Stop redacting dates in GA4 tracking on GOVUK search pages

CHANGELOG.md

@@ -10,6 +10,7 @@
 ## Unreleased
 
 * Set aria-label text in govuk_logo.html to "GOV.UK" ([PR #4217](https://github.com/alphagov/govuk_publishing_components/pull/4217))
+* Stop redacting dates in GA4 tracking on GOVUK search pages ([PR #4223](https://github.com/alphagov/govuk_publishing_components/pull/4223))
 
 ## 43.1.1
 

app/assets/javascripts/govuk_publishing_components/analytics-ga4/ga4-core.js

@@ -295,7 +295,7 @@ window.GOVUK.analyticsGa4 = window.GOVUK.analyticsGa4 || {};
         var PIIRemover = new window.GOVUK.analyticsGa4.PIIRemover()
         searchTerm = searchTerm.replace(/\++|(%2B)+/gm, ' ') // Turn + characters or unicode + characters (%2B) into a space.
         searchTerm = window.GOVUK.analyticsGa4.core.trackFunctions.removeLinesAndExtraSpaces(searchTerm)
-        searchTerm = PIIRemover.stripPIIWithOverride(searchTerm, true, true)
+        searchTerm = PIIRemover.stripPIIWithOverride(searchTerm, false, true)
         searchTerm = searchTerm.toLowerCase()
         return searchTerm
       }

app/assets/javascripts/govuk_publishing_components/analytics-ga4/ga4-page-views.js

@@ -78,7 +78,9 @@ window.GOVUK.analyticsGa4.analyticsModules = window.GOVUK.analyticsGa4.analytics
     },
 
     getLocation: function () {
-      return this.PIIRemover.stripPIIWithOverride(this.stripGaParam(document.location.href), true, true)
+      // We don't want to remove dates on search pages.
+      var stripDates = !this.getSearchTerm()
+      return this.PIIRemover.stripPIIWithOverride(this.stripGaParam(document.location.href), stripDates, true)
     },
 
     getSearchTerm: function () {
@@ -110,8 +112,10 @@ window.GOVUK.analyticsGa4.analyticsModules = window.GOVUK.analyticsGa4.analytics
     getQueryString: function () {
       var queryString = window.GOVUK.analyticsGa4.core.trackFunctions.getSearch()
       if (queryString) {
+        // We don't want to remove dates on search pages.
+        var stripDates = !this.getSearchTerm()
         queryString = this.stripGaParam(queryString)
-        queryString = this.PIIRemover.stripPIIWithOverride(queryString, true, true)
+        queryString = this.PIIRemover.stripPIIWithOverride(queryString, stripDates, true)
         queryString = queryString.substring(1) // removes the '?' character from the start.
         return queryString
       }

spec/javascripts/govuk_publishing_components/analytics-ga4/ga4-core.spec.js

@@ -297,7 +297,7 @@ describe('GA4 core', function () {
 
     it('standardises search terms for consistency across trackers', function () {
       var searchTerm = 'NO UPPERCASE, NO %2B plus + signs, NO PII [email protected] SW1A 2AA 1st Jan 1990 and    NO    extra    spaces \n \r      '
-      var expected = 'no uppercase, no plus signs, no pii [email] [postcode] [date] and no extra spaces'
+      var expected = 'no uppercase, no plus signs, no pii [email] [postcode] 1st jan 1990 and no extra spaces'
 
       searchTerm = GOVUK.analyticsGa4.core.trackFunctions.standardiseSearchTerm(searchTerm)
       expect(searchTerm).toEqual(expected)

spec/javascripts/govuk_publishing_components/analytics-ga4/ga4-page-views.spec.js

@@ -375,6 +375,34 @@ describe('Google Tag Manager page view tracking', function () {
     expect(window.dataLayer[0]).toEqual(expected)
   })
 
+  it('doesn\'t remove date pii from the location if on a search page', function () {
+    spyOn(GOVUK.analyticsGa4.core.trackFunctions, 'getSearch').and.returnValue('?keywords=2020-01-01')
+
+    // We can't spy on location, so instead we use an anchor link to change the URL temporarily
+
+    // Reset the URL so we can build the expected link string
+    var linkForURLMock = document.createElement('a')
+    linkForURLMock.href = '#'
+    linkForURLMock.click()
+    var location = document.location.href
+
+    expected.page_view.location = location + '[email]/[postcode]?keywords=2020-01-01'
+    expected.page_view.search_term = '2020-01-01'
+    expected.page_view.query_string = 'keywords=2020-01-01'
+
+    // Add personally identifiable information to the current page location
+    linkForURLMock.href = '#[email protected]/SW12AA?keywords=2020-01-01'
+    linkForURLMock.click()
+
+    GOVUK.analyticsGa4.analyticsModules.PageViewTracker.init()
+
+    // Reset the page location for other tests
+    linkForURLMock.href = '#'
+    linkForURLMock.click()
+
+    expect(window.dataLayer[0]).toEqual(expected)
+  })
+
   it('can redact ga parameters as expected', function () {
     var url = 'https://www.gov.uk/welfare?'
     var param1 = '_ga=2.237932419.220854294.1692605006-1618308030.1687790776'
@@ -558,20 +586,28 @@ describe('Google Tag Manager page view tracking', function () {
     expect(window.dataLayer[0]).toEqual(expected)
   })
 
+  it('correctly sets the query_string parameter without PII date redaction if a search term exists', function () {
+    spyOn(GOVUK.analyticsGa4.core.trackFunctions, 'getSearch').and.returnValue('?keywords=hello&query2=world&[email protected]&postcode=SW12AA&birthday=1990-01-01&_ga=19900101.567&_gl=19900101.567')
+    expected.page_view.search_term = 'hello'
+    expected.page_view.query_string = 'keywords=hello&query2=world&email=[email]&postcode=[postcode]&birthday=1990-01-01&_ga=[id]&_gl=[id]'
+    GOVUK.analyticsGa4.analyticsModules.PageViewTracker.init()
+    expect(window.dataLayer[0]).toEqual(expected)
+  })
+
   describe('search_term parameter', function () {
     describe('using the query parameter "keywords"', function () {
       it('correctly sets the parameter using ?keywords= with PII values redacted', function () {
         spyOn(GOVUK.analyticsGa4.core.trackFunctions, 'getSearch').and.returnValue('[email protected]+SW12AA+1990-01-01&another=one')
-        expected.page_view.query_string = 'keywords=hello+world+[email]+[postcode]+[date]&another=one'
-        expected.page_view.search_term = 'hello world [email] [postcode] [date]'
+        expected.page_view.query_string = 'keywords=hello+world+[email]+[postcode]+1990-01-01&another=one'
+        expected.page_view.search_term = 'hello world [email] [postcode] 1990-01-01'
         GOVUK.analyticsGa4.analyticsModules.PageViewTracker.init()
         expect(window.dataLayer[0]).toEqual(expected)
       })
 
-      it('correctly sets the parameter using &keywords= with PII values redacted', function () {
+      it('correctly sets the parameter using &keywords= with PII values redacted except for dates', function () {
         spyOn(GOVUK.analyticsGa4.core.trackFunctions, 'getSearch').and.returnValue('?test=true&[email protected]+SW12AA+1990-01-01&another=one')
-        expected.page_view.query_string = 'test=true&keywords=hello+world+[email]+[postcode]+[date]&another=one'
-        expected.page_view.search_term = 'hello world [email] [postcode] [date]'
+        expected.page_view.query_string = 'test=true&keywords=hello+world+[email]+[postcode]+1990-01-01&another=one'
+        expected.page_view.search_term = 'hello world [email] [postcode] 1990-01-01'
         GOVUK.analyticsGa4.analyticsModules.PageViewTracker.init()
         expect(window.dataLayer[0]).toEqual(expected)
       })
@@ -620,9 +656,9 @@ describe('Google Tag Manager page view tracking', function () {
         document.body.removeChild(div)
       })
 
-      it('correctly sets the parameter with PII values redacted', function () {
+      it('correctly sets the parameter with PII values redacted, except dates', function () {
         div.setAttribute('data-ga4-search-query', 'hello world [email protected] SW12AA 1990-01-01')
-        expected.page_view.search_term = 'hello world [email] [postcode] [date]'
+        expected.page_view.search_term = 'hello world [email] [postcode] 1990-01-01'
         GOVUK.analyticsGa4.analyticsModules.PageViewTracker.init()
         expect(window.dataLayer[0]).toEqual(expected)
       })