[Rework] Urls: rework urls hash structure

diff --git a/src/libserver/url.c b/src/libserver/url.c
index 043f523f0650213e41f09863843c52e0a3b2411c..3449310b282eb5c201b76e6be91decd441fd0ecf 100644 (file)
--- a/src/libserver/url.c
+++ b/src/libserver/url.c
@@ -214,6 +214,12 @@ struct url_matcher static_matchers[] = {
                                URL_FLAG_NOHTML}
 };
 
+/* Hash table implementation */
+__KHASH_IMPL (rspamd_url_hash, kh_inline,struct rspamd_url *, char, false,
+               rspamd_url_hash, rspamd_urls_cmp);
+__KHASH_IMPL (rspamd_url_host_hash, kh_inline,struct rspamd_url *, char, false,
+               rspamd_url_host_hash, rspamd_urls_host_cmp);
+
 struct url_callback_data {
        const gchar *begin;
        gchar *url_str;
@@ -3374,7 +3380,7 @@ rspamd_url_task_subject_callback (struct rspamd_url *url, gsize start_offset,
        return TRUE;
 }
 
-guint
+inline guint
 rspamd_url_hash (gconstpointer u)
 {
        const struct rspamd_url *url = u;
@@ -3387,7 +3393,7 @@ rspamd_url_hash (gconstpointer u)
        return 0;
 }
 
-guint
+inline guint
 rspamd_url_host_hash (gconstpointer u)
 {
        const struct rspamd_url *url = u;
@@ -3401,7 +3407,7 @@ rspamd_url_host_hash (gconstpointer u)
        return 0;
 }
 
-guint
+inline guint
 rspamd_email_hash (gconstpointer u)
 {
        const struct rspamd_url *url = u;
@@ -3421,7 +3427,7 @@ rspamd_email_hash (gconstpointer u)
 }
 
 /* Compare two emails for building emails tree */
-gboolean
+inline gboolean
 rspamd_emails_cmp (gconstpointer a, gconstpointer b)
 {
        const struct rspamd_url *u1 = a, *u2 = b;
@@ -3450,7 +3456,7 @@ rspamd_emails_cmp (gconstpointer a, gconstpointer b)
        return FALSE;
 }
 
-gboolean
+inline gboolean
 rspamd_urls_cmp (gconstpointer a, gconstpointer b)
 {
        const struct rspamd_url *u1 = a, *u2 = b;
@@ -3466,7 +3472,7 @@ rspamd_urls_cmp (gconstpointer a, gconstpointer b)
        return r == 0;
 }
 
-gboolean
+inline gboolean
 rspamd_urls_host_cmp (gconstpointer a, gconstpointer b)
 {
        const struct rspamd_url *u1 = a, *u2 = b;
@@ -3806,3 +3812,38 @@ rspamd_url_protocol_from_string (const gchar *str)
 
        return ret;
 }
+
+
+bool
+rspamd_url_set_add_or_increase (khash_t (rspamd_url_hash) *set,
+                                                                        struct rspamd_url *u)
+{
+       khiter_t k;
+       gint r;
+
+       k = kh_put (rspamd_url_hash, set, u, &r);
+
+       if (r == 0) {
+               struct rspamd_url *ex = kh_key (set, k);
+
+               ex->count ++;
+
+               return false;
+       }
+
+       return true;
+}
+
+bool
+rspamd_url_set_has (khash_t (rspamd_url_hash) *set, struct rspamd_url *u)
+{
+       khiter_t k;
+
+       k = kh_get (rspamd_url_hash, set, u);
+
+       if (k == kh_end (set)) {
+               return false;
+       }
+
+       return true;
+}
\ No newline at end of file

diff --git a/src/libserver/url.h b/src/libserver/url.h
index 00f09ac30362f625cd5c1b6f70c9e838b549de6f..358c61e16fe29b8e4df6fc5ab0013a7341fe40f1 100644 (file)
--- a/src/libserver/url.h
+++ b/src/libserver/url.h
@@ -4,6 +4,7 @@
 
 #include "config.h"
 #include "mem_pool.h"
+#include "khash.h"
 #include "fstring.h"
 
 #ifdef  __cplusplus
@@ -281,6 +282,27 @@ const gchar *rspamd_url_protocol_name (enum rspamd_url_protocol proto);
  */
 enum rspamd_url_protocol rspamd_url_protocol_from_string (const gchar *str);
 
+/* Defines sets of urls indexed by url as is */
+KHASH_DECLARE (rspamd_url_hash, struct rspamd_url *, char);
+KHASH_DECLARE (rspamd_url_host_hash, struct rspamd_url *, char);
+
+/* Convenience functions for url sets */
+/**
+ * Add an url to set or increase the existing url count
+ * @param set
+ * @param u
+ * @return true if a new url has been added
+ */
+bool rspamd_url_set_add_or_increase (khash_t (rspamd_url_hash) *set,
+               struct rspamd_url *u);
+/**
+ * Checks if a url is in set
+ * @param set
+ * @param u
+ * @return
+ */
+bool rspamd_url_set_has (khash_t (rspamd_url_hash) *set, struct rspamd_url *u);
+
 #ifdef  __cplusplus
 }
 #endif