@rws-framework/ai-tools 3.2.0 → 3.2.2

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@rws-framework/ai-tools",
   "private": false,
-  "version": "3.2.0",
+  "version": "3.2.2",
   "description": "",
   "main": "src/index.ts",
   "scripts": {},
@@ -16,7 +16,8 @@
     "@rws-framework/console": "*",
     "uuid": "^9.0.0",
     "xml2js": "^0.6.2",
-    "zod": "^3.25.76"
+    "zod": "^3.25.76",
+    "tiktoken": "^1.0.22"
   },
   "devDependencies": {
     "@types/xml2js": "^0.4.14"

package/src/services/LangChainRAGService.ts

@@ -161,7 +161,7 @@ export class LangChainRAGService {
             const knowledgeVectorPromises = knowledgeIds.map(async (knowledgeId) => {
                 const vectorData = await this.loadKnowledgeVectorWithEmbeddings(knowledgeId);
                 return {
-                    knowledgeId,
+                    knowledgeId,                    
                     chunks: vectorData.chunks
                 };
             });
@@ -178,13 +178,14 @@ export class LangChainRAGService {
             
             // Convert results to expected format
             const results: ISearchResult[] = searchResponse.results.map(result => ({
+                knowledgeId: result.metadata.knowledgeId,
                 content: result.content,
                 score: result.score,
                 metadata: result.metadata,
-                chunkId: result.chunkId
+                chunkId: result.chunkId,                
             }));
 
-            this.log('log', `[SEARCH] Found ${results.length} relevant chunks for query: "${request.query}"`);
+            this.log('log', `[SEARCH] Found ${results.length} relevant chunks for query: "${request.query}"\n ${JSON.stringify({ results})}`);
 
             return {
                 success: true,
@@ -334,7 +335,7 @@ export class LangChainRAGService {
     /**
      * Load vector data for a specific knowledge item with embeddings
      */
-    private async loadKnowledgeVectorWithEmbeddings(knowledgeId: string | number): Promise<{ chunks: Array<{ content: string; embedding: number[]; metadata: any }> }> {
+    private async loadKnowledgeVectorWithEmbeddings(knowledgeId: string | number): Promise<{ knowledgeId?: string | number, chunks: Array<{ content: string; embedding: number[]; metadata: any }> }> {
         const vectorFilePath = this.getKnowledgeVectorPath(knowledgeId);
         
         if (!fs.existsSync(vectorFilePath)) {
@@ -347,7 +348,8 @@ export class LangChainRAGService {
             const vectorData = JSON.parse(fs.readFileSync(vectorFilePath, 'utf8'));
             
             return {
-                chunks: vectorData.chunks || []
+                chunks: vectorData.chunks || [],
+                knowledgeId
             };
         } catch (error) {
             this.log('error', `[LOAD] Failed to load vector data for knowledge ${knowledgeId}:`, error);

package/src/services/OptimizedVectorSearchService.ts

@@ -292,7 +292,8 @@ export class OptimizedVectorSearchService {
                 content: result.content,
                 score: result.score,
                 metadata: result.metadata,
-                chunkId: result.chunkId
+                chunkId: result.chunkId,
+                knowledgeId: result.knowledgeId
             }));
 
             return {

package/src/types/search.types.ts

@@ -5,6 +5,7 @@ export interface ISearchResult {
     content: string;
     score: number;
     metadata: any;
+    knowledgeId: string | number;
     chunkId: string;
 }
 
@@ -12,8 +13,7 @@ export interface IVectorSearchRequest {
     query: string;
     maxResults?: number;
     similarityThreshold?: number;
-    filter?: {
-        knowledgeIds?: string[];
+    filter?: {        
         documentIds?: string[];
         [key: string]: any;
     };