☰

Ximalaya Opus Data Collector

自动采集喜马拉雅“个人中心”专辑“播放”和“订阅”数据。

  1. // ==UserScript==
  2. // @name         Ximalaya Opus Data Collector
  3. // @namespace    http://tampermonkey.net/
  4. // @version      2.9
  5. // @description  自动采集喜马拉雅“个人中心”专辑“播放”和“订阅”数据。
  6. // @author       You
  7. // @match        https://studio.ximalaya.com/opus
  8. // @grant        none
  9. // @license      MIT
  10. // ==/UserScript==
  11. (function() {
  12.     'use strict';
  13.  
  14.     let isCollecting = false;
  15.     let allData = [];
  16.     let timeoutId = null;
  17.  
  18.     // 创建按钮和输入框
  19.     function createControls() {
  20.         // 停止采集按钮
  21.         const stopButton = createButton('停止采集', '210px', '#dc3545', () => {
  22.             if (isCollecting) {
  23.                 isCollecting = false;
  24.                 if (timeoutId) {
  25.                     clearTimeout(timeoutId);
  26.                     timeoutId = null;
  27.                 }
  28.                 stopButton.disabled = true;
  29.                 stopButton.textContent = '已停止';
  30.                 saveAndDownloadCurrentData();
  31.                 alert('采集已停止，当前数据已保存为文件！');
  32.                 returnToFirstPage();
  33.                 setTimeout(() => {
  34.                     stopButton.disabled = false;
  35.                     stopButton.textContent = '停止采集';
  36.                     allData = [];
  37.                 }, 2000);
  38.             } else {
  39.                 alert('当前未在采集数据！');
  40.             }
  41.         });
  42.         document.body.appendChild(stopButton);
  43.  
  44.         // 指定页码范围采集区域
  45.         const rangeContainer = document.createElement('div');
  46.         rangeContainer.style.position = 'fixed';
  47.         rangeContainer.style.top = '260px';  // 原230px + 30px
  48.         rangeContainer.style.right = '0px';
  49.         rangeContainer.style.zIndex = '9999';
  50.  
  51.         const startLabel = document.createElement('label');
  52.         startLabel.textContent = '起始页: ';
  53.         startLabel.style.marginRight = '5px';
  54.         startLabel.style.backgroundColor = '#666';
  55.         startLabel.style.color = 'white';
  56.         startLabel.style.padding = '2px 5px';
  57.         const startInput = document.createElement('input');
  58.         startInput.type = 'number';
  59.         startInput.value = '1';
  60.         startInput.min = '1';
  61.         startInput.id = 'startPage';
  62.         startInput.style.width = '60px';
  63.         startInput.style.marginRight = '10px';
  64.         startInput.style.backgroundColor = '#666';
  65.         startInput.style.color = 'white';
  66.  
  67.         const endLabel = document.createElement('label');
  68.         endLabel.textContent = '结束页: ';
  69.         endLabel.style.marginRight = '5px';
  70.         endLabel.style.backgroundColor = '#666';
  71.         endLabel.style.color = 'white';
  72.         endLabel.style.padding = '2px 5px';
  73.         const endInput = document.createElement('input');
  74.         endInput.type = 'number';
  75.         endInput.value = '57';
  76.         endInput.min = '1';
  77.         endInput.max = '100';
  78.         endInput.id = 'endPage';
  79.         endInput.style.width = '60px';
  80.         endInput.style.marginRight = '0px';
  81.         endInput.style.backgroundColor = '#666';
  82.         endInput.style.color = 'white';
  83.  
  84.         const rangeButton = createButton('开始采集', '150px', '#17a2b8', () => {
  85.             // [保持原有逻辑不变]
  86.             const startInputEl = document.getElementById('startPage');
  87.             const endInputEl = document.getElementById('endPage');
  88.             if (!startInputEl || !endInputEl) {
  89.                 alert('输入框未找到，请刷新页面重试！');
  90.                 return;
  91.             }
  92.             const start = parseInt(startInputEl.value, 10);
  93.             const end = parseInt(endInputEl.value, 10);
  94.             if (isNaN(start) || isNaN(end) || start < 1 || end > 100 || start > end) {
  95.                 alert('输入的页码无效！请确保起始页大于等于 1，结束页小于等于 100，且起始页小于等于结束页。');
  96.                 return;
  97.             }
  98.             if (!isCollecting) {
  99.                 isCollecting = true;
  100.                 allData = [];
  101.                 rangeButton.textContent = '正在采集';
  102.                 rangeButton.disabled = true;
  103.                 stopButton.disabled = false;
  104.                 collectRangePages(start, end).then(() => {
  105.                     isCollecting = false;
  106.                     rangeButton.textContent = '开始采集';
  107.                     rangeButton.disabled = false;
  108.                     stopButton.disabled = false;
  109.                     checkAndTriggerNextCollection();
  110.                 }).catch(err => {
  111.                     isCollecting = false;
  112.                     rangeButton.textContent = '开始采集';
  113.                     rangeButton.disabled = false;
  114.                     console.error(err);
  115.                     alert('采集失败: ' + err.message);
  116.                     stopButton.disabled = false;
  117.                 });
  118.             } else {
  119.                 alert('正在采集中，请先停止当前任务！');
  120.             }
  121.         });
  122.         rangeContainer.appendChild(startLabel);
  123.         rangeContainer.appendChild(startInput);
  124.         rangeContainer.appendChild(endLabel);
  125.         rangeContainer.appendChild(endInput);
  126.         rangeContainer.appendChild(rangeButton);
  127.         document.body.appendChild(rangeContainer);
  128.  
  129.         // 循环控制和时间间隔区域
  130.         const loopContainer = document.createElement('div');
  131.         loopContainer.style.position = 'fixed';
  132.         loopContainer.style.top = '300px';  // 原280px + 30px
  133.         loopContainer.style.right = '0px';
  134.         loopContainer.style.zIndex = '9999';
  135.  
  136.         const loopCheckbox = document.createElement('input');
  137.         loopCheckbox.type = 'checkbox';
  138.         loopCheckbox.id = 'loopEnabled';
  139.         loopCheckbox.checked = false;
  140.         const loopLabel = document.createElement('label');
  141.         loopLabel.textContent = ' 启用循环采集';
  142.         loopLabel.style.marginRight = '10px';
  143.         loopLabel.style.backgroundColor = '#666';
  144.         loopLabel.style.color = 'white';
  145.         loopLabel.style.padding = '2px 5px';
  146.         loopLabel.htmlFor = 'loopEnabled';
  147.  
  148.         const intervalLabel = document.createElement('label');
  149.         intervalLabel.textContent = '循环间隔 (分钟): ';
  150.         intervalLabel.style.marginRight = '5px';
  151.         intervalLabel.style.backgroundColor = '#666';
  152.         intervalLabel.style.color = 'white';
  153.         intervalLabel.style.padding = '2px 5px';
  154.         const intervalInput = document.createElement('input');
  155.         intervalInput.type = 'number';
  156.         intervalInput.value = '30';
  157.         intervalInput.min = '1';
  158.         intervalInput.id = 'intervalMinutes';
  159.         intervalInput.style.width = '60px';
  160.         intervalInput.style.backgroundColor = '#666';
  161.         intervalInput.style.color = 'white';
  162.         loopContainer.appendChild(loopCheckbox);
  163.         loopContainer.appendChild(loopLabel);
  164.         loopContainer.appendChild(intervalLabel);
  165.         loopContainer.appendChild(intervalInput);
  166.         document.body.appendChild(loopContainer);
  167.     }
  168.  
  169.     // 创建按钮的辅助函数
  170.     function createButton(text, top, color, onClick) {
  171.         const button = document.createElement('button');
  172.         button.textContent = text;
  173.         button.style.position = 'fixed';
  174.         button.style.top = top;
  175.         button.style.right = '0px';
  176.         button.style.zIndex = '9999';
  177.         button.style.padding = '8px 16px';
  178.         button.style.backgroundColor = color;
  179.         button.style.color = 'white';
  180.         button.style.border = 'none';
  181.         button.style.borderRadius = '4px';
  182.         button.style.cursor = 'pointer';
  183.         button.addEventListener('click', onClick);
  184.         return button;
  185.     }
  186.  
  187.  
  188.     // 等待 iframe 加载（3秒超时）
  189.     function waitForIframe(timeout = 3000) {
  190.         return new Promise((resolve, reject) => {
  191.             const iframe = document.getElementById('contentWrapper');
  192.             if (!iframe) {
  193.                 reject(new Error('未找到 iframe 元素'));
  194.                 return;
  195.             }
  196.  
  197.             const checkIframe = () => {
  198.                 if (iframe.contentDocument && iframe.contentDocument.body) {
  199.                     clearInterval(interval);
  200.                     clearTimeout(timeoutId);
  201.                     resolve(iframe.contentDocument);
  202.                 }
  203.             };
  204.  
  205.             const interval = setInterval(checkIframe, 100);
  206.             const timeoutId = setTimeout(() => {
  207.                 clearInterval(interval);
  208.                 reject(new Error('等待 iframe 加载超时（3秒）'));
  209.             }, timeout);
  210.         });
  211.     }
  212.  
  213.     // 采集指定页码范围的数据
  214.     async function collectRangePages(startPage, endPage) {
  215.         console.log(`开始采集从第 ${startPage} 页到第 ${endPage} 页的数据...`);
  216.         allData = []; // 清空之前的数据
  217.         let currentPage = startPage;
  218.  
  219.         try {
  220.             // 刷新 iframe，确保内容重置但脚本状态不变
  221.             await refreshIframe();
  222.             const iframeDoc = await waitForIframe(3000); // 3 秒等待，确保加载
  223.             let hasNextPage = true;
  224.  
  225.             while (hasNextPage && currentPage <= endPage && isCollecting) {
  226.                 console.log(`正在采集第 ${currentPage} 页...`);
  227.                 try {
  228.                     const pageData = await collectPageData(iframeDoc, currentPage);
  229.                     allData = allData.concat(pageData.filter(item => item.plays !== '0')); // 过滤播放数为 0 的数据
  230.  
  231.                     // 等待 3 秒，确保数据加载完成
  232.                     await new Promise(resolve => setTimeout(resolve, 3000));
  233.  
  234.                     if (currentPage === endPage) {
  235.                         hasNextPage = false;
  236.                         saveToCsv(allData, 'all');
  237.                         returnToFirstPage();
  238.                     } else {
  239.                         // 检查是否有下一页
  240.                         const nextButton = iframeDoc.querySelector('.ant-pagination-next:not(.ant-pagination-disabled)');
  241.                         const paginationItems = iframeDoc.querySelectorAll('.ant-pagination-item');
  242.                         const lastPageItem = Array.from(paginationItems).pop();
  243.                         const isLastPage = !nextButton || nextButton.getAttribute('aria-disabled') === 'true' ||
  244.                                          (lastPageItem && lastPageItem.textContent.trim() === currentPage.toString());
  245.  
  246.                         if (isLastPage) {
  247.                             hasNextPage = false;
  248.                             saveToCsv(allData, 'all');
  249.                             returnToFirstPage();
  250.                         } else {
  251.                             // 模拟点击下一页按钮
  252.                             nextButton.click();
  253.                             currentPage++;
  254.  
  255.                             // 等待 3 秒，确保页面翻转并加载完成
  256.                             await new Promise(resolve => setTimeout(resolve, 3000));
  257.                         }
  258.                     }
  259.                 } catch (err) {
  260.                     console.warn(`第 ${currentPage} 页采集失败: ${err.message}，继续下一页`);
  261.                     currentPage++;
  262.                     continue;
  263.                 }
  264.  
  265.                 if (!isCollecting) break; // 检查是否需要停止
  266.             }
  267.  
  268.             if (allData.length === 0) {
  269.                 throw new Error('未采集到任何数据，请确保页面已加载专辑列表！');
  270.             }
  271.         } catch (err) {
  272.             throw err;
  273.         }
  274.     }
  275.  
  276.     // 采集单页数据
  277.     async function collectPageData(iframeDoc, pageNumber) {
  278.         console.log(`采集第 ${pageNumber} 页数据...`);
  279.         const albumItems = iframeDoc.querySelectorAll('[class*="AlbumItem_listItem"]');
  280.         console.log(`找到 ${albumItems.length} 个专辑项`);
  281.  
  282.         if (albumItems.length === 0) {
  283.             throw new Error(`第 ${pageNumber} 页未找到专辑数据！`);
  284.         }
  285.  
  286.         const pageData = [];
  287.         const currentTime = new Date().toLocaleString('zh-CN', {
  288.             year: 'numeric',
  289.             month: '2-digit',
  290.             day: '2-digit',
  291.             hour: '2-digit',
  292.             minute: '2-digit'
  293.         }).replace(/\//g, '-').replace(' ', '-'); // 格式化为 "YYYY-MM-DD-HH:MM"
  294.  
  295.         albumItems.forEach((item, index) => {
  296.             let title = '未知专辑';
  297.             let episodes = '0';
  298.             let plays = '0';
  299.             let subscribes = '0';
  300.  
  301.             // 提取标题，增加容错
  302.             const titleEl = item.querySelector('.AlbumItem_title__1q92X, [class*="title"]');
  303.             if (titleEl) {
  304.                 title = titleEl.textContent.trim().replace('公开', '') || '未知专辑';
  305.             }
  306.  
  307.             // 提取集数，增加容错
  308.             const episodeEl = item.querySelector('.AlbumItem_number__2lJS9 span, [class*="number"] span');
  309.             if (episodeEl) {
  310.                 episodes = episodeEl.textContent.replace(/[^\d]/g, '').trim() || '0';
  311.             }
  312.  
  313.             // 提取播放和订阅，增加容错并处理“万”、“亿”格式
  314.             const dataEls = item.querySelectorAll('.AlbumItem_data__upkkX span, [class*="data"] span');
  315.             dataEls.forEach(data => {
  316.                 const text = data.textContent.trim();
  317.                 if (text.includes('播放')) {
  318.                     plays = convertNumericFormat(text) || '0';
  319.                 } else if (text.includes('订阅')) {
  320.                     subscribes = convertNumericFormat(text) || '0';
  321.                 }
  322.             });
  323.  
  324.             pageData.push({ title, episodes, plays, subscribes, time: currentTime });
  325.             console.log(`采集第 ${pageNumber} 页的第 ${index + 1} 个专辑: ${title}, 集数: ${episodes}, 播放: ${plays}, 订阅: ${subscribes}, 时间: ${currentTime}`);
  326.         });
  327.  
  328.         return pageData;
  329.     }
  330.  
  331.     // 转换数值格式（如“2.64万”转换为 26400）
  332.     function convertNumericFormat(text) {
  333.         if (!text) return '0';
  334.         const match = text.match(/([\d.]+)(万|亿)?/);
  335.         if (!match) return '0';
  336.         const value = parseFloat(match[1]);
  337.         const unit = match[2];
  338.         if (unit === '亿') {
  339.             return Math.round(value * 100000000).toString(); // 亿 -> 整数
  340.         } else if (unit === '万') {
  341.             return Math.round(value * 10000).toString(); // 万 -> 整数
  342.         } else {
  343.             return Math.round(value).toString(); // 纯数字
  344.         }
  345.     }
  346.  
  347.     // 保存并下载当前数据
  348.     function saveAndDownloadCurrentData() {
  349.         if (allData.length === 0) return;
  350.  
  351.         let csvContent = '\ufeff' + '专辑名,集数,播放,订阅,时间\n'; // 确保 UTF-8 BOM 避免中文乱码
  352.         allData.forEach(item => {
  353.             csvContent += `"${item.title}",${item.episodes},${item.plays},${item.subscribes},"${item.time}"\n`;
  354.         });
  355.  
  356.         const blob = new Blob([csvContent], { type: 'text/csv;charset=utf-8;' });
  357.         const link = document.createElement('a');
  358.         const timestamp = new Date().toLocaleString('zh-CN', {
  359.             year: 'numeric',
  360.             month: '2-digit',
  361.             day: '2-digit',
  362.             hour: '2-digit',
  363.             minute: '2-digit'
  364.         }).replace(/\//g, '-').replace(' ', '-'); // 格式化为 "YYYY-MM-DD-HH:MM"
  365.         link.setAttribute('href', URL.createObjectURL(blob));
  366.         link.setAttribute('download', `中心数据${timestamp}.csv`);
  367.         document.body.appendChild(link);
  368.         link.click();
  369.         document.body.removeChild(link);
  370.     }
  371.  
  372.     // 保存数据到 CSV 文件
  373.     function saveToCsv(data, pageNumber) {
  374.         let csvContent = '\ufeff' + '专辑名,集数,播放,订阅,时间\n'; // 确保 UTF-8 BOM 避免中文乱码
  375.         data.forEach(item => {
  376.             csvContent += `"${item.title}",${item.episodes},${item.plays},${item.subscribes},"${item.time}"\n`;
  377.         });
  378.  
  379.         const blob = new Blob([csvContent], { type: 'text/csv;charset=utf-8;' });
  380.         const link = document.createElement('a');
  381.         const timestamp = new Date().toLocaleString('zh-CN', {
  382.             year: 'numeric',
  383.             month: '2-digit',
  384.             day: '2-digit',
  385.             hour: '2-digit',
  386.             minute: '2-digit'
  387.         }).replace(/\//g, '-').replace(' ', '-'); // 格式化为 "YYYY-MM-DD-HH:MM"
  388.         link.setAttribute('href', URL.createObjectURL(blob));
  389.         link.setAttribute('download', `中心数据${timestamp}.csv`);
  390.         document.body.appendChild(link);
  391.         link.click();
  392.         document.body.removeChild(link);
  393.     }
  394.  
  395.     // 返回到第 1 页
  396.     function returnToFirstPage() {
  397.         const iframe = document.getElementById('contentWrapper');
  398.         if (iframe && iframe.contentDocument) {
  399.             const firstPageButton = iframe.contentDocument.querySelector('.ant-pagination-item-1:not(.ant-pagination-item-active)');
  400.             if (firstPageButton) {
  401.                 firstPageButton.click();
  402.                 console.log('已返回到第 1 页');
  403.             }
  404.         }
  405.     }
  406.  
  407.     // 刷新 iframe
  408.     function refreshIframe() {
  409.         return new Promise((resolve) => {
  410.             const iframe = document.getElementById('contentWrapper');
  411.             if (iframe) {
  412.                 iframe.contentWindow.location.reload(); // 仅刷新 iframe
  413.                 setTimeout(() => {
  414.                     waitForIframe(3000).then(iframeDoc => {
  415.                         resolve(iframeDoc);
  416.                     }).catch(err => {
  417.                         console.error('刷新 iframe 后等待超时:', err);
  418.                         resolve(null); // 即使超时也继续
  419.                     });
  420.                 }, 1000); // 等待 1 秒后检查，总计 3 秒
  421.             } else {
  422.                 resolve(waitForIframe(3000)); // 如果 iframe 还未加载，直接等待
  423.             }
  424.         });
  425.     }
  426.  
  427.     // 检查是否启用循环并触发下一次采集
  428.     function checkAndTriggerNextCollection() {
  429.         const loopCheckbox = document.getElementById('loopEnabled');
  430.         if (loopCheckbox && loopCheckbox.checked) {
  431.             const intervalInput = document.getElementById('intervalMinutes');
  432.             if (!intervalInput) {
  433.                 console.error('循环间隔输入框未找到');
  434.                 return;
  435.             }
  436.             const intervalMinutes = parseInt(intervalInput.value, 10) || 30; // 默认30分钟
  437.             timeoutId = setTimeout(() => {
  438.                 refreshPageAndTriggerCollection();
  439.             }, intervalMinutes * 60 * 1000); // 转换为毫秒
  440.         }
  441.     }
  442.  
  443.     // 刷新 iframe 并触发下一次采集，增加 5 秒间隔
  444.     function refreshPageAndTriggerCollection() {
  445.         return new Promise((resolve) => {
  446.             const iframe = document.getElementById('contentWrapper');
  447.             if (iframe) {
  448.                 iframe.contentWindow.location.reload(); // 仅刷新 iframe
  449.                 setTimeout(() => {
  450.                     // 增加 5 秒间隔后模拟按下“开始采集”按钮
  451.                     setTimeout(() => {
  452.                         const rangeButton = document.querySelector('button[style*="top: 0px"][style*="right: 0px"]');
  453.                         if (rangeButton) {
  454.                             rangeButton.click(); // 模拟按下“开始采集”按钮
  455.                             console.log('已模拟按下“开始采集”按钮');
  456.                         } else {
  457.                             console.error('未找到“开始采集”按钮');
  458.                         }
  459.                         resolve();
  460.                     }, 5000); // 5 秒间隔
  461.                 }, 3000); // 等待 3 秒确保 iframe 加载
  462.             } else {
  463.                 console.error('未找到 iframe 元素');
  464.                 resolve();
  465.             }
  466.         });
  467.     }
  468.  
  469.     // 页面加载完成后添加按钮和输入框
  470.     window.addEventListener('load', () => {
  471.         createControls();
  472.     });
  473. })();

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