[Feature] Distinguish URLs found in Subject

src/libmime/message.c

@@ -632,8 +632,6 @@ rspamd_message_from_data (struct rspamd_task *task, const guchar *start,
 gboolean
 rspamd_message_parse (struct rspamd_task *task)
 {
-	GPtrArray *hdrs;
-	struct rspamd_mime_header *rh;
 	struct rspamd_mime_text_part *p1, *p2;
 	struct received_header *recv, *trecv;
 	const gchar *p;
@@ -824,13 +822,11 @@ rspamd_message_parse (struct rspamd_task *task)
 	}
 
 	/* Parse urls inside Subject header */
-	hdrs = rspamd_message_get_header_array (task, "Subject", FALSE);
-
-	PTR_ARRAY_FOREACH (hdrs, i, rh) {
-		p = rh->decoded;
+	if (task->subject) {
+		p = task->subject;
 		len = strlen (p);
 		rspamd_url_find_multiple (task->task_pool, p, len, FALSE, NULL,
-				rspamd_url_task_callback, task);
+				rspamd_url_task_subject_callback, task);
 	}
 
 	/* Calculate distance for 2-parts messages */

src/libserver/url.c

@@ -2486,7 +2486,7 @@ rspamd_url_find_single (rspamd_mempool_t *pool, const gchar *in,
 
 
 void
-rspamd_url_task_callback (struct rspamd_url *url, gsize start_offset,
+rspamd_url_task_subject_callback (struct rspamd_url *url, gsize start_offset,
 		gsize end_offset, gpointer ud)
 {
 	struct rspamd_task *task = ud;
@@ -2495,7 +2495,7 @@ rspamd_url_task_callback (struct rspamd_url *url, gsize start_offset,
 	gint rc;
 
 	/* It is just a displayed URL, we should not check it for certain things */
-	url->flags |= RSPAMD_URL_FLAG_HTML_DISPLAYED;
+	url->flags |= RSPAMD_URL_FLAG_HTML_DISPLAYED|RSPAMD_URL_FLAG_SUBJECT;
 
 	if (url->protocol == PROTOCOL_MAILTO) {
 		if (url->userlen > 0) {

src/libserver/url.h

@@ -16,6 +16,7 @@ enum rspamd_url_flags {
 	RSPAMD_URL_FLAG_REDIRECTED = 1 << 3,
 	RSPAMD_URL_FLAG_HTML_DISPLAYED = 1 << 4,
 	RSPAMD_URL_FLAG_FROM_TEXT = 1 << 5,
+	RSPAMD_URL_FLAG_SUBJECT = 1 << 6,
 };
 
 struct rspamd_url_tag {
@@ -163,7 +164,8 @@ void rspamd_url_find_single (rspamd_mempool_t *pool, const gchar *in,
  * @param end_offset
  * @param ud
  */
-void rspamd_url_task_callback (struct rspamd_url *url, gsize start_offset,
+void rspamd_url_task_subject_callback (struct rspamd_url *url,
+		gsize start_offset,
 		gsize end_offset, gpointer ud);
 
 /**

src/lua/lua_url.c

@@ -53,6 +53,7 @@ LUA_FUNCTION_DEF (url, is_phished);
 LUA_FUNCTION_DEF (url, is_redirected);
 LUA_FUNCTION_DEF (url, is_obscured);
 LUA_FUNCTION_DEF (url, is_html_displayed);
+LUA_FUNCTION_DEF (url, is_subject);
 LUA_FUNCTION_DEF (url, get_phished);
 LUA_FUNCTION_DEF (url, get_tag);
 LUA_FUNCTION_DEF (url, get_tags);
@@ -76,6 +77,7 @@ static const struct luaL_reg urllib_m[] = {
 	LUA_INTERFACE_DEF (url, is_redirected),
 	LUA_INTERFACE_DEF (url, is_obscured),
 	LUA_INTERFACE_DEF (url, is_html_displayed),
+	LUA_INTERFACE_DEF (url, is_subject),
 	LUA_INTERFACE_DEF (url, get_phished),
 	LUA_INTERFACE_DEF (url, get_tag),
 	LUA_INTERFACE_DEF (url, get_tags),
@@ -339,6 +341,26 @@ lua_url_is_html_displayed (lua_State *L)
 	return 1;
 }
 
+/***
+ * @method url:is_subject()
+ * Check whether URL is found in subject
+ * @return {boolean} `true` if URL is found in subject
+ */
+static gint
+lua_url_is_subject (lua_State *L)
+{
+	struct rspamd_lua_url *url = lua_check_url (L, 1);
+
+	if (url != NULL) {
+		lua_pushboolean (L, url->url->flags & RSPAMD_URL_FLAG_SUBJECT);
+	}
+	else {
+		lua_pushnil (L);
+	}
+
+	return 1;
+}
+
 /***
  * @method url:get_tag(tag)
  * Returns list of string for a specific tagname for an url