From 6aedc80226e3cbcb040f79c936933093d92de9da Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Fri, 16 Dec 2016 16:38:59 +0000
Subject: [PATCH] [Rework|Minor] More updates for mime parsing

---
 src/libmime/message.h     | 10 +++++-----
 src/libmime/mime_parser.c |  6 +++---
 src/libserver/protocol.c  |  8 ++++++--
 src/libserver/re_cache.c  | 10 +++++-----
 src/libserver/task.c      |  7 +++++--
 5 files changed, 24 insertions(+), 17 deletions(-)

diff --git a/src/libmime/message.h b/src/libmime/message.h
index 66981ca9e..c4f0e7a8e 100644
--- a/src/libmime/message.h
+++ b/src/libmime/message.h
@@ -36,6 +36,10 @@ enum rspamd_cte {
 
 struct rspamd_mime_text_part;
 
+struct rspamd_mime_multipart {
+	GPtrArray *children;
+};
+
 struct rspamd_mime_part {
 	struct rspamd_content_type *ct;
 	struct rspamd_content_disposition *cd;
@@ -48,7 +52,7 @@ struct rspamd_mime_part {
 	enum rspamd_cte cte;
 
 	union {
-		struct rspamd_mime_multipart *mp;
+		struct rspamd_mime_multipart mp;
 		struct rspamd_mime_text_part *txt;
 		struct rspamd_image *img;
 		struct rspamd_archive *arch;
@@ -58,10 +62,6 @@ struct rspamd_mime_part {
 	guchar digest[rspamd_cryptobox_HASHBYTES];
 };
 
-struct rspamd_mime_multipart {
-	GPtrArray *children;
-};
-
 #define RSPAMD_MIME_TEXT_PART_FLAG_UTF (1 << 0)
 #define RSPAMD_MIME_TEXT_PART_FLAG_BALANCED (1 << 1)
 #define RSPAMD_MIME_TEXT_PART_FLAG_EMPTY (1 << 2)
diff --git a/src/libmime/mime_parser.c b/src/libmime/mime_parser.c
index 7586a803e..0b192b8d5 100644
--- a/src/libmime/mime_parser.c
+++ b/src/libmime/mime_parser.c
@@ -341,8 +341,8 @@ rspamd_mime_process_multipart_node (struct rspamd_task *task,
 
 	hdr_pos = rspamd_string_find_eoh (&str, &body_pos);
 
-	if (multipart->children == NULL) {
-		multipart->children = g_ptr_array_sized_new (2);
+	if (multipart->specific.mp.children == NULL) {
+		multipart->specific.mp.children = g_ptr_array_sized_new (2);
 	}
 
 	npart = rspamd_mempool_alloc0 (task->task_pool,
@@ -350,7 +350,7 @@ rspamd_mime_process_multipart_node (struct rspamd_task *task,
 	npart->parent_part = multipart;
 	npart->raw_headers =  g_hash_table_new_full (rspamd_strcase_hash,
 			rspamd_strcase_equal, NULL, rspamd_ptr_array_free_hard);
-	g_ptr_array_add (multipart->children, npart);
+	g_ptr_array_add (multipart->specific.mp.children, npart);
 
 	if (hdr_pos > 0 && hdr_pos < str.len) {
 			npart->raw_headers_str = str.str;
diff --git a/src/libserver/protocol.c b/src/libserver/protocol.c
index a1a1ae8fe..cdc7f263a 100644
--- a/src/libserver/protocol.c
+++ b/src/libserver/protocol.c
@@ -778,7 +778,7 @@ make_rewritten_subject (struct rspamd_metric *metric, struct rspamd_task *task)
 		c = SPAM_SUBJECT;
 	}
 
-	s = g_mime_message_get_subject (task->message);
+	s = task->subject;
 
 	while (p < end) {
 		if (*c == '\0') {
@@ -794,6 +794,7 @@ make_rewritten_subject (struct rspamd_metric *metric, struct rspamd_task *task)
 		}
 		p++;
 	}
+
 	res = g_mime_utils_header_encode_text (subj_buf);
 
 	rspamd_mempool_add_destructor (task->task_pool,
@@ -897,8 +898,11 @@ rspamd_metric_result_ucl (struct rspamd_task *task,
 
 	if (action == METRIC_ACTION_REWRITE_SUBJECT) {
 		subject = make_rewritten_subject (m, task);
-		ucl_object_insert_key (obj, ucl_object_fromstring (subject),
+
+		if (subject) {
+			ucl_object_insert_key (obj, ucl_object_fromstring (subject),
 				"subject", 0, false);
+		}
 	}
 	/* Now handle symbols */
 	g_hash_table_iter_init (&hiter, mres->symbols);
diff --git a/src/libserver/re_cache.c b/src/libserver/re_cache.c
index 1e713dc86..e5298ff04 100644
--- a/src/libserver/re_cache.c
+++ b/src/libserver/re_cache.c
@@ -884,8 +884,8 @@ rspamd_re_cache_exec_re (struct rspamd_task *task,
 				}
 				/* Select data for regexp */
 				if (re_class->type == RSPAMD_RE_RAWMIME) {
-					in = part->raw->data;
-					len = part->raw->len;
+					in = part->raw.begin;
+					len = part->raw.len;
 					raw = TRUE;
 				}
 				else {
@@ -1022,9 +1022,9 @@ rspamd_re_cache_exec_re (struct rspamd_task *task,
 			for (i = 0; i < task->text_parts->len; i++) {
 				part = g_ptr_array_index (task->text_parts, i);
 
-				if (part->orig) {
-					scvec[i] = (guchar *)part->orig->data;
-					lenvec[i] = part->orig->len;
+				if (part->parsed.len > 0) {
+					scvec[i] = (guchar *)part->parsed.begin;
+					lenvec[i] = part->parsed.len;
 				}
 				else {
 					scvec[i] = (guchar *)"";
diff --git a/src/libserver/task.c b/src/libserver/task.c
index cee604acd..656e177aa 100644
--- a/src/libserver/task.c
+++ b/src/libserver/task.c
@@ -203,8 +203,11 @@ rspamd_task_free (struct rspamd_task *task)
 			if (p->raw_headers) {
 				g_hash_table_unref (p->raw_headers);
 			}
-			if (p->children) {
-				g_ptr_array_free (p->children, TRUE);
+
+			if (IS_CT_MULTIPART (p->ct)) {
+				if (p->specific.mp.children) {
+					g_ptr_array_free (p->specific.mp.children, TRUE);
+				}
 			}
 		}
 
-- 
2.39.5