[Feature] Implement url tags concept

src/libserver/html.c

@@ -1297,7 +1297,7 @@ rspamd_html_process_url (rspamd_mempool_t *pool, const gchar *start, guint len,
 
 	*d = '\0';
 
-	url = rspamd_mempool_alloc (pool, sizeof (*url));
+	url = rspamd_mempool_alloc0 (pool, sizeof (*url));
 	rc = rspamd_url_parse (url, decoded, d - decoded, pool);
 
 	if (rc == URI_ERRNO_OK) {

src/libserver/url.c

@@ -2537,3 +2537,18 @@ rspamd_url_task_callback (struct rspamd_url *url, gsize start_offset,
 		}
 	}
 }
+
+void
+rspamd_url_add_tag (struct rspamd_url *url, const gchar *tag,
+		rspamd_mempool_t *pool)
+{
+	g_assert (url != NULL && tag != NULL);
+
+	if (url->tags == NULL) {
+		url->tags = g_ptr_array_sized_new (2);
+		rspamd_mempool_add_destructor (pool, rspamd_ptr_array_free_hard, url->tags);
+	}
+
+	g_ptr_array_add (url->tags, rspamd_mempool_strdup (pool, tag));
+
+}

src/libserver/url.h

@@ -42,6 +42,7 @@ struct rspamd_url {
 	guint urllen;
 
 	enum rspamd_url_flags flags;
+	GPtrArray *tags;
 };
 
 enum uri_errno {
@@ -161,4 +162,13 @@ void rspamd_url_find_single (rspamd_mempool_t *pool, const gchar *in,
 void rspamd_url_task_callback (struct rspamd_url *url, gsize start_offset,
 		gsize end_offset, gpointer ud);
 
+/**
+ * Adds a tag for url
+ * @param url
+ * @param tag
+ * @param pool
+ */
+void rspamd_url_add_tag (struct rspamd_url *url, const gchar *tag,
+		rspamd_mempool_t *pool);
+
 #endif

src/lua/lua_url.c

@@ -52,6 +52,8 @@ LUA_FUNCTION_DEF (url, is_phished);
 LUA_FUNCTION_DEF (url, is_redirected);
 LUA_FUNCTION_DEF (url, is_obscured);
 LUA_FUNCTION_DEF (url, get_phished);
+LUA_FUNCTION_DEF (url, get_tags);
+LUA_FUNCTION_DEF (url, add_tag);
 LUA_FUNCTION_DEF (url, create);
 LUA_FUNCTION_DEF (url, init);
 LUA_FUNCTION_DEF (url, all);
@@ -71,6 +73,8 @@ static const struct luaL_reg urllib_m[] = {
 	LUA_INTERFACE_DEF (url, is_redirected),
 	LUA_INTERFACE_DEF (url, is_obscured),
 	LUA_INTERFACE_DEF (url, get_phished),
+	LUA_INTERFACE_DEF (url, get_tags),
+	LUA_INTERFACE_DEF (url, add_tag),
 	{"get_redirected", lua_url_get_phished},
 	{"__tostring", lua_url_get_text},
 	{NULL, NULL}
@@ -309,6 +313,61 @@ lua_url_is_obscured (lua_State *L)
 	return 1;
 }
 
+/***
+ * @method url:get_tags()
+ * Returns list of string tags for an url
+ * @return {table/strings} list of tags for an url
+ */
+static gint
+lua_url_get_tags (lua_State *L)
+{
+	struct rspamd_lua_url *url = lua_check_url (L, 1);
+	guint i;
+	const gchar *tag;
+
+	if (url != NULL) {
+		if (url->url->tags == NULL) {
+			lua_createtable (L, 0, 0);
+		}
+		else {
+			lua_createtable (L, url->url->tags->len, 0);
+
+			PTR_ARRAY_FOREACH (url->url->tags, i, tag) {
+				lua_pushstring (L, tag);
+				lua_rawseti (L, -2, i + 1);
+			}
+		}
+	}
+	else {
+		lua_pushnil (L);
+	}
+
+	return 1;
+}
+
+/***
+ * @method url:add_tag(tag, mempool)
+ * Adds a new tag for url
+ * @param {string} tag new tag to add
+ * @param {mempool} mempool memory pool (e.g. `task:get_pool()`)
+ */
+static gint
+lua_url_add_tag (lua_State *L)
+{
+	struct rspamd_lua_url *url = lua_check_url (L, 1);
+	rspamd_mempool_t *mempool = rspamd_lua_check_mempool (L, 3);
+	const gchar *tag = luaL_checkstring (L, 2);
+
+	if (url != NULL && mempool != NULL && tag != NULL) {
+		rspamd_url_add_tag (url->url, tag, mempool);
+	}
+	else {
+		return luaL_error (L, "invalid arguments");
+	}
+
+	return 0;
+}
+
 /***
  * @method url:get_phished()
  * Get another URL that pretends to be this URL (e.g. used in phishing)

src/plugins/surbl.c

@@ -1308,7 +1308,7 @@ surbl_redirector_finish (struct rspamd_http_connection *conn,
 			urllen = hdr->len;
 			urlstr = rspamd_mempool_alloc (task->task_pool,
 					urllen + 1);
-			redirected_url = rspamd_mempool_alloc (task->task_pool,
+			redirected_url = rspamd_mempool_alloc0 (task->task_pool,
 					sizeof (*redirected_url));
 			rspamd_strlcpy (urlstr, hdr->begin, urllen + 1);
 			r = rspamd_url_parse (redirected_url, urlstr, urllen,