diff options
author | Vsevolod Stakhov <vsevolod@highsecure.ru> | 2017-03-23 17:14:07 +0000 |
---|---|---|
committer | Vsevolod Stakhov <vsevolod@highsecure.ru> | 2017-03-23 17:14:07 +0000 |
commit | 11021601e04634213a31df0c16aa4d1c064201ac (patch) | |
tree | 12f5fac32c4b5747ea45c967fe9f0d0de19f1a45 /src/plugins/chartable.c | |
parent | 8e865ebfcb0fe584841c2e082acd894646569213 (diff) | |
download | rspamd-11021601e04634213a31df0c16aa4d1c064201ac.tar.gz rspamd-11021601e04634213a31df0c16aa4d1c064201ac.zip |
[Feature] Process subject for mixed characters
Diffstat (limited to 'src/plugins/chartable.c')
-rw-r--r-- | src/plugins/chartable.c | 35 |
1 files changed, 35 insertions, 0 deletions
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 |