☰

Bachngocsach Vip scraper

Scrape novel content from bachngocsach.net.vn. Ctrl+Alt+S to start, Ctrl+Alt+C to stop.

  1. // ==UserScript==
  2. // @name        Bachngocsach Vip scraper
  3. // @name:en     Bachngocsach Vip scraper
  4. // @name:vi     Tải truyện vip bachngocsach
  5. // @namespace   Violentmonkey Scripts
  6. // @match       *://bachngocsach.info/*
  7. // @grant       GM_setValue
  8. // @grant       GM_getValue
  9. // @grant       GM_deleteValue
  10. // @grant       GM_listValues
  11. // @version     1.1.0
  12. // @author      Tác giả = tác = làm, giả là giả (dối) = làm giả
  13. // @description Scrape novel content from bachngocsach.net.vn. Ctrl+Alt+S to start, Ctrl+Alt+C to stop.
  14. // @description:vi Tải truyện từ bachngocsach.net.vn. Ctrl+Alt+S để bắt đầu, Ctrl+Alt+C để ngừng
  15. // @description:en Scrape novel content from bachngocsach.net.vn. Ctrl+Alt+S to start, Ctrl+Alt+C to stop
  16. // @license     MIT2
  17. // ==/UserScript==
  18. const startKey={key:'KeyS', ctrlKey:true, altKey: true, shiftKey:false };
  19. const stopKey= {key:'KeyC', ctrlKey:true, altKey: true, shiftKey:false };
  20.  
  21. const useEvent=false;
  22.  
  23. const sleep = (ms) => new Promise(rs => setTimeout(rs, ms));
  24.  
  25. const pressKey = (key) => window.dispatchEvent(new KeyboardEvent('keydown', { key: key, code: key, bubbles: true }));
  26.  
  27. function addUrlChangeEvent() {
  28.   const urlchangeEvent= new Event('urlchange');
  29.   history._pushState=history.pushState;
  30.   history.pushState=(...args)=>{
  31.     history._pushState(...args);
  32.     if(!!window.onurlchange && typeof window.onurlchange =='function') window.onurlchange(); else
  33.     window.dispatchEvent(urlchangeEvent);
  34.   }
  35.  
  36.   window.addEventListener('popstate',()=>{
  37.     if(!!window.onurlchange && typeof window.onurlchange =='function') window.onurlchange(); else
  38.     window.dispatchEvent(urlchangeEvent);
  39.   })
  40. }
  41.  
  42. function reEnableConsoleLog(c) {
  43.   switch (c) {
  44.     case 1: console.log = console.dir; break;
  45.     case 2: console.log = console.info; break;
  46.     case 3: console.log = console.debug; break;
  47.     case 4: console.log = console.warn; break;
  48.     default: {
  49.       const iF = document.createElement('iframe');
  50.       document.body.appendChild(iF);
  51.       iF.style.display = 'none';
  52.       window.console.log = iF.contentWindow.console.log;
  53.     }
  54.   }
  55. }
  56.  
  57. function getStyles(useRegex=true) {
  58.   const style={};
  59.   const styleStr=document.querySelector('style.dynamic-styles').textContent;
  60.   if (!styleStr) return style;
  61.   if(useRegex) {
  62.     const reg=/\.?([0-9a-zA-Z]+?){order:([0-9]+?)}/g
  63.     styleStr.matchAll(reg).forEach(m=>style[m[1]]=parseInt(m[2]));
  64.   } else {
  65.     styleStr.split('}').forEach(m=>{
  66.       if(m=='') return;
  67.       let s=m.split('{order:');
  68.       if(s[0].startsWith('.')) style[s[0].slice(1)]=parseInt(s[1]); else style[s[0]]=parseInt(s[1]);
  69.     })
  70.   }
  71.   return style;
  72. }
  73.  
  74. function chapterContentByTreeWalker(el = document.body) {  //dang chay sai o day
  75.   const textList = [];
  76.   textList.toString = () => { return textList.reduce((s, n) => s += n.nodeValue, '') }
  77.   const treeWalker = document.createTreeWalker(el, NodeFilter.SHOW_TEXT, (node) => {
  78.     if (['META', 'SCRIPT', 'NOSCRIPT', 'STYLE', 'AREA', 'BASE', 'CANVAS', 'CODE', 'EMBED', 'LINK', 'MAP', 'PARAM', 'SOURCE', 'VIDEO', 'IMG', 'PICTURE', 'INPUT', 'TEXTAREA'].includes(node.parentNode?.tagName))
  79.       return NodeFilter.FILTER_REJECT;
  80.     return NodeFilter.FILTER_ACCEPT;
  81.   });
  82.  
  83.   let node;
  84.   while (node = treeWalker.nextNode())
  85.     textList.push(node);
  86.   return textList.toString();
  87. }
  88.  
  89. function chapterContentByStyle() {
  90.   console.log('Get chapter content');
  91.   const result={};
  92.   let t= document.querySelectorAll('main>div>.container>div>.line-clamp-1');
  93.   result.storyName=t[0].textContent;
  94.   result.chapterName=t[1].textContent;
  95.   result.chapterNumber=parseInt(location.href.match(/.\/chuong-(\d+)/)[1]);
  96.  
  97.   const badText = document.querySelector('div.published-content');
  98.   const goodText = badText?.previousElementSibling?.innerText||'';
  99.   if (!badText) {result.chapterContent=goodText; return result;}
  100.  
  101.   let badLines = Array(badText.children.length);
  102.   let style = getStyles();
  103.  
  104.   for (const eights of badText.children) {
  105.     let className = eights.className.toLowerCase();
  106.     if (!style[className]) style[className] = parseInt(getComputedStyle(eights).order);
  107.     let badLine = Array(eights.children.length);
  108.     for (const sixes of eights.children) {
  109.       let tagName = sixes.tagName.toLowerCase();
  110.       if (!style[tagName]) style[tagName] = parseInt(getComputedStyle(sixes).order);
  111.       badLine[style[tagName]] = sixes.textContent;
  112.     }
  113.     badLines[style[className]] = badLine.join('');
  114.   }
  115.   result.chapterContent = (goodText.trim() + badLines.join('\n\n')).replaceAll(/\n{3,}/g, '\n\n').replaceAll('·', '');
  116.   return result;
  117. }
  118.  
  119. const getChapterContent=chapterContentByStyle;
  120. // const getChapterContent=chapterContentByTreeWalker;
  121.  
  122. function startDownload(url) {
  123.   console.log('Start downloading');
  124.   GM_setValue('downloading', true);
  125.   if (/^https:\/\/bachngocsach\.net\.vn\/truyen\/[a-z\-\d]+\/?$/.test(window.location.href)) window.location.assign(window.location.href + '/chuong-1');
  126.   else window.location.reload();
  127. }
  128.  
  129. async function stopDownload() {
  130.   console.log('Stop downloading');
  131.   let storyName = GM_getValue('storyName');
  132.   GM_deleteValue('storyName');
  133.   GM_deleteValue('downloading');
  134.   let chapters = GM_listValues();
  135.   chapters.sort((a, b) => { parseInt(a) - parseInt(b) });
  136.   let content = await Promise.all(chapters.map(chapter => GM_getValue(`${chapter}`)));
  137.   chapters.forEach(ch => GM_deleteValue(ch));
  138.   content = content.join('\n\n').replaceAll(/\n{1,1}/g, '\n\n').replaceAll(/\n{3,}/g, '\n\n');
  139.  
  140.   let download = document.createElement('a');
  141.   download.href = 'data:attachment/text,' + encodeURI(content);
  142.   download.target = '_blank';
  143.   download.download = storyName + `(c${chapters[0]}-c${chapters.at(-1)})` + '.txt';
  144.   download.click();
  145.   return;
  146. }
  147.  
  148. function nextChapter() {
  149.   console.log('Go to next chapter');
  150.   if(Math.random() > .4) pressKey('ArrowRight'); //ArrowRight
  151.   else {
  152.     const t=document.querySelector('.container>div:nth-last-child(2)>div:first-child>a:last-of-type');
  153.     t.scrollIntoView({behavior:'smooth'})
  154.     t.click();
  155.   }
  156. }
  157.  
  158. function isLastChapter() {
  159.   console.log('Is last chapter?');
  160.   return document.querySelector('.container>div:nth-last-child(2)>div:first-child>a:last-of-type').href.endsWith('#');
  161. }
  162.  
  163. async function download() {
  164.   console.log('Downloading...');
  165.   scrollTo({left:0,top:10000, behavior:'smooth'});
  166.   await sleep(500+ Math.random()*600); //350 is necessary time the chapter content being loaded
  167.   const chapter=getChapterContent();
  168.   GM_setValue('storyName',chapter.storyName);
  169.   GM_setValue(chapter.chapterNumber,chapter.chapterName+'\n\n'+chapter.chapterContent);
  170.   scrollTo({left:0,top:0, behavior:'smooth'});
  171.   await sleep(100+ Math.random()*300);
  172.   if (isLastChapter()) await stopDownload();
  173.   else nextChapter();
  174. }
  175.  
  176. (async function(){
  177.   if(window!==window.top) return;
  178.  
  179.   if (useEvent) {
  180.     addUrlChangeEvent();
  181.     //await sleep(1000);
  182.     let txt=getChapterContent();
  183.     console.log(txt.chapterContent);
  184.  
  185.     window.addEventListener('urlchange',async (e)=>{
  186.       if (GM_getValue('downloading', undefined)) await download();
  187.       else {
  188.         console.log(getChapterContent().chapterContent); }
  189.     })
  190.   } else {
  191.     let oldURL='';
  192.     const observer= new MutationObserver(async (mList)=>{
  193.       mList.forEach(async (m)=>{
  194.         if (m.target.className?.includes('published-content')&& window.location.href!=oldURL) {
  195.             oldURL=window.location.href;
  196.             if (GM_getValue('downloading', false)) await download();
  197.             else console.log(getChapterContent().chapterContent);
  198.         }
  199.       });
  200.     });
  201.     observer.observe(document.querySelector('body'), { childList: true, subtree:true });
  202.   }
  203.  
  204.   window.addEventListener('keydown',async(e)=>{
  205.     if (e.ctrlKey==startKey.ctrlKey && e.altKey==startKey.altKey && e.shiftKey==startKey.shiftKey && (e.key==startKey.key||e.code==startKey.key)) startDownload(location.href);
  206.     if (e.ctrlKey==stopKey.ctrlKey && e.altKey==stopKey.altKey && e.shiftKey==stopKey.shiftKey && (e.key==stopKey.key||e.code==stopKey.key)) await stopDownload();
  207.   })
  208. })();