☰

信息抓取

文章、笔记信息抓取

  1. // ==UserScript==
  2. // @name         信息抓取
  3. // @namespace    http://tampermonkey.net/
  4. // @version      0.0.4
  5. // @description  文章、笔记信息抓取
  6. // @author       dabennn
  7. // @match        https://*.xiaohongshu.com/explore/*
  8. // @match        https://*.douyin.com/video/*
  9. // @match        https://baijiahao.baidu.com/*
  10. // @match        https://mp.weixin.qq.com/*
  11. // @match        https://*.sohu.com/a/*
  12. // @match        https://*.toutiao.com/article/*
  13. // @match        https://*.toutiao.com/w/*
  14. // @icon         https://www.google.com/s2/favicons?sz=64&domain=github.com
  15. // @license      MIT
  16. // @grant        GM_notification
  17. // @grant        unsafeWindow
  18. // @run-at       document-start
  19. // ==/UserScript==
  20.  
  21. (function () {
  22.   "use strict";
  23.   function formatTimestamp(timestamp) {
  24.     const date = new Date(timestamp);
  25.     const year = date.getFullYear();
  26.     const month = String(date.getMonth() + 1).padStart(2, "0");
  27.     const day = String(date.getDate()).padStart(2, "0");
  28.     const hours = String(date.getHours()).padStart(2, "0");
  29.     const minutes = String(date.getMinutes()).padStart(2, "0");
  30.     const seconds = String(date.getSeconds()).padStart(2, "0");
  31.     return `${year}-${month}-${day} ${hours}:${minutes}:${seconds}`;
  32.   }
  33.   function getElementText(domSelector) {
  34.     const el = document.querySelector(domSelector);
  35.     return el ? el.textContent.trim() : "";
  36.   }
  37.   function getElementTextByXPath(xpath) {
  38.     const el = document.evaluate(
  39.       xpath,
  40.       document,
  41.       null,
  42.       XPathResult.FIRST_ORDERED_NODE_TYPE,
  43.       null
  44.     ).singleNodeValue;
  45.     return el ? el.textContent.trim() : "";
  46.   }
  47.   function createInfoModal(infoText, copyText) {
  48.     // 创建一个用于显示信息的div元素作为浮窗
  49.     const infoDiv = document.createElement("div");
  50.     infoDiv.style.position = "fixed";
  51.     infoDiv.style.top = "0";
  52.     infoDiv.style.right = "0";
  53.     infoDiv.style.backgroundColor = "rgba(255, 255, 255, 0.8)";
  54.     infoDiv.style.padding = "10px";
  55.     infoDiv.style.border = "1px solid gray";
  56.     infoDiv.style.zIndex = "9999";
  57.  
  58.     // 创建关闭按钮元素
  59.     const closeButton = document.createElement("span");
  60.     closeButton.id = "__close_btn__";
  61.     closeButton.textContent = "×";
  62.     closeButton.style.cursor = "pointer";
  63.     closeButton.style.float = "right";
  64.     closeButton.style.fontSize = "18px";
  65.     closeButton.style.color = "red";
  66.  
  67.     // 创建复制按钮元素
  68.     const copyButton = document.createElement("span");
  69.     copyButton.id = "__copy__";
  70.     copyButton.textContent = "复制";
  71.     copyButton.style.cursor = "pointer";
  72.     copyButton.style.float = "right";
  73.     copyButton.style.marginRight = "5px";
  74.     copyButton.style.fontSize = "18px";
  75.     copyButton.style.color = "blue";
  76.  
  77.     // 先将关闭按钮添加到浮窗中
  78.     infoDiv.appendChild(closeButton);
  79.     // 再将复制按钮添加到浮窗中
  80.     infoDiv.appendChild(copyButton);
  81.  
  82.     // 将信息拼接成字符串并设置到浮窗的innerHTML中
  83.     infoDiv.innerHTML += infoText;
  84.  
  85.     // 将浮窗添加到页面的body元素中
  86.     document.body.appendChild(infoDiv);
  87.  
  88.     document
  89.       .querySelector("#__close_btn__")
  90.       .addEventListener("click", function () {
  91.         infoDiv.parentNode.removeChild(infoDiv);
  92.       });
  93.  
  94.     document.querySelector("#__copy__").onclick = function () {
  95.       navigator.clipboard
  96.         .writeText(copyText)
  97.         .then(function () {
  98.           GM_notification("已复制到剪贴板");
  99.         })
  100.         .catch(function (err) {
  101.           GM_notification("复制失败");
  102.           console.error("复制失败：", err);
  103.         });
  104.     };
  105.   }
  106.   function createFloatButton() {
  107.     const button = document.createElement("div");
  108.     button.textContent = "信息抓取";
  109.     button.id = "__float_btn__";
  110.     button.style.position = "fixed";
  111.     button.style.top = "150px";
  112.     button.style.right = "0";
  113.     button.style.zIndex = "9999";
  114.     button.style.backgroundColor = "rgba(255, 100, 100, 0.9)";
  115.     button.style.color = "#fff";
  116.     button.style.borderRadius = "4px";
  117.     button.style.padding = "10px";
  118.     button.style.cursor = "pointer";
  119.  
  120.     // 关闭按钮
  121.     const closeButton = document.createElement("span");
  122.     closeButton.textContent = "×";
  123.     closeButton.style.position = "absolute";
  124.     closeButton.style.top = "-3px";
  125.     closeButton.style.right = "2px";
  126.     closeButton.style.fontSize = "16px";
  127.     closeButton.style.lineHeight = "1";
  128.  
  129.     button.appendChild(closeButton);
  130.     closeButton.addEventListener("click", (e) => {
  131.       e.stopPropagation();
  132.       button.parentNode.removeChild(button);
  133.     });
  134.     document.body.appendChild(button);
  135.   }
  136.   const formatTextInfo = (texts) =>
  137.     texts.join("<br>").replace(/undefined/g, "未获取到");
  138.   const formatNumInfo = (texts) => texts.join(" ").replace(/undefined/g, "0");
  139.   const formatCopyNum = (num) => (/\d/.test(num) ? num : 0);
  140.   const getCopyTime = (time) => time.replace(/[年月]/g, "/").replace(/日/, "");
  141.   const getInfoTexts = () => {
  142.     let textInfo = "";
  143.     let copyInfo = "";
  144.     if (window.location.href.includes("xiaohongshu.com/explore")) {
  145.       const state = unsafeWindow.__INITIAL_STATE__;
  146.       const title = getElementText(".note-content .title");
  147.       const author = getElementText(".username");
  148.       let time = "";
  149.       try {
  150.         time = formatTimestamp(
  151.           state.note.noteDetailMap[state.note.firstNoteId.value].note
  152.             .lastUpdateTime
  153.         );
  154.       } catch (e) {
  155.         console.error(e);
  156.       }
  157.       const wordCount = getElementText(".note-content").length;
  158.       const likeNum = getElementText(
  159.         ".interact-container .like-wrapper .count"
  160.       );
  161.       const commentNum = getElementText(
  162.         ".interact-container .chat-wrapper .count"
  163.       );
  164.       const collectNum = getElementText(
  165.         ".interact-container .collect-wrapper .count"
  166.       );
  167.       textInfo = formatTextInfo([
  168.         `标题：${title}`,
  169.         `作者：${author}`,
  170.         `发布时间：${time}`,
  171.         `字数：${wordCount}`,
  172.         `点赞数：${likeNum}`,
  173.         `评论数：${commentNum}`,
  174.         `收藏数：${collectNum}`,
  175.       ]);
  176.       copyInfo = [
  177.         getCopyTime(time),
  178.         "小红书",
  179.         author,
  180.         title,
  181.         window.location.href,
  182.         formatNumInfo([
  183.           `点赞${formatCopyNum(likeNum)}`,
  184.           `评论${formatCopyNum(commentNum)}`,
  185.           `收藏${formatCopyNum(collectNum)}`,
  186.         ]),
  187.         wordCount,
  188.       ].join("\t");
  189.     } else if (window.location.href.includes("douyin.com/video")) {
  190.       const title = getElementText(
  191.         "h1 span span + span span span span span span"
  192.       );
  193.       const author = getElementTextByXPath(
  194.         '//*[@id="douyin-right-container"]/div[2]/div/div/div[1]/div[4]/div/div[1]/div[2]/a/div/span/span/span/span/span/span'
  195.       );
  196.       const time = getElementTextByXPath(
  197.         '//*[@id="douyin-right-container"]/div[2]/div/div/div[1]/div[3]/div/div[2]/div[2]/span/text()[2]'
  198.       );
  199.       const duration = getElementText(".time-duration");
  200.       const likeNum = getElementTextByXPath(
  201.         '//*[@id="douyin-right-container"]/div[2]/div/div/div[1]/div[3]/div/div[2]/div[1]/div[1]/span'
  202.       );
  203.       const commentNum = getElementTextByXPath(
  204.         '//*[@id="douyin-right-container"]/div[2]/div/div/div[1]/div[3]/div/div[2]/div[1]/div[2]/span'
  205.       );
  206.       const collectNum = getElementTextByXPath(
  207.         '//*[@id="douyin-right-container"]/div[2]/div/div/div[1]/div[3]/div/div[2]/div[1]/div[3]/span'
  208.       );
  209.       const shareNum = getElementTextByXPath(
  210.         '//*[@id="douyin-right-container"]/div[2]/div/div/div[1]/div[3]/div/div[2]/div[1]/div[4]/span'
  211.       );
  212.       textInfo = formatTextInfo([
  213.         `标题：${title}`,
  214.         `作者：${author}`,
  215.         `发布时间：${time}`,
  216.         `时长：${duration}`,
  217.         `点赞数：${likeNum}`,
  218.         `评论数：${commentNum}`,
  219.         `收藏数：${collectNum}`,
  220.         `分享数：${shareNum}`,
  221.       ]);
  222.       copyInfo = [
  223.         getCopyTime(time),
  224.         "抖音视频",
  225.         author,
  226.         title,
  227.         window.location.href,
  228.         formatNumInfo([
  229.           `点赞${formatCopyNum(likeNum)}`,
  230.           `评论${formatCopyNum(commentNum)}`,
  231.           `收藏${formatCopyNum(collectNum)}`,
  232.           `分享${formatCopyNum(shareNum)}`,
  233.         ]),
  234.         duration,
  235.       ].join("\t");
  236.     } else if (window.location.href.includes("sohu.com/a")) {
  237.       const title = getElementText("h1");
  238.       const author = getElementText("#user-info h4 a");
  239.       const time = getElementText("#news-time");
  240.       const wordCount = getElementText("mp-editor").length;
  241.       const readNum = getElementText(".read-num em");
  242.       const likeNum = getElementText(".like-c .count");
  243.       const commentNum = getElementText(".comment-count");
  244.       const collectNum = getElementText(".collection-c .count");
  245.       const shareNum = getElementText(".share-c .count");
  246.       textInfo = formatTextInfo([
  247.         `标题：${title}`,
  248.         `作者：${author}`,
  249.         `发布时间：${time}`,
  250.         `字数：${wordCount}`,
  251.         `阅读数：${readNum}`,
  252.         `点赞数：${likeNum}`,
  253.         `评论数：${commentNum}`,
  254.         `收藏数：${collectNum}`,
  255.         `分享数：${shareNum}`,
  256.       ]);
  257.       copyInfo = [
  258.         getCopyTime(time),
  259.         "搜狐",
  260.         author,
  261.         title,
  262.         window.location.href,
  263.         formatNumInfo([
  264.           `阅读${formatCopyNum(readNum)}`,
  265.           `点赞${formatCopyNum(likeNum)}`,
  266.           `评论${formatCopyNum(commentNum)}`,
  267.           `收藏${formatCopyNum(collectNum)}`,
  268.           `分享${formatCopyNum(shareNum)}`,
  269.         ]),
  270.         wordCount,
  271.       ].join("\t");
  272.     } else if (window.location.href.includes("toutiao.com/article")) {
  273.       const title = getElementText("h1");
  274.       const author = getElementText(".article-meta .name");
  275.       const time = getElementText(".article-meta span");
  276.       const wordCount = getElementText(".tt-article-content").length;
  277.       const likeNum = getElementText(".detail-like span");
  278.       const commentNum = getElementText(".detail-interaction-comment span");
  279.       const collectNum = getElementText(".detail-interaction-collect span");
  280.       textInfo = formatTextInfo([
  281.         `标题：${title}`,
  282.         `作者：${author}`,
  283.         `发布时间：${time}`,
  284.         `字数：${wordCount}`,
  285.         `点赞数：${likeNum}`,
  286.         `评论数：${commentNum}`,
  287.         `收藏数：${collectNum}`,
  288.       ]);
  289.       copyInfo = [
  290.         getCopyTime(time),
  291.         "今日头条",
  292.         author,
  293.         title,
  294.         window.location.href,
  295.         formatNumInfo([
  296.           `点赞${formatCopyNum(likeNum)}`,
  297.           `评论${formatCopyNum(commentNum)}`,
  298.           `收藏${formatCopyNum(collectNum)}`,
  299.         ]),
  300.         wordCount,
  301.       ].join("\t");
  302.     } else if (window.location.href.includes("toutiao.com/w")) {
  303.       const title = getElementText("h1");
  304.       const author = getElementText(".desc .name");
  305.       const time = getElementText(".abstract .time");
  306.       const wordCount = getElementText("article").length;
  307.       const likeNum = getElementText(".detail-like span");
  308.       const commentNum = getElementText(".detail-interaction-comment span");
  309.       const collectNum = getElementText(".detail-interaction-collect span");
  310.       textInfo = formatTextInfo([
  311.         `标题：${title}`,
  312.         `作者：${author}`,
  313.         `发布时间：${time}`,
  314.         `字数：${wordCount}`,
  315.         `点赞数：${likeNum}`,
  316.         `评论数：${commentNum}`,
  317.         `收藏数：${collectNum}`,
  318.       ]);
  319.       copyInfo = [
  320.         getCopyTime(time),
  321.         "今日头条",
  322.         author,
  323.         title,
  324.         window.location.href,
  325.         formatNumInfo([
  326.           `点赞${formatCopyNum(likeNum)}`,
  327.           `评论${formatCopyNum(commentNum)}`,
  328.           `收藏${formatCopyNum(collectNum)}`,
  329.         ]),
  330.         wordCount,
  331.       ].join("\t");
  332.     } else if (window.location.href.includes("baijiahao.baidu.com")) {
  333.       const title = getElementText("#header div");
  334.       const author = getElementText("#header [data-testid=author-name]");
  335.       const time = getElementText("#header [data-testid=updatetime]");
  336.       const wordCount = getElementText("[data-testid=article]").length;
  337.       const likeNum = getElementText("[data-testid=like-btn] .interact-desc");
  338.       const commentNum = getElementText(
  339.         "[data-testid=comment-btn] .interact-desc"
  340.       );
  341.       const collectNum = getElementText(
  342.         "[data-testid=favor-btn] .interact-desc"
  343.       );
  344.       const shareNum = getElementText("[data-testid=share-btn] .interact-desc");
  345.       textInfo = formatTextInfo([
  346.         `标题：${title}`,
  347.         `作者：${author}`,
  348.         `发布时间：${time}`,
  349.         `字数：${wordCount}`,
  350.         `点赞数：${likeNum}`,
  351.         `评论数：${commentNum}`,
  352.         `收藏数：${collectNum}`,
  353.         `分享数：${shareNum}`,
  354.       ]);
  355.       copyInfo = [
  356.         getCopyTime(time),
  357.         "百度",
  358.         author,
  359.         title,
  360.         window.location.href,
  361.         formatNumInfo([
  362.           `点赞${formatCopyNum(likeNum)}`,
  363.           `评论${formatCopyNum(commentNum)}`,
  364.           `收藏${formatCopyNum(collectNum)}`,
  365.           `分享${formatCopyNum(shareNum)}`,
  366.         ]),
  367.         wordCount,
  368.       ].join("\t");
  369.     } else if (window.location.href.includes("mp.weixin.qq.com")) {
  370.       const title = getElementText("h1");
  371.       const author = getElementText("#js_name");
  372.       const time = getElementText("#publish_time");
  373.       const wordCount = getElementText("#js_content").length;
  374.       textInfo = formatTextInfo([
  375.         `标题：${title}`,
  376.         `作者：${author}`,
  377.         `发布时间：${time}`,
  378.         `字数：${wordCount}`,
  379.       ]);
  380.       copyInfo = [
  381.         getCopyTime(time),
  382.         "公众号",
  383.         author,
  384.         title,
  385.         window.location.href,
  386.         formatNumInfo([`点赞`, `转发`, `喜欢`, `评论`]),
  387.         wordCount,
  388.       ].join("\t");
  389.     }
  390.     return {
  391.       textInfo,
  392.       copyInfo,
  393.     };
  394.   };
  395.  
  396.   window.addEventListener("load", function () {
  397.     createFloatButton();
  398.     document.querySelector("#__float_btn__").addEventListener("click", () => {
  399.       const { textInfo, copyInfo } = getInfoTexts();
  400.       createInfoModal(textInfo, copyInfo);
  401.     });
  402.   });
  403. })();

欢迎访问Greasy Fork镜像

土豆服务器，请按需使用

镜像地址随时可能被墙，建议加群获取最新地址

加入Q群

QingJ © 2025

镜像随时可能失效，请加Q群300939539或关注我们的公众号极客氢云获取最新地址