@soulcraft/brainy 5.6.2 → 5.7.0

package/dist/import/BackgroundDeduplicator.js

@@ -0,0 +1,359 @@
+/**
+ * Background Deduplicator
+ *
+ * Performs 3-tier entity deduplication in background after imports:
+ * - Tier 1: ID-based (O(1)) - Uses entity metadata for deterministic IDs
+ * - Tier 2: Name-based (O(log n)) - Exact name matching (case-insensitive)
+ * - Tier 3: Similarity-based (O(n log n)) - Vector similarity via TypeAware HNSW
+ *
+ * NO MOCKS - Production-ready implementation using existing indexes
+ */
+import { prodLog } from '../utils/logger.js';
+/**
+ * BackgroundDeduplicator - Auto-runs deduplication 5 minutes after imports
+ *
+ * Architecture:
+ * - Debounced trigger (5 min after last import)
+ * - Import-scoped deduplication (no cross-contamination)
+ * - 3-tier strategy (ID → Name → Similarity)
+ * - Uses existing indexes (EntityIdMapper, MetadataIndexManager, TypeAware HNSW)
+ */
+export class BackgroundDeduplicator {
+    constructor(brain) {
+        this.pendingImports = new Set();
+        this.isProcessing = false;
+        this.brain = brain;
+    }
+    /**
+     * Schedule deduplication for an import (debounced 5 minutes)
+     * Called by ImportCoordinator after each import completes
+     */
+    scheduleDedup(importId) {
+        prodLog.info(`[BackgroundDedup] Scheduled deduplication for import ${importId}`);
+        // Add to pending queue
+        this.pendingImports.add(importId);
+        // Clear existing timer (debouncing)
+        if (this.debounceTimer) {
+            clearTimeout(this.debounceTimer);
+        }
+        // Schedule for 5 minutes from now
+        this.debounceTimer = setTimeout(() => {
+            this.runBatchDedup().catch(error => {
+                prodLog.error('[BackgroundDedup] Batch dedup failed:', error);
+            });
+        }, 5 * 60 * 1000);
+    }
+    /**
+     * Run deduplication for all pending imports
+     * @private
+     */
+    async runBatchDedup() {
+        if (this.isProcessing) {
+            prodLog.warn('[BackgroundDedup] Already processing, skipping');
+            return;
+        }
+        this.isProcessing = true;
+        try {
+            const imports = Array.from(this.pendingImports);
+            prodLog.info(`[BackgroundDedup] Processing ${imports.length} pending import(s)`);
+            for (const importId of imports) {
+                await this.deduplicateImport(importId);
+            }
+            this.pendingImports.clear();
+            prodLog.info('[BackgroundDedup] Batch deduplication complete');
+        }
+        finally {
+            this.isProcessing = false;
+        }
+    }
+    /**
+     * Deduplicate entities from a specific import
+     * Uses 3-tier strategy: ID → Name → Similarity
+     */
+    async deduplicateImport(importId) {
+        const startTime = performance.now();
+        prodLog.info(`[BackgroundDedup] Starting deduplication for import ${importId}`);
+        const stats = {
+            totalEntities: 0,
+            tier1Matches: 0,
+            tier2Matches: 0,
+            tier3Matches: 0,
+            totalMerged: 0,
+            processingTime: 0
+        };
+        try {
+            // Get all entities from this import using brain.find()
+            const results = await this.brain.find({
+                where: { importId },
+                limit: 100000 // Large limit to get all entities from import
+            });
+            const entities = results.map(r => r.entity);
+            stats.totalEntities = entities.length;
+            if (entities.length === 0) {
+                prodLog.info(`[BackgroundDedup] No entities found for import ${importId}`);
+                return stats;
+            }
+            prodLog.info(`[BackgroundDedup] Processing ${entities.length} entities from import ${importId}`);
+            // Tier 1: ID-based deduplication (O(1) per entity)
+            const tier1Merged = await this.tier1_IdBased(entities, importId);
+            stats.tier1Matches = tier1Merged;
+            stats.totalMerged += tier1Merged;
+            // Re-check which entities still exist after Tier 1
+            let remainingEntities = entities;
+            if (tier1Merged > 0) {
+                remainingEntities = await this.filterExisting(entities);
+                prodLog.info(`[BackgroundDedup] After Tier 1: ${entities.length} → ${remainingEntities.length} entities`);
+            }
+            // Tier 2: Name-based deduplication on reduced set
+            const tier2Merged = await this.tier2_NameBased(remainingEntities, importId);
+            stats.tier2Matches = tier2Merged;
+            stats.totalMerged += tier2Merged;
+            // Re-check which entities still exist after Tier 2
+            if (tier2Merged > 0) {
+                remainingEntities = await this.filterExisting(remainingEntities);
+                prodLog.info(`[BackgroundDedup] After Tier 2: ${remainingEntities.length} entities remaining`);
+            }
+            // Tier 3: Similarity-based deduplication on final reduced set
+            const tier3Merged = await this.tier3_SimilarityBased(remainingEntities, importId);
+            stats.tier3Matches = tier3Merged;
+            stats.totalMerged += tier3Merged;
+            stats.processingTime = performance.now() - startTime;
+            prodLog.info(`[BackgroundDedup] Completed for import ${importId}: ` +
+                `${stats.totalMerged} merged (T1: ${stats.tier1Matches}, T2: ${stats.tier2Matches}, T3: ${stats.tier3Matches}) ` +
+                `in ${stats.processingTime.toFixed(0)}ms`);
+            return stats;
+        }
+        catch (error) {
+            prodLog.error(`[BackgroundDedup] Error deduplicating import ${importId}:`, error);
+            stats.processingTime = performance.now() - startTime;
+            return stats;
+        }
+    }
+    /**
+     * Tier 1: ID-based deduplication
+     * Uses entity metadata sourceId field for deterministic matching
+     * Complexity: O(n) where n = number of entities in import
+     */
+    async tier1_IdBased(entities, importId) {
+        const startTime = performance.now();
+        let merged = 0;
+        // Group entities by sourceId (if available)
+        const sourceIdGroups = new Map();
+        for (const entity of entities) {
+            const sourceId = entity.metadata?.sourceId || entity.metadata?.sourceRow;
+            if (sourceId) {
+                const key = `${sourceId}`;
+                if (!sourceIdGroups.has(key)) {
+                    sourceIdGroups.set(key, []);
+                }
+                sourceIdGroups.get(key).push(entity);
+            }
+        }
+        // Merge duplicates with same sourceId
+        for (const [sourceId, group] of sourceIdGroups) {
+            if (group.length > 1) {
+                await this.mergeEntities(group, 'ID');
+                merged += group.length - 1;
+            }
+        }
+        const elapsed = performance.now() - startTime;
+        if (merged > 0) {
+            prodLog.info(`[BackgroundDedup] Tier 1 (ID): Merged ${merged} duplicates in ${elapsed.toFixed(0)}ms`);
+        }
+        return merged;
+    }
+    /**
+     * Tier 2: Name-based deduplication
+     * Exact name matching (case-insensitive, normalized)
+     * Complexity: O(n) where n = number of entities in import
+     */
+    async tier2_NameBased(entities, importId) {
+        const startTime = performance.now();
+        let merged = 0;
+        // Group entities by normalized name
+        const nameGroups = new Map();
+        for (const entity of entities) {
+            const name = entity.metadata?.name;
+            if (name && typeof name === 'string') {
+                const normalized = this.normalizeName(name);
+                if (!nameGroups.has(normalized)) {
+                    nameGroups.set(normalized, []);
+                }
+                nameGroups.get(normalized).push(entity);
+            }
+        }
+        // Merge duplicates with same normalized name and type
+        for (const [name, group] of nameGroups) {
+            if (group.length > 1) {
+                // Further group by type (only merge same types)
+                const typeGroups = new Map();
+                for (const entity of group) {
+                    const type = entity.type || 'unknown';
+                    if (!typeGroups.has(type)) {
+                        typeGroups.set(type, []);
+                    }
+                    typeGroups.get(type).push(entity);
+                }
+                // Merge within each type group
+                for (const [type, typeGroup] of typeGroups) {
+                    if (typeGroup.length > 1) {
+                        await this.mergeEntities(typeGroup, 'Name');
+                        merged += typeGroup.length - 1;
+                    }
+                }
+            }
+        }
+        const elapsed = performance.now() - startTime;
+        if (merged > 0) {
+            prodLog.info(`[BackgroundDedup] Tier 2 (Name): Merged ${merged} duplicates in ${elapsed.toFixed(0)}ms`);
+        }
+        return merged;
+    }
+    /**
+     * Tier 3: Similarity-based deduplication
+     * Uses TypeAware HNSW for vector similarity matching
+     * Complexity: O(n log n) where n = number of entities in import
+     */
+    async tier3_SimilarityBased(entities, importId) {
+        const startTime = performance.now();
+        let merged = 0;
+        // Process in batches to avoid memory spikes
+        const batchSize = 100;
+        const similarityThreshold = 0.85;
+        for (let i = 0; i < entities.length; i += batchSize) {
+            const batch = entities.slice(i, i + batchSize);
+            // Batch vector searches using brain.find() (uses TypeAware HNSW)
+            const searches = batch.map(entity => {
+                const query = `${entity.metadata?.name || ''} ${entity.metadata?.description || ''}`.trim();
+                if (!query)
+                    return Promise.resolve([]);
+                return this.brain.find({
+                    query,
+                    limit: 5,
+                    where: { type: entity.type } // Type-aware search
+                });
+            });
+            const results = await Promise.all(searches);
+            // Process matches
+            for (let j = 0; j < batch.length; j++) {
+                const entity = batch[j];
+                const matches = results[j];
+                for (const match of matches) {
+                    // Skip self-matches
+                    if (match.id === entity.id)
+                        continue;
+                    // Only merge high-similarity matches from same import
+                    if (match.score >= similarityThreshold && match.entity.metadata?.importId === importId) {
+                        // Check if not already merged
+                        const stillExists = await this.brain.get(entity.id);
+                        if (stillExists) {
+                            // Cast match.entity to HNSWNounWithMetadata (it comes from brain.find results)
+                            const matchEntity = match.entity;
+                            await this.mergeEntities([entity, matchEntity], 'Similarity');
+                            merged++;
+                            break; // Only merge with first high-similarity match
+                        }
+                    }
+                }
+            }
+        }
+        const elapsed = performance.now() - startTime;
+        if (merged > 0) {
+            prodLog.info(`[BackgroundDedup] Tier 3 (Similarity): Merged ${merged} duplicates in ${elapsed.toFixed(0)}ms`);
+        }
+        return merged;
+    }
+    /**
+     * Merge multiple entities into one
+     * Keeps entity with highest confidence, merges metadata, deletes duplicates
+     */
+    async mergeEntities(entities, reason) {
+        if (entities.length < 2)
+            return;
+        // Find entity with highest confidence
+        const primary = entities.reduce((best, curr) => {
+            const bestConf = best.metadata?.confidence || 0.5;
+            const currConf = curr.metadata?.confidence || 0.5;
+            return currConf > bestConf ? curr : best;
+        });
+        // Merge metadata from all entities
+        const primaryMeta = primary.metadata || {};
+        const mergedMetadata = {
+            ...primaryMeta,
+            // Merge import IDs
+            importIds: Array.from(new Set([
+                ...(Array.isArray(primaryMeta.importIds) ? primaryMeta.importIds : []),
+                ...entities.flatMap(e => Array.isArray(e.metadata?.importIds) ? e.metadata.importIds : [])
+            ])),
+            // Merge VFS paths
+            vfsPaths: Array.from(new Set([
+                ...(Array.isArray(primaryMeta.vfsPaths) ? primaryMeta.vfsPaths : []),
+                ...entities.flatMap(e => Array.isArray(e.metadata?.vfsPaths) ? e.metadata.vfsPaths : [])
+            ])),
+            // Merge concepts
+            concepts: Array.from(new Set([
+                ...(Array.isArray(primaryMeta.concepts) ? primaryMeta.concepts : []),
+                ...entities.flatMap(e => Array.isArray(e.metadata?.concepts) ? e.metadata.concepts : [])
+            ])),
+            // Track merge
+            mergeCount: (typeof primaryMeta.mergeCount === 'number' ? primaryMeta.mergeCount : 0) + (entities.length - 1),
+            mergedWith: entities.filter(e => e.id !== primary.id).map(e => e.id),
+            lastMerged: Date.now(),
+            mergeReason: reason
+        };
+        // Update primary entity with merged metadata
+        await this.brain.update({
+            id: primary.id,
+            metadata: mergedMetadata,
+            merge: true
+        });
+        // Delete duplicate entities
+        for (const entity of entities) {
+            if (entity.id !== primary.id) {
+                try {
+                    await this.brain.delete(entity.id);
+                }
+                catch (error) {
+                    // Entity might already be deleted, continue
+                    prodLog.debug(`[BackgroundDedup] Could not delete ${entity.id}:`, error);
+                }
+            }
+        }
+    }
+    /**
+     * Filter entities to only those that still exist (not deleted)
+     * @private
+     */
+    async filterExisting(entities) {
+        const existing = [];
+        for (const entity of entities) {
+            const stillExists = await this.brain.get(entity.id);
+            if (stillExists) {
+                existing.push(entity);
+            }
+        }
+        return existing;
+    }
+    /**
+     * Normalize string for comparison
+     * Lowercase, trim, remove special characters
+     */
+    normalizeName(str) {
+        return str
+            .toLowerCase()
+            .trim()
+            .replace(/[^a-z0-9\s]/g, '')
+            .replace(/\s+/g, ' ');
+    }
+    /**
+     * Cancel pending deduplication (for cleanup)
+     */
+    cancelPending() {
+        if (this.debounceTimer) {
+            clearTimeout(this.debounceTimer);
+            this.debounceTimer = undefined;
+        }
+        this.pendingImports.clear();
+    }
+}
+//# sourceMappingURL=BackgroundDeduplicator.js.map

package/dist/import/ImportCoordinator.d.ts

@@ -248,8 +248,8 @@ export interface ImportResult {
 export declare class ImportCoordinator {
     private brain;
     private detector;
-    private deduplicator;
     private history;
+    private backgroundDedup;
     private excelImporter;
     private pdfImporter;
     private csvImporter;

package/dist/import/ImportCoordinator.js

@@ -10,8 +10,8 @@
  * NO MOCKS - Production-ready implementation
  */
 import { FormatDetector } from './FormatDetector.js';
-import { EntityDeduplicator } from './EntityDeduplicator.js';
 import { ImportHistory } from './ImportHistory.js';
+import { BackgroundDeduplicator } from './BackgroundDeduplicator.js';
 import { SmartExcelImporter } from '../importers/SmartExcelImporter.js';
 import { SmartPDFImporter } from '../importers/SmartPDFImporter.js';
 import { SmartCSVImporter } from '../importers/SmartCSVImporter.js';
@@ -31,8 +31,8 @@ export class ImportCoordinator {
     constructor(brain) {
         this.brain = brain;
         this.detector = new FormatDetector();
-        this.deduplicator = new EntityDeduplicator(brain);
         this.history = new ImportHistory(brain);
+        this.backgroundDedup = new BackgroundDeduplicator(brain);
         this.excelImporter = new SmartExcelImporter(brain);
         this.pdfImporter = new SmartPDFImporter(brain);
         this.csvImporter = new SmartCSVImporter(brain);
@@ -683,20 +683,20 @@ export class ImportCoordinator {
                 try {
                     const importSource = vfsResult.rootPath;
                     let entityId;
-                    let wasMerged = false;
-                    // Use deduplicator to check for existing entities
-                    const mergeResult = await this.deduplicator.createOrMerge({
-                        id: entity.id,
-                        name: entity.name,
+                    // v5.7.0: No deduplication during import (12-24x speedup)
+                    // Background deduplication runs 5 minutes after import completes
+                    entityId = await this.brain.add({
+                        data: entity.description || entity.name,
                         type: entity.type,
-                        description: entity.description || entity.name,
-                        confidence: entity.confidence,
                         metadata: {
                             ...entity.metadata,
+                            name: entity.name,
+                            confidence: entity.confidence,
                             vfsPath: vfsFile?.path,
                             importedFrom: 'import-coordinator',
                             // v4.10.0: Import tracking metadata
                             ...(trackingContext && {
+                                importId: trackingContext.importId, // Used for background dedup
                                 importIds: [trackingContext.importId],
                                 projectId: trackingContext.projectId,
                                 importedAt: trackingContext.importedAt,
@@ -707,19 +707,8 @@ export class ImportCoordinator {
                                 ...trackingContext.customMetadata
                             })
                         }
-                    }, importSource, {
-                        similarityThreshold: options.deduplicationThreshold || 0.85,
-                        strictTypeMatching: true,
-                        enableFuzzyMatching: true
                     });
-                    entityId = mergeResult.mergedEntityId;
-                    wasMerged = mergeResult.wasMerged;
-                    if (wasMerged) {
-                        mergedCount++;
-                    }
-                    else {
-                        newCount++;
-                    }
+                    newCount++;
                     // Update entity ID in extraction result
                     entity.id = entityId;
                     entities.push({
@@ -943,6 +932,10 @@ export class ImportCoordinator {
                 // Continue - relationships are optional
             }
         }
+        // v5.7.0: Schedule background deduplication (debounced 5 minutes)
+        if (trackingContext && trackingContext.importId) {
+            this.backgroundDedup.scheduleDedup(trackingContext.importId);
+        }
         return {
             entities,
             relationships,

package/dist/import/index.d.ts

@@ -10,7 +10,9 @@
 export { ImportCoordinator } from './ImportCoordinator.js';
 export { FormatDetector, SupportedFormat, DetectionResult } from './FormatDetector.js';
 export { EntityDeduplicator } from './EntityDeduplicator.js';
+export { BackgroundDeduplicator } from './BackgroundDeduplicator.js';
 export { ImportHistory } from './ImportHistory.js';
 export type { ImportSource, ImportOptions, ImportProgress, ImportResult } from './ImportCoordinator.js';
 export type { EntityCandidate, DuplicateMatch, EntityDeduplicationOptions, MergeResult } from './EntityDeduplicator.js';
+export type { DeduplicationStats } from './BackgroundDeduplicator.js';
 export type { ImportHistoryEntry, RollbackResult } from './ImportHistory.js';

package/dist/import/index.js

@@ -10,5 +10,6 @@
 export { ImportCoordinator } from './ImportCoordinator.js';
 export { FormatDetector } from './FormatDetector.js';
 export { EntityDeduplicator } from './EntityDeduplicator.js';
+export { BackgroundDeduplicator } from './BackgroundDeduplicator.js';
 export { ImportHistory } from './ImportHistory.js';
 //# sourceMappingURL=index.js.map

package/dist/storage/baseStorage.d.ts

@@ -51,6 +51,7 @@ export declare function getDirectoryPath(entityType: 'noun' | 'verb', dataType:
 export declare abstract class BaseStorage extends BaseStorageAdapter {
     protected isInitialized: boolean;
     protected graphIndex?: GraphAdjacencyIndex;
+    protected graphIndexPromise?: Promise<GraphAdjacencyIndex>;
     protected readOnly: boolean;
     refManager?: RefManager;
     blobStorage?: BlobStorage;
@@ -311,9 +312,15 @@ export declare abstract class BaseStorage extends BaseStorageAdapter {
      */
     deleteVerb(id: string): Promise<void>;
     /**
-     * Get graph index (lazy initialization)
+     * Get graph index (lazy initialization with concurrent access protection)
+     * v5.7.1: Fixed race condition where concurrent calls could trigger multiple rebuilds
      */
     getGraphIndex(): Promise<GraphAdjacencyIndex>;
+    /**
+     * Internal method to initialize graph index (called once by getGraphIndex)
+     * @private
+     */
+    private _initializeGraphIndex;
     /**
      * Clear all data from storage
      * This method should be implemented by each specific adapter
@@ -481,7 +488,7 @@ export declare abstract class BaseStorage extends BaseStorageAdapter {
     protected getVerbsBySource_internal(sourceId: string): Promise<HNSWVerbWithMetadata[]>;
     /**
      * Get verbs by target (COW-aware implementation)
-     * v5.4.0: Fixed to directly list verb files instead of directories
+     * v5.7.0: BILLION-SCALE OPTIMIZATION - Use GraphAdjacencyIndex for O(log n) lookup
      */
     protected getVerbsByTarget_internal(targetId: string): Promise<HNSWVerbWithMetadata[]>;
     /**