From 46acf7be5791a164f8fa5dbc16494d8175e5b850 Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Fri, 14 Nov 2014 15:08:02 +0000
Subject: [PATCH] Ignore SGML style tags in html.

---
 src/libserver/html.c | 6 +++++-
 src/libserver/html.h | 2 ++
 2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/src/libserver/html.c b/src/libserver/html.c
index fdb6a11db..edf831238 100644
--- a/src/libserver/html.c
+++ b/src/libserver/html.c
@@ -493,6 +493,10 @@ construct_html_node (rspamd_mempool_t * pool, gchar *text, gsize tag_len)
 		html->flags |= FL_XML;
 		html->tag = NULL;
 	}
+	else if (*text == '!') {
+		html->flags |= FL_SGML;
+		html->tag = NULL;
+	}
 	else {
 		if (*text == '/') {
 			html->flags |= FL_CLOSING;
@@ -972,7 +976,7 @@ add_html_node (struct rspamd_task *task,
 				part->is_balanced = FALSE;
 			}
 		}
-		else {
+		else if ((data->flags & (FL_XML|FL_SGML)) == 0) {
 
 			g_node_append (*cur_level, new);
 			if ((data->flags & FL_CLOSED) == 0) {
diff --git a/src/libserver/html.h b/src/libserver/html.h
index 6709ed151..1760ad0ca 100644
--- a/src/libserver/html.h
+++ b/src/libserver/html.h
@@ -192,6 +192,8 @@ typedef enum
 #define FL_CLOSING      (1 << 1)
 /* Fully closed tag (e.g. <a attrs />) */
 #define FL_CLOSED       (1 << 2)
+/* <! SGML tag */
+#define FL_SGML			(1 << 3)
 
 struct html_tag {
 	tag_id_t id;
-- 
2.39.5