Add custom conversion ID patterns for Verifiable Advertiser Conversions

To increase adoption of Verifiable Advertiser Conversions we provide a mechanism to extract a conversion ID from an advertiser's conversion site w/o the need to modify its DOM. To this end a regex pattern can be configured via the conversions resource to extract the ID from either the URL or the DOM. The default pattern has been made configurable via griffin.
brave · Apr 26, 2021 · 978142a · 978142a
1 parent 1a0d5c4
commit 978142a
Show file tree

Hide file tree

Showing 16 changed files with 641 additions and 37 deletions.
diff --git a/components/brave_ads/test/BUILD.gn b/components/brave_ads/test/BUILD.gn
@@ -63,6 +63,7 @@ source_set("brave_ads_unit_tests") {
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/features/ad_serving/ad_serving_features_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/features/anti_targeting/anti_targeting_features_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/features/bandits/epsilon_greedy_bandit_features_unittest.cc",
+      "//brave/vendor/bat-native-ads/src/bat/ads/internal/features/conversions/conversions_features_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/features/purchase_intent/purchase_intent_features_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/features/text_classification/text_classification_features_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/features/user_activity/user_activity_features_unittest.cc",
@@ -128,6 +129,7 @@ source_set("brave_ads_unit_tests") {
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/resources/behavioral/bandits/epsilon_greedy_bandit_resource_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/resources/behavioral/purchase_intent/purchase_intent_resource_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/resources/contextual/text_classification/text_classification_resource_unittest.cc",
+      "//brave/vendor/bat-native-ads/src/bat/ads/internal/resources/conversions/conversions_resource_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/resources/frequency_capping/anti_targeting_resource_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/security/conversions/conversions_util_unittest.cc",
       "//brave/vendor/bat-native-ads/src/bat/ads/internal/security/crypto_util_unittest.cc",

diff --git a/vendor/bat-native-ads/BUILD.gn b/vendor/bat-native-ads/BUILD.gn
@@ -391,6 +391,8 @@ source_set("ads") {
     "src/bat/ads/internal/features/anti_targeting/anti_targeting_features.h",
     "src/bat/ads/internal/features/bandits/epsilon_greedy_bandit_features.cc",
     "src/bat/ads/internal/features/bandits/epsilon_greedy_bandit_features.h",
+    "src/bat/ads/internal/features/conversions/conversions_features.cc",
+    "src/bat/ads/internal/features/conversions/conversions_features.h",
     "src/bat/ads/internal/features/features.cc",
     "src/bat/ads/internal/features/features.h",
     "src/bat/ads/internal/features/features_util.cc",
@@ -552,6 +554,10 @@ source_set("ads") {
     "src/bat/ads/internal/resources/behavioral/purchase_intent/purchase_intent_resource.h",
     "src/bat/ads/internal/resources/contextual/text_classification/text_classification_resource.cc",
     "src/bat/ads/internal/resources/contextual/text_classification/text_classification_resource.h",
+    "src/bat/ads/internal/resources/conversions/conversion_id_pattern_info.cc",
+    "src/bat/ads/internal/resources/conversions/conversion_id_pattern_info.h",
+    "src/bat/ads/internal/resources/conversions/conversions_resource.cc",
+    "src/bat/ads/internal/resources/conversions/conversions_resource.h",
     "src/bat/ads/internal/resources/country_components.h",
     "src/bat/ads/internal/resources/frequency_capping/anti_targeting_info.cc",
     "src/bat/ads/internal/resources/frequency_capping/anti_targeting_info.h",

diff --git a/vendor/bat-native-ads/data/test/resources/nnqccijfhvzwyrxpxwjrpmynaiazctqb b/vendor/bat-native-ads/data/test/resources/nnqccijfhvzwyrxpxwjrpmynaiazctqb
@@ -0,0 +1,15 @@
+{
+  "locale": "us",
+  "version": 1,
+  "timestamp": "2021-04-26 09:00:00",
+  "conversion_id_patterns": {
+    "https://brave.com/foobar": {
+      "id_pattern": "<div.*id=\"conversion-id\".*>(.*)<\/div>",
+      "search_in": "html"
+    },
+    "https://brave.com/foobar?conversion_id=*": {
+      "id_pattern": "conversion_id\\=(.*)",
+      "search_in": "url"
+    }
+  }
+}
diff --git a/vendor/bat-native-ads/src/bat/ads/internal/ads_impl.cc b/vendor/bat-native-ads/src/bat/ads/internal/ads_impl.cc
@@ -45,6 +45,7 @@
 #include "bat/ads/internal/resources/behavioral/bandits/epsilon_greedy_bandit_resource.h"
 #include "bat/ads/internal/resources/behavioral/purchase_intent/purchase_intent_resource.h"
 #include "bat/ads/internal/resources/contextual/text_classification/text_classification_resource.h"
+#include "bat/ads/internal/resources/conversions/conversions_resource.h"
 #include "bat/ads/internal/resources/country_components.h"
 #include "bat/ads/internal/resources/frequency_capping/anti_targeting_resource.h"
 #include "bat/ads/internal/resources/language_components.h"
@@ -126,6 +127,7 @@ void AdsImpl::ChangeLocale(const std::string& locale) {
   text_classification_resource_->Load();
   purchase_intent_resource_->Load();
   anti_targeting_resource_->Load();
+  conversions_resource_->Load();
 }
 
 void AdsImpl::OnAdsSubdivisionTargetingCodeHasChanged() {
@@ -150,7 +152,8 @@ void AdsImpl::OnHtmlLoaded(const int32_t tab_id,
 
   const std::string original_url = redirect_chain.front();
   ad_transfer_->MaybeTransferAd(tab_id, original_url);
-  conversions_->MaybeConvert(redirect_chain, html);
+  conversions_->MaybeConvert(redirect_chain, html,
+                             conversions_resource_->get());
 }
 
 void AdsImpl::OnTextLoaded(const int32_t tab_id,
@@ -274,6 +277,7 @@ void AdsImpl::OnResourceComponentUpdated(const std::string& id) {
   } else if (kComponentCountryIds.find(id) != kComponentCountryIds.end()) {
     purchase_intent_resource_->Load();
     anti_targeting_resource_->Load();
+    conversions_resource_->Load();
   } else {
     BLOG(0, "Unknown resource for " << id);
   }
@@ -424,6 +428,8 @@ void AdsImpl::set(privacy::TokenGeneratorInterface* token_generator) {
 
   anti_targeting_resource_ = std::make_unique<resource::AntiTargeting>();
 
+  conversions_resource_ = std::make_unique<resource::Conversions>();
+
   ad_targeting_ = std::make_unique<AdTargeting>();
   subdivision_targeting_ =
       std::make_unique<ad_targeting::geographic::SubdivisionTargeting>();

diff --git a/vendor/bat-native-ads/src/bat/ads/internal/ads_impl.h b/vendor/bat-native-ads/src/bat/ads/internal/ads_impl.h
@@ -47,6 +47,7 @@ class SubdivisionTargeting;
 
 namespace resource {
 class AntiTargeting;
+class Conversions;
 class EpsilonGreedyBandit;
 class PurchaseIntent;
 class TextClassification;
@@ -202,6 +203,7 @@ class AdsImpl : public Ads,
   std::unique_ptr<ad_targeting::processor::PurchaseIntent>
       purchase_intent_processor_;
   std::unique_ptr<resource::AntiTargeting> anti_targeting_resource_;
+  std::unique_ptr<resource::Conversions> conversions_resource_;
   std::unique_ptr<ad_targeting::geographic::SubdivisionTargeting>
       subdivision_targeting_;
   std::unique_ptr<AdTargeting> ad_targeting_;

diff --git a/vendor/bat-native-ads/src/bat/ads/internal/conversions/conversions.cc b/vendor/bat-native-ads/src/bat/ads/internal/conversions/conversions.cc
@@ -20,6 +20,7 @@
 #include "bat/ads/internal/database/tables/ad_events_database_table.h"
 #include "bat/ads/internal/database/tables/conversion_queue_database_table.h"
 #include "bat/ads/internal/database/tables/conversions_database_table.h"
+#include "bat/ads/internal/features/conversions/conversions_features.h"
 #include "bat/ads/internal/logging.h"
 #include "bat/ads/internal/time_formatting_util.h"
 #include "bat/ads/internal/url_util.h"
@@ -34,8 +35,8 @@ namespace {
 const int64_t kConvertAfterSeconds =
     base::Time::kHoursPerDay * base::Time::kSecondsPerHour;
 const int64_t kDebugConvertAfterSeconds = 10 * base::Time::kSecondsPerMinute;
-
 const int64_t kExpiredConvertAfterSeconds = 1 * base::Time::kSecondsPerMinute;
+const char kSearchInUrl[] = "url";
 
 bool HasObservationWindowForAdEventExpired(const int observation_window,
                                            const AdEventInfo& ad_event) {
@@ -83,14 +84,41 @@ bool DoesConfirmationTypeMatchConversionType(
   }
 }
 
-std::string ExtractVerifiableConversionIdFromHtml(const std::string& html) {
-  re2::StringPiece text_string_piece(html);
-  RE2 r("<meta.*name=\"ad-conversion-id\".*content=\"(.*)\".*>");
+std::string ExtractConversionIdFromText(
+    const std::string& html,
+    const std::vector<std::string>& redirect_chain,
+    const std::string& conversion_url_pattern,
+    const ConversionIdPatternMap& conversion_id_patterns) {
+  std::string conversion_id;
+  std::string conversion_id_pattern =
+      features::GetGetDefaultConversionIdPattern();
+  std::string text = html;
+
+  const auto iter = conversion_id_patterns.find(conversion_url_pattern);
+  if (iter != conversion_id_patterns.end()) {
+    const ConversionIdPatternInfo conversion_id_pattern_info = iter->second;
+    if (conversion_id_pattern_info.search_in == kSearchInUrl) {
+      const auto url_iter = std::find_if(
+          redirect_chain.begin(), redirect_chain.end(),
+          [=](const std::string& url) {
+            return DoesUrlMatchPattern(url, conversion_url_pattern);
+          });
+
+      if (url_iter == redirect_chain.end()) {
+        return conversion_id;
+      }
+
+      text = *url_iter;
+    }
+
+    conversion_id_pattern = conversion_id_pattern_info.id_pattern;
+  }
 
-  std::string verifiable_conversion_id;
-  RE2::FindAndConsume(&text_string_piece, r, &verifiable_conversion_id);
+  re2::StringPiece text_string_piece(text);
+  RE2 r(conversion_id_pattern);
+  RE2::FindAndConsume(&text_string_piece, r, &conversion_id);
 
-  return verifiable_conversion_id;
+  return conversion_id;
 }
 
 std::set<std::string> GetConvertedCreativeSets(const AdEventList& ad_events) {
@@ -155,8 +183,10 @@ void Conversions::RemoveObserver(ConversionsObserver* observer) {
   observers_.RemoveObserver(observer);
 }
 
-void Conversions::MaybeConvert(const std::vector<std::string>& redirect_chain,
-                               const std::string& html) {
+void Conversions::MaybeConvert(
+    const std::vector<std::string>& redirect_chain,
+    const std::string& html,
+    const ConversionIdPatternMap& conversion_id_patterns) {
   if (!ShouldAllow()) {
     BLOG(1, "Conversions are not allowed");
     return;
@@ -168,7 +198,7 @@ void Conversions::MaybeConvert(const std::vector<std::string>& redirect_chain,
     return;
   }
 
-  CheckRedirectChain(redirect_chain, html);
+  CheckRedirectChain(redirect_chain, html, conversion_id_patterns);
 }
 
 void Conversions::StartTimerIfReady() {
@@ -202,7 +232,8 @@ bool Conversions::ShouldAllow() const {
 
 void Conversions::CheckRedirectChain(
     const std::vector<std::string>& redirect_chain,
-    const std::string& html) {
+    const std::string& html,
+    const ConversionIdPatternMap& conversion_id_patterns) {
   BLOG(1, "Checking URL for conversions");
 
   database::table::AdEvents ad_events_database_table;
@@ -254,8 +285,9 @@ void Conversions::CheckRedirectChain(
           creative_set_ids.insert(ad_event.creative_set_id);
 
           VerifiableConversionInfo verifiable_conversion;
-          verifiable_conversion.id =
-              ExtractVerifiableConversionIdFromHtml(html);
+          verifiable_conversion.id = ExtractConversionIdFromText(
+              html, redirect_chain, conversion.url_pattern,
+              conversion_id_patterns);
           verifiable_conversion.public_key = conversion.advertiser_public_key;
 
           Convert(ad_event, verifiable_conversion);

diff --git a/vendor/bat-native-ads/src/bat/ads/internal/conversions/conversions.h b/vendor/bat-native-ads/src/bat/ads/internal/conversions/conversions.h
@@ -17,6 +17,7 @@
 #include "bat/ads/internal/conversions/conversion_queue_item_info.h"
 #include "bat/ads/internal/conversions/conversions_observer.h"
 #include "bat/ads/internal/conversions/verifiable_conversion_info.h"
+#include "bat/ads/internal/resources/conversions/conversion_id_pattern_info.h"
 #include "bat/ads/internal/security/conversions/verifiable_conversion_envelope_info.h"
 #include "bat/ads/internal/timer.h"
 
@@ -34,7 +35,8 @@ class Conversions {
   bool ShouldAllow() const;
 
   void MaybeConvert(const std::vector<std::string>& redirect_chain,
-                    const std::string& html);
+                    const std::string& html,
+                    const ConversionIdPatternMap& conversion_id_patterns);
 
   void StartTimerIfReady();
 
@@ -44,7 +46,8 @@ class Conversions {
   Timer timer_;
 
   void CheckRedirectChain(const std::vector<std::string>& redirect_chain,
-                          const std::string& html);
+                          const std::string& html,
+                          const ConversionIdPatternMap& conversion_id_patterns);
 
   void Convert(const AdEventInfo& ad_event,
                const VerifiableConversionInfo& verifiable_conversion);