@retrivora-ai/rag-engine 1.8.9 → 1.9.1

package/dist/server.d.ts

@@ -1,7 +1,7 @@
-import { n as VectorDBConfig, L as LLMConfig, e as EmbeddingConfig, k as RagConfig, I as ILLMProvider, g as IngestDocument, C as ChatMessage, d as ChatResponse, p as RetrievalResult, m as UpsertDocument, V as VectorMatch, G as GraphDBConfig, q as GraphNode, r as Edge, s as GraphSearchResult, o as VectorDBProvider, h as LLMProvider, f as EmbeddingProvider, j as RAGConfig, U as UIConfig, c as ChatOptions, E as EmbedOptions } from './ILLMProvider-CbUtvWAW.js';
+import { n as VectorDBConfig, L as LLMConfig, e as EmbeddingConfig, k as RagConfig, I as ILLMProvider, g as IngestDocument, C as ChatMessage, d as ChatResponse, p as RetrievalResult, m as UpsertDocument, V as VectorMatch, G as GraphDBConfig, q as GraphNode, r as Edge, s as GraphSearchResult, o as VectorDBProvider, h as LLMProvider, f as EmbeddingProvider, j as RAGConfig, U as UIConfig, c as ChatOptions, E as EmbedOptions } from './ILLMProvider-BQyKZLnd.js';
 export { C as Chunk, a as ChunkOptions, D as DocumentChunker } from './DocumentChunker-Dh9TvmGG.js';
-import { H as HealthCheckResult, I as IProviderValidator, a as IProviderHealthChecker } from './index-DgzcnqZs.js';
-export { C as ConfigValidator, V as ValidationError, b as VectorPlugin, c as createChatHandler, d as createHealthHandler, e as createIngestHandler, f as createStreamHandler, g as createUploadHandler, s as sseErrorFrame, h as sseFrame, i as sseMetaFrame, j as sseTextFrame } from './index-DgzcnqZs.js';
+import { H as HealthCheckResult, I as IProviderValidator, a as IProviderHealthChecker } from './index-A0GqPsaG.js';
+export { C as ConfigValidator, V as ValidationError, b as VectorPlugin, c as createChatHandler, d as createHealthHandler, e as createIngestHandler, f as createStreamHandler, g as createUploadHandler, s as sseErrorFrame, h as sseFrame, i as sseMetaFrame, j as sseTextFrame } from './index-A0GqPsaG.js';
 import 'next/server';
 
 /**
@@ -113,6 +113,7 @@ declare class Pipeline {
     private graphDB?;
     private llmProvider;
     private embeddingProvider;
+    private llmRouter;
     private chunker;
     private llamaIngestor?;
     private entityExtractor?;
@@ -150,6 +151,12 @@ declare class Pipeline {
     /**
      * High-performance streaming RAG flow.
      * Yields text chunks first, then the retrieval metadata + observability trace at the end.
+     *
+     * Latency optimizations:
+     * - Strategy classification runs in parallel with query embedding (saves ~400ms)
+     * - Hallucination scoring is fire-and-forget (doesn't block metadata yield)
+     * - UITransformation is computed after text streaming and emitted with metadata
+     * - SchemaMapper.train runs while answer generation streams
      */
     askStream(question: string, history?: ChatMessage[], namespace?: string): AsyncIterable<string | ChatResponse>;
     /**
@@ -157,6 +164,14 @@ declare class Pipeline {
      * Uses an LRU-bounded embedding cache to avoid re-embedding the same query.
      */
     private generateUiTransformation;
+    private applyStructuredFilters;
+    private resolveNumericPredicateValue;
+    private extractNumericValueFromContent;
+    private matchesNumericPredicate;
+    private normalizeComparableField;
+    private fieldSimilarityScore;
+    private fieldTokens;
+    private toFiniteNumber;
     retrieve(query: string, options: {
         namespace?: string;
         topK?: number;
@@ -811,8 +826,8 @@ declare class RedisProvider extends BaseVectorProvider {
     delete(id: string, namespace?: string): Promise<void>;
     deleteNamespace(namespace: string): Promise<void>;
     /**
-     * Redis is TCP-based and has no HTTP health endpoint.
-     * Returns true; actual connectivity is validated on the first operation.
+     * Check reachability via a PING command.
+     * Returns false on connection failure so health checks surface real problems.
      */
     ping(): Promise<boolean>;
     disconnect(): Promise<void>;
@@ -838,6 +853,9 @@ declare class WeaviateProvider extends BaseVectorProvider {
     deleteNamespace(namespace: string): Promise<void>;
     ping(): Promise<boolean>;
     disconnect(): Promise<void>;
+    private extractPrimitiveMetadata;
+    private buildWhereFilter;
+    private weaviateOperand;
 }
 
 /**
@@ -969,8 +987,17 @@ declare class UniversalLLMAdapter implements ILLMProvider {
     private readonly systemPrompt;
     private readonly maxTokens;
     private readonly temperature;
+    private readonly baseUrl;
+    private readonly apiKey?;
+    private readonly resolvedHeaders;
     constructor(config: LLMConfig | EmbeddingConfig);
     chat(messages: ChatMessage[], context?: string): Promise<string>;
+    /**
+     * Streaming chat using native fetch + ReadableStream.
+     * Parses OpenAI-compatible SSE frames: `data: {...}\n\n`
+     * Works with vLLM, LMStudio, Together AI, Fireworks, and any OpenAI-compatible API.
+     */
+    chatStream(messages: ChatMessage[], context?: string): AsyncIterable<string>;
     embed(text: string): Promise<number[]>;
     batchEmbed(texts: string[]): Promise<number[][]>;
     ping(): Promise<boolean>;