From c2c0d9f0847ed304bbeba3544c7744b6b4ec6bab Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Mon, 6 Mar 2017 13:44:13 +0000
Subject: [PATCH] [Fix] Distinguish missing and broken mandatory headers

---
 src/libmime/content_type.h |  1 +
 src/libmime/message.h      |  3 ++-
 src/libmime/mime_parser.c  | 22 ++++++++++++++++++++--
 3 files changed, 23 insertions(+), 3 deletions(-)

diff --git a/src/libmime/content_type.h b/src/libmime/content_type.h
index 25283d365..57fe0e6a3 100644
--- a/src/libmime/content_type.h
+++ b/src/libmime/content_type.h
@@ -27,6 +27,7 @@ enum rspamd_content_type_flags {
 	RSPAMD_CONTENT_TYPE_TEXT = 1 << 2,
 	RSPAMD_CONTENT_TYPE_MESSAGE = 1 << 3,
 	RSPAMD_CONTENT_TYPE_DSN = 1 << 4,
+	RSPAMD_CONTENT_TYPE_MISSING = 1 << 5,
 };
 
 #define IS_CT_MULTIPART(ct) ((ct)->flags & RSPAMD_CONTENT_TYPE_MULTIPART)
diff --git a/src/libmime/message.h b/src/libmime/message.h
index 15fcfcccc..6004d1759 100644
--- a/src/libmime/message.h
+++ b/src/libmime/message.h
@@ -24,7 +24,8 @@ enum rspamd_mime_part_flags {
 	RSPAMD_MIME_PART_ATTACHEMENT = (1 << 1),
 	RSPAMD_MIME_PART_IMAGE = (1 << 2),
 	RSPAMD_MIME_PART_ARCHIVE = (1 << 3),
-	RSPAMD_MIME_PART_BAD_CTE = (1 << 4)
+	RSPAMD_MIME_PART_BAD_CTE = (1 << 4),
+	RSPAMD_MIME_PART_MISSING_CTE = (1 << 5)
 };
 
 enum rspamd_cte {
diff --git a/src/libmime/mime_parser.c b/src/libmime/mime_parser.c
index 54c490109..b16983287 100644
--- a/src/libmime/mime_parser.c
+++ b/src/libmime/mime_parser.c
@@ -243,7 +243,9 @@ rspamd_mime_part_get_cte (struct rspamd_task *task, struct rspamd_mime_part *par
 
 	if (hdrs == NULL) {
 		part->cte = rspamd_mime_part_get_cte_heuristic (task, part);
-		part->flags |= RSPAMD_MIME_PART_BAD_CTE;
+		msg_info_task ("detected missing CTE for part as: %s",
+				rspamd_cte_to_string (part->cte));
+		part->flags |= RSPAMD_MIME_PART_MISSING_CTE;
 	}
 	else {
 		for (i = 0; i < hdrs->len; i ++) {
@@ -262,6 +264,9 @@ rspamd_mime_part_get_cte (struct rspamd_task *task, struct rspamd_mime_part *par
 
 		if (part->cte == RSPAMD_CTE_UNKNOWN) {
 			part->cte = rspamd_mime_part_get_cte_heuristic (task, part);
+
+			msg_info_task ("corrected bad CTE for part to: %s",
+					rspamd_cte_to_string (part->cte));
 		}
 		else if (part->cte == RSPAMD_CTE_B64 || part->cte == RSPAMD_CTE_QP) {
 			/* Additionally check sanity */
@@ -357,6 +362,19 @@ rspamd_mime_parse_normal_part (struct rspamd_task *task,
 	case RSPAMD_CTE_7BIT:
 	case RSPAMD_CTE_8BIT:
 	case RSPAMD_CTE_UNKNOWN:
+		if (part->ct->flags & RSPAMD_CONTENT_TYPE_MISSING) {
+			if (part->cte != RSPAMD_CTE_7BIT) {
+				/* We have something that has a missing content-type,
+				 * but it has non-7bit characters.
+				 *
+				 * In theory, it is very unsafe to process it as a text part
+				 * as we unlikely get some sane result
+				 */
+				part->ct->flags &= ~RSPAMD_CONTENT_TYPE_TEXT;
+				part->ct->flags |= RSPAMD_CONTENT_TYPE_BROKEN;
+			}
+		}
+
 		if (IS_CT_TEXT (part->ct)) {
 			/* Need to copy text as we have couple of in-place change functions */
 			parsed = rspamd_fstring_sized_new (part->raw_data.len);
@@ -1059,7 +1077,7 @@ rspamd_mime_parse_message (struct rspamd_task *task,
 		/* For messages we automatically assume plaintext */
 		msg_info_task ("cannot find content-type for a message, assume text/plain");
 		sel = rspamd_mempool_alloc0 (task->task_pool, sizeof (*sel));
-		sel->flags = RSPAMD_CONTENT_TYPE_TEXT|RSPAMD_CONTENT_TYPE_BROKEN;
+		sel->flags = RSPAMD_CONTENT_TYPE_TEXT|RSPAMD_CONTENT_TYPE_MISSING;
 		RSPAMD_FTOK_ASSIGN (&sel->type, "text");
 		RSPAMD_FTOK_ASSIGN (&sel->subtype, "plain");
 	}
-- 
2.39.5