@librechat/agents 2.4.317 → 2.4.318

package/src/tools/search/search.ts

@@ -2,8 +2,9 @@
 import axios from 'axios';
 import { RecursiveCharacterTextSplitter } from '@langchain/textsplitters';
 import type * as t from './types';
-import { getAttribution, FirecrawlScraper } from './firecrawl';
+import { FirecrawlScraper } from './firecrawl';
 import { BaseReranker } from './rerankers';
+import { getAttribution } from './utils';
 
 const chunker = {
   cleanText: (text: string): string => {
@@ -116,16 +117,11 @@ const getHighlights = async ({
 const createSerperAPI = (
   apiKey?: string
 ): {
-  getSources: (
-    query: string,
-    numResults?: number,
-    storedLocation?: string
-  ) => Promise<t.SearchResult>;
+  getSources: (params: t.GetSourcesParams) => Promise<t.SearchResult>;
 } => {
   const config = {
     apiKey: apiKey ?? process.env.SERPER_API_KEY,
     apiUrl: 'https://google.serper.dev/search',
-    defaultLocation: 'us',
     timeout: 10000,
   };
 
@@ -133,43 +129,46 @@ const createSerperAPI = (
     throw new Error('SERPER_API_KEY is required for SerperAPI');
   }
 
-  const getSources = async (
-    query: string,
-    numResults: number = 8,
-    storedLocation?: string
-  ): Promise<t.SearchResult> => {
+  const getSources = async ({
+    query,
+    country,
+    numResults = 8,
+  }: t.GetSourcesParams): Promise<t.SearchResult> => {
     if (!query.trim()) {
       return { success: false, error: 'Query cannot be empty' };
     }
 
     try {
-      const searchLocation = (
-        storedLocation ?? config.defaultLocation
-      ).toLowerCase();
-
-      const payload = {
+      const payload: t.SerperSearchPayload = {
         q: query,
         num: Math.min(Math.max(1, numResults), 10),
-        gl: searchLocation,
       };
 
-      const response = await axios.post(config.apiUrl, payload, {
-        headers: {
-          'X-API-KEY': config.apiKey,
-          'Content-Type': 'application/json',
-        },
-        timeout: config.timeout,
-      });
+      if (country != null && country !== '') {
+        payload['gl'] = country.toLowerCase();
+      }
+
+      const response = await axios.post<t.SerperResultData>(
+        config.apiUrl,
+        payload,
+        {
+          headers: {
+            'X-API-KEY': config.apiKey,
+            'Content-Type': 'application/json',
+          },
+          timeout: config.timeout,
+        }
+      );
 
       const data = response.data;
       const results: t.SearchResultData = {
         organic: data.organic,
         images: data.images ?? [],
+        answerBox: data.answerBox,
         topStories: data.topStories ?? [],
-        knowledgeGraph: data.knowledgeGraph as t.KnowledgeGraphResult,
-        answerBox: data.answerBox as t.AnswerBoxResult,
-        peopleAlsoAsk: data.peopleAlsoAsk as t.PeopleAlsoAskResult[],
-        relatedSearches: data.relatedSearches as string[],
+        peopleAlsoAsk: data.peopleAlsoAsk,
+        knowledgeGraph: data.knowledgeGraph,
+        relatedSearches: data.relatedSearches,
       };
 
       return { success: true, data: results };
@@ -187,11 +186,7 @@ const createSearXNGAPI = (
   instanceUrl?: string,
   apiKey?: string
 ): {
-  getSources: (
-    query: string,
-    numResults?: number,
-    storedLocation?: string
-  ) => Promise<t.SearchResult>;
+  getSources: (params: t.GetSourcesParams) => Promise<t.SearchResult>;
 } => {
   const config = {
     instanceUrl: instanceUrl ?? process.env.SEARXNG_INSTANCE_URL,
@@ -204,11 +199,10 @@ const createSearXNGAPI = (
     throw new Error('SEARXNG_INSTANCE_URL is required for SearXNG API');
   }
 
-  const getSources = async (
-    query: string,
-    numResults: number = 8,
-    storedLocation?: string
-  ): Promise<t.SearchResult> => {
+  const getSources = async ({
+    query,
+    numResults = 8,
+  }: t.GetSourcesParams): Promise<t.SearchResult> => {
     if (!query.trim()) {
       return { success: false, error: 'Query cannot be empty' };
     }
@@ -225,7 +219,7 @@ const createSearXNGAPI = (
       }
 
       // Prepare parameters for SearXNG
-      const params: Record<string, string | number> = {
+      const params: t.SearxNGSearchPayload = {
         q: query,
         format: 'json',
         pageno: 1,
@@ -233,13 +227,8 @@ const createSearXNGAPI = (
         language: 'all',
         safesearch: 0,
         engines: 'google,bing,duckduckgo',
-        max_results: Math.min(Math.max(1, numResults), 20),
       };
 
-      if (storedLocation != null && storedLocation !== 'all') {
-        params.language = storedLocation;
-      }
-
       const headers: Record<string, string> = {
         'Content-Type': 'application/json',
       };
@@ -301,11 +290,7 @@ const createSearXNGAPI = (
 export const createSearchAPI = (
   config: t.SearchConfig
 ): {
-  getSources: (
-    query: string,
-    numResults?: number,
-    storedLocation?: string
-  ) => Promise<t.SearchResult>;
+  getSources: (params: t.GetSourcesParams) => Promise<t.SearchResult>;
 } => {
   const {
     searchProvider = 'serper',
@@ -366,12 +351,14 @@ export const createSourceProcessor = (
             .then(([url, response]) => {
               const attribution = getAttribution(url, response.data?.metadata);
               if (response.success && response.data) {
-                const content = firecrawlScraper.extractContent(response);
+                const [content, references] =
+                  firecrawlScraper.extractContent(response);
                 return {
                   url,
+                  references,
                   attribution,
                   content: chunker.cleanText(content),
-                };
+                } as t.ScrapeResult;
               }
 
               return {
@@ -379,7 +366,7 @@ export const createSourceProcessor = (
                 attribution,
                 error: true,
                 content: `Failed to scrape ${url}: ${response.error ?? 'Unknown error'}`,
-              };
+              } as t.ScrapeResult;
             })
             .then(async (result) => {
               try {
@@ -443,10 +430,11 @@ export const createSourceProcessor = (
       if (result.error === true) {
         continue;
       }
-      const { url, content, attribution, highlights } = result;
+      const { url, content, attribution, references, highlights } = result;
       onContentScraped?.(url, {
         content,
         attribution,
+        references,
         highlights,
       });
     }