add indexName and embeddingFieldPath

Author: mrlynn

src/core/IndexManager.js

@@ -3,110 +3,108 @@ import debug from 'debug';
 const log = debug('mongodb-rag:index');
 
 class IndexManager {
-  constructor(collection, options = {}) {
-    this.collection = collection;
-    this.options = {
-      indexName: options.indexName || 'default',
-      dimensions: options.dimensions || 1536,
-      similarity: options.similarity || 'cosine',
-      ...options
-    };
-  }
-
-  async ensureIndexes() {
-    try {
-      log("Checking indexes...");
-      await this.createSupportingIndexes(); // Ensure only supporting indexes
-      log("All indexes verified");
-    } catch (error) {
-      throw new Error(`Failed to ensure indexes: ${error.message}`);
+    constructor(collection, options = {}) {
+        this.collection = collection;
+        this.options = {
+            indexName: options.indexName || 'default',
+            dimensions: options.dimensions || 1536,
+            similarity: options.similarity || 'cosine',
+            ...options
+        };
     }
-  }
-
-  async createSupportingIndexes() {
-    try {
-      // Index for fast document lookups
-      await this.collection.createIndex(
-        { documentId: 1 },
-        { name: 'document_lookup' }
-      );
-
-      // Index for metadata filtering
-      await this.collection.createIndex(
-        { 'metadata.source': 1, 'metadata.category': 1 },
-        { name: 'metadata_filter', sparse: true }
-      );
-
-      log('Supporting indexes created successfully');
-    } catch (error) {
-      throw new Error(`Failed to create supporting indexes: ${error.message}`);
-    }
-  }
-
-  buildSearchQuery(embedding, filter = {}, options = {}) {
-    const vectorSearchQuery = {
-      index: this.options.indexName,
-      path: "embedding",
-      queryVector: embedding,
-      limit: options.maxResults || 10,
-      exact: options.exact || false  // Allow exact to be configurable
-    };
-  
-    // Only add numCandidates if exact is false
-    if (!vectorSearchQuery.exact) {
-      vectorSearchQuery.numCandidates = options.numCandidates || 100;
+
+    async ensureIndexes() {
+        try {
+            log("Checking indexes...");
+            await this.createSupportingIndexes(); // Ensure only supporting indexes
+            log("All indexes verified");
+        } catch (error) {
+            throw new Error(`Failed to ensure indexes: ${error.message}`);
+        }
     }
-  
-    const pipeline = [
-      {
-        $vectorSearch: vectorSearchQuery
-      },
-      {
-        $addFields: {
-          score: { $meta: "searchScore" }
+
+    async createSupportingIndexes() {
+        try {
+            // Index for fast document lookups
+            await this.collection.createIndex(
+                { documentId: 1 },
+                { name: 'document_lookup' }
+            );
+
+            // Index for metadata filtering
+            await this.collection.createIndex(
+                { 'metadata.source': 1, 'metadata.category': 1 },
+                { name: 'metadata_filter', sparse: true }
+            );
+
+            log('Supporting indexes created successfully');
+        } catch (error) {
+            throw new Error(`Failed to create supporting indexes: ${error.message}`);
         }
-      }
-    ];
-  
-    // Add filter if present
-    if (Object.keys(filter).length > 0) {
-      pipeline[0].$vectorSearch.filter = filter;
     }
-  
-    if (options.includeMetadata) {
-      pipeline.push({
-        $project: {
-          documentId: 1,
-          content: 1,
-          metadata: 1,
-          score: 1
+
+    buildSearchQuery(embedding, filter = {}, options = {}) {
+        const vectorSearchQuery = {
+            index: this.indexName,  // <-- Use configured index name
+            path: this.embeddingFieldPath,  // <-- Use configured embedding field path
+            queryVector: embedding,
+            limit: options.maxResults || 10,
+            exact: options.exact || false
+        };
+
+        if (!vectorSearchQuery.exact) {
+            vectorSearchQuery.numCandidates = options.numCandidates || 100;
+        }
+
+        const pipeline = [
+            {
+                $vectorSearch: vectorSearchQuery
+            },
+            {
+                $addFields: {
+                    score: { $meta: "searchScore" }
+                }
+            }
+        ];
+
+        if (Object.keys(filter).length > 0) {
+            pipeline[0].$vectorSearch.filter = filter;
         }
-      });
+
+        if (options.includeMetadata) {
+            pipeline.push({
+                $project: {
+                    documentId: 1,
+                    content: 1,
+                    metadata: 1,
+                    score: 1
+                }
+            });
+        }
+
+        console.log('Generated search query:', JSON.stringify(pipeline, null, 2));
+        return pipeline;
     }
-  
-    console.log('Generated search query:', JSON.stringify(pipeline, null, 2));
-    return pipeline;
-  }
-  
-
-  async getIndexStats() {
-    try {
-      const stats = await this.collection.stats();
-      const indexes = await this.collection.listIndexes().toArray();
-      
-      return {
-        documentCount: stats.count,
-        indexSize: stats.totalIndexSize,
-        indexes: indexes.map(index => ({
-          name: index.name,
-          size: stats.indexSizes[index.name],
-          fields: Object.keys(index.key)
-        }))
-      };
-    } catch (error) {
-      throw new Error(`Failed to get index stats: ${error.message}`);
+
+
+    async getIndexStats() {
+        try {
+            const stats = await this.collection.stats();
+            const indexes = await this.collection.listIndexes().toArray();
+
+            return {
+                documentCount: stats.count,
+                indexSize: stats.totalIndexSize,
+                indexes: indexes.map(index => ({
+                    name: index.name,
+                    size: stats.indexSizes[index.name],
+                    fields: Object.keys(index.key)
+                }))
+            };
+        } catch (error) {
+            throw new Error(`Failed to get index stats: ${error.message}`);
+        }
     }
-  }
 }
 
 export default IndexManager;

src/core/MongoRAG.js

@@ -37,6 +37,8 @@ class MongoRAG {
             mongoUrl: config.mongoUrl,
             defaultDatabase: config.database,
             defaultCollection: config.collection,
+            indexName: config.indexName || "vector_index",  // <-- Ensure we allow index configuration
+            embeddingFieldPath: config.embeddingFieldPath || "embedding", // <-- Allow setting embedding field path
             embedding: {
                 provider: config.embedding?.provider || 'openai',
                 apiKey: config.embedding.apiKey,
@@ -50,6 +52,7 @@ class MongoRAG {
                 maxResults: config.search?.maxResults || 5
             }
         };
+        
 
         this.client = null;
         this.indexManager = null;
@@ -135,6 +138,8 @@ class MongoRAG {
         const embedding = await this._getEmbedding(query);
 
         const indexManager = new IndexManager(col, {
+            indexName: this.config.indexName,  // <-- Pass configured index name
+            embeddingFieldPath: this.config.embeddingFieldPath,  // <-- Pass embedding field path
             dimensions: this.config.embedding.dimensions
         });
 
@@ -193,7 +198,7 @@ class MongoRAG {
      * @param {string} text - The text to embed.
      * @returns {Promise<Array<number>>} The embedded text.
      */
-    
+
     async _getEmbedding(text) {
         if (!this.embeddingProvider) {
             await this._initializeEmbeddingProvider();