Home-Toolbox-Plugin/contents/read.ts

import { formatHTML } from "~utils/html";

const getTitle = () => {
  const title = document.title;
  const metaTitle = document.querySelector<HTMLMetaElement>(`meta[name="title"]`);
  const metaOgTitle = document.querySelector<HTMLMetaElement>(`meta[name="og:title"]`);

  if (metaOgTitle) {
    return metaOgTitle.getAttribute("content");;
  }

  if (metaTitle) {
    return metaTitle.getAttribute("content");;
  }

  if (title.includes(" - ")) {
    const end = title.lastIndexOf(" - ");

    return title.substring(0, end);
  }

  if (title.includes(" | ")) {
    const end = title.lastIndexOf(" | ");

    return title.substring(0, end);
  }

  return title;
}

const getDesc = () => {
  const articleFirstParagraph = document.querySelector<HTMLParagraphElement>("article p") || document.querySelector<HTMLParagraphElement>("p");
  const metaDescription = document.querySelector<HTMLMetaElement>(`meta[name="description"]`);
  const metaOgDescription = document.querySelector<HTMLMetaElement>(`meta[property="og:description"]`);

  if (metaOgDescription) {
    return metaOgDescription.getAttribute("content");
  }

  if (metaDescription) {
    return metaDescription.getAttribute("content");
  }

  if (articleFirstParagraph) {
    return articleFirstParagraph.innerText;
  }

  return "";
}

const getAuthor = () => {
  const metaAuthor = document.querySelector<HTMLMetaElement>(`meta[name="author"]`);
  const metaOgAuthor = document.querySelector<HTMLMetaElement>(`meta[property="og:article:author"]`);
  const itemPropAuthor = document.querySelector<HTMLMetaElement>(`[itemprop="author"] [itemprop="name"]`);

  if (metaAuthor) {
    return metaAuthor.getAttribute("content");
  }

  if (metaOgAuthor) {
    return metaOgAuthor.getAttribute("content");
  }

  if (itemPropAuthor) {
    return itemPropAuthor.getAttribute("content");
  }

  return "";
}

const getImage = () => {
  const metaOgImage = document.querySelector<HTMLMetaElement>(`meta[property="og:image"]`);

  if (metaOgImage) {
    return metaOgImage.getAttribute("content");
  }

  return "";
}

const getSiteName = () => {
  const title = document.title;
  const metaSiteName = document.querySelector<HTMLMetaElement>(`meta[property="og:site_name"]`);

  if (metaSiteName) {
    return metaSiteName.getAttribute("content");
  }

  if (title.includes(" - ")) {
    const start = title.lastIndexOf(" - ") + 3;

    return title.substring(start);
  }

  if (title.includes(" | ")) {
    const start = title.lastIndexOf(" | ") + 3;

    return title.substring(start);
  }

  return "";
}

const getTags = () => {
  const metaKeywords = document.querySelector<HTMLMetaElement>(`meta[name="keywords"]`);
  const itemPropKeywords = document.querySelector<HTMLMetaElement>(`meta[itemprop="keywords"]`);

  if (metaKeywords) {
    return metaKeywords.getAttribute("content");
  }

  if (itemPropKeywords) {
    return itemPropKeywords.getAttribute("content");
  }

  return "";
}

const getHTML = () => {
  // 微信
  if (location.host === "mp.weixin.qq.com") {
    console.log("微信");

    return document.querySelector(".rich_media_content").innerHTML;
  }

  // 腾讯云开发者社区
  if (location.host === "cloud.tencent.com") {
    console.log("腾讯云");

    return formatHTML(document.querySelector(".rno-markdown").innerHTML, (doc) => {
      // 替换掉腾讯云自己的链接
      const qcloudLinkEl = doc.querySelectorAll("a");
      qcloudLinkEl.forEach((el) => {
        if (el.href.includes("qcloud")) {
          el.parentNode.replaceChild(document.createTextNode(el.innerText), el);
        }
      });

      const mdTextEl = doc.querySelectorAll(".rno-markdown__textlink-new") as NodeListOf<HTMLElement>;
      mdTextEl.forEach((el) => {
        el.parentNode.replaceChild(document.createTextNode(el.innerText), el);
      });
    });
  }

  // 掘金
  const itemPropArticle = document.querySelector("[itemprop=articleBody] .markdown-body");

  if (itemPropArticle) {
    console.log("掘金");

    return formatHTML(itemPropArticle.innerHTML);
  }

  // CSDN
  const contentViews = document.getElementById("content_views");

  if (contentViews) {
    console.log("CSDN");

    return formatHTML(contentViews.innerHTML);
  }

  // 通用模式
  const articleEl = document.querySelector("article");

  if (articleEl) {
    console.log("文章标签");

    return formatHTML(articleEl.innerHTML);
  }

  // 尝试匹配到文章
  const firstParagraph = document.querySelector("p");

  let continueFindCorrectParent = true;
  let parentEl;
  let limitCount = 0;

  if (firstParagraph) {
    while (continueFindCorrectParent) {
      if (limitCount > 20) {
        continueFindCorrectParent = false;
        throw new Error("DOM 层级检测遍历次数过多");
      }

      limitCount += 1;

      parentEl = parentEl ? parentEl.parentElement : firstParagraph.parentElement;

      // 这个父下面有多个疑似正文的元素
      if (Array.from(parentEl.querySelectorAll("h2, h3, h4, p")).length > 1) {
        console.log(parentEl, `匹配结束，向上遍历 ${limitCount} 次`);
        continueFindCorrectParent = false;
      }
    }

    return formatHTML(parentEl.innerHTML);
  }

  return "";
}

chrome.runtime.onMessage.addListener((req, sender, send) => {
  if (req.type === "toolbox:getInfo") {
    send({
      title: getTitle(),
      link: `${location.origin}${location.pathname}`,
      desc: getDesc(),
      author: getAuthor(),
      image: getImage(),
      sitename: getSiteName(),
      tags: getTags(),
      raw_content: getHTML(),
    });
  }
});