one-search-mcp/dist/index.cjs.map

One Search MCP Server, Web Search & Crawl & Scraper & Extract, support Firecrawl, SearXNG, Tavily, DuckDuckGo, Bing, etc.

{"version":3,"sources":["../src/index.ts","../src/search/bing.ts","../src/search/duckduckgo.ts","../src/search/searxng.ts","../src/search/tavily.ts","../src/libs/browser/types.ts","../src/libs/browser/finder.ts","../src/libs/browser/base.ts","../src/libs/browser/local.ts","../src/libs/browser/remote.ts","../src/libs/browser-search/readability.ts","../src/libs/browser-search/search.ts","../src/libs/browser-search/utils.ts","../src/libs/browser-search/queue.ts","../src/libs/browser-search/engines/bing.ts","../src/libs/browser-search/engines/baidu.ts","../src/libs/browser-search/engines/sogou.ts","../src/libs/browser-search/engines/google.ts","../src/libs/browser-search/engines/get.ts","../src/search/local.ts","../src/tools.ts"],"sourcesContent":["#!/usr/bin/env node\n\nimport { Server } from '@modelcontextprotocol/sdk/server/index.js';\nimport  { CallToolRequestSchema, ListToolsRequestSchema } from '@modelcontextprotocol/sdk/types.js';\nimport { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio.js';\nimport { ISearchRequestOptions, ISearchResponse, SearchProvider } from './interface.js';\nimport { bingSearch, duckDuckGoSearch, searxngSearch, tavilySearch, localSearch } from './search/index.js';\nimport { SEARCH_TOOL, EXTRACT_TOOL, SCRAPE_TOOL, MAP_TOOL } from './tools.js';\nimport FirecrawlApp, { MapParams, ScrapeParams } from '@mendable/firecrawl-js';\nimport dotenvx from '@dotenvx/dotenvx';\nimport { SafeSearchType } from 'duck-duck-scrape';\n\ndotenvx.config();\n\n// search api\nconst SEARCH_API_URL = process.env.SEARCH_API_URL;\nconst SEARCH_API_KEY = process.env.SEARCH_API_KEY;\nconst SEARCH_PROVIDER: SearchProvider = process.env.SEARCH_PROVIDER as SearchProvider ?? 'local';\n\n// search query params\nconst SAFE_SEARCH = process.env.SAFE_SEARCH ?? 0;\nconst LIMIT = process.env.LIMIT ?? 10;\nconst CATEGORIES = process.env.CATEGORIES ?? 'general';\nconst ENGINES = process.env.ENGINES ?? 'all';\nconst FORMAT = process.env.FORMAT ?? 'json';\nconst LANGUAGE = process.env.LANGUAGE ?? 'auto';\nconst TIME_RANGE = process.env.TIME_RANGE ?? '';\nconst DEFAULT_TIMEOUT = process.env.TIMEOUT ?? 10000;\n\n// firecrawl api\nconst FIRECRAWL_API_KEY = process.env.FIRECRAWL_API_KEY;\nconst FIRECRAWL_API_URL = process.env.FIRECRAWL_API_URL;\n\n// firecrawl client\nconst firecrawl = new FirecrawlApp({\n  apiKey: FIRECRAWL_API_KEY ?? '',\n  ...(FIRECRAWL_API_URL ? { apiUrl: FIRECRAWL_API_URL } : {}),\n});\n\n// Server implementation\nconst server = new Server(\n  {\n    name: 'one-search-mcp',\n    version: '0.0.1',\n  },\n  {\n    capabilities: {\n      tools: {},\n      logging: {},\n    },\n  },\n);\n\nconst searchDefaultConfig = {\n  limit: Number(LIMIT),\n  categories: CATEGORIES,\n  format: FORMAT,\n  safesearch: SAFE_SEARCH,\n  language: LANGUAGE,\n  engines: ENGINES,\n  time_range: TIME_RANGE,\n  timeout: DEFAULT_TIMEOUT,\n};\n\n// Tool handlers\nserver.setRequestHandler(ListToolsRequestSchema, async () => ({\n  tools: [\n    SEARCH_TOOL,\n    EXTRACT_TOOL,\n    SCRAPE_TOOL,\n    MAP_TOOL,\n  ],\n}));\n\nserver.setRequestHandler(CallToolRequestSchema, async (request) => {\n  const startTime = Date.now();\n\n  try {\n    const { name, arguments: args } = request.params;\n\n    if (!args) {\n      throw new Error('No arguments provided');\n    }\n  \n    server.sendLoggingMessage({\n      level: 'info',\n      data: `[${new Date().toISOString()}] Received request for tool: [${name}]`,\n    });\n  \n    switch (name) {\n      case 'one_search': {\n        // check args.\n        if (!checkSearchArgs(args)) {\n          throw new Error(`Invalid arguments for tool: [${name}]`);\n        }\n        try {\n          const { results, success } = await processSearch({\n            ...args,\n            apiKey: SEARCH_API_KEY ?? '',\n            apiUrl: SEARCH_API_URL,\n          });\n          if (!success) {\n            throw new Error('Failed to search');\n          }\n          const resultsText = results.map((result) => (\n            `Title: ${result.title}\nURL: ${result.url}\nDescription: ${result.snippet}\n${result.markdown ? `Content: ${result.markdown}` : ''}`\n          ));\n          return {\n            content: [\n              {\n                type: 'text',\n                text: resultsText.join('\\n\\n'),\n              },\n            ],\n            results,\n            success,\n          };\n        } catch (error) {\n          server.sendLoggingMessage({\n            level: 'error',\n            data: `[${new Date().toISOString()}] Error searching: ${error}`,\n          });\n          const msg = error instanceof Error ? error.message : 'Unknown error';\n          return {\n            success: false,\n            content: [\n              {\n                type: 'text',\n                text: msg,\n              },\n            ],\n          };\n        }\n      }\n      case 'one_scrape': {\n        if (!checkScrapeArgs(args)) {\n          throw new Error(`Invalid arguments for tool: [${name}]`);\n        }\n        try {\n          const startTime = Date.now();\n          server.sendLoggingMessage({\n            level: 'info',\n            data: `[${new Date().toISOString()}] Scraping started for url: [${args.url}]`,\n          });\n\n          const { url, ...scrapeArgs } = args;\n          const { content, success, result } = await processScrape(url, scrapeArgs);\n\n          server.sendLoggingMessage({\n            level: 'info',\n            data: `[${new Date().toISOString()}] Scraping completed in ${Date.now() - startTime}ms`,\n          });\n\n          return {\n            content,\n            result,\n            success,\n          };\n        } catch (error) {\n          server.sendLoggingMessage({\n            level: 'error',\n            data: `[${new Date().toISOString()}] Error scraping: ${error}`,\n          });\n          const msg = error instanceof Error ? error.message : 'Unknown error';\n          return {\n            success: false,\n            content: [\n              {\n                type: 'text',\n                text: msg,\n              },\n            ],\n          };\n        }\n      }\n      case 'one_map': {\n        if (!checkMapArgs(args)) {\n          throw new Error(`Invalid arguments for tool: [${name}]`);\n        }\n        try {\n          const { content, success, result } = await processMapUrl(args.url, args);\n          return {\n            content,\n            result,\n            success,\n          };\n        } catch (error) {\n          server.sendLoggingMessage({\n            level: 'error',\n            data: `[${new Date().toISOString()}] Error mapping: ${error}`,\n          });\n          const msg = error instanceof Error ? error.message : String(error);\n          return {\n            success: false,\n            content: [\n              {\n                type: 'text',\n                text: msg,\n              },\n            ],\n          };\n        }\n      }\n      default: {\n        throw new Error(`Unknown tool: ${name}`);\n      }\n    }\n  } catch(error) {\n    const msg = error instanceof Error ? error.message : String(error);\n    server.sendLoggingMessage({\n      level: 'error',\n      data: {\n        message: `[${new Date().toISOString()}] Error processing request: ${msg}`,\n        tool: request.params.name,\n        arguments: request.params.arguments,\n        timestamp: new Date().toISOString(),\n        duration: Date.now() - startTime,\n      },\n    });\n    return {\n      success: false,\n      content: [\n        {\n          type: 'text',\n          text: msg,\n        },\n      ],\n    };\n  } finally {\n    server.sendLoggingMessage({\n      level: 'info',\n      data: `[${new Date().toISOString()}] Request completed in ${Date.now() - startTime}ms`,\n    });\n  }\n});\n\nasync function processSearch(args: ISearchRequestOptions): Promise<ISearchResponse> {\n  switch (SEARCH_PROVIDER) {\n    case 'searxng': {\n      // merge default config with args\n      const params = {\n        ...searchDefaultConfig,\n        ...args,\n        apiKey: SEARCH_API_KEY,\n      };\n\n      // but categories and language have higher priority (ENV > args).\n      const { categories, language } = searchDefaultConfig;\n\n      if (categories) {\n        params.categories = categories;\n      }\n      if (language) {\n        params.language = language;\n      }\n      return await searxngSearch(params);\n    }\n    case 'tavily': {\n      return await tavilySearch({\n        ...searchDefaultConfig,\n        ...args,\n        apiKey: SEARCH_API_KEY,\n      });\n    }\n    case 'bing': {\n      return await bingSearch({\n        ...searchDefaultConfig,\n        ...args,\n        apiKey: SEARCH_API_KEY,\n      });\n    }\n    case 'duckduckgo': {\n      const safeSearch = args.safeSearch ?? 0;\n      const safeSearchOptions = [SafeSearchType.STRICT, SafeSearchType.MODERATE, SafeSearchType.OFF];\n      return await duckDuckGoSearch({\n        ...searchDefaultConfig,\n        ...args,\n        apiKey: SEARCH_API_KEY,\n        safeSearch: safeSearchOptions[safeSearch],\n      });\n    }\n    case 'local': {\n      return await localSearch({\n        ...searchDefaultConfig,\n        ...args,\n      });\n    }\n    default:\n      throw new Error(`Unsupported search provider: ${SEARCH_PROVIDER}`);\n  }\n}\n\nasync function processScrape(url: string, args: ScrapeParams) {\n  const res = await firecrawl.scrapeUrl(url, {\n    ...args,\n  });\n\n  if (!res.success) {\n    throw new Error(`Failed to scrape: ${res.error}`);\n  }\n\n  const content: string[] = [];\n\n  if (res.markdown) {\n    content.push(res.markdown);\n  }\n\n  if (res.rawHtml) {\n    content.push(res.rawHtml);\n  }\n\n  if (res.links) {\n    content.push(res.links.join('\\n'));\n  }\n\n  if (res.screenshot) {\n    content.push(res.screenshot);\n  }\n\n  if (res.html) {\n    content.push(res.html);\n  }\n\n  if (res.extract) {\n    content.push(res.extract);\n  }\n\n  return {\n    content: [\n      {\n        type: 'text',\n        text: content.join('\\n\\n') || 'No content found',\n      },\n    ],\n    result: res,\n    success: true,\n  };\n}\n\nasync function processMapUrl(url: string, args: MapParams) {\n  const res = await firecrawl.mapUrl(url, {\n    ...args,\n  });\n\n  if ('error' in res) {\n    throw new Error(`Failed to map: ${res.error}`);\n  }\n\n  if (!res.links) {\n    throw new Error(`No links found from: ${url}`);\n  }\n\n  return {\n    content: [\n      {\n        type: 'text',\n        text: res.links.join('\\n').trim(),\n      },\n    ],\n    result: res.links,\n    success: true,\n  };\n}\n\nfunction checkSearchArgs(args: unknown): args is ISearchRequestOptions {\n  return (\n    typeof args === 'object' &&\n    args !== null &&\n    'query' in args &&\n    typeof args.query === 'string'\n  );\n}\n\nfunction checkScrapeArgs(args: unknown): args is ScrapeParams & { url: string } {\n  return (\n    typeof args === 'object' &&\n    args !== null &&\n    'url' in args &&\n    typeof args.url === 'string'\n  );\n}\n\nfunction checkMapArgs(args: unknown): args is MapParams & { url: string } {\n  return (\n    typeof args === 'object' &&\n    args !== null &&\n    'url' in args &&\n    typeof args.url === 'string'\n  );\n}\n\nasync function runServer() {\n  try {\n    process.stdout.write('Starting OneSearch MCP server...\\n');\n\n    const transport = new StdioServerTransport();\n    await server.connect(transport);\n\n    server.sendLoggingMessage({\n      level: 'info',\n      data: 'OneSearch MCP server started',\n    });\n\n  } catch (error) {\n    const msg = error instanceof Error ? error.message : String(error);\n    process.stderr.write(`Error starting server: ${msg}\\n`);\n    process.exit(1);\n  }\n}\n\n// run server\nrunServer().catch((error) => {\n  const msg = error instanceof Error ? error.message : String(error);\n  process.stderr.write(`Error running server: ${msg}\\n`);\n  process.exit(1);\n});\n\n// export types\nexport * from './interface.js';\n","/**\n * Bing Search API\n */\nimport { ISearchRequestOptions, ISearchResponse } from '../interface.js';\n\n\n/**\n * Options for performing a Bing search\n */\nexport interface BingSearchOptions {\n  /**\n   * Search query string\n   */\n  q: string;\n\n  /**\n   * Number of results to return\n   */\n  count?: number;\n\n  /**\n   * Result offset for pagination\n   */\n  offset?: number;\n\n  /**\n   * Market code (e.g., 'en-US')\n   */\n  mkt?: string;\n\n  /**\n   * Safe search filtering level\n   */\n  safeSearch?: 'Off' | 'Moderate' | 'Strict';\n\n  /**\n   * Bing API key\n   */\n  apiKey: string;\n\n  /**\n   * Bing Search API URL\n   */\n  apiUrl?: string;\n\n  /**\n   * Additional parameters supported by Bing Search API\n   */\n  [key: string]: any;\n}\n\n/**\n * Represents a web page result from Bing Search\n */\nexport interface BingSearchWebPage {\n  /**\n   * Title of the web page\n   */\n  name: string;\n\n  /**\n   * URL of the web page\n   */\n  url: string;\n\n  /**\n   * Text snippet from the web page\n   */\n  snippet: string;\n\n  /**\n   * Date the page was last crawled by Bing\n   */\n  dateLastCrawled?: string;\n\n  /**\n   * Display URL for the web page\n   */\n  displayUrl?: string;\n\n  /**\n   * Unique identifier for the result\n   */\n  id?: string;\n\n  /**\n   * Indicates if the content is family friendly\n   */\n  isFamilyFriendly?: boolean;\n\n  /**\n   * Indicates if the result is navigational\n   */\n  isNavigational?: boolean;\n\n  /**\n   * Language of the web page\n   */\n  language?: string;\n\n  /**\n   * Indicates if caching should be disabled\n   */\n  noCache?: boolean;\n\n  /**\n   * Name of the website\n   */\n  siteName?: string;\n\n  /**\n   * URL to a thumbnail image\n   */\n  thumbnailUrl?: string;\n}\n\n/**\n * Represents an image result from Bing Search\n */\nexport interface BingSearchImage {\n  contentSize: string;\n  contentUrl: string;\n  datePublished: string;\n  encodingFormat: string;\n  height: number;\n  width: number;\n  hostPageDisplayUrl: string;\n  hostPageUrl: string;\n  name: string;\n  thumbnail: {\n    height: number;\n    width: number;\n  };\n  thumbnailUrl: string;\n  webSearchUrl: string;\n}\n\n/**\n * Represents a video result from Bing Search\n */\nexport interface BingSearchVideo {\n  allowHttpsEmbed: boolean;\n  allowMobileEmbed: boolean;\n  contentUrl: string;\n  creator?: {\n    name: string;\n  };\n  datePublished: string;\n  description: string;\n  duration: string;\n  embedHtml: string;\n  encodingFormat: string;\n  height: number;\n  width: number;\n  hostPageDisplayUrl: string;\n  hostPageUrl: string;\n  name: string;\n  publisher?: {\n    name: string;\n  }[];\n  thumbnail: {\n    height: number;\n    width: number;\n  };\n  thumbnailUrl: string;\n  viewCount?: number;\n  webSearchUrl: string;\n}\n\nexport interface BingSearchResponse {\n  _type?: string;\n  queryContext?: {\n    originalQuery: string;\n  };\n  webPages?: {\n    value: BingSearchWebPage[];\n    totalEstimatedMatches?: number;\n    someResultsRemoved?: boolean;\n    webSearchUrl?: string;\n  };\n  images?: {\n    value: BingSearchImage[];\n    isFamilyFriendly?: boolean;\n    readLink?: string;\n    webSearchUrl?: string;\n    id?: string;\n  };\n  videos?: {\n    value: BingSearchVideo[];\n    isFamilyFriendly?: boolean;\n    readLink?: string;\n    webSearchUrl?: string;\n    id?: string;\n    scenario?: string;\n  };\n  rankingResponse?: {\n    mainline?: {\n      items: {\n        answerType: string;\n        resultIndex?: number;\n        value: {\n          id: string;\n        };\n      }[];\n    };\n  };\n  [key: string]: any; // Allow other response fields\n}\n\nexport async function bingSearch(options: ISearchRequestOptions): Promise<ISearchResponse> {\n  const { query, limit = 10, safeSearch = 0, page = 1, apiUrl = 'https://api.bing.microsoft.com/v7.0/search', apiKey, language } = options;\n\n  const bingSafeSearchOptions = ['Off', 'Moderate', 'Strict'];\n\n  if (!apiKey) {\n    throw new Error('Bing API key is required');\n  }\n\n  const searchOptions = {\n    q: query,\n    count: limit,\n    offset: (page - 1) * limit,\n    mkt: language,\n    safeSearch: bingSafeSearchOptions[safeSearch] as 'Off' | 'Moderate' | 'Strict',\n  };\n\n  try {\n    const queryParams = new URLSearchParams();\n    Object.entries(searchOptions).forEach(([key, value]) => {\n      if (value !== undefined) {\n        queryParams.set(key, value.toString());\n      }\n    });\n\n    const res = await fetch(`${apiUrl}?${queryParams}`, {\n      method: 'GET',\n      headers: {\n        'Content-Type': 'application/json',\n        'Ocp-Apim-Subscription-Key': apiKey,\n      },\n    });\n\n    if (!res.ok) {\n      throw new Error(`Bing search error: ${res.status} ${res.statusText}`);\n    }\n\n    const data = await res.json();\n    const serp = data.webPages?.value as Array<BingSearchWebPage>;\n    const results = serp?.map((item: BingSearchWebPage) => ({\n      title: item.name,\n      snippet: item.snippet,\n      url: item.url,\n      source: item.siteName,\n      thumbnailUrl: item.thumbnailUrl,\n      language: item.language,\n      image: null,\n      video: null,\n      engine: 'bing',\n    })) ?? [];\n\n    return {\n      results,\n      success: true,\n    };\n  } catch (err: unknown) {\n    const msg = err instanceof Error ? err.message : 'Bing search error.';\n    process.stdout.write(msg);\n    throw err;\n  }\n}","import * as DDG from 'duck-duck-scrape';\nimport asyncRetry from 'async-retry';\nimport type { SearchOptions } from 'duck-duck-scrape';\nimport { ISearchRequestOptions, ISearchResponse } from '../interface.js';\n\n\nexport async function duckDuckGoSearch(options: Omit<ISearchRequestOptions, 'safeSearch'> & SearchOptions): Promise<ISearchResponse> {\n  try {\n    const { query, timeout = 10000, safeSearch = DDG.SafeSearchType.OFF, retry = { retries: 3 }, ...searchOptions } = options;\n  \n    const res = await asyncRetry(\n      () => {\n        return DDG.search(query, {\n          ...searchOptions,\n          safeSearch,\n        }, {\n          // needle options\n          response_timeout: timeout,\n        });\n      },\n      retry,\n    );\n\n    const results = res ? {\n      noResults: res.noResults,\n      vqd: res.vqd,\n      results: res.results,\n    } : {\n      noResults: true,\n      vqd: '',\n      results: [],\n    };\n\n    return {\n      results: results.results.map((result) => ({\n        title: result.title,\n        snippet: result.description,\n        url: result.url,\n        source: result.hostname,\n        image: null,\n        video: null,\n        engine: 'duckduckgo',\n      })),\n      success: true,\n    };\n  } catch (error) {\n    const msg = error instanceof Error ? error.message : 'DuckDuckGo search error.';\n    process.stdout.write(msg);\n    throw error;\n  }\n}\n","import url from 'node:url';\nimport { ISearchRequestOptions, ISearchResponse, ISearchResponseResult } from '../interface.js';\n\n/**\n * SearxNG Search API\n * - https://docs.searxng.org/dev/search_api.html\n */\nexport async function searxngSearch(params: ISearchRequestOptions): Promise<ISearchResponse> {\n  try {\n    const {\n      query,\n      page = 1,\n      limit = 10,\n      categories = 'general',\n      engines = 'all',\n      safeSearch = 0,\n      format = 'json',\n      language = 'auto',\n      timeRange = '',\n      timeout = 10000,\n      apiKey,\n      apiUrl,\n    } = params;\n\n    if (!apiUrl) {\n      throw new Error('SearxNG API URL is required');\n    }\n\n    const controller = new AbortController();\n    const timeoutId = setTimeout(() => controller.abort(), Number(timeout));\n\n    const config = {\n      q: query,\n      pageno: page,\n      categories,\n      format,\n      safesearch: safeSearch,\n      language,\n      engines,\n      time_range: timeRange,\n    };\n\n    const endpoint = `${apiUrl}/search`;\n\n    const queryParams = url.format({ query: config });\n\n    const headers: HeadersInit = {\n      'Content-Type': 'application/json',\n    };\n\n    if (apiKey) {\n      headers['Authorization'] = `Bearer ${apiKey}`;\n    }\n\n    const res = await fetch(`${endpoint}${queryParams}`, {\n      method: 'POST',\n      headers,\n      signal: controller.signal,\n    });\n\n    clearTimeout(timeoutId);\n    const response = await res.json();\n    if (response.results) {\n      const list = (response.results as Array<Record<string, any>>).slice(0, limit);\n      const results: ISearchResponseResult[] = list.map((item: Record<string, any>) => {\n        const image = item.img_src ? {\n          thumbnail: item.thumbnail_src,\n          src: item.img_src,\n        } : null;\n        const video = item.iframe_src ? {\n          thumbnail: item.thumbnail_src,\n          src: item.iframe_src,\n        } : null;\n        return {\n          title: item.title,\n          snippet: item.content,\n          url: item.url,\n          source: item.source,\n          image,\n          video,\n          engine: item.engine,\n        };\n      });\n      return {\n        results,\n        success: true,\n      };\n    }\n    return {\n      results: [],\n      success: false,\n    };\n  } catch (err: unknown) {\n    const msg = err instanceof Error ? err.message : 'Searxng search error.';\n    process.stdout.write(msg);\n    throw err;\n  }\n}\n","import { tavily, TavilySearchOptions } from '@tavily/core';\nimport { ISearchRequestOptions, ISearchResponse } from '../interface.js';\n\n/**\n * Tavily Search API\n * - https://docs.tavily.com/documentation/quickstart\n */\nexport async function tavilySearch(options: ISearchRequestOptions): Promise<ISearchResponse> {\n  const {\n    query,\n    limit = 10,\n    categories = 'general',\n    timeRange,\n    apiKey,\n  } = options;\n\n  if (!apiKey) {\n    throw new Error('Tavily API key is required');\n  }\n\n  try {\n    const tvly = tavily({\n      apiKey,\n    });\n  \n    const params: TavilySearchOptions = {\n      topic: categories as TavilySearchOptions['topic'],\n      timeRange: timeRange as TavilySearchOptions['timeRange'],\n      maxResults: limit,\n    };\n  \n    const res = await tvly.search(query, params);\n    const results = res.results.map(item => ({\n      title: item.title,\n      url: item.url,\n      snippet: item.content,\n      engine: 'tavily',\n    }));\n  \n    return {\n      results,\n      success: true,\n    };\n  } catch (error) {\n    const msg = error instanceof Error ? error.message : 'Tavily search error.';\n    process.stdout.write(msg);\n    throw error;\n  }\n}","/**\n * The following code is based on\n * https://github.com/bytedance/UI-TARS-desktop/tree/main/packages/agent-infra/browser\n * \n * Copyright (c) 2025 Bytedance, Inc. and its affiliates.\n * SPDX-License-Identifier: Apache-2.0\n */\nimport { Page, WaitForOptions } from 'puppeteer-core';\n\n/**\n * Options for launching a browser instance\n * @interface LaunchOptions\n */\nexport interface LaunchOptions {\n  /**\n   * Whether to run browser in headless mode\n   * @default false\n   */\n  headless?: boolean;\n\n  /**\n   * Maximum time in milliseconds to wait for the browser to start\n   * @default 0 (no timeout)\n   */\n  timeout?: number;\n\n  /**\n   * The viewport dimensions\n   * @property {number} width - Viewport width in pixels\n   * @property {number} height - Viewport height in pixels\n   */\n  defaultViewport?: {\n    width: number;\n    height: number;\n  };\n\n  /**\n   * Path to a browser executable to use instead of the automatically detected one\n   * If not provided, the system will attempt to find an installed browser\n   */\n  executablePath?: string;\n\n  /**\n   * Path to a specific browser profile to use\n   * Allows using existing browser profiles with cookies, extensions, etc.\n   */\n  profilePath?: string;\n\n  /**\n   * Proxy server URL, e.g. 'http://proxy.example.com:8080'\n   * Used to route browser traffic through a proxy server\n   */\n  proxy?: string;\n}\n\n/**\n * Options for evaluating JavaScript in a new page\n * @template T - Array of parameters to pass to the page function\n * @template R - Return type of the page function\n * @interface EvaluateOnNewPageOptions\n */\nexport interface EvaluateOnNewPageOptions<T extends any[], R> {\n  /**\n   * URL to navigate to before evaluating the function\n   * The page will load this URL before executing the pageFunction\n   */\n  url: string;\n\n  /**\n   * Options for waiting for the page to load\n   */\n  waitForOptions?: WaitForOptions;\n\n  /**\n   * Function to be evaluated in the page context\n   * This function runs in the context of the browser page, not Node.js\n   * @param {Window} window - The window object of the page\n   * @param {...T} args - Additional arguments passed to the function\n   * @returns {R} Result of the function execution\n   */\n  pageFunction: (window: Window, ...args: T) => R;\n\n  /**\n   * Parameters to pass to the page function\n   * These values will be serialized and passed to the pageFunction\n   */\n  pageFunctionParams: T;\n\n  /**\n   * Optional function to execute before page navigation\n   * Useful for setting up page configuration before loading the URL\n   * @param {Page} page - Puppeteer page instance\n   * @returns {void | Promise<void>}\n   */\n  beforePageLoad?: (page: Page) => void | Promise<void>;\n\n  /**\n   * Optional function to execute after page navigation\n   * Useful for setting up page configuration after loading the URL\n   * @param {Page} page - Puppeteer page instance\n   * @returns {void | Promise<void>}\n   */\n  afterPageLoad?: (page: Page) => void | Promise<void>;\n\n  /**\n   * Optional function to process the result before returning\n   * Can be used to transform or validate the result from page evaluation\n   * @param {Page} page - Puppeteer page instance\n   * @param {R} result - Result from page function evaluation\n   * @returns {R | Promise<R>} Processed result\n   */\n  beforeSendResult?: (page: Page, result: R) => R | Promise<R>;\n}\n\n/**\n * Core browser interface that all browser implementations must implement\n * Defines the standard API for browser automation\n * @interface BrowserInterface\n */\nexport interface BrowserInterface {\n  /**\n   * Launch a new browser instance\n   * @param {LaunchOptions} [options] - Launch configuration options\n   * @returns {Promise<void>} Promise resolving when browser is launched\n   */\n  launch(options?: LaunchOptions): Promise<void>;\n\n  /**\n   * Close the browser instance and all its pages\n   * @returns {Promise<void>} Promise resolving when browser is closed\n   */\n  close(): Promise<void>;\n\n  /**\n   * Create a new page in the browser\n   * @returns {Promise<Page>} Promise resolving to the new page instance\n   */\n  createPage(): Promise<Page>;\n\n  /**\n   * Evaluate a function in a new page context\n   * Creates a new page, navigates to URL, executes function, and returns result\n   * @template T - Array of parameters to pass to the page function\n   * @template R - Return type of the page function\n   * @param {EvaluateOnNewPageOptions<T, R>} options - Evaluation options\n   * @returns {Promise<R | null>} Promise resolving to the function result or null\n   */\n  evaluateOnNewPage<T extends any[], R>(\n    options: EvaluateOnNewPageOptions<T, R>,\n  ): Promise<R | null>;\n\n  /**\n   * Get the currently active page or create one if none exists\n   * @returns {Promise<Page>} Promise resolving to the active page instance\n   */\n  getActivePage(): Promise<Page>;\n}\n\nexport { Page };","/**\n * The following code is modified based on\n * https://github.com/egoist/local-web-search/blob/main/src/find-browser.ts\n * Copy from\n * https://github.com/bytedance/UI-TARS-desktop/blob/main/packages/agent-infra/browser/src/browser-finder.ts\n * \n * MIT Licensed\n * Copyright (c) 2025 ChatWise (https://chatwise.app) <kevin@chatwise.app>\n * https://github.com/egoist/local-web-search/blob/main/LICENSE\n */\n\nimport * as fs from 'fs';\nimport * as path from 'path';\nimport * as os from 'os';\nimport { Logger, defaultLogger } from '@agent-infra/logger';\n\n/**\n * Interface defining browser locations and configurations\n * Contains paths and settings for different operating systems\n * @interface Browser\n */\ninterface Browser {\n  /**\n   * Browser name identifier\n   */\n  name: string;\n\n  /**\n   * Executable paths by platform\n   * @property {string} win32 - Windows executable path\n   * @property {string} darwin - macOS executable path\n   * @property {string} linux - Linux executable path\n   */\n  executable: {\n    win32: string;\n    darwin: string;\n    linux: string;\n  };\n\n  /**\n   * User data directory paths by platform\n   * @property {string} win32 - Windows user data directory\n   * @property {string} darwin - macOS user data directory\n   * @property {string} linux - Linux user data directory\n   */\n  userDataDir: {\n    win32: string;\n    darwin: string;\n    linux: string;\n  };\n}\n\n/**\n * Class responsible for finding and managing browser installations\n * Detects installed browsers and their profiles across different platforms\n */\nexport class BrowserFinder {\n  /**\n   * Logger instance for diagnostic output\n   */\n  private logger: Logger;\n\n  /**\n   * Creates a new BrowserFinder instance\n   * @param {Logger} [logger] - Optional custom logger\n   */\n  constructor(logger?: Logger) {\n    this.logger = logger ?? defaultLogger;\n  }\n\n  /**\n   * Getter that returns the list of supported browsers with their platform-specific paths\n   * @returns {Browser[]} Array of browser configurations\n   * @private\n   */\n  private get browsers(): Browser[] {\n    // Get HOME_DIR inside the getter to ensure it's always current\n    const HOME_DIR = os.homedir();\n    const LOCAL_APP_DATA = process.env.LOCALAPPDATA;\n\n    return [\n      {\n        name: 'Chromium',\n        executable: {\n          win32: 'C:\\\\Program Files\\\\Chromium\\\\Application\\\\chrome.exe',\n          darwin: '/Applications/Chromium.app/Contents/MacOS/Chromium',\n          linux: '/usr/bin/chromium',\n        },\n        userDataDir: {\n          win32: `${LOCAL_APP_DATA}\\\\Chromium\\\\User Data`,\n          darwin: `${HOME_DIR}/Library/Application Support/Chromium`,\n          linux: `${HOME_DIR}/.config/chromium`,\n        },\n      },\n      {\n        name: 'Google Chrome',\n        executable: {\n          win32: 'C:\\\\Program Files\\\\Google\\\\Chrome\\\\Application\\\\chrome.exe',\n          darwin:\n            '/Applications/Google Chrome.app/Contents/MacOS/Google Chrome',\n          linux: '/usr/bin/google-chrome',\n        },\n        userDataDir: {\n          win32: `${LOCAL_APP_DATA}\\\\Google\\\\Chrome\\\\User Data`,\n          darwin: `${HOME_DIR}/Library/Application Support/Google/Chrome`,\n          linux: `${HOME_DIR}/.config/google-chrome`,\n        },\n      },\n      {\n        name: 'Google Chrome Canary',\n        executable: {\n          win32:\n            'C:\\\\Program Files\\\\Google\\\\Chrome Canary\\\\Application\\\\chrome.exe',\n          darwin:\n            '/Applications/Google Chrome Canary.app/Contents/MacOS/Google Chrome Canary',\n          linux: '/usr/bin/google-chrome-canary',\n        },\n        userDataDir: {\n          win32: `${LOCAL_APP_DATA}\\\\Google\\\\Chrome Canary\\\\User Data`,\n          darwin: `${HOME_DIR}/Library/Application Support/Google/Chrome Canary`,\n          linux: `${HOME_DIR}/.config/google-chrome-canary`,\n        },\n      },\n    ];\n  }\n\n  /**\n   * Find a specific browser or the first available browser\n   * @param {string} [name] - Optional browser name to find\n   * @returns {{ executable: string; userDataDir: string }} Browser executable and user data paths\n   * @throws {Error} If no supported browser is found or the platform is unsupported\n   */\n  findBrowser(name?: string): {\n    executable: string;\n    userDataDir: string;\n  } {\n    const platform = process.platform;\n    this.logger.info('Finding browser on platform:', platform);\n\n    if (platform !== 'darwin' && platform !== 'win32' && platform !== 'linux') {\n      const error = new Error(`Unsupported platform: ${platform}`);\n      this.logger.error(error.message);\n      throw error;\n    }\n\n    const browser = name\n      ? this.browsers.find(\n        (b) => b.name === name && fs.existsSync(b.executable[platform]),\n      )\n      : this.browsers.find((b) => fs.existsSync(b.executable[platform]));\n\n    this.logger.log('browser', browser);\n\n    if (!browser) {\n      const error = name\n        ? new Error(`Cannot find browser: ${name}`)\n        : new Error(\n          'Cannot find a supported browser on your system. Please install Chrome, Edge, or Brave.',\n        );\n      this.logger.error(error.message);\n      throw error;\n    }\n\n    const result = {\n      executable: browser.executable[platform],\n      userDataDir: browser.userDataDir[platform],\n    };\n\n    this.logger.success(`Found browser: ${browser.name}`);\n    this.logger.info('Browser details:', result);\n\n    return result;\n  }\n\n  /**\n   * Get browser profiles for a specific browser\n   * Reads the Local State file to extract profile information\n   * @param {string} [browserName] - Optional browser name to get profiles for\n   * @returns {Array<{ displayName: string; path: string }>} Array of profile objects with display names and paths\n   */\n  getBrowserProfiles(\n    browserName?: string,\n  ): Array<{ displayName: string; path: string }> {\n    const browser = this.findBrowser(browserName);\n\n    try {\n      const localState = JSON.parse(\n        fs.readFileSync(path.join(browser.userDataDir, 'Local State'), 'utf8'),\n      );\n      const profileInfo = localState.profile.info_cache;\n\n      return Object.entries(profileInfo).map(\n        ([profileName, info]: [string, any]) => ({\n          displayName: info.name,\n          path: path.join(browser.userDataDir, profileName),\n        }),\n      );\n    } catch (error) {\n      return [];\n    }\n  }\n\n  /**\n   * Legacy method for backwards compatibility\n   * Finds Chrome browser executable path\n   * @deprecated Use findBrowser instead\n   * @returns {string | null} Chrome executable path or null if not found\n   */\n  findChrome(): string | null {\n    try {\n      const { executable } = this.findBrowser('Google Chrome');\n      return executable;\n    } catch {\n      return null;\n    }\n  }\n}\n","/**\n * The following code is based on\n * https://github.com/bytedance/UI-TARS-desktop/tree/main/packages/agent-infra/browser\n * \n * Copyright (c) 2025 Bytedance, Inc. and its affiliates.\n * SPDX-License-Identifier: Apache-2.0\n */\nimport * as puppeteer from 'puppeteer-core';\nimport { Logger, defaultLogger } from '@agent-infra/logger';\nimport {\n  BrowserInterface,\n  EvaluateOnNewPageOptions,\n  LaunchOptions,\n  Page,\n} from './types.js';\n\n/**\n * Configuration options for the BaseBrowser class\n * @interface BaseBrowserOptions\n * @property {Logger} [logger] - Custom logger instance to use for browser logging\n */\nexport interface BaseBrowserOptions {\n  logger?: Logger;\n}\n\n/**\n * Abstract base class that implements common browser automation functionality\n * Provides a foundation for specific browser implementations with shared capabilities\n * @abstract\n * @implements {BrowserInterface}\n */\nexport abstract class BaseBrowser implements BrowserInterface {\n  /**\n   * The underlying Puppeteer browser instance\n   * @protected\n   */\n  protected browser: puppeteer.Browser | null = null;\n\n  /**\n   * Logger instance for browser-related logging\n   * @protected\n   */\n  protected logger: Logger;\n\n  /**\n   * Reference to the currently active browser page\n   * @protected\n   */\n  protected activePage: Page | null = null;\n\n  /**\n   * Creates an instance of BaseBrowser\n   * @param {BaseBrowserOptions} [options] - Configuration options\n   */\n  constructor(options?: BaseBrowserOptions) {\n    this.logger = options?.logger ?? defaultLogger;\n    this.logger.info('Browser Options:', options);\n  }\n\n  /**\n   * Get the underlying Puppeteer browser instance\n   * @throws Error if browser is not launched\n\n   * @returns {puppeteer.Browser} Puppeteer browser instance\n   */\n  getBrowser(): puppeteer.Browser {\n    if (!this.browser) {\n      throw new Error('Browser not launched');\n    }\n    return this.browser;\n  }\n\n  /**\n   * Sets up listeners for browser page events\n   * Tracks page creation and updates active page reference\n   * @protected\n   */\n  protected async setupPageListener() {\n    if (!this.browser) return;\n\n    this.browser.on('targetcreated', async (target) => {\n      const page = await target.page();\n      if (page) {\n        this.logger.info('New page created:', await page.url());\n        this.activePage = page;\n\n        page.once('close', () => {\n          if (this.activePage === page) {\n            this.activePage = null;\n          }\n        });\n\n        page.once('error', () => {\n          if (this.activePage === page) {\n            this.activePage = null;\n          }\n        });\n      }\n    });\n  }\n\n  /**\n   * Launches the browser with specified options\n   * @abstract\n   * @param {LaunchOptions} [options] - Browser launch configuration options\n   * @returns {Promise<void>} Promise that resolves when browser is launched\n   */\n  abstract launch(options?: LaunchOptions): Promise<void>;\n\n  /**\n   * Closes the browser instance and cleans up resources\n   * @returns {Promise<void>} Promise that resolves when browser is closed\n   * @throws {Error} If browser fails to close properly\n   */\n  async close(): Promise<void> {\n    this.logger.info('Closing browser');\n    try {\n      await this.browser?.close();\n      this.browser = null;\n      this.logger.success('Browser closed successfully');\n    } catch (error) {\n      this.logger.error('Failed to close browser:', error);\n      throw error;\n    }\n  }\n\n  /**\n   * Creates a new page, navigates to the specified URL, executes a function in the page context, and returns the result\n   * This method is inspired and modified from https://github.com/egoist/local-web-search/blob/04608ed09aa103e2fff6402c72ca12edfb692d19/src/browser.ts#L74\n   * @template T - Type of parameters passed to the page function\n   * @template R - Return type of the page function\n   * @param {EvaluateOnNewPageOptions<T, R>} options - Configuration options for the page evaluation\n   * @returns {Promise<R | null>} Promise resolving to the result of the page function or null\n   * @throws {Error} If page creation or evaluation fails\n   */\n  async evaluateOnNewPage<T extends any[], R>(\n    options: EvaluateOnNewPageOptions<T, R>,\n  ): Promise<R | null> {\n    const {\n      url,\n      pageFunction,\n      pageFunctionParams,\n      beforePageLoad,\n      afterPageLoad,\n      beforeSendResult,\n      waitForOptions,\n    } = options;\n    const page = await this.browser!.newPage();\n    try {\n      await beforePageLoad?.(page);\n      await page.goto(url, {\n        waitUntil: 'networkidle2',\n        ...waitForOptions,\n      });\n      await afterPageLoad?.(page);\n      const _window = await page.evaluateHandle(() => window);\n      const result = await page.evaluate(\n        pageFunction,\n        _window,\n        ...pageFunctionParams,\n      );\n      await beforeSendResult?.(page, result);\n      await _window.dispose();\n      await page.close();\n      return result;\n    } catch (error) {\n      await page.close();\n      throw error;\n    }\n  }\n\n  /**\n   * Creates a new browser page\n   * @returns {Promise<Page>} Promise resolving to the newly created page\n   * @throws {Error} If browser is not launched or page creation fails\n   */\n  async createPage(): Promise<Page> {\n    if (!this.browser) {\n      this.logger.error('No active browser');\n      throw new Error('Browser not launched');\n    }\n    const page = await this.browser.newPage();\n    return page;\n  }\n\n  /**\n   * Gets the currently active page or finds an active page if none is currently tracked\n   * If no active pages exist, creates a new page\n   * @returns {Promise<Page>} Promise resolving to the active page\n   * @throws {Error} If browser is not launched or no active page can be found/created\n   */\n  async getActivePage(): Promise<Page> {\n    if (!this.browser) {\n      throw new Error('Browser not launched');\n    }\n\n    // If activePage exists and is still available, return directly\n    if (this.activePage) {\n      try {\n        // Verify that the page is still available\n        await this.activePage.evaluate(() => document.readyState);\n        return this.activePage;\n      } catch (e) {\n        this.logger.warn('Active page no longer available:', e);\n        this.activePage = null;\n      }\n    }\n\n    // Get all pages and find the last active page\n    const pages = await this.browser.pages();\n\n    if (pages.length === 0) {\n      this.activePage = await this.createPage();\n      return this.activePage;\n    }\n\n    // Find the last responding page\n    for (let i = pages.length - 1; i >= 0; i--) {\n      const page = pages[i];\n      try {\n        await page.evaluate(() => document.readyState);\n        this.activePage = page;\n        return page;\n      } catch (e) {\n        continue;\n      }\n    }\n\n    throw new Error('No active page found');\n  }\n}","/*\n * Copyright (c) 2025 Bytedance, Inc. and its affiliates.\n * SPDX-License-Identifier: Apache-2.0\n */\nimport * as puppeteer from 'puppeteer-core';\nimport { LaunchOptions } from './types.js';\nimport { BrowserFinder } from './finder.js';\nimport { BaseBrowser } from './base.js';\n\n/**\n * LocalBrowser class for controlling locally installed browsers\n * Extends the BaseBrowser with functionality specific to managing local browser instances\n * @extends BaseBrowser\n */\nexport class LocalBrowser extends BaseBrowser {\n  /**\n   * Browser finder instance to detect and locate installed browsers\n   * @private\n   */\n  private browserFinder = new BrowserFinder();\n\n  /**\n   * Launches a local browser instance with specified options\n   * Automatically detects installed browsers if no executable path is provided\n   * @param {LaunchOptions} options - Configuration options for launching the browser\n   * @returns {Promise<void>} Promise that resolves when the browser is successfully launched\n   * @throws {Error} If the browser cannot be launched\n   */\n  async launch(options: LaunchOptions = {}): Promise<void> {\n    this.logger.info('Launching browser with options:', options);\n\n    const executablePath =\n      options?.executablePath || this.browserFinder.findBrowser().executable;\n\n    this.logger.info('Using executable path:', executablePath);\n\n    const viewportWidth = options?.defaultViewport?.width ?? 1280;\n    const viewportHeight = options?.defaultViewport?.height ?? 800;\n\n    const puppeteerLaunchOptions: puppeteer.LaunchOptions = {\n      executablePath,\n      headless: options?.headless ?? false,\n      defaultViewport: {\n        width: viewportWidth,\n        height: viewportHeight,\n      },\n      args: [\n        '--no-sandbox',\n        '--mute-audio',\n        '--disable-gpu',\n        '--disable-http2',\n        '--disable-blink-features=AutomationControlled',\n        '--disable-infobars',\n        '--disable-background-timer-throttling',\n        '--disable-popup-blocking',\n        '--disable-backgrounding-occluded-windows',\n        '--disable-renderer-backgrounding',\n        '--disable-window-activation',\n        '--disable-focus-on-load',\n        '--no-default-browser-check', // disable default browser check\n        '--disable-web-security', // disable CORS\n        '--disable-features=IsolateOrigins,site-per-process',\n        '--disable-site-isolation-trials',\n        `--window-size=${viewportWidth},${viewportHeight + 90}`,\n        options?.proxy ? `--proxy-server=${options.proxy}` : '',\n        options?.profilePath\n          ? `--profile-directory=${options.profilePath}`\n          : '',\n      ].filter(Boolean),\n      ignoreDefaultArgs: ['--enable-automation'],\n      timeout: options.timeout ?? 0,\n      downloadBehavior: {\n        policy: 'deny',\n      },\n    };\n\n    this.logger.info('Launch options:', puppeteerLaunchOptions);\n\n    try {\n      this.browser = await puppeteer.launch(puppeteerLaunchOptions);\n      await this.setupPageListener();\n      this.logger.success('Browser launched successfully');\n    } catch (error) {\n      this.logger.error('Failed to launch browser:', error);\n      throw error;\n    }\n  }\n}","/*\n * Copyright (c) 2025 Bytedance, Inc. and its affiliates.\n * SPDX-License-Identifier: Apache-2.0\n */\nimport * as puppeteer from 'puppeteer-core';\nimport { BaseBrowser, BaseBrowserOptions } from './base.js';\nimport { LaunchOptions } from './types.js';\n\n/**\n * Configuration options for RemoteBrowser\n * @extends BaseBrowserOptions\n * @interface RemoteBrowserOptions\n * @property {string} [wsEndpoint] - WebSocket endpoint URL for direct connection\n * @property {string} [host] - Remote host address (default: 'localhost')\n * @property {number} [port] - Remote debugging port (default: 9222)\n */\nexport interface RemoteBrowserOptions extends BaseBrowserOptions {\n  wsEndpoint?: string;\n  host?: string;\n  port?: number;\n}\n\n/**\n * RemoteBrowser class for connecting to remote browser instances\n *\n * Currently, this RemoteBrowser is not production ready,\n * mainly because it still relies on `puppeteer-core`,\n * which can only run on Node.js.\n *\n * At the same time, Chrome instances built with\n * `--remote-debugging-address` on Linux have security risks\n *\n * @see https://issues.chromium.org/issues/41487252\n * @see https://issues.chromium.org/issues/40261787\n * @see https://github.com/pyppeteer/pyppeteer/pull/379\n * @see https://stackoverflow.com/questions/72760355/chrome-remote-debugging-not-working-computer-to-computer\n *\n * @extends BaseBrowser\n */\nexport class RemoteBrowser extends BaseBrowser {\n  /**\n   * Creates a new RemoteBrowser instance\n   * @param {RemoteBrowserOptions} [options] - Configuration options for remote browser connection\n   */\n  constructor(private options?: RemoteBrowserOptions) {\n    super(options);\n  }\n\n  /**\n   * Connects to a remote browser instance using WebSocket\n   * If no WebSocket endpoint is provided, attempts to discover it using the DevTools Protocol\n   * @param {LaunchOptions} [options] - Launch configuration options\n   * @returns {Promise<void>} Promise that resolves when connected to the remote browser\n   * @throws {Error} If connection to the remote browser fails\n   */\n  async launch(options?: LaunchOptions): Promise<void> {\n    this.logger.info('Browser Launch options:', options);\n\n    let browserWSEndpoint = this.options?.wsEndpoint;\n\n    if (!browserWSEndpoint) {\n      const host = this.options?.host || 'localhost';\n      const port = this.options?.port || 9222;\n      const response = await fetch(`http://${host}:${port}/json/version`);\n      const { webSocketDebuggerUrl } = await response.json();\n      browserWSEndpoint = webSocketDebuggerUrl;\n    }\n\n    this.logger.info('Using WebSocket endpoint:', browserWSEndpoint);\n\n    const puppeteerConnectOptions: puppeteer.ConnectOptions = {\n      browserWSEndpoint,\n      defaultViewport: options?.defaultViewport ?? { width: 1280, height: 800 },\n    };\n\n    try {\n      this.browser = await puppeteer.connect(puppeteerConnectOptions);\n      await this.setupPageListener();\n      this.logger.success('Connected to remote browser successfully');\n    } catch (error) {\n      this.logger.error('Failed to connect to remote browser:', error);\n      throw error;\n    }\n  }\n}","/**\n * PLEASE DO NOT MODIFY IT as it is generated by the build script\n *\n * Build: scripts/build-readability.ts\n * Source: https://github.com/mozilla/readability/blob/main/Readability.js\n */\n\n/**\n * Copyright (c) 2010 Arc90 Inc\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *    http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n\nexport const READABILITY_SCRIPT =\n  'function q(t,e){if(e&&e.documentElement)t=e,e=arguments[2];else if(!t||!t.documentElement)throw new Error(\"First argument to Readability constructor should be a document object.\");if(e=e||{},this._doc=t,this._docJSDOMParser=this._doc.firstChild.__JSDOMParser__,this._articleTitle=null,this._articleByline=null,this._articleDir=null,this._articleSiteName=null,this._attempts=[],this._debug=!!e.debug,this._maxElemsToParse=e.maxElemsToParse||this.DEFAULT_MAX_ELEMS_TO_PARSE,this._nbTopCandidates=e.nbTopCandidates||this.DEFAULT_N_TOP_CANDIDATES,this._charThreshold=e.charThreshold||this.DEFAULT_CHAR_THRESHOLD,this._classesToPreserve=this.CLASSES_TO_PRESERVE.concat(e.classesToPreserve||[]),this._keepClasses=!!e.keepClasses,this._serializer=e.serializer||function(i){return i.innerHTML},this._disableJSONLD=!!e.disableJSONLD,this._allowedVideoRegex=e.allowedVideoRegex||this.REGEXPS.videos,this._flags=this.FLAG_STRIP_UNLIKELYS|this.FLAG_WEIGHT_CLASSES|this.FLAG_CLEAN_CONDITIONALLY,this._debug){let i=function(r){if(r.nodeType==r.TEXT_NODE)return`${r.nodeName} (\"${r.textContent}\")`;l