CleanURLs (Clean URL Improved)

CleanURLs

Remove tracking parameters and redirect to original URL. This Userscript uses the URL Interface instread of RegEx.
目前为 2023-06-01 提交的版本。查看最新版本。
换行
// ==UserScript==
// @name        CleanURLs
// @namespace   i2p.schimon.cleanurl
// @description Remove tracking parameters and redirect to original URL. This Userscript uses the URL Interface instread of RegEx.
// @homepageURL https://greasyfork.org/en/scripts/465933-clean-url-improved
// @supportURL  https://greasyfork.org/en/scripts/465933-clean-url-improved/feedback
// @copyright   2023, Schimon Jehudah (http://schimon.i2p)
// @license     MIT; https://opensource.org/licenses/MIT
// @grant       none
// @run-at      document-end
// @match       *://*/*
// @version     23.06.01
// @icon        data:image/svg+xml;base64,PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIHZpZXdCb3g9IjAgMCAxMDAgMTAwIj48dGV4dCB5PSIuOWVtIiBmb250LXNpemU9IjkwIj7wn5qlPC90ZXh0Pjwvc3ZnPgo=
 
// ==/UserScript==
 
/*
 
Simple version of this Userscript
let url = new URL(location.href);
if (url.hash || url.search) {
  location.href = url.origin + url.pathname
};
 
*/
 
// Check whether HTML; otherwise, exit.
//if (!document.contentType == 'text/html')
if (document.doctype == null) return;
 
//let point = [];
const namespace = 'i2p.schimon.cleanurl';
 
// List of url parameters
const urls = [
  'ref',
  'source',
  'src',
  'url',
  'utm_source'];
 
// List of reserved parameters
const whitelist = [
  'art',                  // article
  'action',               // wiki
  'bill',                 // law
  'c',                    // cdn
  'category',             // id
  'code',                 // code
  'content',              // id
  'dark',                 // yorik.uncreated.net
  'date',                 // date
  'days',                 // wiki
  'district',             // house.mo.gov
  'exp_time',             // cdn
  'expires',              // cdn
  'ezimgfmt',             // cdn image processor
  'feedformat',           // wiki
  'fid',                  // mybb
  'file_host',            // cdn
  'filename',             // filename
  'for',                  // cdn
  'format',               // file type
  'guid',                 // guid
  'hash',                 // cdn
  'hidebots',             // wiki
  'hl',                   // language
  'id',                   // id
  'ie',                   // character encoding
  'ip',                   // ip address
  'item_class',           // greasyfork
  'item_id',              // greasyfork
  'jid',                  // jabber id (xmpp)
  'key',                  // cdn
  'limit',                // wiki
  'lang',                 // language
  'language',             // language
  'library',              // oujs
  'locale',               // locale
  'lr',                   // cdn
  'lra',                  // cdn
  'mobileaction',         // wiki
  'news_id',              // post
  'order',                // bugzilla
  'orderBy',              // oujs
  'orderDir',             // oujs
  'p',                    // search query / page number
  'page',                 // mybb
  'preferencesReturnUrl', // return url
  'product',              // bugzilla
  'q',                    // search query
  'query',                // search query
  'query_format',         // bugzilla
//'referer',              // signin <-- provided pathname contains login (log-in) or signin (sign-in)
  'resolution',           // bugzilla
  'return_to',            // signin
  's',                    // search query
  'search',               // search query
  'show_all_versions',    // greasyfork
  'sign',                 // cdn
  'signature',            // cdn
  'sort',                 // greasyfork
  'speed',                // cdn
  'start_time',           // media playback
  'state',                // cdn
  '__switch_theme',       // theme (theanarchistlibrary.org)
  'tag',                  // id
  'tid',                  // mybb
  'title',                // send (share) links and wiki
  'type',                 // file type
//'url',                  // url <-- not whitelisted nor blacklisted
  'utf8',                 // encoding
  'urlversion',           // wiki
  'v',                    // video
  'version',              // greasyfork
  //'_x_tr_sl', // translate online service
  //'_x_tr_tl=', // translate online service
  //'_x_tr_hl=', // translate online service
  //'_x_tr_pto', // translate online service
  //'_x_tr_hist', // translate online service
  'year'                  // year
  ];
 
// List of useless hash
const hash = [
  'back-url',
  'intcid',
  'niche-',
//'searchinput',
  'src'];
 
// List of useless parameters
const blacklist = [
  'ad',
  'ad_medium',
  'ad_name',
  'ad_pvid',
  'ad_sub',
  //'ad_tags',
  'advertising-id',
  //'aem_p4p_detail',
  'af',
  'aff',
  'aff_fcid',
  'aff_fsk',
  'aff_platform',
  'aff_trace_key',
  'affparams',
  'afSmartRedirect',
  //'aid',
  'algo_exp_id',
  'algo_pvid',
  'ar',
  //'ascsubtag',
  //'asc_contentid',
  'asgtbndr',
  'ats',
  //'b64e', // breaks yandex
  'bizType',
  //'block',
  'bta',
  'businessType',
  'campaign',
  'campaignId',
  //'__cf_chl_rt_tk',
  'cid',
  'ck',
  //'clickid',
  //'client_id',
  //'cm_ven',
  'content-id',
  'crid',
  'cst',
  'cts',
  'curPageLogUid',
  //'data', // breaks yandex
  //'dchild',
  //'dclid',
  'deals-widget',
  'dicbo',
  //'dt',
  'edd',
  'edm_click_module',
  //'ei',
  //'embed',
  '_encoding',
  //'etext', // breaks yandex
  'fbclid',
  'feature',
  'forced_click',
  //'fr',
  'frs',
  //'from', // breaks yandex
  '_ga',
  'ga_order',
  'ga_search_query',
  'ga_search_type',
  'ga_view_type',
  'gatewayAdapt',
  //'gclid',
  //'gclsrc',
  'gh_jid',
  'gps-id',
  //'gs_lcp',
  'gt',
  'guccounter',
  'hdtime',
  'ICID',
  'ico',
  'ig_rid',
  //'idzone',
  //'iflsig',
  //'irgwc',
  //'irpid',
  'itid',
  //'itok',
  //'katds_labels',
  //'keywords',
  'keyno',
  'l10n',
  'linkCode',
  'mc',
  'mid',
  'mp',
  'nats',
  'nci',
  'obOrigUrl',
  'optout',
  'oq',
  'organic_search_click',
  'pa',
  'Partner',
  'partner',
  'partner_id',
  'pcampaignid',
  'pd_rd_i',
  'pd_rd_r',
  'pd_rd_w',
  'pd_rd_wg',
  'pdp_npi',
  'pf_rd_i',
  'pf_rd_m',
  'pf_rd_p',
  'pf_rd_r',
  'pf_rd_s',
  'pf_rd_t',
  'pg',
  'PHPSESSID',
  'pk_campaign',
  'pdp_ext_f',
  'pkey',
  'platform',
  'plkey',
  'pqr',
  'pr',
  'pro',
  'prod',
  'promo',
  'promocode',
  'promoid',
  'psc',
  'psprogram',
  'pvid',
  'qid',
  //'r',
  'realDomain',
  'redirect',
  'ref',
  'ref_',
  'ref_src',
  'refcode',
  'referrer',
  'refinements',
  'reftag',
  'rowan_id1',
  'rowan_msg_id',
  //'sCh',
  'sclient',
  'scm',
  'scm_id',
  'scm-url',
  //'sd',
  'si',
  '___SID',
  '_src',
  'src_cmp',
  'src_player',
  'src_src',
  'shareId',
  'showVariations',
  'sid',
  //'site_id',
  'sk',
  'smid',
  'social_params',
  'source',
  'sourceId',
  'sp_csd',
  'spLa',
  'spm',
  'spreadType',
  //'sprefix',
  'sr',
  'src',
  'srcSns',
  'su',
  '_t',
  //'tag',
  'tcampaign',
  'td',
  'terminal_id',
  //'text',
  'th', // Sometimes restored after page load
  //'title',
  'tracelog',
  'traffic_id',
  'traffic_type',
  'tt',
  'uact',
  'ug_edm_item_id',
  //'utm1',
  //'utm2',
  //'utm3',
  //'utm4',
  //'utm5',
  //'utm6',
  //'utm7',
  //'utm8',
  //'utm9',
  'utm_campaign',
  'utm_content',
  'utm_medium',
  'utm_source',
  'utm_term',
  'uuid',
  //'utype',
  //'ve',
  //'ved',
  //'zone'
  ];
 
// URL Indexers
const paraIDX = [
  'algo_exp_id',
  'algo_pvid',
  'b64e',
  'cst',
  'cts',
  'data',
  'ei',
  //'etext',
  'from',
  'iflsig',
  'gbv',
  'gs_lcp',
  'hdtime',
  'keyno',
  'l10n',
  'mc',
  'oq',
  //'q',
  'sei',
  'sclient',
  'sign',
  'source',
  'state',
  //'text',
  'uact',
  'uuid',
  'ved'];
 
// Market Places 
const paraMKT = [
  '___SID',
  '_t',
  'ad_pvid',
  'af',
  'aff_fsk',
  'aff_platform',
  'aff_trace_key',
  'afSmartRedirect',
  'bizType',
  'businessType',
  'ck',
  'content-id',
  'crid',
  'curPageLogUid',
  'deals-widget',
  'edm_click_module',
  'gatewayAdapt',
  'gps-id',
  'keywords',
  'pd_rd_i',
  'pd_rd_r',
  'pd_rd_w',
  'pd_rd_wg',
  'pdp_npi',
  'pf_rd_i',
  'pf_rd_m',
  'pf_rd_p',
  'pf_rd_r',
  'pf_rd_s',
  'pf_rd_t',
  'platform',
  'pdp_ext_f',
  'ref_',
  'refinements',
  'rowan_id1',
  'rowan_msg_id',
  'scm',
  'scm_id',
  'scm-url',
  'shareId',
  //'showVariations',
  'sk',
  'smid',
  'social_params',
  'spLa',
  'spm',
  'spreadType',
  'sr',
  'srcSns',
  'terminal_id',
  'th', // Sometimes restored after page load
  'tracelog',
  'tt',
  'ug_edm_item_id'];
 
// IL
const paraIL = [
  'dicbo',
  'obOrigUrl'];
 
// General
const paraWWW = [
  'aff',
  'promo',
  'promoid',
  'ref',
  'utm_campaign',
  'utm_content',
  'utm_medium',
  'utm_source',
  'utm_term'];
 
// For URL of the Address bar
// Check and modify page address
// TODO Add bar and ask to clean address bar
(function modifyURL() {
 
  let
    check = [],
    url = new URL(location.href);
 
  // TODO turn into boolean function
  for (let i = 0; i < blacklist.length; i++) {
    if (url.searchParams.get(blacklist[i])) {
      check.push(blacklist[i]);
      url.searchParams.delete(blacklist[i]);
      //newURL = url.origin + url.pathname + url.search + url.hash;
    }
  }
 
  // TODO turn into boolean function
  for (let i = 0; i < hash.length; i++) {
    if (url.hash.startsWith('#' + hash[i])) {
      check.push(hash[i]);
      //newURL = url.origin + url.pathname + url.search;
    }
  }
 
  if (check.length > 0) {
    let newURL = url.origin + url.pathname + url.search;
    window.history.pushState(null, null, newURL);
    //location.href = newURL;
  }
 
})();
 
(function scanAllURLs() {
  for (let i = 0; i < document.links.length; i++) {
    let url = new URL(document.links[i].href);
    if (url.search) {
    //if (url.search || url.hash) {
      document.links[i].setAttribute('href-data', document.links[i].href);
    }
  }
})();
 
(function scanBadURLs() {
  for (let i = 0; i < document.links.length; i++) {
    // TODO callback, Mutation Observer, and Event Listener
    hash.forEach(j => cleanLink(document.links[i], j, 'hash'));
    blacklist.forEach(j => cleanLink(document.links[i], j, 'para'));
  }
})();
 
// TODO Add an Event Listener
function cleanLink(link, target, type) {
  let url = new URL(link.href);
  switch (type) {
    case 'hash':
      //console.log('hash ' + i)
      if (url.hash.startsWith('#' + target)) {
        //link.setAttribute('href-data', link.href);
        link.href = url.origin + url.pathname + url.search;
      }
      break;
    case 'para':
      //console.log('para ' + i)
      if (url.searchParams.get(target)) { 
        url.searchParams.delete(target);
        //link.setAttribute('href-data', link.href);
        link.href = url.origin + url.pathname + url.search;
      }
      break;
  }
 
  /*
  // EXTRA
  // For URL of hyperlinks
  for (const a of document.querySelectorAll('a')) {
    try{
      let url = new URL(a.href);
      for (let i = 0; i < blacklist.length; i++) {
        if (url.searchParams.get(blacklist[i])) {
          url.searchParams.delete(blacklist[i]);
        }
      }
      a.href = url;
    } catch (err) {
      //console.warn('Found no href for element: ' + a);
      //console.error(err);
    }
  } */
 
}
 
// TODO Hunt (for any) links within attributes using getAttributeNames()[i]
 
// Event Listener
// TODO Scan 'e.target.childNodes' until 'href-data' (link) is found
document.body.addEventListener("mouseover", function(e) { // mouseover works with keyboard too
  //if (e.target && e.target.nodeName == "A") {
  hrefData = e.target.getAttribute('href-data');
  //if (e.target && hrefData && !document.getElementById(namespace)) {
  if (e.target && hrefData && hrefData != document.getElementById('url-original')) {
    if (document.getElementById(namespace)) {
      document.getElementById(namespace).remove();
    }
    selectionItem = createButton(e.pageX, e.pageY, hrefData);
    hrefData = new URL(hrefData);
    selectionItem.append(purgeURL(hrefData));
    let types = ['whitelist', 'blacklist', 'original']
    for (let i = 0; i < types.length; i++) {
      let button = purgeURL(hrefData, types[i]);
      let exist;
      selectionItem.childNodes.forEach(
        node => {
          if (button.href == node.href) {
            exist = true;
          }
        }
      )
      if (!exist) {
        selectionItem.append(button);
      }
    }
    // Check for URLs
    for (let i = 0; i < urls.length; i++) {
      if (hrefData.searchParams.get(urls[i])) { // hrefData.includes('url=')
        urlParameter = hrefData.searchParams.get(urls[i]);
        try {
          urlParameter = new URL (urlParameter);
        } catch {
          if (urlParameter.includes('.')) {  // NOTE It is a guess
            try {
              urlParameter = new URL ('http:' + urlParameter);
            } catch {}
          }
        }
        if (typeof urlParameter == 'object') {
          newURLItem = extractURL(urlParameter);
          selectionItem.prepend(newURLItem);
        }
      }
    }
    // do not add element, if url has only whitelisted parameters and no potential url
    // add element, only if a potential url or non-whitelisted parameters are found
    urlTypes = ['url-extracted', 'url-original', 'url-purged'];
    for (let i = 0; i < urlTypes.length; i++) {
      if (selectionItem.querySelector(`#${urlTypes[i]}`)) {
        document.body.append(selectionItem);
        return;
      }
    }
  }
});
 
function createButton(x, y, url) {
  // create element
  let item = document.createElement(namespace);
  // set content
  item.id = namespace;
  // set position
  item.style.all = 'unset';
  item.style.position = 'absolute';
  item.style.left = x+5 + 'px';
  item.style.top = y-3 + 'px';
  // set appearance
  item.style.fontFamily = 'none'; // emoji
  item.style.background = '#333';
  item.style.borderRadius = '5%';
  item.style.padding = '3px';
  item.style.zIndex = 10000;
  //item.style.opacity = 0.7;
  item.style.filter = 'brightness(0.7)'
  // center character
  item.style.justifyContent = 'center';
  item.style.alignItems = 'center';
  item.style.display = 'flex';
  // disable selection marks
  item.style.userSelect = 'none';
  item.style.cursor = 'default';
  // set button behaviour
  item.onmouseover = () => {
    //item.style.opacity = 1;
    item.style.filter = 'unset';
  };
  item.onmouseleave = () => { // onmouseout
    // TODO Wait a few seconds
    item.remove();
  };
  return item;
}
 
function extractURL(url) {
  let item = document.createElement('a');
  item.textContent = '🔗'; // 🧧 🏷️ 🔖
  item.id = 'url-extracted';
  item.style.all = 'unset';
  item.style.outline = 'none';
  item.style.height = '15px';
  item.style.width = '15px';
  item.style.padding = '3px';
  item.style.margin = '3px';
  //item.style.fontSize = '0.9rem' // 90%
  item.style.lineHeight = 'normal'; // initial
  //item.style.height = 'fit-content';
  item.href = url;
  return item;
}
 
// TODO Use icons (with shapes) for cases when color is not optimal
function purgeURL(url, listType) {
  let itemTitle, itemId, resUrl;
  let item = document.createElement('a');
  item.style.all = 'unset';
  switch (listType) {
    case 'blacklist':
      itemColor = 'yellow';
      //itemTextContent = '🟡';
      itemTitle = 'Clean link'; // Purged URL
      itemId = 'url-purged';
      resUrl = hrefDataHandler(url, blacklist);
      break;
    case 'original': // TODO dbclick (double-click)
      itemColor = 'orangered';
      //itemTextContent = '🔴';
      itemTitle = 'Unsafe link'; // Original URL
      itemId = 'url-original';
      resUrl = url;
      item.style.cursor = `not-allowed`; // no-drop
      item.onmouseenter = () => {
        item.style.filter = `drop-shadow(2px 4px 6px ${itemColor})`;
      };
      item.onmouseout = () => {
        item.style.filter = 'unset';
      };
      break;
    case 'whitelist':
      itemColor = 'lawngreen';
      //itemTextContent = '🟢';
      itemTitle = 'Safe link'; // Link with whitelisted parameters
      itemId = 'url-known';
      resUrl = hrefDataHandler(url, whitelist);
      break;
    default:
      itemColor = 'antiquewhite';
      //itemTextContent = '⚪';
      itemTitle = 'Base link'; // Link without parameters
      itemId = 'url-base';
      resUrl = url.origin + url.pathname;
      break;
  }
  item.id = itemId;
  item.title = itemTitle;
  item.style.background = itemColor;
  //item.textContent = itemTextContent;
  item.style.borderRadius = '50%';
  item.style.outline = 'none';
  item.style.height = '15px';
  item.style.width = '15px';
  item.style.padding = '3px';
  item.style.margin = '3px';
  item.href = resUrl;
  return item;
}
 
function hrefDataHandler(url, listType) {
  url = new URL(url.href);
  url.searchParams.sort();
  switch (listType) {
    case whitelist:
      let newURL = new URL (url.origin + url.pathname);
      for (let i = 0; i < whitelist.length; i++) {
        if (url.searchParams.get(whitelist[i])) {
          newURL.searchParams.set(
            whitelist[i],
            url.searchParams.get(whitelist[i]) // catchedValue
          );
        }
      }
      url = newURL;
      break;
    case blacklist:
      for (let i = 0; i < blacklist.length; i++) {
        if (url.searchParams.get(blacklist[i])) {
          url.searchParams.delete(blacklist[i]);
        }
      }
      break;
  }
  return url;
}