From fa277f69b14446b47a3dfa63ed7fc1a8a16ef7f8 Mon Sep 17 00:00:00 2001
From: Dan Goodliffe <dan.goodliffe@octal.co.uk>
Date: Thu, 9 Jul 2026 13:09:34 +0100
Subject: Include type in existingEntities cache

Fixes the issues of entities having the same value (hash) but occurring
in different places. This is most often when the client states the
virtual host name as the referrer.
---
 src/ingestor.cpp | 14 ++++++++------
 src/ingestor.hpp |  2 +-
 src/logTypes.hpp |  9 +++++++--
 3 files changed, 16 insertions(+), 9 deletions(-)

(limited to 'src')

diff --git a/src/ingestor.cpp b/src/ingestor.cpp
index 41dd24f..ebff00b 100644
--- a/src/ingestor.cpp
+++ b/src/ingestor.cpp
@@ -53,9 +53,11 @@ namespace WebStat {
 			operator()(const std::string_view value) const
 			{
 				return {
-						.hash = makeHash(value),
+						.key = {
+								.hash = makeHash(value),
+								.type = Type,
+						},
 						.id = std::nullopt,
-						.type = Type,
 						.value = value,
 				};
 			}
@@ -331,7 +333,7 @@ namespace WebStat {
 	}
 
 	constexpr auto ENTITY_IDS = std::views::transform([](auto && value) {
-		return std::make_pair(value->hash, *value->id);
+		return std::tie(value->key, *value->id);
 	});
 
 	Ingestor::Job::Result
@@ -610,7 +612,7 @@ namespace WebStat {
 	{
 		auto lockedEntities = existingEntities.shared();
 		for (const auto entity : entities) {
-			if (auto existing = lockedEntities->find(entity->hash); existing != lockedEntities->end()) {
+			if (auto existing = lockedEntities->find(entity->key); existing != lockedEntities->end()) {
 				entity->id = existing->second;
 			}
 		}
@@ -644,7 +646,7 @@ namespace WebStat {
 				}};
 
 		assert(!entity.id);
-		const auto & [typeName, onInsert] = ENTITY_TYPE_VALUES[std::to_underlying(entity.type)];
+		const auto & [typeName, onInsert] = ENTITY_TYPE_VALUES[std::to_underlying(entity.key.type)];
 		bool entityNullDetail = true;
 		std::tie(entity.id, entityNullDetail)
 				= insert<EntityId, bool>(dbconn, SQL::ENTITY_INSERT, SQL::ENTITY_INSERT_OPTS, entity.value, typeName);
@@ -671,7 +673,7 @@ namespace WebStat {
 	void
 	Ingestor::onNewUserAgent(const Entity & entity) const
 	{
-		const auto & [entityHash, entityId, type, value] = entity;
+		const auto & [_, entityId, value] = entity;
 		auto curlOp = curlGetUserAgentDetail(*entityId, value, settings.userAgentAPI.c_str());
 		{
 			std::lock_guard curlOperationsLock {curlOperationsMutex};
diff --git a/src/ingestor.hpp b/src/ingestor.hpp
index e250072..7082dc0 100644
--- a/src/ingestor.hpp
+++ b/src/ingestor.hpp
@@ -101,7 +101,7 @@ namespace WebStat {
 		DB::ConnectionPoolPtr dbpool;
 		mutable Stats stats {};
 
-		ThreadSafeT<std::map<EntityHash, EntityId>> existingEntities;
+		ThreadSafeT<std::map<EntityKey, EntityId>> existingEntities;
 		LineBatch queuedLines, processingLines;
 
 		bool terminated = false;
diff --git a/src/logTypes.hpp b/src/logTypes.hpp
index 7f0473e..d1bda3f 100644
--- a/src/logTypes.hpp
+++ b/src/logTypes.hpp
@@ -38,10 +38,15 @@ namespace WebStat {
 	using EntityId = int32_t;
 	using EntityHash = std::array<uint8_t, MD5_DIGEST_LENGTH>;
 
-	struct Entity {
+	struct EntityKey {
 		EntityHash hash;
-		std::optional<EntityId> id;
 		EntityType type;
+		constexpr auto operator<=>(const EntityKey &) const noexcept = default;
+	};
+
+	struct Entity {
+		EntityKey key;
+		std::optional<EntityId> id;
 		std::string_view value;
 	};
 }
-- 
cgit v1.3