From 9fe90fcd255b044358998355dea37d43c7136d57 Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@rspamd.com>
Date: Wed, 4 Sep 2024 10:53:55 +0100
Subject: [PATCH] [Rework] Change the logic of skipping symbols

We now do not skip pre/post filters even if the task result has reached
threshold.
---
 src/libmime/lang_detection.c                |  2 +-
 src/libserver/symcache/symcache_runtime.cxx | 43 ++++++++++++---------
 src/libserver/symcache/symcache_runtime.hxx |  7 +++-
 3 files changed, 32 insertions(+), 20 deletions(-)

diff --git a/src/libmime/lang_detection.c b/src/libmime/lang_detection.c
index 4796e4834..6e180ea66 100644
--- a/src/libmime/lang_detection.c
+++ b/src/libmime/lang_detection.c
@@ -1828,7 +1828,7 @@ rspamd_language_detector_detect(struct rspamd_task *task,
 	unsigned int cand_len;
 	enum rspamd_language_category cat;
 	struct rspamd_lang_detector_res *cand;
-	enum rspamd_language_detected_type r;
+	enum rspamd_language_detected_type r = rs_detect_none;
 	struct rspamd_frequency_sort_cbdata cbd;
 	/* Check if we have sorted candidates based on frequency */
 	gboolean frequency_heuristic_applied = FALSE, ret = FALSE, internal_heuristic_applied = FALSE;
diff --git a/src/libserver/symcache/symcache_runtime.cxx b/src/libserver/symcache/symcache_runtime.cxx
index dc7066b32..ddfbdf1ae 100644
--- a/src/libserver/symcache/symcache_runtime.cxx
+++ b/src/libserver/symcache/symcache_runtime.cxx
@@ -333,8 +333,8 @@ auto symcache_runtime::process_pre_postfilters(struct rspamd_task *task,
 		 */
 		if (stage != RSPAMD_TASK_STAGE_IDEMPOTENT &&
 			!(item->flags & SYMBOL_TYPE_IGNORE_PASSTHROUGH)) {
-			if (check_metric_limit(task)) {
-				msg_debug_cache_task_lambda("task has already the result being set, ignore further checks");
+			if (check_process_status(task) == check_status::passthrough) {
+				msg_debug_cache_task_lambda("task has already the passthrough result being set, ignore further checks");
 
 				return true;
 			}
@@ -407,13 +407,20 @@ auto symcache_runtime::process_filters(struct rspamd_task *task, symcache &cache
 			break;
 		}
 
+		auto check_result = check_process_status(task);
+
 		if (!(item->flags & (SYMBOL_TYPE_FINE | SYMBOL_TYPE_IGNORE_PASSTHROUGH))) {
-			if (has_passtrough || check_metric_limit(task)) {
-				msg_debug_cache_task_lambda("task has already the result being set, ignore further checks");
+			if (has_passtrough || check_result == check_status::passthrough) {
+				msg_debug_cache_task_lambda("task has already the passthrough result being set, ignore further checks");
 				has_passtrough = true;
 				/* Skip this item */
 				continue;
 			}
+			else if (check_result == check_status::limit_reached) {
+				msg_debug_cache_task_lambda("task has already the limit reached result being set, ignore further checks");
+				/* Skip this item */
+				continue;
+			}
 		}
 
 		auto dyn_item = &dynamic_items[idx];
@@ -531,19 +538,8 @@ auto symcache_runtime::process_symbol(struct rspamd_task *task, symcache &cache,
 	return true;
 }
 
-auto symcache_runtime::check_metric_limit(struct rspamd_task *task) -> bool
+auto symcache_runtime::check_process_status(struct rspamd_task *task) -> symcache_runtime::check_status
 {
-	if (task->flags & RSPAMD_TASK_FLAG_PASS_ALL) {
-		return false;
-	}
-
-	/* Check score limit */
-	if (!std::isnan(lim)) {
-		if (task->result->score > lim) {
-			return true;
-		}
-	}
-
 	if (task->result->passthrough_result != nullptr) {
 		/* We also need to check passthrough results */
 		auto *pr = task->result->passthrough_result;
@@ -563,11 +559,22 @@ auto symcache_runtime::check_metric_limit(struct rspamd_task *task) -> bool
 			}
 
 			/* Immediately stop on non least passthrough action */
-			return true;
+			return check_status::passthrough;
 		}
 	}
 
-	return false;
+	if (task->flags & RSPAMD_TASK_FLAG_PASS_ALL) {
+		return check_status::allow;
+	}
+
+	/* Check score limit */
+	if (!std::isnan(lim)) {
+		if (task->result->score > lim) {
+			return check_status::limit_reached;
+		}
+	}
+
+	return check_status::allow;
 }
 
 auto symcache_runtime::check_item_deps(struct rspamd_task *task, symcache &cache, cache_item *item,
diff --git a/src/libserver/symcache/symcache_runtime.hxx b/src/libserver/symcache/symcache_runtime.hxx
index 7e4a41269..d1dc7ac24 100644
--- a/src/libserver/symcache/symcache_runtime.hxx
+++ b/src/libserver/symcache/symcache_runtime.hxx
@@ -60,6 +60,11 @@ class symcache_runtime {
 		enabled = 1,
 		disabled = 2,
 	} slow_status;
+	enum class check_status {
+		allow,
+		limit_reached,
+		passthrough,
+	};
 	bool profile;
 
 	double profile_start;
@@ -77,7 +82,7 @@ class symcache_runtime {
 	/* Specific stages of the processing */
 	auto process_pre_postfilters(struct rspamd_task *task, symcache &cache, int start_events, unsigned int stage) -> bool;
 	auto process_filters(struct rspamd_task *task, symcache &cache, int start_events) -> bool;
-	auto check_metric_limit(struct rspamd_task *task) -> bool;
+	auto check_process_status(struct rspamd_task *task) -> check_status;
 	auto check_item_deps(struct rspamd_task *task, symcache &cache, cache_item *item,
 						 cache_dynamic_item *dyn_item, bool check_only) -> bool;
 
-- 
2.39.5