☰

No Spam Results in Google Search

No more spam results in your google search!

  1. // ==UserScript==
  2. // @name         No Spam Results in Google Search
  3. // @namespace    http://tampermonkey.net/
  4. // @version      0.8
  5. // @description  No more spam results in your google search!
  6. // @author       CY Fung
  7. // @match        https://www.google.com/search?*
  8. // @match        https://www.google.*.*/search?*
  9. // @match        http://*/*
  10. // @match        https://*/*
  11. // @icon         https://greatest.deepsurf.us/rails/active_storage/representations/redirect/eyJfcmFpbHMiOnsibWVzc2FnZSI6IkJBaHBBeGswQVE9PSIsImV4cCI6bnVsbCwicHVyIjoiYmxvYl9pZCJ9fQ==--d840d20219ec465d39691a24251f7affa09443a5/eyJfcmFpbHMiOnsibWVzc2FnZSI6IkJBaDdCem9MWm05eWJXRjBTU0lJY0c1bkJqb0dSVlE2RkhKbGMybDZaVjkwYjE5c2FXMXBkRnNIYVFISWFRSEkiLCJleHAiOm51bGwsInB1ciI6InZhcmlhdGlvbiJ9fQ==--e4f27e4605e5535222e2c2f9dcbe36f4bd1deb29/nospam.png
  12. // @grant       GM.setValue
  13. // @grant       GM.getValue
  14. // @grant       GM.deleteValue
  15. // @grant       GM_getTab
  16. // @grant       GM_saveTab
  17. // @grant       GM_registerMenuCommand
  18. // @license MIT
  19. // @noframes
  20.  
  21. // ==/UserScript==
  22.  
  23. 'use strict';
  24.  
  25. const cssTxt = `
  26. a[href^="http"].gs-spam-link-disable, a[href^="http"].gs-spam-link-disable>h3 {
  27.     text-decoration: initial;
  28.     color: #d88a8a;
  29.     background: #605766;
  30.     cursor: no-drop;
  31. }
  32. `;
  33.  
  34. (function () {
  35.     'use strict';
  36.  
  37.     const [window, document, hostname, location, history] = new Function('return [window, document, location.hostname, location, history];')(); // real window & document object
  38.     let isGoogleSearchPage = hostname.includes('.google')?/\w+\.(google|googleusercontent)\.com(\.\w{0,2}){0,2}/.test(hostname):false;
  39.  
  40.     function getOrigin(url){
  41.         let nt=null;
  42.         url = `${url}`;
  43.         try{
  44.             nt=new URL(url);
  45.         }catch(e){
  46.             return url;
  47.         }
  48.         return `${nt.origin}`;
  49.     }
  50.  
  51.     function purify(url){
  52.         let nt=null;
  53.         url = `${url}`;
  54.         try{
  55.             nt=new URL(url);
  56.         }catch(e){
  57.             return url;
  58.         }
  59.         return `${nt.origin}${nt.pathname}`;
  60.     }
  61.     function elementCount(document){
  62.         if(!document) return null;
  63.         try{
  64.             return document.all.length;
  65.         }catch(e){
  66.             return document.getElementsByTagName("*").length;
  67.         }
  68.     }
  69.     function consoleJSON(object){
  70.         try{
  71.             console.table(JSON.stringify(object,null,2))
  72.         }catch(e){
  73.             console.log('consoleJSON is not supported');
  74.         }
  75.     }
  76.     function getTopDomain(str){
  77.         let m = /((?<![^\/]\/)\b[a-zA-Z0-9][a-zA-Z0-9-]{1,61}[a-zA-Z0-9]\.\b[a-zA-Z]{2,3}(?:\.\b[a-zA-Z]{2})?)(?:$|\/)/.exec(str);
  78.         if(m) return m[1];
  79.         return str;
  80.     }
  81.     function checkSpam(tabObject){
  82.  
  83.         if( tabObject.hitAt>1 && Date.now()- tabObject.hitAt<30*1000){
  84.             let filteredTimeline = tabObject.timeline.filter(entry=>entry.timeAt>=tabObject.hitAt);
  85.                 consoleJSON(filteredTimeline);
  86.             let [entryA, entryB] = [filteredTimeline[0],filteredTimeline[1]]
  87.             if(filteredTimeline.length<10 && filteredTimeline.length>=2 && entryA.type=='click' && entryB.type=='start'){
  88.  
  89.                 let networkTime = entryB.timeAt - entryA.timeAt;
  90.                 let [entryZ, entryY] = [filteredTimeline[filteredTimeline.length-1],filteredTimeline[filteredTimeline.length-2]]
  91.  
  92.                 function noPageContentGenerated(){
  93.                     // not reliable.
  94.                     return filteredTimeline.filter(entry=>entry.type=='complete').every(entry=>entry.elementCount<20);
  95.                 }
  96.  
  97.                 function noRenderOfActualPage(){
  98.                     // check whether there is any rendering of actual page
  99.                     let renderOK= true;
  100.                     filteredTimeline.reduce((a,b)=>{
  101.                         if(a && b && a.type=='start' && b.type=='complete' && a.pageHref === b.pageHref && b.timeAt-a.timeAt>5000){
  102.                             // assume >5000 rendering time is not spam page
  103.                             renderOK = false;
  104.                         }
  105.                         return b;
  106.                     },null);
  107.                     return renderOK;
  108.                 }
  109.  
  110.  
  111.                 function jsHistCounter(){
  112.  
  113.                     if(entryA.type=='click' && entryB.type=='start' && entryA.histCount >= entryB.histCount) entryA.histCount = entryB.histCount-1;
  114.  
  115.                     let jsHistCountInOrder = true;
  116.                     let endHist = -1;
  117.                     filteredTimeline.map(entry=>entry.histCount).reduce((a,b)=>{
  118.                         if(a>-1 && b>-1 && a<=b){
  119.                             endHist = b
  120.                         }else{
  121.                             jsHistCountInOrder=false;
  122.                         }
  123.                         return b;
  124.                     },0);
  125.  
  126.                     if(jsHistCountInOrder){
  127.                         return endHist - entryA.histCount;
  128.                     }
  129.  
  130.                     return 0;
  131.  
  132.                 }
  133.  
  134.  
  135.                 function countDomains(filteredTimeline){
  136.                     let domains = {}
  137.                     for(const entry of filteredTimeline){
  138.                         if(entry.type==='click') continue;
  139.                         let domain = getTopDomain(new URL(entry.pageHref).hostname)
  140.                         domains[domain]=(domains[domain]||0)+1
  141.                     }
  142.                     domains._arr=Object.keys(domains)
  143.                     domains._count=domains._arr.length
  144.                     return domains;
  145.                 }
  146.  
  147.  
  148.                 let redirectFlag = false;
  149.  
  150.                 if(entryA.type=='click' && entryB.type=='start' && getTopDomain(entryA.clickedHref) !== getTopDomain(entryB.pageHref) ){
  151.                     // server side redirect
  152.                     redirectFlag = true;
  153.  
  154.                 }else if(entryZ.type=='start' && (entryY.type=='complete' || entryY.type=='start')){
  155.                     let [domainZ, domainY] =[ getTopDomain(new URL(entryZ.pageHref).hostname), getTopDomain(new URL(entryY.pageHref).hostname)]
  156.                     if(domainZ!==domainY && entryZ.timeAt-entryY.timeAt<=1.25*networkTime && entryZ.timeAt-entryB.timeAt<=3*networkTime){
  157.                         redirectFlag = true;
  158.                     }
  159.                 }
  160.  
  161.                 consoleJSON(filteredTimeline);
  162.  
  163.                 function startCount(){
  164.                     return filteredTimeline.filter(entry=>entry.type=='start').length
  165.                 }
  166.  
  167.                 if(redirectFlag || (noPageContentGenerated() && noRenderOfActualPage() && startCount()>=2)){
  168.  
  169.                     let jhc = jsHistCounter();
  170.                     if(jhc>0){
  171.  
  172.                         // in case redirect on the same domain
  173.                         // it is allowed
  174.                         let domains = countDomains(filteredTimeline)
  175.                         if(domains._count===1) return;
  176.  
  177.                         return {filteredTimeline, jhc};
  178.                     }
  179.  
  180.                 }
  181.             }
  182.         }
  183.     }
  184.  
  185.     GM_getTab(tabObject=>{
  186.  
  187.         if(!tabObject.timeline)tabObject.timeline=[];
  188.  
  189.         if(isGoogleSearchPage){
  190.             if(tabObject.hitAt<0){
  191.                 history.pushState({},null);
  192.             }
  193.             tabObject.timeline.length=0;
  194.             tabObject.hitAt=0;
  195.         }else if(!tabObject.hitAt){
  196.             return;
  197.         }else if(Date.now()-tabObject.hitAt>30*1000){
  198.             tabObject.timeline.length=0;
  199.             tabObject.hitAt=0;
  200.             GM_saveTab(tabObject);
  201.             return;
  202.         }
  203.  
  204.         // either isGoogleSearchPage or hitTime < 30s
  205.  
  206.         tabObject.timeline.push({type:'start', pageHref:purify(location.href), timeAt:Date.now() , histCount: history.length});
  207.  
  208.         let resCheckSpam = checkSpam(tabObject);
  209.         if(resCheckSpam){
  210.             let {filteredTimeline, jhc}=resCheckSpam;
  211.             tabObject.hitAt=-1;
  212.             GM_saveTab(tabObject);
  213.  
  214.             let clickEntry = filteredTimeline[0]
  215.             // let entryZ = filteredTimeline[filteredTimeline.length-1];
  216.  
  217.             ;(async ()=>{
  218.                 try{
  219.                     let spamDomains = (await GM.getValue('spamDomains', "")).split(',');
  220.  
  221.                     // Note awaiting the set -- required so the next get sees this set.
  222.  
  223.                     await GM.setValue('spamDomains', spamDomains+','+ new URL(clickEntry.clickedHref).origin);
  224.  
  225.                 }catch(e){}
  226.             })();
  227.  
  228.             alert(`Spam Website Detected.\n - Spam Domain: "${getOrigin(clickEntry.clickedHref)}"\n - Go back: "${getOrigin(clickEntry.pageHref)}"`);
  229.             history.go(-jhc);
  230.         }else{
  231.             GM_saveTab(tabObject);
  232.         }
  233.  
  234.  
  235.  
  236.         function addStyle (styleText) {
  237.             const styleNode = document.createElement('style');
  238.             styleNode.type = 'text/css';
  239.             styleNode.textContent = styleText;
  240.             document.documentElement.appendChild(styleNode);
  241.             return styleNode;
  242.         }
  243.  
  244.  
  245.         function onReady(){
  246.  
  247.             let emc = elementCount(document);
  248.             if(emc<40){ // including html, body, script, div, ... //7 16 33
  249.                 emc-=document.querySelectorAll('script, style, meta, title, head, link, noscript').length; //remove count for non-visual elements //6 12
  250.             }
  251.             tabObject.timeline.push({type:'complete', pageHref:purify(location.href), timeAt:Date.now(), elementCount: emc , histCount: history.length });
  252.             GM_saveTab(tabObject);
  253.  
  254.             let hElems = null
  255.             if(isGoogleSearchPage){
  256.                 hElems = [...document.querySelectorAll('a[href^="http"]')].filter(elm=>!/^https?\:\/\/\w+\.(google|googleusercontent)\.com(\.\w{0,2}){0,2}\//.test(elm.href))
  257.                 for(const hElem of hElems){
  258.                     hElem.addEventListener('click',function(){
  259.                         tabObject.hitAt = Date.now();
  260.                         tabObject.timeline.push({type:'click', clickedHref:purify(hElem.href), pageHref:purify(location.href), timeAt:tabObject.hitAt , histCount: history.length});
  261.                         GM_saveTab(tabObject);
  262.                     })
  263.                 }
  264.  
  265.                 ;(async () => {
  266.  
  267.                     let spamDomains = (await GM.getValue('spamDomains', "")).split(',');
  268.                     console.log(spamDomains)
  269.                     let spamLinks = hElems.filter( elm=> spamDomains.includes(new URL(elm.href).origin) )
  270.  
  271.                     if(spamLinks.length>0){
  272.  
  273.                         let noClickFunc=function(evt){if(evt){evt.returnValue = false; evt.preventDefault();} return false;}
  274.                         for(const spamLink of spamLinks) {
  275.                             spamLink.classList.add('gs-spam-link-disable');
  276.                             spamLink.onclick=noClickFunc
  277.                         }
  278.                         addStyle (cssTxt);
  279.  
  280.                     }
  281.  
  282.                 })();
  283.  
  284.                 async function clearSpamDomainList(){
  285.                     await GM.deleteValue('spamDomains');
  286.                     for(const spamLink of hElems) {
  287.                         spamLink.classList.remove('gs-spam-link-disable');
  288.                         spamLink.onclick=null
  289.                     }
  290.                 }
  291.                 GM_registerMenuCommand('Clear Spam Domain List', clearSpamDomainList);
  292.  
  293.             }else{
  294.  
  295.                 function fx(){
  296.                     document.removeEventListener('mousedown',fx,true);
  297.                     document.removeEventListener('keydown',fx,true);
  298.                     tabObject.timeline.length=0;
  299.                     tabObject.hitAt=0;
  300.                     GM_saveTab(tabObject);
  301.                 }
  302.  
  303.                 document.addEventListener('mousedown',fx,true);
  304.                 document.addEventListener('keydown',fx,true);
  305.             }
  306.  
  307.         }
  308.  
  309.         ;(function uiMain(){
  310.             if(document.documentElement == null) return window.requestAnimationFrame(uiMain)
  311.             if (document.readyState !== 'loading') {
  312.                 onReady();
  313.             } else {
  314.                 document.addEventListener('DOMContentLoaded', onReady);
  315.             }
  316.         })();
  317.  
  318.     })
  319. })();