Use flags for urls

diff --git a/src/libserver/html.c b/src/libserver/html.c
index 2d1896d8459706deda7f015db2b29be2474a650c..daea7f7ff8dc208878a73f0bb2d0f9681e40c3ed 100644 (file)
--- a/src/libserver/html.c
+++ b/src/libserver/html.c
@@ -859,7 +859,7 @@ rspamd_html_url_is_phished (rspamd_mempool_t *pool,
 
                                if (href_url->tldlen != text_url->tldlen || memcmp (href_url->tld,
                                                text_url->tld, href_url->tldlen) != 0) {
-                                       href_url->is_phished = TRUE;
+                                       href_url->flags |= RSPAMD_URL_FLAG_PHISHED;
                                        href_url->phished_url = text_url;
                                }
                        }

diff --git a/src/libserver/protocol.c b/src/libserver/protocol.c
index 7e91f6a1ea43eeb8fa120bedf95faa38517a408b..d20c50ff8381a90b415dac86f50fdc30b5e1c9c5 100644 (file)
--- a/src/libserver/protocol.c
+++ b/src/libserver/protocol.c
@@ -645,7 +645,7 @@ urls_protocol_cb (gpointer key, gpointer value, gpointer ud)
                        ucl_object_insert_key (obj, elt, "surbl", 0, false);
                }
 
-               elt = ucl_object_frombool (url->is_phished);
+               elt = ucl_object_frombool (url->flags & RSPAMD_URL_FLAG_PHISHED);
                ucl_object_insert_key (obj, elt, "phished", 0, false);
        }
        ucl_array_append (cb->top, obj);

diff --git a/src/libserver/url.c b/src/libserver/url.c
index 18669748349a2c5e3a2339709668e709c6811986..f0a65df8cfd486901415e52de9202ed0fae44ca1 100644 (file)
--- a/src/libserver/url.c
+++ b/src/libserver/url.c
@@ -1093,7 +1093,7 @@ rspamd_url_is_ip (struct rspamd_url *uri, rspamd_mempool_t *pool)
                uri->hostlen = strlen (uri->host);
                uri->tld = uri->host;
                uri->tldlen = uri->hostlen;
-               uri->is_numeric = TRUE;
+               uri->flags |= RSPAMD_URL_FLAG_NUMERIC;
                ret = TRUE;
        }
        else if (inet_pton (AF_INET6, buf, &in6) == 1) {
@@ -1103,7 +1103,7 @@ rspamd_url_is_ip (struct rspamd_url *uri, rspamd_mempool_t *pool)
                uri->hostlen = strlen (uri->host);
                uri->tld = uri->host;
                uri->tldlen = uri->hostlen;
-               uri->is_numeric = TRUE;
+               uri->flags |= RSPAMD_URL_FLAG_NUMERIC;
                ret = TRUE;
        }
        else {
@@ -1175,7 +1175,7 @@ rspamd_url_is_ip (struct rspamd_url *uri, rspamd_mempool_t *pool)
                        uri->hostlen = strlen (uri->host);
                        uri->tld = uri->host;
                        uri->tldlen = uri->hostlen;
-                       uri->is_numeric = TRUE;
+                       uri->flags |= RSPAMD_URL_FLAG_NUMERIC|RSPAMD_URL_FLAG_OBSCURED;
                        ret = TRUE;
                }
        }

diff --git a/src/libserver/url.h b/src/libserver/url.h
index f1c850b8096e2cd5d74b8948630dc20845bc9b3d..90b6d5c61d66fba429dbbc831f5ee0ea3d680e22 100644 (file)
--- a/src/libserver/url.h
+++ b/src/libserver/url.h
@@ -8,6 +8,12 @@
 struct rspamd_task;
 struct mime_text_part;
 
+enum rspamd_url_flags {
+       RSPAMD_URL_FLAG_PHISHED = 1 << 0,
+       RSPAMD_URL_FLAG_NUMERIC = 1 << 1,
+       RSPAMD_URL_FLAG_OBSCURED = 1 << 2,
+};
+
 struct rspamd_url {
        gchar *string;
        gint protocol;
@@ -35,8 +41,7 @@ struct rspamd_url {
        guint tldlen;
        guint urllen;
 
-       gboolean is_phished; /* URI maybe phishing */
-       gboolean is_numeric; /* URI contains IP address */
+       enum rspamd_url_flags flags;
 };
 
 enum uri_errno {

diff --git a/src/libutil/util.c b/src/libutil/util.c
index 8afac64c0199f9c1166f3c211a0bfa47206e94d8..3b95a1190ccbc54a42249d92845d7b56fce8d960 100644 (file)
--- a/src/libutil/util.c
+++ b/src/libutil/util.c
@@ -1383,7 +1383,7 @@ rspamd_url_hash (gconstpointer u)
        if (url->userlen > 0) {
                XXH64_update (&st, url->user, url->userlen);
        }
-       XXH64_update (&st, &url->is_phished, sizeof (url->is_phished));
+       XXH64_update (&st, &url->flags, sizeof (url->flags));
 
        return XXH64_digest (&st);
 }
@@ -1426,7 +1426,7 @@ rspamd_urls_cmp (gconstpointer a, gconstpointer b)
        }
        else {
                r = g_ascii_strncasecmp (u1->host, u2->host, u1->hostlen);
-               if (r == 0 && u1->is_phished != u2->is_phished) {
+               if (r == 0 && u1->flags != u2->flags) {
                        /* Always insert phished urls to the tree */
                        return FALSE;
                }

diff --git a/src/lua/lua_url.c b/src/lua/lua_url.c
index 3e74ceadc304483bde7bdf86c714f474bd49eae6..8a9c90216d102d1b894429a6941acc52659892ac 100644 (file)
--- a/src/lua/lua_url.c
+++ b/src/lua/lua_url.c
@@ -261,7 +261,7 @@ lua_url_is_phished (lua_State *L)
        struct rspamd_lua_url *url = lua_check_url (L, 1);
 
        if (url != NULL) {
-               lua_pushboolean (L, url->url->is_phished);
+               lua_pushboolean (L, url->url->flags & RSPAMD_URL_FLAG_PHISHED);
        }
        else {
                lua_pushnil (L);
@@ -281,7 +281,8 @@ lua_url_get_phished (lua_State *L)
        struct rspamd_lua_url *purl, *url = lua_check_url (L, 1);
 
        if (url) {
-               if (url->url->is_phished && url->url->phished_url != NULL) {
+               if ((url->url->flags & RSPAMD_URL_FLAG_PHISHED)
+                               && url->url->phished_url != NULL) {
                        purl = lua_newuserdata (L, sizeof (struct rspamd_lua_url));
                        rspamd_lua_setclass (L, "rspamd{url}", -1);
                        purl->url = url->url->phished_url;
@@ -476,7 +477,7 @@ lua_url_all (lua_State *L)
                        pos = text;
                        end = text + length;
                        lua_newtable (L);
-                       
+
                        while (pos <= end) {
                                url = rspamd_url_get_next (pool, text, &pos, NULL);
 

diff --git a/src/plugins/surbl.c b/src/plugins/surbl.c
index f3f09c29b0bbcec247e955b24a893e91db785f0d..40ab9f6681667abae5933ff6431b4e0995a60037 100644 (file)
--- a/src/plugins/surbl.c
+++ b/src/plugins/surbl.c
@@ -678,7 +678,7 @@ format_surbl_request (rspamd_mempool_t * pool,
        len = hostname->len + slen + 2;
 
        p = hostname->begin;
-       is_numeric = url->is_numeric;
+       is_numeric = url->flags & RSPAMD_URL_FLAG_NUMERIC;
 
        while (p - hostname->begin < (gint)hostname->len && dots_num < MAX_LEVELS) {
                if (*p == '.') {