commit 971025c: [Rework] Rework options hash structure

[Vsevolod Stakhov]

Author: Vsevolod Stakhov
Date: 2020-01-29 15:03:08 +0000
URL: https://github.com/rspamd/rspamd/commit/971025ceca5af31e11a1d079302365bb6f3d7f8c

[Rework] Rework options hash structure

---
 src/libmime/scan_result.c         | 24 ++++++++++++++----------
 src/libmime/scan_result.h         |  5 ++++-
 src/libmime/scan_result_private.h | 18 +++++++++++++++++-
 3 files changed, 35 insertions(+), 12 deletions(-)

diff --git a/src/libmime/scan_result.c b/src/libmime/scan_result.c
index 7a8451f9c..671e99a9b 100644
--- a/src/libmime/scan_result.c
+++ b/src/libmime/scan_result.c
@@ -293,7 +293,7 @@ insert_metric_result (struct rspamd_task *task,
 		s->nshots ++;
 
 		if (opt) {
-			rspamd_task_add_result_option (task, s, opt);
+			rspamd_task_add_result_option (task, s, opt, strlen (opt));
 		}
 
 		/* Adjust diff */
@@ -459,7 +459,9 @@ insert_metric_result (struct rspamd_task *task,
 			s->score = 0;
 		}
 
-		rspamd_task_add_result_option (task, s, opt);
+		if (opt) {
+			rspamd_task_add_result_option (task, s, opt, strlen (opt));
+		}
 	}
 
 	msg_debug_metric ("final insertion for symbol %s, score %.2f, factor: %f",
@@ -503,12 +505,13 @@ rspamd_task_insert_result_full (struct rspamd_task *task,
 
 gboolean
 rspamd_task_add_result_option (struct rspamd_task *task,
-		struct rspamd_symbol_result *s, const gchar *val)
+							   struct rspamd_symbol_result *s,
+							   const gchar *val,
+							   gsize vlen)
 {
-	struct rspamd_symbol_option *opt;
+	struct rspamd_symbol_option *opt, srch;
 	gboolean ret = FALSE;
 	gchar *opt_cpy = NULL;
-	gsize vlen;
 	khiter_t k;
 	gint r;
 
@@ -524,8 +527,6 @@ rspamd_task_add_result_option (struct rspamd_task *task,
 			s->options = kh_init (rspamd_options_hash);
 		}
 
-		vlen = strlen (val);
-
 		if (vlen + s->opts_len > task->cfg->max_opts_len) {
 			/* Add truncated option */
 			msg_info_task ("cannot add more options to symbol %s when adding option %s",
@@ -544,7 +545,9 @@ rspamd_task_add_result_option (struct rspamd_task *task,
 		if (!(s->sym && (s->sym->flags & RSPAMD_SYMBOL_FLAG_ONEPARAM)) &&
 				kh_size (s->options) < task->cfg->default_max_shots) {
 			/* Append new options */
-			k = kh_get (rspamd_options_hash, s->options, val);
+			srch.option = (gchar *)val;
+			srch.optlen = vlen;
+			k = kh_get (rspamd_options_hash, s->options, &srch);
 
 			if (k == kh_end (s->options)) {
 				opt = rspamd_mempool_alloc0 (task->task_pool, sizeof (*opt));
@@ -553,9 +556,10 @@ rspamd_task_add_result_option (struct rspamd_task *task,
 					opt_cpy = rspamd_mempool_strdup (task->task_pool, val);
 				}
 
-				k = kh_put (rspamd_options_hash, s->options, opt_cpy, &r);
+				opt->optlen = vlen;
+				opt->option = opt_cpy;
 
-				kh_value (s->options, k) = opt;
+				k = kh_put (rspamd_options_hash, s->options, opt, &r);
 				opt->option = opt_cpy;
 				DL_APPEND (s->opts_head, opt);
 
diff --git a/src/libmime/scan_result.h b/src/libmime/scan_result.h
index 3b222fffb..74d0d81b9 100644
--- a/src/libmime/scan_result.h
+++ b/src/libmime/scan_result.h
@@ -20,6 +20,7 @@ struct rspamd_classifier_config;
 
 struct rspamd_symbol_option {
 	gchar *option;
+	gsize optlen;
 	struct rspamd_symbol_option *prev, *next;
 };
 
@@ -142,7 +143,9 @@ struct rspamd_symbol_result *rspamd_task_insert_result_full (struct rspamd_task
  * @param opt
  */
 gboolean rspamd_task_add_result_option (struct rspamd_task *task,
-										struct rspamd_symbol_result *s, const gchar *opt);
+										struct rspamd_symbol_result *s,
+										const gchar *opt,
+										gsize vlen);
 
 /**
  * Finds symbol result
diff --git a/src/libmime/scan_result_private.h b/src/libmime/scan_result_private.h
index aa90430e1..cb4ff4cda 100644
--- a/src/libmime/scan_result_private.h
+++ b/src/libmime/scan_result_private.h
@@ -12,7 +12,23 @@
 extern "C" {
 #endif
 
-KHASH_MAP_INIT_STR (rspamd_options_hash, struct rspamd_symbol_option *);
+#define RSPAMD_OPTS_SEED 0x9f1f608628a4fefbULL
+#define rspamd_symopt_hash(opt) (rspamd_cryptobox_fast_hash ( \
+		((struct rspamd_symbol_option *)opt)->option, \
+		((struct rspamd_symbol_option *)opt)->optlen, RSPAMD_OPTS_SEED))
+static inline bool
+rspamd_symopt_equal (const struct rspamd_symbol_option *o1,
+		const struct rspamd_symbol_option *o2)
+{
+	if (o1->optlen == o2->optlen) {
+		return (memcmp (o1->option, o2->option, o1->optlen) == 0);
+	}
+
+	return false;
+}
+
+KHASH_INIT (rspamd_options_hash, struct rspamd_symbol_option *, char,
+		0, rspamd_symopt_hash, rspamd_symopt_equal);
 /**
  * Result of metric processing
  */