@danielsimonjr/memory-mcp 12.2.3 → 12.5.2

package/dist/search/TFIDFIndexManager.js

@@ -1,215 +0,0 @@
-/**
- * TF-IDF Index Manager
- *
- * Manages pre-calculated TF-IDF indexes for fast ranked search.
- * Handles index building, incremental updates, and persistence.
- *
- * @module search/TFIDFIndexManager
- */
-import * as fs from 'fs/promises';
-import * as path from 'path';
-import { calculateIDFFromTokenSets, tokenize } from '../utils/index.js';
-const INDEX_VERSION = '1.0';
-const INDEX_FILENAME = 'tfidf-index.json';
-/**
- * Manages TF-IDF index lifecycle: building, updating, and persistence.
- */
-export class TFIDFIndexManager {
-    indexPath;
-    index = null;
-    constructor(storageDir) {
-        this.indexPath = path.join(storageDir, '.indexes', INDEX_FILENAME);
-    }
-    /**
-     * Build a complete TF-IDF index from a knowledge graph.
-     *
-     * @param graph - Knowledge graph to index
-     * @returns Newly built TF-IDF index
-     */
-    async buildIndex(graph) {
-        const documents = new Map();
-        const allTokenSets = [];
-        // Build document vectors - tokenize once per document
-        for (const entity of graph.entities) {
-            const documentText = [
-                entity.name,
-                entity.entityType,
-                ...entity.observations,
-            ].join(' ');
-            const tokens = tokenize(documentText);
-            const tokenSet = new Set(tokens);
-            allTokenSets.push(tokenSet);
-            // Calculate term frequencies
-            const termFreq = {};
-            for (const term of tokens) {
-                termFreq[term] = (termFreq[term] || 0) + 1;
-            }
-            documents.set(entity.name, {
-                entityName: entity.name,
-                terms: termFreq,
-                documentText,
-            });
-        }
-        // Calculate IDF for all terms using pre-tokenized sets (O(1) lookup per document)
-        const idf = new Map();
-        const allTerms = new Set(allTokenSets.flatMap(s => Array.from(s)));
-        for (const term of allTerms) {
-            const idfScore = calculateIDFFromTokenSets(term, allTokenSets);
-            idf.set(term, idfScore);
-        }
-        this.index = {
-            version: INDEX_VERSION,
-            lastUpdated: new Date().toISOString(),
-            documents,
-            idf,
-        };
-        return this.index;
-    }
-    /**
-     * Update the index incrementally when entities change.
-     *
-     * More efficient than rebuilding the entire index.
-     *
-     * @param graph - Updated knowledge graph
-     * @param changedEntityNames - Names of entities that changed
-     */
-    async updateIndex(graph, changedEntityNames) {
-        if (!this.index) {
-            // No existing index, build from scratch
-            return this.buildIndex(graph);
-        }
-        // Rebuild document vectors for changed entities
-        const allTokenSets = [];
-        const updatedDocuments = new Map(this.index.documents);
-        // Remove deleted entities
-        for (const entityName of changedEntityNames) {
-            const entity = graph.entities.find(e => e.name === entityName);
-            if (!entity) {
-                updatedDocuments.delete(entityName);
-            }
-        }
-        // Update/add changed entities - tokenize once per document
-        for (const entity of graph.entities) {
-            const documentText = [
-                entity.name,
-                entity.entityType,
-                ...entity.observations,
-            ].join(' ');
-            const tokens = tokenize(documentText);
-            const tokenSet = new Set(tokens);
-            allTokenSets.push(tokenSet);
-            if (changedEntityNames.has(entity.name)) {
-                // Calculate term frequencies for changed entity
-                const termFreq = {};
-                for (const term of tokens) {
-                    termFreq[term] = (termFreq[term] || 0) + 1;
-                }
-                updatedDocuments.set(entity.name, {
-                    entityName: entity.name,
-                    terms: termFreq,
-                    documentText,
-                });
-            }
-        }
-        // Recalculate IDF using pre-tokenized sets (O(1) lookup per document)
-        const idf = new Map();
-        const allTerms = new Set(allTokenSets.flatMap(s => Array.from(s)));
-        for (const term of allTerms) {
-            const idfScore = calculateIDFFromTokenSets(term, allTokenSets);
-            idf.set(term, idfScore);
-        }
-        this.index = {
-            version: INDEX_VERSION,
-            lastUpdated: new Date().toISOString(),
-            documents: updatedDocuments,
-            idf,
-        };
-        return this.index;
-    }
-    /**
-     * Load index from disk.
-     *
-     * @returns Loaded index or null if not found
-     */
-    async loadIndex() {
-        try {
-            const data = await fs.readFile(this.indexPath, 'utf-8');
-            const serialized = JSON.parse(data);
-            this.index = {
-                version: serialized.version,
-                lastUpdated: serialized.lastUpdated,
-                documents: new Map(serialized.documents),
-                idf: new Map(serialized.idf),
-            };
-            return this.index;
-        }
-        catch (error) {
-            // Index doesn't exist or is invalid
-            return null;
-        }
-    }
-    /**
-     * Save index to disk.
-     *
-     * @param index - Index to save (uses cached index if not provided)
-     */
-    async saveIndex(index) {
-        const indexToSave = index || this.index;
-        if (!indexToSave) {
-            throw new Error('No index to save');
-        }
-        // Ensure index directory exists
-        const indexDir = path.dirname(this.indexPath);
-        await fs.mkdir(indexDir, { recursive: true });
-        // Serialize Map objects to arrays for JSON
-        const serialized = {
-            version: indexToSave.version,
-            lastUpdated: indexToSave.lastUpdated,
-            documents: Array.from(indexToSave.documents.entries()),
-            idf: Array.from(indexToSave.idf.entries()),
-        };
-        await fs.writeFile(this.indexPath, JSON.stringify(serialized, null, 2), 'utf-8');
-    }
-    /**
-     * Get the current cached index.
-     *
-     * @returns Cached index or null if not loaded
-     */
-    getIndex() {
-        return this.index;
-    }
-    /**
-     * Clear the cached index and delete from disk.
-     */
-    async clearIndex() {
-        this.index = null;
-        try {
-            await fs.unlink(this.indexPath);
-        }
-        catch {
-            // Index file doesn't exist, nothing to delete
-        }
-    }
-    /**
-     * Check if the index needs rebuilding based on graph state.
-     *
-     * @param graph - Current knowledge graph
-     * @returns True if index should be rebuilt
-     */
-    needsRebuild(graph) {
-        if (!this.index) {
-            return true;
-        }
-        // Check if entity count matches
-        if (this.index.documents.size !== graph.entities.length) {
-            return true;
-        }
-        // Check if all entities are in index
-        for (const entity of graph.entities) {
-            if (!this.index.documents.has(entity.name)) {
-                return true;
-            }
-        }
-        return false;
-    }
-}

package/dist/search/VectorStore.d.ts

@@ -1,235 +0,0 @@
-/**
- * Vector Store
- *
- * Phase 4 Sprint 11: Vector storage and retrieval for semantic search.
- * Provides in-memory and SQLite-backed implementations.
- *
- * @module search/VectorStore
- */
-import type { IVectorStore, VectorSearchResult } from '../types/index.js';
-/**
- * Calculate cosine similarity between two vectors.
- *
- * Uses an optimized inner loop without array methods for maximum performance.
- *
- * @param a - First vector
- * @param b - Second vector
- * @returns Cosine similarity score (0.0 to 1.0)
- */
-export declare function cosineSimilarity(a: number[], b: number[]): number;
-/**
- * In-Memory Vector Store
- *
- * Stores vectors in memory using a Map for O(1) add/remove operations.
- * Search uses brute-force cosine similarity which is O(n) but fast for
- * small to medium graphs (<10K entities).
- *
- * @example
- * ```typescript
- * const store = new InMemoryVectorStore();
- * store.add("entity1", [0.1, 0.2, 0.3]);
- * store.add("entity2", [0.4, 0.5, 0.6]);
- * const results = store.search([0.1, 0.2, 0.3], 5);
- * console.log(results); // [{ name: "entity1", score: 1.0 }, ...]
- * ```
- */
-export declare class InMemoryVectorStore implements IVectorStore {
-    /** Map of entity name to embedding vector */
-    private vectors;
-    /**
-     * Add a vector for an entity.
-     *
-     * @param entityName - Name of the entity
-     * @param vector - Embedding vector
-     */
-    add(entityName: string, vector: number[]): void;
-    /**
-     * Search for similar vectors using cosine similarity.
-     *
-     * @param queryVector - Query embedding vector
-     * @param k - Number of results to return
-     * @returns Array of results with entity name and similarity score
-     */
-    search(queryVector: number[], k: number): VectorSearchResult[];
-    /**
-     * Remove a vector by entity name.
-     *
-     * @param entityName - Name of the entity to remove
-     * @returns True if found and removed
-     */
-    remove(entityName: string): boolean;
-    /**
-     * Get the number of vectors stored.
-     *
-     * @returns Number of vectors
-     */
-    size(): number;
-    /**
-     * Clear all vectors from the store.
-     */
-    clear(): void;
-    /**
-     * Check if a vector exists for an entity.
-     *
-     * @param entityName - Name of the entity
-     * @returns True if vector exists
-     */
-    has(entityName: string): boolean;
-    /**
-     * Get the vector for an entity.
-     *
-     * @param entityName - Name of the entity
-     * @returns Vector if found, undefined otherwise
-     */
-    get(entityName: string): number[] | undefined;
-    /**
-     * Get all entity names with stored vectors.
-     *
-     * @returns Array of entity names
-     */
-    getEntityNames(): string[];
-    /**
-     * Load vectors from an iterable source.
-     *
-     * @param entries - Iterable of [entityName, vector] pairs
-     */
-    loadFrom(entries: Iterable<[string, number[]]>): void;
-}
-/**
- * SQLite Vector Store
- *
- * Persists vectors to SQLite storage while maintaining an in-memory cache
- * for fast search operations. Combines persistence with performance.
- *
- * Uses SQLiteStorage's embedding storage methods for persistence.
- *
- * @example
- * ```typescript
- * const store = new SQLiteVectorStore(sqliteStorage);
- * await store.initialize();
- * store.add("entity1", [0.1, 0.2, 0.3]);
- * const results = store.search([0.1, 0.2, 0.3], 5);
- * ```
- */
-export declare class SQLiteVectorStore implements IVectorStore {
-    /** In-memory cache for fast search */
-    private memoryStore;
-    /** SQLite storage reference for persistence */
-    private storage;
-    /** Whether the store has been initialized */
-    private initialized;
-    /** Model name used for embeddings */
-    private embeddingModel;
-    /**
-     * Create a SQLite vector store.
-     *
-     * @param storage - SQLite storage instance with embedding support
-     * @param embeddingModel - Model name used for embeddings
-     */
-    constructor(storage?: SQLiteStorageWithEmbeddings, embeddingModel?: string);
-    /**
-     * Initialize the store by loading vectors from SQLite.
-     */
-    initialize(): Promise<void>;
-    /**
-     * Add a vector for an entity.
-     * Stores in both memory and SQLite for persistence.
-     *
-     * @param entityName - Name of the entity
-     * @param vector - Embedding vector
-     */
-    add(entityName: string, vector: number[]): void;
-    /**
-     * Search for similar vectors using cosine similarity.
-     *
-     * @param queryVector - Query embedding vector
-     * @param k - Number of results to return
-     * @returns Array of results with entity name and similarity score
-     */
-    search(queryVector: number[], k: number): VectorSearchResult[];
-    /**
-     * Remove a vector by entity name.
-     *
-     * @param entityName - Name of the entity to remove
-     * @returns True if found and removed
-     */
-    remove(entityName: string): boolean;
-    /**
-     * Get the number of vectors stored.
-     *
-     * @returns Number of vectors
-     */
-    size(): number;
-    /**
-     * Clear all vectors from the store.
-     */
-    clear(): void;
-    /**
-     * Check if a vector exists for an entity.
-     *
-     * @param entityName - Name of the entity
-     * @returns True if vector exists
-     */
-    has(entityName: string): boolean;
-    /**
-     * Get the vector for an entity.
-     *
-     * @param entityName - Name of the entity
-     * @returns Vector if found, undefined otherwise
-     */
-    get(entityName: string): number[] | undefined;
-    /**
-     * Set the SQLite storage reference.
-     *
-     * @param storage - SQLite storage instance
-     */
-    setStorage(storage: SQLiteStorageWithEmbeddings): void;
-    /**
-     * Set the embedding model name.
-     *
-     * @param model - Model name
-     */
-    setEmbeddingModel(model: string): void;
-}
-/**
- * Interface for SQLite storage with embedding support.
- *
- * This is a subset of SQLiteStorage that only includes embedding-related methods.
- * Allows for loose coupling between VectorStore and SQLiteStorage.
- */
-export interface SQLiteStorageWithEmbeddings {
-    /**
-     * Store an embedding for an entity.
-     *
-     * @param entityName - Name of the entity
-     * @param vector - Embedding vector
-     * @param model - Model name used for the embedding
-     */
-    storeEmbedding(entityName: string, vector: number[], model: string): void;
-    /**
-     * Load all embeddings from storage.
-     *
-     * @returns Array of [entityName, vector] pairs
-     */
-    loadAllEmbeddings(): Promise<[string, number[]][]>;
-    /**
-     * Remove an embedding for an entity.
-     *
-     * @param entityName - Name of the entity
-     */
-    removeEmbedding(entityName: string): void;
-    /**
-     * Clear all embeddings from storage.
-     */
-    clearAllEmbeddings(): void;
-}
-/**
- * Create a vector store based on storage type.
- *
- * @param storageType - Storage type: 'jsonl' or 'sqlite'
- * @param storage - Optional SQLite storage reference for 'sqlite' type
- * @param embeddingModel - Optional model name for embedding tracking
- * @returns Vector store instance
- */
-export declare function createVectorStore(storageType?: 'jsonl' | 'sqlite', storage?: SQLiteStorageWithEmbeddings, embeddingModel?: string): IVectorStore;
-//# sourceMappingURL=VectorStore.d.ts.map

package/dist/search/VectorStore.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"VectorStore.d.ts","sourceRoot":"","sources":["../../src/search/VectorStore.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,KAAK,EAAE,YAAY,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AAE1E;;;;;;;;GAQG;AACH,wBAAgB,gBAAgB,CAAC,CAAC,EAAE,MAAM,EAAE,EAAE,CAAC,EAAE,MAAM,EAAE,GAAG,MAAM,CA2BjE;AAED;;;;;;;;;;;;;;;GAeG;AACH,qBAAa,mBAAoB,YAAW,YAAY;IACtD,6CAA6C;IAC7C,OAAO,CAAC,OAAO,CAAoC;IAEnD;;;;;OAKG;IACH,GAAG,CAAC,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,IAAI;IAI/C;;;;;;OAMG;IACH,MAAM,CAAC,WAAW,EAAE,MAAM,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,kBAAkB,EAAE;IAuB9D;;;;;OAKG;IACH,MAAM,CAAC,UAAU,EAAE,MAAM,GAAG,OAAO;IAInC;;;;OAIG;IACH,IAAI,IAAI,MAAM;IAId;;OAEG;IACH,KAAK,IAAI,IAAI;IAIb;;;;;OAKG;IACH,GAAG,CAAC,UAAU,EAAE,MAAM,GAAG,OAAO;IAIhC;;;;;OAKG;IACH,GAAG,CAAC,UAAU,EAAE,MAAM,GAAG,MAAM,EAAE,GAAG,SAAS;IAI7C;;;;OAIG;IACH,cAAc,IAAI,MAAM,EAAE;IAI1B;;;;OAIG;IACH,QAAQ,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC,MAAM,EAAE,MAAM,EAAE,CAAC,CAAC,GAAG,IAAI;CAKtD;AAED;;;;;;;;;;;;;;;GAeG;AACH,qBAAa,iBAAkB,YAAW,YAAY;IACpD,sCAAsC;IACtC,OAAO,CAAC,WAAW,CAAkD;IAErE,+CAA+C;IAC/C,OAAO,CAAC,OAAO,CAA4C;IAE3D,6CAA6C;IAC7C,OAAO,CAAC,WAAW,CAAS;IAE5B,qCAAqC;IACrC,OAAO,CAAC,cAAc,CAAc;IAEpC;;;;;OAKG;gBACS,OAAO,CAAC,EAAE,2BAA2B,EAAE,cAAc,GAAE,MAAkB;IAKrF;;OAEG;IACG,UAAU,IAAI,OAAO,CAAC,IAAI,CAAC;IAYjC;;;;;;OAMG;IACH,GAAG,CAAC,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,IAAI;IAU/C;;;;;;OAMG;IACH,MAAM,CAAC,WAAW,EAAE,MAAM,EAAE,EAAE,CAAC,EAAE,MAAM,GAAG,kBAAkB,EAAE;IAI9D;;;;;OAKG;IACH,MAAM,CAAC,UAAU,EAAE,MAAM,GAAG,OAAO;IAWnC;;;;OAIG;IACH,IAAI,IAAI,MAAM;IAId;;OAEG;IACH,KAAK,IAAI,IAAI;IAQb;;;;;OAKG;IACH,GAAG,CAAC,UAAU,EAAE,MAAM,GAAG,OAAO;IAIhC;;;;;OAKG;IACH,GAAG,CAAC,UAAU,EAAE,MAAM,GAAG,MAAM,EAAE,GAAG,SAAS;IAI7C;;;;OAIG;IACH,UAAU,CAAC,OAAO,EAAE,2BAA2B,GAAG,IAAI;IAItD;;;;OAIG;IACH,iBAAiB,CAAC,KAAK,EAAE,MAAM,GAAG,IAAI;CAGvC;AAED;;;;;GAKG;AACH,MAAM,WAAW,2BAA2B;IAC1C;;;;;;OAMG;IACH,cAAc,CAAC,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,EAAE,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IAE1E;;;;OAIG;IACH,iBAAiB,IAAI,OAAO,CAAC,CAAC,MAAM,EAAE,MAAM,EAAE,CAAC,EAAE,CAAC,CAAC;IAEnD;;;;OAIG;IACH,eAAe,CAAC,UAAU,EAAE,MAAM,GAAG,IAAI,CAAC;IAE1C;;OAEG;IACH,kBAAkB,IAAI,IAAI,CAAC;CAC5B;AAED;;;;;;;GAOG;AACH,wBAAgB,iBAAiB,CAC/B,WAAW,GAAE,OAAO,GAAG,QAAkB,EACzC,OAAO,CAAC,EAAE,2BAA2B,EACrC,cAAc,CAAC,EAAE,MAAM,GACtB,YAAY,CAQd"}