commit add9595: [Feature] Export visible part of url to lua

[Miecio Za]

Author: Miecio Za
Date: 2019-02-27 11:56:35 +0100
URL: https://github.com/rspamd/rspamd/commit/add9595913c3f456c6328a53801739cdca70da9d

[Feature] Export visible part of url to lua
Export visible part of url to lua

---
 src/libserver/html.c |  6 ++++++
 src/libserver/url.h  |  2 ++
 src/lua/lua_url.c    | 23 +++++++++++++++++++++++
 3 files changed, 31 insertions(+)

diff --git a/src/libserver/html.c b/src/libserver/html.c
index 31438ddad..70b66ccc0 100644
--- a/src/libserver/html.c
+++ b/src/libserver/html.c
@@ -2373,6 +2373,12 @@ rspamd_html_check_displayed_url (rspamd_mempool_t *pool,
 		return;
 	}
 
+	gint visible_part_len = dest->len - href_offset;
+	url->visible_part = rspamd_mempool_alloc0(pool, visible_part_len +1);
+	url->visible_partlen = visible_part_len;
+	gchar *visible_part = g_strndup(dest->data + href_offset, visible_part_len);
+	g_stpcpy(url->visible_part, visible_part);
+
 	rspamd_html_url_is_phished (pool, url,
 			dest->data + href_offset,
 			dest->len - href_offset,
diff --git a/src/libserver/url.h b/src/libserver/url.h
index 4d1948921..571424099 100644
--- a/src/libserver/url.h
+++ b/src/libserver/url.h
@@ -48,6 +48,7 @@ struct rspamd_url {
 	gchar *fragment;
 	gchar *surbl;
 	gchar *tld;
+	gchar *visible_part;
 
 	struct rspamd_url *phished_url;
 
@@ -61,6 +62,7 @@ struct rspamd_url {
 	guint tldlen;
 	guint urllen;
 	guint rawlen;
+	guint visible_partlen;
 
 	enum rspamd_url_flags flags;
 	guint count;
diff --git a/src/lua/lua_url.c b/src/lua/lua_url.c
index 0a301e96d..7bad50359 100644
--- a/src/lua/lua_url.c
+++ b/src/lua/lua_url.c
@@ -63,6 +63,7 @@ LUA_FUNCTION_DEF (url, get_tag);
 LUA_FUNCTION_DEF (url, get_count);
 LUA_FUNCTION_DEF (url, get_tags);
 LUA_FUNCTION_DEF (url, add_tag);
+LUA_FUNCTION_DEF (url, get_visible);
 LUA_FUNCTION_DEF (url, create);
 LUA_FUNCTION_DEF (url, init);
 LUA_FUNCTION_DEF (url, all);
@@ -89,6 +90,7 @@ static const struct luaL_reg urllib_m[] = {
 	LUA_INTERFACE_DEF (url, get_tag),
 	LUA_INTERFACE_DEF (url, get_tags),
 	LUA_INTERFACE_DEF (url, add_tag),
+	LUA_INTERFACE_DEF (url, get_visible),
 	LUA_INTERFACE_DEF (url, get_count),
 	LUA_INTERFACE_DEF (url, get_flags),
 	{"get_redirected", lua_url_get_phished},
@@ -650,6 +652,27 @@ lua_url_get_count (lua_State *L)
 	return 1;
 }
 
+ /***
+* @method url:get_visible()
+* Get visible part of the url with html tags stripped
+* @return {string} url string
+*/
+static gint
+lua_url_get_visible (lua_State *L)
+{
+	LUA_TRACE_POINT;
+	struct rspamd_lua_url *url = lua_check_url (L, 1);
+
+	if (url != NULL) {
+		lua_pushlstring (L, url->url->visible_part, url->url->visible_partlen);
+	}
+	else {
+		lua_pushnil (L);
+	}
+
+return 1;
+}
+
 /***
  * @method url:to_table()
  * Return url as a table with the following fields: