☰

Greasy Fork is available in English.

百度指数

百度指数爬虫工具

このスクリプトの質問や評価の投稿はこちら、通報はこちらへお寄せください。 
  1. // ==UserScript==
  2. // @name         百度指数
  3. // @namespace    http://tampermonkey.net/
  4. // @version      0.2
  5. // @description  百度指数爬虫工具
  6. // @author       siji-Xian
  7. // @match        *://index.baidu.com/v2/main/index.html*
  8. // @icon         https://www.google.com/s2/favicons?domain=oceanengine.com
  9. // @grant        none
  10. // @license      MIT
  11. // @require      https://lf26-cdn-tos.bytecdntp.com/cdn/expire-1-M/jquery/3.2.1/jquery.min.js
  12. // @require      https://cdn.bootcss.com/moment.js/2.20.1/moment.min.js
  13. // @require      https://greatest.deepsurf.us/scripts/404478-jsonexportexcel-min/code/JsonExportExcelmin.js?version=811266
  14. // @require      https://greatest.deepsurf.us/scripts/455576-qmsg/code/Qmsg.js?version=1122361
  15. // ==/UserScript==
  16.  
  17. (function () {
  18.   "use strict";
  19.   var new_element = document.createElement("link");
  20.   new_element.setAttribute("rel", "stylesheet");
  21.   new_element.setAttribute("href", "https://qmsg.refrain.xyz/message.min.css");
  22.   document.body.appendChild(new_element);
  23.  
  24.   const button = document.createElement("div");
  25.   button.textContent = "导出数据";
  26.   Object.assign(button.style, {
  27.     height: "34px",
  28.     lineHeight: "var(--line-height, 34px)",
  29.     width: "90px",
  30.     textAlign: "center",
  31.     alignItems: "center",
  32.     color: "white",
  33.     background: "linear-gradient(90deg, rgba(0, 239, 253), rgba(64, 166, 254))",
  34.     borderRadius: "5px",
  35.     marginLeft: "10px",
  36.     fontSize: "13px",
  37.     padding: "0 10px",
  38.     cursor: "pointer",
  39.     fontWeight: "500",
  40.   });
  41.   button.addEventListener("click", urlClick);
  42.  
  43.   //message.js
  44.   let loadingMsg = null;
  45.  
  46.   //加密数据
  47.   let target_data = [];
  48.   //密钥
  49.   let ptbk = null;
  50.  
  51.   (function listen() {
  52.     var origin = {
  53.       open: XMLHttpRequest.prototype.open,
  54.       send: XMLHttpRequest.prototype.send,
  55.     };
  56.     XMLHttpRequest.prototype.open = function (a, b) {
  57.       this.addEventListener("load", replaceFn);
  58.       origin.open.apply(this, arguments);
  59.     };
  60.     XMLHttpRequest.prototype.send = function (a, b) {
  61.       origin.send.apply(this, arguments);
  62.     };
  63.     function replaceFn(obj) {
  64.       if (
  65.         this?.responseURL?.slice(0, 43) ==
  66.         "https://index.baidu.com/api/SearchApi/index"
  67.       ) {
  68.         let res = JSON.parse(obj?.target?.response);
  69.         target_data = res.data.userIndexes;
  70.       }
  71.       if (
  72.         this?.responseURL?.slice(0, 38) ==
  73.         "https://index.baidu.com/Interface/ptbk"
  74.       ) {
  75.         let res = JSON.parse(obj?.target?.response);
  76.         ptbk = res.data;
  77.       }
  78.     }
  79.   })();
  80.  
  81.   function appendDoc() {
  82.     const likeComment = document.querySelector(".index-trend-words");
  83.     if (likeComment) {
  84.       likeComment.append(button);
  85.       return;
  86.     }
  87.     setTimeout(appendDoc, 1000);
  88.   }
  89.   appendDoc();
  90.  
  91.   let decrypt = function (t, e) {
  92.     for (
  93.       var n = t.split(""), i = e.split(""), a = {}, r = [], o = 0;
  94.       o < n.length / 2;
  95.       o++
  96.     )
  97.       a[n[o]] = n[n.length / 2 + o];
  98.     for (var s = 0; s < e.length; s++) r.push(a[i[s]]);
  99.     return r.join("");
  100.   };
  101.  
  102.   /**
  103.    * 得到开始和结束日期，得到中间所有天返回数组
  104.    * @param {String} startDay 开始日期'2021-7-1'
  105.    * @param {String} endDay 结束日期'2021-8-1'
  106.    * @return {Array} ['2021-07-01', '2021-07-01'...., '2021-08-01']
  107.    */
  108.   function getDayArr(startDay, endDay) {
  109.     let startVal = moment(startDay).format("YYYY-MM-DD");
  110.     let dayArr = [];
  111.     while (moment(startVal).isBefore(endDay)) {
  112.       dayArr.push(startVal);
  113.       // 自增
  114.       startVal = moment(startVal).add(1, "day").format("YYYY-MM-DD");
  115.     }
  116.     // 将结束日期的天放进数组
  117.     dayArr.push(moment(endDay).format("YYYY-MM-DD"));
  118.     return dayArr;
  119.   }
  120.  
  121.   function expExcel() {
  122.     let dates = getDayArr(target_data[0].all.startDate,target_data[0].all.endDate)
  123.     let data = target_data?.map((v,i)=>{
  124.       return {
  125.         key:v.word[0].name,
  126.         value:decrypt(ptbk, v.all.data)?.split(',')?.map((y,z)=>{return {date:dates[z],value:y}}),
  127.       }
  128.     });
  129.     
  130.     let contrast = {
  131.       date: "日期",
  132.       value: "值"
  133.     };
  134.  
  135.     let option = {};
  136.     option.fileName = "百度指数"; //文件名
  137.     option.datas = data.map(v=>{
  138.       return {
  139.         sheetName: v.key,
  140.         sheetData: v.value,
  141.         sheetHeader: Object.values(contrast),
  142.         sheetFilter: Object.keys(contrast),
  143.         columnWidths: [], // 列宽
  144.       }
  145.     });
  146.     var toExcel = new ExportJsonExcel(option);
  147.     toExcel.saveExcel();
  148.     loadingMsg.close();
  149.   }
  150.  
  151.   function urlClick() {
  152.     if (target_data.length) {
  153.       loadingMsg = Qmsg.loading("正在导出，请勿重复点击！");
  154.       expExcel(target_data);
  155.     } else {
  156.       loadingMsg = Qmsg.error("数据加载失败，请重试");
  157.     }
  158.   }
  159. })();