link-previu/dist/index.mjs

A Node.js library for getting link previews

// src/index.ts
import axios from "axios";
import NodeCache from "node-cache";
import * as cheerio from "cheerio";

// src/helpers.ts
import urlRegexSafe from "url-regex-safe";
function isValidUrl(url) {
  if (!url) return false;
  try {
    new URL(url);
    return urlRegexSafe({ exact: true }).test(url);
  } catch {
    return false;
  }
}
function getVideoIdFromYoutubeUrl(url) {
  const match = url.match(/^(?:https?:\/\/)?(?:www\.)?(?:m\.)?(?:youtu\.be\/|youtube\.com\/(?:embed\/|v\/|watch\?(?:.*&)?v=|shorts\/|live\/))([^"&?\/\s]{11})/);
  return match ? match[1] : null;
}
function hasAllMetadata(data) {
  return !!(data.title && data.desc && data.image && data.url);
}

// src/parsers/html.parser.ts
var HtmlParser = class {
  constructor(document) {
    this.document = document;
  }
  parse() {
    const baseDto = {};
    baseDto.title = this.getTitle();
    baseDto.desc = this.getDescription();
    baseDto.image = this.getImage();
    baseDto.siteName = this.getSiteName();
    return baseDto;
  }
  getTitle() {
    return this.document("head title").text();
  }
  getDescription() {
    return this.document("head meta[name='description']").attr("content") || this.document("head meta[property='description']").attr("content");
  }
  getImage() {
    return this.document("body img").first().attr("src");
  }
  getSiteName() {
    return this.document("head meta[name='site_name']").attr("content") || this.document("head meta[property='site_name']").attr("content");
  }
};

// src/parsers/open-graph.parser.ts
var OpenGraphParser = class {
  constructor(document) {
    this.document = document;
  }
  parse() {
    const metadata = {};
    metadata.title = this.getMetaProperty("og:title");
    metadata.desc = this.getMetaProperty("og:description");
    metadata.image = this.getMetaProperty("og:image");
    metadata.url = this.getMetaProperty("og:url");
    metadata.siteName = this.getMetaProperty("og:site_name");
    return metadata;
  }
  getMetaProperty(property) {
    return this.document(`meta[property='${property}']`).attr("content") || this.document(`meta[name='${property}']`).attr("content");
  }
};

// src/parsers/json-ld.parser.ts
var JsonLdParser = class {
  constructor(document) {
    this.document = document;
    this.jsonData = this.parseToJson(document);
  }
  jsonData;
  parse() {
    const metadata = {};
    metadata.title = this.getTitle();
    metadata.desc = this.getDescription();
    metadata.image = this.getImage();
    metadata.siteName = new OpenGraphParser(this.document).parse().siteName;
    return metadata;
  }
  parseToJson(document) {
    const data = document("script[type='application/ld+json']").html();
    if (!data) return null;
    try {
      const cleanedData = data.replace(/\n/g, " ").trim();
      return JSON.parse(cleanedData);
    } catch (error) {
      return null;
    }
  }
  getTitle() {
    const data = this.jsonData;
    if (!data) return void 0;
    if (Array.isArray(data) && data.length > 0) {
      return data[0]["name"] || data[0]["headline"];
    } else if (typeof data === "object" && data !== null) {
      return data["name"] || data["headline"];
    }
    return void 0;
  }
  getDescription() {
    const data = this.jsonData;
    if (!data) return void 0;
    if (Array.isArray(data) && data.length > 0) {
      return data[0]["description"] || data[0]["headline"];
    } else if (typeof data === "object" && data !== null) {
      return data["description"] || data["headline"];
    }
    return void 0;
  }
  getImage() {
    const data = this.jsonData;
    if (!data) return void 0;
    if (Array.isArray(data) && data.length > 0) {
      return this.imageResultToString(data[0]["logo"] || data[0]["image"]);
    } else if (typeof data === "object" && data !== null) {
      return this.imageResultToString(data["logo"] || data["image"]);
    }
    return void 0;
  }
  imageResultToString(result) {
    if (Array.isArray(result) && result.length > 0) {
      result = result[0];
    }
    if (typeof result === "string") {
      return result;
    } else if (result && typeof result === "object") {
      return result["url"] || result["contentUrl"];
    }
    return void 0;
  }
};

// src/parsers/other.parser.ts
var OtherParser = class {
  constructor(document) {
    this.document = document;
  }
  parse() {
    const metadata = {};
    metadata.title = this.getTitle();
    metadata.desc = this.getDescription();
    metadata.image = this.getImage();
    metadata.siteName = this.getSiteName();
    metadata.url = this.getUrl();
    return metadata;
  }
  getTitle() {
    return this.getMetaProperty("name", "title");
  }
  getDescription() {
    return this.getMetaProperty("name", "description");
  }
  getImage() {
    return this.getMetaProperty("name", "image");
  }
  getSiteName() {
    return this.getMetaProperty("name", "site_name");
  }
  getUrl() {
    return this.getMetaProperty("name", "url");
  }
  getMetaProperty(attribute, property) {
    return this.document(`meta[${attribute}='${property}']`).attr("content");
  }
};

// src/parsers/twitter.parser.ts
var TwitterParser = class {
  constructor(document) {
    this.document = document;
  }
  parse() {
    const metadata = {};
    const ogParser = new OpenGraphParser(this.document).parse();
    metadata.title = this.getTitle();
    metadata.desc = this.getDescription();
    metadata.image = this.getImage();
    metadata.title = metadata.title || ogParser.title;
    metadata.desc = metadata.desc || ogParser.desc;
    metadata.image = metadata.image || ogParser.image;
    metadata.siteName = ogParser.siteName;
    metadata.url = ogParser.url;
    if (!metadata.title && ogParser.siteName?.includes("Twitter")) {
      metadata.title = "Twitter Post";
    }
    if (metadata.image?.endsWith(".svg") && !metadata.image.includes("twimg.com/media")) {
      metadata.image = void 0;
    }
    return metadata;
  }
  getTitle() {
    return this.getMetaProperty("name", "twitter:title") || this.getMetaProperty("property", "twitter:title") || this.document('meta[name="title"]').attr("content");
  }
  getDescription() {
    return this.getMetaProperty("name", "twitter:description") || this.getMetaProperty("property", "twitter:description") || this.document('meta[name="description"]').attr("content");
  }
  getImage() {
    const image = this.getMetaProperty("name", "twitter:image") || this.getMetaProperty("property", "twitter:image") || this.getMetaProperty("name", "twitter:image:src") || this.getMetaProperty("property", "twitter:image:src");
    if (image?.endsWith(".svg") && !image.includes("media")) {
      return void 0;
    }
    return image;
  }
  getMetaProperty(attribute, property) {
    return this.document(`meta[${attribute}='${property}']`).attr("content");
  }
};

// src/parsers/youtube.parser.ts
var YoutubeParser = class {
  constructor(document) {
    this.document = document;
    this.jsonData = this.parseToJson(document);
  }
  jsonData;
  parse() {
    const metadata = {};
    metadata.title = this.getTitle();
    metadata.image = this.getImage();
    metadata.siteName = this.getSiteName();
    metadata.url = this.getUrl();
    return metadata;
  }
  parseToJson(document) {
    try {
      const html = document.html() || "";
      const cleanedData = html.replace("<html><head></head><body>", "").replace("</body></html>", "").replace(/\n/g, " ").trim();
      return JSON.parse(cleanedData);
    } catch (error) {
      return null;
    }
  }
  getTitle() {
    const data = this.jsonData;
    if (!data) return void 0;
    if (Array.isArray(data) && data.length > 0) {
      return data[0]["title"];
    } else if (typeof data === "object" && data !== null) {
      return data["title"];
    }
    return void 0;
  }
  getImage() {
    const data = this.jsonData;
    if (!data) return void 0;
    if (Array.isArray(data) && data.length > 0) {
      return this.imageResultToString(data[0]["thumbnail_url"]);
    } else if (typeof data === "object" && data !== null) {
      return this.imageResultToString(data["thumbnail_url"]);
    }
    return void 0;
  }
  getSiteName() {
    const data = this.jsonData;
    if (!data) return void 0;
    if (Array.isArray(data) && data.length > 0) {
      return data[0]["provider_name"];
    } else if (typeof data === "object" && data !== null) {
      return data["provider_name"];
    }
    return void 0;
  }
  getUrl() {
    const data = this.jsonData;
    if (!data) return void 0;
    if (Array.isArray(data) && data.length > 0) {
      return data[0]["provider_url"];
    } else if (typeof data === "object" && data !== null) {
      return data["provider_url"];
    }
    return void 0;
  }
  imageResultToString(result) {
    if (Array.isArray(result) && result.length > 0) {
      result = result[0];
    }
    if (typeof result === "string") {
      return result;
    } else if (result && typeof result === "object") {
      return result["url"] || result["contentUrl"];
    }
    return void 0;
  }
};

// src/index.ts
var LinkPreview = class {
  redis;
  nodeCache;
  cacheMaxAge;
  cacheEnabled;
  requestTimeout;
  maxRedirects;
  httpHeaders = {
    "User-Agent": "WhatsApp/2.23.4.79 A"
    // default option
  };
  /**
   * Creates a new LinkPreview instance
   * @param options Configuration options for link preview fetching and caching
   */
  constructor(options) {
    this.requestTimeout = options.requestTimeout ?? 5e3;
    this.cacheMaxAge = options.cacheMaxAge ?? 0;
    this.cacheEnabled = Boolean(options.cacheMaxAge);
    this.maxRedirects = options.maxRedirects ?? 5;
    this.httpHeaders = {
      ...this.httpHeaders,
      ...options.httpHeaders ?? {}
    };
    if (this.cacheEnabled) {
      this.redis = options.redis;
      this.nodeCache = options.redis ? void 0 : new NodeCache();
    }
  }
  /**
   * Fetches preview data for a given URL
   * @param url The URL to fetch preview data for
   * @returns Promise containing the preview data
   */
  async getLinkPreview(url) {
    const cacheKey = `link-preview-node:${url}`;
    if (!isValidUrl(url)) {
      return null;
    }
    if (this.cacheEnabled) {
      if (this.redis) {
        const cachedData = await this.redis.get(cacheKey);
        if (cachedData) {
          return JSON.parse(cachedData);
        }
      } else {
        const cachedData = this.nodeCache?.get(cacheKey);
        if (cachedData) {
          return cachedData;
        }
      }
    }
    const videoId = getVideoIdFromYoutubeUrl(url);
    const response = videoId ? await this.getYoutubeData(videoId) : await this.fetchWithRedirects(url);
    const contentType = response.headers["content-type"];
    if (contentType?.startsWith("image/")) {
      return {
        title: "",
        desc: "",
        siteName: "",
        image: url,
        url
      };
    }
    const document = this.responseToDocument(response.data);
    if (!document) {
      return null;
    }
    const metadata = this.extractMetadata(document, url);
    if (this.cacheEnabled) {
      if (this.redis) {
        await this.redis.set(cacheKey, JSON.stringify(metadata), "EX", this.cacheMaxAge);
      } else {
        this.nodeCache?.set(cacheKey, metadata, this.cacheMaxAge);
      }
    }
    return metadata;
  }
  /**
   * Extracts metadata from a given document
   * @param document The document to extract metadata from
   * @param url The URL of the document
   * @returns The extracted metadata
   */
  extractMetadata(document, url) {
    const output = {};
    const parsers = [
      new OpenGraphParser(document).parse(),
      new TwitterParser(document).parse(),
      new YoutubeParser(document).parse(),
      new JsonLdParser(document).parse(),
      new HtmlParser(document).parse(),
      new OtherParser(document).parse()
    ];
    for (const p of parsers) {
      if (!p) continue;
      output.title = output.title ?? p.title;
      output.desc = output.desc ?? p.desc;
      output.image = output.image ?? p.image;
      output.siteName = output.siteName ?? p.siteName;
      output.url = output.url ?? p.url ?? url;
      if (hasAllMetadata(output)) break;
    }
    if (output.url && output.image) {
      try {
        const baseUrl = new URL(output.url);
        output.image = new URL(output.image, baseUrl).toString();
      } catch (error) {
        output.image = void 0;
      }
    }
    return output;
  }
  /**
   * Converts an HTML string to a cheerio document
   * @param html The HTML string to convert
   * @returns The converted cheerio document or null if conversion fails
   */
  responseToDocument(html) {
    try {
      return cheerio.load(html);
    } catch (err) {
      throw err;
    }
  }
  /**
   * Fetches data from a given URL with redirects
   * @param url The URL to fetch data from
   * @returns The fetched data
   */
  async fetchWithRedirects(url) {
    try {
      return await axios.get(url, {
        headers: { ...this.httpHeaders },
        maxRedirects: this.maxRedirects,
        timeout: this.requestTimeout
      });
    } catch (err) {
      throw err;
    }
  }
  /**
   * Fetches data from a given YouTube video ID
   * @param videoId The YouTube video ID to fetch data from
   * @returns The fetched data
   */
  async getYoutubeData(videoId) {
    const url = `https://www.youtube.com/watch?v=${videoId}`;
    return await this.fetchWithRedirects(url);
  }
};
var index_default = LinkPreview;
export {
  index_default as default
};