- // ==UserScript==
- // @name Wenku Doc Downloader
- // @namespace http://tampermonkey.net/
- // @version 1.3.7
- // @description 下载“百度文库”文档,导出txt或pdf。支持①豆丁网②爱问共享资料(新浪文档)③得力文库④道客巴巴,文档导出pdf。在文档页面最最底部有蓝/绿色长方形按钮,说明脚本生效了。2021/12/17百度文档更新了技术,【脚本对大部分百度文档无效了】,请等待以后更新,抱歉。
- // @author allenlv2690@gmail.com
- // @note 紧急修复了部分豆丁网文档无效的bug。
- // @match https://wenku.baidu.com/view/*
- // @match https://*.docin.com/p-*
- // @match https://ishare.iask.sina.com.cn/f/*
- // @match https://www.deliwenku.com/p-*
- // @match *://www.doc88.com/p-*
- // @match https://wk.baidu.com/view/*
- // @require https://cdn.bootcdn.net/ajax/libs/FileSaver.js/2.0.5/FileSaver.js
- // @require https://cdn.bootcdn.net/ajax/libs/jszip/3.6.0/jszip.js
- // @require https://cdn.bootcdn.net/ajax/libs/jspdf/2.3.1/jspdf.umd.min.js
- // @icon https://wenku.baidu.com/favicon.ico
- // @grant none
- // @license GPL-3.0-only
- // @create 2021-11-22
- // ==/UserScript==
-
- "use strict";
-
- let options = {
- "fast_mode": false
- }
-
- const jsPDF = jspdf.jsPDF;
-
- /*
- * 附属功能函数部分
- */
-
- /**
- * 创建并下载文件
- * @param {string} fileName
- * @param {string} content
- */
- function createAndDownloadFile(fileName, content) {
- let aTag = document.createElement('a');
- let blob = new Blob([content]);
- aTag.download = fileName;
- aTag.href = URL.createObjectURL(blob);
- aTag.click();
- URL.revokeObjectURL(blob);
- }
-
- /**
- * 允许打印页面
- */
- function allowPrint() {
- let style = document.createElement("style");
- style.innerHTML = `@media print {
- body{
- display:block;
- }
- }`;
- document.getElementsByTagName("head")[0].appendChild(style);
- }
-
- /**
- * 隐藏按钮,打印页面,显示按钮
- */
- function hideBtnThenPrint() {
- // 隐藏按钮,然后打印页面
- let section = document.getElementsByClassName("btns_section")[0];
- section.style.display = "none";
- window.print();
- // 打印结束,显示按钮
- section.style.removeProperty("display");
- }
-
- /**
- * 增强按钮(默认为蓝色按钮:展开文档)的点击效果
- * @param {String} custom_btn 按钮变量名
- */
- function enhanceBtnClickReaction(custom_btn = null) {
- let aim_btn;
- // 如果不使用自定义按钮元素,则默认为使用蓝色展开文档按钮
- if (!custom_btn || custom_btn === "btn_1") {
- aim_btn = document.getElementsByClassName("init-btn")[0];
- } else if (custom_btn === "btn_2") {
- aim_btn = document.querySelector(".save-doc-btn");
- } else if (custom_btn === "btn_3") {
- aim_btn = document.querySelector(".save-html-btn");
- console.log(aim_btn);
- }
-
- let old_color = aim_btn.style.color; // 保存旧的颜色
- let old_text = aim_btn.textContent; // 保存旧的文字内容
- // 暗红缩小
- aim_btn.style.color = "#c90000";
- aim_btn.style.fontWeight = "normal";
- aim_btn.textContent = `——>[ ${old_text} ]<——`;
- // 复原加粗
- let changeColor = function() {
- aim_btn.style.color = old_color;
- aim_btn.style.fontWeight = "bold";
- aim_btn.textContent = old_text;
- };
- setTimeout(changeColor, 1500);
- }
-
- /**
- * 用input框跳转到对应页码
- * @param {Element} cur_page 当前页码
- * @param {string} aim_page 目标页码
- * @param {string} event_type 键盘事件类型:"keyup" | "keypress" | "keydown"
- */
- function jump2pageNo(cur_page, aim_page, event_type) {
- // 设置跳转页码为目标页码
- cur_page.value = aim_page;
- // 模拟回车事件来跳转
- let keyboard_event_enter = new KeyboardEvent(event_type, {
- bubbles: true,
- cancelable: true,
- keyCode: 13
- });
- cur_page.dispatchEvent(keyboard_event_enter);
- }
-
- /**
- * 滚动到页面底部
- */
- function scrollToBottom() {
- window.scrollTo({
- top: document.body.scrollHeight,
- behavior: "smooth"
- });
- }
-
- /**
- * 用于纯文本文档的文本美化,适用性:百度文库
- * @param {string} text
- * @returns 美化后的文本
- */
- function formatText(text) {
- let reg_exp_1 = new RegExp(" ?[(]?=[\u4e00-\u9fa5] ?[)]");
- let reg_exp_2 = new RegExp("(?<=TEMP[\u4e00-\u9fa5]) ?");
-
- let text_1 = text.replace(reg_exp_1, "TEMP");
- let text_2 = text_1.replace(reg_exp_2, "");
- let text_3 = text_2.replace("TEMP", "");
- let text_final = text_3.replace(/ /g, " ");
- return text_final;
- }
-
- /**
- * 用于图形文字混合型文档的文本美化,适用性:百度文库
- * @param {string} text
- * @returns 美化后的文本
- */
- function formatText2(text) {
- let reg_exp = new RegExp("[ ]{2,}");
- let content_1 = text.replace(reg_exp, "\n");
-
- let content_2 = content_1.replace(/[ ]\n/g, "\n");
-
- let reg_exp_2 = new RegExp("\n[ ]*\n*\n");
- let content_3 = content_2.replace(reg_exp_2, "\n");
-
- let reg_exp_3 = new RegExp(" *\n * ");
- let content_4 = content_3.replace(reg_exp_3, "\n");
-
- let content_5 = content_4.replace(/[ ]/g, " ");
- let final_content = content_5.replace(/[ \n]精选文档[ \n]/g).replace(/\n{2,}/g, "\n");
-
- return final_content;
- }
-
- /**
- * 判断文档的详细类型,适用性:百度文库
- * @returns 源文档类型-实际内容类型,如: doc-pic-word
- */
- function detectType() {
- // 获取文档类型名称
- let file_type, type;
- try {
- let doc_title_wrap = document.getElementsByClassName("doc-title-wrap")[0];
- file_type = doc_title_wrap.children[0].className;
- } catch (e) {
- alert("请刷新页面以激活该按钮。\n先点击【导出pdf】橙色按钮后该按钮将无法使用。");
- return "safe_quit";
- }
- // 判断文档类型
- if (file_type.search("word") !== -1) {
- type = "word";
- } else if (file_type.search("ppt") !== -1) {
- type = "ppt";
- } else if (file_type.search("excel") !== -1) {
- type = "excel";
- } else if (file_type.search("pdf") !== -1) {
- type = "pdf";
- } else if (file_type.search("txt" !== -1)) {
- type = "txt";
- } else {
- type = file_type;
- }
- // 分别尝试获取相应元素列表,若列表长度为0则不存在相应元素,否则存在
- let pic_nums = document.getElementsByClassName("reader-pic-item").length;
- let word_nums = document.getElementsByClassName("reader-word-layer").length;
- let ppt_img_nums = document.getElementsByClassName("ppt-image-wrap").length;
-
- // 判断文档类型、文字和图片的数量状况
- if ((type === "ppt" && ppt_img_nums > 1) || (type === "pdf" && !word_nums && !pic_nums && ppt_img_nums)) {
- // ppt: 包含至少2页内容 / 纯ppt图形页面构成
- return "ppt";
- } else {
- return type;
- }
- }
-
- /**
- * 用try移除元素
- * @param {Element} element 要移除的元素
- */
- function tryToRemoveElement(element) {
- try {
- element.remove();
- } catch (e) {
- console.log();
- }
- }
-
- /**
- * 用try移除 [元素列表1, 元素列表2, ...] 中的元素
- * @param {Array} elem_list_box 要移除的元素列表构成的列表
- */
- function tryToRemoveSameElem(elem_list_box) {
- for (let elem_list of elem_list_box) {
- if (!elem_list) {
- continue;
- }
- for (let elem of elem_list) {
- try {
- elem.remove();
- } catch (e) {
- console.log();
- }
- }
- }
- }
-
- /**
- * 使文档在页面上居中
- * @param {String} class_name 文档元素的class
- * @param {String} default_offset 文档部分向右偏移的百分比(0-59)
- * @returns 偏移值是否合法
- */
- function centerDoc(class_name, default_offset) {
- let doc_main = document.getElementsByClassName(class_name)[0];
- let offset = window.prompt("请输入偏移百分位:", default_offset);
- // 如果输入的数字不在 0-59 内,提醒用户重新设置
- if (offset.length === 1 && offset.search(/[0-9]/) !== -1) {
- doc_main.style.marginLeft = offset + "%";
- return true;
- } else if (offset.length === 2 && offset.search(/[1-5][0-9]/) !== -1) {
- doc_main.style.marginLeft = offset + "%";
- return true
- } else {
- alert("请输入一个正整数,范围在0至59之间,用来使文档居中\n(不同文档偏移量不同,所以需要手动调整)");
- return false;
- }
- }
-
- /**
- * 提示文档已经展开,改变按钮形态:隐去蓝色按钮,显示绿色按钮
- * @param {Boolean} shrink_btn2 是否缩小绿色按钮
- * @param {Boolean} recommend_btn2 是否推荐绿色按钮
- * @param {String} btn2_new_text (如果需要)替换绿色按钮的文本为新文本
- * @param {Boolean} use_hint 是否提示文档已经展开
- */
- function transformButtons(shrink_btn2 = false, recommend_btn2 = false, btn2_new_text = "", use_hint = true) {
- // 提示文档已经展开
- if (use_hint) {
- let hint = "文档已经完全展开,可以导出";
- alert(hint);
- }
- // 准备调整按钮,先获取按钮
- let btn_1 = document.getElementsByClassName("init-btn")[0];
- let btn_2 = document.getElementsByClassName("save-doc-btn")[0];
- // 调整按钮显示状况
- btn_1.style.display = "none";
- btn_2.style.removeProperty("display");
- if (shrink_btn2) {
- btn_2.style.width = "34.8%";
- }
- if (btn2_new_text) {
- btn_2.textContent = btn2_new_text;
- }
- if (recommend_btn2) {
- btn_2.textContent += "(推荐)";
- }
- }
-
- /*
- * 主要功能函数部分
- */
-
- /**
- * 清理并打印得力文库的文档页
- */
- function printPageDeliwenku() {
- // 移除页面上的无关元素
- let selector = ".hr-wrap, #readshop, .nav_uis, .bookdesc, #boxright, .QQ_S1, .QQ_S, #outer_page_more, .works-manage-box.shenshu, .works-intro, .mt10.related-pic-box, .mt10.works-comment, .foot_nav, .siteInner";
- let elem_list = document.querySelectorAll(selector);
- for (let elem of elem_list) {
- tryToRemoveElement(elem);
- }
- // 修改页间距
- let outer_pages = document.getElementsByClassName("outer_page");
- for (let page of outer_pages) {
- page.style.marginBottom = "20px";
- }
- // 使文档居中
- alert("建议使用:\n偏移量: 3\n缩放: 112\n请上下滚动页面,确保每页内容都加载完成以避免空白页\n如果预览时有空白页或文末有绿色按钮,请取消打印重试");
- if (!centerDoc("boxleft", "3")) {
- return; // 如果输入非法,终止函数调用
- }
- // 打印文档
- hideBtnThenPrint();
- }
-
- /**
- * 清理并打印爱问共享资料的文档页
- * @returns 如果输入偏移量非法,返回空值以终止函数
- */
- function printPageiShare() {
- // # 清理并打印爱问共享资料的文档页
- // ## 移除页面上无关的元素
- // ### 移除单个元素
- let topbanner = document.getElementsByClassName("detail-topbanner")[0];
- let header = document.getElementsByClassName("new-detail-header")[0];
- let fixright = document.getElementById("fix-right");
- let redpacket = document.getElementsByClassName("loginRedPacket-dialog")[0];
- let fixedrightfull = document.getElementsByClassName("fixed-right-full")[0];
- let footer = document.getElementsByClassName("website-footer")[0];
- let guess = document.getElementsByClassName("guess-you-like-warpper")[0];
- let detailtopbox = document.getElementsByClassName("detail-top-box")[0];
- let fullscreen = document.getElementsByClassName("reader-fullScreen")[0];
- let endhint = document.getElementsByClassName("endof-trial-reading")[0];
- let crumb_arrow;
- try { crumb_arrow = document.getElementsByClassName("crumb-arrow")[0].parentElement; } catch (e) { console.log(); }
- let copyright = document.getElementsByClassName("copyright-container")[0];
- let state_btn = document.getElementsByClassName("state-bottom")[0];
- let comments = document.getElementsByClassName("user-comments-wrapper")[0];
- // ### 执行移除
- let elem_list = [
- topbanner,
- header,
- fixright,
- redpacket,
- fixedrightfull,
- footer,
- guess,
- detailtopbox,
- fullscreen,
- endhint,
- crumb_arrow,
- copyright,
- state_btn,
- comments
- ];
- for (let elem of elem_list) {
- tryToRemoveElement(elem);
- }
- // ### 移除全部同类元素
- let elem_list_2 = document.querySelectorAll(".tui-detail, .adv-container")
- for (let elem_2 of elem_list_2) {
- tryToRemoveElement(elem_2);
- }
- // 使文档居中
- alert("建议使用:\n偏移量: 18\n缩放: 默认\n如果预览中有广告,就取消打印\n再点一次按钮,预览中应该就没有广告了");
- if (!centerDoc("doc-main", "18")) {
- return; // 如果输入非法,终止函数调用
- }
- // 隐藏按钮,然后打印页面
- hideBtnThenPrint();
- }
-
- /**
- * 清理并打百度文库的文档页
- * @returns 如果输入偏移量非法,返回空值以终止函数
- */
- function printPageBaidu() {
- // # 清理并打百度文库的文档页
- // ## 移除无关页面元素
- // ### 要移除的单个元素
- let header_wrapper = document.getElementsByClassName("header-wrapper")[0];
- let right_wrapper = document.getElementById("right-wrapper-id");
- let reader_topbar = document.getElementsByClassName("reader-topbar")[0];
- let end_fold_page = document.getElementsByClassName("try-end-fold-page")[0];
- let catalog = document.querySelector(".catalog");
- let bottom = document.querySelector(".journal");
-
- let elem_list = [
- header_wrapper,
- right_wrapper,
- reader_topbar,
- end_fold_page,
- catalog,
- bottom
- ]
- for (let elem of elem_list) {
- tryToRemoveElement(elem);
- }
- // ### 移除全部同类元素
- let lazy_load_list = document.getElementsByClassName("lazy-load");
- let no_full_screen_list = document.getElementsByClassName("no-full-screen");
- let ads = document.getElementsByClassName("hx-warp");
-
- tryToRemoveSameElem([lazy_load_list, ads, no_full_screen_list]);
-
- // ## 文档类型测试
- // ### 常规文档测试
- let normal = Boolean(document.getElementsByClassName("left-wrapper")[0]);
- // ### 文献/期刊文档测试
- let journal = Boolean(document.getElementById("#journal-view"))
- let class_name;
- if (normal) {
- class_name = "left-wrapper";
- } else if (journal) {
- class_name = "right-wrapper";
- } else {
- alert("文档部分元素class不为left-wrapper,且id不为journal-view\n是未知文档类型,无法处理,请联系作者\nQQ: 2690874578\nEmail: allenlv2690@gmail.com");
- }
-
- // 使文档居中
- alert("建议使用:\n偏移量: 0\n缩放: 118%\n请上下滚动页面,确保每页内容都加载完成以避免空白页\n如果预览时有空白页或文末有按钮,请取消打印重试");
- if (!centerDoc(class_name, "0")) {
- return; // 如果输入非法,退出函数调用
- }
- // 隐藏按钮,然后打印页面
- hideBtnThenPrint();
- }
-
- function printPageWK() {
- // 去水印
- document.querySelectorAll("[class*=__wm]").forEach((elem) => {
- elem.className += " __web-inspector-hide-shortcut__";
- });
- // 调整页面边距、圆角、背景色
- let pages = document.querySelectorAll(".retype-page");
- pages.forEach((page) => {
- page.style.borderRadius = "0";
- page.style.background = "";
- page.style.padding = "0 0 0 0";
- });
- // 页面间隔调为0
- document.querySelectorAll(".gap").forEach((gap) => {
- gap.style.height = "0";
- });
- hideBtnThenPrint();
- }
-
- /**
- * 清理WK页面的无效元素
- */
- function clearPageWK() {
- // 去除元素
- let selector = ".vip-cashier-dialog-wrap, .top-card, .college-strong-guide-contain, .reader-pop-manager-view-containter, .middle-box-root";
- document.querySelectorAll(selector).forEach((elem) => {
- tryToRemoveElement(elem);
- });
- // 去除父级元素
- let child_selector = ".bartop, .barbottom";
- document.querySelectorAll(child_selector).forEach((elem) => {
- tryToRemoveElement(elem.parentElement);
- });
- // 前项元素去除
- let previous_selector = ".reader-pop-manager-view-containter";
- document.querySelectorAll(previous_selector).forEach((elem) => {
- tryToRemoveElement(elem.previousElementSibling);
- });
- }
-
- /**
- * 创建“打印页面到PDF”按钮
- * @param {Function} printPageWebsite
- * @returns btn_3元素引用
- */
- function createPrintPageBtn(printPageWebsite) {
- let btn_3 = document.createElement("button");
- // 样式设定
- btn_3.setAttribute("class", "save-html-btn");
- btn_3.style.height = "25px";
- btn_3.style.width = "15%";
- btn_3.style.marginLeft = "0.2%";
- btn_3.style.backgroundColor = "orange";
- btn_3.style.border = "none";
- btn_3.textContent = "导出pdf";
- btn_3.style.color = "black";
- btn_3.style.fontWeight = "bold";
- btn_3.style.borderRadius = "10%";
- btn_3.style.zIndex = "99999";
- // 绑定事件,添加到页面上
- btn_3.onclick = () => {
- enhanceBtnClickReaction("btn_3");
- printPageWebsite();
- };
- let section = document.getElementsByClassName("btns_section")[0];
- section.appendChild(btn_3);
- return btn_3;
- }
-
- /**
- * 点击“继续阅读”,适用性:得力文库
- */
- function readAllDeliwenku() {
- // 点击“同意并开始预览全文”
- let start_btn = document.getElementsByClassName("pre_button")[0];
- let display = start_btn.parentElement.parentElement.style.display;
- // 如果该按钮显示着,则点击,然后滚动至页面底部,最后终止函数
- if (!display) {
- start_btn.children[0].click();
- setTimeout("scroll(0, document.body.scrollHeight)", 200);
- return;
- }
- // 增强按钮点击效果
- enhanceBtnClickReaction();
-
- let read_all_btn = document.getElementsByClassName("fc2e")[0];
- let display2 = read_all_btn.parentElement.parentElement.style.display
- // 继续阅读
- if (display2 !== "none") {
- // 获取input元素
- let cur_page = document.querySelector("#pageNumInput");
- let page_old = cur_page.value;
- let page_max = cur_page.parentElement.nextElementSibling.textContent.replace(" / ", "");
- // 跳转到尾页
- jump2pageNo(cur_page, page_max, "keydown");
- // 跳转回来
- jump2pageNo(cur_page, page_old, "keydown");
-
- // 切换按钮准备导出
- } else {
- // 推荐导出全部图片链接
- transformButtons(true, true);
- // btn_3 橙色按钮
- createPrintPageBtn(printPageDeliwenku);
- }
- }
-
- /**
- * 点击“展开继续阅读”,适用性:爱尚共享资料
- */
- function readAlliShare() {
- // 获取“继续阅读”元素
- let red_btn = document.getElementsByClassName("red-color")[0];
- let red_text = red_btn.textContent;
- // 增强按钮点击效果
- enhanceBtnClickReaction();
- // 如果可以展开,则展开
- if (red_text.search("点击可继续阅读") !== -1) {
- red_btn.click();
- setTimeout(readAlliShare, 1000);
- }
- // 否则启动按钮2,准备清理页面然后打印为PDF
- else {
- // 平滑往返页面顶部和底部
- window.scrollTo({
- top: 0,
- behavior: "smooth"
- });
-
- // 推荐导出全部图片链接
- transformButtons(true, true);
- // btn_3,橙色按钮
- createPrintPageBtn(printPageiShare);
-
- // 显示svg图片的链接
- let page1 = document.querySelector('[data-num="1"] .data-detail embed');
- if (!page1) {
- // 如果不存在svg图形,终止后续代码
- return;
- }
- let page2 = document.querySelector('[data-num="2"] .data-detail embed');
- let [svg1_src_div, svg2_src_div] = [document.createElement("div"), document.createElement("div")];
- svg1_src_div.innerHTML = `<div id="src-1"
- style="font-weight: bold;font-size: 20px; height: 100px; width: 100%">
- 访问以下链接以复制文字:<br>${page1.src}
- </div>`;
- svg2_src_div.innerHTML = `<div id="src-1"
- style="font-weight: bold;font-size: 20px; height: 100px; width: 100%">
- 访问以下链接以复制文字:<br>${page2.src}
- </div>`;
- // 添加到页面上
- page1.parentElement.parentElement.parentElement.append(svg1_src_div);
- page2.parentElement.parentElement.parentElement.append(svg2_src_div);
- }
- }
-
-
- /**
- * 点击“继续阅读”,适用性:百度文库
- * @returns 特殊文档类型不予下载,返回null
- */
- function readAll() {
- // 获取“继续阅读”按钮
- let read_all_btn = document.getElementsByClassName("read-all")[0];
- // 如果存在“继续阅读”的按钮
- if (read_all_btn) {
- // 跳转到文末(等同于展开全文)
- let cur_page = document.getElementsByClassName("cur-page")[0];
- // 取得最大页码
- let page_max = cur_page.parentElement.children[2];
- // 设置跳转页码为最大页码
- cur_page.value = page_max;
- // 跳转到尾页
- jump2pageNo(cur_page, page_max, "keyup");
-
- // 否则认为已经展开了文档,判断文档类型,然后显示对应的按钮
- } else {
- let type;
- try {
- // 判断文档类型
- type = detectType();
- console.log(type);
- } catch (e) {
- alert("未知/特殊文档类型,例如学术文献,暂不支持下载\n也可与作者反馈或联系:\nallenlv2690@gmail.com");
- return null;
- }
- // 添加 btn_3
- createPrintPageBtn(printPageBaidu);
- // ppt、pdf有部分文档是纯图片的
- if (type === "ppt") {
- transformButtons(true, true, "导出全部图片链接");
- }
- // 其他均为canvas图形
- else {
- transformButtons(true);
- // 禁用按钮
- document.querySelector(".save-doc-btn").disabled = true;
- document.querySelector(".save-html-btn").disabled = true;
- // 监听scroll,捕获canvas
- let storeCanvases_Baidu = () => {
- let selector = ".creader-canvas";
- let id_format = "creader-canvas-@";
- storeCanvases(selector, id_format, max_id);
- }
- window.onscroll = storeCanvases_Baidu;
- }
- }
- }
-
- function readAllDoc88() {
- // 获取“继续阅读”按钮
- let continue_btn = document.querySelector("#continueButton");
- // 增强按钮点击效果
- enhanceBtnClickReaction();
- // 如果存在“继续阅读”按钮
- if (continue_btn) {
- // 跳转到文末(等同于展开全文)
- let cur_page = document.querySelector("#pageNumInput");
- // 取得最大页码
- let page_max = cur_page.parentElement.textContent.replace(" / ", "");
- // 跳转到尾页
- jump2pageNo(cur_page, page_max, "keypress");
- // 返回顶部
- setTimeout(jump2pageNo(cur_page, "1", "keypress"), 1000);
- }
- // 否则启动按钮2
- else {
- transformButtons(true, false, "", true);
- // 显示btn_3
- document.querySelector(".save-html-btn").style.removeProperty("display");
- }
- }
-
- function readAllWK() {
- enhanceBtnClickReaction();
- let first_open = document.querySelector(".open-arrow");
- // 第一次展开文档
- if (first_open) {
- first_open.click();
- setTimeout(readAllWK, 1000);
- }
- // 非第一次展开文档
- else {
- let read_all = document.querySelector(".pagerwg-button");
- // 如果“继续阅读”按钮不存在或已经被隐藏
- if (!read_all || read_all.style.display === "none") {
- // 尝试关闭弹窗(如果存在)
- try {
- functiondocument.querySelector(".btn-cancel[id*=wui]").click();
- } catch (e) { console.log("继续阅读按钮引起的弹窗不存在,无需关闭"); }
- // 尝试移除底部无关元素
- let bottom_part = document.querySelector(".live-broadcast-pop-wrap");
- tryToRemoveElement(bottom_part.previousElementSibling);
- // 换为按钮2
- transformButtons(false, false, "", true);
- }
- // 否则文档尚未完全展开,继续点击按钮展开文档
- else {
- setTimeout(() => {
- read_all.click();
- clearPageWK();
- readAllWK();
- }, 1000);
- }
- }
- }
-
- /**
- * 存储非PPT文档的png图形链接,适用性:百度文库
- */
- function savePDFData() {
- let pic_urls = document.getElementsByClassName("reader-pic-item");
- let text_list = [];
- // 去掉前缀
- let reg_exp_1 = new RegExp(": ?url[(]");
- // 去掉后缀
- let reg_exp_2 = new RegExp("[)]; ?background-position");
-
- for (let i = 0; i < pic_urls.length; i++) {
- let whole_text = pic_urls[i].getAttribute("style");
- let de_pretext = whole_text.split(reg_exp_1)[1];
- let url = de_pretext.split(reg_exp_2)[0];
- text_list.push(url);
- }
-
- text_list[0] = text_list[0].replace(/"/g, "");
- let content = text_list.join("\n");
- // 启动下载
- createAndDownloadFile("urls.csv", content);
- }
-
- /**
- * 存储纯文本,适用性:百度文库
- */
- function saveDocData() {
- // 获取文本
- let text_elements = document.getElementsByClassName("reader-word-layer");
- let texts = [];
- for (let elem of text_elements) {
- texts.push(elem.textContent);
- }
- // 美化后导出文本
- let origin_content = texts.join("");
- let content = formatText(origin_content);
- createAndDownloadFile("纯文本文档.txt", content);
- }
-
- /**
- * 存储PPT图像链接,适用性:百度文库
- */
- function savePPTData() {
- let pic_elements = document.getElementsByClassName("ppt-image-wrap");
- let pic_urls = [];
-
- for (let elem of pic_elements) {
- let pic_obj = elem.children[0];
- let url = pic_obj.src;
- pic_urls.push(url);
- }
- let content = pic_urls.join("\n");
- // 启动下载
- createAndDownloadFile("urls.csv", content);
- }
-
- /**
- * 存储文字型表格,适用性:百度文库
- */
- function saveExcelData() {
- // 1. 拿到表格
- let table_pic = document.getElementsByClassName("reader-pic-item")[0];
- let url = table_pic.style.getPropertyValue("background-image");
- // 获取图片地址
- let pure_url = url.slice(5, -2);
-
- // 2. 拿到表格内文字信息
- let text_elems = document.getElementsByClassName("reader-word-layer");
- let text_list = [];
- for (let elem of text_elems) {
- text_list.push(elem.textContent);
- }
- let _text = text_list.join("\n");
- // 替换奇怪的空格
- let text = _text.replace(/ /g, " ");
-
- // 3. 合并至一个字符串,然后导出
- let head = "表格图形链接如下(复制到浏览器中打开):";
- let content = head + "\n\n" + pure_url + "\n\n" + text;
- createAndDownloadFile("图片地址和表格内容.txt", content);
- }
-
- /**
- * 对于文字和图形混合型的data只能存储其中的纯文字\
- * 适用性:百度文库
- */
- function saveDocAndPicData() {
- // 获取文本
- let text_elements = document.getElementsByClassName("reader-word-layer");
- let texts = [];
- for (let elem of text_elements) {
- texts.push(elem.textContent);
- }
- let origin_content = texts.join("");
- // 美化后导出文本
- let content = formatText2(origin_content);
- createAndDownloadFile("纯文本文档.txt", content);
- }
-
- /**
- * 存储纯文本到本地,适用性:百度文库
- */
- function saveTxtData() {
- let text_elements = document.getElementsByClassName("p-txt");
- let texts = [];
- for (let elem of text_elements) {
- texts.push(elem.textContent);
- }
- let content = texts.join("");
- createAndDownloadFile("纯文本文档.txt", content);
- }
-
- /**
- * 按文档类型,用对应方法储存数据到本地,适用性:百度文库
- * @returns 特殊情况下返回null,表示安全退出
- */
- function saveData() {
- let type = detectType();
- if (type === "ppt") {
- // ppt按类似于纯图文档的方法处理
- savePPTData();
- } else if (type === "safe_quit") {
- // 安全退出
- return null;
- } else {
- let data = {};
- }
- }
-
- /**
- * 下载全部图片链接,适用性:爱问共享资料、得力文库
- * @param {string} selector 图形元素的父级元素
- */
- function savePicUrls(selector) {
- let pages = document.querySelectorAll(selector);
- let pic_urls = [];
-
- for (let elem of pages) {
- let pic_obj = elem.children[0];
- let url = pic_obj.src;
- pic_urls.push(url);
- }
- let content = pic_urls.join("\n");
- // 启动下载
- createAndDownloadFile("urls.csv", content);
- }
-
- /**
- * 存储所有canvas图形为png到一个压缩包
- * @param {Array} node_list canvas元素列表
- * @param {String} title 文档标题
- */
- function saveCanvasesToZip(node_list, title) {
- // canvas元素转为png图像
- // 所有png合并为一个zip压缩包
- let zip = new JSZip();
- let n = node_list.length;
-
- for (let i = 0; i < n; i++) {
- let canvas = node_list[i];
- let data_base64 = canvas.toDataURL();
- let blob = atob(data_base64.split(",")[1]);
- zip.file(`page-${i+1}.png`, blob, { binary: true });
- }
-
- // 导出zip
- // promise.then(onCompleted, onRejected);
- zip.generateAsync({ type: "blob" }).then(function(content) {
- // see filesaver.js
- console.log(content);
- saveAs(content, `${title}.zip`);
- });
- }
-
- /**
- * 将canvas转为jpeg,然后导出pdf
- * @param {Array} node_list canvas元素列表
- * @param {String} title 文档标题
- * @param {Number} quality 图片质量,浮点数,范围 (0,1],默认值0.92
- */
- function saveCanvasesToPDF(node_list, title, quality = 0.92) {
- let first_canvas = node_list[0];
- // 如果style的长宽不存在,则直接用canvas的元素长宽
- let width_str, height_str;
- if (!first_canvas.style.width) {
- [width_str, height_str] = [first_canvas.style.width.replace(/(px)|(rem)|(em)/, ""), first_canvas.style.height.replace(/(px)|(rem)|(em)/, "")];
- } else {
- [width_str, height_str] = [first_canvas.width, first_canvas.height];
- }
- // jsPDF的第三个参数为format,当自定义时,参数为数字数组。
- let [width, height] = [parseFloat(width_str), parseFloat(height_str)];
- // 如果文档第一页的宽比长更大,则landscape,否则portrait
- let orientation = width > height ? 'l' : 'p';
- let pdf = new jsPDF(orientation, 'px', [height, width]);
-
- // 保存每一页文档到每一页pdf
- node_list.forEach(function(canvas, index) {
- pdf.addImage(canvas.toDataURL("image/jpeg", quality), 'JPEG', 0, 0, width, height);
- // 如果当前不是文档最后一页,则需要添加下一个空白页
- if (index !== node_list.length - 1) {
- pdf.addPage();
- }
- });
-
- // 导出文件
- pdf.save(`${title}.pdf`);
- }
-
-
- /**
- * 创建两个初始按钮:展开文档、存储文档
- * @returns
- */
- function create2btns() {
- // 创建脚本启动按钮1、2
- let btn_1 = document.createElement("button");
- let btn_2 = document.createElement("button");
-
- // 设定按钮1、2样式
- btn_1.setAttribute("class", "init-btn");
- btn_1.style.height = "25px";
- btn_1.style.width = "50%";
- btn_1.style.marginLeft = "25%";
- btn_1.style.border = "none";
- btn_1.style.backgroundColor = "blue";
- btn_1.style.color = "white";
- btn_1.style.fontWeight = "bold";
- btn_1.textContent = "展开文档";
- btn_1.style.zIndex = "99999";
-
- btn_2.setAttribute("class", "save-doc-btn");
- btn_2.style.height = "25px";
- btn_2.style.width = "50%";
- btn_2.style.marginLeft = "25%";
- btn_2.style.backgroundColor = "green";
- btn_2.style.border = "none";
- btn_2.style.display = "none";
- btn_2.style.color = "white";
- btn_2.style.fontWeight = "bold";
- btn_2.style.zIndex = "99999";
-
- // 添加按钮元素到页面
- let section = document.createElement("section");
- section.setAttribute("class", "btns_section");
- section.appendChild(btn_1);
- section.appendChild(btn_2);
- document.body.appendChild(section);
- // 返回元素引用
- return [btn_1, btn_2]
- }
-
- /*
- * 主函数部分
- */
-
- /**
- * 百度文库文档下载策略
- */
- function baiduWenku() {
- // 创建脚本启动按钮1、2
- let [btn_1, btn_2] = create2btns();
- btn_2.textContent = "导出全部图片";
-
- // 绑定主函数
- btn_1.onclick = () => {
- enhanceBtnClickReaction();
- readAll();
- };
- btn_2.onclick = () => {
- enhanceBtnClickReaction("btn_2");
- saveData();
- };
-
- // 解除打印限制
- allowPrint();
- }
-
- /**
- * 豆丁文档下载策略
- */
- function docin() {
- // 创建脚本启动按钮
- let [btn_1, btn_2] = create2btns();
- btn_2.textContent = "导出全部图片";
-
- // 隐藏底部工具栏
- document.querySelector("#j_select").click(); // 选择指针
- let tool_bar = document.querySelector(".reader_tools_bar_wrap.tools_bar_small.clear");
- tool_bar.style.display = "none";
-
- // 绑定主函数
- let getCanvasList = function() {
- // 获取全部canvas元素,用于传递canvas元素列表给 btn_2 和 btn_3
- let parent_node_list = document.querySelectorAll(".hkswf-content");
- let node_list = [];
- for (let node of parent_node_list) {
- node_list.push(node.firstElementChild);
- }
- return node_list;
- };
-
- let prepare = function() {
- // 获取canvas元素列表
- let node_list = getCanvasList();
- // 获取文档标题
- let title;
- if (document.querySelector("h1 [title=doc]")) {
- title = document.querySelector("h1 [title=doc]").nextElementSibling.textContent;
- } else if (document.querySelector(".doc_title")) {
- title = document.querySelector(".doc_title").textContent;
- } else {
- title = "文档";
- }
- // 根据页数决定图形质量
- let quality, page_num_str, page_num;
- try {
- page_num_str = document.querySelectorAll(".info_txt")[1].children[0].textContent;}
- catch {page_num_str = "26";}
- page_num = parseInt(page_num_str);
-
- if (page_num <= 25) {
- quality = 1.0;
- } else if (25 < page_num <= 50) {
- quality = 0.85;
- } else {
- quality = 0.7;
- }
-
- return [node_list, title, quality];
- }
-
- // 判断是否有canvas元素
- let detectCanvas = function() {
- let btn_2 = document.querySelector(".save-doc-btn");
- let haveCanvas = getCanvasList().length === 0 ? false : true;
- // 如果没有canvas元素,则认为文档页面由外链图片构成
- if (!haveCanvas) {
- // 调整按钮显示
- transformButtons(false, false, "导出全部图片链接", false);
- // 绑定主函数
- btn_2.onclick = function() {
- enhanceBtnClickReaction("btn_2");
- if (confirm("确定每页内容都加载完成了吗?")) {
- savePicUrls("[id*=img_]");
- }
- }
- } else {
- // 调整按钮显示
- transformButtons(true, false, "", false);
- // btn_2: 导出zip
- btn_2.onclick = function() {
- enhanceBtnClickReaction("btn_2");
- if (confirm("确定每页内容都加载完成了吗?")) {
- saveCanvasesToZip(...prepare());
- }
- };
- // btn_3: 导出pdf
- createPrintPageBtn(function() {
- if (confirm("确定每页内容都加载完成了吗?")) {
- saveCanvasesToPDF(...prepare());
- }
- });
- }
- }
- btn_1.textContent = "判断文档类型";
- btn_1.onclick = detectCanvas;
- }
-
- /**
- * 爱问共享资料文档下载策略
- */
- function ishare() {
- // 创建脚本启动按钮1、2
- let [btn_1, btn_2] = create2btns();
- btn_2.textContent = "导出全部图片链接";
-
- // 绑定主函数
- btn_1.onclick = readAlliShare;
- btn_2.onclick = function() { savePicUrls(".data-detail"); };
-
- // 移除底部下载条
- let detailfixed = document.getElementsByClassName("detail-fixed")[0];
- detailfixed.remove();
- }
-
- /**
- * 得力文库文档下载策略
- */
- function deliwenku() {
- // 创建脚本启动按钮1、2
- let [btn_1, btn_2] = create2btns();
- btn_2.textContent = "导出全部图片链接";
-
- // 绑定主函数
- btn_1.onclick = readAllDeliwenku;
- btn_2.onclick = function() { savePicUrls('.inner_page div'); };
-
- // 尝试关闭页面弹窗
- try { document.querySelector("div[title=点击关闭]").click(); } catch (e) { console.log(0); }
- // 解除打印限制
- allowPrint();
- }
-
- /**
- * 道客巴巴文档下载策略
- */
- function doc88() {
- // 创建脚本启动按钮1、2
- let [btn_1, btn_2] = create2btns();
- btn_2.textContent = "导出全部图片";
-
- // 绑定主函数
- let prepare = function() {
- // 获取canvas元素列表
- let node_list = document.querySelectorAll(".inner_page");
- // 获取文档标题
- let title;
- if (document.querySelector(".doctopic h1")) {
- title = document.querySelector(".doctopic h1").title;
- } else {
- title = "文档";
- }
- // 根据页数决定图形质量
- let quality, page_num_str, page_num;
- page_num_str = document.querySelector("#pageNumInput").parentElement.textContent.replace(" / ", "");
- page_num = parseInt(page_num_str);
-
- if (page_num <= 25) {
- quality = 1.0;
- } else if (25 < page_num <= 50) {
- quality = 0.85;
- } else {
- quality = 0.7;
- }
-
- return [node_list, title, quality];
- }
-
- // btn_1: 展开文档
- btn_1.onclick = readAllDoc88;
- // btn_2: 导出zip
- btn_2.onclick = function() {
- enhanceBtnClickReaction("btn_2");
- if (confirm("确定每页内容都加载完成了吗?")) {
- saveCanvasesToZip(...prepare());
- }
- };
- // btn_3: 导出pdf
- let btn_3 = createPrintPageBtn(function() {
- if (confirm("确定每页内容都加载完成了吗?")) {
- saveCanvasesToPDF(...prepare());
- }
- });
- btn_3.style.display = "none";
- }
-
- /**
- * wk文档下载策略
- */
- function baiduWenkuMobile() {
- // 创建初始按钮
- clearPageWK();
- let [btn_1, btn_2] = create2btns();
- btn_2.textContent = "打印页面到PDF";
- // 绑定主函数
- btn_1.onclick = readAllWK;
- btn_2.onclick = () => {
- enhanceBtnClickReaction("btn_2");
- printPageWK();
- }
- }
-
- /**
- * 主函数:识别网站,执行对应文档下载策略
- */
- function main() {
- let host = window.location.host;
- if (host === "wenku.baidu.com") {
- baiduWenku();
- } else if (host === "wk.baidu.com") {
- baiduWenkuMobile();
- } else if (host.includes("docin.com")) {
- docin();
- } else if (host === "ishare.iask.sina.com.cn") {
- ishare();
- } else if (host === "www.deliwenku.com") {
- deliwenku();
- } else if (host === "www.doc88.com") {
- doc88();
- } else {
- console.log("匹配到了无效网页");
- }
- }
-
- if (options["fast_mode"]) {
- main();
- } else {
- window.onload = main;
- }