From: Dmitriy Alekseev <1865999+dragoangel@users.noreply.github.com>
Date: Tue, 24 Feb 2026 17:19:40 +0000 (+0100)
Subject: fix: properly set bayes class labels to S and H for class spam and ham class names... 
X-Git-Tag: 4.0.0~66^2~1
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=d8fc7df68c14953eb4d737d9b863ed6aece366cc;p=thirdparty%2Frspamd.git

fix: properly set bayes class labels to S and H for class spam and ham class names, adjust bayes expiry to write occurrences as spam and ham instead of S and H

Signed-off-by: Dmitriy Alekseev <1865999+dragoangel@users.noreply.github.com>
---

diff --git a/lualib/redis_scripts/bayes_learn.lua b/lualib/redis_scripts/bayes_learn.lua
index 3f8ee1cbc7..8fa984e08b 100644
--- a/lualib/redis_scripts/bayes_learn.lua
+++ b/lualib/redis_scripts/bayes_learn.lua
@@ -18,17 +18,17 @@ if ARGV[5] then
   text_tokens = cmsgpack.unpack(ARGV[5])
 end
 
--- Handle backward compatibility for boolean values
-if class_label == 'true' then
-  class_label = 'S' -- spam
-elseif class_label == 'false' then
-  class_label = 'H' -- ham
+-- Handle RS_<ID> HASH booleans and full class name strings for backward compatibility
+if class_label == 'true' or class_label == 'spam' then
+  class_label = 'S'
+elseif class_label == 'false' or class_label == 'ham' then
+  class_label = 'H'
 end
 
 local hash_key = class_label
 local learned_key = 'learns_' .. string.lower(class_label)
 
--- Handle legacy keys for backward compatibility
+-- Handle RS HASH keys for backward compatibility
 if class_label == 'S' then
   learned_key = 'learns_spam'
 elseif class_label == 'H' then
diff --git a/src/libstat/backends/redis_backend.cxx b/src/libstat/backends/redis_backend.cxx
index 8489947e36..9e2867f263 100644
--- a/src/libstat/backends/redis_backend.cxx
+++ b/src/libstat/backends/redis_backend.cxx
@@ -191,6 +191,10 @@ rspamd_redis_stat_quark(void)
 
 /*
  * Get the class label for a statfile (for multi-class support)
+ * Returns the Redis hash field name to use for this class.  "spam" and "ham"
+ * always map to the legacy single-char labels "S" and "H" so that existing
+ * Redis token data (stored under those field names) remains readable after a
+ * config migration from `spam = true/false` to explicit `class = "spam"/"ham"`.
  */
 static const char *
 get_class_label(struct rspamd_statfile_config *stcf)
@@ -201,16 +205,25 @@ get_class_label(struct rspamd_statfile_config *stcf)
 		if (label) {
 			return label;
 		}
-		/* If no label mapping found, use class name directly */
-		return stcf->class_name;
+		/* No mapping found — fall through to backward-compat checks below */
 	}
 
-	/* For multiclass without explicit label mapping, use class_name directly */
-	if (stcf->class_name && !stcf->is_spam_converted) {
-		return stcf->class_name;
+	if (stcf->class_name) {
+		/* Map canonical binary class names to their legacy Redis field names */
+		if (strcmp(stcf->class_name, "spam") == 0) {
+			return "S";
+		}
+		if (strcmp(stcf->class_name, "ham") == 0) {
+			return "H";
+		}
+
+		/* True multiclass name (not spam/ham) — use the class name as the field */
+		if (!stcf->is_spam_converted) {
+			return stcf->class_name;
+		}
 	}
 
-	/* Fallback to legacy binary classification */
+	/* Final fallback: legacy binary is_spam flag */
 	return stcf->is_spam ? "S" : "H";
 }
 
diff --git a/src/plugins/lua/bayes_expiry.lua b/src/plugins/lua/bayes_expiry.lua
index 69b0f93266..10ac8d2a30 100644
--- a/src/plugins/lua/bayes_expiry.lua
+++ b/src/plugins/lua/bayes_expiry.lua
@@ -73,6 +73,8 @@ local function check_redis_classifier(cls, cfg)
 
       if class_name then
         class_symbols[class_name] = symbol
+        logger.debugm(N, rspamd_config,
+          'check_statfile_table: found class_name=%s and its symbol=%s', class_name, symbol)
       end
     end
 
@@ -284,6 +286,12 @@ local expiry_script = [[
       -- Parse hash data into class counts
       for i = 1, #hash_data, 2 do
         local class_label = hash_data[i]
+        -- Handle RS_<ID> HASH short class names for backward compatibility
+        if class_label == 'S' then
+          class_label = 'spam'
+        elseif class_label == 'H' then
+          class_label = 'ham'
+        end
         local count = tonumber(hash_data[i + 1]) or 0
         class_counts[class_label] = count
         total = total + count