Match consent vendors by host+path instead of host only

Allow URL matchers to include path segments so that vendors
sharing a domain (e.g. google.com/maps/embed vs google.com)
can be distinguished. The old consent_vendor_host_matchers
setter is preserved for backwards compatibility, automatically
rewriting trailing `$` anchors to `/`.

REDMINE-21240

Author: tf

entry_types/scrolled/app/helpers/pageflow_scrolled/editor/entry_json_seed_helper.rb

@@ -11,7 +11,7 @@ def scrolled_entry_editor_json_seed(json, scrolled_entry)
         entry_config = Pageflow.config_for(scrolled_entry)
 
         scrolled_entry_editor_legacy_typography_variants_seed(json, entry_config)
-        scrolled_entry_editor_consent_vendor_host_matchers_seed(json, entry_config)
+        scrolled_entry_editor_consent_vendor_url_matchers_seed(json, entry_config)
 
         scrolled_entry_json_seed(json,
                                  scrolled_entry,
@@ -32,10 +32,10 @@ def scrolled_entry_editor_legacy_typography_variants_seed(json, entry_config)
         )
       end
 
-      def scrolled_entry_editor_consent_vendor_host_matchers_seed(json, entry_config)
-        json.consent_vendor_host_matchers(
+      def scrolled_entry_editor_consent_vendor_url_matchers_seed(json, entry_config)
+        json.consent_vendor_url_matchers(
           entry_config
-            .consent_vendor_host_matchers
+            .consent_vendor_url_matchers
             .transform_keys { |regexp| regexp.inspect[1..-2] }
         )
       end

entry_types/scrolled/lib/pageflow_scrolled/configuration.rb

@@ -80,18 +80,32 @@ class Configuration
     # @since 16.1
     attr_reader :content_element_consent_vendors
 
-    # Mapping from URL hosts to consent vendor names. Used for iframe
-    # embed opt-in.
+    # Mapping from URL host+path patterns to consent vendor names.
+    # Used for iframe embed opt-in.
     #
     # @exmaple
     #
-    #     entry_type_config.consent_vendor_host_matchers = {
-    #       /\.some-vendor\.com$/ => 'someVendor'
+    #     entry_type_config.consent_vendor_url_matchers = {
+    #       /\.some-vendor\.com\// => 'someVendor',
+    #       /google\.com\/maps\/embed/ => 'googleMaps'
     #     }
     #
     # @return [Hash<RegExp, String>]
+    # @since edge
+    attr_accessor :consent_vendor_url_matchers
+
+    # @deprecated Use {#consent_vendor_url_matchers=} instead.
+    #   Rewrites trailing `$` in patterns to `/` and merges
+    #   into {#consent_vendor_url_matchers}.
     # @since 16.1
-    attr_accessor :consent_vendor_host_matchers
+    def consent_vendor_host_matchers=(matchers)
+      rewritten = matchers.transform_keys do |key|
+        source = key.source.sub(/\$$/, '/')
+        Regexp.new(source, key.options)
+      end
+
+      @consent_vendor_url_matchers.merge!(rewritten)
+    end
 
     # Migrate typography variants to palette colors. Before palette
     # colors for text blocks and headings were introduced, it was
@@ -155,7 +169,7 @@ def initialize(*)
       @additional_frontend_seed_data = AdditionalSeedData.new
       @additional_theme_assets = AdditionalThemeAssets.new
       @content_element_consent_vendors = ContentElementConsentVendors.new
-      @consent_vendor_host_matchers = {}
+      @consent_vendor_url_matchers = {}
 
       @legacy_typography_variants = {}
     end

entry_types/scrolled/lib/pageflow_scrolled/plugin.rb

@@ -227,10 +227,10 @@ def configure(config)
       return unless configuration['requireConsent']
 
       uri = URI.parse(configuration['source'])
-      host_matchers = Pageflow.config_for(entry).consent_vendor_host_matchers
+      url_matchers = Pageflow.config_for(entry).consent_vendor_url_matchers
 
-      host_matchers.detect { |matcher, _|
-        uri.host =~ matcher
+      url_matchers.detect { |matcher, _|
+        (uri.host + uri.path) =~ matcher
       }&.last
     rescue URI::InvalidURIError
       nil

entry_types/scrolled/package/spec/editor/models/ConsentVendors-spec.js

@@ -2,10 +2,10 @@ import {ConsentVendors} from 'editor/models/ConsentVendors';
 
 describe('ConsentVendors', () => {
   describe('fromUrl', () => {
-    it('detects vendor from seed data host matcher', () => {
+    it('detects vendor from seed data url matcher', () => {
       const consentVendors = new ConsentVendors({
-        hostMatchers: {
-          '\\.some-vendor.com$': 'someVendor'
+        urlMatchers: {
+          '\\.some-vendor\\.com/': 'someVendor'
         }
       });
 
@@ -14,5 +14,18 @@ describe('ConsentVendors', () => {
       expect(consentVendors.fromUrl('https://other.com/abc'))
         .toBeUndefined();
     });
+
+    it('detects vendor from path-based url matcher', () => {
+      const consentVendors = new ConsentVendors({
+        urlMatchers: {
+          'google\\.com/maps/embed': 'googleMaps'
+        }
+      });
+
+      expect(consentVendors.fromUrl('https://www.google.com/maps/embed?pb=1234'))
+        .toEqual('googleMaps');
+      expect(consentVendors.fromUrl('https://www.google.com/search?q=test'))
+        .toBeUndefined();
+    });
   })
 })

entry_types/scrolled/package/src/editor/models/ConsentVendors.js

@@ -1,10 +1,10 @@
-export function ConsentVendors({hostMatchers}) {
+export function ConsentVendors({urlMatchers}) {
   return {
     fromUrl(url) {
       url = new URL(url);
 
-      return Object.entries(hostMatchers).find(([matcher]) =>
-        new RegExp(matcher).test(url.host)
+      return Object.entries(urlMatchers).find(([matcher]) =>
+        new RegExp(matcher).test(url.host + url.pathname)
       )?.[1];
     }
   }

entry_types/scrolled/package/src/editor/models/ScrolledEntry/index.js

@@ -47,7 +47,7 @@ const defaultAspectRatios = [{
 
 export const ScrolledEntry = Entry.extend({
   setupFromEntryTypeSeed(seed) {
-    this.consentVendors = new ConsentVendors({hostMatchers: seed.consentVendorHostMatchers});
+    this.consentVendors = new ConsentVendors({urlMatchers: seed.consentVendorUrlMatchers});
 
     this.contentElements = new ContentElementsCollection(seed.collections.contentElements);
     this.sections = new SectionsCollection(seed.collections.sections,

entry_types/scrolled/spec/helpers/pageflow_scrolled/editor/entry_json_seed_helper_spec.rb

@@ -115,11 +115,11 @@ def render(helper, entry)
                                          })
         end
 
-        it 'renders consent vendor host matchers' do
+        it 'renders consent vendor url matchers' do
           pageflow_configure do |config|
             config.for_entry_type(PageflowScrolled.entry_type) do |entry_type_config|
-              entry_type_config.consent_vendor_host_matchers = {
-                /\.some-vendor\.com$/ => 'someVendor'
+              entry_type_config.consent_vendor_url_matchers = {
+                /\.some-vendor\.com\// => 'someVendor'
               }
             end
           end
@@ -128,8 +128,8 @@ def render(helper, entry)
 
           result = render(helper, entry)
 
-          expect(result).to include_json(consentVendorHostMatchers: {
-                                           '\\.some-vendor\\.com$' => 'someVendor'
+          expect(result).to include_json(consentVendorUrlMatchers: {
+                                           '\\.some-vendor\\.com\\/' => 'someVendor'
                                          })
         end
 

entry_types/scrolled/spec/pageflow_scrolled/plugin_spec.rb

@@ -41,8 +41,8 @@ def request(uri = 'https://example.com/some-entry')
       it 'returns nil if consent not required' do
         pageflow_configure do |config|
           config.for_entry_type(PageflowScrolled.entry_type) do |entry_type_config|
-            entry_type_config.consent_vendor_host_matchers = {
-              /\.typeform\.com$/ => 'typeform'
+            entry_type_config.consent_vendor_url_matchers = {
+              /\.typeform\.com\// => 'typeform'
             }
           end
         end
@@ -59,6 +59,46 @@ def request(uri = 'https://example.com/some-entry')
       end
 
       it 'detects vendor from source when consent is required' do
+        pageflow_configure do |config|
+          config.for_entry_type(PageflowScrolled.entry_type) do |entry_type_config|
+            entry_type_config.consent_vendor_url_matchers = {
+              /\.typeform\.com\// => 'typeform'
+            }
+          end
+        end
+
+        result = Plugin::IFRAME_EMBED_CONSENT_VENDOR.call(
+          entry: create(:published_entry, type_name: 'scrolled'),
+          configuration: {
+            'requireConsent' => true,
+            'source' => 'https://foo.typeform.com/to/1234'
+          }
+        )
+
+        expect(result).to eq('typeform')
+      end
+
+      it 'detects vendor from path-based url matcher' do
+        pageflow_configure do |config|
+          config.for_entry_type(PageflowScrolled.entry_type) do |entry_type_config|
+            entry_type_config.consent_vendor_url_matchers = {
+              /google\.com\/maps\/embed/ => 'googleMaps'
+            }
+          end
+        end
+
+        result = Plugin::IFRAME_EMBED_CONSENT_VENDOR.call(
+          entry: create(:published_entry, type_name: 'scrolled'),
+          configuration: {
+            'requireConsent' => true,
+            'source' => 'https://www.google.com/maps/embed?pb=1234'
+          }
+        )
+
+        expect(result).to eq('googleMaps')
+      end
+
+      it 'supports deprecated consent_vendor_host_matchers setter' do
         pageflow_configure do |config|
           config.for_entry_type(PageflowScrolled.entry_type) do |entry_type_config|
             entry_type_config.consent_vendor_host_matchers = {