@retrivora-ai/rag-engine 1.0.0 → 1.0.2

package/dist/server.d.mts

@@ -2,8 +2,8 @@ import { e as VectorDBConfig, L as LLMConfig, E as EmbeddingConfig, c as RagConf
 export { R as RAGConfig, U as UIConfig } from './RagConfig-DRJO4hGU.mjs';
 import { C as ChatMessage, I as ILLMProvider, a as ChatOptions, E as EmbedOptions } from './DocumentChunker-C-sCZPhi.mjs';
 export { b as Chunk, c as ChunkOptions, D as DocumentChunker } from './DocumentChunker-C-sCZPhi.mjs';
-import { H as HealthCheckResult, I as IProviderValidator, a as IProviderHealthChecker } from './index-v669iV-k.mjs';
-export { C as ConfigValidator, V as ValidationError, c as createChatHandler, b as createHealthHandler, d as createIngestHandler, e as createUploadHandler } from './index-v669iV-k.mjs';
+import { H as HealthCheckResult, I as IProviderValidator, a as IProviderHealthChecker } from './index-Bjy0es5a.mjs';
+export { C as ConfigValidator, V as ValidationError, c as createChatHandler, b as createHealthHandler, d as createIngestHandler, e as createUploadHandler } from './index-Bjy0es5a.mjs';
 import 'next/server';
 
 /**
@@ -111,26 +111,7 @@ declare class ProviderHealthCheck {
 declare class VectorPlugin {
     private config;
     private pipeline;
-    /**
-     * Initializes the plugin with the host configuration.
-     * @param hostConfig - Configuration object passed from the host application.
-     * @throws Error if configuration is invalid or providers are unhealthy
-     *
-     * @example
-     * const plugin = new VectorPlugin({
-     *   projectId: 'my-app',
-     *   vectorDb: {
-     *     provider: 'pinecone',
-     *     indexName: 'my-index',
-     *     options: { apiKey: process.env.PINECONE_API_KEY }
-     *   },
-     *   llm: {
-     *     provider: 'openai',
-     *     model: 'gpt-4o',
-     *     apiKey: process.env.OPENAI_API_KEY
-     *   }
-     * });
-     */
+    private validationPromise;
     constructor(hostConfig?: Partial<RagConfig>);
     /**
      * Get the current resolved configuration.
@@ -152,6 +133,10 @@ declare class VectorPlugin {
      * Run a chat query.
      */
     chat(message: string, history?: ChatMessage[], namespace?: string): Promise<ChatResponse>;
+    /**
+     * Run a streaming chat query.
+     */
+    chatStream(message: string, history?: ChatMessage[], namespace?: string): AsyncGenerator<string | ChatResponse, void, any>;
     /**
      * Ingest documents into the vector database.
      */

package/dist/server.d.ts

@@ -2,8 +2,8 @@ import { e as VectorDBConfig, L as LLMConfig, E as EmbeddingConfig, c as RagConf
 export { R as RAGConfig, U as UIConfig } from './RagConfig-DRJO4hGU.js';
 import { C as ChatMessage, I as ILLMProvider, a as ChatOptions, E as EmbedOptions } from './DocumentChunker-C-sCZPhi.js';
 export { b as Chunk, c as ChunkOptions, D as DocumentChunker } from './DocumentChunker-C-sCZPhi.js';
-import { H as HealthCheckResult, I as IProviderValidator, a as IProviderHealthChecker } from './index-CrGMwXfO.js';
-export { C as ConfigValidator, V as ValidationError, c as createChatHandler, b as createHealthHandler, d as createIngestHandler, e as createUploadHandler } from './index-CrGMwXfO.js';
+import { H as HealthCheckResult, I as IProviderValidator, a as IProviderHealthChecker } from './index-B2mutkgp.js';
+export { C as ConfigValidator, V as ValidationError, c as createChatHandler, b as createHealthHandler, d as createIngestHandler, e as createUploadHandler } from './index-B2mutkgp.js';
 import 'next/server';
 
 /**
@@ -111,26 +111,7 @@ declare class ProviderHealthCheck {
 declare class VectorPlugin {
     private config;
     private pipeline;
-    /**
-     * Initializes the plugin with the host configuration.
-     * @param hostConfig - Configuration object passed from the host application.
-     * @throws Error if configuration is invalid or providers are unhealthy
-     *
-     * @example
-     * const plugin = new VectorPlugin({
-     *   projectId: 'my-app',
-     *   vectorDb: {
-     *     provider: 'pinecone',
-     *     indexName: 'my-index',
-     *     options: { apiKey: process.env.PINECONE_API_KEY }
-     *   },
-     *   llm: {
-     *     provider: 'openai',
-     *     model: 'gpt-4o',
-     *     apiKey: process.env.OPENAI_API_KEY
-     *   }
-     * });
-     */
+    private validationPromise;
     constructor(hostConfig?: Partial<RagConfig>);
     /**
      * Get the current resolved configuration.
@@ -152,6 +133,10 @@ declare class VectorPlugin {
      * Run a chat query.
      */
     chat(message: string, history?: ChatMessage[], namespace?: string): Promise<ChatResponse>;
+    /**
+     * Run a streaming chat query.
+     */
+    chatStream(message: string, history?: ChatMessage[], namespace?: string): AsyncGenerator<string | ChatResponse, void, any>;
     /**
      * Ingest documents into the vector database.
      */

package/dist/server.js

@@ -10,6 +10,9 @@ var __getProtoOf = Object.getPrototypeOf;
 var __hasOwnProp = Object.prototype.hasOwnProperty;
 var __propIsEnum = Object.prototype.propertyIsEnumerable;
 var __knownSymbol = (name, symbol) => (symbol = Symbol[name]) ? symbol : /* @__PURE__ */ Symbol.for("Symbol." + name);
+var __typeError = (msg) => {
+  throw TypeError(msg);
+};
 var __defNormalProp = (obj, key, value) => key in obj ? __defProp(obj, key, { enumerable: true, configurable: true, writable: true, value }) : obj[key] = value;
 var __spreadValues = (a, b) => {
   for (var prop in b || (b = {}))
@@ -74,6 +77,34 @@ var __asyncGenerator = (__this, __arguments, generator) => {
   }, method = (k, call, wait, clear) => it[k] = (x) => (call = new Promise((yes, no, run) => (run = () => resume(k, x, yes, no), q ? q.then(run) : run())), clear = () => q === wait && (q = 0), q = wait = call.then(clear, clear), call), q, it = {};
   return generator = generator.apply(__this, __arguments), it[__knownSymbol("asyncIterator")] = () => it, method("next"), method("throw"), method("return"), it;
 };
+var __yieldStar = (value) => {
+  var obj = value[__knownSymbol("asyncIterator")], isAwait = false, method, it = {};
+  if (obj == null) {
+    obj = value[__knownSymbol("iterator")]();
+    method = (k) => it[k] = (x) => obj[k](x);
+  } else {
+    obj = obj.call(value);
+    method = (k) => it[k] = (v) => {
+      if (isAwait) {
+        isAwait = false;
+        if (k === "throw") throw v;
+        return v;
+      }
+      isAwait = true;
+      return {
+        done: false,
+        value: new __await(new Promise((resolve) => {
+          var x = obj[k](v);
+          if (!(x instanceof Object)) __typeError("Object expected");
+          resolve(x);
+        }), 1)
+      };
+    };
+  }
+  return it[__knownSymbol("iterator")] = () => it, method("next"), "throw" in obj ? method("throw") : it.throw = (x) => {
+    throw x;
+  }, "return" in obj && method("return"), it;
+};
 var __forAwait = (obj, it, method) => (it = obj[__knownSymbol("asyncIterator")]) ? it.call(obj) : (obj = obj[__knownSymbol("iterator")](), it = {}, method = (key, fn) => (fn = obj[key]) && (it[key] = (arg) => new Promise((yes, no, done) => (arg = fn.call(obj, arg), done = arg.done, Promise.resolve(arg.value).then((value) => yes({ value, done }), no)))), method("next"), method("return"), it);
 
 // src/utils/templateUtils.ts
@@ -527,7 +558,7 @@ var init_MongoDBProvider = __esm({
         if (!opts.uri || !opts.database || !opts.collection) {
           throw new Error("[MongoDBProvider] options uri, database, and collection are required.");
         }
-        this.client = new import_mongodb.MongoClient(opts.uri);
+        this.client = new import_mongodb.MongoClient(opts.uri, { serverSelectionTimeoutMS: 5e3 });
         this.dbName = opts.database;
         this.collectionName = opts.collection;
         this.embeddingKey = opts.embeddingKey || "embedding";
@@ -2908,6 +2939,9 @@ Extract entities and relationships from the following text.
 Format the output as JSON with two keys: "nodes" (array of {id, label, properties}) and "edges" (array of {source, target, type, properties}).
 Use the same ID for the same entity.
 
+IMPORTANT: Ensure all property values are valid JSON types (strings, numbers, booleans, or null). 
+DO NOT include mathematical expressions like "4.5/5" as numbers; use strings instead.
+
 Text:
 "${text}"
 
@@ -3488,6 +3522,7 @@ var Pipeline = class {
   async initialize() {
     var _a;
     if (this.initialised) return;
+    console.log(`[Pipeline] Initializing with provider: ${this.config.vectorDb.provider}...`);
     this.vectorDB = await ProviderRegistry.createVectorProvider(this.config.vectorDb);
     const { llmProvider, embeddingProvider } = await EmbeddingStrategyResolver.resolve(
       this.config.llm,
@@ -3872,29 +3907,9 @@ var ProviderHealthCheck = class {
 
 // src/core/VectorPlugin.ts
 var VectorPlugin = class {
-  /**
-   * Initializes the plugin with the host configuration.
-   * @param hostConfig - Configuration object passed from the host application.
-   * @throws Error if configuration is invalid or providers are unhealthy
-   *
-   * @example
-   * const plugin = new VectorPlugin({
-   *   projectId: 'my-app',
-   *   vectorDb: {
-   *     provider: 'pinecone',
-   *     indexName: 'my-index',
-   *     options: { apiKey: process.env.PINECONE_API_KEY }
-   *   },
-   *   llm: {
-   *     provider: 'openai',
-   *     model: 'gpt-4o',
-   *     apiKey: process.env.OPENAI_API_KEY
-   *   }
-   * });
-   */
   constructor(hostConfig) {
     this.config = ConfigResolver.resolve(hostConfig);
-    ConfigValidator.validateAndThrow(this.config);
+    this.validationPromise = ConfigValidator.validateAndThrow(this.config);
     this.pipeline = new Pipeline(this.config);
   }
   /**
@@ -3920,12 +3935,23 @@ var VectorPlugin = class {
    * Run a chat query.
    */
   async chat(message, history = [], namespace) {
+    await this.validationPromise;
     return this.pipeline.ask(message, history, namespace);
   }
+  /**
+   * Run a streaming chat query.
+   */
+  chatStream(_0) {
+    return __asyncGenerator(this, arguments, function* (message, history = [], namespace) {
+      yield new __await(this.validationPromise);
+      yield* __yieldStar(this.pipeline.askStream(message, history, namespace));
+    });
+  }
   /**
    * Ingest documents into the vector database.
    */
   async ingest(documents, namespace) {
+    await this.validationPromise;
     return this.pipeline.ingest(documents, namespace);
   }
 };

package/dist/server.mjs

@@ -1,15 +1,15 @@
 import {
   ChromaDBProvider
-} from "./chunk-WGSZNY3X.mjs";
+} from "./chunk-CFVEZTBJ.mjs";
 import {
   RedisProvider
-} from "./chunk-NXUCKY5L.mjs";
+} from "./chunk-5YGUXK7Z.mjs";
 import {
   WeaviateProvider
-} from "./chunk-OOQXNLXD.mjs";
+} from "./chunk-M6JSPGAR.mjs";
 import {
   UniversalVectorProvider
-} from "./chunk-U6KHVZLF.mjs";
+} from "./chunk-LR3VMDVK.mjs";
 import {
   AnthropicProvider,
   BatchProcessor,
@@ -34,23 +34,23 @@ import {
   createIngestHandler,
   createUploadHandler,
   getRagConfig
-} from "./chunk-P4HAQ7KB.mjs";
-import "./chunk-67AJ6SMD.mjs";
+} from "./chunk-OJNAKSQ2.mjs";
+import "./chunk-YLTMFW4M.mjs";
 import {
   PineconeProvider
-} from "./chunk-ZNBKHNJ4.mjs";
+} from "./chunk-VUQJVIJT.mjs";
 import {
   PostgreSQLProvider
 } from "./chunk-FLOSGE6A.mjs";
 import {
   MongoDBProvider
-} from "./chunk-7SOSCZGS.mjs";
+} from "./chunk-73I6VWU3.mjs";
 import {
   MilvusProvider
-} from "./chunk-TYHTZIDP.mjs";
+} from "./chunk-U55XRW3U.mjs";
 import {
   QdrantProvider
-} from "./chunk-4A47RCG2.mjs";
+} from "./chunk-3DSHW676.mjs";
 import {
   BaseVectorProvider
 } from "./chunk-IMP6FUCY.mjs";
@@ -58,7 +58,7 @@ import {
   __objRest,
   __spreadProps,
   __spreadValues
-} from "./chunk-QMIKLALV.mjs";
+} from "./chunk-X4TOT24V.mjs";
 
 // src/config/ConfigBuilder.ts
 var ConfigBuilder = class {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@retrivora-ai/rag-engine",
-  "version": "1.0.0",
+  "version": "1.0.2",
   "description": "Retrivora AI is a plug-and-play AI engine for RAG chat experiences — generic vector DB + LLM provider, embeddable or standalone.",
   "author": "Abhinav Alkuchi",
   "license": "MIT",

package/src/components/ChatWindow.tsx

@@ -170,19 +170,20 @@ export function ChatWindow({ className = '', style, onClose, showClose = false }
             </div>
           </div>
         ) : (
-          messages.map((msg) => (
+          messages.map((msg, index) => (
             <MessageBubble
               key={msg.id}
               message={msg}
               sources={msg.sources}
+              isStreaming={isLoading && index === messages.length - 1 && msg.role === 'assistant'}
               primaryColor={ui.primaryColor}
               accentColor={ui.accentColor}
             />
           ))
         )}
 
-        {/* Typing indicator */}
-        {isLoading && (
+        {/* Typing indicator (only if assistant hasn't started replying yet) */}
+        {isLoading && messages[messages.length - 1]?.role !== 'assistant' && (
           <MessageBubble
             message={{ role: 'assistant', content: '' }}
             isStreaming

package/src/components/MessageBubble.tsx

@@ -55,8 +55,8 @@ export function MessageBubble({
               : {}
           }
         >
-          {isStreaming ? (
-            <span className="flex items-center gap-1">
+          {isStreaming && !message.content ? (
+            <span className="flex items-center gap-1 py-1">
               <span className="w-1.5 h-1.5 rounded-full bg-slate-400 dark:bg-white/60 animate-bounce [animation-delay:0ms]" />
               <span className="w-1.5 h-1.5 rounded-full bg-slate-400 dark:bg-white/60 animate-bounce [animation-delay:150ms]" />
               <span className="w-1.5 h-1.5 rounded-full bg-slate-400 dark:bg-white/60 animate-bounce [animation-delay:300ms]" />
@@ -66,6 +66,9 @@ export function MessageBubble({
               <ReactMarkdown remarkPlugins={[remarkGfm]}>
                 {message.content}
               </ReactMarkdown>
+              {isStreaming && message.content && (
+                <span className="inline-block w-1.5 h-3.5 ml-1 bg-emerald-500 animate-pulse align-middle" />
+              )}
             </div>
           )}
         </div>

package/src/core/Pipeline.ts

@@ -50,6 +50,7 @@ export class Pipeline {
   async initialize(): Promise<void> {
     if (this.initialised) return;
 
+    console.log(`[Pipeline] Initializing with provider: ${this.config.vectorDb.provider}...`);
     // Resolve vector DB provider
     this.vectorDB = await ProviderRegistry.createVectorProvider(this.config.vectorDb);
 

package/src/core/VectorPlugin.ts

@@ -20,32 +20,13 @@ import { ChatMessage } from '../llm/ILLMProvider';
 export class VectorPlugin {
   private config: RagConfig;
   private pipeline: Pipeline;
+  private validationPromise: Promise<void>;
 
-  /**
-   * Initializes the plugin with the host configuration.
-   * @param hostConfig - Configuration object passed from the host application.
-   * @throws Error if configuration is invalid or providers are unhealthy
-   *
-   * @example
-   * const plugin = new VectorPlugin({
-   *   projectId: 'my-app',
-   *   vectorDb: {
-   *     provider: 'pinecone',
-   *     indexName: 'my-index',
-   *     options: { apiKey: process.env.PINECONE_API_KEY }
-   *   },
-   *   llm: {
-   *     provider: 'openai',
-   *     model: 'gpt-4o',
-   *     apiKey: process.env.OPENAI_API_KEY
-   *   }
-   * });
-   */
   constructor(hostConfig?: Partial<RagConfig>) {
     this.config = ConfigResolver.resolve(hostConfig);
     
-    // Validate configuration early with detailed error messages
-    ConfigValidator.validateAndThrow(this.config);
+    // Start validation early
+    this.validationPromise = ConfigValidator.validateAndThrow(this.config);
     
     this.pipeline = new Pipeline(this.config);
   }
@@ -80,13 +61,23 @@ export class VectorPlugin {
    * Run a chat query.
    */
   async chat(message: string, history: ChatMessage[] = [], namespace?: string): Promise<ChatResponse> {
+    await this.validationPromise;
     return this.pipeline.ask(message, history, namespace);
   }
 
+  /**
+   * Run a streaming chat query.
+   */
+  async *chatStream(message: string, history: ChatMessage[] = [], namespace?: string) {
+    await this.validationPromise;
+    yield* this.pipeline.askStream(message, history, namespace);
+  }
+
   /**
    * Ingest documents into the vector database.
    */
   async ingest(documents: IngestDocument[], namespace?: string): Promise<Array<{ docId: string | number; chunksIngested: number }>> {
+    await this.validationPromise;
     return this.pipeline.ingest(documents, namespace);
   }
 

package/src/handlers/index.ts

@@ -32,6 +32,59 @@ export function createChatHandler(config?: Partial<RagConfig>) {
   };
 }
 
+/**
+ * createStreamHandler — factory that returns a streaming Next.js App Router POST handler
+ */
+export function createStreamHandler(config?: Partial<RagConfig>) {
+  const plugin = new VectorPlugin(config);
+
+  return async function POST(req: NextRequest) {
+    try {
+      const body = await req.json();
+      const { message, history = [], namespace } = body;
+      
+      const encoder = new TextEncoder();
+      const stream = new ReadableStream({
+        async start(controller) {
+          try {
+            const pipelineStream = plugin.chatStream(message, history, namespace);
+            
+            for await (const chunk of pipelineStream) {
+              if (typeof chunk === 'string') {
+                controller.enqueue(encoder.encode(chunk));
+              } else {
+                // Yield metadata/sources at the end
+                controller.enqueue(encoder.encode(`\n\n__METADATA__${JSON.stringify(chunk)}`));
+              }
+            }
+            controller.close();
+          } catch (streamError) {
+            console.error('[createStreamHandler] Stream processing error:', streamError);
+            const errorMessage = streamError instanceof Error ? streamError.message : String(streamError);
+            controller.enqueue(encoder.encode(`\n\n__ERROR__${JSON.stringify({ error: errorMessage })}`));
+            controller.close();
+          }
+        },
+      });
+
+      return new Response(stream, {
+        headers: {
+          'Content-Type': 'text/event-stream',
+          'Cache-Control': 'no-cache',
+          'Connection': 'keep-alive',
+        },
+      });
+    } catch (err) {
+      const message = err instanceof Error ? err.message : 'Internal server error';
+      return new Response(JSON.stringify({ error: message }), {
+        status: 500,
+        headers: { 'Content-Type': 'application/json' },
+      });
+    }
+  };
+}
+
+
 /**
  * createIngestHandler — factory for the document ingestion endpoint.
  */

package/src/hooks/useRagChat.ts

@@ -14,7 +14,6 @@
  */
 
 import { useState, useCallback, useRef, useEffect } from 'react';
-import axios from 'axios';
 import { VectorMatch } from '@/types';
 import { useStoredMessages } from './useStoredMessages';
 
@@ -111,26 +110,91 @@ export function useRagChat(
       try {
         const history = messagesRef.current.map(({ role, content }) => ({ role, content }));
 
-        const { data } = await axios.post(apiUrl, {
-          message: trimmed,
-          history,
-          namespace: namespace ?? projectId,
+        const response = await fetch(apiUrl, {
+          method: 'POST',
+          headers: { 'Content-Type': 'application/json' },
+          body: JSON.stringify({
+            message: trimmed,
+            history,
+            namespace: namespace ?? projectId,
+          }),
         });
 
-        const assistantMessage: RagMessage = {
-          id: `assistant_${Date.now()}`,
+        if (!response.ok) {
+          const errorData = await response.json().catch(() => ({}));
+          throw new Error(errorData.error || `Server returned ${response.status}`);
+        }
+
+        if (!response.body) {
+          throw new Error('No response body received from server');
+        }
+
+        const reader = response.body.getReader();
+        const decoder = new TextDecoder();
+        let assistantContent = '';
+        let sources: VectorMatch[] = [];
+
+        // Add a placeholder assistant message that we will update
+        const assistantMessageId = `assistant_${Date.now()}`;
+        setMessages((prev) => [
+          ...prev,
+          {
+            id: assistantMessageId,
+            role: 'assistant',
+            content: '',
+            createdAt: new Date().toISOString(),
+          },
+        ]);
+
+        while (true) {
+          const { done, value } = await reader.read();
+          if (done) break;
+
+          const chunk = decoder.decode(value, { stream: true });
+          
+          // Handle protocol tokens
+          if (chunk.includes('__METADATA__')) {
+            const [text, metadataStr] = chunk.split('__METADATA__');
+            if (text) assistantContent += text;
+            try {
+              const meta = JSON.parse(metadataStr);
+              sources = meta.sources || [];
+            } catch {
+              console.warn('[useRagChat] Failed to parse metadata');
+            }
+          } else if (chunk.includes('__ERROR__')) {
+            const [text, errorStr] = chunk.split('__ERROR__');
+            if (text) assistantContent += text;
+            try {
+              const err = JSON.parse(errorStr);
+              setError(err.error || 'Stream error');
+            } catch {
+              setError('An error occurred during streaming');
+            }
+          } else {
+            assistantContent += chunk;
+          }
+
+          // Update the specific message
+          setMessages((prev) => 
+            prev.map((msg) => 
+              msg.id === assistantMessageId 
+                ? { ...msg, content: assistantContent, sources: sources.length > 0 ? sources : msg.sources } 
+                : msg
+            )
+          );
+        }
+
+        const finalAssistantMessage: RagMessage = {
+          id: assistantMessageId,
           role: 'assistant',
-          content: data.reply,
-          sources: data.sources,
+          content: assistantContent,
+          sources: sources.length > 0 ? sources : undefined,
           createdAt: new Date().toISOString(),
         };
-
-        setMessages((prev) => [...prev, assistantMessage]);
-        onReply?.(assistantMessage);
+        onReply?.(finalAssistantMessage);
       } catch (err) {
-        const msg = axios.isAxiosError(err)
-          ? (err.response?.data?.error ?? err.message)
-          : 'Something went wrong. Please try again.';
+        const msg = err instanceof Error ? err.message : 'Something went wrong. Please try again.';
         setError(msg);
         onError?.(msg);
       } finally {

package/src/providers/vectordb/MongoDBProvider.ts

@@ -25,7 +25,7 @@ export class MongoDBProvider extends BaseVectorProvider {
     if (!opts.uri || !opts.database || !opts.collection) {
       throw new Error('[MongoDBProvider] options uri, database, and collection are required.');
     }
-    this.client = new MongoClient(opts.uri as string);
+    this.client = new MongoClient(opts.uri as string, { serverSelectionTimeoutMS: 5000 });
     this.dbName = opts.database as string;
     this.collectionName = opts.collection as string;
     this.embeddingKey = (opts.embeddingKey as string) || 'embedding';

package/src/rag/EntityExtractor.ts

@@ -20,6 +20,9 @@ Extract entities and relationships from the following text.
 Format the output as JSON with two keys: "nodes" (array of {id, label, properties}) and "edges" (array of {source, target, type, properties}).
 Use the same ID for the same entity.
 
+IMPORTANT: Ensure all property values are valid JSON types (strings, numbers, booleans, or null). 
+DO NOT include mathematical expressions like "4.5/5" as numbers; use strings instead.
+
 Text:
 "${text}"