@librechat/agents 2.4.319 → 2.4.320

package/src/tools/search/tool.ts

@@ -1,6 +1,7 @@
 /* eslint-disable no-console */
 import { z } from 'zod';
 import { tool, DynamicStructuredTool } from '@langchain/core/tools';
+import type { RunnableConfig } from '@langchain/core/runnables';
 import type * as t from './types';
 import { createSearchAPI, createSourceProcessor } from './search';
 import { createFirecrawlScraper } from './firecrawl';
@@ -38,6 +39,129 @@ Examples:
 - "in" for India
 `.trim();
 
+function createSearchProcessor({
+  searchAPI,
+  sourceProcessor,
+  onGetHighlights,
+}: {
+  searchAPI: ReturnType<typeof createSearchAPI>;
+  sourceProcessor: ReturnType<typeof createSourceProcessor>;
+  onGetHighlights: t.SearchToolConfig['onGetHighlights'];
+}) {
+  return async function ({
+    query,
+    country,
+    proMode = true,
+    maxSources = 5,
+    onSearchResults,
+  }: {
+    query: string;
+    country?: string;
+    maxSources?: number;
+    proMode?: boolean;
+    onSearchResults: t.SearchToolConfig['onSearchResults'];
+  }): Promise<t.SearchResultData> {
+    try {
+      const result = await searchAPI.getSources({ query, country });
+      onSearchResults?.(result);
+
+      if (!result.success) {
+        throw new Error(result.error ?? 'Search failed');
+      }
+
+      const processedSources = await sourceProcessor.processSources({
+        query,
+        result,
+        proMode,
+        onGetHighlights,
+        numElements: maxSources,
+      });
+      return expandHighlights(processedSources);
+    } catch (error) {
+      console.error('Error in search:', error);
+      return {
+        organic: [],
+        topStories: [],
+        images: [],
+        relatedSearches: [],
+        error: error instanceof Error ? error.message : String(error),
+      };
+    }
+  };
+}
+
+function createOnSearchResults({
+  runnableConfig,
+  onSearchResults,
+}: {
+  runnableConfig: RunnableConfig;
+  onSearchResults: t.SearchToolConfig['onSearchResults'];
+}) {
+  return function (results: t.SearchResult): void {
+    if (!onSearchResults) {
+      return;
+    }
+    onSearchResults(results, runnableConfig);
+  };
+}
+
+function createTool({
+  schema,
+  search,
+  onSearchResults: _onSearchResults,
+}: {
+  schema: t.SearchToolSchema;
+  search: ReturnType<typeof createSearchProcessor>;
+  onSearchResults: t.SearchToolConfig['onSearchResults'];
+}): DynamicStructuredTool<typeof schema> {
+  return tool<typeof schema>(
+    async (params, runnableConfig) => {
+      const { query, country: _c } = params;
+      const country = typeof _c === 'string' && _c ? _c : undefined;
+      const searchResult = await search({
+        query,
+        country,
+        onSearchResults: createOnSearchResults({
+          runnableConfig,
+          onSearchResults: _onSearchResults,
+        }),
+      });
+      const turn = runnableConfig.toolCall?.turn ?? 0;
+      const { output, references } = formatResultsForLLM(turn, searchResult);
+      const data: t.SearchResultData = { turn, ...searchResult, references };
+      return [output, { [Constants.WEB_SEARCH]: data }];
+    },
+    {
+      name: Constants.WEB_SEARCH,
+      description: `Real-time search. Results have required citation anchors.
+
+Note: Use ONCE per reply unless instructed otherwise.
+
+Anchors:
+- \\ue202turnXtypeY
+- X = turn idx, type = 'search' | 'news' | 'image' | 'ref', Y = item idx
+
+Special Markers:
+- \\ue203...\\ue204 — highlight start/end of cited text (for Standalone or Group citations)
+- \\ue200...\\ue201 — group block (e.g. \\ue200\\ue202turn0search1\\ue202turn0news2\\ue201)
+
+**CITE EVERY NON-OBVIOUS FACT/QUOTE:**
+Use anchor marker(s) immediately after the statement:
+- Standalone: "Pure functions produce same output. \\ue202turn0search0"
+- Standalone (multiple): "Today's News \\ue202turn0search0\\ue202turn0news0"
+- Highlight: "\\ue203Highlight text.\\ue204\\ue202turn0news1"
+- Group: "Sources. \\ue200\\ue202turn0search0\\ue202turn0news1\\ue201"
+- Group Highlight: "\\ue203Highlight for group.\\ue204 \\ue200\\ue202turn0search0\\ue202turn0news1\\ue201"
+- Image: "See photo \\ue202turn0image0."
+
+**NEVER use markdown links, [1], or footnotes. CITE ONLY with anchors provided.**
+`.trim(),
+      schema: schema,
+      responseFormat: Constants.CONTENT_AND_ARTIFACT,
+    }
+  );
+}
+
 /**
  * Creates a search tool with a schema that dynamically includes the country field
  * only when the searchProvider is 'serper'.
@@ -47,7 +171,7 @@ Examples:
  */
 export const createSearchTool = (
   config: t.SearchToolConfig = {}
-): DynamicStructuredTool<typeof SearchToolSchema> => {
+): DynamicStructuredTool<typeof toolSchema> => {
   const {
     searchProvider = 'serper',
     serperApiKey,
@@ -63,6 +187,7 @@ export const createSearchTool = (
     jinaApiKey,
     cohereApiKey,
     onSearchResults: _onSearchResults,
+    onGetHighlights,
   } = config;
 
   const querySchema = z.string().describe(DEFAULT_QUERY_DESCRIPTION);
@@ -80,7 +205,7 @@ export const createSearchTool = (
       .describe(DEFAULT_COUNTRY_DESCRIPTION);
   }
 
-  const SearchToolSchema = z.object(schemaObject);
+  const toolSchema = z.object(schemaObject);
 
   const searchAPI = createSearchAPI({
     searchProvider,
@@ -115,92 +240,15 @@ export const createSearchTool = (
     firecrawlScraper
   );
 
-  const search = async ({
-    query,
-    country,
-    proMode = true,
-    maxSources = 5,
-    onSearchResults,
-  }: {
-    query: string;
-    country?: string;
-    maxSources?: number;
-    proMode?: boolean;
-    onSearchResults?: (sources: t.SearchResult) => void;
-  }): Promise<t.SearchResultData> => {
-    try {
-      const sources = await searchAPI.getSources({ query, country });
-      onSearchResults?.(sources);
-
-      if (!sources.success) {
-        throw new Error(sources.error ?? 'Search failed');
-      }
-
-      const processedSources = await sourceProcessor.processSources(
-        sources,
-        maxSources,
-        query,
-        proMode
-      );
-      return expandHighlights(processedSources);
-    } catch (error) {
-      console.error('Error in search:', error);
-      return {
-        organic: [],
-        topStories: [],
-        images: [],
-        relatedSearches: [],
-        error: error instanceof Error ? error.message : String(error),
-      };
-    }
-  };
-
-  return tool<typeof SearchToolSchema>(
-    async (params, runnableConfig) => {
-      const { query, country: _c } = params;
-      const country = typeof _c === 'string' && _c ? _c : undefined;
-      const searchResult = await search({
-        query,
-        country,
-        onSearchResults: _onSearchResults
-          ? (result): void => {
-            _onSearchResults(result, runnableConfig);
-          }
-          : undefined,
-      });
-      const turn = runnableConfig.toolCall?.turn ?? 0;
-      const { output, references } = formatResultsForLLM(turn, searchResult);
-      const data: t.SearchResultData = { turn, ...searchResult, references };
-      return [output, { [Constants.WEB_SEARCH]: data }];
-    },
-    {
-      name: Constants.WEB_SEARCH,
-      description: `
-Real-time search. Results have required citation anchors.
-
-Note: Use ONCE per reply unless instructed otherwise.
-
-Anchors:
-- \\ue202turnXtypeY
-- X = turn idx, type = 'search' | 'news' | 'image' | 'ref', Y = item idx
-
-Special Markers:
-- \\ue203...\\ue204 — highlight start/end of cited text (for Standalone or Group citations)
-- \\ue200...\\ue201 — group block (e.g. \\ue200\\ue202turn0search1\\ue202turn0news2\\ue201)
-
-**CITE EVERY NON-OBVIOUS FACT/QUOTE:**
-Use anchor marker(s) immediately after the statement:
-- Standalone: "Pure functions produce same output. \\ue202turn0search0"
-- Standalone (multiple): "Today's News \\ue202turn0search0\\ue202turn0news0"
-- Highlight: "\\ue203Highlight text.\\ue204\\ue202turn0news1"
-- Group: "Sources. \\ue200\\ue202turn0search0\\ue202turn0news1\\ue201"
-- Group Highlight: "\\ue203Highlight for group.\\ue204 \\ue200\\ue202turn0search0\\ue202turn0news1\\ue201"
-- Image: "See photo \\ue202turn0image0."
+  const search = createSearchProcessor({
+    searchAPI,
+    sourceProcessor,
+    onGetHighlights,
+  });
 
-**NEVER use markdown links, [1], or footnotes. CITE ONLY with anchors provided.**
-`.trim(),
-      schema: SearchToolSchema,
-      responseFormat: Constants.CONTENT_AND_ARTIFACT,
-    }
-  );
+  return createTool({
+    search,
+    schema: toolSchema,
+    onSearchResults: _onSearchResults,
+  });
 };

package/src/tools/search/types.ts

@@ -1,3 +1,4 @@
+import { z } from 'zod';
 import type { RunnableConfig } from '@langchain/core/runnables';
 import type { BaseReranker } from './rerankers';
 
@@ -143,6 +144,7 @@ export interface SearchToolConfig
     results: SearchResult,
     runnableConfig?: RunnableConfig
   ) => void;
+  onGetHighlights?: (link: string) => void;
 }
 export interface MediaReference {
   originalUrl: string;
@@ -205,7 +207,7 @@ export interface ScrapeMetadata {
   publishedTime?: string;
   modifiedTime?: string;
   // Twitter metadata
-  'twitter:site'?: string;
+  'twitter:site'?: string | boolean | number | null;
   'twitter:creator'?: string;
   'twitter:card'?: string;
   'twitter:image'?: string;
@@ -555,3 +557,28 @@ export interface SearXNGResult {
   publishedDate?: string;
   img_src?: string;
 }
+
+export type ProcessSourcesFields = {
+  result: SearchResult;
+  numElements: number;
+  query: string;
+  proMode: boolean;
+  onGetHighlights: SearchToolConfig['onGetHighlights'];
+};
+
+export type SearchToolSchema = z.ZodObject<
+  {
+    query: z.ZodString;
+    country?: z.ZodOptional<z.ZodString>;
+  },
+  'strip',
+  z.ZodTypeAny,
+  {
+    query: string;
+    country?: unknown;
+  },
+  {
+    query: string;
+    country?: unknown;
+  }
+>;

package/src/tools/search/utils.ts

@@ -25,11 +25,15 @@ export function getAttribution(
 ): string | undefined {
   if (!metadata) return getDomainName(link, metadata);
 
+  const twitterSite = metadata['twitter:site'];
+  const twitterSiteFormatted =
+    typeof twitterSite === 'string' ? twitterSite.replace(/^@/, '') : undefined;
+
   const possibleAttributions = [
     metadata.ogSiteName,
     metadata['og:site_name'],
     metadata.title?.split('|').pop()?.trim(),
-    metadata['twitter:site']?.replace(/^@/, ''),
+    twitterSiteFormatted,
   ];
 
   const attribution = possibleAttributions.find(