From 091e84951a2b032bb2930b300ffe43eaf01a304e Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@rambler-co.ru>
Date: Wed, 13 Jul 2011 13:07:45 +0400
Subject: [PATCH] Add validity detector for statfiles inside classifier. Add
 euristic to detect spam/ham classes based on statfile symbol.

---
 src/cfg_file.h  |  1 +
 src/cfg_utils.c | 56 +++++++++++++++++++++++++++++++++++++++++++++++++
 src/cfg_xml.c   |  4 ++--
 3 files changed, 59 insertions(+), 2 deletions(-)

diff --git a/src/cfg_file.h b/src/cfg_file.h
index d15923639..64a46d149 100644
--- a/src/cfg_file.h
+++ b/src/cfg_file.h
@@ -430,6 +430,7 @@ gboolean parse_normalizer (struct config_file *cfg, struct statfile *st, const g
 gboolean read_xml_config (struct config_file *cfg, const gchar *filename);
 gboolean check_modules_config (struct config_file *cfg);
 void insert_classifier_symbols (struct config_file *cfg);
+gboolean check_classifier_statfiles (struct classifier_config *cf);
 
 struct classifier_config* find_classifier_conf (struct config_file *cfg, const gchar *name);
 
diff --git a/src/cfg_utils.c b/src/cfg_utils.c
index 6bd16d620..56202d4f1 100644
--- a/src/cfg_utils.c
+++ b/src/cfg_utils.c
@@ -1044,6 +1044,62 @@ find_classifier_conf (struct config_file *cfg, const gchar *name)
 	return NULL;
 }
 
+gboolean
+check_classifier_statfiles (struct classifier_config *cf)
+{
+	struct statfile                *st;
+	gboolean                        has_other = FALSE, cur_class;
+	GList                          *cur;
+
+	/* First check classes directly */
+	cur = cf->statfiles;
+	while (cur) {
+		st = cur->data;
+		if (!has_other) {
+			cur_class = st->is_spam;
+			has_other = TRUE;
+		}
+		else {
+			if (cur_class != st->is_spam) {
+				return TRUE;
+			}
+		}
+
+		cur = g_list_next (cur);
+	}
+
+	if (!has_other) {
+		/* We have only one statfile */
+		return FALSE;
+	}
+	/* We have not detected any statfile that has different class, so turn on euristic based on symbol's name */
+	has_other = FALSE;
+	cur = cf->statfiles;
+	while (cur) {
+		st = cur->data;
+		if (rspamd_strncasestr (st->symbol, "spam", -1) != NULL) {
+			st->is_spam = TRUE;
+		}
+		else if (rspamd_strncasestr (st->symbol, "ham", -1) != NULL) {
+			st->is_spam = FALSE;
+		}
+
+		if (!has_other) {
+			cur_class = st->is_spam;
+			has_other = TRUE;
+		}
+		else {
+			if (cur_class != st->is_spam) {
+				return TRUE;
+			}
+		}
+
+		cur = g_list_next (cur);
+	}
+
+	return FALSE;
+}
+
 /*
  * vi:ts=4
  */
diff --git a/src/cfg_xml.c b/src/cfg_xml.c
index 6953edb3a..bb23dd641 100644
--- a/src/cfg_xml.c
+++ b/src/cfg_xml.c
@@ -1790,8 +1790,8 @@ rspamd_xml_end_element (GMarkupParseContext	*context, const gchar *element_name,
 			CHECK_TAG ("classifier", FALSE);
 			if (res) {
 				ccf = ud->section_pointer;
-				if (ccf->statfiles == NULL) {
-					*error = g_error_new (xml_error_quark (), XML_PARAM_MISSING, "classifier cannot contains no statfiles");
+				if (ccf->statfiles == NULL || !check_classifier_statfiles (ccf)) {
+					*error = g_error_new (xml_error_quark (), XML_PARAM_MISSING, "classifier cannot contains no statfiles or statfiles of the same class");
 					ud->state = XML_ERROR;
 					return;
 				}
-- 
2.39.5