Search for urls in query parts of html urls as well.

1 files changed, 43 insertions, 0 deletions

diff --git a/src/libserver/html.c b/src/libserver/html.c
index 0a46cef02..520c49d39 100644
--- a/src/libserver/html.c
+++ b/src/libserver/html.c
@@ -1308,6 +1308,43 @@ rspamd_html_process_url_tag (rspamd_mempool_t *pool, struct html_tag *tag)
 }
 
 static void
+rspamd_process_html_url (rspamd_mempool_t *pool, struct rspamd_url *url,
+		GHashTable *target)
+{
+	gint nstate = 0;
+	struct rspamd_url *query_url;
+	gchar *url_str;
+	gint rc;
+
+	if (url->querylen > 0) {
+
+		if (rspamd_url_find (pool, url->query, url->querylen, NULL, NULL,
+				&url_str, TRUE, &nstate)) {
+			query_url = rspamd_mempool_alloc0 (pool,
+					sizeof (struct rspamd_url));
+
+			rc = rspamd_url_parse (query_url,
+					url_str,
+					strlen (url_str),
+					pool);
+
+			if (rc == URI_ERRNO_OK &&
+					url->hostlen > 0) {
+				msg_debug_pool ("found url %s in query of url"
+						" %*s", url_str, url->querylen, url->query);
+
+				if (!g_hash_table_lookup (target,
+						query_url)) {
+					g_hash_table_insert (target,
+							query_url,
+							query_url);
+				}
+			}
+		}
+	}
+}
+
+static void
 rspamd_html_process_img_tag (rspamd_mempool_t *pool, struct html_tag *tag,
 		struct html_content *hc)
 {
@@ -1923,6 +1960,12 @@ rspamd_html_process_part_full (rspamd_mempool_t *pool, struct html_content *hc,
 								else {
 									url = NULL;
 								}
+
+								if (turl == NULL && url != NULL) {
+									rspamd_process_html_url (pool,
+											url,
+											target_tbl);
+								}
 							}
 
 							href_offset = dest->len;