@tanstack/db-ivm
Version:
Incremental View Maintenance for TanStack DB based on Differential Dataflow
1 lines • 10.2 kB
Source Map (JSON)
{"version":3,"file":"multiset.cjs","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 getInner(): MultiSetArray<T> {\n return this.#inner\n }\n}\n"],"names":["chunkedArrayPush","globalObjectIdGenerator","DefaultMap","hash"],"mappings":";;;;;;;;;;;;;AAaO,MAAM,YAAN,MAAM,UAAY;AAAA,EAGvB,YAAY,OAAyB,IAAI;AAHpC;AACL;AAGE,uBAAK,QAAS;AAAA,EAChB;AAAA,EAEA,SAAS,SAAS,OAAe;AAC/B,WAAO,YAAY,KAAK,UAAU,mBAAK,SAAQ,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,UAAS,KAAK,MAAM,IAAI,CAAC;AAAA,EACtC;AAAA;AAAA;AAAA;AAAA,EAKA,IAAO,GAAgC;AACrC,WAAO,IAAI;AAAA,MACT,mBAAK,QAAO,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,UAAS,mBAAK,QAAO,OAAO,CAAC,CAAC,MAAM,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;AAAA,EAChE;AAAA;AAAA;AAAA;AAAA,EAKA,SAAsB;AACpB,WAAO,IAAI;AAAA,MACT,mBAAK,QAAO,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;AAC9BA,2BAAiB,KAAK,mBAAK,OAAM;AACjCA,UAAAA,iBAAiB,KAAK,MAAM,UAAU;AACtC,WAAO,IAAI,UAAS,GAAG;AAAA,EACzB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,cAA2B;;AAEzB,QAAI,mBAAK,QAAO,SAAS,GAAG;AAC1B,YAAM,aAAY,wBAAK,QAAO,CAAC,MAAb,mBAAiB;AACnC,UAAI,MAAM,QAAQ,SAAS,KAAK,UAAU,WAAW,GAAG;AACtD,eAAO,sBAAK,0CAAL;AAAA,MACT;AAAA,IACF;AAGA,WAAO,sBAAK,4CAAL;AAAA,EACT;AAAA,EA2HA,OAAO,OAA6C;AAClD,UAAM,aAAa,iBAAiB,YAAW,MAAM,aAAa;AAClEA,2BAAiB,mBAAK,SAAQ,UAAU;AAAA,EAC1C;AAAA,EAEA,WAA6B;AAC3B,WAAO,mBAAK;AAAA,EACd;AACF;AAxME;AADK;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAkFL,sBAAA,WAAiC;AAC/B,QAAM,mCAAmB,IAAA;AACzB,QAAM,6BAAa,IAAA;AASnB,QAAM,aAAa,CAAC,UAA8B;AAChD,QAAI,MAAM,WAAW,GAAG;AACtB,YAAM,IAAI,MAAM,4BAA4B;AAAA,IAC9C;AACA,UAAM,CAAC,OAAO,MAAM,IAAI;AACxB,WAAO,GAAGC,8BAAwB,YAAY,KAAK,CAAC,IAAIA,8BAAwB,YAAY,MAAM,CAAC;AAAA,EACrG;AAGA,aAAW,CAAC,MAAM,YAAY,KAAK,mBAAK,SAAQ;AAE9C,QAAI,CAAC,MAAM,QAAQ,IAAI,KAAK,KAAK,WAAW,GAAG;AAE7C,aAAO,sBAAK,4CAAL;AAAA,IACT;AAEA,UAAM,CAAC,KAAK,KAAK,IAAI;AAGrB,QAAI,OAAO,QAAQ,YAAY,OAAO,QAAQ,UAAU;AAEtD,aAAO,sBAAK,4CAAL;AAAA,IACT;AAGA,QAAI;AACJ,QAAI,MAAM,QAAQ,KAAK,KAAK,MAAM,WAAW,GAAG;AAE9C,gBAAU,WAAW,KAAK;AAAA,IAC5B,OAAO;AAEL,gBAAUA,MAAAA,wBAAwB,YAAY,KAAK;AAAA,IACrD;AAGA,UAAM,eAAe,MAAM,MAAM;AACjC,iBAAa;AAAA,MACX;AAAA,OACC,aAAa,IAAI,YAAY,KAAK,KAAK;AAAA,IAAA;AAI1C,QAAI,CAAC,OAAO,IAAI,YAAY,GAAG;AAC7B,aAAO,IAAI,cAAc,IAAS;AAAA,IACpC;AAAA,EACF;AAGA,QAAM,SAA2B,CAAA;AACjC,aAAW,CAAC,cAAc,YAAY,KAAK,cAAc;AACvD,QAAI,iBAAiB,GAAG;AACtB,aAAO,KAAK,CAAC,OAAO,IAAI,YAAY,GAAI,YAAY,CAAC;AAAA,IACvD;AAAA,EACF;AAEA,SAAO,IAAI,UAAS,MAAM;AAC5B;AAAA;AAAA;AAAA;AAKA,wBAAA,WAAmC;AACjC,QAAM,eAAe,IAAIC,iBAAoC,MAAM,CAAC;AACpE,QAAM,6BAAa,IAAA;AAEnB,MAAI,YAAY;AAChB,MAAI,YAAY;AAChB,MAAI,WAAW;AACf,aAAW,CAAC,MAAM,CAAC,KAAK,mBAAK,SAAQ;AACnC,QAAI,OAAO,SAAS,UAAU;AAC5B,kBAAY;AAAA,IACd,WAAW,OAAO,SAAS,UAAU;AACnC,kBAAY;AAAA,IACd,OAAO;AACL,iBAAW;AACX;AAAA,IACF;AAAA,EACF;AAEA,QAAM,cAAc,YAAa,aAAa;AAE9C,aAAW,CAAC,MAAM,YAAY,KAAK,mBAAK,SAAQ;AAC9C,UAAM,MAAM,cAAcC,UAAK,IAAI,IAAK;AACxC,QAAI,eAAe,CAAC,OAAO,IAAI,GAAa,GAAG;AAC7C,aAAO,IAAI,KAAe,IAAI;AAAA,IAChC;AACA,iBAAa,OAAO,KAAK,CAAC,UAAU,QAAQ,YAAY;AAAA,EAC1D;AAEA,QAAM,SAA2B,CAAA;AACjC,aAAW,CAAC,KAAK,YAAY,KAAK,aAAa,WAAW;AACxD,QAAI,iBAAiB,GAAG;AACtB,YAAM,YAAY,cAAc,OAAO,IAAI,GAAa,IAAI;AAC5D,aAAO,KAAK,CAAC,WAAgB,YAAY,CAAC;AAAA,IAC5C;AAAA,EACF;AAEA,SAAO,IAAI,UAAS,MAAM;AAC5B;AA/LK,IAAM,WAAN;;"}