SA_Internal_Ads_Only.user.js

// ==UserScript==
// @name        SA Internal Ads Only
// @namespace   nyo.dk
// @description Get rid of Something Awful Forums user ads that link outside the forum
// @include     /forums.somethingawful.com/
// @version     1
// @grant       GM_getValue
// @grant       GM_setValue
// @grant       GM_xmlhttpRequest
// @grant       GM_log
// ==/UserScript==

var maxCacheAge = 60*60*24*1000; // one day

function GetDomainFromUrl(url) {
  var a = document.createElement("a");
  a.href = url;
  return a.hostname;
}

function GetResponseDocument(rsp) {
  if (rsp.responseXML)
    return rsp.responseXML;
  return (new DOMParser()).parseFromString(rsp.responseText, "text/html");
}

function IsAdGood(href) {
  if (href.match(/^https?:\/\/forums.somethingawful.com/)) {
    return true;
  } else {
    return false;
  }
}

function CheckExtraLinks(extraLinks) {
  // ad quality checker modules
  var checkers = [
    {
      name: "bitly",
      handles: function(domain) {
        return !!domain.match(/^(bit\.ly|j\.mp)/);
      },
      check: function(url, continuation) {
        GM_xmlhttpRequest({
          method: "GET",
          url: url + "+",
          onload: function(rsp) {
            var rspDoc = GetResponseDocument(rsp);
            if (rspDoc) {
              var realUrl = rspDoc.getElementById("bitmark_long_url");
              if (realUrl) {
                continuation(IsAdGood(realUrl.href), realUrl.href);
                return;
              }
            }
            continuation(false, null);
          },
          onerror: function(rsp) {
            continuation(false, null);
          },
        });
      }
    }
  ];
  
  // urls to check quality of
  var urls = [];
  // iteration index for continuations
  var urlsIndex = 0;
  
  // when all urls are quality-checked
  function Finish() {
    // extra ads to keep
    var filteredAds = [];
    // check each original ad for having good/bad url
    extraLinks.forEach(function(ad) {
      if (urls.indexOf(ad.href) >= 0) {
        if (urls[ad.href]) {
          ad.originalHref = ad.href;
          ad.href = urls[ad.href];
        }
        // add the good ones to the list
        filteredAds.push(ad);
      }
    });
    // and append those to the cache
    if (filteredAds.length > 0) {
      GM_log("Finished checking external ad links, adding " + filteredAds.length + " more to cache");
      var adcache = JSON.parse(GM_getValue("adcache", "[]"));
      adcache = adcache.concat(filteredAds);
      GM_setValue("adcache", JSON.stringify(adcache));
    }
  }
  
  // url quality check step function
  function CheckNextExtraLink() {
    // check end of iteration
    if (urls.length <= urlsIndex) {
      return Finish();
    }
    
    var url = urls[urlsIndex];
    var checker = checkers.find(function(c) { return c.handles(GetDomainFromUrl(url)); });
    
    // continuation for quality check module
    function WhenChecked(checkResult, realUrl) {
      if (checkResult) {
        // good url: keep it, and store the wrapped url
        urls[urls[urlsIndex]] = realUrl;
        urlsIndex++;
      } else {
        // bad url: remove it from the array
        urls.splice(urlsIndex, 1);
      }
      CheckNextExtraLink();
    }
    GM_log("Check external ad link: " + url)
    if (checker) {
      // has a checker module for domain, call it
      checker.check(url, WhenChecked);
    } else {
      // no checker module, assume bad
      WhenChecked(false);
    }
  }
  
  // get unique urls from list of ads
  extraLinks.forEach(function (ad) {
    if (urls.indexOf(ad.href) < 0) {
      urls.push(ad.href);
    }
  });
  
  // begin check
  CheckNextExtraLink();
}

function CacheAdListFromDocument(adlistDoc) {
  // storage for found good ads
  var ads = [];
  // storage for bad ads that need an extra check
  var doublecheckLinks = [];
  
  // get adlist container
  var adlist = adlistDoc.getElementById("adlist");
  if (!adlist) {
    return null;
  }
  
  // iterate through it
  var curAd = adlist.firstElementChild;
  while (curAd) {
    var theLink = curAd.firstElementChild;
    var theImage = theLink.firstElementChild;
    if (IsAdGood(theLink.href)) {
      ads.push({href: theLink.href, src: theImage.src, hrefDomain: theLink.domain});
    } else {
      doublecheckLinks.push({href: theLink.href, src: theImage.src, hrefDomain: theLink.hostname});
    }
    curAd = curAd.nextElementSibling;
  }

  // update cache
  GM_setValue("adcache", JSON.stringify(ads));
  GM_setValue("adcache_updated", Date.now());
  GM_log("Loaded "+ads.length+" ads and stored in cache");
  
  // kick off check of bad-unknown ads
  CheckExtraLinks(doublecheckLinks);

  return ads;
}

function GetAdList(continuation) {
  // check if the cache exists and is up-to-date
  var cacheUpdated = GM_getValue("adcache_updated");
  var now = Date.now();
  var cache;
  if (cacheUpdated && (now - cacheUpdated) < maxCacheAge) {
    // cache good, load and use it
    cache = JSON.parse(GM_getValue("adcache"));
    return continuation(cache);
  }
  
  // otherwise kick off a cache update
  GM_log("Reloading SA ad list");
  var xhr = GM_xmlhttpRequest({
    method: "GET",
    url: window.location.protocol + "//forums.somethingawful.com/adlist.php",
    onload: function (rsp) {
      var adlistDoc = GetResponseDocument(rsp);
      if (adlistDoc) {
        continuation(CacheAdListFromDocument(adlistDoc));
      } else {
        continuation(null);
      }
    },
    onerror: function (rsp) {
      GM_log("Failed loading ad list");
      continuation(null);
    },
  });
}

function FadeOutAdlistBads() {
  // arguably this should also check the ad cache for ads with wrapped links that are still good
  GM_log("Fading out on adlist");
  var adlist = document.getElementById("adlist");
  var ad = adlist.firstElementChild;
  while (ad) {
    var theLink = ad.firstElementChild;
    if (!IsAdGood(theLink.href)) {
      ad.style.opacity = 0.2;
    }
    ad = ad.nextElementSibling;
  }
  
  // also, since we have the full ad list anyway, update the cache
  CacheAdListFromDocument(document);
}

function ReplaceAd() {
  // find ad container
  var containerDiv = document.getElementById("ad_banner_user");
  if (!containerDiv) {
    // not found, check if this is the adlist page instead
    if (document.getElementById("adlist")) {
      FadeOutAdlistBads();
    }
    return;
  }
  
  // check if the ad needs replacing
  var theLink = containerDiv.firstElementChild;
  var theImage = theLink.firstElementChild;
  if (!IsAdGood(theLink.href)) {
    // get list of possible replacements
    GetAdList(function(availableAds) {
      if (!availableAds || availableAds.length == 0) {
        // no replacement, don't replace
        return;
      }
      // pick a replacement
      var i = Math.floor(Math.random() * availableAds.length);
      var newAd = availableAds[i];
      // replace link and image
      theLink.href = newAd.href;
      theImage.src = newAd.src;
      // and add our signature
      containerDiv.appendChild(document.createTextNode(" | Ad replaced by userscript"));
      // if the ad had a wrapped/shortened link, also add the original one
      if (newAd.originalHref) {
        var orgLink = document.createElement("a");
        orgLink.href = newAd.originalHref;
        orgLink.appendChild(document.createTextNode("Original ad link"));
        containerDiv.appendChild(document.createTextNode(" | "));
        containerDiv.appendChild(orgLink);
      }
    });
  }
}

ReplaceAd();