@librechat/agents 2.4.319 → 2.4.320

package/dist/cjs/tools/search/search.cjs

@@ -40,7 +40,7 @@ const chunker = {
         return Promise.all(promises);
     },
 };
-const createSourceUpdateCallback = (sourceMap) => {
+function createSourceUpdateCallback(sourceMap) {
     return (link, update) => {
         const source = sourceMap.get(link);
         if (source) {
@@ -50,7 +50,7 @@ const createSourceUpdateCallback = (sourceMap) => {
             });
         }
     };
-};
+}
 const getHighlights = async ({ query, content, reranker, topResults = 5, }) => {
     if (!content) {
         console.warn('No content provided for highlights');
@@ -225,11 +225,12 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
     reranker, } = config;
     const firecrawlScraper = scraperInstance;
     const webScraper = {
-        scrapeMany: async ({ query, links, }) => {
+        scrapeMany: async ({ query, links, onGetHighlights, }) => {
             console.log(`Scraping ${links.length} links with Firecrawl`);
             const promises = [];
             try {
-                for (const currentLink of links) {
+                for (let i = 0; i < links.length; i++) {
+                    const currentLink = links[i];
                     const promise = firecrawlScraper
                         .scrapeUrl(currentLink, {})
                         .then(([url, response]) => {
@@ -263,6 +264,9 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
                                 reranker,
                                 content: result.content,
                             });
+                            if (onGetHighlights) {
+                                onGetHighlights(result.url);
+                            }
                             return {
                                 ...result,
                                 highlights,
@@ -293,10 +297,14 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
             }
         },
     };
-    const fetchContents = async ({ links, query, target, onContentScraped, }) => {
+    const fetchContents = async ({ links, query, target, onGetHighlights, onContentScraped, }) => {
         const initialLinks = links.slice(0, target);
         // const remainingLinks = links.slice(target).reverse();
-        const results = await webScraper.scrapeMany({ query, links: initialLinks });
+        const results = await webScraper.scrapeMany({
+            query,
+            links: initialLinks,
+            onGetHighlights,
+        });
         for (const result of results) {
             if (result.error === true) {
                 continue;
@@ -310,7 +318,7 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
             });
         }
     };
-    const processSources = async (result, numElements, query, proMode = false) => {
+    const processSources = async ({ result, numElements, query, proMode = true, onGetHighlights, }) => {
         try {
             if (!result.data) {
                 return {
@@ -334,6 +342,7 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
                 await fetchContents({
                     query,
                     target: 1,
+                    onGetHighlights,
                     onContentScraped,
                     links: [wikiSources[0].link],
                 });
@@ -362,8 +371,9 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
             }
             const onContentScraped = createSourceUpdateCallback(sourceMap);
             await fetchContents({
-                links: allLinks,
                 query,
+                links: allLinks,
+                onGetHighlights,
                 onContentScraped,
                 target: numElements,
             });

package/dist/cjs/tools/search/search.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"search.cjs","sources":["../../../../src/tools/search/search.ts"],"sourcesContent":["/* eslint-disable no-console */\nimport axios from 'axios';\nimport { RecursiveCharacterTextSplitter } from '@langchain/textsplitters';\nimport type * as t from './types';\nimport { FirecrawlScraper } from './firecrawl';\nimport { BaseReranker } from './rerankers';\nimport { getAttribution } from './utils';\n\nconst chunker = {\n  cleanText: (text: string): string => {\n    if (!text) return '';\n\n    /** Normalized all line endings to '\\n' */\n    const normalizedText = text.replace(/\\r\\n/g, '\\n').replace(/\\r/g, '\\n');\n\n    /** Handle multiple backslashes followed by newlines\n     * This replaces patterns like '\\\\\\\\\\\\n' with a single newline */\n    const fixedBackslashes = normalizedText.replace(/\\\\+\\n/g, '\\n');\n\n    /** Cleaned up consecutive newlines, tabs, and spaces around newlines */\n    const cleanedNewlines = fixedBackslashes.replace(/[\\t ]*\\n[\\t \\n]*/g, '\\n');\n\n    /** Cleaned up excessive spaces and tabs */\n    const cleanedSpaces = cleanedNewlines.replace(/[ \\t]+/g, ' ');\n\n    return cleanedSpaces.trim();\n  },\n  splitText: async (\n    text: string,\n    options?: {\n      chunkSize?: number;\n      chunkOverlap?: number;\n      separators?: string[];\n    }\n  ): Promise<string[]> => {\n    const chunkSize = options?.chunkSize ?? 150;\n    const chunkOverlap = options?.chunkOverlap ?? 50;\n    const separators = options?.separators || ['\\n\\n', '\\n'];\n\n    const splitter = new RecursiveCharacterTextSplitter({\n      separators,\n      chunkSize,\n      chunkOverlap,\n    });\n\n    return await splitter.splitText(text);\n  },\n\n  splitTexts: async (\n    texts: string[],\n    options?: {\n      chunkSize?: number;\n      chunkOverlap?: number;\n      separators?: string[];\n    }\n  ): Promise<string[][]> => {\n    // Split multiple texts\n    const promises = texts.map((text) =>\n      chunker.splitText(text, options).catch((error) => {\n        console.error('Error splitting text:', error);\n        return [text];\n      })\n    );\n    return Promise.all(promises);\n  },\n};\n\nconst createSourceUpdateCallback = (sourceMap: Map<string, t.ValidSource>) => {\n  return (link: string, update?: Partial<t.ValidSource>): void => {\n    const source = sourceMap.get(link);\n    if (source) {\n      sourceMap.set(link, {\n        ...source,\n        ...update,\n      });\n    }\n  };\n};\n\nconst getHighlights = async ({\n  query,\n  content,\n  reranker,\n  topResults = 5,\n}: {\n  content: string;\n  query: string;\n  reranker?: BaseReranker;\n  topResults?: number;\n}): Promise<t.Highlight[] | undefined> => {\n  if (!content) {\n    console.warn('No content provided for highlights');\n    return;\n  }\n  if (!reranker) {\n    console.warn('No reranker provided for highlights');\n    return;\n  }\n\n  try {\n    const documents = await chunker.splitText(content);\n    if (Array.isArray(documents)) {\n      return await reranker.rerank(query, documents, topResults);\n    } else {\n      console.error(\n        'Expected documents to be an array, got:',\n        typeof documents\n      );\n      return;\n    }\n  } catch (error) {\n    console.error('Error in content processing:', error);\n    return;\n  }\n};\n\nconst createSerperAPI = (\n  apiKey?: string\n): {\n  getSources: (params: t.GetSourcesParams) => Promise<t.SearchResult>;\n} => {\n  const config = {\n    apiKey: apiKey ?? process.env.SERPER_API_KEY,\n    apiUrl: 'https://google.serper.dev/search',\n    timeout: 10000,\n  };\n\n  if (config.apiKey == null || config.apiKey === '') {\n    throw new Error('SERPER_API_KEY is required for SerperAPI');\n  }\n\n  const getSources = async ({\n    query,\n    country,\n    numResults = 8,\n  }: t.GetSourcesParams): Promise<t.SearchResult> => {\n    if (!query.trim()) {\n      return { success: false, error: 'Query cannot be empty' };\n    }\n\n    try {\n      const payload: t.SerperSearchPayload = {\n        q: query,\n        num: Math.min(Math.max(1, numResults), 10),\n      };\n\n      if (country != null && country !== '') {\n        payload['gl'] = country.toLowerCase();\n      }\n\n      const response = await axios.post<t.SerperResultData>(\n        config.apiUrl,\n        payload,\n        {\n          headers: {\n            'X-API-KEY': config.apiKey,\n            'Content-Type': 'application/json',\n          },\n          timeout: config.timeout,\n        }\n      );\n\n      const data = response.data;\n      const results: t.SearchResultData = {\n        organic: data.organic,\n        images: data.images ?? [],\n        answerBox: data.answerBox,\n        topStories: data.topStories ?? [],\n        peopleAlsoAsk: data.peopleAlsoAsk,\n        knowledgeGraph: data.knowledgeGraph,\n        relatedSearches: data.relatedSearches,\n      };\n\n      return { success: true, data: results };\n    } catch (error) {\n      const errorMessage =\n        error instanceof Error ? error.message : String(error);\n      return { success: false, error: `API request failed: ${errorMessage}` };\n    }\n  };\n\n  return { getSources };\n};\n\nconst createSearXNGAPI = (\n  instanceUrl?: string,\n  apiKey?: string\n): {\n  getSources: (params: t.GetSourcesParams) => Promise<t.SearchResult>;\n} => {\n  const config = {\n    instanceUrl: instanceUrl ?? process.env.SEARXNG_INSTANCE_URL,\n    apiKey: apiKey ?? process.env.SEARXNG_API_KEY,\n    defaultLocation: 'all',\n    timeout: 10000,\n  };\n\n  if (config.instanceUrl == null || config.instanceUrl === '') {\n    throw new Error('SEARXNG_INSTANCE_URL is required for SearXNG API');\n  }\n\n  const getSources = async ({\n    query,\n    numResults = 8,\n  }: t.GetSourcesParams): Promise<t.SearchResult> => {\n    if (!query.trim()) {\n      return { success: false, error: 'Query cannot be empty' };\n    }\n\n    try {\n      // Ensure the instance URL ends with /search\n      if (config.instanceUrl == null || config.instanceUrl === '') {\n        return { success: false, error: 'Instance URL is not defined' };\n      }\n\n      let searchUrl = config.instanceUrl;\n      if (!searchUrl.endsWith('/search')) {\n        searchUrl = searchUrl.replace(/\\/$/, '') + '/search';\n      }\n\n      // Prepare parameters for SearXNG\n      const params: t.SearxNGSearchPayload = {\n        q: query,\n        format: 'json',\n        pageno: 1,\n        categories: 'general',\n        language: 'all',\n        safesearch: 0,\n        engines: 'google,bing,duckduckgo',\n      };\n\n      const headers: Record<string, string> = {\n        'Content-Type': 'application/json',\n      };\n\n      if (config.apiKey != null && config.apiKey !== '') {\n        headers['X-API-Key'] = config.apiKey;\n      }\n\n      const response = await axios.get(searchUrl, {\n        headers,\n        params,\n        timeout: config.timeout,\n      });\n\n      const data = response.data;\n\n      // Transform SearXNG results to match SerperAPI format\n      const organicResults = (data.results ?? [])\n        .slice(0, numResults)\n        .map((result: t.SearXNGResult) => ({\n          title: result.title ?? '',\n          link: result.url ?? '',\n          snippet: result.content ?? '',\n          date: result.publishedDate ?? '',\n        }));\n\n      // Extract image results if available\n      const imageResults = (data.results ?? [])\n        .filter((result: t.SearXNGResult) => result.img_src)\n        .slice(0, 6)\n        .map((result: t.SearXNGResult) => ({\n          title: result.title ?? '',\n          imageUrl: result.img_src ?? '',\n        }));\n\n      // Format results to match SerperAPI structure\n      const results: t.SearchResultData = {\n        organic: organicResults,\n        images: imageResults,\n        topStories: [],\n        // Use undefined instead of null for optional properties\n        relatedSearches: data.suggestions ?? [],\n      };\n\n      return { success: true, data: results };\n    } catch (error) {\n      const errorMessage =\n        error instanceof Error ? error.message : String(error);\n      return {\n        success: false,\n        error: `SearXNG API request failed: ${errorMessage}`,\n      };\n    }\n  };\n\n  return { getSources };\n};\n\nexport const createSearchAPI = (\n  config: t.SearchConfig\n): {\n  getSources: (params: t.GetSourcesParams) => Promise<t.SearchResult>;\n} => {\n  const {\n    searchProvider = 'serper',\n    serperApiKey,\n    searxngInstanceUrl,\n    searxngApiKey,\n  } = config;\n\n  if (searchProvider.toLowerCase() === 'serper') {\n    return createSerperAPI(serperApiKey);\n  } else if (searchProvider.toLowerCase() === 'searxng') {\n    return createSearXNGAPI(searxngInstanceUrl, searxngApiKey);\n  } else {\n    throw new Error(\n      `Invalid search provider: ${searchProvider}. Must be 'serper' or 'searxng'`\n    );\n  }\n};\n\nexport const createSourceProcessor = (\n  config: t.ProcessSourcesConfig = {},\n  scraperInstance?: FirecrawlScraper\n): {\n  processSources: (\n    result: t.SearchResult,\n    numElements: number,\n    query: string,\n    proMode?: boolean\n  ) => Promise<t.SearchResultData>;\n  topResults: number;\n} => {\n  if (!scraperInstance) {\n    throw new Error('Firecrawl scraper instance is required');\n  }\n  const {\n    topResults = 5,\n    // strategies = ['no_extraction'],\n    // filterContent = true,\n    reranker,\n  } = config;\n\n  const firecrawlScraper = scraperInstance;\n\n  const webScraper = {\n    scrapeMany: async ({\n      query,\n      links,\n    }: {\n      query: string;\n      links: string[];\n    }): Promise<Array<t.ScrapeResult>> => {\n      console.log(`Scraping ${links.length} links with Firecrawl`);\n      const promises: Array<Promise<t.ScrapeResult>> = [];\n      try {\n        for (const currentLink of links) {\n          const promise: Promise<t.ScrapeResult> = firecrawlScraper\n            .scrapeUrl(currentLink, {})\n            .then(([url, response]) => {\n              const attribution = getAttribution(url, response.data?.metadata);\n              if (response.success && response.data) {\n                const [content, references] =\n                  firecrawlScraper.extractContent(response);\n                return {\n                  url,\n                  references,\n                  attribution,\n                  content: chunker.cleanText(content),\n                } as t.ScrapeResult;\n              }\n\n              return {\n                url,\n                attribution,\n                error: true,\n                content: `Failed to scrape ${url}: ${response.error ?? 'Unknown error'}`,\n              } as t.ScrapeResult;\n            })\n            .then(async (result) => {\n              try {\n                if (result.error != null) {\n                  console.error(\n                    `Error scraping ${result.url}: ${result.content}`\n                  );\n                  return {\n                    ...result,\n                  };\n                }\n                const highlights = await getHighlights({\n                  query,\n                  reranker,\n                  content: result.content,\n                });\n                return {\n                  ...result,\n                  highlights,\n                };\n              } catch (error) {\n                console.error('Error processing scraped content:', error);\n                return {\n                  ...result,\n                };\n              }\n            })\n            .catch((error) => {\n              console.error(`Error scraping ${currentLink}:`, error);\n              return {\n                url: currentLink,\n                error: true,\n                content: `Failed to scrape ${currentLink}: ${error.message ?? 'Unknown error'}`,\n              };\n            });\n          promises.push(promise);\n        }\n        return await Promise.all(promises);\n      } catch (error) {\n        console.error('Error in scrapeMany:', error);\n        return [];\n      }\n    },\n  };\n\n  const fetchContents = async ({\n    links,\n    query,\n    target,\n    onContentScraped,\n  }: {\n    links: string[];\n    query: string;\n    target: number;\n    onContentScraped?: (link: string, update?: Partial<t.ValidSource>) => void;\n  }): Promise<void> => {\n    const initialLinks = links.slice(0, target);\n    // const remainingLinks = links.slice(target).reverse();\n    const results = await webScraper.scrapeMany({ query, links: initialLinks });\n    for (const result of results) {\n      if (result.error === true) {\n        continue;\n      }\n      const { url, content, attribution, references, highlights } = result;\n      onContentScraped?.(url, {\n        content,\n        attribution,\n        references,\n        highlights,\n      });\n    }\n  };\n\n  const processSources = async (\n    result: t.SearchResult,\n    numElements: number,\n    query: string,\n    proMode: boolean = false\n  ): Promise<t.SearchResultData> => {\n    try {\n      if (!result.data) {\n        return {\n          organic: [],\n          topStories: [],\n          images: [],\n          relatedSearches: [],\n        };\n      } else if (!result.data.organic) {\n        return result.data;\n      }\n\n      if (!proMode) {\n        const wikiSources = result.data.organic.filter((source) =>\n          source.link.includes('wikipedia.org')\n        );\n\n        if (!wikiSources.length) {\n          return result.data;\n        }\n\n        const wikiSourceMap = new Map<string, t.ValidSource>();\n        wikiSourceMap.set(wikiSources[0].link, wikiSources[0]);\n        const onContentScraped = createSourceUpdateCallback(wikiSourceMap);\n        await fetchContents({\n          query,\n          target: 1,\n          onContentScraped,\n          links: [wikiSources[0].link],\n        });\n\n        for (let i = 0; i < result.data.organic.length; i++) {\n          const source = result.data.organic[i];\n          const updatedSource = wikiSourceMap.get(source.link);\n          if (updatedSource) {\n            result.data.organic[i] = {\n              ...source,\n              ...updatedSource,\n            };\n          }\n        }\n\n        return result.data;\n      }\n\n      const sourceMap = new Map<string, t.ValidSource>();\n      const allLinks: string[] = [];\n\n      for (const source of result.data.organic) {\n        if (source.link) {\n          allLinks.push(source.link);\n          sourceMap.set(source.link, source);\n        }\n      }\n\n      if (allLinks.length === 0) {\n        return result.data;\n      }\n\n      const onContentScraped = createSourceUpdateCallback(sourceMap);\n      await fetchContents({\n        links: allLinks,\n        query,\n        onContentScraped,\n        target: numElements,\n      });\n\n      for (let i = 0; i < result.data.organic.length; i++) {\n        const source = result.data.organic[i];\n        const updatedSource = sourceMap.get(source.link);\n        if (updatedSource) {\n          result.data.organic[i] = {\n            ...source,\n            ...updatedSource,\n          };\n        }\n      }\n\n      const successfulSources = result.data.organic\n        .filter(\n          (source) =>\n            source.content != null && !source.content.startsWith('Failed')\n        )\n        .slice(0, numElements);\n\n      if (successfulSources.length > 0) {\n        result.data.organic = successfulSources;\n      }\n      return result.data;\n    } catch (error) {\n      console.error('Error in processSources:', error);\n      return {\n        organic: [],\n        topStories: [],\n        images: [],\n        relatedSearches: [],\n        ...result.data,\n        error: error instanceof Error ? error.message : String(error),\n      };\n    }\n  };\n\n  return {\n    processSources,\n    topResults,\n  };\n};\n"],"names":["RecursiveCharacterTextSplitter","getAttribution"],"mappings":";;;;;;AAAA;AAQA,MAAM,OAAO,GAAG;AACd,IAAA,SAAS,EAAE,CAAC,IAAY,KAAY;AAClC,QAAA,IAAI,CAAC,IAAI;AAAE,YAAA,OAAO,EAAE;;AAGpB,QAAA,MAAM,cAAc,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,IAAI,CAAC;AAEvE;AACiE;QACjE,MAAM,gBAAgB,GAAG,cAAc,CAAC,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC;;QAG/D,MAAM,eAAe,GAAG,gBAAgB,CAAC,OAAO,CAAC,mBAAmB,EAAE,IAAI,CAAC;;QAG3E,MAAM,aAAa,GAAG,eAAe,CAAC,OAAO,CAAC,SAAS,EAAE,GAAG,CAAC;AAE7D,QAAA,OAAO,aAAa,CAAC,IAAI,EAAE;KAC5B;AACD,IAAA,SAAS,EAAE,OACT,IAAY,EACZ,OAIC,KACoB;AACrB,QAAA,MAAM,SAAS,GAAG,OAAO,EAAE,SAAS,IAAI,GAAG;AAC3C,QAAA,MAAM,YAAY,GAAG,OAAO,EAAE,YAAY,IAAI,EAAE;QAChD,MAAM,UAAU,GAAG,OAAO,EAAE,UAAU,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC;AAExD,QAAA,MAAM,QAAQ,GAAG,IAAIA,4CAA8B,CAAC;YAClD,UAAU;YACV,SAAS;YACT,YAAY;AACb,SAAA,CAAC;AAEF,QAAA,OAAO,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC;KACtC;AAED,IAAA,UAAU,EAAE,OACV,KAAe,EACf,OAIC,KACsB;;QAEvB,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,KAC9B,OAAO,CAAC,SAAS,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,KAAI;AAC/C,YAAA,OAAO,CAAC,KAAK,CAAC,uBAAuB,EAAE,KAAK,CAAC;YAC7C,OAAO,CAAC,IAAI,CAAC;SACd,CAAC,CACH;AACD,QAAA,OAAO,OAAO,CAAC,GAAG,CAAC,QAAQ,CAAC;KAC7B;CACF;AAED,MAAM,0BAA0B,GAAG,CAAC,SAAqC,KAAI;AAC3E,IAAA,OAAO,CAAC,IAAY,EAAE,MAA+B,KAAU;QAC7D,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC;QAClC,IAAI,MAAM,EAAE;AACV,YAAA,SAAS,CAAC,GAAG,CAAC,IAAI,EAAE;AAClB,gBAAA,GAAG,MAAM;AACT,gBAAA,GAAG,MAAM;AACV,aAAA,CAAC;;AAEN,KAAC;AACH,CAAC;AAED,MAAM,aAAa,GAAG,OAAO,EAC3B,KAAK,EACL,OAAO,EACP,QAAQ,EACR,UAAU,GAAG,CAAC,GAMf,KAAwC;IACvC,IAAI,CAAC,OAAO,EAAE;AACZ,QAAA,OAAO,CAAC,IAAI,CAAC,oCAAoC,CAAC;QAClD;;IAEF,IAAI,CAAC,QAAQ,EAAE;AACb,QAAA,OAAO,CAAC,IAAI,CAAC,qCAAqC,CAAC;QACnD;;AAGF,IAAA,IAAI;QACF,MAAM,SAAS,GAAG,MAAM,OAAO,CAAC,SAAS,CAAC,OAAO,CAAC;AAClD,QAAA,IAAI,KAAK,CAAC,OAAO,CAAC,SAAS,CAAC,EAAE;YAC5B,OAAO,MAAM,QAAQ,CAAC,MAAM,CAAC,KAAK,EAAE,SAAS,EAAE,UAAU,CAAC;;aACrD;YACL,OAAO,CAAC,KAAK,CACX,yCAAyC,EACzC,OAAO,SAAS,CACjB;YACD;;;IAEF,OAAO,KAAK,EAAE;AACd,QAAA,OAAO,CAAC,KAAK,CAAC,8BAA8B,EAAE,KAAK,CAAC;QACpD;;AAEJ,CAAC;AAED,MAAM,eAAe,GAAG,CACtB,MAAe,KAGb;AACF,IAAA,MAAM,MAAM,GAAG;AACb,QAAA,MAAM,EAAE,MAAM,IAAI,OAAO,CAAC,GAAG,CAAC,cAAc;AAC5C,QAAA,MAAM,EAAE,kCAAkC;AAC1C,QAAA,OAAO,EAAE,KAAK;KACf;AAED,IAAA,IAAI,MAAM,CAAC,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,MAAM,KAAK,EAAE,EAAE;AACjD,QAAA,MAAM,IAAI,KAAK,CAAC,0CAA0C,CAAC;;AAG7D,IAAA,MAAM,UAAU,GAAG,OAAO,EACxB,KAAK,EACL,OAAO,EACP,UAAU,GAAG,CAAC,GACK,KAA6B;AAChD,QAAA,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE;YACjB,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,uBAAuB,EAAE;;AAG3D,QAAA,IAAI;AACF,YAAA,MAAM,OAAO,GAA0B;AACrC,gBAAA,CAAC,EAAE,KAAK;AACR,gBAAA,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,UAAU,CAAC,EAAE,EAAE,CAAC;aAC3C;YAED,IAAI,OAAO,IAAI,IAAI,IAAI,OAAO,KAAK,EAAE,EAAE;gBACrC,OAAO,CAAC,IAAI,CAAC,GAAG,OAAO,CAAC,WAAW,EAAE;;AAGvC,YAAA,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,MAAM,CAAC,MAAM,EACb,OAAO,EACP;AACE,gBAAA,OAAO,EAAE;oBACP,WAAW,EAAE,MAAM,CAAC,MAAM;AAC1B,oBAAA,cAAc,EAAE,kBAAkB;AACnC,iBAAA;gBACD,OAAO,EAAE,MAAM,CAAC,OAAO;AACxB,aAAA,CACF;AAED,YAAA,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI;AAC1B,YAAA,MAAM,OAAO,GAAuB;gBAClC,OAAO,EAAE,IAAI,CAAC,OAAO;AACrB,gBAAA,MAAM,EAAE,IAAI,CAAC,MAAM,IAAI,EAAE;gBACzB,SAAS,EAAE,IAAI,CAAC,SAAS;AACzB,gBAAA,UAAU,EAAE,IAAI,CAAC,UAAU,IAAI,EAAE;gBACjC,aAAa,EAAE,IAAI,CAAC,aAAa;gBACjC,cAAc,EAAE,IAAI,CAAC,cAAc;gBACnC,eAAe,EAAE,IAAI,CAAC,eAAe;aACtC;YAED,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE;;QACvC,OAAO,KAAK,EAAE;AACd,YAAA,MAAM,YAAY,GAChB,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;YACxD,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,CAAuB,oBAAA,EAAA,YAAY,CAAE,CAAA,EAAE;;AAE3E,KAAC;IAED,OAAO,EAAE,UAAU,EAAE;AACvB,CAAC;AAED,MAAM,gBAAgB,GAAG,CACvB,WAAoB,EACpB,MAAe,KAGb;AACF,IAAA,MAAM,MAAM,GAAG;AACb,QAAA,WAAW,EAAE,WAAW,IAAI,OAAO,CAAC,GAAG,CAAC,oBAAoB;AAC5D,QAAA,MAAM,EAAE,MAAM,IAAI,OAAO,CAAC,GAAG,CAAC,eAAe;AAC7C,QACA,OAAO,EAAE,KAAK;KACf;AAED,IAAA,IAAI,MAAM,CAAC,WAAW,IAAI,IAAI,IAAI,MAAM,CAAC,WAAW,KAAK,EAAE,EAAE;AAC3D,QAAA,MAAM,IAAI,KAAK,CAAC,kDAAkD,CAAC;;AAGrE,IAAA,MAAM,UAAU,GAAG,OAAO,EACxB,KAAK,EACL,UAAU,GAAG,CAAC,GACK,KAA6B;AAChD,QAAA,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE;YACjB,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,uBAAuB,EAAE;;AAG3D,QAAA,IAAI;;AAEF,YAAA,IAAI,MAAM,CAAC,WAAW,IAAI,IAAI,IAAI,MAAM,CAAC,WAAW,KAAK,EAAE,EAAE;gBAC3D,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,6BAA6B,EAAE;;AAGjE,YAAA,IAAI,SAAS,GAAG,MAAM,CAAC,WAAW;YAClC,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE;gBAClC,SAAS,GAAG,SAAS,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,GAAG,SAAS;;;AAItD,YAAA,MAAM,MAAM,GAA2B;AACrC,gBAAA,CAAC,EAAE,KAAK;AACR,gBAAA,MAAM,EAAE,MAAM;AACd,gBAAA,MAAM,EAAE,CAAC;AACT,gBAAA,UAAU,EAAE,SAAS;AACrB,gBAAA,QAAQ,EAAE,KAAK;AACf,gBAAA,UAAU,EAAE,CAAC;AACb,gBAAA,OAAO,EAAE,wBAAwB;aAClC;AAED,YAAA,MAAM,OAAO,GAA2B;AACtC,gBAAA,cAAc,EAAE,kBAAkB;aACnC;AAED,YAAA,IAAI,MAAM,CAAC,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,MAAM,KAAK,EAAE,EAAE;AACjD,gBAAA,OAAO,CAAC,WAAW,CAAC,GAAG,MAAM,CAAC,MAAM;;YAGtC,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,GAAG,CAAC,SAAS,EAAE;gBAC1C,OAAO;gBACP,MAAM;gBACN,OAAO,EAAE,MAAM,CAAC,OAAO;AACxB,aAAA,CAAC;AAEF,YAAA,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI;;YAG1B,MAAM,cAAc,GAAG,CAAC,IAAI,CAAC,OAAO,IAAI,EAAE;AACvC,iBAAA,KAAK,CAAC,CAAC,EAAE,UAAU;AACnB,iBAAA,GAAG,CAAC,CAAC,MAAuB,MAAM;AACjC,gBAAA,KAAK,EAAE,MAAM,CAAC,KAAK,IAAI,EAAE;AACzB,gBAAA,IAAI,EAAE,MAAM,CAAC,GAAG,IAAI,EAAE;AACtB,gBAAA,OAAO,EAAE,MAAM,CAAC,OAAO,IAAI,EAAE;AAC7B,gBAAA,IAAI,EAAE,MAAM,CAAC,aAAa,IAAI,EAAE;AACjC,aAAA,CAAC,CAAC;;YAGL,MAAM,YAAY,GAAG,CAAC,IAAI,CAAC,OAAO,IAAI,EAAE;iBACrC,MAAM,CAAC,CAAC,MAAuB,KAAK,MAAM,CAAC,OAAO;AAClD,iBAAA,KAAK,CAAC,CAAC,EAAE,CAAC;AACV,iBAAA,GAAG,CAAC,CAAC,MAAuB,MAAM;AACjC,gBAAA,KAAK,EAAE,MAAM,CAAC,KAAK,IAAI,EAAE;AACzB,gBAAA,QAAQ,EAAE,MAAM,CAAC,OAAO,IAAI,EAAE;AAC/B,aAAA,CAAC,CAAC;;AAGL,YAAA,MAAM,OAAO,GAAuB;AAClC,gBAAA,OAAO,EAAE,cAAc;AACvB,gBAAA,MAAM,EAAE,YAAY;AACpB,gBAAA,UAAU,EAAE,EAAE;;AAEd,gBAAA,eAAe,EAAE,IAAI,CAAC,WAAW,IAAI,EAAE;aACxC;YAED,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE;;QACvC,OAAO,KAAK,EAAE;AACd,YAAA,MAAM,YAAY,GAChB,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;YACxD,OAAO;AACL,gBAAA,OAAO,EAAE,KAAK;gBACd,KAAK,EAAE,CAA+B,4BAAA,EAAA,YAAY,CAAE,CAAA;aACrD;;AAEL,KAAC;IAED,OAAO,EAAE,UAAU,EAAE;AACvB,CAAC;AAEY,MAAA,eAAe,GAAG,CAC7B,MAAsB,KAGpB;AACF,IAAA,MAAM,EACJ,cAAc,GAAG,QAAQ,EACzB,YAAY,EACZ,kBAAkB,EAClB,aAAa,GACd,GAAG,MAAM;AAEV,IAAA,IAAI,cAAc,CAAC,WAAW,EAAE,KAAK,QAAQ,EAAE;AAC7C,QAAA,OAAO,eAAe,CAAC,YAAY,CAAC;;AAC/B,SAAA,IAAI,cAAc,CAAC,WAAW,EAAE,KAAK,SAAS,EAAE;AACrD,QAAA,OAAO,gBAAgB,CAAC,kBAAkB,EAAE,aAAa,CAAC;;SACrD;AACL,QAAA,MAAM,IAAI,KAAK,CACb,4BAA4B,cAAc,CAAA,+BAAA,CAAiC,CAC5E;;AAEL;AAEa,MAAA,qBAAqB,GAAG,CACnC,SAAiC,EAAE,EACnC,eAAkC,KAShC;IACF,IAAI,CAAC,eAAe,EAAE;AACpB,QAAA,MAAM,IAAI,KAAK,CAAC,wCAAwC,CAAC;;IAE3D,MAAM,EACJ,UAAU,GAAG,CAAC;;;IAGd,QAAQ,GACT,GAAG,MAAM;IAEV,MAAM,gBAAgB,GAAG,eAAe;AAExC,IAAA,MAAM,UAAU,GAAG;QACjB,UAAU,EAAE,OAAO,EACjB,KAAK,EACL,KAAK,GAIN,KAAoC;YACnC,OAAO,CAAC,GAAG,CAAC,CAAA,SAAA,EAAY,KAAK,CAAC,MAAM,CAAuB,qBAAA,CAAA,CAAC;YAC5D,MAAM,QAAQ,GAAmC,EAAE;AACnD,YAAA,IAAI;AACF,gBAAA,KAAK,MAAM,WAAW,IAAI,KAAK,EAAE;oBAC/B,MAAM,OAAO,GAA4B;AACtC,yBAAA,SAAS,CAAC,WAAW,EAAE,EAAE;yBACzB,IAAI,CAAC,CAAC,CAAC,GAAG,EAAE,QAAQ,CAAC,KAAI;AACxB,wBAAA,MAAM,WAAW,GAAGC,oBAAc,CAAC,GAAG,EAAE,QAAQ,CAAC,IAAI,EAAE,QAAQ,CAAC;wBAChE,IAAI,QAAQ,CAAC,OAAO,IAAI,QAAQ,CAAC,IAAI,EAAE;AACrC,4BAAA,MAAM,CAAC,OAAO,EAAE,UAAU,CAAC,GACzB,gBAAgB,CAAC,cAAc,CAAC,QAAQ,CAAC;4BAC3C,OAAO;gCACL,GAAG;gCACH,UAAU;gCACV,WAAW;AACX,gCAAA,OAAO,EAAE,OAAO,CAAC,SAAS,CAAC,OAAO,CAAC;6BAClB;;wBAGrB,OAAO;4BACL,GAAG;4BACH,WAAW;AACX,4BAAA,KAAK,EAAE,IAAI;4BACX,OAAO,EAAE,oBAAoB,GAAG,CAAA,EAAA,EAAK,QAAQ,CAAC,KAAK,IAAI,eAAe,CAAE,CAAA;yBACvD;AACrB,qBAAC;AACA,yBAAA,IAAI,CAAC,OAAO,MAAM,KAAI;AACrB,wBAAA,IAAI;AACF,4BAAA,IAAI,MAAM,CAAC,KAAK,IAAI,IAAI,EAAE;AACxB,gCAAA,OAAO,CAAC,KAAK,CACX,CAAA,eAAA,EAAkB,MAAM,CAAC,GAAG,CAAA,EAAA,EAAK,MAAM,CAAC,OAAO,CAAA,CAAE,CAClD;gCACD,OAAO;AACL,oCAAA,GAAG,MAAM;iCACV;;AAEH,4BAAA,MAAM,UAAU,GAAG,MAAM,aAAa,CAAC;gCACrC,KAAK;gCACL,QAAQ;gCACR,OAAO,EAAE,MAAM,CAAC,OAAO;AACxB,6BAAA,CAAC;4BACF,OAAO;AACL,gCAAA,GAAG,MAAM;gCACT,UAAU;6BACX;;wBACD,OAAO,KAAK,EAAE;AACd,4BAAA,OAAO,CAAC,KAAK,CAAC,mCAAmC,EAAE,KAAK,CAAC;4BACzD,OAAO;AACL,gCAAA,GAAG,MAAM;6BACV;;AAEL,qBAAC;AACA,yBAAA,KAAK,CAAC,CAAC,KAAK,KAAI;wBACf,OAAO,CAAC,KAAK,CAAC,CAAA,eAAA,EAAkB,WAAW,CAAG,CAAA,CAAA,EAAE,KAAK,CAAC;wBACtD,OAAO;AACL,4BAAA,GAAG,EAAE,WAAW;AAChB,4BAAA,KAAK,EAAE,IAAI;4BACX,OAAO,EAAE,oBAAoB,WAAW,CAAA,EAAA,EAAK,KAAK,CAAC,OAAO,IAAI,eAAe,CAAE,CAAA;yBAChF;AACH,qBAAC,CAAC;AACJ,oBAAA,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC;;AAExB,gBAAA,OAAO,MAAM,OAAO,CAAC,GAAG,CAAC,QAAQ,CAAC;;YAClC,OAAO,KAAK,EAAE;AACd,gBAAA,OAAO,CAAC,KAAK,CAAC,sBAAsB,EAAE,KAAK,CAAC;AAC5C,gBAAA,OAAO,EAAE;;SAEZ;KACF;AAED,IAAA,MAAM,aAAa,GAAG,OAAO,EAC3B,KAAK,EACL,KAAK,EACL,MAAM,EACN,gBAAgB,GAMjB,KAAmB;QAClB,MAAM,YAAY,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,MAAM,CAAC;;AAE3C,QAAA,MAAM,OAAO,GAAG,MAAM,UAAU,CAAC,UAAU,CAAC,EAAE,KAAK,EAAE,KAAK,EAAE,YAAY,EAAE,CAAC;AAC3E,QAAA,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE;AAC5B,YAAA,IAAI,MAAM,CAAC,KAAK,KAAK,IAAI,EAAE;gBACzB;;AAEF,YAAA,MAAM,EAAE,GAAG,EAAE,OAAO,EAAE,WAAW,EAAE,UAAU,EAAE,UAAU,EAAE,GAAG,MAAM;YACpE,gBAAgB,GAAG,GAAG,EAAE;gBACtB,OAAO;gBACP,WAAW;gBACX,UAAU;gBACV,UAAU;AACX,aAAA,CAAC;;AAEN,KAAC;AAED,IAAA,MAAM,cAAc,GAAG,OACrB,MAAsB,EACtB,WAAmB,EACnB,KAAa,EACb,OAAmB,GAAA,KAAK,KACO;AAC/B,QAAA,IAAI;AACF,YAAA,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE;gBAChB,OAAO;AACL,oBAAA,OAAO,EAAE,EAAE;AACX,oBAAA,UAAU,EAAE,EAAE;AACd,oBAAA,MAAM,EAAE,EAAE;AACV,oBAAA,eAAe,EAAE,EAAE;iBACpB;;AACI,iBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,EAAE;gBAC/B,OAAO,MAAM,CAAC,IAAI;;YAGpB,IAAI,CAAC,OAAO,EAAE;gBACZ,MAAM,WAAW,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,MAAM,KACpD,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,eAAe,CAAC,CACtC;AAED,gBAAA,IAAI,CAAC,WAAW,CAAC,MAAM,EAAE;oBACvB,OAAO,MAAM,CAAC,IAAI;;AAGpB,gBAAA,MAAM,aAAa,GAAG,IAAI,GAAG,EAAyB;AACtD,gBAAA,aAAa,CAAC,GAAG,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,WAAW,CAAC,CAAC,CAAC,CAAC;AACtD,gBAAA,MAAM,gBAAgB,GAAG,0BAA0B,CAAC,aAAa,CAAC;AAClE,gBAAA,MAAM,aAAa,CAAC;oBAClB,KAAK;AACL,oBAAA,MAAM,EAAE,CAAC;oBACT,gBAAgB;oBAChB,KAAK,EAAE,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;AAC7B,iBAAA,CAAC;AAEF,gBAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;oBACnD,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;oBACrC,MAAM,aAAa,GAAG,aAAa,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC;oBACpD,IAAI,aAAa,EAAE;AACjB,wBAAA,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG;AACvB,4BAAA,GAAG,MAAM;AACT,4BAAA,GAAG,aAAa;yBACjB;;;gBAIL,OAAO,MAAM,CAAC,IAAI;;AAGpB,YAAA,MAAM,SAAS,GAAG,IAAI,GAAG,EAAyB;YAClD,MAAM,QAAQ,GAAa,EAAE;YAE7B,KAAK,MAAM,MAAM,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,EAAE;AACxC,gBAAA,IAAI,MAAM,CAAC,IAAI,EAAE;AACf,oBAAA,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC;oBAC1B,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC;;;AAItC,YAAA,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE;gBACzB,OAAO,MAAM,CAAC,IAAI;;AAGpB,YAAA,MAAM,gBAAgB,GAAG,0BAA0B,CAAC,SAAS,CAAC;AAC9D,YAAA,MAAM,aAAa,CAAC;AAClB,gBAAA,KAAK,EAAE,QAAQ;gBACf,KAAK;gBACL,gBAAgB;AAChB,gBAAA,MAAM,EAAE,WAAW;AACpB,aAAA,CAAC;AAEF,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBACnD,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;gBACrC,MAAM,aAAa,GAAG,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC;gBAChD,IAAI,aAAa,EAAE;AACjB,oBAAA,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG;AACvB,wBAAA,GAAG,MAAM;AACT,wBAAA,GAAG,aAAa;qBACjB;;;AAIL,YAAA,MAAM,iBAAiB,GAAG,MAAM,CAAC,IAAI,CAAC;iBACnC,MAAM,CACL,CAAC,MAAM,KACL,MAAM,CAAC,OAAO,IAAI,IAAI,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,QAAQ,CAAC;AAEjE,iBAAA,KAAK,CAAC,CAAC,EAAE,WAAW,CAAC;AAExB,YAAA,IAAI,iBAAiB,CAAC,MAAM,GAAG,CAAC,EAAE;AAChC,gBAAA,MAAM,CAAC,IAAI,CAAC,OAAO,GAAG,iBAAiB;;YAEzC,OAAO,MAAM,CAAC,IAAI;;QAClB,OAAO,KAAK,EAAE;AACd,YAAA,OAAO,CAAC,KAAK,CAAC,0BAA0B,EAAE,KAAK,CAAC;YAChD,OAAO;AACL,gBAAA,OAAO,EAAE,EAAE;AACX,gBAAA,UAAU,EAAE,EAAE;AACd,gBAAA,MAAM,EAAE,EAAE;AACV,gBAAA,eAAe,EAAE,EAAE;gBACnB,GAAG,MAAM,CAAC,IAAI;AACd,gBAAA,KAAK,EAAE,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;aAC9D;;AAEL,KAAC;IAED,OAAO;QACL,cAAc;QACd,UAAU;KACX;AACH;;;;;"}
+{"version":3,"file":"search.cjs","sources":["../../../../src/tools/search/search.ts"],"sourcesContent":["/* eslint-disable no-console */\nimport axios from 'axios';\nimport { RecursiveCharacterTextSplitter } from '@langchain/textsplitters';\nimport type * as t from './types';\nimport { FirecrawlScraper } from './firecrawl';\nimport { BaseReranker } from './rerankers';\nimport { getAttribution } from './utils';\n\nconst chunker = {\n  cleanText: (text: string): string => {\n    if (!text) return '';\n\n    /** Normalized all line endings to '\\n' */\n    const normalizedText = text.replace(/\\r\\n/g, '\\n').replace(/\\r/g, '\\n');\n\n    /** Handle multiple backslashes followed by newlines\n     * This replaces patterns like '\\\\\\\\\\\\n' with a single newline */\n    const fixedBackslashes = normalizedText.replace(/\\\\+\\n/g, '\\n');\n\n    /** Cleaned up consecutive newlines, tabs, and spaces around newlines */\n    const cleanedNewlines = fixedBackslashes.replace(/[\\t ]*\\n[\\t \\n]*/g, '\\n');\n\n    /** Cleaned up excessive spaces and tabs */\n    const cleanedSpaces = cleanedNewlines.replace(/[ \\t]+/g, ' ');\n\n    return cleanedSpaces.trim();\n  },\n  splitText: async (\n    text: string,\n    options?: {\n      chunkSize?: number;\n      chunkOverlap?: number;\n      separators?: string[];\n    }\n  ): Promise<string[]> => {\n    const chunkSize = options?.chunkSize ?? 150;\n    const chunkOverlap = options?.chunkOverlap ?? 50;\n    const separators = options?.separators || ['\\n\\n', '\\n'];\n\n    const splitter = new RecursiveCharacterTextSplitter({\n      separators,\n      chunkSize,\n      chunkOverlap,\n    });\n\n    return await splitter.splitText(text);\n  },\n\n  splitTexts: async (\n    texts: string[],\n    options?: {\n      chunkSize?: number;\n      chunkOverlap?: number;\n      separators?: string[];\n    }\n  ): Promise<string[][]> => {\n    // Split multiple texts\n    const promises = texts.map((text) =>\n      chunker.splitText(text, options).catch((error) => {\n        console.error('Error splitting text:', error);\n        return [text];\n      })\n    );\n    return Promise.all(promises);\n  },\n};\n\nfunction createSourceUpdateCallback(sourceMap: Map<string, t.ValidSource>) {\n  return (link: string, update?: Partial<t.ValidSource>): void => {\n    const source = sourceMap.get(link);\n    if (source) {\n      sourceMap.set(link, {\n        ...source,\n        ...update,\n      });\n    }\n  };\n}\n\nconst getHighlights = async ({\n  query,\n  content,\n  reranker,\n  topResults = 5,\n}: {\n  content: string;\n  query: string;\n  reranker?: BaseReranker;\n  topResults?: number;\n}): Promise<t.Highlight[] | undefined> => {\n  if (!content) {\n    console.warn('No content provided for highlights');\n    return;\n  }\n  if (!reranker) {\n    console.warn('No reranker provided for highlights');\n    return;\n  }\n\n  try {\n    const documents = await chunker.splitText(content);\n    if (Array.isArray(documents)) {\n      return await reranker.rerank(query, documents, topResults);\n    } else {\n      console.error(\n        'Expected documents to be an array, got:',\n        typeof documents\n      );\n      return;\n    }\n  } catch (error) {\n    console.error('Error in content processing:', error);\n    return;\n  }\n};\n\nconst createSerperAPI = (\n  apiKey?: string\n): {\n  getSources: (params: t.GetSourcesParams) => Promise<t.SearchResult>;\n} => {\n  const config = {\n    apiKey: apiKey ?? process.env.SERPER_API_KEY,\n    apiUrl: 'https://google.serper.dev/search',\n    timeout: 10000,\n  };\n\n  if (config.apiKey == null || config.apiKey === '') {\n    throw new Error('SERPER_API_KEY is required for SerperAPI');\n  }\n\n  const getSources = async ({\n    query,\n    country,\n    numResults = 8,\n  }: t.GetSourcesParams): Promise<t.SearchResult> => {\n    if (!query.trim()) {\n      return { success: false, error: 'Query cannot be empty' };\n    }\n\n    try {\n      const payload: t.SerperSearchPayload = {\n        q: query,\n        num: Math.min(Math.max(1, numResults), 10),\n      };\n\n      if (country != null && country !== '') {\n        payload['gl'] = country.toLowerCase();\n      }\n\n      const response = await axios.post<t.SerperResultData>(\n        config.apiUrl,\n        payload,\n        {\n          headers: {\n            'X-API-KEY': config.apiKey,\n            'Content-Type': 'application/json',\n          },\n          timeout: config.timeout,\n        }\n      );\n\n      const data = response.data;\n      const results: t.SearchResultData = {\n        organic: data.organic,\n        images: data.images ?? [],\n        answerBox: data.answerBox,\n        topStories: data.topStories ?? [],\n        peopleAlsoAsk: data.peopleAlsoAsk,\n        knowledgeGraph: data.knowledgeGraph,\n        relatedSearches: data.relatedSearches,\n      };\n\n      return { success: true, data: results };\n    } catch (error) {\n      const errorMessage =\n        error instanceof Error ? error.message : String(error);\n      return { success: false, error: `API request failed: ${errorMessage}` };\n    }\n  };\n\n  return { getSources };\n};\n\nconst createSearXNGAPI = (\n  instanceUrl?: string,\n  apiKey?: string\n): {\n  getSources: (params: t.GetSourcesParams) => Promise<t.SearchResult>;\n} => {\n  const config = {\n    instanceUrl: instanceUrl ?? process.env.SEARXNG_INSTANCE_URL,\n    apiKey: apiKey ?? process.env.SEARXNG_API_KEY,\n    defaultLocation: 'all',\n    timeout: 10000,\n  };\n\n  if (config.instanceUrl == null || config.instanceUrl === '') {\n    throw new Error('SEARXNG_INSTANCE_URL is required for SearXNG API');\n  }\n\n  const getSources = async ({\n    query,\n    numResults = 8,\n  }: t.GetSourcesParams): Promise<t.SearchResult> => {\n    if (!query.trim()) {\n      return { success: false, error: 'Query cannot be empty' };\n    }\n\n    try {\n      // Ensure the instance URL ends with /search\n      if (config.instanceUrl == null || config.instanceUrl === '') {\n        return { success: false, error: 'Instance URL is not defined' };\n      }\n\n      let searchUrl = config.instanceUrl;\n      if (!searchUrl.endsWith('/search')) {\n        searchUrl = searchUrl.replace(/\\/$/, '') + '/search';\n      }\n\n      // Prepare parameters for SearXNG\n      const params: t.SearxNGSearchPayload = {\n        q: query,\n        format: 'json',\n        pageno: 1,\n        categories: 'general',\n        language: 'all',\n        safesearch: 0,\n        engines: 'google,bing,duckduckgo',\n      };\n\n      const headers: Record<string, string> = {\n        'Content-Type': 'application/json',\n      };\n\n      if (config.apiKey != null && config.apiKey !== '') {\n        headers['X-API-Key'] = config.apiKey;\n      }\n\n      const response = await axios.get(searchUrl, {\n        headers,\n        params,\n        timeout: config.timeout,\n      });\n\n      const data = response.data;\n\n      // Transform SearXNG results to match SerperAPI format\n      const organicResults = (data.results ?? [])\n        .slice(0, numResults)\n        .map((result: t.SearXNGResult) => ({\n          title: result.title ?? '',\n          link: result.url ?? '',\n          snippet: result.content ?? '',\n          date: result.publishedDate ?? '',\n        }));\n\n      // Extract image results if available\n      const imageResults = (data.results ?? [])\n        .filter((result: t.SearXNGResult) => result.img_src)\n        .slice(0, 6)\n        .map((result: t.SearXNGResult) => ({\n          title: result.title ?? '',\n          imageUrl: result.img_src ?? '',\n        }));\n\n      // Format results to match SerperAPI structure\n      const results: t.SearchResultData = {\n        organic: organicResults,\n        images: imageResults,\n        topStories: [],\n        // Use undefined instead of null for optional properties\n        relatedSearches: data.suggestions ?? [],\n      };\n\n      return { success: true, data: results };\n    } catch (error) {\n      const errorMessage =\n        error instanceof Error ? error.message : String(error);\n      return {\n        success: false,\n        error: `SearXNG API request failed: ${errorMessage}`,\n      };\n    }\n  };\n\n  return { getSources };\n};\n\nexport const createSearchAPI = (\n  config: t.SearchConfig\n): {\n  getSources: (params: t.GetSourcesParams) => Promise<t.SearchResult>;\n} => {\n  const {\n    searchProvider = 'serper',\n    serperApiKey,\n    searxngInstanceUrl,\n    searxngApiKey,\n  } = config;\n\n  if (searchProvider.toLowerCase() === 'serper') {\n    return createSerperAPI(serperApiKey);\n  } else if (searchProvider.toLowerCase() === 'searxng') {\n    return createSearXNGAPI(searxngInstanceUrl, searxngApiKey);\n  } else {\n    throw new Error(\n      `Invalid search provider: ${searchProvider}. Must be 'serper' or 'searxng'`\n    );\n  }\n};\n\nexport const createSourceProcessor = (\n  config: t.ProcessSourcesConfig = {},\n  scraperInstance?: FirecrawlScraper\n): {\n  processSources: (\n    fields: t.ProcessSourcesFields\n  ) => Promise<t.SearchResultData>;\n  topResults: number;\n} => {\n  if (!scraperInstance) {\n    throw new Error('Firecrawl scraper instance is required');\n  }\n  const {\n    topResults = 5,\n    // strategies = ['no_extraction'],\n    // filterContent = true,\n    reranker,\n  } = config;\n\n  const firecrawlScraper = scraperInstance;\n\n  const webScraper = {\n    scrapeMany: async ({\n      query,\n      links,\n      onGetHighlights,\n    }: {\n      query: string;\n      links: string[];\n      onGetHighlights: t.SearchToolConfig['onGetHighlights'];\n    }): Promise<Array<t.ScrapeResult>> => {\n      console.log(`Scraping ${links.length} links with Firecrawl`);\n      const promises: Array<Promise<t.ScrapeResult>> = [];\n      try {\n        for (let i = 0; i < links.length; i++) {\n          const currentLink = links[i];\n          const promise: Promise<t.ScrapeResult> = firecrawlScraper\n            .scrapeUrl(currentLink, {})\n            .then(([url, response]) => {\n              const attribution = getAttribution(url, response.data?.metadata);\n              if (response.success && response.data) {\n                const [content, references] =\n                  firecrawlScraper.extractContent(response);\n                return {\n                  url,\n                  references,\n                  attribution,\n                  content: chunker.cleanText(content),\n                } as t.ScrapeResult;\n              }\n\n              return {\n                url,\n                attribution,\n                error: true,\n                content: `Failed to scrape ${url}: ${response.error ?? 'Unknown error'}`,\n              } as t.ScrapeResult;\n            })\n            .then(async (result) => {\n              try {\n                if (result.error != null) {\n                  console.error(\n                    `Error scraping ${result.url}: ${result.content}`\n                  );\n                  return {\n                    ...result,\n                  };\n                }\n                const highlights = await getHighlights({\n                  query,\n                  reranker,\n                  content: result.content,\n                });\n                if (onGetHighlights) {\n                  onGetHighlights(result.url);\n                }\n                return {\n                  ...result,\n                  highlights,\n                };\n              } catch (error) {\n                console.error('Error processing scraped content:', error);\n                return {\n                  ...result,\n                };\n              }\n            })\n            .catch((error) => {\n              console.error(`Error scraping ${currentLink}:`, error);\n              return {\n                url: currentLink,\n                error: true,\n                content: `Failed to scrape ${currentLink}: ${error.message ?? 'Unknown error'}`,\n              };\n            });\n          promises.push(promise);\n        }\n        return await Promise.all(promises);\n      } catch (error) {\n        console.error('Error in scrapeMany:', error);\n        return [];\n      }\n    },\n  };\n\n  const fetchContents = async ({\n    links,\n    query,\n    target,\n    onGetHighlights,\n    onContentScraped,\n  }: {\n    links: string[];\n    query: string;\n    target: number;\n    onGetHighlights: t.SearchToolConfig['onGetHighlights'];\n    onContentScraped?: (link: string, update?: Partial<t.ValidSource>) => void;\n  }): Promise<void> => {\n    const initialLinks = links.slice(0, target);\n    // const remainingLinks = links.slice(target).reverse();\n    const results = await webScraper.scrapeMany({\n      query,\n      links: initialLinks,\n      onGetHighlights,\n    });\n    for (const result of results) {\n      if (result.error === true) {\n        continue;\n      }\n      const { url, content, attribution, references, highlights } = result;\n      onContentScraped?.(url, {\n        content,\n        attribution,\n        references,\n        highlights,\n      });\n    }\n  };\n\n  const processSources = async ({\n    result,\n    numElements,\n    query,\n    proMode = true,\n    onGetHighlights,\n  }: t.ProcessSourcesFields): Promise<t.SearchResultData> => {\n    try {\n      if (!result.data) {\n        return {\n          organic: [],\n          topStories: [],\n          images: [],\n          relatedSearches: [],\n        };\n      } else if (!result.data.organic) {\n        return result.data;\n      }\n\n      if (!proMode) {\n        const wikiSources = result.data.organic.filter((source) =>\n          source.link.includes('wikipedia.org')\n        );\n\n        if (!wikiSources.length) {\n          return result.data;\n        }\n\n        const wikiSourceMap = new Map<string, t.ValidSource>();\n        wikiSourceMap.set(wikiSources[0].link, wikiSources[0]);\n        const onContentScraped = createSourceUpdateCallback(wikiSourceMap);\n        await fetchContents({\n          query,\n          target: 1,\n          onGetHighlights,\n          onContentScraped,\n          links: [wikiSources[0].link],\n        });\n\n        for (let i = 0; i < result.data.organic.length; i++) {\n          const source = result.data.organic[i];\n          const updatedSource = wikiSourceMap.get(source.link);\n          if (updatedSource) {\n            result.data.organic[i] = {\n              ...source,\n              ...updatedSource,\n            };\n          }\n        }\n\n        return result.data;\n      }\n\n      const sourceMap = new Map<string, t.ValidSource>();\n      const allLinks: string[] = [];\n\n      for (const source of result.data.organic) {\n        if (source.link) {\n          allLinks.push(source.link);\n          sourceMap.set(source.link, source);\n        }\n      }\n\n      if (allLinks.length === 0) {\n        return result.data;\n      }\n\n      const onContentScraped = createSourceUpdateCallback(sourceMap);\n      await fetchContents({\n        query,\n        links: allLinks,\n        onGetHighlights,\n        onContentScraped,\n        target: numElements,\n      });\n\n      for (let i = 0; i < result.data.organic.length; i++) {\n        const source = result.data.organic[i];\n        const updatedSource = sourceMap.get(source.link);\n        if (updatedSource) {\n          result.data.organic[i] = {\n            ...source,\n            ...updatedSource,\n          };\n        }\n      }\n\n      const successfulSources = result.data.organic\n        .filter(\n          (source) =>\n            source.content != null && !source.content.startsWith('Failed')\n        )\n        .slice(0, numElements);\n\n      if (successfulSources.length > 0) {\n        result.data.organic = successfulSources;\n      }\n      return result.data;\n    } catch (error) {\n      console.error('Error in processSources:', error);\n      return {\n        organic: [],\n        topStories: [],\n        images: [],\n        relatedSearches: [],\n        ...result.data,\n        error: error instanceof Error ? error.message : String(error),\n      };\n    }\n  };\n\n  return {\n    processSources,\n    topResults,\n  };\n};\n"],"names":["RecursiveCharacterTextSplitter","getAttribution"],"mappings":";;;;;;AAAA;AAQA,MAAM,OAAO,GAAG;AACd,IAAA,SAAS,EAAE,CAAC,IAAY,KAAY;AAClC,QAAA,IAAI,CAAC,IAAI;AAAE,YAAA,OAAO,EAAE;;AAGpB,QAAA,MAAM,cAAc,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,IAAI,CAAC;AAEvE;AACiE;QACjE,MAAM,gBAAgB,GAAG,cAAc,CAAC,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC;;QAG/D,MAAM,eAAe,GAAG,gBAAgB,CAAC,OAAO,CAAC,mBAAmB,EAAE,IAAI,CAAC;;QAG3E,MAAM,aAAa,GAAG,eAAe,CAAC,OAAO,CAAC,SAAS,EAAE,GAAG,CAAC;AAE7D,QAAA,OAAO,aAAa,CAAC,IAAI,EAAE;KAC5B;AACD,IAAA,SAAS,EAAE,OACT,IAAY,EACZ,OAIC,KACoB;AACrB,QAAA,MAAM,SAAS,GAAG,OAAO,EAAE,SAAS,IAAI,GAAG;AAC3C,QAAA,MAAM,YAAY,GAAG,OAAO,EAAE,YAAY,IAAI,EAAE;QAChD,MAAM,UAAU,GAAG,OAAO,EAAE,UAAU,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC;AAExD,QAAA,MAAM,QAAQ,GAAG,IAAIA,4CAA8B,CAAC;YAClD,UAAU;YACV,SAAS;YACT,YAAY;AACb,SAAA,CAAC;AAEF,QAAA,OAAO,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC;KACtC;AAED,IAAA,UAAU,EAAE,OACV,KAAe,EACf,OAIC,KACsB;;QAEvB,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,KAC9B,OAAO,CAAC,SAAS,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,KAAI;AAC/C,YAAA,OAAO,CAAC,KAAK,CAAC,uBAAuB,EAAE,KAAK,CAAC;YAC7C,OAAO,CAAC,IAAI,CAAC;SACd,CAAC,CACH;AACD,QAAA,OAAO,OAAO,CAAC,GAAG,CAAC,QAAQ,CAAC;KAC7B;CACF;AAED,SAAS,0BAA0B,CAAC,SAAqC,EAAA;AACvE,IAAA,OAAO,CAAC,IAAY,EAAE,MAA+B,KAAU;QAC7D,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC;QAClC,IAAI,MAAM,EAAE;AACV,YAAA,SAAS,CAAC,GAAG,CAAC,IAAI,EAAE;AAClB,gBAAA,GAAG,MAAM;AACT,gBAAA,GAAG,MAAM;AACV,aAAA,CAAC;;AAEN,KAAC;AACH;AAEA,MAAM,aAAa,GAAG,OAAO,EAC3B,KAAK,EACL,OAAO,EACP,QAAQ,EACR,UAAU,GAAG,CAAC,GAMf,KAAwC;IACvC,IAAI,CAAC,OAAO,EAAE;AACZ,QAAA,OAAO,CAAC,IAAI,CAAC,oCAAoC,CAAC;QAClD;;IAEF,IAAI,CAAC,QAAQ,EAAE;AACb,QAAA,OAAO,CAAC,IAAI,CAAC,qCAAqC,CAAC;QACnD;;AAGF,IAAA,IAAI;QACF,MAAM,SAAS,GAAG,MAAM,OAAO,CAAC,SAAS,CAAC,OAAO,CAAC;AAClD,QAAA,IAAI,KAAK,CAAC,OAAO,CAAC,SAAS,CAAC,EAAE;YAC5B,OAAO,MAAM,QAAQ,CAAC,MAAM,CAAC,KAAK,EAAE,SAAS,EAAE,UAAU,CAAC;;aACrD;YACL,OAAO,CAAC,KAAK,CACX,yCAAyC,EACzC,OAAO,SAAS,CACjB;YACD;;;IAEF,OAAO,KAAK,EAAE;AACd,QAAA,OAAO,CAAC,KAAK,CAAC,8BAA8B,EAAE,KAAK,CAAC;QACpD;;AAEJ,CAAC;AAED,MAAM,eAAe,GAAG,CACtB,MAAe,KAGb;AACF,IAAA,MAAM,MAAM,GAAG;AACb,QAAA,MAAM,EAAE,MAAM,IAAI,OAAO,CAAC,GAAG,CAAC,cAAc;AAC5C,QAAA,MAAM,EAAE,kCAAkC;AAC1C,QAAA,OAAO,EAAE,KAAK;KACf;AAED,IAAA,IAAI,MAAM,CAAC,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,MAAM,KAAK,EAAE,EAAE;AACjD,QAAA,MAAM,IAAI,KAAK,CAAC,0CAA0C,CAAC;;AAG7D,IAAA,MAAM,UAAU,GAAG,OAAO,EACxB,KAAK,EACL,OAAO,EACP,UAAU,GAAG,CAAC,GACK,KAA6B;AAChD,QAAA,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE;YACjB,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,uBAAuB,EAAE;;AAG3D,QAAA,IAAI;AACF,YAAA,MAAM,OAAO,GAA0B;AACrC,gBAAA,CAAC,EAAE,KAAK;AACR,gBAAA,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,UAAU,CAAC,EAAE,EAAE,CAAC;aAC3C;YAED,IAAI,OAAO,IAAI,IAAI,IAAI,OAAO,KAAK,EAAE,EAAE;gBACrC,OAAO,CAAC,IAAI,CAAC,GAAG,OAAO,CAAC,WAAW,EAAE;;AAGvC,YAAA,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,IAAI,CAC/B,MAAM,CAAC,MAAM,EACb,OAAO,EACP;AACE,gBAAA,OAAO,EAAE;oBACP,WAAW,EAAE,MAAM,CAAC,MAAM;AAC1B,oBAAA,cAAc,EAAE,kBAAkB;AACnC,iBAAA;gBACD,OAAO,EAAE,MAAM,CAAC,OAAO;AACxB,aAAA,CACF;AAED,YAAA,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI;AAC1B,YAAA,MAAM,OAAO,GAAuB;gBAClC,OAAO,EAAE,IAAI,CAAC,OAAO;AACrB,gBAAA,MAAM,EAAE,IAAI,CAAC,MAAM,IAAI,EAAE;gBACzB,SAAS,EAAE,IAAI,CAAC,SAAS;AACzB,gBAAA,UAAU,EAAE,IAAI,CAAC,UAAU,IAAI,EAAE;gBACjC,aAAa,EAAE,IAAI,CAAC,aAAa;gBACjC,cAAc,EAAE,IAAI,CAAC,cAAc;gBACnC,eAAe,EAAE,IAAI,CAAC,eAAe;aACtC;YAED,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE;;QACvC,OAAO,KAAK,EAAE;AACd,YAAA,MAAM,YAAY,GAChB,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;YACxD,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,CAAuB,oBAAA,EAAA,YAAY,CAAE,CAAA,EAAE;;AAE3E,KAAC;IAED,OAAO,EAAE,UAAU,EAAE;AACvB,CAAC;AAED,MAAM,gBAAgB,GAAG,CACvB,WAAoB,EACpB,MAAe,KAGb;AACF,IAAA,MAAM,MAAM,GAAG;AACb,QAAA,WAAW,EAAE,WAAW,IAAI,OAAO,CAAC,GAAG,CAAC,oBAAoB;AAC5D,QAAA,MAAM,EAAE,MAAM,IAAI,OAAO,CAAC,GAAG,CAAC,eAAe;AAC7C,QACA,OAAO,EAAE,KAAK;KACf;AAED,IAAA,IAAI,MAAM,CAAC,WAAW,IAAI,IAAI,IAAI,MAAM,CAAC,WAAW,KAAK,EAAE,EAAE;AAC3D,QAAA,MAAM,IAAI,KAAK,CAAC,kDAAkD,CAAC;;AAGrE,IAAA,MAAM,UAAU,GAAG,OAAO,EACxB,KAAK,EACL,UAAU,GAAG,CAAC,GACK,KAA6B;AAChD,QAAA,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE;YACjB,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,uBAAuB,EAAE;;AAG3D,QAAA,IAAI;;AAEF,YAAA,IAAI,MAAM,CAAC,WAAW,IAAI,IAAI,IAAI,MAAM,CAAC,WAAW,KAAK,EAAE,EAAE;gBAC3D,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,6BAA6B,EAAE;;AAGjE,YAAA,IAAI,SAAS,GAAG,MAAM,CAAC,WAAW;YAClC,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE;gBAClC,SAAS,GAAG,SAAS,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,GAAG,SAAS;;;AAItD,YAAA,MAAM,MAAM,GAA2B;AACrC,gBAAA,CAAC,EAAE,KAAK;AACR,gBAAA,MAAM,EAAE,MAAM;AACd,gBAAA,MAAM,EAAE,CAAC;AACT,gBAAA,UAAU,EAAE,SAAS;AACrB,gBAAA,QAAQ,EAAE,KAAK;AACf,gBAAA,UAAU,EAAE,CAAC;AACb,gBAAA,OAAO,EAAE,wBAAwB;aAClC;AAED,YAAA,MAAM,OAAO,GAA2B;AACtC,gBAAA,cAAc,EAAE,kBAAkB;aACnC;AAED,YAAA,IAAI,MAAM,CAAC,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,MAAM,KAAK,EAAE,EAAE;AACjD,gBAAA,OAAO,CAAC,WAAW,CAAC,GAAG,MAAM,CAAC,MAAM;;YAGtC,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,GAAG,CAAC,SAAS,EAAE;gBAC1C,OAAO;gBACP,MAAM;gBACN,OAAO,EAAE,MAAM,CAAC,OAAO;AACxB,aAAA,CAAC;AAEF,YAAA,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI;;YAG1B,MAAM,cAAc,GAAG,CAAC,IAAI,CAAC,OAAO,IAAI,EAAE;AACvC,iBAAA,KAAK,CAAC,CAAC,EAAE,UAAU;AACnB,iBAAA,GAAG,CAAC,CAAC,MAAuB,MAAM;AACjC,gBAAA,KAAK,EAAE,MAAM,CAAC,KAAK,IAAI,EAAE;AACzB,gBAAA,IAAI,EAAE,MAAM,CAAC,GAAG,IAAI,EAAE;AACtB,gBAAA,OAAO,EAAE,MAAM,CAAC,OAAO,IAAI,EAAE;AAC7B,gBAAA,IAAI,EAAE,MAAM,CAAC,aAAa,IAAI,EAAE;AACjC,aAAA,CAAC,CAAC;;YAGL,MAAM,YAAY,GAAG,CAAC,IAAI,CAAC,OAAO,IAAI,EAAE;iBACrC,MAAM,CAAC,CAAC,MAAuB,KAAK,MAAM,CAAC,OAAO;AAClD,iBAAA,KAAK,CAAC,CAAC,EAAE,CAAC;AACV,iBAAA,GAAG,CAAC,CAAC,MAAuB,MAAM;AACjC,gBAAA,KAAK,EAAE,MAAM,CAAC,KAAK,IAAI,EAAE;AACzB,gBAAA,QAAQ,EAAE,MAAM,CAAC,OAAO,IAAI,EAAE;AAC/B,aAAA,CAAC,CAAC;;AAGL,YAAA,MAAM,OAAO,GAAuB;AAClC,gBAAA,OAAO,EAAE,cAAc;AACvB,gBAAA,MAAM,EAAE,YAAY;AACpB,gBAAA,UAAU,EAAE,EAAE;;AAEd,gBAAA,eAAe,EAAE,IAAI,CAAC,WAAW,IAAI,EAAE;aACxC;YAED,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE;;QACvC,OAAO,KAAK,EAAE;AACd,YAAA,MAAM,YAAY,GAChB,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;YACxD,OAAO;AACL,gBAAA,OAAO,EAAE,KAAK;gBACd,KAAK,EAAE,CAA+B,4BAAA,EAAA,YAAY,CAAE,CAAA;aACrD;;AAEL,KAAC;IAED,OAAO,EAAE,UAAU,EAAE;AACvB,CAAC;AAEY,MAAA,eAAe,GAAG,CAC7B,MAAsB,KAGpB;AACF,IAAA,MAAM,EACJ,cAAc,GAAG,QAAQ,EACzB,YAAY,EACZ,kBAAkB,EAClB,aAAa,GACd,GAAG,MAAM;AAEV,IAAA,IAAI,cAAc,CAAC,WAAW,EAAE,KAAK,QAAQ,EAAE;AAC7C,QAAA,OAAO,eAAe,CAAC,YAAY,CAAC;;AAC/B,SAAA,IAAI,cAAc,CAAC,WAAW,EAAE,KAAK,SAAS,EAAE;AACrD,QAAA,OAAO,gBAAgB,CAAC,kBAAkB,EAAE,aAAa,CAAC;;SACrD;AACL,QAAA,MAAM,IAAI,KAAK,CACb,4BAA4B,cAAc,CAAA,+BAAA,CAAiC,CAC5E;;AAEL;AAEa,MAAA,qBAAqB,GAAG,CACnC,SAAiC,EAAE,EACnC,eAAkC,KAMhC;IACF,IAAI,CAAC,eAAe,EAAE;AACpB,QAAA,MAAM,IAAI,KAAK,CAAC,wCAAwC,CAAC;;IAE3D,MAAM,EACJ,UAAU,GAAG,CAAC;;;IAGd,QAAQ,GACT,GAAG,MAAM;IAEV,MAAM,gBAAgB,GAAG,eAAe;AAExC,IAAA,MAAM,UAAU,GAAG;QACjB,UAAU,EAAE,OAAO,EACjB,KAAK,EACL,KAAK,EACL,eAAe,GAKhB,KAAoC;YACnC,OAAO,CAAC,GAAG,CAAC,CAAA,SAAA,EAAY,KAAK,CAAC,MAAM,CAAuB,qBAAA,CAAA,CAAC;YAC5D,MAAM,QAAQ,GAAmC,EAAE;AACnD,YAAA,IAAI;AACF,gBAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;AACrC,oBAAA,MAAM,WAAW,GAAG,KAAK,CAAC,CAAC,CAAC;oBAC5B,MAAM,OAAO,GAA4B;AACtC,yBAAA,SAAS,CAAC,WAAW,EAAE,EAAE;yBACzB,IAAI,CAAC,CAAC,CAAC,GAAG,EAAE,QAAQ,CAAC,KAAI;AACxB,wBAAA,MAAM,WAAW,GAAGC,oBAAc,CAAC,GAAG,EAAE,QAAQ,CAAC,IAAI,EAAE,QAAQ,CAAC;wBAChE,IAAI,QAAQ,CAAC,OAAO,IAAI,QAAQ,CAAC,IAAI,EAAE;AACrC,4BAAA,MAAM,CAAC,OAAO,EAAE,UAAU,CAAC,GACzB,gBAAgB,CAAC,cAAc,CAAC,QAAQ,CAAC;4BAC3C,OAAO;gCACL,GAAG;gCACH,UAAU;gCACV,WAAW;AACX,gCAAA,OAAO,EAAE,OAAO,CAAC,SAAS,CAAC,OAAO,CAAC;6BAClB;;wBAGrB,OAAO;4BACL,GAAG;4BACH,WAAW;AACX,4BAAA,KAAK,EAAE,IAAI;4BACX,OAAO,EAAE,oBAAoB,GAAG,CAAA,EAAA,EAAK,QAAQ,CAAC,KAAK,IAAI,eAAe,CAAE,CAAA;yBACvD;AACrB,qBAAC;AACA,yBAAA,IAAI,CAAC,OAAO,MAAM,KAAI;AACrB,wBAAA,IAAI;AACF,4BAAA,IAAI,MAAM,CAAC,KAAK,IAAI,IAAI,EAAE;AACxB,gCAAA,OAAO,CAAC,KAAK,CACX,CAAA,eAAA,EAAkB,MAAM,CAAC,GAAG,CAAA,EAAA,EAAK,MAAM,CAAC,OAAO,CAAA,CAAE,CAClD;gCACD,OAAO;AACL,oCAAA,GAAG,MAAM;iCACV;;AAEH,4BAAA,MAAM,UAAU,GAAG,MAAM,aAAa,CAAC;gCACrC,KAAK;gCACL,QAAQ;gCACR,OAAO,EAAE,MAAM,CAAC,OAAO;AACxB,6BAAA,CAAC;4BACF,IAAI,eAAe,EAAE;AACnB,gCAAA,eAAe,CAAC,MAAM,CAAC,GAAG,CAAC;;4BAE7B,OAAO;AACL,gCAAA,GAAG,MAAM;gCACT,UAAU;6BACX;;wBACD,OAAO,KAAK,EAAE;AACd,4BAAA,OAAO,CAAC,KAAK,CAAC,mCAAmC,EAAE,KAAK,CAAC;4BACzD,OAAO;AACL,gCAAA,GAAG,MAAM;6BACV;;AAEL,qBAAC;AACA,yBAAA,KAAK,CAAC,CAAC,KAAK,KAAI;wBACf,OAAO,CAAC,KAAK,CAAC,CAAA,eAAA,EAAkB,WAAW,CAAG,CAAA,CAAA,EAAE,KAAK,CAAC;wBACtD,OAAO;AACL,4BAAA,GAAG,EAAE,WAAW;AAChB,4BAAA,KAAK,EAAE,IAAI;4BACX,OAAO,EAAE,oBAAoB,WAAW,CAAA,EAAA,EAAK,KAAK,CAAC,OAAO,IAAI,eAAe,CAAE,CAAA;yBAChF;AACH,qBAAC,CAAC;AACJ,oBAAA,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC;;AAExB,gBAAA,OAAO,MAAM,OAAO,CAAC,GAAG,CAAC,QAAQ,CAAC;;YAClC,OAAO,KAAK,EAAE;AACd,gBAAA,OAAO,CAAC,KAAK,CAAC,sBAAsB,EAAE,KAAK,CAAC;AAC5C,gBAAA,OAAO,EAAE;;SAEZ;KACF;AAED,IAAA,MAAM,aAAa,GAAG,OAAO,EAC3B,KAAK,EACL,KAAK,EACL,MAAM,EACN,eAAe,EACf,gBAAgB,GAOjB,KAAmB;QAClB,MAAM,YAAY,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,MAAM,CAAC;;AAE3C,QAAA,MAAM,OAAO,GAAG,MAAM,UAAU,CAAC,UAAU,CAAC;YAC1C,KAAK;AACL,YAAA,KAAK,EAAE,YAAY;YACnB,eAAe;AAChB,SAAA,CAAC;AACF,QAAA,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE;AAC5B,YAAA,IAAI,MAAM,CAAC,KAAK,KAAK,IAAI,EAAE;gBACzB;;AAEF,YAAA,MAAM,EAAE,GAAG,EAAE,OAAO,EAAE,WAAW,EAAE,UAAU,EAAE,UAAU,EAAE,GAAG,MAAM;YACpE,gBAAgB,GAAG,GAAG,EAAE;gBACtB,OAAO;gBACP,WAAW;gBACX,UAAU;gBACV,UAAU;AACX,aAAA,CAAC;;AAEN,KAAC;AAED,IAAA,MAAM,cAAc,GAAG,OAAO,EAC5B,MAAM,EACN,WAAW,EACX,KAAK,EACL,OAAO,GAAG,IAAI,EACd,eAAe,GACQ,KAAiC;AACxD,QAAA,IAAI;AACF,YAAA,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE;gBAChB,OAAO;AACL,oBAAA,OAAO,EAAE,EAAE;AACX,oBAAA,UAAU,EAAE,EAAE;AACd,oBAAA,MAAM,EAAE,EAAE;AACV,oBAAA,eAAe,EAAE,EAAE;iBACpB;;AACI,iBAAA,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,EAAE;gBAC/B,OAAO,MAAM,CAAC,IAAI;;YAGpB,IAAI,CAAC,OAAO,EAAE;gBACZ,MAAM,WAAW,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,MAAM,KACpD,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,eAAe,CAAC,CACtC;AAED,gBAAA,IAAI,CAAC,WAAW,CAAC,MAAM,EAAE;oBACvB,OAAO,MAAM,CAAC,IAAI;;AAGpB,gBAAA,MAAM,aAAa,GAAG,IAAI,GAAG,EAAyB;AACtD,gBAAA,aAAa,CAAC,GAAG,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,WAAW,CAAC,CAAC,CAAC,CAAC;AACtD,gBAAA,MAAM,gBAAgB,GAAG,0BAA0B,CAAC,aAAa,CAAC;AAClE,gBAAA,MAAM,aAAa,CAAC;oBAClB,KAAK;AACL,oBAAA,MAAM,EAAE,CAAC;oBACT,eAAe;oBACf,gBAAgB;oBAChB,KAAK,EAAE,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;AAC7B,iBAAA,CAAC;AAEF,gBAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;oBACnD,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;oBACrC,MAAM,aAAa,GAAG,aAAa,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC;oBACpD,IAAI,aAAa,EAAE;AACjB,wBAAA,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG;AACvB,4BAAA,GAAG,MAAM;AACT,4BAAA,GAAG,aAAa;yBACjB;;;gBAIL,OAAO,MAAM,CAAC,IAAI;;AAGpB,YAAA,MAAM,SAAS,GAAG,IAAI,GAAG,EAAyB;YAClD,MAAM,QAAQ,GAAa,EAAE;YAE7B,KAAK,MAAM,MAAM,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,EAAE;AACxC,gBAAA,IAAI,MAAM,CAAC,IAAI,EAAE;AACf,oBAAA,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC;oBAC1B,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC;;;AAItC,YAAA,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE;gBACzB,OAAO,MAAM,CAAC,IAAI;;AAGpB,YAAA,MAAM,gBAAgB,GAAG,0BAA0B,CAAC,SAAS,CAAC;AAC9D,YAAA,MAAM,aAAa,CAAC;gBAClB,KAAK;AACL,gBAAA,KAAK,EAAE,QAAQ;gBACf,eAAe;gBACf,gBAAgB;AAChB,gBAAA,MAAM,EAAE,WAAW;AACpB,aAAA,CAAC;AAEF,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBACnD,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;gBACrC,MAAM,aAAa,GAAG,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC;gBAChD,IAAI,aAAa,EAAE;AACjB,oBAAA,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG;AACvB,wBAAA,GAAG,MAAM;AACT,wBAAA,GAAG,aAAa;qBACjB;;;AAIL,YAAA,MAAM,iBAAiB,GAAG,MAAM,CAAC,IAAI,CAAC;iBACnC,MAAM,CACL,CAAC,MAAM,KACL,MAAM,CAAC,OAAO,IAAI,IAAI,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,QAAQ,CAAC;AAEjE,iBAAA,KAAK,CAAC,CAAC,EAAE,WAAW,CAAC;AAExB,YAAA,IAAI,iBAAiB,CAAC,MAAM,GAAG,CAAC,EAAE;AAChC,gBAAA,MAAM,CAAC,IAAI,CAAC,OAAO,GAAG,iBAAiB;;YAEzC,OAAO,MAAM,CAAC,IAAI;;QAClB,OAAO,KAAK,EAAE;AACd,YAAA,OAAO,CAAC,KAAK,CAAC,0BAA0B,EAAE,KAAK,CAAC;YAChD,OAAO;AACL,gBAAA,OAAO,EAAE,EAAE;AACX,gBAAA,UAAU,EAAE,EAAE;AACd,gBAAA,MAAM,EAAE,EAAE;AACV,gBAAA,eAAe,EAAE,EAAE;gBACnB,GAAG,MAAM,CAAC,IAAI;AACd,gBAAA,KAAK,EAAE,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;aAC9D;;AAEL,KAAC;IAED,OAAO;QACL,cAAc;QACd,UAAU;KACX;AACH;;;;;"}

package/dist/cjs/tools/search/tool.cjs

@@ -37,56 +37,21 @@ Examples:
 - "de" for Germany
 - "in" for India
 `.trim();
-/**
- * Creates a search tool with a schema that dynamically includes the country field
- * only when the searchProvider is 'serper'.
- *
- * @param config - The search tool configuration
- * @returns A DynamicStructuredTool with a schema that depends on the searchProvider
- */
-const createSearchTool = (config = {}) => {
-    const { searchProvider = 'serper', serperApiKey, searxngInstanceUrl, searxngApiKey, rerankerType = 'cohere', topResults = 5, strategies = ['no_extraction'], filterContent = true, firecrawlApiKey, firecrawlApiUrl, firecrawlFormats = ['markdown', 'html'], jinaApiKey, cohereApiKey, onSearchResults: _onSearchResults, } = config;
-    const querySchema = zod.z.string().describe(DEFAULT_QUERY_DESCRIPTION);
-    const schemaObject = {
-        query: querySchema,
-    };
-    if (searchProvider === 'serper') {
-        schemaObject.country = zod.z
-            .string()
-            .optional()
-            .describe(DEFAULT_COUNTRY_DESCRIPTION);
-    }
-    const SearchToolSchema = zod.z.object(schemaObject);
-    const searchAPI = search.createSearchAPI({
-        searchProvider,
-        serperApiKey,
-        searxngInstanceUrl,
-        searxngApiKey,
-    });
-    const firecrawlScraper = firecrawl.createFirecrawlScraper({
-        apiKey: firecrawlApiKey ?? process.env.FIRECRAWL_API_KEY,
-        apiUrl: firecrawlApiUrl,
-        formats: firecrawlFormats,
-    });
-    const selectedReranker = rerankers.createReranker({
-        rerankerType,
-        jinaApiKey,
-        cohereApiKey,
-    });
-    if (!selectedReranker) {
-        console.warn('No reranker selected. Using default ranking.');
-    }
-    const sourceProcessor = search.createSourceProcessor({
-        reranker: selectedReranker,
-        topResults}, firecrawlScraper);
-    const search$1 = async ({ query, country, proMode = true, maxSources = 5, onSearchResults, }) => {
+function createSearchProcessor({ searchAPI, sourceProcessor, onGetHighlights, }) {
+    return async function ({ query, country, proMode = true, maxSources = 5, onSearchResults, }) {
         try {
-            const sources = await searchAPI.getSources({ query, country });
-            onSearchResults?.(sources);
-            if (!sources.success) {
-                throw new Error(sources.error ?? 'Search failed');
+            const result = await searchAPI.getSources({ query, country });
+            onSearchResults?.(result);
+            if (!result.success) {
+                throw new Error(result.error ?? 'Search failed');
             }
-            const processedSources = await sourceProcessor.processSources(sources, maxSources, query, proMode);
+            const processedSources = await sourceProcessor.processSources({
+                query,
+                result,
+                proMode,
+                onGetHighlights,
+                numElements: maxSources,
+            });
             return highlights.expandHighlights(processedSources);
         }
         catch (error) {
@@ -100,17 +65,26 @@ const createSearchTool = (config = {}) => {
             };
         }
     };
+}
+function createOnSearchResults({ runnableConfig, onSearchResults, }) {
+    return function (results) {
+        if (!onSearchResults) {
+            return;
+        }
+        onSearchResults(results, runnableConfig);
+    };
+}
+function createTool({ schema, search, onSearchResults: _onSearchResults, }) {
     return tools.tool(async (params, runnableConfig) => {
         const { query, country: _c } = params;
         const country = typeof _c === 'string' && _c ? _c : undefined;
-        const searchResult = await search$1({
+        const searchResult = await search({
             query,
             country,
-            onSearchResults: _onSearchResults
-                ? (result) => {
-                    _onSearchResults(result, runnableConfig);
-                }
-                : undefined,
+            onSearchResults: createOnSearchResults({
+                runnableConfig,
+                onSearchResults: _onSearchResults,
+            }),
         });
         const turn = runnableConfig.toolCall?.turn ?? 0;
         const { output, references } = format.formatResultsForLLM(turn, searchResult);
@@ -118,8 +92,7 @@ const createSearchTool = (config = {}) => {
         return [output, { [_enum.Constants.WEB_SEARCH]: data }];
     }, {
         name: _enum.Constants.WEB_SEARCH,
-        description: `
-Real-time search. Results have required citation anchors.
+        description: `Real-time search. Results have required citation anchors.
 
 Note: Use ONCE per reply unless instructed otherwise.
 
@@ -142,9 +115,62 @@ Use anchor marker(s) immediately after the statement:
 
 **NEVER use markdown links, [1], or footnotes. CITE ONLY with anchors provided.**
 `.trim(),
-        schema: SearchToolSchema,
+        schema: schema,
         responseFormat: _enum.Constants.CONTENT_AND_ARTIFACT,
     });
+}
+/**
+ * Creates a search tool with a schema that dynamically includes the country field
+ * only when the searchProvider is 'serper'.
+ *
+ * @param config - The search tool configuration
+ * @returns A DynamicStructuredTool with a schema that depends on the searchProvider
+ */
+const createSearchTool = (config = {}) => {
+    const { searchProvider = 'serper', serperApiKey, searxngInstanceUrl, searxngApiKey, rerankerType = 'cohere', topResults = 5, strategies = ['no_extraction'], filterContent = true, firecrawlApiKey, firecrawlApiUrl, firecrawlFormats = ['markdown', 'html'], jinaApiKey, cohereApiKey, onSearchResults: _onSearchResults, onGetHighlights, } = config;
+    const querySchema = zod.z.string().describe(DEFAULT_QUERY_DESCRIPTION);
+    const schemaObject = {
+        query: querySchema,
+    };
+    if (searchProvider === 'serper') {
+        schemaObject.country = zod.z
+            .string()
+            .optional()
+            .describe(DEFAULT_COUNTRY_DESCRIPTION);
+    }
+    const toolSchema = zod.z.object(schemaObject);
+    const searchAPI = search.createSearchAPI({
+        searchProvider,
+        serperApiKey,
+        searxngInstanceUrl,
+        searxngApiKey,
+    });
+    const firecrawlScraper = firecrawl.createFirecrawlScraper({
+        apiKey: firecrawlApiKey ?? process.env.FIRECRAWL_API_KEY,
+        apiUrl: firecrawlApiUrl,
+        formats: firecrawlFormats,
+    });
+    const selectedReranker = rerankers.createReranker({
+        rerankerType,
+        jinaApiKey,
+        cohereApiKey,
+    });
+    if (!selectedReranker) {
+        console.warn('No reranker selected. Using default ranking.');
+    }
+    const sourceProcessor = search.createSourceProcessor({
+        reranker: selectedReranker,
+        topResults}, firecrawlScraper);
+    const search$1 = createSearchProcessor({
+        searchAPI,
+        sourceProcessor,
+        onGetHighlights,
+    });
+    return createTool({
+        search: search$1,
+        schema: toolSchema,
+        onSearchResults: _onSearchResults,
+    });
 };
 
 exports.createSearchTool = createSearchTool;

package/dist/cjs/tools/search/tool.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"tool.cjs","sources":["../../../../src/tools/search/tool.ts"],"sourcesContent":["/* eslint-disable no-console */\nimport { z } from 'zod';\nimport { tool, DynamicStructuredTool } from '@langchain/core/tools';\nimport type * as t from './types';\nimport { createSearchAPI, createSourceProcessor } from './search';\nimport { createFirecrawlScraper } from './firecrawl';\nimport { expandHighlights } from './highlights';\nimport { formatResultsForLLM } from './format';\nimport { createReranker } from './rerankers';\nimport { Constants } from '@/common';\n\nconst DEFAULT_QUERY_DESCRIPTION = `\nGUIDELINES:\n- Start broad, then narrow: Begin with key concepts, then refine with specifics\n- Think like sources: Use terminology experts would use in the field\n- Consider perspective: Frame queries from different viewpoints for better results\n- Quality over quantity: A precise 3-4 word query often beats lengthy sentences\n\nTECHNIQUES (combine for power searches):\n- EXACT PHRASES: Use quotes (\"climate change report\")\n- EXCLUDE TERMS: Use minus to remove unwanted results (-wikipedia)\n- SITE-SPECIFIC: Restrict to websites (site:edu research)\n- FILETYPE: Find specific documents (filetype:pdf study)\n- OR OPERATOR: Find alternatives (electric OR hybrid cars)\n- DATE RANGE: Recent information (data after:2020)\n- WILDCARDS: Use * for unknown terms (how to * bread)\n- SPECIFIC QUESTIONS: Use who/what/when/where/why/how\n- DOMAIN TERMS: Include technical terminology for specialized topics\n- CONCISE TERMS: Prioritize keywords over sentences\n`.trim();\n\nconst DEFAULT_COUNTRY_DESCRIPTION = `Country code to localize search results.\nUse standard 2-letter country codes: \"us\", \"uk\", \"ca\", \"de\", \"fr\", \"jp\", \"br\", etc.\nProvide this when the search should return results specific to a particular country.\nExamples:\n- \"us\" for United States (default)\n- \"de\" for Germany\n- \"in\" for India\n`.trim();\n\n/**\n * Creates a search tool with a schema that dynamically includes the country field\n * only when the searchProvider is 'serper'.\n *\n * @param config - The search tool configuration\n * @returns A DynamicStructuredTool with a schema that depends on the searchProvider\n */\nexport const createSearchTool = (\n  config: t.SearchToolConfig = {}\n): DynamicStructuredTool<typeof SearchToolSchema> => {\n  const {\n    searchProvider = 'serper',\n    serperApiKey,\n    searxngInstanceUrl,\n    searxngApiKey,\n    rerankerType = 'cohere',\n    topResults = 5,\n    strategies = ['no_extraction'],\n    filterContent = true,\n    firecrawlApiKey,\n    firecrawlApiUrl,\n    firecrawlFormats = ['markdown', 'html'],\n    jinaApiKey,\n    cohereApiKey,\n    onSearchResults: _onSearchResults,\n  } = config;\n\n  const querySchema = z.string().describe(DEFAULT_QUERY_DESCRIPTION);\n  const schemaObject: {\n    query: z.ZodString;\n    country?: z.ZodOptional<z.ZodString>;\n  } = {\n    query: querySchema,\n  };\n\n  if (searchProvider === 'serper') {\n    schemaObject.country = z\n      .string()\n      .optional()\n      .describe(DEFAULT_COUNTRY_DESCRIPTION);\n  }\n\n  const SearchToolSchema = z.object(schemaObject);\n\n  const searchAPI = createSearchAPI({\n    searchProvider,\n    serperApiKey,\n    searxngInstanceUrl,\n    searxngApiKey,\n  });\n\n  const firecrawlScraper = createFirecrawlScraper({\n    apiKey: firecrawlApiKey ?? process.env.FIRECRAWL_API_KEY,\n    apiUrl: firecrawlApiUrl,\n    formats: firecrawlFormats,\n  });\n\n  const selectedReranker = createReranker({\n    rerankerType,\n    jinaApiKey,\n    cohereApiKey,\n  });\n\n  if (!selectedReranker) {\n    console.warn('No reranker selected. Using default ranking.');\n  }\n\n  const sourceProcessor = createSourceProcessor(\n    {\n      reranker: selectedReranker,\n      topResults,\n      strategies,\n      filterContent,\n    },\n    firecrawlScraper\n  );\n\n  const search = async ({\n    query,\n    country,\n    proMode = true,\n    maxSources = 5,\n    onSearchResults,\n  }: {\n    query: string;\n    country?: string;\n    maxSources?: number;\n    proMode?: boolean;\n    onSearchResults?: (sources: t.SearchResult) => void;\n  }): Promise<t.SearchResultData> => {\n    try {\n      const sources = await searchAPI.getSources({ query, country });\n      onSearchResults?.(sources);\n\n      if (!sources.success) {\n        throw new Error(sources.error ?? 'Search failed');\n      }\n\n      const processedSources = await sourceProcessor.processSources(\n        sources,\n        maxSources,\n        query,\n        proMode\n      );\n      return expandHighlights(processedSources);\n    } catch (error) {\n      console.error('Error in search:', error);\n      return {\n        organic: [],\n        topStories: [],\n        images: [],\n        relatedSearches: [],\n        error: error instanceof Error ? error.message : String(error),\n      };\n    }\n  };\n\n  return tool<typeof SearchToolSchema>(\n    async (params, runnableConfig) => {\n      const { query, country: _c } = params;\n      const country = typeof _c === 'string' && _c ? _c : undefined;\n      const searchResult = await search({\n        query,\n        country,\n        onSearchResults: _onSearchResults\n          ? (result): void => {\n            _onSearchResults(result, runnableConfig);\n          }\n          : undefined,\n      });\n      const turn = runnableConfig.toolCall?.turn ?? 0;\n      const { output, references } = formatResultsForLLM(turn, searchResult);\n      const data: t.SearchResultData = { turn, ...searchResult, references };\n      return [output, { [Constants.WEB_SEARCH]: data }];\n    },\n    {\n      name: Constants.WEB_SEARCH,\n      description: `\nReal-time search. Results have required citation anchors.\n\nNote: Use ONCE per reply unless instructed otherwise.\n\nAnchors:\n- \\\\ue202turnXtypeY\n- X = turn idx, type = 'search' | 'news' | 'image' | 'ref', Y = item idx\n\nSpecial Markers:\n- \\\\ue203...\\\\ue204 — highlight start/end of cited text (for Standalone or Group citations)\n- \\\\ue200...\\\\ue201 — group block (e.g. \\\\ue200\\\\ue202turn0search1\\\\ue202turn0news2\\\\ue201)\n\n**CITE EVERY NON-OBVIOUS FACT/QUOTE:**\nUse anchor marker(s) immediately after the statement:\n- Standalone: \"Pure functions produce same output. \\\\ue202turn0search0\"\n- Standalone (multiple): \"Today's News \\\\ue202turn0search0\\\\ue202turn0news0\"\n- Highlight: \"\\\\ue203Highlight text.\\\\ue204\\\\ue202turn0news1\"\n- Group: \"Sources. \\\\ue200\\\\ue202turn0search0\\\\ue202turn0news1\\\\ue201\"\n- Group Highlight: \"\\\\ue203Highlight for group.\\\\ue204 \\\\ue200\\\\ue202turn0search0\\\\ue202turn0news1\\\\ue201\"\n- Image: \"See photo \\\\ue202turn0image0.\"\n\n**NEVER use markdown links, [1], or footnotes. CITE ONLY with anchors provided.**\n`.trim(),\n      schema: SearchToolSchema,\n      responseFormat: Constants.CONTENT_AND_ARTIFACT,\n    }\n  );\n};\n"],"names":["z","createSearchAPI","createFirecrawlScraper","createReranker","createSourceProcessor","search","expandHighlights","tool","formatResultsForLLM","Constants"],"mappings":";;;;;;;;;;;AAAA;AAWA,MAAM,yBAAyB,GAAG;;;;;;;;;;;;;;;;;;CAkBjC,CAAC,IAAI,EAAE;AAER,MAAM,2BAA2B,GAAG,CAAA;;;;;;;CAOnC,CAAC,IAAI,EAAE;AAER;;;;;;AAMG;MACU,gBAAgB,GAAG,CAC9B,MAA6B,GAAA,EAAE,KACmB;IAClD,MAAM,EACJ,cAAc,GAAG,QAAQ,EACzB,YAAY,EACZ,kBAAkB,EAClB,aAAa,EACb,YAAY,GAAG,QAAQ,EACvB,UAAU,GAAG,CAAC,EACd,UAAU,GAAG,CAAC,eAAe,CAAC,EAC9B,aAAa,GAAG,IAAI,EACpB,eAAe,EACf,eAAe,EACf,gBAAgB,GAAG,CAAC,UAAU,EAAE,MAAM,CAAC,EACvC,UAAU,EACV,YAAY,EACZ,eAAe,EAAE,gBAAgB,GAClC,GAAG,MAAM;IAEV,MAAM,WAAW,GAAGA,KAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,yBAAyB,CAAC;AAClE,IAAA,MAAM,YAAY,GAGd;AACF,QAAA,KAAK,EAAE,WAAW;KACnB;AAED,IAAA,IAAI,cAAc,KAAK,QAAQ,EAAE;QAC/B,YAAY,CAAC,OAAO,GAAGA;AACpB,aAAA,MAAM;AACN,aAAA,QAAQ;aACR,QAAQ,CAAC,2BAA2B,CAAC;;IAG1C,MAAM,gBAAgB,GAAGA,KAAC,CAAC,MAAM,CAAC,YAAY,CAAC;IAE/C,MAAM,SAAS,GAAGC,sBAAe,CAAC;QAChC,cAAc;QACd,YAAY;QACZ,kBAAkB;QAClB,aAAa;AACd,KAAA,CAAC;IAEF,MAAM,gBAAgB,GAAGC,gCAAsB,CAAC;AAC9C,QAAA,MAAM,EAAE,eAAe,IAAI,OAAO,CAAC,GAAG,CAAC,iBAAiB;AACxD,QAAA,MAAM,EAAE,eAAe;AACvB,QAAA,OAAO,EAAE,gBAAgB;AAC1B,KAAA,CAAC;IAEF,MAAM,gBAAgB,GAAGC,wBAAc,CAAC;QACtC,YAAY;QACZ,UAAU;QACV,YAAY;AACb,KAAA,CAAC;IAEF,IAAI,CAAC,gBAAgB,EAAE;AACrB,QAAA,OAAO,CAAC,IAAI,CAAC,8CAA8C,CAAC;;IAG9D,MAAM,eAAe,GAAGC,4BAAqB,CAC3C;AACE,QAAA,QAAQ,EAAE,gBAAgB;QAC1B,WAGD,EACD,gBAAgB,CACjB;IAED,MAAMC,QAAM,GAAG,OAAO,EACpB,KAAK,EACL,OAAO,EACP,OAAO,GAAG,IAAI,EACd,UAAU,GAAG,CAAC,EACd,eAAe,GAOhB,KAAiC;AAChC,QAAA,IAAI;AACF,YAAA,MAAM,OAAO,GAAG,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC;AAC9D,YAAA,eAAe,GAAG,OAAO,CAAC;AAE1B,YAAA,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE;gBACpB,MAAM,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,IAAI,eAAe,CAAC;;AAGnD,YAAA,MAAM,gBAAgB,GAAG,MAAM,eAAe,CAAC,cAAc,CAC3D,OAAO,EACP,UAAU,EACV,KAAK,EACL,OAAO,CACR;AACD,YAAA,OAAOC,2BAAgB,CAAC,gBAAgB,CAAC;;QACzC,OAAO,KAAK,EAAE;AACd,YAAA,OAAO,CAAC,KAAK,CAAC,kBAAkB,EAAE,KAAK,CAAC;YACxC,OAAO;AACL,gBAAA,OAAO,EAAE,EAAE;AACX,gBAAA,UAAU,EAAE,EAAE;AACd,gBAAA,MAAM,EAAE,EAAE;AACV,gBAAA,eAAe,EAAE,EAAE;AACnB,gBAAA,KAAK,EAAE,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;aAC9D;;AAEL,KAAC;IAED,OAAOC,UAAI,CACT,OAAO,MAAM,EAAE,cAAc,KAAI;QAC/B,MAAM,EAAE,KAAK,EAAE,OAAO,EAAE,EAAE,EAAE,GAAG,MAAM;AACrC,QAAA,MAAM,OAAO,GAAG,OAAO,EAAE,KAAK,QAAQ,IAAI,EAAE,GAAG,EAAE,GAAG,SAAS;AAC7D,QAAA,MAAM,YAAY,GAAG,MAAMF,QAAM,CAAC;YAChC,KAAK;YACL,OAAO;AACP,YAAA,eAAe,EAAE;AACf,kBAAE,CAAC,MAAM,KAAU;AACjB,oBAAA,gBAAgB,CAAC,MAAM,EAAE,cAAc,CAAC;;AAE1C,kBAAE,SAAS;AACd,SAAA,CAAC;QACF,MAAM,IAAI,GAAG,cAAc,CAAC,QAAQ,EAAE,IAAI,IAAI,CAAC;AAC/C,QAAA,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,GAAGG,0BAAmB,CAAC,IAAI,EAAE,YAAY,CAAC;QACtE,MAAM,IAAI,GAAuB,EAAE,IAAI,EAAE,GAAG,YAAY,EAAE,UAAU,EAAE;AACtE,QAAA,OAAO,CAAC,MAAM,EAAE,EAAE,CAACC,eAAS,CAAC,UAAU,GAAG,IAAI,EAAE,CAAC;AACnD,KAAC,EACD;QACE,IAAI,EAAEA,eAAS,CAAC,UAAU;AAC1B,QAAA,WAAW,EAAE;;;;;;;;;;;;;;;;;;;;;;;AAuBlB,CAAA,CAAC,IAAI,EAAE;AACF,QAAA,MAAM,EAAE,gBAAgB;QACxB,cAAc,EAAEA,eAAS,CAAC,oBAAoB;AAC/C,KAAA,CACF;AACH;;;;"}
+{"version":3,"file":"tool.cjs","sources":["../../../../src/tools/search/tool.ts"],"sourcesContent":["/* eslint-disable no-console */\nimport { z } from 'zod';\nimport { tool, DynamicStructuredTool } from '@langchain/core/tools';\nimport type { RunnableConfig } from '@langchain/core/runnables';\nimport type * as t from './types';\nimport { createSearchAPI, createSourceProcessor } from './search';\nimport { createFirecrawlScraper } from './firecrawl';\nimport { expandHighlights } from './highlights';\nimport { formatResultsForLLM } from './format';\nimport { createReranker } from './rerankers';\nimport { Constants } from '@/common';\n\nconst DEFAULT_QUERY_DESCRIPTION = `\nGUIDELINES:\n- Start broad, then narrow: Begin with key concepts, then refine with specifics\n- Think like sources: Use terminology experts would use in the field\n- Consider perspective: Frame queries from different viewpoints for better results\n- Quality over quantity: A precise 3-4 word query often beats lengthy sentences\n\nTECHNIQUES (combine for power searches):\n- EXACT PHRASES: Use quotes (\"climate change report\")\n- EXCLUDE TERMS: Use minus to remove unwanted results (-wikipedia)\n- SITE-SPECIFIC: Restrict to websites (site:edu research)\n- FILETYPE: Find specific documents (filetype:pdf study)\n- OR OPERATOR: Find alternatives (electric OR hybrid cars)\n- DATE RANGE: Recent information (data after:2020)\n- WILDCARDS: Use * for unknown terms (how to * bread)\n- SPECIFIC QUESTIONS: Use who/what/when/where/why/how\n- DOMAIN TERMS: Include technical terminology for specialized topics\n- CONCISE TERMS: Prioritize keywords over sentences\n`.trim();\n\nconst DEFAULT_COUNTRY_DESCRIPTION = `Country code to localize search results.\nUse standard 2-letter country codes: \"us\", \"uk\", \"ca\", \"de\", \"fr\", \"jp\", \"br\", etc.\nProvide this when the search should return results specific to a particular country.\nExamples:\n- \"us\" for United States (default)\n- \"de\" for Germany\n- \"in\" for India\n`.trim();\n\nfunction createSearchProcessor({\n  searchAPI,\n  sourceProcessor,\n  onGetHighlights,\n}: {\n  searchAPI: ReturnType<typeof createSearchAPI>;\n  sourceProcessor: ReturnType<typeof createSourceProcessor>;\n  onGetHighlights: t.SearchToolConfig['onGetHighlights'];\n}) {\n  return async function ({\n    query,\n    country,\n    proMode = true,\n    maxSources = 5,\n    onSearchResults,\n  }: {\n    query: string;\n    country?: string;\n    maxSources?: number;\n    proMode?: boolean;\n    onSearchResults: t.SearchToolConfig['onSearchResults'];\n  }): Promise<t.SearchResultData> {\n    try {\n      const result = await searchAPI.getSources({ query, country });\n      onSearchResults?.(result);\n\n      if (!result.success) {\n        throw new Error(result.error ?? 'Search failed');\n      }\n\n      const processedSources = await sourceProcessor.processSources({\n        query,\n        result,\n        proMode,\n        onGetHighlights,\n        numElements: maxSources,\n      });\n      return expandHighlights(processedSources);\n    } catch (error) {\n      console.error('Error in search:', error);\n      return {\n        organic: [],\n        topStories: [],\n        images: [],\n        relatedSearches: [],\n        error: error instanceof Error ? error.message : String(error),\n      };\n    }\n  };\n}\n\nfunction createOnSearchResults({\n  runnableConfig,\n  onSearchResults,\n}: {\n  runnableConfig: RunnableConfig;\n  onSearchResults: t.SearchToolConfig['onSearchResults'];\n}) {\n  return function (results: t.SearchResult): void {\n    if (!onSearchResults) {\n      return;\n    }\n    onSearchResults(results, runnableConfig);\n  };\n}\n\nfunction createTool({\n  schema,\n  search,\n  onSearchResults: _onSearchResults,\n}: {\n  schema: t.SearchToolSchema;\n  search: ReturnType<typeof createSearchProcessor>;\n  onSearchResults: t.SearchToolConfig['onSearchResults'];\n}): DynamicStructuredTool<typeof schema> {\n  return tool<typeof schema>(\n    async (params, runnableConfig) => {\n      const { query, country: _c } = params;\n      const country = typeof _c === 'string' && _c ? _c : undefined;\n      const searchResult = await search({\n        query,\n        country,\n        onSearchResults: createOnSearchResults({\n          runnableConfig,\n          onSearchResults: _onSearchResults,\n        }),\n      });\n      const turn = runnableConfig.toolCall?.turn ?? 0;\n      const { output, references } = formatResultsForLLM(turn, searchResult);\n      const data: t.SearchResultData = { turn, ...searchResult, references };\n      return [output, { [Constants.WEB_SEARCH]: data }];\n    },\n    {\n      name: Constants.WEB_SEARCH,\n      description: `Real-time search. Results have required citation anchors.\n\nNote: Use ONCE per reply unless instructed otherwise.\n\nAnchors:\n- \\\\ue202turnXtypeY\n- X = turn idx, type = 'search' | 'news' | 'image' | 'ref', Y = item idx\n\nSpecial Markers:\n- \\\\ue203...\\\\ue204 — highlight start/end of cited text (for Standalone or Group citations)\n- \\\\ue200...\\\\ue201 — group block (e.g. \\\\ue200\\\\ue202turn0search1\\\\ue202turn0news2\\\\ue201)\n\n**CITE EVERY NON-OBVIOUS FACT/QUOTE:**\nUse anchor marker(s) immediately after the statement:\n- Standalone: \"Pure functions produce same output. \\\\ue202turn0search0\"\n- Standalone (multiple): \"Today's News \\\\ue202turn0search0\\\\ue202turn0news0\"\n- Highlight: \"\\\\ue203Highlight text.\\\\ue204\\\\ue202turn0news1\"\n- Group: \"Sources. \\\\ue200\\\\ue202turn0search0\\\\ue202turn0news1\\\\ue201\"\n- Group Highlight: \"\\\\ue203Highlight for group.\\\\ue204 \\\\ue200\\\\ue202turn0search0\\\\ue202turn0news1\\\\ue201\"\n- Image: \"See photo \\\\ue202turn0image0.\"\n\n**NEVER use markdown links, [1], or footnotes. CITE ONLY with anchors provided.**\n`.trim(),\n      schema: schema,\n      responseFormat: Constants.CONTENT_AND_ARTIFACT,\n    }\n  );\n}\n\n/**\n * Creates a search tool with a schema that dynamically includes the country field\n * only when the searchProvider is 'serper'.\n *\n * @param config - The search tool configuration\n * @returns A DynamicStructuredTool with a schema that depends on the searchProvider\n */\nexport const createSearchTool = (\n  config: t.SearchToolConfig = {}\n): DynamicStructuredTool<typeof toolSchema> => {\n  const {\n    searchProvider = 'serper',\n    serperApiKey,\n    searxngInstanceUrl,\n    searxngApiKey,\n    rerankerType = 'cohere',\n    topResults = 5,\n    strategies = ['no_extraction'],\n    filterContent = true,\n    firecrawlApiKey,\n    firecrawlApiUrl,\n    firecrawlFormats = ['markdown', 'html'],\n    jinaApiKey,\n    cohereApiKey,\n    onSearchResults: _onSearchResults,\n    onGetHighlights,\n  } = config;\n\n  const querySchema = z.string().describe(DEFAULT_QUERY_DESCRIPTION);\n  const schemaObject: {\n    query: z.ZodString;\n    country?: z.ZodOptional<z.ZodString>;\n  } = {\n    query: querySchema,\n  };\n\n  if (searchProvider === 'serper') {\n    schemaObject.country = z\n      .string()\n      .optional()\n      .describe(DEFAULT_COUNTRY_DESCRIPTION);\n  }\n\n  const toolSchema = z.object(schemaObject);\n\n  const searchAPI = createSearchAPI({\n    searchProvider,\n    serperApiKey,\n    searxngInstanceUrl,\n    searxngApiKey,\n  });\n\n  const firecrawlScraper = createFirecrawlScraper({\n    apiKey: firecrawlApiKey ?? process.env.FIRECRAWL_API_KEY,\n    apiUrl: firecrawlApiUrl,\n    formats: firecrawlFormats,\n  });\n\n  const selectedReranker = createReranker({\n    rerankerType,\n    jinaApiKey,\n    cohereApiKey,\n  });\n\n  if (!selectedReranker) {\n    console.warn('No reranker selected. Using default ranking.');\n  }\n\n  const sourceProcessor = createSourceProcessor(\n    {\n      reranker: selectedReranker,\n      topResults,\n      strategies,\n      filterContent,\n    },\n    firecrawlScraper\n  );\n\n  const search = createSearchProcessor({\n    searchAPI,\n    sourceProcessor,\n    onGetHighlights,\n  });\n\n  return createTool({\n    search,\n    schema: toolSchema,\n    onSearchResults: _onSearchResults,\n  });\n};\n"],"names":["expandHighlights","tool","formatResultsForLLM","Constants","z","createSearchAPI","createFirecrawlScraper","createReranker","createSourceProcessor","search"],"mappings":";;;;;;;;;;;AAAA;AAYA,MAAM,yBAAyB,GAAG;;;;;;;;;;;;;;;;;;CAkBjC,CAAC,IAAI,EAAE;AAER,MAAM,2BAA2B,GAAG,CAAA;;;;;;;CAOnC,CAAC,IAAI,EAAE;AAER,SAAS,qBAAqB,CAAC,EAC7B,SAAS,EACT,eAAe,EACf,eAAe,GAKhB,EAAA;AACC,IAAA,OAAO,gBAAgB,EACrB,KAAK,EACL,OAAO,EACP,OAAO,GAAG,IAAI,EACd,UAAU,GAAG,CAAC,EACd,eAAe,GAOhB,EAAA;AACC,QAAA,IAAI;AACF,YAAA,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,UAAU,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC;AAC7D,YAAA,eAAe,GAAG,MAAM,CAAC;AAEzB,YAAA,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE;gBACnB,MAAM,IAAI,KAAK,CAAC,MAAM,CAAC,KAAK,IAAI,eAAe,CAAC;;AAGlD,YAAA,MAAM,gBAAgB,GAAG,MAAM,eAAe,CAAC,cAAc,CAAC;gBAC5D,KAAK;gBACL,MAAM;gBACN,OAAO;gBACP,eAAe;AACf,gBAAA,WAAW,EAAE,UAAU;AACxB,aAAA,CAAC;AACF,YAAA,OAAOA,2BAAgB,CAAC,gBAAgB,CAAC;;QACzC,OAAO,KAAK,EAAE;AACd,YAAA,OAAO,CAAC,KAAK,CAAC,kBAAkB,EAAE,KAAK,CAAC;YACxC,OAAO;AACL,gBAAA,OAAO,EAAE,EAAE;AACX,gBAAA,UAAU,EAAE,EAAE;AACd,gBAAA,MAAM,EAAE,EAAE;AACV,gBAAA,eAAe,EAAE,EAAE;AACnB,gBAAA,KAAK,EAAE,KAAK,YAAY,KAAK,GAAG,KAAK,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;aAC9D;;AAEL,KAAC;AACH;AAEA,SAAS,qBAAqB,CAAC,EAC7B,cAAc,EACd,eAAe,GAIhB,EAAA;AACC,IAAA,OAAO,UAAU,OAAuB,EAAA;QACtC,IAAI,CAAC,eAAe,EAAE;YACpB;;AAEF,QAAA,eAAe,CAAC,OAAO,EAAE,cAAc,CAAC;AAC1C,KAAC;AACH;AAEA,SAAS,UAAU,CAAC,EAClB,MAAM,EACN,MAAM,EACN,eAAe,EAAE,gBAAgB,GAKlC,EAAA;IACC,OAAOC,UAAI,CACT,OAAO,MAAM,EAAE,cAAc,KAAI;QAC/B,MAAM,EAAE,KAAK,EAAE,OAAO,EAAE,EAAE,EAAE,GAAG,MAAM;AACrC,QAAA,MAAM,OAAO,GAAG,OAAO,EAAE,KAAK,QAAQ,IAAI,EAAE,GAAG,EAAE,GAAG,SAAS;AAC7D,QAAA,MAAM,YAAY,GAAG,MAAM,MAAM,CAAC;YAChC,KAAK;YACL,OAAO;YACP,eAAe,EAAE,qBAAqB,CAAC;gBACrC,cAAc;AACd,gBAAA,eAAe,EAAE,gBAAgB;aAClC,CAAC;AACH,SAAA,CAAC;QACF,MAAM,IAAI,GAAG,cAAc,CAAC,QAAQ,EAAE,IAAI,IAAI,CAAC;AAC/C,QAAA,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,GAAGC,0BAAmB,CAAC,IAAI,EAAE,YAAY,CAAC;QACtE,MAAM,IAAI,GAAuB,EAAE,IAAI,EAAE,GAAG,YAAY,EAAE,UAAU,EAAE;AACtE,QAAA,OAAO,CAAC,MAAM,EAAE,EAAE,CAACC,eAAS,CAAC,UAAU,GAAG,IAAI,EAAE,CAAC;AACnD,KAAC,EACD;QACE,IAAI,EAAEA,eAAS,CAAC,UAAU;AAC1B,QAAA,WAAW,EAAE,CAAA;;;;;;;;;;;;;;;;;;;;;;AAsBlB,CAAA,CAAC,IAAI,EAAE;AACF,QAAA,MAAM,EAAE,MAAM;QACd,cAAc,EAAEA,eAAS,CAAC,oBAAoB;AAC/C,KAAA,CACF;AACH;AAEA;;;;;;AAMG;MACU,gBAAgB,GAAG,CAC9B,MAA6B,GAAA,EAAE,KACa;IAC5C,MAAM,EACJ,cAAc,GAAG,QAAQ,EACzB,YAAY,EACZ,kBAAkB,EAClB,aAAa,EACb,YAAY,GAAG,QAAQ,EACvB,UAAU,GAAG,CAAC,EACd,UAAU,GAAG,CAAC,eAAe,CAAC,EAC9B,aAAa,GAAG,IAAI,EACpB,eAAe,EACf,eAAe,EACf,gBAAgB,GAAG,CAAC,UAAU,EAAE,MAAM,CAAC,EACvC,UAAU,EACV,YAAY,EACZ,eAAe,EAAE,gBAAgB,EACjC,eAAe,GAChB,GAAG,MAAM;IAEV,MAAM,WAAW,GAAGC,KAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,yBAAyB,CAAC;AAClE,IAAA,MAAM,YAAY,GAGd;AACF,QAAA,KAAK,EAAE,WAAW;KACnB;AAED,IAAA,IAAI,cAAc,KAAK,QAAQ,EAAE;QAC/B,YAAY,CAAC,OAAO,GAAGA;AACpB,aAAA,MAAM;AACN,aAAA,QAAQ;aACR,QAAQ,CAAC,2BAA2B,CAAC;;IAG1C,MAAM,UAAU,GAAGA,KAAC,CAAC,MAAM,CAAC,YAAY,CAAC;IAEzC,MAAM,SAAS,GAAGC,sBAAe,CAAC;QAChC,cAAc;QACd,YAAY;QACZ,kBAAkB;QAClB,aAAa;AACd,KAAA,CAAC;IAEF,MAAM,gBAAgB,GAAGC,gCAAsB,CAAC;AAC9C,QAAA,MAAM,EAAE,eAAe,IAAI,OAAO,CAAC,GAAG,CAAC,iBAAiB;AACxD,QAAA,MAAM,EAAE,eAAe;AACvB,QAAA,OAAO,EAAE,gBAAgB;AAC1B,KAAA,CAAC;IAEF,MAAM,gBAAgB,GAAGC,wBAAc,CAAC;QACtC,YAAY;QACZ,UAAU;QACV,YAAY;AACb,KAAA,CAAC;IAEF,IAAI,CAAC,gBAAgB,EAAE;AACrB,QAAA,OAAO,CAAC,IAAI,CAAC,8CAA8C,CAAC;;IAG9D,MAAM,eAAe,GAAGC,4BAAqB,CAC3C;AACE,QAAA,QAAQ,EAAE,gBAAgB;QAC1B,WAGD,EACD,gBAAgB,CACjB;IAED,MAAMC,QAAM,GAAG,qBAAqB,CAAC;QACnC,SAAS;QACT,eAAe;QACf,eAAe;AAChB,KAAA,CAAC;AAEF,IAAA,OAAO,UAAU,CAAC;gBAChBA,QAAM;AACN,QAAA,MAAM,EAAE,UAAU;AAClB,QAAA,eAAe,EAAE,gBAAgB;AAClC,KAAA,CAAC;AACJ;;;;"}

package/dist/cjs/tools/search/utils.cjs

@@ -17,11 +17,13 @@ const getDomainName = (link, metadata) => {
 function getAttribution(link, metadata) {
     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((attr) => attr != null && typeof attr === 'string' && attr.trim() !== '');
     if (attribution != null) {

package/dist/cjs/tools/search/utils.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"utils.cjs","sources":["../../../../src/tools/search/utils.ts"],"sourcesContent":["/* eslint-disable no-console */\nimport type * as t from './types';\n\nexport const getDomainName = (\n  link: string,\n  metadata?: t.ScrapeMetadata\n): string | undefined => {\n  try {\n    const url = metadata?.sourceURL ?? metadata?.url ?? (link || '');\n    const domain = new URL(url).hostname.replace(/^www\\./, '');\n    if (domain) {\n      return domain;\n    }\n  } catch (e) {\n    // URL parsing failed\n    console.error('Error parsing URL:', e);\n  }\n\n  return;\n};\n\nexport function getAttribution(\n  link: string,\n  metadata?: t.ScrapeMetadata\n): string | undefined {\n  if (!metadata) return getDomainName(link, metadata);\n\n  const possibleAttributions = [\n    metadata.ogSiteName,\n    metadata['og:site_name'],\n    metadata.title?.split('|').pop()?.trim(),\n    metadata['twitter:site']?.replace(/^@/, ''),\n  ];\n\n  const attribution = possibleAttributions.find(\n    (attr) => attr != null && typeof attr === 'string' && attr.trim() !== ''\n  );\n  if (attribution != null) {\n    return attribution;\n  }\n\n  return getDomainName(link, metadata);\n}\n"],"names":[],"mappings":";;MAGa,aAAa,GAAG,CAC3B,IAAY,EACZ,QAA2B,KACL;AACtB,IAAA,IAAI;AACF,QAAA,MAAM,GAAG,GAAG,QAAQ,EAAE,SAAS,IAAI,QAAQ,EAAE,GAAG,KAAK,IAAI,IAAI,EAAE,CAAC;AAChE,QAAA,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC;QAC1D,IAAI,MAAM,EAAE;AACV,YAAA,OAAO,MAAM;;;IAEf,OAAO,CAAC,EAAE;;AAEV,QAAA,OAAO,CAAC,KAAK,CAAC,oBAAoB,EAAE,CAAC,CAAC;;IAGxC;AACF;AAEgB,SAAA,cAAc,CAC5B,IAAY,EACZ,QAA2B,EAAA;AAE3B,IAAA,IAAI,CAAC,QAAQ;AAAE,QAAA,OAAO,aAAa,CAAC,IAAI,EAAE,QAAQ,CAAC;AAEnD,IAAA,MAAM,oBAAoB,GAAG;AAC3B,QAAA,QAAQ,CAAC,UAAU;QACnB,QAAQ,CAAC,cAAc,CAAC;AACxB,QAAA,QAAQ,CAAC,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,IAAI,EAAE;QACxC,QAAQ,CAAC,cAAc,CAAC,EAAE,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC;KAC5C;IAED,MAAM,WAAW,GAAG,oBAAoB,CAAC,IAAI,CAC3C,CAAC,IAAI,KAAK,IAAI,IAAI,IAAI,IAAI,OAAO,IAAI,KAAK,QAAQ,IAAI,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CACzE;AACD,IAAA,IAAI,WAAW,IAAI,IAAI,EAAE;AACvB,QAAA,OAAO,WAAW;;AAGpB,IAAA,OAAO,aAAa,CAAC,IAAI,EAAE,QAAQ,CAAC;AACtC;;;;;"}
+{"version":3,"file":"utils.cjs","sources":["../../../../src/tools/search/utils.ts"],"sourcesContent":["/* eslint-disable no-console */\nimport type * as t from './types';\n\nexport const getDomainName = (\n  link: string,\n  metadata?: t.ScrapeMetadata\n): string | undefined => {\n  try {\n    const url = metadata?.sourceURL ?? metadata?.url ?? (link || '');\n    const domain = new URL(url).hostname.replace(/^www\\./, '');\n    if (domain) {\n      return domain;\n    }\n  } catch (e) {\n    // URL parsing failed\n    console.error('Error parsing URL:', e);\n  }\n\n  return;\n};\n\nexport function getAttribution(\n  link: string,\n  metadata?: t.ScrapeMetadata\n): string | undefined {\n  if (!metadata) return getDomainName(link, metadata);\n\n  const twitterSite = metadata['twitter:site'];\n  const twitterSiteFormatted =\n    typeof twitterSite === 'string' ? twitterSite.replace(/^@/, '') : undefined;\n\n  const possibleAttributions = [\n    metadata.ogSiteName,\n    metadata['og:site_name'],\n    metadata.title?.split('|').pop()?.trim(),\n    twitterSiteFormatted,\n  ];\n\n  const attribution = possibleAttributions.find(\n    (attr) => attr != null && typeof attr === 'string' && attr.trim() !== ''\n  );\n  if (attribution != null) {\n    return attribution;\n  }\n\n  return getDomainName(link, metadata);\n}\n"],"names":[],"mappings":";;MAGa,aAAa,GAAG,CAC3B,IAAY,EACZ,QAA2B,KACL;AACtB,IAAA,IAAI;AACF,QAAA,MAAM,GAAG,GAAG,QAAQ,EAAE,SAAS,IAAI,QAAQ,EAAE,GAAG,KAAK,IAAI,IAAI,EAAE,CAAC;AAChE,QAAA,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC;QAC1D,IAAI,MAAM,EAAE;AACV,YAAA,OAAO,MAAM;;;IAEf,OAAO,CAAC,EAAE;;AAEV,QAAA,OAAO,CAAC,KAAK,CAAC,oBAAoB,EAAE,CAAC,CAAC;;IAGxC;AACF;AAEgB,SAAA,cAAc,CAC5B,IAAY,EACZ,QAA2B,EAAA;AAE3B,IAAA,IAAI,CAAC,QAAQ;AAAE,QAAA,OAAO,aAAa,CAAC,IAAI,EAAE,QAAQ,CAAC;AAEnD,IAAA,MAAM,WAAW,GAAG,QAAQ,CAAC,cAAc,CAAC;IAC5C,MAAM,oBAAoB,GACxB,OAAO,WAAW,KAAK,QAAQ,GAAG,WAAW,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,GAAG,SAAS;AAE7E,IAAA,MAAM,oBAAoB,GAAG;AAC3B,QAAA,QAAQ,CAAC,UAAU;QACnB,QAAQ,CAAC,cAAc,CAAC;AACxB,QAAA,QAAQ,CAAC,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,IAAI,EAAE;QACxC,oBAAoB;KACrB;IAED,MAAM,WAAW,GAAG,oBAAoB,CAAC,IAAI,CAC3C,CAAC,IAAI,KAAK,IAAI,IAAI,IAAI,IAAI,OAAO,IAAI,KAAK,QAAQ,IAAI,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CACzE;AACD,IAAA,IAAI,WAAW,IAAI,IAAI,EAAE;AACvB,QAAA,OAAO,WAAW;;AAGpB,IAAA,OAAO,aAAa,CAAC,IAAI,EAAE,QAAQ,CAAC;AACtC;;;;;"}

package/dist/esm/tools/search/search.mjs

@@ -38,7 +38,7 @@ const chunker = {
         return Promise.all(promises);
     },
 };
-const createSourceUpdateCallback = (sourceMap) => {
+function createSourceUpdateCallback(sourceMap) {
     return (link, update) => {
         const source = sourceMap.get(link);
         if (source) {
@@ -48,7 +48,7 @@ const createSourceUpdateCallback = (sourceMap) => {
             });
         }
     };
-};
+}
 const getHighlights = async ({ query, content, reranker, topResults = 5, }) => {
     if (!content) {
         console.warn('No content provided for highlights');
@@ -223,11 +223,12 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
     reranker, } = config;
     const firecrawlScraper = scraperInstance;
     const webScraper = {
-        scrapeMany: async ({ query, links, }) => {
+        scrapeMany: async ({ query, links, onGetHighlights, }) => {
             console.log(`Scraping ${links.length} links with Firecrawl`);
             const promises = [];
             try {
-                for (const currentLink of links) {
+                for (let i = 0; i < links.length; i++) {
+                    const currentLink = links[i];
                     const promise = firecrawlScraper
                         .scrapeUrl(currentLink, {})
                         .then(([url, response]) => {
@@ -261,6 +262,9 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
                                 reranker,
                                 content: result.content,
                             });
+                            if (onGetHighlights) {
+                                onGetHighlights(result.url);
+                            }
                             return {
                                 ...result,
                                 highlights,
@@ -291,10 +295,14 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
             }
         },
     };
-    const fetchContents = async ({ links, query, target, onContentScraped, }) => {
+    const fetchContents = async ({ links, query, target, onGetHighlights, onContentScraped, }) => {
         const initialLinks = links.slice(0, target);
         // const remainingLinks = links.slice(target).reverse();
-        const results = await webScraper.scrapeMany({ query, links: initialLinks });
+        const results = await webScraper.scrapeMany({
+            query,
+            links: initialLinks,
+            onGetHighlights,
+        });
         for (const result of results) {
             if (result.error === true) {
                 continue;
@@ -308,7 +316,7 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
             });
         }
     };
-    const processSources = async (result, numElements, query, proMode = false) => {
+    const processSources = async ({ result, numElements, query, proMode = true, onGetHighlights, }) => {
         try {
             if (!result.data) {
                 return {
@@ -332,6 +340,7 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
                 await fetchContents({
                     query,
                     target: 1,
+                    onGetHighlights,
                     onContentScraped,
                     links: [wikiSources[0].link],
                 });
@@ -360,8 +369,9 @@ const createSourceProcessor = (config = {}, scraperInstance) => {
             }
             const onContentScraped = createSourceUpdateCallback(sourceMap);
             await fetchContents({
-                links: allLinks,
                 query,
+                links: allLinks,
+                onGetHighlights,
                 onContentScraped,
                 target: numElements,
             });