[Feature] Insert redirected URL to the urls list

3 files changed, 43 insertions, 17 deletions

diff --git a/src/libserver/protocol.c b/src/libserver/protocol.c
index 35c912f59..e32dfc6e5 100644
--- a/src/libserver/protocol.c
+++ b/src/libserver/protocol.c
@@ -595,6 +595,39 @@ struct tree_cb_data {
 	struct rspamd_task *task;
 };
 
+static ucl_object_t *
+rspamd_protocol_extended_url (struct rspamd_url *url)
+{
+	ucl_object_t *obj, *elt;
+
+	obj = ucl_object_typed_new (UCL_OBJECT);
+
+	elt = ucl_object_fromlstring (url->string, url->urllen);
+	ucl_object_insert_key (obj, elt, "url", 0, false);
+
+	if (url->surbllen > 0) {
+		elt = ucl_object_fromlstring (url->surbl, url->surbllen);
+		ucl_object_insert_key (obj, elt, "surbl", 0, false);
+	}
+	if (url->hostlen > 0) {
+		elt = ucl_object_fromlstring (url->host, url->hostlen);
+		ucl_object_insert_key (obj, elt, "host", 0, false);
+	}
+
+	elt = ucl_object_frombool (url->flags & RSPAMD_URL_FLAG_PHISHED);
+	ucl_object_insert_key (obj, elt, "phished", 0, false);
+
+	elt = ucl_object_frombool (url->flags & RSPAMD_URL_FLAG_REDIRECTED);
+	ucl_object_insert_key (obj, elt, "redirected", 0, false);
+
+	if (url->phished_url) {
+		elt = rspamd_protocol_extended_url (url->phished_url);
+		ucl_object_insert_key (obj, elt, "orig_url", 0, false);
+	}
+
+	return obj;
+}
+
 /*
  * Callback for writing urls
  */
@@ -603,7 +636,7 @@ urls_protocol_cb (gpointer key, gpointer value, gpointer ud)
 {
 	struct tree_cb_data *cb = ud;
 	struct rspamd_url *url = value;
-	ucl_object_t *obj, *elt;
+	ucl_object_t *obj;
 	struct rspamd_task *task = cb->task;
 	const gchar *user_field = "unknown";
 	gboolean has_user = FALSE;
@@ -612,22 +645,7 @@ urls_protocol_cb (gpointer key, gpointer value, gpointer ud)
 		obj = ucl_object_fromlstring (url->string, url->urllen);
 	}
 	else {
-		obj = ucl_object_typed_new (UCL_OBJECT);
-
-		elt = ucl_object_fromlstring (url->string, url->urllen);
-		ucl_object_insert_key (obj, elt, "url", 0, false);
-
-		if (url->surbllen > 0) {
-			elt = ucl_object_fromlstring (url->surbl, url->surbllen);
-			ucl_object_insert_key (obj, elt, "surbl", 0, false);
-		}
-		if (url->hostlen > 0) {
-			elt = ucl_object_fromlstring (url->host, url->hostlen);
-			ucl_object_insert_key (obj, elt, "host", 0, false);
-		}
-
-		elt = ucl_object_frombool (url->flags & RSPAMD_URL_FLAG_PHISHED);
-		ucl_object_insert_key (obj, elt, "phished", 0, false);
+		obj = rspamd_protocol_extended_url (url);
 	}
 
 	ucl_array_append (cb->top, obj);
diff --git a/src/libserver/url.h b/src/libserver/url.h
index 4d38e3d79..7dfcb05af 100644
--- a/src/libserver/url.h
+++ b/src/libserver/url.h
@@ -13,6 +13,7 @@ enum rspamd_url_flags {
 	RSPAMD_URL_FLAG_PHISHED = 1 << 0,
 	RSPAMD_URL_FLAG_NUMERIC = 1 << 1,
 	RSPAMD_URL_FLAG_OBSCURED = 1 << 2,
+	RSPAMD_URL_FLAG_REDIRECTED = 1 << 3,
 };
 
 struct rspamd_url {
diff --git a/src/plugins/surbl.c b/src/plugins/surbl.c
index dcf69f8f2..ad2c3be9c 100644
--- a/src/plugins/surbl.c
+++ b/src/plugins/surbl.c
@@ -1281,6 +1281,13 @@ surbl_redirector_finish (struct rspamd_http_connection *conn,
 					task->task_pool);
 
 			if (r == URI_ERRNO_OK) {
+				if (!g_hash_table_lookup (task->urls, redirected_url)) {
+					g_hash_table_insert (task->urls, redirected_url,
+							redirected_url);
+					redirected_url->phished_url = param->url;
+					redirected_url->flags |= RSPAMD_URL_FLAG_REDIRECTED;
+				}
+
 				make_surbl_requests (redirected_url,
 						param->task,
 						param->suffix,