Bug 1835321 - Allow any link on SERPs that don't match an ad expression to be considered a non-ad - r=Standard8

Differential Revision: https://phabricator.services.mozilla.com/D179367

browser/actors/SearchSERPTelemetryChild.sys.mjs

@@ -67,9 +67,6 @@ class SearchProviders {
           extraAdServersRegexps: p.extraAdServersRegexps.map(
             r => new RegExp(r)
           ),
-          nonAdsLinkRegexps: p.nonAdsLinkRegexps?.length
-            ? p.nonAdsLinkRegexps.map(r => new RegExp(r))
-            : [],
         };
       });
 
@@ -388,13 +385,6 @@ class SearchAdImpression {
         if (result.relatedElements?.length) {
           this.#addEventListenerToElements(result.relatedElements, result.type);
         }
-        // If an anchor doesn't match any component, and it doesn't have a non
-        // ads link regexp, cache the anchor so the parent process can observe
-        // them.
-      } else if (!this.#providerInfo.nonAdsLinkRegexps.length) {
-        this.#recordElementData(anchor, {
-          type: "non_ads_link",
-        });
       }
     }
   }

browser/components/search/SearchSERPTelemetry.sys.mjs

@@ -925,9 +925,16 @@ class ContentHandler {
     }
 
     let wrappedChannel = ChannelWrapper.get(channel);
+    // The channel we're observing might be a redirect of a channel we've
+    // observed before.
     if (wrappedChannel._adClickRecorded) {
       lazy.logConsole.debug("Ad click already recorded");
       return;
+      // When _adClickRecorded is false but _recordedClick is true, it means we
+      // recorded a non-ad link click, and it is being re-directed.
+    } else if (wrappedChannel._recordedClick) {
+      lazy.logConsole.debug("Non ad-click already recorded");
+      return;
     }
 
     Services.tm.dispatchToMainThread(() => {
@@ -953,15 +960,35 @@ class ContentHandler {
         return provider.telemetryId == providerInfo;
       });
 
-      // The SERP "clicked" action is implied if a user loads another page from
-      // the context of a SERP. At this point, we don't know if the request is
-      // from a SERP but we want to avoid inspecting requests that are not
-      // documents, or not a top level load.
+      // Some channels re-direct by loading pages that return 200. The result
+      // is the channel will have an originURL that changes from the SERP to
+      // either a nonAdsRegexp or an extraAdServersRegexps. This is typical
+      // for loading a page in a new tab. The channel will have changed so any
+      // properties attached to them to record state (e.g. _recordedClick)
+      // won't be present.
+      if (
+        info.nonAdsLinkRegexps.some(r => r.test(originURL)) ||
+        info.extraAdServersRegexps.some(r => r.test(originURL))
+      ) {
+        return;
+      }
+
+      // A click event is recorded if a user loads a resource from an
+      // originURL that is a SERP.
+      //
+      // Typically, we only want top level loads containing documents to avoid
+      // recording any event on an in-page resource a SERP might load
+      // (e.g. CSS files).
+      //
+      // The exception to this is if a subframe loads a resource that matches
+      // a non ad link. Some SERPs encode non ad search results with a URL
+      // that gets loaded into an iframe, which then tells the container of
+      // the iframe to change the location of the page.
       if (
         lazy.serpEventsEnabled &&
         channel.isDocument &&
-        channel.loadInfo.isTopLevelLoad &&
-        !wrappedChannel._countedClick
+        (channel.loadInfo.isTopLevelLoad ||
+          info.nonAdsLinkRegexps.some(r => r.test(URL)))
       ) {
         let start = Cu.now();
 
@@ -1018,27 +1045,12 @@ class ContentHandler {
               }
             }
           }
-          // Check if the href matches a non-ads link. Do this after looking at
-          // hrefToComponentMap because a link that looks like a non-ad might
-          // have a more specific component type.
+
+          // Default value for URLs that don't match any components categorized
+          // on the page.
           if (!type) {
-            type = info.nonAdsLinkRegexps.some(r => r.test(URL))
-              ? SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK
-              : "";
-          }
-          // The SERP may have moved onto another page that matches a SERP page
-          // e.g. Related Search
-          if (!type && isSerp) {
             type = SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK;
           }
-          // There might be other types of pages on a SERP that don't fall
-          // neatly into expected non-ad expressions or SERPs, such as Image
-          // Search, Maps, etc.
-          if (!type) {
-            type = info.extraPageRegexps?.some(r => r.test(URL))
-              ? SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK
-              : "";
-          }
 
           if (isSerp && isFromNewtab) {
             SearchSERPTelemetry.setBrowserContentSource(
@@ -1047,35 +1059,22 @@ class ContentHandler {
             );
           }
 
-          // Step 3: If we have a type, record an engagement.
-          // Exceptions:
-          // - Related searches on some SERPs can be encoded with a URL that
-          //   match a nonAdsLinkRegexp. This means we'll have seen the link
-          //   twice, once with the nonAdsLinkRegexp and again with a SERP URL
-          //   matching a searchPageRegexp. We don't want to record the
-          //   engagement twice, so if the origin of the request was
-          //   nonAdsLinkRegexp, skip the categorization. The reason why we
-          //   don't do this check earlier is because if the final URL is a
-          //   SERP, we'll want to define the source property of the subsequent
-          //   SERP impression.
-          if (type && !info.nonAdsLinkRegexps.some(r => r.test(originURL))) {
-            impressionIdsWithoutEngagementsSet.delete(
-              telemetryState.impressionId
-            );
-            Glean.serp.engagement.record({
-              impression_id: telemetryState.impressionId,
-              action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
-              target: type,
-            });
-            lazy.logConsole.debug("Counting click:", {
-              impressionId: telemetryState.impressionId,
-              type,
-              URL,
-            });
-            wrappedChannel._countedClick = true;
-          } else if (!type) {
-            lazy.logConsole.warn(`Could not find a component type for ${URL}`);
-          }
+          // Step 3: Record the engagement.
+          impressionIdsWithoutEngagementsSet.delete(
+            telemetryState.impressionId
+          );
+          Glean.serp.engagement.record({
+            impression_id: telemetryState.impressionId,
+            action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
+            target: type,
+          });
+          lazy.logConsole.debug("Counting click:", {
+            impressionId: telemetryState.impressionId,
+            type,
+            URL,
+          });
+          // Prevent re-directed channels from being examined more than once.
+          wrappedChannel._recordedClick = true;
         }
         ChromeUtils.addProfilerMarker(
           "SearchSERPTelemetry._observeActivity",

browser/components/search/test/browser/browser.ini

@@ -99,13 +99,30 @@ tags = search-telemetry
 support-files =
   searchTelemetryAd_searchbox_with_content.html
   searchTelemetryAd_searchbox_with_content.html^headers^
+[browser_search_telemetry_engagement_non_ad.js]
+tags = search-telemetry
+support-files =
+  searchTelemetryAd_searchbox_with_content.html
+  searchTelemetryAd_searchbox_with_content.html^headers^
+  serp.css
+[browser_search_telemetry_engagement_redirect.js]
+tags = search-telemetry
+support-files =
+  redirect_final.sjs
+  redirect_once.sjs
+  redirect_thrice.sjs
+  redirect_twice.sjs
+  searchTelemetryAd_adsLink_redirect.html
+  searchTelemetryAd_components_text.html
+  searchTelemetryAd_nonAdsLink_redirect.html
+  searchTelemetryAd_nonAdsLink_redirect.html^headers^
+  searchTelemetryAd_nonAdsLink_redirect_nonTopLoaded.html
+  searchTelemetryAd_nonAdsLink_redirect_nonTopLoaded.html^headers^
   serp.css
 [browser_search_telemetry_engagement_target.js]
 tags = search-telemetry
 support-files =
   searchTelemetryAd_components_text.html
-  searchTelemetryAd_nonAdsLink_redirect.html
-  searchTelemetryAd_nonAdsLink_redirect.html^headers^
   searchTelemetryAd_searchbox.html
   searchTelemetryAd_searchbox.html^headers^
   serp.css

browser/components/search/test/browser/browser_search_telemetry_engagement_cached.js

@@ -19,9 +19,7 @@ const TEST_PROVIDER_INFO = [
     codeParamName: "abc",
     taggedCodes: ["ff"],
     adServerAttributes: ["mozAttr"],
-    nonAdsLinkRegexps: [
-      /^https:\/\/example.com\/browser\/browser\/components\/search\/test\/browser\//,
-    ],
+    nonAdsLinkRegexps: [],
     extraAdServersRegexps: [/^https:\/\/example\.com\/ad/],
     components: [
       {

browser/components/search/test/browser/browser_search_telemetry_engagement_content.js

@@ -323,11 +323,8 @@ add_task(async function test_click_related_search_in_new_tab() {
   BrowserTestUtils.removeTab(tab2);
 });
 
-// We consider regular expressions in nonAdsLinkRegexps and
-// searchPageRegexp/extraPageRegexps as valid non ads links when recording
-// an engagement event. However, if a nonAdsLinkRegexp leads to a
-// searchPageRegexp/extraPageRegexps, than we risk double counting in the case
-// of a re-direct occuring in a new tab.
+// We consider regular expressions in nonAdsLinkRegexps and searchPageRegexp
+// as valid non ads links when recording an engagement event.
 add_task(async function test_click_redirect_search_in_newtab() {
   resetTelemetry();
   let url = getSERPUrl("searchTelemetryAd_searchbox_with_content.html");

browser/components/search/test/browser/browser_search_telemetry_engagement_non_ad.js

@@ -0,0 +1,164 @@
+/* Any copyright is dedicated to the Public Domain.
+ * http://creativecommons.org/publicdomain/zero/1.0/ */
+
+/**
+ * These tests load SERPs and click on links that are non ads. Non ads can have
+ * slightly different behavior from ads.
+ */
+
+"use strict";
+
+const { SearchSERPTelemetry, SearchSERPTelemetryUtils } =
+  ChromeUtils.importESModule("resource:///modules/SearchSERPTelemetry.sys.mjs");
+
+const TEST_PROVIDER_INFO = [
+  {
+    telemetryId: "example",
+    searchPageRegexp:
+      /^https:\/\/example.org\/browser\/browser\/components\/search\/test\/browser\/searchTelemetryAd_/,
+    queryParamName: "s",
+    codeParamName: "abc",
+    taggedCodes: ["ff"],
+    adServerAttributes: ["mozAttr"],
+    nonAdsLinkRegexps: [],
+    extraAdServersRegexps: [/^https:\/\/example\.com\/ad/],
+    components: [
+      {
+        type: SearchSERPTelemetryUtils.COMPONENTS.AD_LINK,
+        default: true,
+      },
+    ],
+  },
+];
+
+function getSERPUrl(page, organic = false) {
+  let url =
+    getRootDirectory(gTestPath).replace(
+      "chrome://mochitests/content",
+      "https://example.org"
+    ) + page;
+  return `${url}?s=test${organic ? "" : "&abc=ff"}`;
+}
+
+async function promiseImpressionReceived() {
+  return TestUtils.waitForCondition(() => {
+    let adImpressions = Glean.serp.adImpression.testGetValue() ?? [];
+    return adImpressions.length;
+  }, "Should have received an ad impression.");
+}
+
+async function waitForIdle() {
+  for (let i = 0; i < 10; i++) {
+    await new Promise(resolve => Services.tm.idleDispatchToMainThread(resolve));
+  }
+}
+
+add_setup(async function () {
+  SearchSERPTelemetry.overrideSearchTelemetryForTests(TEST_PROVIDER_INFO);
+  await waitForIdle();
+  // Enable local telemetry recording for the duration of the tests.
+  await SpecialPowers.pushPrefEnv({
+    set: [
+      ["browser.search.log", true],
+      ["browser.search.serpEventTelemetry.enabled", true],
+    ],
+  });
+
+  registerCleanupFunction(async () => {
+    SearchSERPTelemetry.overrideSearchTelemetryForTests();
+    resetTelemetry();
+  });
+});
+
+// If an anchor is a non_ads_link and it doesn't match a non-ads regular
+// expression, it should still be categorize it as a non ad.
+add_task(async function test_click_non_ads_link() {
+  await waitForIdle();
+
+  resetTelemetry();
+  let url = getSERPUrl("searchTelemetryAd_components_text.html");
+  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
+  await waitForPageWithAdImpressions();
+
+  // Click a non ad.
+  let pageLoadPromise = BrowserTestUtils.waitForLocationChange(gBrowser);
+  await BrowserTestUtils.synthesizeMouseAtCenter(
+    "#non_ads_link",
+    {},
+    tab.linkedBrowser
+  );
+  await pageLoadPromise;
+
+  assertImpressionEvents([
+    {
+      impression: {
+        provider: "example",
+        tagged: "true",
+        partner_code: "ff",
+        source: "unknown",
+        is_shopping_page: "false",
+        shopping_tab_displayed: "false",
+      },
+      engagements: [
+        {
+          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
+          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
+        },
+      ],
+    },
+  ]);
+
+  BrowserTestUtils.removeTab(tab);
+
+  // Reset state for other tests.
+  SearchSERPTelemetry.overrideSearchTelemetryForTests(TEST_PROVIDER_INFO);
+  await waitForIdle();
+});
+
+// Click on an non-ad element while no ads are present.
+add_task(async function test_click_non_ad_with_no_ads() {
+  await waitForIdle();
+
+  resetTelemetry();
+
+  let url = getSERPUrl("searchTelemetryAd_searchbox.html");
+  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
+  await waitForPageWithAdImpressions();
+
+  let browserLoadedPromise = BrowserTestUtils.browserLoaded(
+    tab.linkedBrowser,
+    true,
+    "https://example.com/hello_world"
+  );
+  await BrowserTestUtils.synthesizeMouseAtCenter(
+    "#non_ads_link",
+    {},
+    tab.linkedBrowser
+  );
+  await browserLoadedPromise;
+
+  assertImpressionEvents([
+    {
+      impression: {
+        provider: "example",
+        tagged: "true",
+        partner_code: "ff",
+        source: "unknown",
+        is_shopping_page: "false",
+        shopping_tab_displayed: "false",
+      },
+      engagements: [
+        {
+          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
+          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
+        },
+      ],
+    },
+  ]);
+
+  BrowserTestUtils.removeTab(tab);
+
+  // Reset state for other tests.
+  SearchSERPTelemetry.overrideSearchTelemetryForTests(TEST_PROVIDER_INFO);
+  await waitForIdle();
+});

browser/components/search/test/browser/browser_search_telemetry_engagement_redirect.js

@@ -0,0 +1,346 @@
+/* Any copyright is dedicated to the Public Domain.
+ * http://creativecommons.org/publicdomain/zero/1.0/ */
+
+/**
+ * These tests load SERPs and click on both ad and non-ad links that can be
+ * redirected.
+ */
+
+"use strict";
+
+const { SearchSERPTelemetry, SearchSERPTelemetryUtils } =
+  ChromeUtils.importESModule("resource:///modules/SearchSERPTelemetry.sys.mjs");
+
+const TEST_PROVIDER_INFO = [
+  {
+    telemetryId: "example",
+    searchPageRegexp:
+      /^https:\/\/example.org\/browser\/browser\/components\/search\/test\/browser\/searchTelemetryAd_/,
+    queryParamName: "s",
+    codeParamName: "abc",
+    taggedCodes: ["ff"],
+    adServerAttributes: ["mozAttr"],
+    nonAdsLinkRegexps: [
+      /^https:\/\/example.org\/browser\/browser\/components\/search\/test\/browser\/searchTelemetryAd_nonAdsLink_redirect.html/,
+    ],
+    extraAdServersRegexps: [
+      /^https:\/\/example\.com\/ad/,
+      /^https:\/\/example.org\/browser\/browser\/components\/search\/test\/browser\/searchTelemetryAd_adsLink_redirect.html/,
+    ],
+    components: [
+      {
+        type: SearchSERPTelemetryUtils.COMPONENTS.AD_LINK,
+        default: true,
+      },
+    ],
+  },
+];
+
+function getSERPUrl(page, organic = false) {
+  let url =
+    getRootDirectory(gTestPath).replace(
+      "chrome://mochitests/content",
+      "https://example.org"
+    ) + page;
+  return `${url}?s=test${organic ? "" : "&abc=ff"}`;
+}
+
+async function promiseImpressionReceived() {
+  return TestUtils.waitForCondition(() => {
+    let adImpressions = Glean.serp.adImpression.testGetValue() ?? [];
+    return adImpressions.length;
+  }, "Should have received an ad impression.");
+}
+
+async function waitForIdle() {
+  for (let i = 0; i < 10; i++) {
+    await new Promise(resolve => Services.tm.idleDispatchToMainThread(resolve));
+  }
+}
+
+add_setup(async function () {
+  SearchSERPTelemetry.overrideSearchTelemetryForTests(TEST_PROVIDER_INFO);
+  await waitForIdle();
+  // Enable local telemetry recording for the duration of the tests.
+  await SpecialPowers.pushPrefEnv({
+    set: [
+      ["browser.search.log", true],
+      ["browser.search.serpEventTelemetry.enabled", true],
+    ],
+  });
+
+  registerCleanupFunction(async () => {
+    SearchSERPTelemetry.overrideSearchTelemetryForTests();
+    resetTelemetry();
+  });
+});
+
+add_task(async function test_click_non_ads_link_redirected() {
+  resetTelemetry();
+
+  let url = getSERPUrl("searchTelemetryAd_components_text.html");
+  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
+  await waitForPageWithAdImpressions();
+
+  let targetUrl = "https://example.com/hello_world";
+
+  let browserLoadedPromise = BrowserTestUtils.browserLoaded(
+    tab.linkedBrowser,
+    true,
+    targetUrl
+  );
+  await BrowserTestUtils.synthesizeMouseAtCenter(
+    "#non_ads_link_redirected",
+    {},
+    tab.linkedBrowser
+  );
+
+  await browserLoadedPromise;
+
+  assertImpressionEvents([
+    {
+      impression: {
+        provider: "example",
+        tagged: "true",
+        partner_code: "ff",
+        source: "unknown",
+        is_shopping_page: "false",
+        shopping_tab_displayed: "false",
+      },
+      engagements: [
+        {
+          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
+          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
+        },
+      ],
+    },
+  ]);
+
+  BrowserTestUtils.removeTab(tab);
+});
+
+// If a provider does a re-direct and we open it in a new tab, we should
+// record the click and have the correct number of engagements.
+add_task(async function test_click_non_ads_link_redirected_new_tab() {
+  resetTelemetry();
+
+  let url = getSERPUrl("searchTelemetryAd_components_text.html");
+  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
+  await waitForPageWithAdImpressions();
+
+  let redirectUrl =
+    getRootDirectory(gTestPath).replace(
+      "chrome://mochitests/content",
+      "https://example.org"
+    ) + "searchTelemetryAd_nonAdsLink_redirect.html";
+  let targetUrl = "https://example.com/hello_world";
+
+  let tabPromise = BrowserTestUtils.waitForNewTab(gBrowser, targetUrl, true);
+
+  await SpecialPowers.spawn(tab.linkedBrowser, [redirectUrl], urls => {
+    content.document
+      .getElementById(["non_ads_link"])
+      .addEventListener("click", e => {
+        e.preventDefault();
+        content.window.open([urls], "_blank");
+      });
+    content.document.getElementById("non_ads_link").click();
+  });
+  let tab2 = await tabPromise;
+
+  assertImpressionEvents([
+    {
+      impression: {
+        provider: "example",
+        tagged: "true",
+        partner_code: "ff",
+        source: "unknown",
+        is_shopping_page: "false",
+        shopping_tab_displayed: "false",
+      },
+      engagements: [
+        {
+          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
+          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
+        },
+      ],
+    },
+  ]);
+
+  BrowserTestUtils.removeTab(tab);
+  BrowserTestUtils.removeTab(tab2);
+});
+
+// Some providers load a URL of a non ad within a subframe before loading the
+// target website in the top level frame.
+add_task(async function test_click_non_ads_link_redirect_non_top_level() {
+  resetTelemetry();
+
+  let url = getSERPUrl("searchTelemetryAd_components_text.html");
+  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
+  await waitForPageWithAdImpressions();
+
+  let targetUrl = "https://example.com/hello_world";
+
+  let browserPromise = BrowserTestUtils.browserLoaded(
+    tab.linkedBrowser,
+    true,
+    targetUrl
+  );
+  await BrowserTestUtils.synthesizeMouseAtCenter(
+    "#non_ads_link_redirected_no_top_level",
+    {},
+    tab.linkedBrowser
+  );
+
+  await browserPromise;
+
+  assertImpressionEvents([
+    {
+      impression: {
+        provider: "example",
+        tagged: "true",
+        partner_code: "ff",
+        source: "unknown",
+        is_shopping_page: "false",
+        shopping_tab_displayed: "false",
+      },
+      engagements: [
+        {
+          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
+          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
+        },
+      ],
+    },
+  ]);
+
+  BrowserTestUtils.removeTab(tab);
+});
+
+add_task(async function test_multiple_redirects_non_ad_link() {
+  resetTelemetry();
+
+  let url = getSERPUrl("searchTelemetryAd_components_text.html");
+  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
+  await waitForPageWithAdImpressions();
+
+  let targetUrl = "https://example.com/hello_world";
+
+  let browserLoadedPromise = BrowserTestUtils.browserLoaded(
+    tab.linkedBrowser,
+    true,
+    targetUrl
+  );
+  await BrowserTestUtils.synthesizeMouseAtCenter(
+    "#non_ads_link_multiple_redirects",
+    {},
+    tab.linkedBrowser
+  );
+
+  await browserLoadedPromise;
+
+  assertImpressionEvents([
+    {
+      impression: {
+        provider: "example",
+        tagged: "true",
+        partner_code: "ff",
+        source: "unknown",
+        is_shopping_page: "false",
+        shopping_tab_displayed: "false",
+      },
+      engagements: [
+        {
+          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
+          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
+        },
+      ],
+    },
+  ]);
+
+  BrowserTestUtils.removeTab(tab);
+});
+
+add_task(async function test_click_ad_link_redirected() {
+  resetTelemetry();
+
+  let url = getSERPUrl("searchTelemetryAd_components_text.html");
+  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
+  await waitForPageWithAdImpressions();
+
+  let targetUrl = "https://example.com/hello_world";
+
+  let browserLoadedPromise = BrowserTestUtils.browserLoaded(
+    tab.linkedBrowser,
+    true,
+    targetUrl
+  );
+  await BrowserTestUtils.synthesizeMouseAtCenter(
+    "#ad_link_redirect",
+    {},
+    tab.linkedBrowser
+  );
+
+  await browserLoadedPromise;
+
+  assertImpressionEvents([
+    {
+      impression: {
+        provider: "example",
+        tagged: "true",
+        partner_code: "ff",
+        source: "unknown",
+        is_shopping_page: "false",
+        shopping_tab_displayed: "false",
+      },
+      engagements: [
+        {
+          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
+          target: SearchSERPTelemetryUtils.COMPONENTS.AD_LINK,
+        },
+      ],
+    },
+  ]);
+
+  BrowserTestUtils.removeTab(tab);
+});
+
+add_task(async function test_click_ad_link_redirected_new_tab() {
+  resetTelemetry();
+
+  let url = getSERPUrl("searchTelemetryAd_components_text.html");
+  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
+  await waitForPageWithAdImpressions();
+
+  let targetUrl = "https://example.com/hello_world";
+
+  let tabPromise = BrowserTestUtils.waitForNewTab(gBrowser, targetUrl, true);
+  await BrowserTestUtils.synthesizeMouseAtCenter(
+    "#ad_link_redirect",
+    { button: 1 },
+    tab.linkedBrowser
+  );
+  let tab2 = await tabPromise;
+
+  assertImpressionEvents([
+    {
+      impression: {
+        provider: "example",
+        tagged: "true",
+        partner_code: "ff",
+        source: "unknown",
+        is_shopping_page: "false",
+        shopping_tab_displayed: "false",
+      },
+      engagements: [
+        {
+          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
+          target: SearchSERPTelemetryUtils.COMPONENTS.AD_LINK,
+        },
+      ],
+    },
+  ]);
+
+  BrowserTestUtils.removeTab(tab);
+  BrowserTestUtils.removeTab(tab2);
+});

browser/components/search/test/browser/browser_search_telemetry_engagement_target.js

@@ -22,7 +22,7 @@ const TEST_PROVIDER_INFO = [
     taggedCodes: ["ff"],
     adServerAttributes: ["mozAttr"],
     nonAdsLinkRegexps: [
-      /^https:\/\/example.org\/browser\/browser\/components\/search\/test\/browser\/searchTelemetryAd_nonAdsLink_redirect/,
+      /^https:\/\/example.org\/browser\/browser\/components\/search\/test\/browser\/searchTelemetryAd_nonAdsLink_redirect.html/,
     ],
     extraAdServersRegexps: [/^https:\/\/example\.com\/ad/],
     components: [
@@ -190,112 +190,6 @@ add_task(async function test_click_second_ad_in_component() {
   BrowserTestUtils.removeTab(tab);
 });
 
-// If a provider does a re-direct and we open it in a new tab, we should
-// record the click and have the correct number of engagements.
-add_task(async function test_click_non_ads_link_redirected_new_tab() {
-  resetTelemetry();
-
-  let url = getSERPUrl("searchTelemetryAd_components_text.html");
-  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
-  await waitForPageWithAdImpressions();
-
-  let redirectUrl =
-    getRootDirectory(gTestPath).replace(
-      "chrome://mochitests/content",
-      "https://example.org"
-    ) + "searchTelemetryAd_nonAdsLink_redirect.html";
-  let targetUrl = "https://example.com/hello_world";
-
-  let tabPromise = BrowserTestUtils.waitForNewTab(gBrowser, targetUrl, true);
-
-  await SpecialPowers.spawn(tab.linkedBrowser, [redirectUrl], urls => {
-    content.document
-      .getElementById(["non_ads_link"])
-      .addEventListener("click", e => {
-        e.preventDefault();
-        content.window.open([urls], "_blank");
-      });
-    content.document.getElementById("non_ads_link").click();
-  });
-  let tab2 = await tabPromise;
-
-  assertImpressionEvents([
-    {
-      impression: {
-        provider: "example",
-        tagged: "true",
-        partner_code: "ff",
-        source: "unknown",
-        is_shopping_page: "false",
-        shopping_tab_displayed: "false",
-      },
-      engagements: [
-        {
-          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
-          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
-        },
-      ],
-    },
-  ]);
-
-  BrowserTestUtils.removeTab(tab2);
-  BrowserTestUtils.removeTab(tab);
-});
-
-// If a provider does a re-direct in the same tab, we may not be able to
-// determine the component based on the URLs on the page, because the initial
-// redirect URL won't be known, and by the time the page is loaded, the cached
-// data might be invalidated. So this should use regular expressions to make
-// an inference.
-add_task(async function test_click_non_ads_link_redirected() {
-  resetTelemetry();
-
-  let url = getSERPUrl("searchTelemetryAd_components_text.html");
-  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
-  await waitForPageWithAdImpressions();
-
-  let redirectUrl =
-    getRootDirectory(gTestPath).replace(
-      "chrome://mochitests/content",
-      "https://example.org"
-    ) + "searchTelemetryAd_nonAdsLink_redirect.html";
-  let targetUrl = "https://example.com/hello_world";
-
-  let browserLoadedPromise = BrowserTestUtils.browserLoaded(
-    tab.linkedBrowser,
-    true,
-    targetUrl
-  );
-  await SpecialPowers.spawn(tab.linkedBrowser, [redirectUrl], urls => {
-    content.document
-      .getElementById(["non_ads_link"])
-      .setAttribute("href", [urls]);
-    content.document.getElementById("non_ads_link").click();
-  });
-  await browserLoadedPromise;
-
-  assertImpressionEvents([
-    {
-      impression: {
-        provider: "example",
-        tagged: "true",
-        partner_code: "ff",
-        source: "unknown",
-        is_shopping_page: "false",
-        shopping_tab_displayed: "false",
-      },
-      engagements: [
-        {
-          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
-          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
-        },
-      ],
-    },
-  ]);
-
-  BrowserTestUtils.removeTab(tab);
-});
-
 // If a provider appends query parameters to a link after the page has been
 // parsed, we should still be able to record the click.
 add_task(async function test_click_ads_link_modified() {
@@ -337,54 +231,6 @@ add_task(async function test_click_ads_link_modified() {
   BrowserTestUtils.removeTab(tab);
 });
 
-// If an anchor is a non_ads_link and we don't have a matching regular
-// expression, we should still be categorize it as non ads.
-add_task(async function test_click_non_ads_link() {
-  SearchSERPTelemetry.overrideSearchTelemetryForTests(
-    TEST_PROVIDER_INFO_NO_NON_ADS_REGEXP
-  );
-  await waitForIdle();
-
-  resetTelemetry();
-  let url = getSERPUrl("searchTelemetryAd_components_text.html");
-  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
-  await waitForPageWithAdImpressions();
-
-  // Click a non ad.
-  let pageLoadPromise = BrowserTestUtils.waitForLocationChange(gBrowser);
-  await BrowserTestUtils.synthesizeMouseAtCenter(
-    "#non_ads_link",
-    {},
-    tab.linkedBrowser
-  );
-  await pageLoadPromise;
-
-  assertImpressionEvents([
-    {
-      impression: {
-        provider: "example",
-        tagged: "true",
-        partner_code: "ff",
-        source: "unknown",
-        is_shopping_page: "false",
-        shopping_tab_displayed: "false",
-      },
-      engagements: [
-        {
-          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
-          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
-        },
-      ],
-    },
-  ]);
-
-  BrowserTestUtils.removeTab(tab);
-
-  // Reset state for other tests.
-  SearchSERPTelemetry.overrideSearchTelemetryForTests(TEST_PROVIDER_INFO);
-  await waitForIdle();
-});
-
 // Search box is a special case which has to be tracked in the child process.
 add_task(async function test_click_and_submit_incontent_searchbox() {
   resetTelemetry();
@@ -531,58 +377,6 @@ add_task(async function test_click_carousel_expand() {
   BrowserTestUtils.removeTab(tab);
 });
 
-// Click on an non-ad element while no ads are present.
-add_task(async function test_click_non_ad_with_no_ads() {
-  // Use a provider that doesn't a stored non-ads regexp.
-  SearchSERPTelemetry.overrideSearchTelemetryForTests(
-    TEST_PROVIDER_INFO_NO_NON_ADS_REGEXP
-  );
-  await waitForIdle();
-
-  resetTelemetry();
-
-  let url = getSERPUrl("searchTelemetryAd_searchbox.html");
-  let tab = await BrowserTestUtils.openNewForegroundTab(gBrowser, url);
-  await waitForPageWithAdImpressions();
-
-  let browserLoadedPromise = BrowserTestUtils.browserLoaded(
-    tab.linkedBrowser,
-    true,
-    "https://example.com/hello_world"
-  );
-  await BrowserTestUtils.synthesizeMouseAtCenter(
-    "#non_ads_link",
-    {},
-    tab.linkedBrowser
-  );
-  await browserLoadedPromise;
-
-  assertImpressionEvents([
-    {
-      impression: {
-        provider: "example",
-        tagged: "true",
-        partner_code: "ff",
-        source: "unknown",
-        is_shopping_page: "false",
-        shopping_tab_displayed: "false",
-      },
-      engagements: [
-        {
-          action: SearchSERPTelemetryUtils.ACTIONS.CLICKED,
-          target: SearchSERPTelemetryUtils.COMPONENTS.NON_ADS_LINK,
-        },
-      ],
-    },
-  ]);
-
-  BrowserTestUtils.removeTab(tab);
-
-  // Reset state for other tests.
-  SearchSERPTelemetry.overrideSearchTelemetryForTests(TEST_PROVIDER_INFO);
-  await waitForIdle();
-});
-
 // This test clicks a link that has apostrophes in the both the path and list
 // of query parameters, and uses search telemetry with no nonAdsRegexps defined,
 // which will force us to cache every non ads link in a map and pass it back to

browser/components/search/test/browser/redirect_final.sjs

@@ -0,0 +1,9 @@
+/**
+ * Any copyright is dedicated to the Public Domain.
+ * http://creativecommons.org/publicdomain/zero/1.0/
+ */
+
+function handleRequest(request, response) {
+  response.setStatusLine("1.1", 302, "Found");
+  response.setHeader("Location", "https://example.com/hello_world", false);
+}

browser/components/search/test/browser/redirect_once.sjs

@@ -0,0 +1,9 @@
+/**
+ * Any copyright is dedicated to the Public Domain.
+ * http://creativecommons.org/publicdomain/zero/1.0/
+ */
+
+function handleRequest(request, response) {
+  response.setStatusLine("1.1", 302, "Found");
+  response.setHeader("Location", "redirect_final.sjs", false);
+}

browser/components/search/test/browser/redirect_thrice.sjs

@@ -0,0 +1,9 @@
+/**
+ * Any copyright is dedicated to the Public Domain.
+ * http://creativecommons.org/publicdomain/zero/1.0/
+ */
+
+function handleRequest(request, response) {
+  response.setStatusLine("1.1", 302, "Found");
+  response.setHeader("Location", "redirect_twice.sjs", false);
+}

browser/components/search/test/browser/redirect_twice.sjs

@@ -0,0 +1,9 @@
+/**
+ * Any copyright is dedicated to the Public Domain.
+ * http://creativecommons.org/publicdomain/zero/1.0/
+ */
+
+function handleRequest(request, response) {
+  response.setStatusLine("1.1", 302, "Found");
+  response.setHeader("Location", "redirect_once.sjs", false);
+}

browser/components/search/test/browser/searchTelemetryAd_adsLink_redirect.html

@@ -0,0 +1,12 @@
+<!DOCTYPE html>
+<html lang="en">
+  <head>
+    <meta charset="UTF-8">
+    <meta http-equiv="X-UA-Compatible" content="IE=edge">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Page will do a redirect</title>
+    <meta content="0;url=https://example.com/hello_world" http-equiv="refresh">
+  </head>
+  <body>
+  </body>
+</html>

browser/components/search/test/browser/searchTelemetryAd_components_text.html

@@ -15,7 +15,6 @@
         <a href="https://example.com/ad/1">
           <h2>Example Result</h2>
         </a>
-        <p>Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Cras ac velit sed tellus facilisis euismod.</p>
         <span><a href="https://example.com/ad/2">Ad link that says there are 10 Locations nearby</a></span>
         <div class="multi-col">
           <div>
@@ -47,10 +46,9 @@
       </div>
       <div class="moz_ad">
         <h5 test-label>ad_link</h5>
-        <a href="https://example.com/ad/7">
+        <a id="ad_link_redirect" href="https://example.org/browser/browser/components/search/test/browser/searchTelemetryAd_adsLink_redirect.html">
           <h2>Example Shop</h2>
         </a>
-        <p>Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Cras ac velit sed tellus facilisis euismod.</p>
         <div class="factrow">
           <a href="https://example.com/ad/8">Home Page</a>
           <a href="https://example.com/ad/9">Products</a>
@@ -62,23 +60,23 @@
         <a href="https://example.com/ad/11">
           <h2>Example Shop</h2>
         </a>
-        <p>Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Cras ac velit sed tellus facilisis euismod.</p>
-        <div class="factrow">
-          <a href="https://example.com/ad/12">Home Page</a>
-          <a href="https://example.com/ad/13">Products</a>
-          <a href="https://example.com/ad/14">Sales</a>
-        </div>
       </div>
       <div>
         <h5 test-label>non_ads_link</h5>
         <a id="non_ads_link" href="https://example.com/browser/browser/components/search/test/browser/cacheable.html">
-          <h2>Example of a cached non ad</h2>
-        </a>
-      </div>
-      <div>
-        <h5 test-label>non_ads_link</h5>
+          Example of a cached non ad
+        </a><br />
+        <a id="non_ads_link_redirected" href="https://example.org/browser/browser/components/search/test/browser/searchTelemetryAd_nonAdsLink_redirect.html">
+          Example of a redirected non ad link
+        </a><br />
+        <a id="non_ads_link_redirected_no_top_level" href="#">
+          Example of a redirected non ad link that isn't initially top level loaded
+        </a><br />
+        <a id="non_ads_link_multiple_redirects" href="https://example.com/browser/browser/components/search/test/browser/redirect_thrice.sjs">
+          Example of a redirected non ad link that's redirected multiple times
+        </a><br />
         <a id="non_ads_link_with_special_characters_in_path" href="https://example.com/path'?hello_world&foo=bar's">
-          <h2>Example of a non ad</h2>
+          Example of a non ad with special characters in path
         </a>
       </div>
     </div>
@@ -96,5 +94,19 @@
       </div>
     </div>
   </section>
+  <iframe style="display: none;"></iframe>
+  <script>
+    window.addEventListener("message", (event) => {
+      if (event.origin == "https://example.org") {
+        window.location.href = event.data;
+      }
+    });
+    document.getElementById("non_ads_link_redirected_no_top_level")
+      .addEventListener("click", (event) => {
+        event.preventDefault();
+        let iframe = document.querySelector("iframe");
+        iframe.src = "https://example.org/browser/browser/components/search/test/browser/searchTelemetryAd_nonAdsLink_redirect_nonTopLoaded.html";
+    });
+  </script>
 </body>
 </html>

browser/components/search/test/browser/searchTelemetryAd_nonAdsLink_redirect_nonTopLoaded.html

@@ -0,0 +1,17 @@
+<!DOCTYPE html>
+<html lang="en">
+  <head>
+    <meta charset="UTF-8">
+    <meta http-equiv="X-UA-Compatible" content="IE=edge">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Page will do a redirect without doing it in a top load</title>
+    <!-- <meta content="0;url=https://example.com/hello_world" http-equiv="refresh"> -->
+    <script>
+      let parentWindow = window.parent;
+      let url = "https://example.com/hello_world";
+      parentWindow.postMessage(url, "*");
+    </script>
+  </head>
+  <body>
+  </body>
+</html>

browser/components/search/test/browser/searchTelemetryAd_nonAdsLink_redirect_nonTopLoaded.html^headers^

@@ -0,0 +1,4 @@
+Cache-Control: no-cache, must-revalidate
+Pragma: no-cache
+Expires: Fri, 01 Jan 1990 00:00:00 GMT
+Content-Type: text/html; charset=ISO-8859-1