\n}\n\nfunction isReferenceHashedObject(input: object): boolean {\n  return (\n    input instanceof File ||\n    (isBinaryValue(input) &&\n      input.byteLength > UINT8ARRAY_CONTENT_HASH_THRESHOLD)\n  )\n}\n\nfunction isBinaryValue(input: object): input is Uint8Array {\n  return (\n    (typeof Buffer !== `undefined` && input instanceof Buffer) ||\n    input instanceof Uint8Array\n  )\n}\n\nlet nextRefId = 1\nfunction cachedReferenceHash(fn: object): number {\n  let valueHash = hashCache.get(fn)\n  if (valueHash === undefined) {\n    valueHash = nextRefId ^ FUNCTIONS\n    nextRefId++\n    hashCache.set(fn, valueHash)\n  }\n  return valueHash\n}\n\n/**\n * Strings sorted lexicographically.\n */\nfunction keySort(a: string, b: string): number {\n  return a.localeCompare(b)\n}\n"],"names":["valueHash"],"mappings":";AAQA,MAAM,OAAO,WAAA;AACb,MAAM,QAAQ,WAAA;AACd,MAAM,OAAO,WAAA;AACb,MAAM,YAAY,WAAA;AAClB,MAAM,MAAM,WAAA;AACZ,MAAM,YAAY,WAAA;AAClB,MAAM,cAAc,WAAA;AACpB,MAAM,gBAAgB,WAAA;AACtB,MAAM,eAAe,WAAA;AACrB,MAAM,aAAa,WAAA;AACnB,MAAM,aAAa,WAAA;AACnB,MAAM,oBAAoB,WAAA;AAC1B,MAAM,kBAAkB,WAAA;AAGxB,MAAM,2BAA2B;AACjC,MAAM,4BAA4B;AAElC,MAAM,oCAAoB,IAAI;AAAA,EAC5B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF,CAAC;AAOD,SAAS,WAAW,OAAsC;AACxD,QAAM,MAAO,MAAkC,OAAO,WAAW;AACjE,SAAO,OAAO,QAAQ,YAAY,cAAc,IAAI,GAAG;AACzD;AAKA,MAAM,oCAAoC;AAE1C,MAAM,gCAAgB,QAAA;AAGf,SAAS,mBAAmB,OAAqB;AACtD,sBAAoB,KAAK;AAC3B;AAQO,SAAS,KAAK,OAAoB;AACvC,QAAM,SAAS,IAAI,iBAAA;AACnB,eAAa,QAAQ,KAAK;AAC1B,SAAO,OAAO,OAAA;AAChB;AAEA,SAAS,WAAW,OAAe,SAA8B;AAC/D,MAAI,QAAQ,cAAc,QAAQ,2BAA2B;AAC3D,UAAM,IAAI;AAAA,MACR;AAAA,IAAA;AAAA,EAEJ;AAEA,UAAQ,cAAc,IAAI,KAAK;AAE/B,MAAI;AACJ,MAAI;AACF,QAAI,iBAAiB,MAAM;AACzB,kBAAY,SAAS,KAAK;AAAA,IAC5B,WAAW,cAAc,KAAK,GAAG;AAC/B,kBAAY,eAAe,KAAK;AAAA,IAClC,WAAW,WAAW,KAAK,GAAG;AAC5B,kBAAY,aAAa,KAAK;AAAA,IAChC,OAAO;AACL,UAAI,mBAAmB;AACvB,UAAI,SAAS;AAEb,UAAI,iBAAiB,OAAO;AAC1B,iBAAS;AAAA,MACX;AAEA,UAAI,iBAAiB,KAAK;AACxB,iBAAS;AACT,2BAAmB,CAAC,GAAG,MAAM,SAAS;AAAA,MACxC;AAEA,UAAI,iBAAiB,KAAK;AACxB,iBAAS;AACT,2BAAmB,CAAC,GAAG,MAAM,SAAS;AAAA,MACxC;AAEA,kBAAY,gBAAgB,kBAAkB,QAAQ,OAAO;AAAA,IAC/D;AAAA,EACF,UAAA;AACE,YAAQ,cAAc,OAAO,KAAK;AAAA,EACpC;AAEA,UAAQ,cAAc,IAAI,OAAO,SAAS;AAC1C,SAAO;AACT;AAEA,SAAS,SAAS,OAAqB;AACrC,QAAM,SAAS,IAAI,iBAAA;AACnB,SAAO,OAAO,WAAW;AACzB,SAAO,OAAO,MAAM,SAAS;AAC7B,SAAO,OAAO,OAAA;AAChB;AAEA,SAAS,eAAe,OAA2B;AACjD,QAAM,SAAS,IAAI,iBAAA;AACnB,SAAO,OAAO,iBAAiB;AAE/B,SAAO,OAAO,MAAM,UAAU;AAE9B,WAAS,IAAI,GAAG,IAAI,MAAM,YAAY,KAAK;AACzC,WAAO,UAAU,MAAM,CAAC,CAAE;AAAA,EAC5B;AACA,SAAO,OAAO,OAAA;AAChB;AAEA,SAAS,aAAa,OAA6B;AACjD,QAAM,SAAS,IAAI,iBAAA;AACnB,SAAO,OAAO,eAAe;AAC7B,SAAO,OAAO,MAAM,OAAO,WAAW,CAAC;AACvC,SAAO,OAAO,MAAM,UAAU;AAC9B,SAAO,OAAO,OAAA;AAChB;AAEA,SAAS,gBACP,OACA,QACA,SACQ;AACR,QAAM,SAAS,IAAI,iBAAA;AAGnB,SAAO,OAAO,MAAM;AACpB,QAAM,OAAO,OAAO,KAAK,KAAK;AAC9B,OAAK,KAAK,OAAO;AACjB,aAAW,OAAO,MAAM;AACtB,WAAO,OAAO,GAAG;AACjB,WAAO,OAAO,GAAG;AACjB,iBAAa,QAAQ,MAAM,GAAyB,GAAG,OAAO;AAAA,EAChE;AACA,QAAM,aAAa,OAAO,sBAAsB,KAAK,EAClD,OAAO,CAAC,QAAQ,OAAO,UAAU,qBAAqB,KAAK,OAAO,GAAG,CAAC,EACtE,KAAK,CAAC,MAAM,UAAU,kBAAkB,IAAI,IAAI,kBAAkB,KAAK,CAAC;AAC3E,aAAW,OAAO,YAAY;AAC5B,WAAO,OAAO,GAAG;AACjB,WAAO,OAAO,GAAG;AACjB,iBAAa,QAAQ,MAAM,GAAyB,GAAG,OAAO;AAAA,EAChE;AAEA,SAAO,OAAO,OAAA;AAChB;AAEA,SAAS,aACP,QACA,OACA,SACM;AACN,MAAI,WAAW,EAAE,QAAQ,OAAO,0BAA0B;AACxD,UAAM,IAAI,WAAW,sDAAsD;AAAA,EAC7E;AACA,MAAI,UAAU,MAAM;AAClB,WAAO,OAAO,IAAI;AAClB;AAAA,EACF;AACA,UAAQ,OAAO,OAAA;AAAA,IACb,KAAK;AACH,aAAO,OAAO,SAAS;AACvB;AAAA,IACF,KAAK;AACH,aAAO,OAAO,QAAQ,OAAO,KAAK;AAClC;AAAA,IACF,KAAK;AAEH,aAAO,OAAO,MAAM,KAAK,IAAI,MAAM,UAAU,IAAI,IAAI,KAAK;AAC1D;AAAA,IACF,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AACH,aAAO,OAAO,KAAK;AACnB;AAAA,IACF,KAAK;AACH,aAAO,OAAO,cAAc,OAAO,OAAO,CAAC;AAC3C;AAAA,IACF,KAAK;AAGH,aAAO,OAAO,oBAAoB,KAAK,CAAC;AACxC;AAAA,IACF;AACE,cAAQ;AAAA,QACN,sDAAsD,OAAO,KAAK;AAAA,MAAA;AAAA,EACpE;AAEN;AAEA,SAAS,cAAc,OAAe,SAA+B;AACnE,MAAI,CAAC,SAAS;AACZ,UAAM,SAAS,UAAU,IAAI,KAAK;AAClC,QAAI,WAAW,OAAW,QAAO;AACjC,QAAI,wBAAwB,KAAK,EAAG,QAAO,oBAAoB,KAAK;AAIpE,cAAU;AAAA,MACR,mCAAmB,IAAA;AAAA,MACnB,MAAM;AAAA,MACN,mCAAmB,IAAA;AAAA,IAAI;AAEzB,UAAM,SAAS,WAAW,OAAO,OAAO;AACxC,eAAW,CAAC,QAAQA,UAAS,KAAK,QAAQ,eAAe;AACvD,gBAAU,IAAI,QAAQA,UAAS;AAAA,IACjC;AACA,WAAO;AAAA,EACT;AAEA,MAAI,QAAQ,cAAc,IAAI,KAAK,GAAG;AACpC,UAAM,IAAI,UAAU,sCAAsC;AAAA,EAC5D;AAIA,MAAI,wBAAwB,KAAK,EAAG,QAAO,oBAAoB,KAAK;AAEpE,QAAM,YAAY,UAAU,IAAI,KAAK,KAAK,QAAQ,cAAc,IAAI,KAAK;AACzE,MAAI,cAAc,OAAW,QAAO;AAEpC,SAAO,WAAW,OAAO,OAAO;AAClC;AAEA,SAAS,wBAAwB,OAAwB;AACvD,SACE,iBAAiB,QAChB,cAAc,KAAK,KAClB,MAAM,aAAa;AAEzB;AAEA,SAAS,cAAc,OAAoC;AACzD,SACG,OAAO,WAAW,eAAe,iBAAiB,UACnD,iBAAiB;AAErB;AAEA,IAAI,YAAY;AAChB,SAAS,oBAAoB,IAAoB;AAC/C,MAAI,YAAY,UAAU,IAAI,EAAE;AAChC,MAAI,cAAc,QAAW;AAC3B,gBAAY,YAAY;AACxB;AACA,cAAU,IAAI,IAAI,SAAS;AAAA,EAC7B;AACA,SAAO;AACT;AAKA,SAAS,QAAQ,GAAW,GAAmB;AAC7C,SAAO,EAAE,cAAc,CAAC;AAC1B;"}
+{"version":3,"file":"hash.js","sources":["../../../src/hashing/hash.ts"],"sourcesContent":["import { MurmurHashStream, getSymbolIdentity, randomHash } from './murmur.js'\nimport { isTemporal } from './temporal.js'\nimport type { Hasher } from './murmur.js'\n\n/*\n * Implementation of structural hashing based on the Composites polyfill implementation:\n * https://github.com/tc39/proposal-composites\n */\n\nconst TRUE = randomHash()\nconst FALSE = randomHash()\nconst NULL = randomHash()\nconst UNDEFINED = randomHash()\nconst KEY = randomHash()\nconst FUNCTIONS = randomHash()\n// Identity is distinct even if two random hash values happen to collide.\nconst DATE_MARKER = { hash: randomHash() }\nconst REGEXP_MARKER = { hash: randomHash() }\nconst OBJECT_MARKER = { hash: randomHash() }\nconst ARRAY_MARKER = { hash: randomHash() }\nconst MAP_MARKER = { hash: randomHash() }\nconst SET_MARKER = { hash: randomHash() }\nconst UINT8ARRAY_MARKER = { hash: randomHash() }\nconst TEMPORAL_MARKER = { hash: randomHash() }\n// Bound structural recursion and value visits. Shared acyclic subtrees are\n// cached; cycles are rejected rather than given context-dependent hashes.\nconst MAX_STRUCTURAL_HASH_WORK = 1_000_000\nconst MAX_STRUCTURAL_HASH_DEPTH = 768\n\n// Maximum byte length for Uint8Arrays to hash by content instead of reference\n// Arrays smaller than this will be hashed by content, allowing proper equality comparisons\n// for small arrays like ULIDs (16 bytes) while still avoiding performance costs for large arrays\nconst UINT8ARRAY_CONTENT_HASH_THRESHOLD = 128\n\nconst hashCache = new WeakMap<object, number>()\nconst referenceValues = new WeakSet<object>()\n\n/** @internal Register a mutable handle before it enters a structural value. */\nexport function registerOpaqueHash(value: object): void {\n  cachedReferenceHash(value)\n}\n\ntype HashContext = {\n  activeObjects: Set<object>\n  work: number\n  pendingHashes: Map<object, number>\n}\n\nexport function hash(input: any): number {\n  const hasher = new MurmurHashStream()\n  updateHasher(hasher, input)\n  return hasher.digest()\n}\n\nfunction hashObject(input: object, context: HashContext): number {\n  if (context.activeObjects.size >= MAX_STRUCTURAL_HASH_DEPTH) {\n    throw new RangeError(\n      `Value is too complex to hash safely: structural depth`,\n    )\n  }\n\n  context.activeObjects.add(input)\n\n  let valueHash: number | undefined\n  try {\n    const [marker, header, body] = objectParts(input)\n    const hasher = new MurmurHashStream()\n    hasher.update(marker.hash)\n    // Header values of a type with a body (array length, RegExp fields)\n    // count toward the work cap; Date, binary, and Temporal headers do not.\n    for (const value of header)\n      if (body) updateHasher(hasher, value, context)\n      else hasher.update(value as string | number)\n    if (body) {\n      // Sorted string keys, then enumerable symbol keys by identity.\n      const keys: Array<string | symbol> = Object.keys(body).sort(keySort)\n      keys.push(\n        ...Object.getOwnPropertySymbols(body)\n          .filter((key) =>\n            Object.prototype.propertyIsEnumerable.call(body, key),\n          )\n          .sort(\n            (left, right) => getSymbolIdentity(left) - getSymbolIdentity(right),\n          ),\n      )\n      for (const key of keys) {\n        hasher.update(KEY)\n        hasher.update(key)\n        updateHasher(hasher, body[key as keyof typeof body], context)\n      }\n    }\n    valueHash = hasher.digest()\n  } finally {\n    context.activeObjects.delete(input)\n  }\n\n  context.pendingHashes.set(input, valueHash)\n  return valueHash\n}\n\n/**\n * The one type dispatch shared by structural hashing and equality: a type\n * marker, primitive header values, and an optional object whose enumerable\n * own properties are compared structurally.\n */\ntype ObjectParts = [\n  marker: Readonly<{ hash: number }>,\n  // A RegExp `lastIndex` can hold any value; the other headers are strings\n  // or numbers.\n  header: ReadonlyArray<unknown>,\n  body?: object,\n]\n\nfunction objectParts(input: object): ObjectParts {\n  if (input instanceof Date) {\n    const time = input.getTime()\n    // Invalid dates are equal to each other; headers compare with `===`.\n    return [DATE_MARKER, [Number.isNaN(time) ? `invalid` : time]]\n  }\n  if (isBinaryValue(input))\n    return [\n      UINT8ARRAY_MARKER,\n      [String.fromCharCode.apply(null, input as unknown as Array<number>)],\n    ]\n  if (isTemporal(input))\n    return [TEMPORAL_MARKER, [input[Symbol.toStringTag], input.toString()]]\n  if (input instanceof RegExp)\n    return [REGEXP_MARKER, [input.source, input.flags, input.lastIndex], input]\n  if (input instanceof Map) return [MAP_MARKER, [], [...input.entries()]]\n  if (input instanceof Set) return [SET_MARKER, [], [...input.values()]]\n  return input instanceof Array\n    ? [ARRAY_MARKER, [input.length], input]\n    : [OBJECT_MARKER, [], input]\n}\n\nfunction updateHasher(\n  hasher: Hasher,\n  input: unknown,\n  context?: HashContext,\n): void {\n  if (context && ++context.work > MAX_STRUCTURAL_HASH_WORK) {\n    throw new RangeError(`Value is too complex to hash safely: structural work`)\n  }\n  if (input === null) {\n    hasher.update(NULL)\n    return\n  }\n  switch (typeof input) {\n    case `undefined`:\n      hasher.update(UNDEFINED)\n      return\n    case `boolean`:\n      hasher.update(input ? TRUE : FALSE)\n      return\n    case `number`:\n      // Normalize NaNs and -0\n      hasher.update(isNaN(input) ? NaN : input === 0 ? 0 : input)\n      return\n    case `bigint`:\n    case `string`:\n    case `symbol`:\n      hasher.update(input)\n      return\n    case `object`:\n      hasher.update(getCachedHash(input, context))\n      return\n    case `function`:\n      // Functions are assigned a globally unique ID\n      // and that ID is cached in the weak map\n      hasher.update(cachedReferenceHash(input))\n  }\n}\n\nfunction getCachedHash(input: object, context?: HashContext): number {\n  // Active objects are neither cached nor pending, so the cycle check below\n  // still sees every back-reference.\n  const cached = hashCache.get(input) ?? context?.pendingHashes.get(input)\n  if (cached !== undefined) return cached\n\n  // Opaque leaves cannot contain structural back-references. Resolve them\n  // before entering structural recursion, even when they have user properties.\n  if (isReferenceHashedObject(input)) return cachedReferenceHash(input)\n\n  if (context) {\n    if (context.activeObjects.has(input)) {\n      throw new TypeError(`Cannot hash cyclic structural values`)\n    }\n    return hashObject(input, context)\n  }\n\n  // Only an uncached structural root needs graph traversal state. Commit its\n  // cache entries after success so a failed traversal cannot poison retries.\n  const root: HashContext = {\n    activeObjects: new Set(),\n    work: 0,\n    pendingHashes: new Map(),\n  }\n  const result = hashObject(input, root)\n  for (const [object, valueHash] of root.pendingHashes) {\n    hashCache.set(object, valueHash)\n  }\n  return result\n}\n\nfunction isReferenceHashedObject(input: object): boolean {\n  return (\n    referenceValues.has(input) ||\n    (typeof File !== `undefined` && input instanceof File) ||\n    (isBinaryValue(input) &&\n      input.byteLength > UINT8ARRAY_CONTENT_HASH_THRESHOLD)\n  )\n}\n\n// Buffer can come from another realm than Uint8Array (for example under jsdom).\nfunction isBinaryValue(input: object): input is Uint8Array {\n  return (\n    (typeof Buffer !== `undefined` && input instanceof Buffer) ||\n    input instanceof Uint8Array\n  )\n}\n\n/** @internal Compare immutable structural values without computing a digest.\n * Pair memoization also permits cyclic values that structural hashing rejects.\n * Reference-valued leaves remain opaque, including registered mutable handles.\n */\nexport function equalHashValues(left: unknown, right: unknown): boolean {\n  const compared = new Map<object, Set<object>>()\n  const keys = (value: object) =>\n    Reflect.ownKeys(value).filter((key) =>\n      Object.prototype.propertyIsEnumerable.call(value, key),\n    )\n  function equal(a: unknown, b: unknown): boolean {\n    if (a === b || (Number.isNaN(a) && Number.isNaN(b))) return true\n    if (\n      a === null ||\n      b === null ||\n      typeof a !== `object` ||\n      typeof b !== `object` ||\n      isReferenceHashedObject(a) ||\n      isReferenceHashedObject(b)\n    )\n      return false\n    // Arrays from another realm take the object marker but keep this check.\n    if (Array.isArray(a) && Array.isArray(b) && a.length !== b.length)\n      return false\n\n    // Revisited pairs close cycles and avoid expanding shared subtrees. Check\n    // before objectParts copies Map or Set entries; record only pairs with a\n    // body, after their markers and headers match.\n    const peers = compared.get(a)\n    if (peers?.has(b)) return true\n    const [aMarker, aHeader, aBody] = objectParts(a)\n    const [bMarker, bHeader, bBody] = objectParts(b)\n    if (\n      aMarker !== bMarker ||\n      !aHeader.every((value, index) => value === bHeader[index])\n    )\n      return false\n    if (!aBody || !bBody) return true\n    if (peers) peers.add(b)\n    else compared.set(a, new Set([b]))\n    const aKeys = keys(aBody)\n    return (\n      aKeys.length === keys(bBody).length &&\n      aKeys.every(\n        (key) =>\n          Object.prototype.propertyIsEnumerable.call(bBody, key) &&\n          equal(aBody[key as keyof object], bBody[key as keyof object]),\n      )\n    )\n  }\n  return equal(left, right)\n}\n\nlet nextRefId = 1\nfunction cachedReferenceHash(fn: object): number {\n  let valueHash = hashCache.get(fn)\n  if (valueHash === undefined) {\n    valueHash = nextRefId ^ FUNCTIONS\n    nextRefId++\n    hashCache.set(fn, valueHash)\n    referenceValues.add(fn)\n  }\n  return valueHash\n}\n\n/**\n * Strings sorted lexicographically.\n */\nfunction keySort(a: string, b: string): number {\n  return a.localeCompare(b)\n}\n"],"names":[],"mappings":";;AASA,MAAM,OAAO,WAAA;AACb,MAAM,QAAQ,WAAA;AACd,MAAM,OAAO,WAAA;AACb,MAAM,YAAY,WAAA;AAClB,MAAM,MAAM,WAAA;AACZ,MAAM,YAAY,WAAA;AAElB,MAAM,cAAc,EAAE,MAAM,aAAW;AACvC,MAAM,gBAAgB,EAAE,MAAM,aAAW;AACzC,MAAM,gBAAgB,EAAE,MAAM,aAAW;AACzC,MAAM,eAAe,EAAE,MAAM,aAAW;AACxC,MAAM,aAAa,EAAE,MAAM,aAAW;AACtC,MAAM,aAAa,EAAE,MAAM,aAAW;AACtC,MAAM,oBAAoB,EAAE,MAAM,aAAW;AAC7C,MAAM,kBAAkB,EAAE,MAAM,aAAW;AAG3C,MAAM,2BAA2B;AACjC,MAAM,4BAA4B;AAKlC,MAAM,oCAAoC;AAE1C,MAAM,gCAAgB,QAAA;AACtB,MAAM,sCAAsB,QAAA;AAGrB,SAAS,mBAAmB,OAAqB;AACtD,sBAAoB,KAAK;AAC3B;AAQO,SAAS,KAAK,OAAoB;AACvC,QAAM,SAAS,IAAI,iBAAA;AACnB,eAAa,QAAQ,KAAK;AAC1B,SAAO,OAAO,OAAA;AAChB;AAEA,SAAS,WAAW,OAAe,SAA8B;AAC/D,MAAI,QAAQ,cAAc,QAAQ,2BAA2B;AAC3D,UAAM,IAAI;AAAA,MACR;AAAA,IAAA;AAAA,EAEJ;AAEA,UAAQ,cAAc,IAAI,KAAK;AAE/B,MAAI;AACJ,MAAI;AACF,UAAM,CAAC,QAAQ,QAAQ,IAAI,IAAI,YAAY,KAAK;AAChD,UAAM,SAAS,IAAI,iBAAA;AACnB,WAAO,OAAO,OAAO,IAAI;AAGzB,eAAW,SAAS;AAClB,UAAI,KAAM,cAAa,QAAQ,OAAO,OAAO;AAAA,UACxC,QAAO,OAAO,KAAwB;AAC7C,QAAI,MAAM;AAER,YAAM,OAA+B,OAAO,KAAK,IAAI,EAAE,KAAK,OAAO;AACnE,WAAK;AAAA,QACH,GAAG,OAAO,sBAAsB,IAAI,EACjC;AAAA,UAAO,CAAC,QACP,OAAO,UAAU,qBAAqB,KAAK,MAAM,GAAG;AAAA,QAAA,EAErD;AAAA,UACC,CAAC,MAAM,UAAU,kBAAkB,IAAI,IAAI,kBAAkB,KAAK;AAAA,QAAA;AAAA,MACpE;AAEJ,iBAAW,OAAO,MAAM;AACtB,eAAO,OAAO,GAAG;AACjB,eAAO,OAAO,GAAG;AACjB,qBAAa,QAAQ,KAAK,GAAwB,GAAG,OAAO;AAAA,MAC9D;AAAA,IACF;AACA,gBAAY,OAAO,OAAA;AAAA,EACrB,UAAA;AACE,YAAQ,cAAc,OAAO,KAAK;AAAA,EACpC;AAEA,UAAQ,cAAc,IAAI,OAAO,SAAS;AAC1C,SAAO;AACT;AAeA,SAAS,YAAY,OAA4B;AAC/C,MAAI,iBAAiB,MAAM;AACzB,UAAM,OAAO,MAAM,QAAA;AAEnB,WAAO,CAAC,aAAa,CAAC,OAAO,MAAM,IAAI,IAAI,YAAY,IAAI,CAAC;AAAA,EAC9D;AACA,MAAI,cAAc,KAAK;AACrB,WAAO;AAAA,MACL;AAAA,MACA,CAAC,OAAO,aAAa,MAAM,MAAM,KAAiC,CAAC;AAAA,IAAA;AAEvE,MAAI,WAAW,KAAK;AAClB,WAAO,CAAC,iBAAiB,CAAC,MAAM,OAAO,WAAW,GAAG,MAAM,SAAA,CAAU,CAAC;AACxE,MAAI,iBAAiB;AACnB,WAAO,CAAC,eAAe,CAAC,MAAM,QAAQ,MAAM,OAAO,MAAM,SAAS,GAAG,KAAK;AAC5E,MAAI,iBAAiB,IAAK,QAAO,CAAC,YAAY,CAAA,GAAI,CAAC,GAAG,MAAM,QAAA,CAAS,CAAC;AACtE,MAAI,iBAAiB,IAAK,QAAO,CAAC,YAAY,CAAA,GAAI,CAAC,GAAG,MAAM,OAAA,CAAQ,CAAC;AACrE,SAAO,iBAAiB,QACpB,CAAC,cAAc,CAAC,MAAM,MAAM,GAAG,KAAK,IACpC,CAAC,eAAe,CAAA,GAAI,KAAK;AAC/B;AAEA,SAAS,aACP,QACA,OACA,SACM;AACN,MAAI,WAAW,EAAE,QAAQ,OAAO,0BAA0B;AACxD,UAAM,IAAI,WAAW,sDAAsD;AAAA,EAC7E;AACA,MAAI,UAAU,MAAM;AAClB,WAAO,OAAO,IAAI;AAClB;AAAA,EACF;AACA,UAAQ,OAAO,OAAA;AAAA,IACb,KAAK;AACH,aAAO,OAAO,SAAS;AACvB;AAAA,IACF,KAAK;AACH,aAAO,OAAO,QAAQ,OAAO,KAAK;AAClC;AAAA,IACF,KAAK;AAEH,aAAO,OAAO,MAAM,KAAK,IAAI,MAAM,UAAU,IAAI,IAAI,KAAK;AAC1D;AAAA,IACF,KAAK;AAAA,IACL,KAAK;AAAA,IACL,KAAK;AACH,aAAO,OAAO,KAAK;AACnB;AAAA,IACF,KAAK;AACH,aAAO,OAAO,cAAc,OAAO,OAAO,CAAC;AAC3C;AAAA,IACF,KAAK;AAGH,aAAO,OAAO,oBAAoB,KAAK,CAAC;AAAA,EAAA;AAE9C;AAEA,SAAS,cAAc,OAAe,SAA+B;AAGnE,QAAM,SAAS,UAAU,IAAI,KAAK,KAAK,SAAS,cAAc,IAAI,KAAK;AACvE,MAAI,WAAW,OAAW,QAAO;AAIjC,MAAI,wBAAwB,KAAK,EAAG,QAAO,oBAAoB,KAAK;AAEpE,MAAI,SAAS;AACX,QAAI,QAAQ,cAAc,IAAI,KAAK,GAAG;AACpC,YAAM,IAAI,UAAU,sCAAsC;AAAA,IAC5D;AACA,WAAO,WAAW,OAAO,OAAO;AAAA,EAClC;AAIA,QAAM,OAAoB;AAAA,IACxB,mCAAmB,IAAA;AAAA,IACnB,MAAM;AAAA,IACN,mCAAmB,IAAA;AAAA,EAAI;AAEzB,QAAM,SAAS,WAAW,OAAO,IAAI;AACrC,aAAW,CAAC,QAAQ,SAAS,KAAK,KAAK,eAAe;AACpD,cAAU,IAAI,QAAQ,SAAS;AAAA,EACjC;AACA,SAAO;AACT;AAEA,SAAS,wBAAwB,OAAwB;AACvD,SACE,gBAAgB,IAAI,KAAK,KACxB,OAAO,SAAS,eAAe,iBAAiB,QAChD,cAAc,KAAK,KAClB,MAAM,aAAa;AAEzB;AAGA,SAAS,cAAc,OAAoC;AACzD,SACG,OAAO,WAAW,eAAe,iBAAiB,UACnD,iBAAiB;AAErB;AAMO,SAAS,gBAAgB,MAAe,OAAyB;AACtE,QAAM,+BAAe,IAAA;AACrB,QAAM,OAAO,CAAC,UACZ,QAAQ,QAAQ,KAAK,EAAE;AAAA,IAAO,CAAC,QAC7B,OAAO,UAAU,qBAAqB,KAAK,OAAO,GAAG;AAAA,EAAA;AAEzD,WAAS,MAAM,GAAY,GAAqB;AAC9C,QAAI,MAAM,KAAM,OAAO,MAAM,CAAC,KAAK,OAAO,MAAM,CAAC,EAAI,QAAO;AAC5D,QACE,MAAM,QACN,MAAM,QACN,OAAO,MAAM,YACb,OAAO,MAAM,YACb,wBAAwB,CAAC,KACzB,wBAAwB,CAAC;AAEzB,aAAO;AAET,QAAI,MAAM,QAAQ,CAAC,KAAK,MAAM,QAAQ,CAAC,KAAK,EAAE,WAAW,EAAE;AACzD,aAAO;AAKT,UAAM,QAAQ,SAAS,IAAI,CAAC;AAC5B,QAAI,OAAO,IAAI,CAAC,EAAG,QAAO;AAC1B,UAAM,CAAC,SAAS,SAAS,KAAK,IAAI,YAAY,CAAC;AAC/C,UAAM,CAAC,SAAS,SAAS,KAAK,IAAI,YAAY,CAAC;AAC/C,QACE,YAAY,WACZ,CAAC,QAAQ,MAAM,CAAC,OAAO,UAAU,UAAU,QAAQ,KAAK,CAAC;AAEzD,aAAO;AACT,QAAI,CAAC,SAAS,CAAC,MAAO,QAAO;AAC7B,QAAI,MAAO,OAAM,IAAI,CAAC;AAAA,QACjB,UAAS,IAAI,GAAG,oBAAI,IAAI,CAAC,CAAC,CAAC,CAAC;AACjC,UAAM,QAAQ,KAAK,KAAK;AACxB,WACE,MAAM,WAAW,KAAK,KAAK,EAAE,UAC7B,MAAM;AAAA,MACJ,CAAC,QACC,OAAO,UAAU,qBAAqB,KAAK,OAAO,GAAG,KACrD,MAAM,MAAM,GAAmB,GAAG,MAAM,GAAmB,CAAC;AAAA,IAAA;AAAA,EAGpE;AACA,SAAO,MAAM,MAAM,KAAK;AAC1B;AAEA,IAAI,YAAY;AAChB,SAAS,oBAAoB,IAAoB;AAC/C,MAAI,YAAY,UAAU,IAAI,EAAE;AAChC,MAAI,cAAc,QAAW;AAC3B,gBAAY,YAAY;AACxB;AACA,cAAU,IAAI,IAAI,SAAS;AAC3B,oBAAgB,IAAI,EAAE;AAAA,EACxB;AACA,SAAO;AACT;AAKA,SAAS,QAAQ,GAAW,GAAmB;AAC7C,SAAO,EAAE,cAAc,CAAC;AAC1B;"}

package/dist/esm/hashing/temporal.d.ts

@@ -0,0 +1,5 @@
+export interface TemporalLike {
+    [Symbol.toStringTag]: string;
+    toString: () => string;
+}
+export declare function isTemporal(input: object): input is TemporalLike;

package/dist/esm/hashing/temporal.js

@@ -0,0 +1,18 @@
+const temporalTypes = /* @__PURE__ */ new Set([
+  `Temporal.Duration`,
+  `Temporal.Instant`,
+  `Temporal.PlainDate`,
+  `Temporal.PlainDateTime`,
+  `Temporal.PlainMonthDay`,
+  `Temporal.PlainTime`,
+  `Temporal.PlainYearMonth`,
+  `Temporal.ZonedDateTime`
+]);
+function isTemporal(input) {
+  const tag = input[Symbol.toStringTag];
+  return typeof tag === `string` && temporalTypes.has(tag) && Object.getPrototypeOf(input)?.[Symbol.toStringTag] === tag;
+}
+export {
+  isTemporal
+};
+//# sourceMappingURL=temporal.js.map

package/dist/esm/hashing/temporal.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"temporal.js","sources":["../../../src/hashing/temporal.ts"],"sourcesContent":["const temporalTypes = new Set([\n  `Temporal.Duration`,\n  `Temporal.Instant`,\n  `Temporal.PlainDate`,\n  `Temporal.PlainDateTime`,\n  `Temporal.PlainMonthDay`,\n  `Temporal.PlainTime`,\n  `Temporal.PlainYearMonth`,\n  `Temporal.ZonedDateTime`,\n])\n\nexport interface TemporalLike {\n  [Symbol.toStringTag]: string\n  toString: () => string\n}\n\nexport function isTemporal(input: object): input is TemporalLike {\n  const tag = (input as Record<symbol, unknown>)[Symbol.toStringTag]\n  // An own tag on a plain object does not make it a Temporal value.\n  return (\n    typeof tag === `string` &&\n    temporalTypes.has(tag) &&\n    Object.getPrototypeOf(input)?.[Symbol.toStringTag] === tag\n  )\n}\n"],"names":[],"mappings":"AAAA,MAAM,oCAAoB,IAAI;AAAA,EAC5B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF,CAAC;AAOM,SAAS,WAAW,OAAsC;AAC/D,QAAM,MAAO,MAAkC,OAAO,WAAW;AAEjE,SACE,OAAO,QAAQ,YACf,cAAc,IAAI,GAAG,KACrB,OAAO,eAAe,KAAK,IAAI,OAAO,WAAW,MAAM;AAE3D;"}

package/dist/esm/multiset.d.ts

@@ -29,6 +29,22 @@ export declare class MultiSet<T> {
      * Produce as output a collection that is logically equivalent to the input
      * but which combines identical instances of the same record into one
      * (record, multiplicity) pair.
+     *
+     * Record identity depends on the shape of the whole multiset:
+     *
+     * - Keyed: every record is a `[key, value]` pair with a string or number
+     *   key. Keys and primitive values compare by value, and object values by
+     *   reference. A value that is an array of length 2 is a join tuple, and
+     *   its two elements compare by the same rule. The keyed identity text
+     *   has known collisions: https://github.com/TanStack/db/issues/1948
+     * - Unkeyed, one primitive type: every record is a string, or every record
+     *   is a number. Records compare as `Map` keys do, and a `-0` record is
+     *   returned as `0`.
+     * - Otherwise, records compare by structure (`hash`).
+     *
+     * Each identity keeps its first record in input order with the summed
+     * multiplicity. Identities whose sum is zero are dropped. The input is not
+     * changed.
      */
     consolidate(): MultiSet<T>;
     extend(other: MultiSet<T> | MultiSetArray<T>): void;

package/dist/esm/multiset.js

@@ -1,4 +1,4 @@
-import { chunkedArrayPush, globalObjectIdGenerator, DefaultMap } from "./utils.js";
+import { chunkedArrayPush, getStringId } from "./utils.js";
 import { hash } from "./hashing/hash.js";
 class MultiSet {
   #inner;
@@ -49,102 +49,42 @@ class MultiSet {
    * Produce as output a collection that is logically equivalent to the input
    * but which combines identical instances of the same record into one
    * (record, multiplicity) pair.
-   */
-  consolidate() {
-    if (this.#inner.length > 0) {
-      const firstItem = this.#inner[0]?.[0];
-      if (Array.isArray(firstItem) && firstItem.length === 2) {
-        return this.#consolidateKeyed();
-      }
-    }
-    return this.#consolidateUnkeyed();
-  }
-  /**
-   * Private method for consolidating keyed multisets where keys are strings/numbers
-   * and values are compared by reference equality.
    *
-   * This method provides significant performance improvements over the hash-based approach
-   * by using WeakMap for object reference tracking and avoiding expensive serialization.
+   * Record identity depends on the shape of the whole multiset:
    *
-   * Special handling for join operations: When values are tuples of length 2 (common in joins),
-   * we unpack them and compare each element individually to maintain proper equality semantics.
-   */
-  #consolidateKeyed() {
-    const consolidated = /* @__PURE__ */ new Map();
-    const values = /* @__PURE__ */ new Map();
-    const getTupleId = (tuple) => {
-      if (tuple.length !== 2) {
-        throw new Error(`Expected tuple of length 2`);
-      }
-      const [first, second] = tuple;
-      return `${globalObjectIdGenerator.getStringId(first)}|${globalObjectIdGenerator.getStringId(second)}`;
-    };
-    for (const [data, multiplicity] of this.#inner) {
-      if (!Array.isArray(data) || data.length !== 2) {
-        return this.#consolidateUnkeyed();
-      }
-      const [key, value] = data;
-      if (typeof key !== `string` && typeof key !== `number`) {
-        return this.#consolidateUnkeyed();
-      }
-      let valueId;
-      if (Array.isArray(value) && value.length === 2) {
-        valueId = getTupleId(value);
-      } else {
-        valueId = globalObjectIdGenerator.getStringId(value);
-      }
-      const compositeKey = key + `|` + valueId;
-      consolidated.set(
-        compositeKey,
-        (consolidated.get(compositeKey) || 0) + multiplicity
-      );
-      if (!values.has(compositeKey)) {
-        values.set(compositeKey, data);
-      }
-    }
-    const result = [];
-    for (const [compositeKey, multiplicity] of consolidated) {
-      if (multiplicity !== 0) {
-        result.push([values.get(compositeKey), multiplicity]);
-      }
-    }
-    return new MultiSet(result);
-  }
-  /**
-   * Private method for consolidating unkeyed multisets using the original approach.
+   * - Keyed: every record is a `[key, value]` pair with a string or number
+   *   key. Keys and primitive values compare by value, and object values by
+   *   reference. A value that is an array of length 2 is a join tuple, and
+   *   its two elements compare by the same rule. The keyed identity text
+   *   has known collisions: https://github.com/TanStack/db/issues/1948
+   * - Unkeyed, one primitive type: every record is a string, or every record
+   *   is a number. Records compare as `Map` keys do, and a `-0` record is
+   *   returned as `0`.
+   * - Otherwise, records compare by structure (`hash`).
+   *
+   * Each identity keeps its first record in input order with the summed
+   * multiplicity. Identities whose sum is zero are dropped. The input is not
+   * changed.
    */
-  #consolidateUnkeyed() {
-    const consolidated = new DefaultMap(() => 0);
-    const values = /* @__PURE__ */ new Map();
-    let hasString = false;
-    let hasNumber = false;
-    let hasOther = false;
-    for (const [data, _] of this.#inner) {
-      if (typeof data === `string`) {
-        hasString = true;
-      } else if (typeof data === `number`) {
-        hasNumber = true;
-      } else {
-        hasOther = true;
-        break;
-      }
-    }
-    const requireJson = hasOther || hasString && hasNumber;
-    for (const [data, multiplicity] of this.#inner) {
-      const key = requireJson ? hash(data) : data;
-      if (requireJson && !values.has(key)) {
-        values.set(key, data);
-      }
-      consolidated.update(key, (count) => count + multiplicity);
-    }
-    const result = [];
-    for (const [key, multiplicity] of consolidated.entries()) {
-      if (multiplicity !== 0) {
-        const parsedKey = requireJson ? values.get(key) : key;
-        result.push([parsedKey, multiplicity]);
-      }
+  consolidate() {
+    const inner = this.#inner;
+    if (inner.every(
+      ([data]) => Array.isArray(data) && data.length === 2 && (typeof data[0] === `string` || typeof data[0] === `number`)
+    )) {
+      return consolidateBy(inner, (data) => {
+        const [key, value] = data;
+        const valueId = Array.isArray(value) && value.length === 2 ? `${getStringId(value[0])}|${getStringId(value[1])}` : getStringId(value);
+        return key + `|` + valueId;
+      });
     }
-    return new MultiSet(result);
+    let primitiveType;
+    const requireHash = inner.some(([data]) => {
+      const type = typeof data;
+      if (type !== `string` && type !== `number`) return true;
+      primitiveType ??= type;
+      return type !== primitiveType;
+    });
+    return requireHash ? consolidateBy(inner, hash) : consolidateBy(inner, (data) => data, true);
   }
   extend(other) {
     const otherArray = other instanceof MultiSet ? other.getInner() : other;
@@ -159,6 +99,21 @@ class MultiSet {
     return this.#inner;
   }
 }
+function consolidateBy(inner, identityOf, keepIdentity = false) {
+  const consolidated = /* @__PURE__ */ new Map();
+  for (const [data, multiplicity] of inner) {
+    const identity = identityOf(data);
+    const entry = consolidated.get(identity);
+    if (entry) entry[1] += multiplicity;
+    else consolidated.set(identity, [data, multiplicity]);
+  }
+  const result = [];
+  for (const [identity, [data, multiplicity]] of consolidated) {
+    if (multiplicity !== 0)
+      result.push([keepIdentity ? identity : data, multiplicity]);
+  }
+  return new MultiSet(result);
+}
 export {
   MultiSet
 };

package/dist/esm/multiset.js.map

@@ -1 +1 @@
-{"version":3,"file":"multiset.js","sources":["../../src/multiset.ts"],"sourcesContent":["import {\n  DefaultMap,\n  chunkedArrayPush,\n  globalObjectIdGenerator,\n} from './utils.js'\nimport { hash } from './hashing/index.js'\n\nexport type MultiSetArray<T> = Array<[T, number]>\nexport type KeyedData<T> = [key: string, value: T]\n\n/**\n * A multiset of data.\n */\nexport class MultiSet<T> {\n  #inner: MultiSetArray<T>\n\n  constructor(data: MultiSetArray<T> = []) {\n    this.#inner = data\n  }\n\n  toString(indent = false): string {\n    return `MultiSet(${JSON.stringify(this.#inner, null, indent ? 2 : undefined)})`\n  }\n\n  toJSON(): string {\n    return JSON.stringify(Array.from(this.getInner()))\n  }\n\n  static fromJSON<U>(json: string): MultiSet<U> {\n    return new MultiSet(JSON.parse(json))\n  }\n\n  /**\n   * Apply a function to all records in the collection.\n   */\n  map<U>(f: (data: T) => U): MultiSet<U> {\n    return new MultiSet(\n      this.#inner.map(([data, multiplicity]) => [f(data), multiplicity]),\n    )\n  }\n\n  /**\n   * Filter out records for which a function f(record) evaluates to False.\n   */\n  filter(f: (data: T) => boolean): MultiSet<T> {\n    return new MultiSet(this.#inner.filter(([data, _]) => f(data)))\n  }\n\n  /**\n   * Negate all multiplicities in the collection.\n   */\n  negate(): MultiSet<T> {\n    return new MultiSet(\n      this.#inner.map(([data, multiplicity]) => [data, -multiplicity]),\n    )\n  }\n\n  /**\n   * Concatenate two collections together.\n   */\n  concat(other: MultiSet<T>): MultiSet<T> {\n    const out: MultiSetArray<T> = []\n    chunkedArrayPush(out, this.#inner)\n    chunkedArrayPush(out, other.getInner())\n    return new MultiSet(out)\n  }\n\n  /**\n   * Produce as output a collection that is logically equivalent to the input\n   * but which combines identical instances of the same record into one\n   * (record, multiplicity) pair.\n   */\n  consolidate(): MultiSet<T> {\n    // Check if this looks like a keyed multiset (first item is a tuple of length 2)\n    if (this.#inner.length > 0) {\n      const firstItem = this.#inner[0]?.[0]\n      if (Array.isArray(firstItem) && firstItem.length === 2) {\n        return this.#consolidateKeyed()\n      }\n    }\n\n    // Fall back to original method for unkeyed data\n    return this.#consolidateUnkeyed()\n  }\n\n  /**\n   * Private method for consolidating keyed multisets where keys are strings/numbers\n   * and values are compared by reference equality.\n   *\n   * This method provides significant performance improvements over the hash-based approach\n   * by using WeakMap for object reference tracking and avoiding expensive serialization.\n   *\n   * Special handling for join operations: When values are tuples of length 2 (common in joins),\n   * we unpack them and compare each element individually to maintain proper equality semantics.\n   */\n  #consolidateKeyed(): MultiSet<T> {\n    const consolidated = new Map<string, number>()\n    const values = new Map<string, T>()\n\n    // Use global object ID generator for consistent reference equality\n\n    /**\n     * Special handler for tuples (arrays of length 2) commonly produced by join operations.\n     * Unpacks the tuple and generates an ID based on both elements to ensure proper\n     * consolidation of join results like ['A', null] and [null, 'X'].\n     */\n    const getTupleId = (tuple: Array<any>): string => {\n      if (tuple.length !== 2) {\n        throw new Error(`Expected tuple of length 2`)\n      }\n      const [first, second] = tuple\n      return `${globalObjectIdGenerator.getStringId(first)}|${globalObjectIdGenerator.getStringId(second)}`\n    }\n\n    // Process each item in the multiset\n    for (const [data, multiplicity] of this.#inner) {\n      // Verify this is still a keyed item (should be [key, value] pair)\n      if (!Array.isArray(data) || data.length !== 2) {\n        // Found non-keyed item, fall back to unkeyed consolidation\n        return this.#consolidateUnkeyed()\n      }\n\n      const [key, value] = data\n\n      // Verify key is string or number as expected for keyed multisets\n      if (typeof key !== `string` && typeof key !== `number`) {\n        // Found non-string/number key, fall back to unkeyed consolidation\n        return this.#consolidateUnkeyed()\n      }\n\n      // Generate value ID with special handling for join tuples\n      let valueId: string\n      if (Array.isArray(value) && value.length === 2) {\n        // Special case: value is a tuple from join operations\n        valueId = getTupleId(value)\n      } else {\n        // Regular case: use reference/value equality\n        valueId = globalObjectIdGenerator.getStringId(value)\n      }\n\n      // Create composite key and consolidate\n      const compositeKey = key + `|` + valueId\n      consolidated.set(\n        compositeKey,\n        (consolidated.get(compositeKey) || 0) + multiplicity,\n      )\n\n      // Store the original data for the first occurrence\n      if (!values.has(compositeKey)) {\n        values.set(compositeKey, data as T)\n      }\n    }\n\n    // Build result array, filtering out zero multiplicities\n    const result: MultiSetArray<T> = []\n    for (const [compositeKey, multiplicity] of consolidated) {\n      if (multiplicity !== 0) {\n        result.push([values.get(compositeKey)!, multiplicity])\n      }\n    }\n\n    return new MultiSet(result)\n  }\n\n  /**\n   * Private method for consolidating unkeyed multisets using the original approach.\n   */\n  #consolidateUnkeyed(): MultiSet<T> {\n    const consolidated = new DefaultMap<string | number, number>(() => 0)\n    const values = new Map<string, any>()\n\n    let hasString = false\n    let hasNumber = false\n    let hasOther = false\n    for (const [data, _] of this.#inner) {\n      if (typeof data === `string`) {\n        hasString = true\n      } else if (typeof data === `number`) {\n        hasNumber = true\n      } else {\n        hasOther = true\n        break\n      }\n    }\n\n    const requireJson = hasOther || (hasString && hasNumber)\n\n    for (const [data, multiplicity] of this.#inner) {\n      const key = requireJson ? hash(data) : (data as string | number)\n      if (requireJson && !values.has(key as string)) {\n        values.set(key as string, data)\n      }\n      consolidated.update(key, (count) => count + multiplicity)\n    }\n\n    const result: MultiSetArray<T> = []\n    for (const [key, multiplicity] of consolidated.entries()) {\n      if (multiplicity !== 0) {\n        const parsedKey = requireJson ? values.get(key as string) : key\n        result.push([parsedKey as T, multiplicity])\n      }\n    }\n\n    return new MultiSet(result)\n  }\n\n  extend(other: MultiSet<T> | MultiSetArray<T>): void {\n    const otherArray = other instanceof MultiSet ? other.getInner() : other\n    chunkedArrayPush(this.#inner, otherArray)\n  }\n\n  add(item: T, multiplicity: number): void {\n    if (multiplicity !== 0) {\n      this.#inner.push([item, multiplicity])\n    }\n  }\n\n  getInner(): MultiSetArray<T> {\n    return this.#inner\n  }\n}\n"],"names":[],"mappings":";;AAaO,MAAM,SAAY;AAAA,EACvB;AAAA,EAEA,YAAY,OAAyB,IAAI;AACvC,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,SAAS,SAAS,OAAe;AAC/B,WAAO,YAAY,KAAK,UAAU,KAAK,QAAQ,MAAM,SAAS,IAAI,MAAS,CAAC;AAAA,EAC9E;AAAA,EAEA,SAAiB;AACf,WAAO,KAAK,UAAU,MAAM,KAAK,KAAK,SAAA,CAAU,CAAC;AAAA,EACnD;AAAA,EAEA,OAAO,SAAY,MAA2B;AAC5C,WAAO,IAAI,SAAS,KAAK,MAAM,IAAI,CAAC;AAAA,EACtC;AAAA;AAAA;AAAA;AAAA,EAKA,IAAO,GAAgC;AACrC,WAAO,IAAI;AAAA,MACT,KAAK,OAAO,IAAI,CAAC,CAAC,MAAM,YAAY,MAAM,CAAC,EAAE,IAAI,GAAG,YAAY,CAAC;AAAA,IAAA;AAAA,EAErE;AAAA;AAAA;AAAA;AAAA,EAKA,OAAO,GAAsC;AAC3C,WAAO,IAAI,SAAS,KAAK,OAAO,OAAO,CAAC,CAAC,MAAM,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;AAAA,EAChE;AAAA;AAAA;AAAA;AAAA,EAKA,SAAsB;AACpB,WAAO,IAAI;AAAA,MACT,KAAK,OAAO,IAAI,CAAC,CAAC,MAAM,YAAY,MAAM,CAAC,MAAM,CAAC,YAAY,CAAC;AAAA,IAAA;AAAA,EAEnE;AAAA;AAAA;AAAA;AAAA,EAKA,OAAO,OAAiC;AACtC,UAAM,MAAwB,CAAA;AAC9B,qBAAiB,KAAK,KAAK,MAAM;AACjC,qBAAiB,KAAK,MAAM,UAAU;AACtC,WAAO,IAAI,SAAS,GAAG;AAAA,EACzB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,cAA2B;AAEzB,QAAI,KAAK,OAAO,SAAS,GAAG;AAC1B,YAAM,YAAY,KAAK,OAAO,CAAC,IAAI,CAAC;AACpC,UAAI,MAAM,QAAQ,SAAS,KAAK,UAAU,WAAW,GAAG;AACtD,eAAO,KAAK,kBAAA;AAAA,MACd;AAAA,IACF;AAGA,WAAO,KAAK,oBAAA;AAAA,EACd;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAYA,oBAAiC;AAC/B,UAAM,mCAAmB,IAAA;AACzB,UAAM,6BAAa,IAAA;AASnB,UAAM,aAAa,CAAC,UAA8B;AAChD,UAAI,MAAM,WAAW,GAAG;AACtB,cAAM,IAAI,MAAM,4BAA4B;AAAA,MAC9C;AACA,YAAM,CAAC,OAAO,MAAM,IAAI;AACxB,aAAO,GAAG,wBAAwB,YAAY,KAAK,CAAC,IAAI,wBAAwB,YAAY,MAAM,CAAC;AAAA,IACrG;AAGA,eAAW,CAAC,MAAM,YAAY,KAAK,KAAK,QAAQ;AAE9C,UAAI,CAAC,MAAM,QAAQ,IAAI,KAAK,KAAK,WAAW,GAAG;AAE7C,eAAO,KAAK,oBAAA;AAAA,MACd;AAEA,YAAM,CAAC,KAAK,KAAK,IAAI;AAGrB,UAAI,OAAO,QAAQ,YAAY,OAAO,QAAQ,UAAU;AAEtD,eAAO,KAAK,oBAAA;AAAA,MACd;AAGA,UAAI;AACJ,UAAI,MAAM,QAAQ,KAAK,KAAK,MAAM,WAAW,GAAG;AAE9C,kBAAU,WAAW,KAAK;AAAA,MAC5B,OAAO;AAEL,kBAAU,wBAAwB,YAAY,KAAK;AAAA,MACrD;AAGA,YAAM,eAAe,MAAM,MAAM;AACjC,mBAAa;AAAA,QACX;AAAA,SACC,aAAa,IAAI,YAAY,KAAK,KAAK;AAAA,MAAA;AAI1C,UAAI,CAAC,OAAO,IAAI,YAAY,GAAG;AAC7B,eAAO,IAAI,cAAc,IAAS;AAAA,MACpC;AAAA,IACF;AAGA,UAAM,SAA2B,CAAA;AACjC,eAAW,CAAC,cAAc,YAAY,KAAK,cAAc;AACvD,UAAI,iBAAiB,GAAG;AACtB,eAAO,KAAK,CAAC,OAAO,IAAI,YAAY,GAAI,YAAY,CAAC;AAAA,MACvD;AAAA,IACF;AAEA,WAAO,IAAI,SAAS,MAAM;AAAA,EAC5B;AAAA;AAAA;AAAA;AAAA,EAKA,sBAAmC;AACjC,UAAM,eAAe,IAAI,WAAoC,MAAM,CAAC;AACpE,UAAM,6BAAa,IAAA;AAEnB,QAAI,YAAY;AAChB,QAAI,YAAY;AAChB,QAAI,WAAW;AACf,eAAW,CAAC,MAAM,CAAC,KAAK,KAAK,QAAQ;AACnC,UAAI,OAAO,SAAS,UAAU;AAC5B,oBAAY;AAAA,MACd,WAAW,OAAO,SAAS,UAAU;AACnC,oBAAY;AAAA,MACd,OAAO;AACL,mBAAW;AACX;AAAA,MACF;AAAA,IACF;AAEA,UAAM,cAAc,YAAa,aAAa;AAE9C,eAAW,CAAC,MAAM,YAAY,KAAK,KAAK,QAAQ;AAC9C,YAAM,MAAM,cAAc,KAAK,IAAI,IAAK;AACxC,UAAI,eAAe,CAAC,OAAO,IAAI,GAAa,GAAG;AAC7C,eAAO,IAAI,KAAe,IAAI;AAAA,MAChC;AACA,mBAAa,OAAO,KAAK,CAAC,UAAU,QAAQ,YAAY;AAAA,IAC1D;AAEA,UAAM,SAA2B,CAAA;AACjC,eAAW,CAAC,KAAK,YAAY,KAAK,aAAa,WAAW;AACxD,UAAI,iBAAiB,GAAG;AACtB,cAAM,YAAY,cAAc,OAAO,IAAI,GAAa,IAAI;AAC5D,eAAO,KAAK,CAAC,WAAgB,YAAY,CAAC;AAAA,MAC5C;AAAA,IACF;AAEA,WAAO,IAAI,SAAS,MAAM;AAAA,EAC5B;AAAA,EAEA,OAAO,OAA6C;AAClD,UAAM,aAAa,iBAAiB,WAAW,MAAM,aAAa;AAClE,qBAAiB,KAAK,QAAQ,UAAU;AAAA,EAC1C;AAAA,EAEA,IAAI,MAAS,cAA4B;AACvC,QAAI,iBAAiB,GAAG;AACtB,WAAK,OAAO,KAAK,CAAC,MAAM,YAAY,CAAC;AAAA,IACvC;AAAA,EACF;AAAA,EAEA,WAA6B;AAC3B,WAAO,KAAK;AAAA,EACd;AACF;"}
+{"version":3,"file":"multiset.js","sources":["../../src/multiset.ts"],"sourcesContent":["import { chunkedArrayPush, getStringId } from './utils.js'\nimport { hash } from './hashing/index.js'\n\nexport type MultiSetArray<T> = Array<[T, number]>\nexport type KeyedData<T> = [key: string, value: T]\n\n/**\n * A multiset of data.\n */\nexport class MultiSet<T> {\n  #inner: MultiSetArray<T>\n\n  constructor(data: MultiSetArray<T> = []) {\n    this.#inner = data\n  }\n\n  toString(indent = false): string {\n    return `MultiSet(${JSON.stringify(this.#inner, null, indent ? 2 : undefined)})`\n  }\n\n  toJSON(): string {\n    return JSON.stringify(Array.from(this.getInner()))\n  }\n\n  static fromJSON<U>(json: string): MultiSet<U> {\n    return new MultiSet(JSON.parse(json))\n  }\n\n  /**\n   * Apply a function to all records in the collection.\n   */\n  map<U>(f: (data: T) => U): MultiSet<U> {\n    return new MultiSet(\n      this.#inner.map(([data, multiplicity]) => [f(data), multiplicity]),\n    )\n  }\n\n  /**\n   * Filter out records for which a function f(record) evaluates to False.\n   */\n  filter(f: (data: T) => boolean): MultiSet<T> {\n    return new MultiSet(this.#inner.filter(([data, _]) => f(data)))\n  }\n\n  /**\n   * Negate all multiplicities in the collection.\n   */\n  negate(): MultiSet<T> {\n    return new MultiSet(\n      this.#inner.map(([data, multiplicity]) => [data, -multiplicity]),\n    )\n  }\n\n  /**\n   * Concatenate two collections together.\n   */\n  concat(other: MultiSet<T>): MultiSet<T> {\n    const out: MultiSetArray<T> = []\n    chunkedArrayPush(out, this.#inner)\n    chunkedArrayPush(out, other.getInner())\n    return new MultiSet(out)\n  }\n\n  /**\n   * Produce as output a collection that is logically equivalent to the input\n   * but which combines identical instances of the same record into one\n   * (record, multiplicity) pair.\n   *\n   * Record identity depends on the shape of the whole multiset:\n   *\n   * - Keyed: every record is a `[key, value]` pair with a string or number\n   *   key. Keys and primitive values compare by value, and object values by\n   *   reference. A value that is an array of length 2 is a join tuple, and\n   *   its two elements compare by the same rule. The keyed identity text\n   *   has known collisions: https://github.com/TanStack/db/issues/1948\n   * - Unkeyed, one primitive type: every record is a string, or every record\n   *   is a number. Records compare as `Map` keys do, and a `-0` record is\n   *   returned as `0`.\n   * - Otherwise, records compare by structure (`hash`).\n   *\n   * Each identity keeps its first record in input order with the summed\n   * multiplicity. Identities whose sum is zero are dropped. The input is not\n   * changed.\n   */\n  consolidate(): MultiSet<T> {\n    const inner = this.#inner\n    // Keyed multisets hold [string | number key, value] pairs. Their values are\n    // compared by reference, which avoids hashing. Join tuples (values of\n    // length 2) are unpacked so ['A', null] and [null, 'X'] stay distinct.\n    if (\n      inner.every(\n        ([data]) =>\n          Array.isArray(data) &&\n          data.length === 2 &&\n          (typeof data[0] === `string` || typeof data[0] === `number`),\n      )\n    ) {\n      return consolidateBy(inner, (data) => {\n        const [key, value] = data as [string | number, unknown]\n        const valueId =\n          Array.isArray(value) && value.length === 2\n            ? `${getStringId(value[0])}|${getStringId(value[1])}`\n            : getStringId(value)\n        return key + `|` + valueId\n      })\n    }\n\n    // Unkeyed data of a single primitive type is its own identity, and the\n    // result holds the Map key, so -0 becomes 0. Anything else is compared\n    // structurally by hash.\n    let primitiveType: string | undefined\n    const requireHash = inner.some(([data]) => {\n      const type = typeof data\n      if (type !== `string` && type !== `number`) return true\n      primitiveType ??= type\n      return type !== primitiveType\n    })\n    return requireHash\n      ? consolidateBy(inner, hash)\n      : consolidateBy(inner, (data) => data, true)\n  }\n\n  extend(other: MultiSet<T> | MultiSetArray<T>): void {\n    const otherArray = other instanceof MultiSet ? other.getInner() : other\n    chunkedArrayPush(this.#inner, otherArray)\n  }\n\n  add(item: T, multiplicity: number): void {\n    if (multiplicity !== 0) {\n      this.#inner.push([item, multiplicity])\n    }\n  }\n\n  getInner(): MultiSetArray<T> {\n    return this.#inner\n  }\n}\n\n/**\n * Sums multiplicities of records with the same identity, keeping the first\n * record seen for each identity (or its Map key, which writes -0 as 0, when\n * `keepIdentity` is set) and dropping records whose sum is zero.\n */\nfunction consolidateBy<T>(\n  inner: MultiSetArray<T>,\n  identityOf: (data: T) => unknown,\n  keepIdentity = false,\n): MultiSet<T> {\n  const consolidated = new Map<unknown, [T, number]>()\n  for (const [data, multiplicity] of inner) {\n    const identity = identityOf(data)\n    const entry = consolidated.get(identity)\n    if (entry) entry[1] += multiplicity\n    else consolidated.set(identity, [data, multiplicity])\n  }\n  const result: MultiSetArray<T> = []\n  for (const [identity, [data, multiplicity]] of consolidated) {\n    if (multiplicity !== 0)\n      result.push([keepIdentity ? (identity as T) : data, multiplicity])\n  }\n  return new MultiSet(result)\n}\n"],"names":[],"mappings":";;AASO,MAAM,SAAY;AAAA,EACvB;AAAA,EAEA,YAAY,OAAyB,IAAI;AACvC,SAAK,SAAS;AAAA,EAChB;AAAA,EAEA,SAAS,SAAS,OAAe;AAC/B,WAAO,YAAY,KAAK,UAAU,KAAK,QAAQ,MAAM,SAAS,IAAI,MAAS,CAAC;AAAA,EAC9E;AAAA,EAEA,SAAiB;AACf,WAAO,KAAK,UAAU,MAAM,KAAK,KAAK,SAAA,CAAU,CAAC;AAAA,EACnD;AAAA,EAEA,OAAO,SAAY,MAA2B;AAC5C,WAAO,IAAI,SAAS,KAAK,MAAM,IAAI,CAAC;AAAA,EACtC;AAAA;AAAA;AAAA;AAAA,EAKA,IAAO,GAAgC;AACrC,WAAO,IAAI;AAAA,MACT,KAAK,OAAO,IAAI,CAAC,CAAC,MAAM,YAAY,MAAM,CAAC,EAAE,IAAI,GAAG,YAAY,CAAC;AAAA,IAAA;AAAA,EAErE;AAAA;AAAA;AAAA;AAAA,EAKA,OAAO,GAAsC;AAC3C,WAAO,IAAI,SAAS,KAAK,OAAO,OAAO,CAAC,CAAC,MAAM,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;AAAA,EAChE;AAAA;AAAA;AAAA;AAAA,EAKA,SAAsB;AACpB,WAAO,IAAI;AAAA,MACT,KAAK,OAAO,IAAI,CAAC,CAAC,MAAM,YAAY,MAAM,CAAC,MAAM,CAAC,YAAY,CAAC;AAAA,IAAA;AAAA,EAEnE;AAAA;AAAA;AAAA;AAAA,EAKA,OAAO,OAAiC;AACtC,UAAM,MAAwB,CAAA;AAC9B,qBAAiB,KAAK,KAAK,MAAM;AACjC,qBAAiB,KAAK,MAAM,UAAU;AACtC,WAAO,IAAI,SAAS,GAAG;AAAA,EACzB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAuBA,cAA2B;AACzB,UAAM,QAAQ,KAAK;AAInB,QACE,MAAM;AAAA,MACJ,CAAC,CAAC,IAAI,MACJ,MAAM,QAAQ,IAAI,KAClB,KAAK,WAAW,MACf,OAAO,KAAK,CAAC,MAAM,YAAY,OAAO,KAAK,CAAC,MAAM;AAAA,IAAA,GAEvD;AACA,aAAO,cAAc,OAAO,CAAC,SAAS;AACpC,cAAM,CAAC,KAAK,KAAK,IAAI;AACrB,cAAM,UACJ,MAAM,QAAQ,KAAK,KAAK,MAAM,WAAW,IACrC,GAAG,YAAY,MAAM,CAAC,CAAC,CAAC,IAAI,YAAY,MAAM,CAAC,CAAC,CAAC,KACjD,YAAY,KAAK;AACvB,eAAO,MAAM,MAAM;AAAA,MACrB,CAAC;AAAA,IACH;AAKA,QAAI;AACJ,UAAM,cAAc,MAAM,KAAK,CAAC,CAAC,IAAI,MAAM;AACzC,YAAM,OAAO,OAAO;AACpB,UAAI,SAAS,YAAY,SAAS,SAAU,QAAO;AACnD,wBAAkB;AAClB,aAAO,SAAS;AAAA,IAClB,CAAC;AACD,WAAO,cACH,cAAc,OAAO,IAAI,IACzB,cAAc,OAAO,CAAC,SAAS,MAAM,IAAI;AAAA,EAC/C;AAAA,EAEA,OAAO,OAA6C;AAClD,UAAM,aAAa,iBAAiB,WAAW,MAAM,aAAa;AAClE,qBAAiB,KAAK,QAAQ,UAAU;AAAA,EAC1C;AAAA,EAEA,IAAI,MAAS,cAA4B;AACvC,QAAI,iBAAiB,GAAG;AACtB,WAAK,OAAO,KAAK,CAAC,MAAM,YAAY,CAAC;AAAA,IACvC;AAAA,EACF;AAAA,EAEA,WAA6B;AAC3B,WAAO,KAAK;AAAA,EACd;AACF;AAOA,SAAS,cACP,OACA,YACA,eAAe,OACF;AACb,QAAM,mCAAmB,IAAA;AACzB,aAAW,CAAC,MAAM,YAAY,KAAK,OAAO;AACxC,UAAM,WAAW,WAAW,IAAI;AAChC,UAAM,QAAQ,aAAa,IAAI,QAAQ;AACvC,QAAI,MAAO,OAAM,CAAC,KAAK;AAAA,sBACL,IAAI,UAAU,CAAC,MAAM,YAAY,CAAC;AAAA,EACtD;AACA,QAAM,SAA2B,CAAA;AACjC,aAAW,CAAC,UAAU,CAAC,MAAM,YAAY,CAAC,KAAK,cAAc;AAC3D,QAAI,iBAAiB;AACnB,aAAO,KAAK,CAAC,eAAgB,WAAiB,MAAM,YAAY,CAAC;AAAA,EACrE;AACA,SAAO,IAAI,SAAS,MAAM;AAC5B;"}

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

@@ -12,5 +12,5 @@ export declare class DistinctOperator<T extends KeyValue<any, any>> extends Unar
 /**
  * Removes duplicate values
  */
-export declare function distinct<T extends KeyValue<any, any>>(by?: (value: T) => any): (stream: IStreamBuilder<T>) => IStreamBuilder<T>;
+export declare function distinct<T extends KeyValue<any, any>>(by?: (value: T) => any): (stream: IStreamBuilder<T>) => IStreamBuilder<KeyValue<number, GetValue<T>>>;
 export {};

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

@@ -1 +1 @@
-{"version":3,"file":"distinct.js","sources":["../../../src/operators/distinct.ts"],"sourcesContent":["import { DifferenceStreamWriter, UnaryOperator } from '../graph.js'\nimport { StreamBuilder } from '../d2.js'\nimport { hash } from '../hashing/index.js'\nimport { MultiSet } from '../multiset.js'\nimport type { Hash } from '../hashing/index.js'\nimport type { DifferenceStreamReader } from '../graph.js'\nimport type { IStreamBuilder, KeyValue } from '../types.js'\n\ntype Multiplicity = number\n\ntype GetValue<T> = T extends KeyValue<any, infer V> ? V : never\n\n/**\n * Operator that removes duplicates\n */\nexport class DistinctOperator<\n  T extends KeyValue<any, any>,\n> extends UnaryOperator<T, KeyValue<number, GetValue<T>>> {\n  #by: (value: T) => any\n  #values: Map<Hash, Multiplicity> // keeps track of the number of times each value has been seen\n\n  constructor(\n    id: number,\n    input: DifferenceStreamReader<T>,\n    output: DifferenceStreamWriter<KeyValue<number, GetValue<T>>>,\n    by: (value: T) => any = (value: T) => value,\n  ) {\n    super(id, input, output)\n    this.#by = by\n    this.#values = new Map()\n  }\n\n  run(): void {\n    const updatedValues = new Map<Hash, [Multiplicity, T]>()\n\n    // Compute the new multiplicity for each value\n    for (const message of this.inputMessages()) {\n      for (const [value, diff] of message.getInner()) {\n        const hashedValue = hash(this.#by(value))\n\n        const oldMultiplicity =\n          updatedValues.get(hashedValue)?.[0] ??\n          this.#values.get(hashedValue) ??\n          0\n        const newMultiplicity = oldMultiplicity + diff\n        updatedValues.set(hashedValue, [newMultiplicity, value])\n      }\n    }\n\n    const result: Array<[KeyValue<number, GetValue<T>>, number]> = []\n\n    // Check which values became visible or disappeared\n    for (const [\n      hashedValue,\n      [newMultiplicity, value],\n    ] of updatedValues.entries()) {\n      const oldMultiplicity = this.#values.get(hashedValue) ?? 0\n\n      if (newMultiplicity === 0) {\n        this.#values.delete(hashedValue)\n      } else {\n        this.#values.set(hashedValue, newMultiplicity)\n      }\n\n      if (oldMultiplicity <= 0 && newMultiplicity > 0) {\n        // The value wasn't present in the stream\n        // but with this change it is now present in the stream\n        result.push([[hash(this.#by(value)), value[1]], 1])\n      } else if (oldMultiplicity > 0 && newMultiplicity <= 0) {\n        // The value was present in the stream\n        // but with this change it is no longer present in the stream\n        result.push([[hash(this.#by(value)), value[1]], -1])\n      }\n    }\n\n    if (result.length > 0) {\n      this.output.sendData(new MultiSet(result))\n    }\n  }\n}\n\n/**\n * Removes duplicate values\n */\nexport function distinct<T extends KeyValue<any, any>>(\n  by: (value: T) => any = (value: T) => value,\n) {\n  return (stream: IStreamBuilder<T>): IStreamBuilder<T> => {\n    const output = new StreamBuilder<T>(\n      stream.graph,\n      new DifferenceStreamWriter<T>(),\n    )\n    const operator = new DistinctOperator<T>(\n      stream.graph.getNextOperatorId(),\n      stream.connectReader(),\n      output.writer,\n      by,\n    )\n    stream.graph.addOperator(operator)\n    return output\n  }\n}\n"],"names":[],"mappings":";;;;AAeO,MAAM,yBAEH,cAAgD;AAAA,EACxD;AAAA,EACA;AAAA;AAAA,EAEA,YACE,IACA,OACA,QACA,KAAwB,CAAC,UAAa,OACtC;AACA,UAAM,IAAI,OAAO,MAAM;AACvB,SAAK,MAAM;AACX,SAAK,8BAAc,IAAA;AAAA,EACrB;AAAA,EAEA,MAAY;AACV,UAAM,oCAAoB,IAAA;AAG1B,eAAW,WAAW,KAAK,iBAAiB;AAC1C,iBAAW,CAAC,OAAO,IAAI,KAAK,QAAQ,YAAY;AAC9C,cAAM,cAAc,KAAK,KAAK,IAAI,KAAK,CAAC;AAExC,cAAM,kBACJ,cAAc,IAAI,WAAW,IAAI,CAAC,KAClC,KAAK,QAAQ,IAAI,WAAW,KAC5B;AACF,cAAM,kBAAkB,kBAAkB;AAC1C,sBAAc,IAAI,aAAa,CAAC,iBAAiB,KAAK,CAAC;AAAA,MACzD;AAAA,IACF;AAEA,UAAM,SAAyD,CAAA;AAG/D,eAAW;AAAA,MACT;AAAA,MACA,CAAC,iBAAiB,KAAK;AAAA,IAAA,KACpB,cAAc,WAAW;AAC5B,YAAM,kBAAkB,KAAK,QAAQ,IAAI,WAAW,KAAK;AAEzD,UAAI,oBAAoB,GAAG;AACzB,aAAK,QAAQ,OAAO,WAAW;AAAA,MACjC,OAAO;AACL,aAAK,QAAQ,IAAI,aAAa,eAAe;AAAA,MAC/C;AAEA,UAAI,mBAAmB,KAAK,kBAAkB,GAAG;AAG/C,eAAO,KAAK,CAAC,CAAC,KAAK,KAAK,IAAI,KAAK,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,MACpD,WAAW,kBAAkB,KAAK,mBAAmB,GAAG;AAGtD,eAAO,KAAK,CAAC,CAAC,KAAK,KAAK,IAAI,KAAK,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC;AAAA,MACrD;AAAA,IACF;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AACF;AAKO,SAAS,SACd,KAAwB,CAAC,UAAa,OACtC;AACA,SAAO,CAAC,WAAiD;AACvD,UAAM,SAAS,IAAI;AAAA,MACjB,OAAO;AAAA,MACP,IAAI,uBAAA;AAAA,IAA0B;AAEhC,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;"}
+{"version":3,"file":"distinct.js","sources":["../../../src/operators/distinct.ts"],"sourcesContent":["import { DifferenceStreamWriter, UnaryOperator } from '../graph.js'\nimport { StreamBuilder } from '../d2.js'\nimport { hash } from '../hashing/index.js'\nimport { MultiSet } from '../multiset.js'\nimport type { Hash } from '../hashing/index.js'\nimport type { DifferenceStreamReader } from '../graph.js'\nimport type { IStreamBuilder, KeyValue } from '../types.js'\n\ntype Multiplicity = number\n\ntype GetValue<T> = T extends KeyValue<any, infer V> ? V : never\n\n/**\n * Operator that removes duplicates\n */\nexport class DistinctOperator<\n  T extends KeyValue<any, any>,\n> extends UnaryOperator<T, KeyValue<number, GetValue<T>>> {\n  #by: (value: T) => any\n  #values: Map<Hash, Multiplicity> // keeps track of the number of times each value has been seen\n\n  constructor(\n    id: number,\n    input: DifferenceStreamReader<T>,\n    output: DifferenceStreamWriter<KeyValue<number, GetValue<T>>>,\n    by: (value: T) => any = (value: T) => value,\n  ) {\n    super(id, input, output)\n    this.#by = by\n    this.#values = new Map()\n  }\n\n  run(): void {\n    const updatedValues = new Map<Hash, [Multiplicity, T]>()\n\n    // Compute the new multiplicity for each value\n    for (const message of this.inputMessages()) {\n      for (const [value, diff] of message.getInner()) {\n        const hashedValue = hash(this.#by(value))\n\n        const oldMultiplicity =\n          updatedValues.get(hashedValue)?.[0] ??\n          this.#values.get(hashedValue) ??\n          0\n        const newMultiplicity = oldMultiplicity + diff\n        updatedValues.set(hashedValue, [newMultiplicity, value])\n      }\n    }\n\n    const result: Array<[KeyValue<number, GetValue<T>>, number]> = []\n\n    // Check which values became visible or disappeared\n    for (const [\n      hashedValue,\n      [newMultiplicity, value],\n    ] of updatedValues.entries()) {\n      const oldMultiplicity = this.#values.get(hashedValue) ?? 0\n\n      if (newMultiplicity === 0) {\n        this.#values.delete(hashedValue)\n      } else {\n        this.#values.set(hashedValue, newMultiplicity)\n      }\n\n      if (oldMultiplicity <= 0 && newMultiplicity > 0) {\n        // The value wasn't present in the stream\n        // but with this change it is now present in the stream\n        result.push([[hash(this.#by(value)), value[1]], 1])\n      } else if (oldMultiplicity > 0 && newMultiplicity <= 0) {\n        // The value was present in the stream\n        // but with this change it is no longer present in the stream\n        result.push([[hash(this.#by(value)), value[1]], -1])\n      }\n    }\n\n    if (result.length > 0) {\n      this.output.sendData(new MultiSet(result))\n    }\n  }\n}\n\n/**\n * Removes duplicate values\n */\nexport function distinct<T extends KeyValue<any, any>>(\n  by: (value: T) => any = (value: T) => value,\n) {\n  return (\n    stream: IStreamBuilder<T>,\n  ): IStreamBuilder<KeyValue<number, GetValue<T>>> => {\n    const output = new StreamBuilder<KeyValue<number, GetValue<T>>>(\n      stream.graph,\n      new DifferenceStreamWriter<KeyValue<number, GetValue<T>>>(),\n    )\n    const operator = new DistinctOperator<T>(\n      stream.graph.getNextOperatorId(),\n      stream.connectReader(),\n      output.writer,\n      by,\n    )\n    stream.graph.addOperator(operator)\n    return output\n  }\n}\n"],"names":[],"mappings":";;;;AAeO,MAAM,yBAEH,cAAgD;AAAA,EACxD;AAAA,EACA;AAAA;AAAA,EAEA,YACE,IACA,OACA,QACA,KAAwB,CAAC,UAAa,OACtC;AACA,UAAM,IAAI,OAAO,MAAM;AACvB,SAAK,MAAM;AACX,SAAK,8BAAc,IAAA;AAAA,EACrB;AAAA,EAEA,MAAY;AACV,UAAM,oCAAoB,IAAA;AAG1B,eAAW,WAAW,KAAK,iBAAiB;AAC1C,iBAAW,CAAC,OAAO,IAAI,KAAK,QAAQ,YAAY;AAC9C,cAAM,cAAc,KAAK,KAAK,IAAI,KAAK,CAAC;AAExC,cAAM,kBACJ,cAAc,IAAI,WAAW,IAAI,CAAC,KAClC,KAAK,QAAQ,IAAI,WAAW,KAC5B;AACF,cAAM,kBAAkB,kBAAkB;AAC1C,sBAAc,IAAI,aAAa,CAAC,iBAAiB,KAAK,CAAC;AAAA,MACzD;AAAA,IACF;AAEA,UAAM,SAAyD,CAAA;AAG/D,eAAW;AAAA,MACT;AAAA,MACA,CAAC,iBAAiB,KAAK;AAAA,IAAA,KACpB,cAAc,WAAW;AAC5B,YAAM,kBAAkB,KAAK,QAAQ,IAAI,WAAW,KAAK;AAEzD,UAAI,oBAAoB,GAAG;AACzB,aAAK,QAAQ,OAAO,WAAW;AAAA,MACjC,OAAO;AACL,aAAK,QAAQ,IAAI,aAAa,eAAe;AAAA,MAC/C;AAEA,UAAI,mBAAmB,KAAK,kBAAkB,GAAG;AAG/C,eAAO,KAAK,CAAC,CAAC,KAAK,KAAK,IAAI,KAAK,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC;AAAA,MACpD,WAAW,kBAAkB,KAAK,mBAAmB,GAAG;AAGtD,eAAO,KAAK,CAAC,CAAC,KAAK,KAAK,IAAI,KAAK,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC;AAAA,MACrD;AAAA,IACF;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AACF;AAKO,SAAS,SACd,KAAwB,CAAC,UAAa,OACtC;AACA,SAAO,CACL,WACkD;AAClD,UAAM,SAAS,IAAI;AAAA,MACjB,OAAO;AAAA,MACP,IAAI,uBAAA;AAAA,IAAsD;AAE5D,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;"}

package/dist/esm/operators/groupBy.js

@@ -122,7 +122,7 @@ function min(valueExtractor) {
     reduce: (values) => {
       let minValue;
       for (const [value, _multiplicity] of values) {
-        if (!minValue || value && value < minValue) {
+        if (value !== void 0 && (minValue === void 0 || value < minValue)) {
           minValue = value;
         }
       }
@@ -137,7 +137,7 @@ function max(valueExtractor) {
     reduce: (values) => {
       let maxValue;
       for (const [value, _multiplicity] of values) {
-        if (!maxValue || value && value > maxValue) {
+        if (value !== void 0 && (maxValue === void 0 || value > maxValue)) {
           maxValue = value;
         }
       }

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

@@ -1 +1 @@
-{"version":3,"file":"groupBy.js","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 (!minValue || (value && value < minValue)) {\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 (!maxValue || (value && value > maxValue)) {\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":[],"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/B,IAAI,CAAC,SAAS;AACZ,cAAM,MAAM,aAAa,IAAI;AAC7B,cAAM,YAAY,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,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,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,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;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,YAAI,CAAC,YAAa,SAAS,QAAQ,UAAW;AAC5C,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,YAAI,CAAC,YAAa,SAAS,QAAQ,UAAW;AAC5C,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;"}
+{"version":3,"file":"groupBy.js","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":[],"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/B,IAAI,CAAC,SAAS;AACZ,cAAM,MAAM,aAAa,IAAI;AAC7B,cAAM,YAAY,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,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,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,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;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;"}

package/dist/esm/operators/groupedTopKWithFractionalIndex.js

@@ -1,7 +1,7 @@
 import { UnaryOperator, DifferenceStreamWriter } from "../graph.js";
 import { StreamBuilder } from "../d2.js";
 import { MultiSet } from "../multiset.js";
-import { TopKState, handleMoveIn, handleMoveOut } from "./topKState.js";
+import { TopKState, handleMoveIn, handleMoveOut, topKBatch } from "./topKState.js";
 import { createKeyedComparator, TopKArray } from "./topKArray.js";
 class GroupedTopKWithFractionalIndexOperator extends UnaryOperator {
   #groupStates = /* @__PURE__ */ new Map();
@@ -73,11 +73,10 @@ class GroupedTopKWithFractionalIndexOperator extends UnaryOperator {
   }
   run() {
     const result = [];
-    for (const message of this.inputMessages()) {
-      for (const [item, multiplicity] of message.getInner()) {
-        const [key, value] = item;
-        this.#processElement(key, value, multiplicity, result);
-      }
+    for (const [[key, value], multiplicity] of topKBatch(
+      this.inputMessages()
+    )) {
+      this.#processElement(key, value, multiplicity, result);
     }
     if (result.length > 0) {
       this.output.sendData(new MultiSet(result));

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

@@ -1 +1 @@
-{"version":3,"file":"groupedTopKWithFractionalIndex.js","sources":["../../../src/operators/groupedTopKWithFractionalIndex.ts"],"sourcesContent":["import { DifferenceStreamWriter, UnaryOperator } from '../graph.js'\nimport { StreamBuilder } from '../d2.js'\nimport { MultiSet } from '../multiset.js'\nimport { TopKState, handleMoveIn, handleMoveOut } from './topKState.js'\nimport { TopKArray, createKeyedComparator } from './topKArray.js'\nimport type { IndexedValue, TopK } from './topKArray.js'\nimport type { DifferenceStreamReader } from '../graph.js'\nimport type { IStreamBuilder, PipedOperator } from '../types.js'\n\nexport interface GroupedTopKWithFractionalIndexOptions<K, T> {\n  limit?: number\n  offset?: number\n  setSizeCallback?: (getSize: () => number) => void\n  setWindowFn?: (\n    windowFn: (options: { offset?: number; limit?: number }) => void,\n  ) => void\n  /**\n   * Function to extract a group key from the element's key and value.\n   * Elements with the same group key will be sorted and limited together.\n   */\n  groupKeyFn: (key: K, value: T) => unknown\n}\n\n/**\n * Operator for grouped fractional indexed topK operations.\n * This operator maintains separate topK windows for each group,\n * allowing per-group limits and ordering.\n *\n * The input is a keyed stream [K, T] and outputs [K, IndexedValue<T>].\n * Elements are grouped by the groupKeyFn, and each group maintains\n * its own sorted collection with independent limit/offset.\n */\nexport class GroupedTopKWithFractionalIndexOperator<\n  K extends string | number,\n  T,\n> extends UnaryOperator<[K, T], [K, IndexedValue<T>]> {\n  #groupStates: Map<unknown, TopKState<K, T>> = new Map()\n  #groupKeyFn: (key: K, value: T) => unknown\n  #comparator: (a: [K, T], b: [K, T]) => number\n  #offset: number\n  #limit: number\n\n  constructor(\n    id: number,\n    inputA: DifferenceStreamReader<[K, T]>,\n    output: DifferenceStreamWriter<[K, IndexedValue<T>]>,\n    comparator: (a: T, b: T) => number,\n    options: GroupedTopKWithFractionalIndexOptions<K, T>,\n  ) {\n    super(id, inputA, output)\n    this.#groupKeyFn = options.groupKeyFn\n    this.#limit = options.limit ?? Infinity\n    this.#offset = options.offset ?? 0\n    this.#comparator = createKeyedComparator(comparator)\n    options.setSizeCallback?.(() => this.#getTotalSize())\n    options.setWindowFn?.(this.#moveTopK.bind(this))\n  }\n\n  /**\n   * Creates a new TopK data structure for a group.\n   * Can be overridden in subclasses to use different implementations (e.g., B+ tree).\n   */\n  protected createTopK(\n    offset: number,\n    limit: number,\n    comparator: (a: [K, T], b: [K, T]) => number,\n  ): TopK<[K, T]> {\n    return new TopKArray(offset, limit, comparator)\n  }\n\n  #getTotalSize(): number {\n    let size = 0\n    for (const state of this.#groupStates.values()) {\n      size += state.size\n    }\n    return size\n  }\n\n  #getOrCreateGroupState(groupKey: unknown): TopKState<K, T> {\n    let state = this.#groupStates.get(groupKey)\n    if (!state) {\n      const topK = this.createTopK(this.#offset, this.#limit, this.#comparator)\n      state = new TopKState(topK)\n      this.#groupStates.set(groupKey, state)\n    }\n    return state\n  }\n\n  #cleanupGroupIfEmpty(groupKey: unknown, state: TopKState<K, T>): void {\n    if (state.isEmpty) {\n      this.#groupStates.delete(groupKey)\n    }\n  }\n\n  /**\n   * Moves the topK window for all groups based on the provided offset and limit.\n   * Any changes to the topK are sent to the output.\n   */\n  #moveTopK({ offset, limit }: { offset?: number; limit?: number }): void {\n    if (offset !== undefined) {\n      this.#offset = offset\n    }\n    if (limit !== undefined) {\n      this.#limit = limit\n    }\n\n    const result: Array<[[K, IndexedValue<T>], number]> = []\n    let hasChanges = false\n\n    for (const state of this.#groupStates.values()) {\n      const diff = state.move({ offset: this.#offset, limit: this.#limit }) // TODO: think we should just pass offset and limit\n\n      diff.moveIns.forEach((moveIn) => handleMoveIn(moveIn, result))\n      diff.moveOuts.forEach((moveOut) => handleMoveOut(moveOut, result))\n\n      if (diff.changes) {\n        hasChanges = true\n      }\n    }\n\n    if (hasChanges) {\n      this.output.sendData(new MultiSet(result))\n    }\n  }\n\n  run(): void {\n    const result: Array<[[K, IndexedValue<T>], number]> = []\n    for (const message of this.inputMessages()) {\n      for (const [item, multiplicity] of message.getInner()) {\n        const [key, value] = item\n        this.#processElement(key, value, multiplicity, result)\n      }\n    }\n\n    if (result.length > 0) {\n      this.output.sendData(new MultiSet(result))\n    }\n  }\n\n  #processElement(\n    key: K,\n    value: T,\n    multiplicity: number,\n    result: Array<[[K, IndexedValue<T>], number]>,\n  ): void {\n    const groupKey = this.#groupKeyFn(key, value)\n    const state = this.#getOrCreateGroupState(groupKey)\n\n    const changes = state.processElement(key, value, multiplicity)\n    handleMoveIn(changes.moveIn, result)\n    handleMoveOut(changes.moveOut, result)\n\n    // Cleanup empty groups to prevent memory leaks\n    this.#cleanupGroupIfEmpty(groupKey, state)\n  }\n}\n\n/**\n * Limits the number of results per group based on a comparator, with optional offset.\n * Uses fractional indexing to minimize the number of changes when elements move positions.\n * Each element is assigned a fractional index that is lexicographically sortable.\n * When elements move, only the indices of the moved elements are updated, not all elements.\n *\n * This operator groups elements by the provided groupKeyFn and applies the limit/offset\n * independently to each group.\n *\n * @param comparator - A function that compares two elements for ordering\n * @param options - Configuration including groupKeyFn, limit, and offset\n * @returns A piped operator that orders elements per group and limits results per group\n */\nexport function groupedTopKWithFractionalIndex<K extends string | number, T>(\n  comparator: (a: T, b: T) => number,\n  options: GroupedTopKWithFractionalIndexOptions<K, T>,\n): PipedOperator<[K, T], [K, IndexedValue<T>]> {\n  return (\n    stream: IStreamBuilder<[K, T]>,\n  ): IStreamBuilder<[K, IndexedValue<T>]> => {\n    const output = new StreamBuilder<[K, IndexedValue<T>]>(\n      stream.graph,\n      new DifferenceStreamWriter<[K, IndexedValue<T>]>(),\n    )\n    const operator = new GroupedTopKWithFractionalIndexOperator<K, T>(\n      stream.graph.getNextOperatorId(),\n      stream.connectReader(),\n      output.writer,\n      comparator,\n      options,\n    )\n    stream.graph.addOperator(operator)\n    return output\n  }\n}\n"],"names":[],"mappings":";;;;;AAgCO,MAAM,+CAGH,cAA4C;AAAA,EACpD,mCAAkD,IAAA;AAAA,EAClD;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YACE,IACA,QACA,QACA,YACA,SACA;AACA,UAAM,IAAI,QAAQ,MAAM;AACxB,SAAK,cAAc,QAAQ;AAC3B,SAAK,SAAS,QAAQ,SAAS;AAC/B,SAAK,UAAU,QAAQ,UAAU;AACjC,SAAK,cAAc,sBAAsB,UAAU;AACnD,YAAQ,kBAAkB,MAAM,KAAK,cAAA,CAAe;AACpD,YAAQ,cAAc,KAAK,UAAU,KAAK,IAAI,CAAC;AAAA,EACjD;AAAA;AAAA;AAAA;AAAA;AAAA,EAMU,WACR,QACA,OACA,YACc;AACd,WAAO,IAAI,UAAU,QAAQ,OAAO,UAAU;AAAA,EAChD;AAAA,EAEA,gBAAwB;AACtB,QAAI,OAAO;AACX,eAAW,SAAS,KAAK,aAAa,OAAA,GAAU;AAC9C,cAAQ,MAAM;AAAA,IAChB;AACA,WAAO;AAAA,EACT;AAAA,EAEA,uBAAuB,UAAoC;AACzD,QAAI,QAAQ,KAAK,aAAa,IAAI,QAAQ;AAC1C,QAAI,CAAC,OAAO;AACV,YAAM,OAAO,KAAK,WAAW,KAAK,SAAS,KAAK,QAAQ,KAAK,WAAW;AACxE,cAAQ,IAAI,UAAU,IAAI;AAC1B,WAAK,aAAa,IAAI,UAAU,KAAK;AAAA,IACvC;AACA,WAAO;AAAA,EACT;AAAA,EAEA,qBAAqB,UAAmB,OAA8B;AACpE,QAAI,MAAM,SAAS;AACjB,WAAK,aAAa,OAAO,QAAQ;AAAA,IACnC;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA,EAMA,UAAU,EAAE,QAAQ,SAAoD;AACtE,QAAI,WAAW,QAAW;AACxB,WAAK,UAAU;AAAA,IACjB;AACA,QAAI,UAAU,QAAW;AACvB,WAAK,SAAS;AAAA,IAChB;AAEA,UAAM,SAAgD,CAAA;AACtD,QAAI,aAAa;AAEjB,eAAW,SAAS,KAAK,aAAa,OAAA,GAAU;AAC9C,YAAM,OAAO,MAAM,KAAK,EAAE,QAAQ,KAAK,SAAS,OAAO,KAAK,QAAQ;AAEpE,WAAK,QAAQ,QAAQ,CAAC,WAAW,aAAa,QAAQ,MAAM,CAAC;AAC7D,WAAK,SAAS,QAAQ,CAAC,YAAY,cAAc,SAAS,MAAM,CAAC;AAEjE,UAAI,KAAK,SAAS;AAChB,qBAAa;AAAA,MACf;AAAA,IACF;AAEA,QAAI,YAAY;AACd,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,MAAY;AACV,UAAM,SAAgD,CAAA;AACtD,eAAW,WAAW,KAAK,iBAAiB;AAC1C,iBAAW,CAAC,MAAM,YAAY,KAAK,QAAQ,YAAY;AACrD,cAAM,CAAC,KAAK,KAAK,IAAI;AACrB,aAAK,gBAAgB,KAAK,OAAO,cAAc,MAAM;AAAA,MACvD;AAAA,IACF;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,gBACE,KACA,OACA,cACA,QACM;AACN,UAAM,WAAW,KAAK,YAAY,KAAK,KAAK;AAC5C,UAAM,QAAQ,KAAK,uBAAuB,QAAQ;AAElD,UAAM,UAAU,MAAM,eAAe,KAAK,OAAO,YAAY;AAC7D,iBAAa,QAAQ,QAAQ,MAAM;AACnC,kBAAc,QAAQ,SAAS,MAAM;AAGrC,SAAK,qBAAqB,UAAU,KAAK;AAAA,EAC3C;AACF;AAeO,SAAS,+BACd,YACA,SAC6C;AAC7C,SAAO,CACL,WACyC;AACzC,UAAM,SAAS,IAAI;AAAA,MACjB,OAAO;AAAA,MACP,IAAI,uBAAA;AAAA,IAA6C;AAEnD,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,OAAO;AAAA,MACP;AAAA,MACA;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;"}
+{"version":3,"file":"groupedTopKWithFractionalIndex.js","sources":["../../../src/operators/groupedTopKWithFractionalIndex.ts"],"sourcesContent":["import { DifferenceStreamWriter, UnaryOperator } from '../graph.js'\nimport { StreamBuilder } from '../d2.js'\nimport { MultiSet } from '../multiset.js'\nimport {\n  TopKState,\n  handleMoveIn,\n  handleMoveOut,\n  topKBatch,\n} from './topKState.js'\nimport { TopKArray, createKeyedComparator } from './topKArray.js'\nimport type { IndexedValue, TopK } from './topKArray.js'\nimport type { DifferenceStreamReader } from '../graph.js'\nimport type { IStreamBuilder, PipedOperator } from '../types.js'\n\nexport interface GroupedTopKWithFractionalIndexOptions<K, T> {\n  limit?: number\n  offset?: number\n  setSizeCallback?: (getSize: () => number) => void\n  setWindowFn?: (\n    windowFn: (options: { offset?: number; limit?: number }) => void,\n  ) => void\n  /**\n   * Function to extract a group key from the element's key and value.\n   * Elements with the same group key will be sorted and limited together.\n   */\n  groupKeyFn: (key: K, value: T) => unknown\n}\n\n/**\n * Operator for grouped fractional indexed topK operations.\n * This operator maintains separate topK windows for each group,\n * allowing per-group limits and ordering.\n *\n * The input is a keyed stream [K, T] and outputs [K, IndexedValue<T>].\n * Elements are grouped by the groupKeyFn, and each group maintains\n * its own sorted collection with independent limit/offset.\n */\nexport class GroupedTopKWithFractionalIndexOperator<\n  K extends string | number,\n  T,\n> extends UnaryOperator<[K, T], [K, IndexedValue<T>]> {\n  #groupStates: Map<unknown, TopKState<K, T>> = new Map()\n  #groupKeyFn: (key: K, value: T) => unknown\n  #comparator: (a: [K, T], b: [K, T]) => number\n  #offset: number\n  #limit: number\n\n  constructor(\n    id: number,\n    inputA: DifferenceStreamReader<[K, T]>,\n    output: DifferenceStreamWriter<[K, IndexedValue<T>]>,\n    comparator: (a: T, b: T) => number,\n    options: GroupedTopKWithFractionalIndexOptions<K, T>,\n  ) {\n    super(id, inputA, output)\n    this.#groupKeyFn = options.groupKeyFn\n    this.#limit = options.limit ?? Infinity\n    this.#offset = options.offset ?? 0\n    this.#comparator = createKeyedComparator(comparator)\n    options.setSizeCallback?.(() => this.#getTotalSize())\n    options.setWindowFn?.(this.#moveTopK.bind(this))\n  }\n\n  /**\n   * Creates a new TopK data structure for a group.\n   * Can be overridden in subclasses to use different implementations (e.g., B+ tree).\n   */\n  protected createTopK(\n    offset: number,\n    limit: number,\n    comparator: (a: [K, T], b: [K, T]) => number,\n  ): TopK<[K, T]> {\n    return new TopKArray(offset, limit, comparator)\n  }\n\n  #getTotalSize(): number {\n    let size = 0\n    for (const state of this.#groupStates.values()) {\n      size += state.size\n    }\n    return size\n  }\n\n  #getOrCreateGroupState(groupKey: unknown): TopKState<K, T> {\n    let state = this.#groupStates.get(groupKey)\n    if (!state) {\n      const topK = this.createTopK(this.#offset, this.#limit, this.#comparator)\n      state = new TopKState(topK)\n      this.#groupStates.set(groupKey, state)\n    }\n    return state\n  }\n\n  #cleanupGroupIfEmpty(groupKey: unknown, state: TopKState<K, T>): void {\n    if (state.isEmpty) {\n      this.#groupStates.delete(groupKey)\n    }\n  }\n\n  /**\n   * Moves the topK window for all groups based on the provided offset and limit.\n   * Any changes to the topK are sent to the output.\n   */\n  #moveTopK({ offset, limit }: { offset?: number; limit?: number }): void {\n    if (offset !== undefined) {\n      this.#offset = offset\n    }\n    if (limit !== undefined) {\n      this.#limit = limit\n    }\n\n    const result: Array<[[K, IndexedValue<T>], number]> = []\n    let hasChanges = false\n\n    for (const state of this.#groupStates.values()) {\n      const diff = state.move({ offset: this.#offset, limit: this.#limit }) // TODO: think we should just pass offset and limit\n\n      diff.moveIns.forEach((moveIn) => handleMoveIn(moveIn, result))\n      diff.moveOuts.forEach((moveOut) => handleMoveOut(moveOut, result))\n\n      if (diff.changes) {\n        hasChanges = true\n      }\n    }\n\n    if (hasChanges) {\n      this.output.sendData(new MultiSet(result))\n    }\n  }\n\n  run(): void {\n    const result: Array<[[K, IndexedValue<T>], number]> = []\n    for (const [[key, value], multiplicity] of topKBatch(\n      this.inputMessages(),\n    )) {\n      this.#processElement(key, value, multiplicity, result)\n    }\n\n    if (result.length > 0) {\n      this.output.sendData(new MultiSet(result))\n    }\n  }\n\n  #processElement(\n    key: K,\n    value: T,\n    multiplicity: number,\n    result: Array<[[K, IndexedValue<T>], number]>,\n  ): void {\n    const groupKey = this.#groupKeyFn(key, value)\n    const state = this.#getOrCreateGroupState(groupKey)\n\n    const changes = state.processElement(key, value, multiplicity)\n    handleMoveIn(changes.moveIn, result)\n    handleMoveOut(changes.moveOut, result)\n\n    // Cleanup empty groups to prevent memory leaks\n    this.#cleanupGroupIfEmpty(groupKey, state)\n  }\n}\n\n/**\n * Limits the number of results per group based on a comparator, with optional offset.\n * Uses fractional indexing to minimize the number of changes when elements move positions.\n * Each element is assigned a fractional index that is lexicographically sortable.\n * When elements move, only the indices of the moved elements are updated, not all elements.\n *\n * This operator groups elements by the provided groupKeyFn and applies the limit/offset\n * independently to each group.\n *\n * @param comparator - A function that compares two elements for ordering\n * @param options - Configuration including groupKeyFn, limit, and offset\n * @returns A piped operator that orders elements per group and limits results per group\n */\nexport function groupedTopKWithFractionalIndex<K extends string | number, T>(\n  comparator: (a: T, b: T) => number,\n  options: GroupedTopKWithFractionalIndexOptions<K, T>,\n): PipedOperator<[K, T], [K, IndexedValue<T>]> {\n  return (\n    stream: IStreamBuilder<[K, T]>,\n  ): IStreamBuilder<[K, IndexedValue<T>]> => {\n    const output = new StreamBuilder<[K, IndexedValue<T>]>(\n      stream.graph,\n      new DifferenceStreamWriter<[K, IndexedValue<T>]>(),\n    )\n    const operator = new GroupedTopKWithFractionalIndexOperator<K, T>(\n      stream.graph.getNextOperatorId(),\n      stream.connectReader(),\n      output.writer,\n      comparator,\n      options,\n    )\n    stream.graph.addOperator(operator)\n    return output\n  }\n}\n"],"names":[],"mappings":";;;;;AAqCO,MAAM,+CAGH,cAA4C;AAAA,EACpD,mCAAkD,IAAA;AAAA,EAClD;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EAEA,YACE,IACA,QACA,QACA,YACA,SACA;AACA,UAAM,IAAI,QAAQ,MAAM;AACxB,SAAK,cAAc,QAAQ;AAC3B,SAAK,SAAS,QAAQ,SAAS;AAC/B,SAAK,UAAU,QAAQ,UAAU;AACjC,SAAK,cAAc,sBAAsB,UAAU;AACnD,YAAQ,kBAAkB,MAAM,KAAK,cAAA,CAAe;AACpD,YAAQ,cAAc,KAAK,UAAU,KAAK,IAAI,CAAC;AAAA,EACjD;AAAA;AAAA;AAAA;AAAA;AAAA,EAMU,WACR,QACA,OACA,YACc;AACd,WAAO,IAAI,UAAU,QAAQ,OAAO,UAAU;AAAA,EAChD;AAAA,EAEA,gBAAwB;AACtB,QAAI,OAAO;AACX,eAAW,SAAS,KAAK,aAAa,OAAA,GAAU;AAC9C,cAAQ,MAAM;AAAA,IAChB;AACA,WAAO;AAAA,EACT;AAAA,EAEA,uBAAuB,UAAoC;AACzD,QAAI,QAAQ,KAAK,aAAa,IAAI,QAAQ;AAC1C,QAAI,CAAC,OAAO;AACV,YAAM,OAAO,KAAK,WAAW,KAAK,SAAS,KAAK,QAAQ,KAAK,WAAW;AACxE,cAAQ,IAAI,UAAU,IAAI;AAC1B,WAAK,aAAa,IAAI,UAAU,KAAK;AAAA,IACvC;AACA,WAAO;AAAA,EACT;AAAA,EAEA,qBAAqB,UAAmB,OAA8B;AACpE,QAAI,MAAM,SAAS;AACjB,WAAK,aAAa,OAAO,QAAQ;AAAA,IACnC;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA,EAMA,UAAU,EAAE,QAAQ,SAAoD;AACtE,QAAI,WAAW,QAAW;AACxB,WAAK,UAAU;AAAA,IACjB;AACA,QAAI,UAAU,QAAW;AACvB,WAAK,SAAS;AAAA,IAChB;AAEA,UAAM,SAAgD,CAAA;AACtD,QAAI,aAAa;AAEjB,eAAW,SAAS,KAAK,aAAa,OAAA,GAAU;AAC9C,YAAM,OAAO,MAAM,KAAK,EAAE,QAAQ,KAAK,SAAS,OAAO,KAAK,QAAQ;AAEpE,WAAK,QAAQ,QAAQ,CAAC,WAAW,aAAa,QAAQ,MAAM,CAAC;AAC7D,WAAK,SAAS,QAAQ,CAAC,YAAY,cAAc,SAAS,MAAM,CAAC;AAEjE,UAAI,KAAK,SAAS;AAChB,qBAAa;AAAA,MACf;AAAA,IACF;AAEA,QAAI,YAAY;AACd,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,MAAY;AACV,UAAM,SAAgD,CAAA;AACtD,eAAW,CAAC,CAAC,KAAK,KAAK,GAAG,YAAY,KAAK;AAAA,MACzC,KAAK,cAAA;AAAA,IAAc,GAClB;AACD,WAAK,gBAAgB,KAAK,OAAO,cAAc,MAAM;AAAA,IACvD;AAEA,QAAI,OAAO,SAAS,GAAG;AACrB,WAAK,OAAO,SAAS,IAAI,SAAS,MAAM,CAAC;AAAA,IAC3C;AAAA,EACF;AAAA,EAEA,gBACE,KACA,OACA,cACA,QACM;AACN,UAAM,WAAW,KAAK,YAAY,KAAK,KAAK;AAC5C,UAAM,QAAQ,KAAK,uBAAuB,QAAQ;AAElD,UAAM,UAAU,MAAM,eAAe,KAAK,OAAO,YAAY;AAC7D,iBAAa,QAAQ,QAAQ,MAAM;AACnC,kBAAc,QAAQ,SAAS,MAAM;AAGrC,SAAK,qBAAqB,UAAU,KAAK;AAAA,EAC3C;AACF;AAeO,SAAS,+BACd,YACA,SAC6C;AAC7C,SAAO,CACL,WACyC;AACzC,UAAM,SAAS,IAAI;AAAA,MACjB,OAAO;AAAA,MACP,IAAI,uBAAA;AAAA,IAA6C;AAEnD,UAAM,WAAW,IAAI;AAAA,MACnB,OAAO,MAAM,kBAAA;AAAA,MACb,OAAO,cAAA;AAAA,MACP,OAAO;AAAA,MACP;AAAA,MACA;AAAA,IAAA;AAEF,WAAO,MAAM,YAAY,QAAQ;AACjC,WAAO;AAAA,EACT;AACF;"}

package/dist/esm/operators/topKArray.js

@@ -78,6 +78,7 @@ class TopKArray {
     const fractionalIndex = generateKeyBetween(indexBefore, indexAfter);
     const val = indexedValue(value, fractionalIndex);
     this.#sortedValues.splice(index, 0, val);
+    if (this.#topKStart === this.#topKEnd) return result;
     if (index < this.#topKEnd) {
       const moveInIndex = Math.max(index, this.#topKStart);
       if (moveInIndex < this.#sortedValues.length) {
@@ -99,6 +100,7 @@ class TopKArray {
     const result = { moveIn: null, moveOut: null };
     const index = this.#findIndex(value);
     const [removedElem] = this.#sortedValues.splice(index, 1);
+    if (this.#topKStart === this.#topKEnd) return result;
     if (index < this.#topKEnd) {
       result.moveOut = removedElem;
       if (index < this.#topKStart) {