From dd061a3fd83860f26522f0cc4e44e5be2488614d Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Tue, 19 Apr 2016 09:32:47 +0100
Subject: [PATCH] [Feature] Add min learns to classifiers

---
 src/libserver/cfg_file.h        |  1 +
 src/libserver/cfg_rcl.c         |  6 ++++++
 src/libstat/classifiers/bayes.c | 18 ++++++++++++++++++
 3 files changed, 25 insertions(+)

diff --git a/src/libserver/cfg_file.h b/src/libserver/cfg_file.h
index 67749c276..01183af35 100644
--- a/src/libserver/cfg_file.h
+++ b/src/libserver/cfg_file.h
@@ -153,6 +153,7 @@ struct rspamd_classifier_config {
 	gchar *name;									/**< unique name of classifier							*/
 	guint32 min_tokens;								/**< minimal number of tokens to process classifier 	*/
 	guint32 max_tokens;								/**< maximum number of tokens							*/
+	guint min_learns;								/**< minimum number of learns for each statfile			*/
 	guint flags;
 };
 
diff --git a/src/libserver/cfg_rcl.c b/src/libserver/cfg_rcl.c
index e93ba1d45..db4d26b76 100644
--- a/src/libserver/cfg_rcl.c
+++ b/src/libserver/cfg_rcl.c
@@ -2085,6 +2085,12 @@ rspamd_rcl_config_init (struct rspamd_config *cfg)
 			G_STRUCT_OFFSET (struct rspamd_classifier_config, max_tokens),
 			RSPAMD_CL_FLAG_INT_32,
 			"Maximum count of tokens (words) to be considered for statistics");
+	rspamd_rcl_add_default_handler (sub,
+			"max_tokens",
+			rspamd_rcl_parse_struct_integer,
+			G_STRUCT_OFFSET (struct rspamd_classifier_config, min_learns),
+			RSPAMD_CL_FLAG_UINT,
+			"Minimum number of learns for each statfile to use this classifier");
 	rspamd_rcl_add_default_handler (sub,
 			"backend",
 			rspamd_rcl_parse_struct_string,
diff --git a/src/libstat/classifiers/bayes.c b/src/libstat/classifiers/bayes.c
index 912fa5c2d..867fe4dc6 100644
--- a/src/libstat/classifiers/bayes.c
+++ b/src/libstat/classifiers/bayes.c
@@ -211,6 +211,24 @@ bayes_classify (struct rspamd_classifier * ctx,
 	memset (&cl, 0, sizeof (cl));
 	cl.task = task;
 
+	/* Check min learns */
+	if (ctx->cfg->min_learns > 0) {
+		if (ctx->ham_learns < ctx->cfg->min_learns) {
+			msg_info_task ("skip classification as ham class has not enough "
+					"learns: %ul, %ud required",
+					ctx->ham_learns, ctx->cfg->min_learns);
+
+			return TRUE;
+		}
+		if (ctx->spam_learns < ctx->cfg->min_learns) {
+			msg_info_task ("skip classification as spam class has not enough "
+					"learns: %ul, %ud required",
+					ctx->spam_learns, ctx->cfg->min_learns);
+
+			return TRUE;
+		}
+	}
+
 	for (i = 0; i < tokens->len; i ++) {
 		tok = g_ptr_array_index (tokens, i);
 
-- 
2.39.5