☰

Greasy Fork is available in English.

Nowcoder Quiz Snatcher

自动化抓取牛客网上一些免费题库，并以CSV格式导出下载

  1. // ==UserScript==
  2. // @name         Nowcoder Quiz Snatcher
  3. // @namespace    https://www.nowcoder.com/
  4. // @version      0.1
  5. // @description  自动化抓取牛客网上一些免费题库，并以CSV格式导出下载
  6. // @author       Yifei Zhow
  7. // @match        https://www.nowcoder.com/*
  8. // @icon         https://www.google.com/s2/favicons?sz=64&domain=tampermonkey.net
  9. // @grant        none
  10. // @license MIT
  11. // ==/UserScript==
  12.  
  13. // 以此页面举例：https://www.nowcoder.com/exam/intelligent?questionJobId=2&tagId=21029
  14. const gKeywords = ['用户研究','原型设计','数据分析','产品常识','产品设计','产品规划','需求分析','竞品研究','文档撰写'];
  15. // 针对某一题库，每次点击出题具有随机性，抓取到达一定的阈值后停止
  16. const gThresholdLvl = 85;
  17.  
  18. /* JSON Object looks like:
  19.  {"用户研究":{"claim_to_have": 56, "already_grabbed": 30, content: [{
  20.    "id": 1261504920, "type": "单选题", "info": "题干", "options": ["A.甲甲甲","B.乙乙乙","C.丙丙丙","D.丁丁丁"]
  21.  }, ..., {...}]}}
  22.  */
  23. const kNowWorkingKey = 'TM_NOWCODER_WORKING_KEY';
  24. const kQuestionDictKey = 'TM_NOWCODER_QUESTION_DICT_KEY';
  25. const kClaimToHave = 'claim_to_have';
  26. const kAlreadyGrabbed = 'already_grabbed';
  27.  
  28.  
  29. // Utilities
  30. function getLocation(href) {
  31.     var match = href.match(/^(https?\:)\/\/(([^:\/?#]*)(?:\:([0-9]+))?)([\/]{0,1}[^?#]*)(\?[^#]*|)(#.*|)$/);
  32.     return match && {
  33.         href: href,
  34.         protocol: match[1],
  35.         host: match[2],
  36.         hostname: match[3],
  37.         port: match[4],
  38.         pathname: match[5],
  39.         search: match[6],
  40.         hash: match[7]
  41.     }
  42. }
  43.  
  44. function hashCode(s) {
  45.     for(var i = 0, h = 0; i < s.length; i++)
  46.         h = Math.imul(31, h) + s.charCodeAt(i) | 0;
  47.     return h;
  48. }
  49.  
  50. function exportToCsv(filename, rows) {
  51.     var processRow = function (row) {
  52.         var finalVal = '';
  53.         if (!Array.isArray(row)) {
  54.             // Mapping
  55.             const mappedRow = [row['id'], row['type'], row['info']];
  56.             Array.prototype.forEach.call(row['options'], function(el) {
  57.                 mappedRow.push(el);
  58.             });
  59.             row = mappedRow; // Deep copy preferred
  60.         }
  61.         for (var j = 0; j < row.length; j++) {
  62.             var innerValue = row[j] === null ? '' : row[j].toString();
  63.             if (row[j] instanceof Date) {
  64.                 innerValue = row[j].toLocaleString();
  65.             };
  66.             var result = innerValue.replace(/"/g, '""');
  67.             if (result.search(/("|,|\n)/g) >= 0)
  68.                 result = '"' + result + '"';
  69.             if (j > 0)
  70.                 finalVal += ',';
  71.             finalVal += result;
  72.         }
  73.         return finalVal + '\n';
  74.     };
  75.  
  76.     var csvFile = '';
  77.     for (var i = 0; i < rows.length; i++) {
  78.         csvFile += processRow(rows[i]);
  79.     }
  80.  
  81.     var blob = new Blob([csvFile], { type: 'text/csv;charset=utf-8;' });
  82.     if (navigator.msSaveBlob) { // IE 10+
  83.         navigator.msSaveBlob(blob, filename);
  84.     } else {
  85.         var link = document.createElement("a");
  86.         if (link.download !== undefined) { // feature detection
  87.             // Browsers that support HTML5 download attribute
  88.             var url = URL.createObjectURL(blob);
  89.             link.setAttribute("href", url);
  90.             link.setAttribute("download", filename);
  91.             link.style.visibility = 'hidden';
  92.             document.body.appendChild(link);
  93.             link.click();
  94.             document.body.removeChild(link);
  95.         }
  96.     }
  97. }
  98.  
  99. // Sub-process
  100. function storeQuestionIfNecessary(key, id, type, info, options = []) {
  101.     let dict;
  102.     try {
  103.         dict = JSON.parse(window.localStorage.getItem(kQuestionDictKey) || '{}');
  104.     } catch (error) {
  105.         // JSON parse error!
  106.         dict = {};
  107.     }
  108.     const keyDict = dict[key] || {};
  109.     const keyContent = keyDict['content'] || [];
  110.     for (let c of keyContent) {
  111.         if (c['id'] == id) { return false; }
  112.     }
  113.     keyContent.push({
  114.         id,
  115.         type,
  116.         info,
  117.         options
  118.     });
  119.     // Write-back
  120.     keyDict['content'] = keyContent;
  121.     keyDict[kAlreadyGrabbed] = (keyDict[kAlreadyGrabbed] || 0) + 1;
  122.     dict[key] = keyDict;
  123.     window.localStorage.setItem(kQuestionDictKey, JSON.stringify(dict));
  124.     return true;
  125. }
  126.  
  127. function willJumpToNextWords(workingKey) {
  128.     let dict;
  129.     try {
  130.         dict = JSON.parse(window.localStorage.getItem(kQuestionDictKey) || '{}');
  131.     } catch (error) {
  132.         return { 'next': false, 'final-word': false };
  133.     }
  134.     const keyDict = dict[workingKey] || {};
  135.     const total = keyDict[kClaimToHave] || 0;
  136.     const portion = keyDict[kAlreadyGrabbed] || 0;
  137.     if (total == 0) {
  138.         console.err('Now working percent: ', 'Unable to calculate.');
  139.         alert('Total count missing!');
  140.         return { 'next': false, 'final-word': false };
  141.     } else {
  142.         console.log('Now working percent: ', ((portion/total) * 100).toFixed(2) + '%');
  143.     }
  144.  
  145.     if ((portion/total) * 100 > gThresholdLvl) {
  146.         // Reach Threshold Level
  147.         var idx = gKeywords.indexOf(workingKey);
  148.         idx += 1;
  149.         if (idx < gKeywords.length) {
  150.             // Has next word
  151.             workingKey = gKeywords[idx];
  152.             window.localStorage.setItem(kNowWorkingKey, workingKey);
  153.             return { 'next': true, 'final-word': false };
  154.         } else {
  155.             // Current is final word
  156.             return { 'next': false, 'final-word': true };
  157.         }
  158.     } else {
  159.         // Not reached
  160.         return { 'next': false, 'final-word': false };
  161.     }
  162. }
  163.  
  164. function init() {
  165.     let key = window.localStorage.getItem(kNowWorkingKey);
  166.     if (!key) {
  167.         key = gKeywords[0];
  168.     }
  169.     window.localStorage.setItem(kNowWorkingKey, key);
  170. }
  171.  
  172. function reset() {
  173.     window.localStorage.removeItem(kNowWorkingKey);
  174.     window.localStorage.removeItem(kQuestionDictKey);
  175. }
  176.  
  177. function storeClaim(workingKey, claim) {
  178.     let dict;
  179.     try {
  180.         dict = JSON.parse(window.localStorage.getItem(kQuestionDictKey) || '{}');
  181.     } catch (error) {
  182.         return false;
  183.     }
  184.     const keyDict = dict[workingKey] || {};
  185.     const total = keyDict[kClaimToHave];
  186.     if (!total || total < 0 || total < claim) {
  187.         // Write-back
  188.         keyDict[kClaimToHave] = claim;
  189.         dict[workingKey] = keyDict;
  190.         window.localStorage.setItem(kQuestionDictKey, JSON.stringify(dict));
  191.     }
  192. }
  193. // Main
  194. (function() {
  195.     'use strict';
  196.     init();
  197.  
  198.     const h = getLocation(window.location.href);
  199.     const workingKey = window.localStorage.getItem(kNowWorkingKey);
  200.  
  201.     let els;
  202.     if (!h || h.hostname != "www.nowcoder.com") return;
  203.     if (h.pathname == "/exam/intelligent") {
  204.         // Index Page
  205.         els = document.getElementsByClassName("exercise-card") || [];
  206.         Array.prototype.forEach.call(els, function(el) {
  207.             const title = el.querySelector(".exercises-card-title").innerText;
  208.             const score = el.querySelector(".tw-text-gray-500").innerText;
  209.             if (title.indexOf(workingKey) != -1) {
  210.                 // Extract kClaimToHave and Save
  211.                 const mth = score.match(/已做(\d+)\/(\d+)题/);
  212.                 if (mth && mth[2]) {
  213.                     storeClaim(workingKey, parseInt(mth[2], 10));
  214.                     setTimeout(function() {
  215.                         el.click();
  216.                     }, 2000);
  217.                 }
  218.             }
  219.         });
  220.     } else if (h.pathname.indexOf('/exam/test/') != -1) {
  221.         els = document.querySelectorAll(".test-paper .paper-question");
  222.         Array.prototype.forEach.call(els, function(el) {
  223.             const qType = el.querySelector(".question-desc-header").innerText;
  224.             const qInfo = el.querySelector(".question-info").innerText;
  225.             let qOptions = [];
  226.             Array.prototype.forEach.call(el.querySelectorAll(".question-select .option-item"), function(e) { qOptions.push(e.innerText); });
  227.             const qID = hashCode(qInfo);
  228.             const isStored = storeQuestionIfNecessary(workingKey, qID, qType, qInfo, qOptions);
  229.             console.log('Is Stored: ', isStored, qType);
  230.         });
  231.         // Exam Page
  232.         let result = willJumpToNextWords(workingKey);
  233.         if (result['final-word']) {
  234.             if (confirm("Scnatched completed, export as CSV?")) {
  235.                  var dict = JSON.parse(window.localStorage.getItem(kQuestionDictKey));
  236.                 Array.prototype.forEach.call(gKeywords, function(k) {
  237.                     exportToCsv(k+'.csv', dict[k]['content']);
  238.                 });
  239.                 // reset();
  240.             }
  241.         } else {
  242.              setTimeout(function() {
  243.                  history.back();
  244.              }, 2000);
  245.         }
  246.     }
  247. })();