Implement min and max tokens.

src/libserver/cfg_file.h

@@ -154,6 +154,8 @@ struct rspamd_classifier_config {
 	GList *pre_callbacks;                           /**< list of callbacks that are called before classification */
 	GList *post_callbacks;                          /**< list of callbacks that are called after classification */
 	gchar *name;									/**< unique name of classifier							*/
+	guint32 min_tokens;								/**< minimal number of tokens to process classifier 	*/
+	guint32 max_tokens;								/**< maximum number of tokens							*/
 };
 
 struct rspamd_worker_bind_conf {

src/libserver/cfg_rcl.c

@@ -930,7 +930,7 @@ rspamd_rcl_classifier_handler (struct rspamd_config *cfg,
 {
 	const ucl_object_t *val, *cur;
 	ucl_object_iter_t it = NULL;
-	const gchar *key, *type;
+	const gchar *key;
 	struct rspamd_classifier_config *ccf;
 	gboolean res = TRUE;
 	struct rspamd_rcl_section *stat_section;
@@ -1361,6 +1361,16 @@ rspamd_rcl_config_init (void)
 		rspamd_rcl_parse_struct_string,
 		G_STRUCT_OFFSET (struct rspamd_classifier_config, tokenizer),
 		0);
+	rspamd_rcl_add_default_handler (sub,
+		"min_tokens",
+		rspamd_rcl_parse_struct_integer,
+		G_STRUCT_OFFSET (struct rspamd_classifier_config, min_tokens),
+		RSPAMD_CL_FLAG_INT_32);
+	rspamd_rcl_add_default_handler (sub,
+		"max_tokens",
+		rspamd_rcl_parse_struct_integer,
+		G_STRUCT_OFFSET (struct rspamd_classifier_config, max_tokens),
+		RSPAMD_CL_FLAG_INT_32);
 
 	/*
 	 * Statfile defaults

src/libstat/stat_process.c

@@ -37,7 +37,9 @@ struct rspamd_tokenizer_runtime {
 };
 
 struct preprocess_cb_data {
+	struct rspamd_task *task;
 	GList *classifier_runtimes;
+	struct rspamd_tokenizer_runtime *tok;
 	guint results_count;
 };
 
@@ -59,6 +61,18 @@ preprocess_init_stat_token (gpointer k, gpointer v, gpointer d)
 
 	while (cur) {
 		cl_runtime = (struct rspamd_classifier_runtime *)cur->data;
+
+		if (cl_runtime->clcf->min_tokens > 0 &&
+				(guint32)g_tree_nnodes (cbdata->tok->tokens) < cl_runtime->clcf->min_tokens) {
+			/* Skip this classifier */
+			msg_debug ("<%s> contains less tokens than required for %s classifier: "
+					"%ud < %ud", cbdata->task->message_id, cl_runtime->clcf->name,
+					g_tree_nnodes (cbdata->tok->tokens),
+					cl_runtime->clcf->min_tokens);
+			cur = g_list_next (cur);
+			continue;
+		}
+
 		res = &g_array_index (t->results, struct rspamd_token_result, i);
 
 		curst = res->cl_runtime->st_runtime;
@@ -72,6 +86,16 @@ preprocess_init_stat_token (gpointer k, gpointer v, gpointer d)
 			if (st_runtime->backend->process_token (t, res,
 					st_runtime->backend->ctx)) {
 				cl_runtime->processed_tokens ++;
+
+				if (cl_runtime->clcf->max_tokens > 0 &&
+						cl_runtime->processed_tokens > cl_runtime->clcf->max_tokens) {
+					msg_debug ("<%s> contains more tokens than allowed for %s classifier: "
+							"%ud > %ud", cbdata->task, cl_runtime->clcf->name,
+							cl_runtime->processed_tokens,
+							cl_runtime->clcf->max_tokens);
+
+					return TRUE;
+				}
 			}
 
 			i ++;
@@ -192,9 +216,11 @@ rspamd_stat_preprocess (struct rspamd_stat_ctx *st_ctx,
 
 		cbdata.results_count = result_size;
 		cbdata.classifier_runtimes = cl_runtimes;
+		cbdata.task = task;
 
 		/* Allocate token results */
 		LL_FOREACH (tklist, tok) {
+			cbdata.tok = tok;
 			g_tree_foreach (tok->tokens, preprocess_init_stat_token, &cbdata);
 		}
 	}