[Feature] Process subject for mixed characters

3 files changed, 36 insertions, 2 deletions

diff --git a/src/libstat/stat_process.c b/src/libstat/stat_process.c
index 8d9717562..e5c17ddff 100644
--- a/src/libstat/stat_process.c
+++ b/src/libstat/stat_process.c
@@ -18,7 +18,6 @@
 #include "rspamd.h"
 #include "stat_internal.h"
 #include "libmime/message.h"
-#include "libmime/filter.h"
 #include "libmime/images.h"
 #include "libserver/html.h"
 #include "lua/lua_common.h"
diff --git a/src/libutil/util.h b/src/libutil/util.h
index cfea5f851..48381ed92 100644
--- a/src/libutil/util.h
+++ b/src/libutil/util.h
@@ -516,5 +516,5 @@ gdouble rspamd_normalize_probability (gdouble x, gdouble bias);
  */
 guint64 rspamd_tm_to_time (const struct tm *tm, glong tz);
 
-#define PTR_ARRAY_FOREACH(ar, i, cur) if ((ar) != NULL && (ar)->len > 0) for ((i) = 0; (i) < (ar)->len && (((cur) = g_ptr_array_index((ar), (i))) || 1); ++(i))
+#define PTR_ARRAY_FOREACH(ar, i, cur) for ((i) = 0; (ar) != NULL && (i) < (ar)->len && (((cur) = g_ptr_array_index((ar), (i))) || 1); ++(i))
 #endif
diff --git a/src/plugins/chartable.c b/src/plugins/chartable.c
index d6efc3ed2..ff3aa480f 100644
--- a/src/plugins/chartable.c
+++ b/src/plugins/chartable.c
@@ -26,6 +26,8 @@
 #include "libmime/message.h"
 #include "rspamd.h"
 #include "libstat/stat_api.h"
+#include "libstat/tokenizers/tokenizers.h"
+
 #include "unicode/utf8.h"
 #include "unicode/uchar.h"
 
@@ -399,6 +401,39 @@ chartable_symbol_callback (struct rspamd_task *task, void *unused)
 		part = g_ptr_array_index (task->text_parts, i);
 		rspamd_chartable_process_part (task, part);
 	}
+
+	if (task->subject != NULL) {
+		GArray *words;
+		rspamd_stat_token_t *w;
+		guint i;
+		gdouble cur_score = 0.0;
+
+		words = rspamd_tokenize_text (task->subject, strlen (task->subject),
+				TRUE,
+				NULL,
+				NULL,
+				FALSE,
+				NULL);
+
+		if (words) {
+			for (i = 0; i < words->len; i++) {
+				w = &g_array_index (words, rspamd_stat_token_t, i);
+				cur_score += rspamd_chartable_process_word_utf (task, w, FALSE);
+			}
+		}
+
+		cur_score /= (gdouble)part->normalized_words->len;
+
+		if (cur_score > 2.0) {
+			cur_score = 2.0;
+		}
+
+		if (cur_score > chartable_module_ctx->threshold) {
+			rspamd_task_insert_result (task, chartable_module_ctx->symbol,
+					cur_score, "subject");
+
+		}
+	}
 }
 
 static void