[Feature] Store headers order

6 files changed, 37 insertions, 16 deletions

diff --git a/src/libmime/message.h b/src/libmime/message.h
index 03e0e8345..22f4fd24d 100644
--- a/src/libmime/message.h
+++ b/src/libmime/message.h
@@ -49,6 +49,7 @@ struct rspamd_mime_part {
 	rspamd_ftok_t parsed_data;
 	struct rspamd_mime_part *parent_part;
 	GHashTable *raw_headers;
+	GQueue *headers_order;
 	gchar *raw_headers_str;
 	gsize raw_headers_len;
 	enum rspamd_cte cte;
diff --git a/src/libmime/mime_headers.c b/src/libmime/mime_headers.c
index 2ea4c0dcb..5044fc013 100644
--- a/src/libmime/mime_headers.c
+++ b/src/libmime/mime_headers.c
@@ -121,7 +121,8 @@ rspamd_mime_header_check_special (struct rspamd_task *task,
 
 static void
 rspamd_mime_header_add (struct rspamd_task *task,
-		GHashTable *target, struct rspamd_mime_header *rh,
+		GHashTable *target, GQueue *order,
+		struct rspamd_mime_header *rh,
 		gboolean check_special)
 {
 	GPtrArray *ar;
@@ -137,6 +138,8 @@ rspamd_mime_header_add (struct rspamd_task *task,
 		msg_debug_task ("add new raw header %s: %s", rh->name, rh->value);
 	}
 
+	g_queue_push_tail (order, rh);
+
 	if (check_special) {
 		rspamd_mime_header_check_special (task, rh);
 	}
@@ -145,7 +148,9 @@ rspamd_mime_header_add (struct rspamd_task *task,
 /* Convert raw headers to a list of struct raw_header * */
 void
 rspamd_mime_headers_process (struct rspamd_task *task, GHashTable *target,
-		const gchar *in, gsize len, gboolean check_newlines)
+		GQueue *order,
+		const gchar *in, gsize len,
+		gboolean check_newlines)
 {
 	struct rspamd_mime_header *nh = NULL;
 	const gchar *p, *c, *end;
@@ -153,7 +158,7 @@ rspamd_mime_headers_process (struct rspamd_task *task, GHashTable *target,
 	gint state = 0, l, next_state = 100, err_state = 100, t_state;
 	gboolean valid_folding = FALSE;
 	guint nlines_count[RSPAMD_TASK_NEWLINES_MAX];
-	guint order = 0;
+	guint norder = 0;
 
 	p = in;
 	end = p + len;
@@ -334,16 +339,16 @@ rspamd_mime_headers_process (struct rspamd_task *task, GHashTable *target,
 
 			/* We also validate utf8 and replace all non-valid utf8 chars */
 			rspamd_mime_charset_utf_enforce (nh->decoded, strlen (nh->decoded));
-			rspamd_mime_header_add (task, target, nh, check_newlines);
-			nh->order = order ++;
+			rspamd_mime_header_add (task, target, order, nh, check_newlines);
+			nh->order = norder ++;
 			state = 0;
 			break;
 		case 5:
 			/* Header has only name, no value */
 			nh->value = "";
 			nh->decoded = "";
-			rspamd_mime_header_add (task, target, nh, check_newlines);
-			nh->order = order ++;
+			rspamd_mime_header_add (task, target, order, nh, check_newlines);
+			nh->order = norder ++;
 			state = 0;
 			break;
 		case 99:
diff --git a/src/libmime/mime_headers.h b/src/libmime/mime_headers.h
index 9e77aa14b..299e84dea 100644
--- a/src/libmime/mime_headers.h
+++ b/src/libmime/mime_headers.h
@@ -47,7 +47,9 @@ struct rspamd_mime_header {
  * @param check_newlines
  */
 void rspamd_mime_headers_process (struct rspamd_task *task, GHashTable *target,
-		const gchar *in, gsize len, gboolean check_newlines);
+		GQueue *order,
+		const gchar *in, gsize len,
+		gboolean check_newlines);
 
 /**
  * Perform rfc2047 decoding of a header
diff --git a/src/libmime/mime_parser.c b/src/libmime/mime_parser.c
index 1626cecbe..9b245e35a 100644
--- a/src/libmime/mime_parser.c
+++ b/src/libmime/mime_parser.c
@@ -517,6 +517,7 @@ 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);
+	npart->headers_order = g_queue_new ();
 	g_ptr_array_add (multipart->specific.mp.children, npart);
 
 	if (hdr_pos > 0 && hdr_pos < str.len) {
@@ -527,6 +528,7 @@ rspamd_mime_process_multipart_node (struct rspamd_task *task,
 
 			if (task->raw_headers_content.len > 0) {
 				rspamd_mime_headers_process (task, npart->raw_headers,
+						npart->headers_order,
 						npart->raw_headers_str,
 						npart->raw_headers_len,
 						FALSE);
@@ -1032,6 +1034,7 @@ rspamd_mime_parse_message (struct rspamd_task *task,
 
 			if (task->raw_headers_content.len > 0) {
 				rspamd_mime_headers_process (task, task->raw_headers,
+						task->headers_order,
 						task->raw_headers_content.begin,
 						task->raw_headers_content.len,
 						TRUE);
@@ -1052,6 +1055,7 @@ rspamd_mime_parse_message (struct rspamd_task *task,
 
 				if (task->raw_headers_content.len > 0) {
 					rspamd_mime_headers_process (task, task->raw_headers,
+							task->headers_order,
 							task->raw_headers_content.begin,
 							task->raw_headers_content.len,
 							TRUE);
@@ -1078,6 +1082,7 @@ rspamd_mime_parse_message (struct rspamd_task *task,
 		hdr_pos = rspamd_string_find_eoh (&str, &body_pos);
 		npart->raw_headers =  g_hash_table_new_full (rspamd_strcase_hash,
 				rspamd_strcase_equal, NULL, rspamd_ptr_array_free_hard);
+		npart->headers_order = g_queue_new ();
 
 		if (hdr_pos > 0 && hdr_pos < str.len) {
 			npart->raw_headers_str = str.str;
@@ -1086,6 +1091,7 @@ rspamd_mime_parse_message (struct rspamd_task *task,
 
 			if (npart->raw_headers_len > 0) {
 				rspamd_mime_headers_process (task, npart->raw_headers,
+						npart->headers_order,
 						npart->raw_headers_str,
 						npart->raw_headers_len,
 						FALSE);
diff --git a/src/libserver/task.c b/src/libserver/task.c
index cbd9a7a31..d2313afd7 100644
--- a/src/libserver/task.c
+++ b/src/libserver/task.c
@@ -90,6 +90,7 @@ rspamd_task_new (struct rspamd_worker *worker, struct rspamd_config *cfg)
 
 	new_task->raw_headers = g_hash_table_new_full (rspamd_strcase_hash,
 			rspamd_strcase_equal, NULL, rspamd_ptr_array_free_hard);
+	new_task->headers_order = g_queue_new ();
 	new_task->request_headers = g_hash_table_new_full (rspamd_ftok_icase_hash,
 			rspamd_ftok_icase_equal, rspamd_fstring_mapped_ftok_free,
 			rspamd_request_header_dtor);
@@ -100,19 +101,22 @@ rspamd_task_new (struct rspamd_worker *worker, struct rspamd_config *cfg)
 			rspamd_ftok_icase_equal, rspamd_fstring_mapped_ftok_free,
 			rspamd_fstring_mapped_ftok_free);
 	rspamd_mempool_add_destructor (new_task->task_pool,
-		(rspamd_mempool_destruct_t) g_hash_table_unref,
-		new_task->reply_headers);
+			(rspamd_mempool_destruct_t) g_hash_table_unref,
+			new_task->reply_headers);
 	rspamd_mempool_add_destructor (new_task->task_pool,
-		(rspamd_mempool_destruct_t) g_hash_table_unref,
-		new_task->raw_headers);
+			(rspamd_mempool_destruct_t) g_hash_table_unref,
+			new_task->raw_headers);
+	rspamd_mempool_add_destructor (new_task->task_pool,
+			(rspamd_mempool_destruct_t) g_queue_free,
+			new_task->headers_order);
 	new_task->emails = g_hash_table_new (rspamd_email_hash, rspamd_emails_cmp);
 	rspamd_mempool_add_destructor (new_task->task_pool,
-		(rspamd_mempool_destruct_t) g_hash_table_unref,
-		new_task->emails);
+			(rspamd_mempool_destruct_t) g_hash_table_unref,
+			new_task->emails);
 	new_task->urls = g_hash_table_new (rspamd_url_hash, rspamd_urls_cmp);
 	rspamd_mempool_add_destructor (new_task->task_pool,
-		(rspamd_mempool_destruct_t) g_hash_table_unref,
-		new_task->urls);
+			(rspamd_mempool_destruct_t) g_hash_table_unref,
+			new_task->urls);
 	new_task->parts = g_ptr_array_sized_new (4);
 	rspamd_mempool_add_destructor (new_task->task_pool,
 			rspamd_ptr_array_free_hard, new_task->parts);
@@ -208,6 +212,8 @@ rspamd_task_free (struct rspamd_task *task)
 				g_hash_table_unref (p->raw_headers);
 			}
 
+			g_queue_free (p->headers_order);
+
 			if (IS_CT_MULTIPART (p->ct)) {
 				if (p->specific.mp.children) {
 					g_ptr_array_free (p->specific.mp.children, TRUE);
diff --git a/src/libserver/task.h b/src/libserver/task.h
index db86a9ac9..e892e776f 100644
--- a/src/libserver/task.h
+++ b/src/libserver/task.h
@@ -157,6 +157,7 @@ struct rspamd_task {
 	GHashTable *urls;								/**< list of parsed urls							*/
 	GHashTable *emails;								/**< list of parsed emails							*/
 	GHashTable *raw_headers;						/**< list of raw headers							*/
+	GQueue *headers_order;							/**< order of raw headers							*/
 	GHashTable *results;							/**< hash table of metric_result indexed by
 													 *    metric's name									*/
 	GHashTable *lua_cache;							/**< cache of lua objects							*/