Bug 1159884 - Implement inadjacency with a hardcoded list of hashed sites [r=adw]

browser/base/jar.mn

@@ -110,6 +110,7 @@ browser.jar:
         content/browser/newtab/newTab.xul             (content/newtab/newTab.xul)
 *       content/browser/newtab/newTab.js              (content/newtab/newTab.js)
         content/browser/newtab/newTab.css             (content/newtab/newTab.css)
+        content/browser/newtab/newTab.inadjacent.json         (content/newtab/newTab.inadjacent.json)
 *       content/browser/pageinfo/pageInfo.xul         (content/pageinfo/pageInfo.xul)
         content/browser/pageinfo/pageInfo.js          (content/pageinfo/pageInfo.js)
         content/browser/pageinfo/pageInfo.css         (content/pageinfo/pageInfo.css)

browser/modules/DirectoryLinksProvider.jsm

@@ -28,9 +28,22 @@ XPCOMUtils.defineLazyModuleGetter(this, "Promise",
   "resource://gre/modules/Promise.jsm");
 XPCOMUtils.defineLazyModuleGetter(this, "UpdateChannel",
   "resource://gre/modules/UpdateChannel.jsm");
+XPCOMUtils.defineLazyServiceGetter(this, "eTLD",
+  "@mozilla.org/network/effective-tld-service;1",
+  "nsIEffectiveTLDService");
 XPCOMUtils.defineLazyGetter(this, "gTextDecoder", () => {
   return new TextDecoder();
 });
+XPCOMUtils.defineLazyGetter(this, "gCryptoHash", function () {
+  return Cc["@mozilla.org/security/hash;1"].createInstance(Ci.nsICryptoHash);
+});
+XPCOMUtils.defineLazyGetter(this, "gUnicodeConverter", function () {
+  let converter = Cc["@mozilla.org/intl/scriptableunicodeconverter"]
+                    .createInstance(Ci.nsIScriptableUnicodeConverter);
+  converter.charset = 'utf8';
+  return converter;
+});
+
 
 // The filename where directory links are stored locally
 const DIRECTORY_LINKS_FILE = "directoryLinks.json";
@@ -108,12 +121,18 @@ const DEFAULT_PRUNE_TIME_DELTA = 10*24*60*60*1000;
 // The min number of visible (not blocked) history tiles to have before showing suggested tiles
 const MIN_VISIBLE_HISTORY_TILES = 8;
 
+// The max number of visible (not blocked) history tiles to test for inadjacency
+const MAX_VISIBLE_HISTORY_TILES = 15;
+
 // Divide frecency by this amount for pings
 const PING_SCORE_DIVISOR = 10000;
 
 // Allowed ping actions remotely stored as columns: case-insensitive [a-z0-9_]
 const PING_ACTIONS = ["block", "click", "pin", "sponsored", "sponsored_link", "unpin", "view"];
 
+// Location of inadjacent sites json
+const INADJACENCY_SOURCE = "chrome://browser/content/newtab/newTab.inadjacent.json";
+
 /**
  * Singleton that serves as the provider of directory links.
  * Directory links are a hard-coded set of links shown if a user's link
@@ -151,6 +170,23 @@ let DirectoryLinksProvider = {
    */
   _topSitesWithSuggestedLinks: new Set(),
 
+  /**
+   * lookup Set of inadjacent domains
+   */
+  _inadjacentSites: new Set(),
+
+  /**
+   * This flag is set if there is a suggested tile configured to avoid
+   * inadjacent sites in new tab
+   */
+  _avoidInadjacentSites: false,
+
+  /**
+   * This flag is set if _avoidInadjacentSites is true and there is
+   * an inadjacent site in the new tab
+   */
+  _newTabHasInadjacentSite: false,
+
   get _observedPrefs() Object.freeze({
     enhanced: PREF_NEWTAB_ENHANCED,
     linksURL: PREF_DIRECTORY_SOURCE,
@@ -274,22 +310,26 @@ let DirectoryLinksProvider = {
     uri = uri.replace("%LOCALE%", this.locale);
     uri = uri.replace("%CHANNEL%", UpdateChannel.get());
 
+    return this._downloadJsonData(uri).then(json => {
+      return OS.File.writeAtomic(this._directoryFilePath, json, {tmpPath: this._directoryFilePath + ".tmp"});
+    });
+  },
+
+  /**
+   * Downloads a links with json content
+   * @param download uri
+   * @return promise resolved to json string, "{}" returned if status != 200
+   */
+  _downloadJsonData: function DirectoryLinksProvider__downloadJsonData(uri) {
     let deferred = Promise.defer();
     let xmlHttp = this._newXHR();
 
-    let self = this;
     xmlHttp.onload = function(aResponse) {
       let json = this.responseText;
       if (this.status && this.status != 200) {
         json = "{}";
       }
-      OS.File.writeAtomic(self._directoryFilePath, json, {tmpPath: self._directoryFilePath + ".tmp"})
+      deferred.resolve(json);
-        .then(() => {
-          deferred.resolve();
-        },
-        () => {
-          deferred.reject("Error writing uri data in profD.");
-        });
     };
 
     xmlHttp.onerror = function(e) {
@@ -616,6 +656,7 @@ let DirectoryLinksProvider = {
       this._enhancedLinks.clear();
       this._suggestedLinks.clear();
       this._clearCampaignTimeout();
+      this._avoidInadjacentSites = false;
 
       // Only check base domain for images when using the default pref
       let checkBase = !this.__linksURLModified;
@@ -640,6 +681,10 @@ let DirectoryLinksProvider = {
         link.explanation = link.explanation ? ParserUtils.convertToPlainText(link.explanation, sanitizeFlags, 0) : "";
         link.targetedName = ParserUtils.convertToPlainText(link.adgroup_name, sanitizeFlags, 0) || name;
         link.lastVisitDate = rawLinks.suggested.length - position;
+        // check if link wants to avoid inadjacent sites
+        if (link.check_inadjacency) {
+          this._avoidInadjacentSites = true;
+        }
 
         // We cache suggested tiles here but do not push any of them in the links list yet.
         // The decision for which suggested tile to include will be made separately.
@@ -689,6 +734,8 @@ let DirectoryLinksProvider = {
 
     // setup frequency cap file path
     this._frequencyCapFilePath = OS.Path.join(OS.Constants.Path.localProfileDir, FREQUENCY_CAP_FILE);
+    // setup inadjacent sites URL
+    this._inadjacentSitesUrl = INADJACENCY_SOURCE;
 
     NewTabUtils.placesProvider.addObserver(this);
     NewTabUtils.links.addObserver(this);
@@ -704,6 +751,8 @@ let DirectoryLinksProvider = {
       yield this._readFrequencyCapFile();
       // fetch directory on startup without force
       yield this._fetchAndCacheLinksIfNecessary();
+      // fecth inadjacent sites on startup
+      yield this._loadInadjacentSites();
     }.bind(this));
   },
 
@@ -736,6 +785,13 @@ let DirectoryLinksProvider = {
       this._topSitesWithSuggestedLinks.add(changedLinkSite);
       return true;
     }
+
+    // always run _updateSuggestedTile if aLink is inadjacent
+    // and there are tiles configured to avoid it
+    if (this._avoidInadjacentSites && this._isInadjacentLink(aLink)) {
+      return true;
+    }
+
     return false;
   },
 
@@ -779,11 +835,17 @@ let DirectoryLinksProvider = {
 
   _getCurrentTopSiteCount: function() {
     let visibleTopSiteCount = 0;
-    for (let link of NewTabUtils.links.getLinks().slice(0, MIN_VISIBLE_HISTORY_TILES)) {
+    let newTabLinks = NewTabUtils.links.getLinks();
+    for (let link of newTabLinks.slice(0, MIN_VISIBLE_HISTORY_TILES)) {
+      // compute visibleTopSiteCount for suggested tiles
       if (link && (link.type == "history" || link.type == "enhanced")) {
         visibleTopSiteCount++;
       }
     }
+    // since newTabLinks are available, set _newTabHasInadjacentSite here
+    // note that _shouldUpdateSuggestedTile is called by _updateSuggestedTile
+    this._newTabHasInadjacentSite = this._avoidInadjacentSites && this._checkForInadjacentSites(newTabLinks);
+
     return visibleTopSiteCount;
   },
 
@@ -872,6 +934,11 @@ let DirectoryLinksProvider = {
           return;
         }
 
+        // Skip link if it avoids inadjacent sites and newtab has one
+        if (suggestedLink.check_inadjacency && this._newTabHasInadjacentSite) {
+          return;
+        }
+
         possibleLinks.set(url, suggestedLink);
 
         // Keep a map of URL to targeted sites. We later use this to show the user
@@ -913,6 +980,66 @@ let DirectoryLinksProvider = {
     return chosenSuggestedLink;
    },
 
+  /**
+   * Loads inadjacent sites
+   * @return a promise resolved when lookup Set for sites is built
+   */
+  _loadInadjacentSites: function DirectoryLinksProvider_loadInadjacentSites() {
+    return this._downloadJsonData(this._inadjacentSitesUrl).then(jsonString => {
+      let jsonObject = {};
+      try {
+        jsonObject = JSON.parse(jsonString);
+      }
+      catch (e) {
+        Cu.reportError(e);
+      }
+
+      this._inadjacentSites = new Set(jsonObject.domains);
+    });
+  },
+
+  /**
+   * Genegrates hash suitable for looking up inadjacent site
+   * @param value to hsh
+   * @return hased value, base64-ed
+   */
+  _generateHash: function DirectoryLinksProvider_generateHash(value) {
+    let byteArr = gUnicodeConverter.convertToByteArray(value);
+    gCryptoHash.init(gCryptoHash.MD5);
+    gCryptoHash.update(byteArr, byteArr.length);
+    return gCryptoHash.finish(true);
+  },
+
+  /**
+   * Checks if link belongs to inadjacent domain
+   * @param link to check
+   * @return true for inadjacent domains, false otherwise
+   */
+  _isInadjacentLink: function DirectoryLinksProvider_isInadjacentLink(link) {
+    let baseDomain = link.baseDomain || NewTabUtils.extractSite(link.url || "");
+    if (!baseDomain) {
+        return false;
+    }
+    // check if hashed domain is inadjacent
+    return this._inadjacentSites.has(this._generateHash(baseDomain));
+  },
+
+  /**
+   * Checks if new tab has inadjacent site
+   * @param new tab links (or nothing, in which case NewTabUtils.links.getLinks() is called
+   * @return true if new tab shows has inadjacent site
+   */
+  _checkForInadjacentSites: function DirectoryLinksProvider_checkForInadjacentSites(newTabLink) {
+    let links = newTabLink || NewTabUtils.links.getLinks();
+    for (let link of links.slice(0, MAX_VISIBLE_HISTORY_TILES)) {
+      // check links against inadjacent list - specifically include ALL link types
+      if (this._isInadjacentLink(link)) {
+        return true;
+      }
+    }
+    return false;
+  },
+
   /**
    * Reads json file, parses its content, and returns resulting object
    * @param json file path

browser/modules/test/xpcshell/test_DirectoryLinksProvider.js

@@ -209,6 +209,7 @@ function promiseCleanDirectoryLinksProvider() {
     yield promiseDirectoryDownloadOnPrefChange(kLocalePref, "en-US");
     yield promiseDirectoryDownloadOnPrefChange(kSourceUrlPref, kTestURL);
     yield DirectoryLinksProvider._clearFrequencyCap();
+    yield DirectoryLinksProvider._loadInadjacentSites();
     DirectoryLinksProvider._lastDownloadMS  = 0;
     DirectoryLinksProvider.reset();
   });
@@ -1745,3 +1746,181 @@ add_task(function test_sanitizeExplanation() {
   do_check_eq(suggestedLink.explanation, "This is an evil tile X muhahaha");
   do_check_eq(suggestedLink.targetedName, "WE ARE EVIL ");
 });
+
+add_task(function test_inadjecentSites() {
+  let suggestedTile = Object.assign({
+    check_inadjacency: true
+  }, suggestedTile1);
+
+  // Initial setup
+  let topSites = ["1040.com", "site2.com", "hrblock.com", "site4.com", "freetaxusa.com", "site6.com"];
+  let data = {"suggested": [suggestedTile], "directory": [someOtherSite]};
+  let dataURI = 'data:application/json,' + JSON.stringify(data);
+
+  let testObserver = new TestFirstRun();
+  DirectoryLinksProvider.addObserver(testObserver);
+
+  let origGetFrecentSitesName = DirectoryLinksProvider.getFrecentSitesName;
+  DirectoryLinksProvider.getFrecentSitesName = () => "";
+
+  yield promiseSetupDirectoryLinksProvider({linksURL: dataURI});
+  let links = yield fetchData();
+
+  let origIsTopPlacesSite = NewTabUtils.isTopPlacesSite;
+  NewTabUtils.isTopPlacesSite = function(site) {
+    return topSites.indexOf(site) >= 0;
+  }
+
+  let origGetProviderLinks = NewTabUtils.getProviderLinks;
+  NewTabUtils.getProviderLinks = function(provider) {
+    return links;
+  }
+
+  let origCurrentTopSiteCount = DirectoryLinksProvider._getCurrentTopSiteCount;
+  DirectoryLinksProvider._getCurrentTopSiteCount = () => {
+    origCurrentTopSiteCount.apply(DirectoryLinksProvider);
+    return 8;
+  };
+
+  // store oroginal inadjacent sites url
+  let origInadjacentSitesUrl = DirectoryLinksProvider._inadjacentSitesUrl;
+
+  // loading inadjacent sites list function
+  function setInadjacentSites(sites) {
+    let badSiteB64 = [];
+    sites.forEach(site => {
+      badSiteB64.push(DirectoryLinksProvider._generateHash(site));
+    });
+    let theList = {"domains": badSiteB64};
+    let dataURI = 'data:application/json,' + JSON.stringify(theList);
+    DirectoryLinksProvider._inadjacentSitesUrl = dataURI;
+    return DirectoryLinksProvider._loadInadjacentSites();
+  };
+
+  // setup gLinks loader
+  let gLinks = NewTabUtils.links;
+  gLinks.addProvider(DirectoryLinksProvider);
+
+  function updateNewTabCache() {
+    gLinks.populateCache();
+    return new Promise(resolve => {
+      NewTabUtils.allPages.register({
+        observe: _ => _,
+        update() {
+          NewTabUtils.allPages.unregister(this);
+          resolve();
+        }
+      });
+  });
+  }
+
+  // no suggested file
+  do_check_eq(DirectoryLinksProvider._updateSuggestedTile(), undefined);
+  // _avoidInadjacentSites should be set, since link.check_inadjacency is on
+  do_check_true(DirectoryLinksProvider._avoidInadjacentSites);
+  // make sure example.com is included in inadjacent sites list
+  do_check_true(DirectoryLinksProvider._isInadjacentLink({baseDomain: "example.com"}));
+
+  function TestFirstRun() {
+    this.promise = new Promise(resolve => {
+      this.onLinkChanged = (directoryLinksProvider, link) => {
+        do_check_eq(link.url, suggestedTile.url);
+        do_check_eq(link.type, "affiliate");
+        resolve();
+      };
+    });
+  }
+
+  // Test first call to '_updateSuggestedTile()', called when fetching directory links.
+  yield testObserver.promise;
+  DirectoryLinksProvider.removeObserver(testObserver);
+
+  // update newtab cache
+  yield updateNewTabCache();
+  // this should have set
+  do_check_true(DirectoryLinksProvider._avoidInadjacentSites);
+
+  // there should be siggested link
+  let link = DirectoryLinksProvider._updateSuggestedTile();
+  do_check_eq(link.url, "http://turbotax.com");
+  // and it should have avoidInadjacentSites flag
+  do_check_true(link.check_inadjacency);
+
+  // make someothersite.com inadjacent
+  yield setInadjacentSites(["someothersite.com"]);
+
+  // there should be no suggested link
+  link = DirectoryLinksProvider._updateSuggestedTile();
+  do_check_false(link);
+  do_check_true(DirectoryLinksProvider._newTabHasInadjacentSite);
+
+  // _handleLinkChanged must return true on inadjacent site
+  do_check_true(DirectoryLinksProvider._handleLinkChanged({
+    url: "http://someothersite.com",
+    type: "history",
+  }));
+  // _handleLinkChanged must return false on ok site
+  do_check_false(DirectoryLinksProvider._handleLinkChanged({
+    url: "http://foobar.com",
+    type: "history",
+  }));
+
+  // change inadjacent list to sites not on newtab page
+  yield setInadjacentSites(["foo.com", "bar.com"]);
+
+  link = DirectoryLinksProvider._updateSuggestedTile();
+  // we should now have a link
+  do_check_true(link);
+  do_check_eq(link.url, "http://turbotax.com");
+
+  // make newtab offending again
+  yield setInadjacentSites(["someothersite.com", "foo.com"]);
+  // there should be no suggested link
+  link = DirectoryLinksProvider._updateSuggestedTile();
+  do_check_false(link);
+  do_check_true(DirectoryLinksProvider._newTabHasInadjacentSite);
+
+  // remove avoidInadjacentSites flag from suggested tile and reload json
+  delete suggestedTile.check_inadjacency;
+  data = {"suggested": [suggestedTile], "directory": [someOtherSite]};
+  dataURI = 'data:application/json,' + JSON.stringify(data);
+  yield promiseDirectoryDownloadOnPrefChange(kSourceUrlPref, dataURI);
+  yield fetchData();
+
+  // inadjacent checking should be disabled
+  do_check_false(DirectoryLinksProvider._avoidInadjacentSites);
+  link = DirectoryLinksProvider._updateSuggestedTile();
+  do_check_true(link);
+  do_check_eq(link.url, "http://turbotax.com");
+  do_check_false(DirectoryLinksProvider._newTabHasInadjacentSite);
+
+  // _handleLinkChanged should return false now, even if newtab has bad site
+  do_check_false(DirectoryLinksProvider._handleLinkChanged({
+    url: "http://someothersite.com",
+    type: "history",
+  }));
+
+  // test _isInadjacentLink
+  do_check_true(DirectoryLinksProvider._isInadjacentLink({baseDomain: "someothersite.com"}));
+  do_check_false(DirectoryLinksProvider._isInadjacentLink({baseDomain: "bar.com"}));
+  do_check_true(DirectoryLinksProvider._isInadjacentLink({url: "http://www.someothersite.com"}));
+  do_check_false(DirectoryLinksProvider._isInadjacentLink({url: "http://www.bar.com"}));
+  // try to crash _isInadjacentLink
+  do_check_false(DirectoryLinksProvider._isInadjacentLink({baseDomain: ""}));
+  do_check_false(DirectoryLinksProvider._isInadjacentLink({url: ""}));
+  do_check_false(DirectoryLinksProvider._isInadjacentLink({url: "http://localhost:8081/"}));
+  do_check_false(DirectoryLinksProvider._isInadjacentLink({url: "abracodabra"}));
+  do_check_false(DirectoryLinksProvider._isInadjacentLink({}));
+
+  // test _checkForInadjacentSites
+  do_check_true(DirectoryLinksProvider._checkForInadjacentSites());
+
+  // Cleanup
+  gLinks.removeProvider(DirectoryLinksProvider);
+  DirectoryLinksProvider._inadjacentSitesUrl = origInadjacentSitesUrl;
+  DirectoryLinksProvider.getFrecentSitesName = origGetFrecentSitesName;
+  NewTabUtils.isTopPlacesSite = origIsTopPlacesSite;
+  NewTabUtils.getProviderLinks = origGetProviderLinks;
+  DirectoryLinksProvider._getCurrentTopSiteCount = origCurrentTopSiteCount;
+  yield promiseCleanDirectoryLinksProvider();
+});