Merge 590a52d38c into 310c8a75fd

Larger instances will benefit from this change massively.
Also QWant was spamming the logs with some chaptcha problem so best to disable it for now.

docker-compose.yaml

@@ -26,9 +26,21 @@ services:
       - ./config.toml:/home/perplexica/config.toml
     restart: unless-stopped
 
+  redict:
+    image: registry.redict.io/redict:latest
+    container_name: perplexica-redict
+    ports:
+      - "6379:6379"
+    volumes:
+      - redict_data:/data
+    networks:
+      - perplexica-network
+    restart: unless-stopped
+
 networks:
   perplexica-network:
 
 volumes:
   backend-dbstore:
   uploads:
+  redict_data:

docs/API/SEARCH.md

@@ -32,8 +32,7 @@ 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
+  ]
 }
 ```
 
@@ -72,13 +71,11 @@ 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.
 
-#### Standard Response (stream: false)
+#### Example Response
 
 ```json
 {
@@ -103,28 +100,6 @@ 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.

searxng/settings.yml

@@ -12,6 +12,11 @@ search:
 server:
   secret_key: 'a2fb23f1b02e6ee83875b09826990de0f6bd908b6638e8c10277d415f6ab852b' # Is overwritten by ${SEARXNG_SECRET}
 
+redis:
+  url: redis://redict:6379/0
+
 engines:
   - name: wolframalpha
     disabled: false
+  - name: qwant
+    disabled: true

src/app/api/search/route.ts

@@ -33,7 +33,6 @@ interface ChatRequestBody {
   embeddingModel?: embeddingModel;
   query: string;
   history: Array<[string, string]>;
-  stream?: boolean;
 }
 
 export const POST = async (req: Request) => {
@@ -49,7 +48,6 @@ 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'
@@ -127,137 +125,40 @@ export const POST = async (req: Request) => {
       [],
     );
 
-    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) {
+    return new Promise(
+      (
+        resolve: (value: Response) => void,
+        reject: (value: Response) => void,
+      ) => {
+        let message = '';
         let sources: any[] = [];
 
-        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;
-
+        emitter.on('data', (data) => {
           try {
             const parsedData = JSON.parse(data);
-
             if (parsedData.type === 'response') {
-              controller.enqueue(
-                encoder.encode(
-                  JSON.stringify({
-                    type: 'response',
-                    data: parsedData.data,
-                  }) + '\n',
-                ),
-              );
+              message += parsedData.data;
             } else if (parsedData.type === 'sources') {
               sources = parsedData.data;
-              controller.enqueue(
-                encoder.encode(
-                  JSON.stringify({
-                    type: 'sources',
-                    data: sources,
-                  }) + '\n',
-                ),
-              );
             }
           } catch (error) {
-            controller.error(error);
+            reject(
+              Response.json({ message: 'Error parsing data' }, { status: 500 }),
+            );
           }
         });
 
         emitter.on('end', () => {
-          if (signal.aborted) return;
+          resolve(Response.json({ message, sources }, { status: 200 }));
+        });
 
-          controller.enqueue(
-            encoder.encode(
-              JSON.stringify({
-                type: 'done',
-              }) + '\n',
-            ),
+        emitter.on('error', (error) => {
+          reject(
+            Response.json({ message: 'Search error', error }, { status: 500 }),
           );
-          controller.close();
-        });
-
-        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/lib/providers/gemini.ts

@@ -8,10 +8,6 @@ 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',
@@ -21,8 +17,8 @@ const geminiChatModels: Record<string, string>[] = [
     key: 'gemini-2.0-flash-lite',
   },
   {
-    displayName: 'Gemini 2.0 Flash Thinking Experimental',
-    key: 'gemini-2.0-flash-thinking-exp-01-21',
+    displayName: 'Gemini 2.0 Pro Experimental',
+    key: 'gemini-2.0-pro-exp-02-05',
   },
   {
     displayName: 'Gemini 1.5 Flash',