From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Thu, 15 Nov 2018 15:43:08 +0000 (+0000)
Subject: [Minor] Check for empty bayes results
X-Git-Tag: 1.8.2~24
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=b5d08799fcadfb314682f4fd29bcb44cce56c802;p=rspamd.git

[Minor] Check for empty bayes results
---

diff --git a/src/libstat/classifiers/bayes.c b/src/libstat/classifiers/bayes.c
index 1898df4fe..1a28a42cb 100644
--- a/src/libstat/classifiers/bayes.c
+++ b/src/libstat/classifiers/bayes.c
@@ -329,6 +329,21 @@ bayes_classify (struct rspamd_classifier * ctx,
 		bayes_classify_token (ctx, tok, &cl);
 	}
 
+	if (cl.processed_tokens == 0) {
+		msg_info_bayes ("no tokens found in bayes database, ignore stats");
+
+		return TRUE;
+	}
+
+	if (ctx->cfg->min_tokens > 0 &&
+		cl.text_tokens < (gint)(ctx->cfg->min_tokens * 0.1)) {
+		msg_info_bayes ("ignore bayes probability since we have "
+						"too few text tokens: %uL, at least %d is required",
+						cl.text_tokens, (gint)(ctx->cfg->min_tokens * 0.1));
+
+		return TRUE;
+	}
+
 	h = 1 - inv_chi_square (task, cl.spam_prob, cl.processed_tokens);
 	s = 1 - inv_chi_square (task, cl.ham_prob, cl.processed_tokens);
 
@@ -368,17 +383,6 @@ bayes_classify (struct rspamd_classifier * ctx,
 		}
 	}
 
-	if (ctx->cfg->min_tokens > 0 &&
-			cl.text_tokens < (gint)(ctx->cfg->min_tokens * 0.1)) {
-		msg_info_bayes ("ignore bayes probability %.2f since we have "
-				"too few text tokens: %uL, at least %d is required",
-				final_prob,
-				cl.text_tokens,
-				(gint)(ctx->cfg->min_tokens * 0.1));
-
-		return TRUE;
-	}
-
 	pprob = rspamd_mempool_alloc (task->task_pool, sizeof (*pprob));
 	*pprob = final_prob;
 	rspamd_mempool_set_variable (task->task_pool, "bayes_prob", pprob, NULL);