feat(chat-window): update message handling

Merge master into feat/deep-research

.github/workflows/docker-build.yaml

@@ -8,15 +8,12 @@ on:
     types: [published]
 
 jobs:
-  build-and-push:
+  build-amd64:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout code
         uses: actions/checkout@v3
 
-      - name: Set up QEMU
-        uses: docker/setup-qemu-action@v2
-
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v2
         with:
@@ -33,28 +30,109 @@ jobs:
         id: version
         run: echo "RELEASE_VERSION=${GITHUB_REF#refs/tags/}" >> $GITHUB_ENV
 
-      - name: Build and push Docker image
+      - name: Build and push AMD64 Docker image
         if: github.ref == 'refs/heads/master' && github.event_name == 'push'
         run: |
-          docker buildx create --use
-          DOCKERFILE=app.dockerfile; \
-          IMAGE_NAME=perplexica; \
-          docker buildx build --platform linux/amd64,linux/arm64 \
-            --cache-from=type=registry,ref=itzcrazykns1337/${IMAGE_NAME}:main \
+          DOCKERFILE=app.dockerfile
+          IMAGE_NAME=perplexica
+          docker buildx build --platform linux/amd64 \
+            --cache-from=type=registry,ref=itzcrazykns1337/${IMAGE_NAME}:amd64 \
             --cache-to=type=inline \
+            --provenance false \
             -f $DOCKERFILE \
-            -t itzcrazykns1337/${IMAGE_NAME}:main \
+            -t itzcrazykns1337/${IMAGE_NAME}:amd64 \
             --push .
 
-      - name: Build and push release Docker image
+      - name: Build and push AMD64 release Docker image
         if: github.event_name == 'release'
         run: |
-          docker buildx create --use
-          DOCKERFILE=app.dockerfile; \
-          IMAGE_NAME=perplexica; \
-          docker buildx build --platform linux/amd64,linux/arm64 \
-            --cache-from=type=registry,ref=itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }} \
+          DOCKERFILE=app.dockerfile
+          IMAGE_NAME=perplexica
+          docker buildx build --platform linux/amd64 \
+            --cache-from=type=registry,ref=itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }}-amd64 \
             --cache-to=type=inline \
+            --provenance false \
             -f $DOCKERFILE \
-            -t itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }} \
+            -t itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }}-amd64 \
             --push .
+
+  build-arm64:
+    runs-on: ubuntu-24.04-arm
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v3
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
+        with:
+          install: true
+
+      - name: Log in to DockerHub
+        uses: docker/login-action@v2
+        with:
+          username: ${{ secrets.DOCKER_USERNAME }}
+          password: ${{ secrets.DOCKER_PASSWORD }}
+
+      - name: Extract version from release tag
+        if: github.event_name == 'release'
+        id: version
+        run: echo "RELEASE_VERSION=${GITHUB_REF#refs/tags/}" >> $GITHUB_ENV
+
+      - name: Build and push ARM64 Docker image
+        if: github.ref == 'refs/heads/master' && github.event_name == 'push'
+        run: |
+          DOCKERFILE=app.dockerfile
+          IMAGE_NAME=perplexica
+          docker buildx build --platform linux/arm64 \
+            --cache-from=type=registry,ref=itzcrazykns1337/${IMAGE_NAME}:arm64 \
+            --cache-to=type=inline \
+            --provenance false \
+            -f $DOCKERFILE \
+            -t itzcrazykns1337/${IMAGE_NAME}:arm64 \
+            --push .
+
+      - name: Build and push ARM64 release Docker image
+        if: github.event_name == 'release'
+        run: |
+          DOCKERFILE=app.dockerfile
+          IMAGE_NAME=perplexica
+          docker buildx build --platform linux/arm64 \
+            --cache-from=type=registry,ref=itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }}-arm64 \
+            --cache-to=type=inline \
+            --provenance false \
+            -f $DOCKERFILE \
+            -t itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }}-arm64 \
+            --push .
+
+  manifest:
+    needs: [build-amd64, build-arm64]
+    runs-on: ubuntu-latest
+    steps:
+      - name: Log in to DockerHub
+        uses: docker/login-action@v2
+        with:
+          username: ${{ secrets.DOCKER_USERNAME }}
+          password: ${{ secrets.DOCKER_PASSWORD }}
+
+      - name: Extract version from release tag
+        if: github.event_name == 'release'
+        id: version
+        run: echo "RELEASE_VERSION=${GITHUB_REF#refs/tags/}" >> $GITHUB_ENV
+
+      - name: Create and push multi-arch manifest for main
+        if: github.ref == 'refs/heads/master' && github.event_name == 'push'
+        run: |
+          IMAGE_NAME=perplexica
+          docker manifest create itzcrazykns1337/${IMAGE_NAME}:main \
+            --amend itzcrazykns1337/${IMAGE_NAME}:amd64 \
+            --amend itzcrazykns1337/${IMAGE_NAME}:arm64
+          docker manifest push itzcrazykns1337/${IMAGE_NAME}:main
+
+      - name: Create and push multi-arch manifest for releases
+        if: github.event_name == 'release'
+        run: |
+          IMAGE_NAME=perplexica
+          docker manifest create itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }} \
+            --amend itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }}-amd64 \
+            --amend itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }}-arm64
+          docker manifest push itzcrazykns1337/${IMAGE_NAME}:${{ env.RELEASE_VERSION }}

README.md

@@ -153,7 +153,7 @@ For more details, check out the full documentation [here](https://github.com/Itz
 
 ## Expose Perplexica to network
 
-You can access Perplexica over your home network by following our networking guide [here](https://github.com/ItzCrazyKns/Perplexica/blob/master/docs/installation/NETWORKING.md).
+Perplexica runs on Next.js and handles all API requests. It works right away on the same network and stays accessible even with port forwarding.
 
 ## One-Click Deployment
 

app.dockerfile

@@ -1,4 +1,4 @@
-FROM node:20.18.0-alpine AS builder
+FROM node:20.18.0-slim AS builder
 
 WORKDIR /home/perplexica
 
@@ -12,7 +12,7 @@ COPY public ./public
 RUN mkdir -p /home/perplexica/data
 RUN yarn build
 
-FROM node:20.18.0-alpine
+FROM node:20.18.0-slim
 
 WORKDIR /home/perplexica
 

docs/API/SEARCH.md

@@ -32,7 +32,8 @@ The API accepts a JSON object in the request body, where you define the focus mo
   "history": [
     ["human", "Hi, how are you?"],
     ["assistant", "I am doing well, how can I help you today?"]
-  ]
+  ],
+  "stream": false
 }
 ```
 
@@ -71,11 +72,13 @@ The API accepts a JSON object in the request body, where you define the focus mo
   ]
   ```
 
+- **`stream`** (boolean, optional): When set to `true`, enables streaming responses. Default is `false`.
+
 ### Response
 
 The response from the API includes both the final message and the sources used to generate that message.
 
-#### Example Response
+#### Standard Response (stream: false)
 
 ```json
 {
@@ -100,6 +103,28 @@ The response from the API includes both the final message and the sources used t
 }
 ```
 
+#### Streaming Response (stream: true)
+
+When streaming is enabled, the API returns a stream of newline-delimited JSON objects. Each line contains a complete, valid JSON object. The response has Content-Type: application/json.
+
+Example of streamed response objects:
+
+```
+{"type":"init","data":"Stream connected"}
+{"type":"sources","data":[{"pageContent":"...","metadata":{"title":"...","url":"..."}},...]}
+{"type":"response","data":"Perplexica is an "}
+{"type":"response","data":"innovative, open-source "}
+{"type":"response","data":"AI-powered search engine..."}
+{"type":"done"}
+```
+
+Clients should process each line as a separate JSON object. The different message types include:
+
+- **`init`**: Initial connection message
+- **`sources`**: All sources used for the response
+- **`response`**: Chunks of the generated answer text
+- **`done`**: Indicates the stream is complete
+
 ### Fields in the Response
 
 - **`message`** (string): The search result, generated based on the query and focus mode.

package.json

@@ -1,6 +1,6 @@
 {
   "name": "perplexica-frontend",
-  "version": "1.10.0",
+  "version": "1.10.1",
   "license": "MIT",
   "author": "ItzCrazyKns",
   "scripts": {
@@ -15,8 +15,10 @@
     "@headlessui/react": "^2.2.0",
     "@iarna/toml": "^2.2.5",
     "@icons-pack/react-simple-icons": "^12.3.0",
+    "@langchain/anthropic": "^0.3.15",
     "@langchain/community": "^0.3.36",
     "@langchain/core": "^0.3.42",
+    "@langchain/google-genai": "^0.1.12",
     "@langchain/openai": "^0.0.25",
     "@langchain/textsplitters": "^0.1.0",
     "@tailwindcss/typography": "^0.5.12",

src/app/api/chat/route.ts

@@ -295,9 +295,9 @@ export const POST = async (req: Request) => {
       },
     });
   } catch (err) {
-    console.error('An error ocurred while processing chat request:', err);
+    console.error('An error occurred while processing chat request:', err);
     return Response.json(
-      { message: 'An error ocurred while processing chat request' },
+      { message: 'An error occurred while processing chat request' },
       { status: 500 },
     );
   }

src/app/api/config/route.ts

@@ -59,9 +59,9 @@ export const GET = async (req: Request) => {
 
     return Response.json({ ...config }, { status: 200 });
   } catch (err) {
-    console.error('An error ocurred while getting config:', err);
+    console.error('An error occurred while getting config:', err);
     return Response.json(
-      { message: 'An error ocurred while getting config' },
+      { message: 'An error occurred while getting config' },
       { status: 500 },
     );
   }
@@ -100,9 +100,9 @@ export const POST = async (req: Request) => {
 
     return Response.json({ message: 'Config updated' }, { status: 200 });
   } catch (err) {
-    console.error('An error ocurred while updating config:', err);
+    console.error('An error occurred while updating config:', err);
     return Response.json(
-      { message: 'An error ocurred while updating config' },
+      { message: 'An error occurred while updating config' },
       { status: 500 },
     );
   }

src/app/api/discover/route.ts

@@ -48,7 +48,7 @@ export const GET = async (req: Request) => {
       },
     );
   } catch (err) {
-    console.error(`An error ocurred in discover route: ${err}`);
+    console.error(`An error occurred in discover route: ${err}`);
     return Response.json(
       {
         message: 'An error has occurred',

src/app/api/images/route.ts

@@ -74,9 +74,9 @@ export const POST = async (req: Request) => {
 
     return Response.json({ images }, { status: 200 });
   } catch (err) {
-    console.error(`An error ocurred while searching images: ${err}`);
+    console.error(`An error occurred while searching images: ${err}`);
     return Response.json(
-      { message: 'An error ocurred while searching images' },
+      { message: 'An error occurred while searching images' },
       { status: 500 },
     );
   }

src/app/api/models/route.ts

@@ -34,7 +34,7 @@ export const GET = async (req: Request) => {
       },
     );
   } catch (err) {
-    console.error('An error ocurred while fetching models', err);
+    console.error('An error occurred while fetching models', err);
     return Response.json(
       {
         message: 'An error has occurred.',

src/app/api/search/route.ts

@@ -33,6 +33,7 @@ interface ChatRequestBody {
   embeddingModel?: embeddingModel;
   query: string;
   history: Array<[string, string]>;
+  stream?: boolean;
 }
 
 export const POST = async (req: Request) => {
@@ -48,6 +49,7 @@ export const POST = async (req: Request) => {
 
     body.history = body.history || [];
     body.optimizationMode = body.optimizationMode || 'balanced';
+    body.stream = body.stream || false;
 
     const history: BaseMessage[] = body.history.map((msg) => {
       return msg[0] === 'human'
@@ -125,40 +127,137 @@ export const POST = async (req: Request) => {
       [],
     );
 
-    return new Promise(
-      (
-        resolve: (value: Response) => void,
-        reject: (value: Response) => void,
-      ) => {
-        let message = '';
+    if (!body.stream) {
+      return new Promise(
+        (
+          resolve: (value: Response) => void,
+          reject: (value: Response) => void,
+        ) => {
+          let message = '';
+          let sources: any[] = [];
+
+          emitter.on('data', (data: string) => {
+            try {
+              const parsedData = JSON.parse(data);
+              if (parsedData.type === 'response') {
+                message += parsedData.data;
+              } else if (parsedData.type === 'sources') {
+                sources = parsedData.data;
+              }
+            } catch (error) {
+              reject(
+                Response.json(
+                  { message: 'Error parsing data' },
+                  { status: 500 },
+                ),
+              );
+            }
+          });
+
+          emitter.on('end', () => {
+            resolve(Response.json({ message, sources }, { status: 200 }));
+          });
+
+          emitter.on('error', (error: any) => {
+            reject(
+              Response.json(
+                { message: 'Search error', error },
+                { status: 500 },
+              ),
+            );
+          });
+        },
+      );
+    }
+
+    const encoder = new TextEncoder();
+
+    const abortController = new AbortController();
+    const { signal } = abortController;
+
+    const stream = new ReadableStream({
+      start(controller) {
         let sources: any[] = [];
 
-        emitter.on('data', (data) => {
+        controller.enqueue(
+          encoder.encode(
+            JSON.stringify({
+              type: 'init',
+              data: 'Stream connected',
+            }) + '\n',
+          ),
+        );
+
+        signal.addEventListener('abort', () => {
+          emitter.removeAllListeners();
+
+          try {
+            controller.close();
+          } catch (error) {}
+        });
+
+        emitter.on('data', (data: string) => {
+          if (signal.aborted) return;
+
           try {
             const parsedData = JSON.parse(data);
+
             if (parsedData.type === 'response') {
-              message += parsedData.data;
+              controller.enqueue(
+                encoder.encode(
+                  JSON.stringify({
+                    type: 'response',
+                    data: parsedData.data,
+                  }) + '\n',
+                ),
+              );
             } else if (parsedData.type === 'sources') {
               sources = parsedData.data;
+              controller.enqueue(
+                encoder.encode(
+                  JSON.stringify({
+                    type: 'sources',
+                    data: sources,
+                  }) + '\n',
+                ),
+              );
             }
           } catch (error) {
-            reject(
-              Response.json({ message: 'Error parsing data' }, { status: 500 }),
-            );
+            controller.error(error);
           }
         });
 
         emitter.on('end', () => {
-          resolve(Response.json({ message, sources }, { status: 200 }));
+          if (signal.aborted) return;
+
+          controller.enqueue(
+            encoder.encode(
+              JSON.stringify({
+                type: 'done',
+              }) + '\n',
+            ),
+          );
+          controller.close();
         });
 
-        emitter.on('error', (error) => {
-          reject(
-            Response.json({ message: 'Search error', error }, { status: 500 }),
-          );
+        emitter.on('error', (error: any) => {
+          if (signal.aborted) return;
+
+          controller.error(error);
         });
       },
-    );
+      cancel() {
+        abortController.abort();
+      },
+    });
+
+    return new Response(stream, {
+      headers: {
+        'Content-Type': 'text/event-stream',
+        'Cache-Control': 'no-cache, no-transform',
+        Connection: 'keep-alive',
+      },
+    });
   } catch (err: any) {
     console.error(`Error in getting search results: ${err.message}`);
     return Response.json(

src/app/api/suggestions/route.ts

@@ -72,9 +72,9 @@ export const POST = async (req: Request) => {
 
     return Response.json({ suggestions }, { status: 200 });
   } catch (err) {
-    console.error(`An error ocurred while generating suggestions: ${err}`);
+    console.error(`An error occurred while generating suggestions: ${err}`);
     return Response.json(
-      { message: 'An error ocurred while generating suggestions' },
+      { message: 'An error occurred while generating suggestions' },
       { status: 500 },
     );
   }

src/app/api/videos/route.ts

@@ -74,9 +74,9 @@ export const POST = async (req: Request) => {
 
     return Response.json({ videos }, { status: 200 });
   } catch (err) {
-    console.error(`An error ocurred while searching videos: ${err}`);
+    console.error(`An error occurred while searching videos: ${err}`);
     return Response.json(
-      { message: 'An error ocurred while searching videos' },
+      { message: 'An error occurred while searching videos' },
       { status: 500 },
     );
   }

src/components/ChatWindow.tsx

@@ -363,20 +363,18 @@ const ChatWindow = ({ id }: { id?: string }) => {
 
       if (data.type === 'sources') {
         sources = data.data;
-        if (!added) {
-          setMessages((prevMessages) => [
-            ...prevMessages,
-            {
-              content: '',
-              messageId: data.messageId,
-              chatId: chatId!,
-              role: 'assistant',
-              sources: sources,
-              createdAt: new Date(),
-            },
-          ]);
-          added = true;
-        }
+        setMessages((prevMessages) => [
+          ...prevMessages,
+          {
+            content: '',
+            messageId: data.messageId,
+            chatId: chatId!,
+            role: 'assistant',
+            sources: sources,
+            createdAt: new Date(),
+          },
+        ]);
+        added = true;
         setMessageAppeared(true);
       }
 
@@ -394,20 +392,20 @@ const ChatWindow = ({ id }: { id?: string }) => {
             },
           ]);
           added = true;
+          setMessageAppeared(true);
+        } else {
+          setMessages((prev) =>
+            prev.map((message) => {
+              if (message.messageId === data.messageId) {
+                return { ...message, content: message.content + data.data };
+              }
+
+              return message;
+            }),
+          );
         }
 
-        setMessages((prev) =>
-          prev.map((message) => {
-            if (message.messageId === data.messageId) {
-              return { ...message, content: message.content + data.data };
-            }
-
-            return message;
-          }),
-        );
-
         recievedMessage += data.data;
-        setMessageAppeared(true);
       }
 
       if (data.type === 'messageEnd') {

src/lib/providers/anthropic.ts

@@ -1,4 +1,4 @@
-import { ChatOpenAI } from '@langchain/openai';
+import { ChatAnthropic } from '@langchain/anthropic';
 import { ChatModel } from '.';
 import { getAnthropicApiKey } from '../config';
 import { BaseChatModel } from '@langchain/core/language_models/chat_models';
@@ -45,13 +45,10 @@ export const loadAnthropicChatModels = async () => {
     anthropicChatModels.forEach((model) => {
       chatModels[model.key] = {
         displayName: model.displayName,
-        model: new ChatOpenAI({
-          openAIApiKey: anthropicApiKey,
+        model: new ChatAnthropic({
+          apiKey: anthropicApiKey,
           modelName: model.key,
           temperature: 0.7,
-          configuration: {
-            baseURL: 'https://api.anthropic.com/v1/',
-          },
         }) as unknown as BaseChatModel,
       };
     });

src/lib/providers/gemini.ts

@@ -1,10 +1,17 @@
-import { ChatOpenAI, OpenAIEmbeddings } from '@langchain/openai';
+import {
+  ChatGoogleGenerativeAI,
+  GoogleGenerativeAIEmbeddings,
+} from '@langchain/google-genai';
 import { getGeminiApiKey } from '../config';
 import { ChatModel, EmbeddingModel } from '.';
 import { BaseChatModel } from '@langchain/core/language_models/chat_models';
 import { Embeddings } from '@langchain/core/embeddings';
 
 const geminiChatModels: Record<string, string>[] = [
+  {
+    displayName: 'Gemini 2.5 Pro Experimental',
+    key: 'gemini-2.5-pro-exp-03-25',
+  },
   {
     displayName: 'Gemini 2.0 Flash',
     key: 'gemini-2.0-flash',
@@ -14,8 +21,8 @@ const geminiChatModels: Record<string, string>[] = [
     key: 'gemini-2.0-flash-lite',
   },
   {
-    displayName: 'Gemini 2.0 Pro Experimental',
-    key: 'gemini-2.0-pro-exp-02-05',
+    displayName: 'Gemini 2.0 Flash Thinking Experimental',
+    key: 'gemini-2.0-flash-thinking-exp-01-21',
   },
   {
     displayName: 'Gemini 1.5 Flash',
@@ -49,13 +56,10 @@ export const loadGeminiChatModels = async () => {
     geminiChatModels.forEach((model) => {
       chatModels[model.key] = {
         displayName: model.displayName,
-        model: new ChatOpenAI({
-          openAIApiKey: geminiApiKey,
+        model: new ChatGoogleGenerativeAI({
+          apiKey: geminiApiKey,
           modelName: model.key,
           temperature: 0.7,
-          configuration: {
-            baseURL: 'https://generativelanguage.googleapis.com/v1beta/openai/',
-          },
         }) as unknown as BaseChatModel,
       };
     });
@@ -78,12 +82,9 @@ export const loadGeminiEmbeddingModels = async () => {
     geminiEmbeddingModels.forEach((model) => {
       embeddingModels[model.key] = {
         displayName: model.displayName,
-        model: new OpenAIEmbeddings({
-          openAIApiKey: geminiApiKey,
+        model: new GoogleGenerativeAIEmbeddings({
+          apiKey: geminiApiKey,
           modelName: model.key,
-          configuration: {
-            baseURL: 'https://generativelanguage.googleapis.com/v1beta/openai/',
-          },
         }) as unknown as Embeddings,
       };
     });

src/lib/search/metaSearchAgent.ts

@@ -6,11 +6,6 @@ import {
   MessagesPlaceholder,
   PromptTemplate,
 } from '@langchain/core/prompts';
-import {
-  RunnableLambda,
-  RunnableMap,
-  RunnableSequence,
-} from '@langchain/core/runnables';
 import { BaseMessage } from '@langchain/core/messages';
 import { StringOutputParser } from '@langchain/core/output_parsers';
 import LineListOutputParser from '../outputParsers/listLineOutputParser';
@@ -24,6 +19,7 @@ import computeSimilarity from '../utils/computeSimilarity';
 import formatChatHistoryAsString from '../utils/formatHistory';
 import eventEmitter from 'events';
 import { StreamEvent } from '@langchain/core/tracers/log_stream';
+import { EventEmitter } from 'node:stream';
 
 export interface MetaSearchAgentType {
   searchAndAnswer: (
@@ -46,7 +42,7 @@ interface Config {
   activeEngines: string[];
 }
 
-type BasicChainInput = {
+type SearchInput = {
   chat_history: BaseMessage[];
   query: string;
 };
@@ -59,235 +55,240 @@ class MetaSearchAgent implements MetaSearchAgentType {
     this.config = config;
   }
 
-  private async createSearchRetrieverChain(llm: BaseChatModel) {
+  private async searchSources(
+    llm: BaseChatModel,
+    input: SearchInput,
+    emitter: EventEmitter,
+  ) {
     (llm as unknown as ChatOpenAI).temperature = 0;
 
-    return RunnableSequence.from([
-      PromptTemplate.fromTemplate(this.config.queryGeneratorPrompt),
-      llm,
-      this.strParser,
-      RunnableLambda.from(async (input: string) => {
-        const linksOutputParser = new LineListOutputParser({
-          key: 'links',
-        });
+    const chatPrompt = PromptTemplate.fromTemplate(
+      this.config.queryGeneratorPrompt,
+    );
 
-        const questionOutputParser = new LineOutputParser({
-          key: 'question',
-        });
+    const processedChatPrompt = await chatPrompt.invoke({
+      chat_history: formatChatHistoryAsString(input.chat_history),
+      query: input.query,
+    });
 
-        const links = await linksOutputParser.parse(input);
-        let question = this.config.summarizer
-          ? await questionOutputParser.parse(input)
-          : input;
+    const llmRes = await llm.invoke(processedChatPrompt);
+    const messageStr = await this.strParser.invoke(llmRes);
 
-        if (question === 'not_needed') {
-          return { query: '', docs: [] };
+    const linksOutputParser = new LineListOutputParser({
+      key: 'links',
+    });
+
+    const questionOutputParser = new LineOutputParser({
+      key: 'question',
+    });
+
+    const links = await linksOutputParser.parse(messageStr);
+    let question = this.config.summarizer
+      ? await questionOutputParser.parse(messageStr)
+      : messageStr;
+
+    if (question === 'not_needed') {
+      return { query: '', docs: [] };
+    }
+
+    if (links.length > 0) {
+      if (question.length === 0) {
+        question = 'summarize';
+      }
+
+      let docs: Document[] = [];
+
+      const linkDocs = await getDocumentsFromLinks({ links });
+
+      const docGroups: Document[] = [];
+
+      linkDocs.map((doc) => {
+        const URLDocExists = docGroups.find(
+          (d) =>
+            d.metadata.url === doc.metadata.url && d.metadata.totalDocs < 10,
+        );
+
+        if (!URLDocExists) {
+          docGroups.push({
+            ...doc,
+            metadata: {
+              ...doc.metadata,
+              totalDocs: 1,
+            },
+          });
         }
 
-        if (links.length > 0) {
-          if (question.length === 0) {
-            question = 'summarize';
-          }
+        const docIndex = docGroups.findIndex(
+          (d) =>
+            d.metadata.url === doc.metadata.url && d.metadata.totalDocs < 10,
+        );
 
-          let docs: Document[] = [];
-
-          const linkDocs = await getDocumentsFromLinks({ links });
-
-          const docGroups: Document[] = [];
-
-          linkDocs.map((doc) => {
-            const URLDocExists = docGroups.find(
-              (d) =>
-                d.metadata.url === doc.metadata.url &&
-                d.metadata.totalDocs < 10,
-            );
-
-            if (!URLDocExists) {
-              docGroups.push({
-                ...doc,
-                metadata: {
-                  ...doc.metadata,
-                  totalDocs: 1,
-                },
-              });
-            }
-
-            const docIndex = docGroups.findIndex(
-              (d) =>
-                d.metadata.url === doc.metadata.url &&
-                d.metadata.totalDocs < 10,
-            );
-
-            if (docIndex !== -1) {
-              docGroups[docIndex].pageContent =
-                docGroups[docIndex].pageContent + `\n\n` + doc.pageContent;
-              docGroups[docIndex].metadata.totalDocs += 1;
-            }
-          });
-
-          await Promise.all(
-            docGroups.map(async (doc) => {
-              const res = await llm.invoke(`
-            You are a web search summarizer, tasked with summarizing a piece of text retrieved from a web search. Your job is to summarize the 
-            text into a detailed, 2-4 paragraph explanation that captures the main ideas and provides a comprehensive answer to the query.
-            If the query is \"summarize\", you should provide a detailed summary of the text. If the query is a specific question, you should answer it in the summary.
-            
-            - **Journalistic tone**: The summary should sound professional and journalistic, not too casual or vague.
-            - **Thorough and detailed**: Ensure that every key point from the text is captured and that the summary directly answers the query.
-            - **Not too lengthy, but detailed**: The summary should be informative but not excessively long. Focus on providing detailed information in a concise format.
-
-            The text will be shared inside the \`text\` XML tag, and the query inside the \`query\` XML tag.
-
-            <example>
-            1. \`<text>
-            Docker is a set of platform-as-a-service products that use OS-level virtualization to deliver software in packages called containers. 
-            It was first released in 2013 and is developed by Docker, Inc. Docker is designed to make it easier to create, deploy, and run applications 
-            by using containers.
-            </text>
-
-            <query>
-            What is Docker and how does it work?
-            </query>
-
-            Response:
-            Docker is a revolutionary platform-as-a-service product developed by Docker, Inc., that uses container technology to make application 
-            deployment more efficient. It allows developers to package their software with all necessary dependencies, making it easier to run in 
-            any environment. Released in 2013, Docker has transformed the way applications are built, deployed, and managed.
-            \`
-            2. \`<text>
-            The theory of relativity, or simply relativity, encompasses two interrelated theories of Albert Einstein: special relativity and general
-            relativity. However, the word "relativity" is sometimes used in reference to Galilean invariance. The term "theory of relativity" was based
-            on the expression "relative theory" used by Max Planck in 1906. The theory of relativity usually encompasses two interrelated theories by
-            Albert Einstein: special relativity and general relativity. Special relativity applies to all physical phenomena in the absence of gravity.
-            General relativity explains the law of gravitation and its relation to other forces of nature. It applies to the cosmological and astrophysical
-            realm, including astronomy.
-            </text>
-
-            <query>
-            summarize
-            </query>
-
-            Response:
-            The theory of relativity, developed by Albert Einstein, encompasses two main theories: special relativity and general relativity. Special
-            relativity applies to all physical phenomena in the absence of gravity, while general relativity explains the law of gravitation and its
-            relation to other forces of nature. The theory of relativity is based on the concept of "relative theory," as introduced by Max Planck in
-            1906. It is a fundamental theory in physics that has revolutionized our understanding of the universe.
-            \`
-            </example>
-
-            Everything below is the actual data you will be working with. Good luck!
-
-            <query>
-            ${question}
-            </query>
-
-            <text>
-            ${doc.pageContent}
-            </text>
-
-            Make sure to answer the query in the summary.
-          `);
-
-              const document = new Document({
-                pageContent: res.content as string,
-                metadata: {
-                  title: doc.metadata.title,
-                  url: doc.metadata.url,
-                },
-              });
-
-              docs.push(document);
-            }),
-          );
-
-          return { query: question, docs: docs };
-        } else {
-          question = question.replace(/<think>.*?<\/think>/g, '');
-
-          const res = await searchSearxng(question, {
-            language: 'en',
-            engines: this.config.activeEngines,
-          });
-
-          const documents = res.results.map(
-            (result) =>
-              new Document({
-                pageContent:
-                  result.content ||
-                  (this.config.activeEngines.includes('youtube')
-                    ? result.title
-                    : '') /* Todo: Implement transcript grabbing using Youtubei (source: https://www.npmjs.com/package/youtubei) */,
-                metadata: {
-                  title: result.title,
-                  url: result.url,
-                  ...(result.img_src && { img_src: result.img_src }),
-                },
-              }),
-          );
-
-          return { query: question, docs: documents };
+        if (docIndex !== -1) {
+          docGroups[docIndex].pageContent =
+            docGroups[docIndex].pageContent + `\n\n` + doc.pageContent;
+          docGroups[docIndex].metadata.totalDocs += 1;
         }
-      }),
-    ]);
+      });
+
+      await Promise.all(
+        docGroups.map(async (doc) => {
+          const res = await llm.invoke(`
+        You are a web search summarizer, tasked with summarizing a piece of text retrieved from a web search. Your job is to summarize the 
+        text into a detailed, 2-4 paragraph explanation that captures the main ideas and provides a comprehensive answer to the query.
+        If the query is \"summarize\", you should provide a detailed summary of the text. If the query is a specific question, you should answer it in the summary.
+        
+        - **Journalistic tone**: The summary should sound professional and journalistic, not too casual or vague.
+        - **Thorough and detailed**: Ensure that every key point from the text is captured and that the summary directly answers the query.
+        - **Not too lengthy, but detailed**: The summary should be informative but not excessively long. Focus on providing detailed information in a concise format.
+
+        The text will be shared inside the \`text\` XML tag, and the query inside the \`query\` XML tag.
+
+        <example>
+        1. \`<text>
+        Docker is a set of platform-as-a-service products that use OS-level virtualization to deliver software in packages called containers. 
+        It was first released in 2013 and is developed by Docker, Inc. Docker is designed to make it easier to create, deploy, and run applications 
+        by using containers.
+        </text>
+
+        <query>
+        What is Docker and how does it work?
+        </query>
+
+        Response:
+        Docker is a revolutionary platform-as-a-service product developed by Docker, Inc., that uses container technology to make application 
+        deployment more efficient. It allows developers to package their software with all necessary dependencies, making it easier to run in 
+        any environment. Released in 2013, Docker has transformed the way applications are built, deployed, and managed.
+        \`
+        2. \`<text>
+        The theory of relativity, or simply relativity, encompasses two interrelated theories of Albert Einstein: special relativity and general
+        relativity. However, the word "relativity" is sometimes used in reference to Galilean invariance. The term "theory of relativity" was based
+        on the expression "relative theory" used by Max Planck in 1906. The theory of relativity usually encompasses two interrelated theories by
+        Albert Einstein: special relativity and general relativity. Special relativity applies to all physical phenomena in the absence of gravity.
+        General relativity explains the law of gravitation and its relation to other forces of nature. It applies to the cosmological and astrophysical
+        realm, including astronomy.
+        </text>
+
+        <query>
+        summarize
+        </query>
+
+        Response:
+        The theory of relativity, developed by Albert Einstein, encompasses two main theories: special relativity and general relativity. Special
+        relativity applies to all physical phenomena in the absence of gravity, while general relativity explains the law of gravitation and its
+        relation to other forces of nature. The theory of relativity is based on the concept of "relative theory," as introduced by Max Planck in
+        1906. It is a fundamental theory in physics that has revolutionized our understanding of the universe.
+        \`
+        </example>
+
+        Everything below is the actual data you will be working with. Good luck!
+
+        <query>
+        ${question}
+        </query>
+
+        <text>
+        ${doc.pageContent}
+        </text>
+
+        Make sure to answer the query in the summary.
+      `);
+
+          const document = new Document({
+            pageContent: res.content as string,
+            metadata: {
+              title: doc.metadata.title,
+              url: doc.metadata.url,
+            },
+          });
+
+          docs.push(document);
+        }),
+      );
+
+      return { query: question, docs: docs };
+    } else {
+      question = question.replace(/<think>.*?<\/think>/g, '');
+
+      const res = await searchSearxng(question, {
+        language: 'en',
+        engines: this.config.activeEngines,
+      });
+
+      const documents = res.results.map(
+        (result) =>
+          new Document({
+            pageContent:
+              result.content ||
+              (this.config.activeEngines.includes('youtube')
+                ? result.title
+                : '') /* Todo: Implement transcript grabbing using Youtubei (source: https://www.npmjs.com/package/youtubei) */,
+            metadata: {
+              title: result.title,
+              url: result.url,
+              ...(result.img_src && { img_src: result.img_src }),
+            },
+          }),
+      );
+
+      return { query: question, docs: documents };
+    }
   }
 
-  private async createAnsweringChain(
+  private async streamAnswer(
     llm: BaseChatModel,
     fileIds: string[],
     embeddings: Embeddings,
     optimizationMode: 'speed' | 'balanced' | 'quality',
+    input: SearchInput,
+    emitter: EventEmitter,
   ) {
-    return RunnableSequence.from([
-      RunnableMap.from({
-        query: (input: BasicChainInput) => input.query,
-        chat_history: (input: BasicChainInput) => input.chat_history,
-        date: () => new Date().toISOString(),
-        context: RunnableLambda.from(async (input: BasicChainInput) => {
-          const processedHistory = formatChatHistoryAsString(
-            input.chat_history,
-          );
+    const chatPrompt = ChatPromptTemplate.fromMessages([
+      ['system', this.config.responsePrompt],
+      new MessagesPlaceholder('chat_history'),
+      ['user', '{query}'],
+    ]);
 
-          let docs: Document[] | null = null;
-          let query = input.query;
+    let docs: Document[] | null = null;
+    let query = input.query;
 
-          if (this.config.searchWeb) {
-            const searchRetrieverChain =
-              await this.createSearchRetrieverChain(llm);
+    if (this.config.searchWeb) {
+      const searchResults = await this.searchSources(llm, input, emitter);
 
-            const searchRetrieverResult = await searchRetrieverChain.invoke({
-              chat_history: processedHistory,
-              query,
-            });
+      query = searchResults.query;
+      docs = searchResults.docs;
+    }
 
-            query = searchRetrieverResult.query;
-            docs = searchRetrieverResult.docs;
-          }
+    const sortedDocs = await this.rerankDocs(
+      query,
+      docs ?? [],
+      fileIds,
+      embeddings,
+      optimizationMode,
+    );
 
-          const sortedDocs = await this.rerankDocs(
-            query,
-            docs ?? [],
-            fileIds,
-            embeddings,
-            optimizationMode,
-          );
+    emitter.emit('data', JSON.stringify({ type: 'sources', data: sortedDocs }));
 
-          return sortedDocs;
-        })
-          .withConfig({
-            runName: 'FinalSourceRetriever',
-          })
-          .pipe(this.processDocs),
-      }),
-      ChatPromptTemplate.fromMessages([
-        ['system', this.config.responsePrompt],
-        new MessagesPlaceholder('chat_history'),
-        ['user', '{query}'],
-      ]),
-      llm,
-      this.strParser,
-    ]).withConfig({
-      runName: 'FinalResponseGenerator',
+    const context = this.processDocs(sortedDocs);
+
+    const formattedChatPrompt = await chatPrompt.invoke({
+      query: input.query,
+      chat_history: input.chat_history,
+      date: new Date().toISOString(),
+      context: context,
     });
+
+    const llmRes = await llm.stream(formattedChatPrompt);
+
+    for await (const data of llmRes) {
+      const messageStr = await this.strParser.invoke(data);
+
+      emitter.emit(
+        'data',
+        JSON.stringify({ type: 'response', data: messageStr }),
+      );
+    }
+
+    emitter.emit('end');
   }
 
   private async rerankDocs(
@@ -428,39 +429,6 @@ class MetaSearchAgent implements MetaSearchAgentType {
       .join('\n');
   }
 
-  private async handleStream(
-    stream: AsyncGenerator<StreamEvent, any, any>,
-    emitter: eventEmitter,
-  ) {
-    for await (const event of stream) {
-      if (
-        event.event === 'on_chain_end' &&
-        event.name === 'FinalSourceRetriever'
-      ) {
-        ``;
-        emitter.emit(
-          'data',
-          JSON.stringify({ type: 'sources', data: event.data.output }),
-        );
-      }
-      if (
-        event.event === 'on_chain_stream' &&
-        event.name === 'FinalResponseGenerator'
-      ) {
-        emitter.emit(
-          'data',
-          JSON.stringify({ type: 'response', data: event.data.chunk }),
-        );
-      }
-      if (
-        event.event === 'on_chain_end' &&
-        event.name === 'FinalResponseGenerator'
-      ) {
-        emitter.emit('end');
-      }
-    }
-  }
-
   async searchAndAnswer(
     message: string,
     history: BaseMessage[],
@@ -471,25 +439,18 @@ class MetaSearchAgent implements MetaSearchAgentType {
   ) {
     const emitter = new eventEmitter();
 
-    const answeringChain = await this.createAnsweringChain(
+    this.streamAnswer(
       llm,
       fileIds,
       embeddings,
       optimizationMode,
-    );
-
-    const stream = answeringChain.streamEvents(
       {
         chat_history: history,
         query: message,
       },
-      {
-        version: 'v1',
-      },
+      emitter,
     );
 
-    this.handleStream(stream, emitter);
-
     return emitter;
   }
 }

yarn.lock

@@ -12,6 +12,19 @@
   resolved "https://registry.yarnpkg.com/@alloc/quick-lru/-/quick-lru-5.2.0.tgz#7bf68b20c0a350f936915fcae06f58e32007ce30"
   integrity sha512-UrcABB+4bUrFABwbluTIBErXwvbsU/V7TZWfmbgJfbkwiBuziS9gxdODUyuiecfdGQ85jglMW6juS3+z5TsKLw==
 
+"@anthropic-ai/sdk@^0.37.0":
+  version "0.37.0"
+  resolved "https://registry.yarnpkg.com/@anthropic-ai/sdk/-/sdk-0.37.0.tgz#0018127404ecb9b8a12968068e0c4b3e8bbd6386"
+  integrity sha512-tHjX2YbkUBwEgg0JZU3EFSSAQPoK4qQR/NFYa8Vtzd5UAyXzZksCw2In69Rml4R/TyHPBfRYaLK35XiOe33pjw==
+  dependencies:
+    "@types/node" "^18.11.18"
+    "@types/node-fetch" "^2.6.4"
+    abort-controller "^3.0.0"
+    agentkeepalive "^4.2.1"
+    form-data-encoder "1.7.2"
+    formdata-node "^4.3.2"
+    node-fetch "^2.6.7"
+
 "@anthropic-ai/sdk@^0.9.1":
   version "0.9.1"
   resolved "https://registry.yarnpkg.com/@anthropic-ai/sdk/-/sdk-0.9.1.tgz#b2d2b7bf05c90dce502c9a2e869066870f69ba88"
@@ -374,6 +387,11 @@
   resolved "https://registry.yarnpkg.com/@floating-ui/utils/-/utils-0.2.8.tgz#21a907684723bbbaa5f0974cf7730bd797eb8e62"
   integrity sha512-kym7SodPp8/wloecOpcmSnWJsK7M0E5Wg8UcFA+uO4B9s5d0ywXOEro/8HM9x0rW+TljRzul/14UYz3TleT3ig==
 
+"@google/generative-ai@^0.24.0":
+  version "0.24.0"
+  resolved "https://registry.yarnpkg.com/@google/generative-ai/-/generative-ai-0.24.0.tgz#4d27af7d944c924a27a593c17ad1336535d53846"
+  integrity sha512-fnEITCGEB7NdX0BhoYZ/cq/7WPZ1QS5IzJJfC3Tg/OwkvBetMiVJciyaan297OvE4B9Jg1xvo0zIazX/9sGu1Q==
+
 "@headlessui/react@^2.2.0":
   version "2.2.0"
   resolved "https://registry.yarnpkg.com/@headlessui/react/-/react-2.2.0.tgz#a8e32f0899862849a1ce1615fa280e7891431ab7"
@@ -575,6 +593,16 @@
     "@jridgewell/resolve-uri" "^3.1.0"
     "@jridgewell/sourcemap-codec" "^1.4.14"
 
+"@langchain/anthropic@^0.3.15":
+  version "0.3.15"
+  resolved "https://registry.yarnpkg.com/@langchain/anthropic/-/anthropic-0.3.15.tgz#0244cdb345cb492eb40aedd681881ebadfbb73f2"
+  integrity sha512-Ar2viYcZ64idgV7EtCBCb36tIkNtPAhQRxSaMTWPHGspFgMfvwRoleVri9e90sCpjpS9xhlHsIQ0LlUS/Atsrw==
+  dependencies:
+    "@anthropic-ai/sdk" "^0.37.0"
+    fast-xml-parser "^4.4.1"
+    zod "^3.22.4"
+    zod-to-json-schema "^3.22.4"
+
 "@langchain/community@^0.3.36":
   version "0.3.36"
   resolved "https://registry.yarnpkg.com/@langchain/community/-/community-0.3.36.tgz#e4c13b8f928b17e0f9257395f43be2246dfada40"
@@ -640,6 +668,14 @@
     zod "^3.22.4"
     zod-to-json-schema "^3.22.3"
 
+"@langchain/google-genai@^0.1.12":
+  version "0.1.12"
+  resolved "https://registry.yarnpkg.com/@langchain/google-genai/-/google-genai-0.1.12.tgz#6727253bda6f0d87cd74cf0bb6b1e0f398f60f32"
+  integrity sha512-0Ea0E2g63ejCuormVxbuoyJQ5BYN53i2/fb6WP8bMKzyh+y43R13V8JqOtr3e/GmgNyv3ou/VeaZjx7KAvu/0g==
+  dependencies:
+    "@google/generative-ai" "^0.24.0"
+    zod-to-json-schema "^3.22.4"
+
 "@langchain/openai@>=0.1.0 <0.5.0", "@langchain/openai@>=0.2.0 <0.5.0":
   version "0.4.5"
   resolved "https://registry.yarnpkg.com/@langchain/openai/-/openai-0.4.5.tgz#d18e207c3ec3f2ecaa4698a5a5888092f643da52"
@@ -2369,6 +2405,13 @@ fast-levenshtein@^2.0.6:
   resolved "https://registry.yarnpkg.com/fast-levenshtein/-/fast-levenshtein-2.0.6.tgz#3d8a5c66883a16a30ca8643e851f19baa7797917"
   integrity sha512-DCXu6Ifhqcks7TZKY3Hxp3y6qphY5SJZmrWMDrKcERSOXWQdMhU9Ig/PYrzyw/ul9jOIyh0N4M0tbC5hodg8dw==
 
+fast-xml-parser@^4.4.1:
+  version "4.5.3"
+  resolved "https://registry.yarnpkg.com/fast-xml-parser/-/fast-xml-parser-4.5.3.tgz#c54d6b35aa0f23dc1ea60b6c884340c006dc6efb"
+  integrity sha512-RKihhV+SHsIUGXObeVy9AXiBbFwkVk7Syp8XgwN5U3JV416+Gwp/GO9i0JYKmikykgz/UHRrrV4ROuZEo/T0ig==
+  dependencies:
+    strnum "^1.1.1"
+
 fastq@^1.6.0:
   version "1.17.1"
   resolved "https://registry.yarnpkg.com/fastq/-/fastq-1.17.1.tgz#2a523f07a4e7b1e81a42b91b8bf2254107753b47"
@@ -4458,6 +4501,11 @@ strip-json-comments@~2.0.1:
   resolved "https://registry.yarnpkg.com/strip-json-comments/-/strip-json-comments-2.0.1.tgz#3c531942e908c2697c0ec344858c286c7ca0a60a"
   integrity sha512-4gB8na07fecVVkOI6Rs4e7T6NOTki5EmL7TUduTs6bu3EdnSycntVJ4re8kgZA+wx9IueI2Y11bfbgwtzuE0KQ==
 
+strnum@^1.1.1:
+  version "1.1.2"
+  resolved "https://registry.yarnpkg.com/strnum/-/strnum-1.1.2.tgz#57bca4fbaa6f271081715dbc9ed7cee5493e28e4"
+  integrity sha512-vrN+B7DBIoTTZjnPNewwhx6cBA/H+IS7rfW68n7XxC1y7uoiGQBxaKzqucGUgavX15dJgiGztLJ8vxuEzwqBdA==
+
 styled-jsx@5.1.6:
   version "5.1.6"
   resolved "https://registry.yarnpkg.com/styled-jsx/-/styled-jsx-5.1.6.tgz#83b90c077e6c6a80f7f5e8781d0f311b2fe41499"
@@ -4955,6 +5003,11 @@ zod-to-json-schema@^3.22.3, zod-to-json-schema@^3.22.5:
   resolved "https://registry.yarnpkg.com/zod-to-json-schema/-/zod-to-json-schema-3.22.5.tgz#3646e81cfc318dbad2a22519e5ce661615418673"
   integrity sha512-+akaPo6a0zpVCCseDed504KBJUQpEW5QZw7RMneNmKw+fGaML1Z9tUNLnHHAC8x6dzVRO1eB2oEMyZRnuBZg7Q==
 
+zod-to-json-schema@^3.22.4:
+  version "3.24.5"
+  resolved "https://registry.yarnpkg.com/zod-to-json-schema/-/zod-to-json-schema-3.24.5.tgz#d1095440b147fb7c2093812a53c54df8d5df50a3"
+  integrity sha512-/AuWwMP+YqiPbsJx5D6TfgRTc4kTLjsh5SOcd4bLsfUg2RcEXrFMJl1DGgdHy2aCfsIA/cr/1JM0xcB2GZji8g==
+
 zod@^3.22.3, zod@^3.22.4:
   version "3.22.4"
   resolved "https://registry.yarnpkg.com/zod/-/zod-3.22.4.tgz#f31c3a9386f61b1f228af56faa9255e845cf3fff"