From e33ea5e4782b4a6e3f706634d85058d1a4f42d6b Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Mon, 15 Aug 2016 18:32:42 +0100
Subject: [PATCH] [Feature] Store task checksum

---
 src/libmime/message.c | 29 +++++++++++++++++++++++++----
 src/libserver/task.h  |  1 +
 2 files changed, 26 insertions(+), 4 deletions(-)

diff --git a/src/libmime/message.c b/src/libmime/message.c
index 09176cfe30..1f63568813 100644
--- a/src/libmime/message.c
+++ b/src/libmime/message.c
@@ -1375,6 +1375,8 @@ rspamd_message_parse (struct rspamd_task *task)
 	gint i;
 	gdouble diff, *pdiff;
 	guint tw, *ptw, dw;
+	rspamd_cryptobox_hash_state_t st;
+	guchar digest_out[rspamd_cryptobox_HASHBYTES];
 
 	if (RSPAMD_TASK_IS_EMPTY (task)) {
 		/* Don't do anything with empty task */
@@ -1430,6 +1432,7 @@ rspamd_message_parse (struct rspamd_task *task)
 	 * pool allocator
 	 */
 	g_mime_stream_mem_set_owner (GMIME_STREAM_MEM (stream), FALSE);
+	rspamd_cryptobox_hash_init (&st, NULL, 0);
 
 	if (task->flags & RSPAMD_TASK_FLAG_MIME) {
 
@@ -1474,6 +1477,10 @@ rspamd_message_parse (struct rspamd_task *task)
 				task->raw_headers_content.len = hdr_pos;
 				task->raw_headers_content.body_start = p + body_pos;
 
+				rspamd_cryptobox_hash_update (&st,
+						task->raw_headers_content.begin,
+						task->raw_headers_content.len);
+
 				if (task->raw_headers_content.len > 0) {
 					process_raw_headers (task, task->raw_headers,
 							task->raw_headers_content.begin,
@@ -1718,13 +1725,27 @@ rspamd_message_parse (struct rspamd_task *task)
 				"them with each other");
 	}
 
+	for (i = 0; i < task->parts->len; i ++) {
+		struct rspamd_mime_part *part;
+
+		part = g_ptr_array_index (task->parts, i);
+		rspamd_cryptobox_hash_update (&st, part->digest, sizeof (part->digest));
+	}
+
+	rspamd_cryptobox_hash_final (&st, digest_out);
+	memcpy (task->digest, digest_out, sizeof (task->digest));
+
 	if (task->queue_id) {
-		msg_info_task ("loaded message; id: <%s>; queue-id: <%s>; size: %z",
-				task->message_id, task->queue_id, task->msg.len);
+		msg_info_task ("loaded message; id: <%s>; queue-id: <%s>; size: %z; "
+				"checksum: <%*xs>",
+				task->message_id, task->queue_id, task->msg.len,
+				(gint)sizeof (task->digest), task->digest);
 	}
 	else {
-		msg_info_task ("loaded message; id: <%s>; size: %z",
-				task->message_id, task->msg.len);
+		msg_info_task ("loaded message; id: <%s>; size: %z; "
+				"checksum: <%*xs>",
+				task->message_id, task->msg.len,
+				(gint)sizeof (task->digest), task->digest);
 	}
 
 	return TRUE;
diff --git a/src/libserver/task.h b/src/libserver/task.h
index c76dd614e4..f436c8275b 100644
--- a/src/libserver/task.h
+++ b/src/libserver/task.h
@@ -188,6 +188,7 @@ struct rspamd_task {
 	ucl_object_t *settings;							/**< Settings applied to task						*/
 
 	const gchar *classifier;						/**< Classifier to learn (if needed)				*/
+	guchar digest[16];
 };
 
 /**
-- 
2.47.3