[Feature] Pre-calculate blake2 digest for all parts

5 files changed, 35 insertions, 14 deletions

diff --git a/src/libmime/images.c b/src/libmime/images.c
index 5a556ce9c..f2c27f865 100644
--- a/src/libmime/images.c
+++ b/src/libmime/images.c
@@ -226,6 +226,7 @@ process_image (struct rspamd_task *task, struct rspamd_mime_part *part)
 			img->width, img->height,
 			task->message_id);
 		img->filename = part->filename;
+		img->parent = part;
 		part->flags |= RSPAMD_MIME_PART_IMAGE;
 		part->specific_data = img;
 
diff --git a/src/libmime/images.h b/src/libmime/images.h
index aa63d92e2..73e00aebd 100644
--- a/src/libmime/images.h
+++ b/src/libmime/images.h
@@ -5,6 +5,7 @@
 
 struct html_image;
 struct rspamd_task;
+struct rspamd_mime_part;
 
 enum rspamd_image_type {
 	IMAGE_TYPE_PNG = 0,
@@ -15,12 +16,13 @@ enum rspamd_image_type {
 };
 
 struct rspamd_image {
-	enum rspamd_image_type type;
+	struct rspamd_mime_part *parent;
 	GByteArray *data;
-	guint32 width;
-	guint32 height;
 	const gchar *filename;
 	struct html_image *html_image;
+	enum rspamd_image_type type;
+	guint32 width;
+	guint32 height;
 };
 
 /*
diff --git a/src/libmime/message.c b/src/libmime/message.c
index a3b673f28..f97de592f 100644
--- a/src/libmime/message.c
+++ b/src/libmime/message.c
@@ -1061,6 +1061,17 @@ mime_foreach_callback (GMimeObject * part, gpointer user_data)
 	GMimeStream *part_stream;
 	GByteArray *part_content;
 	gchar *hdrs;
+	/* Blake2b applied to string 'rspamd' */
+	static const guchar hash_key[] = {
+			0x78,0x6a,0x02,0xf7,0x42,0x01,0x59,0x03,
+			0xc6,0xc6,0xfd,0x85,0x25,0x52,0xd2,0x72,
+			0x91,0x2f,0x47,0x40,0xe1,0x58,0x47,0x61,
+			0x8a,0x86,0xe2,0x17,0xf7,0x1f,0x54,0x19,
+			0xd2,0x5e,0x10,0x31,0xaf,0xee,0x58,0x53,
+			0x13,0x89,0x64,0x44,0x93,0x4e,0xb0,0x4b,
+			0x90,0x3a,0x68,0x5b,0x14,0x48,0xb7,0x55,
+			0xd5,0x6f,0x70,0x1a,0xfe,0x9b,0xe2,0xce
+	};
 
 	task = md->task;
 	/* 'part' points to the current part node that g_mime_message_foreach_part() is iterating over */
@@ -1139,6 +1150,11 @@ mime_foreach_callback (GMimeObject * part, gpointer user_data)
 		mime_part->mime = part;
 		mime_part->boundary = g_mime_multipart_get_boundary (GMIME_MULTIPART (part));
 
+		if (mime_part->boundary) {
+			rspamd_cryptobox_hash (mime_part->digest, mime_part->boundary,
+					strlen (mime_part->boundary), hash_key, sizeof (hash_key));
+		}
+
 		debug_task ("found part with content-type: %s/%s",
 				type->type,
 				type->subtype);
@@ -1203,6 +1219,12 @@ mime_foreach_callback (GMimeObject * part, gpointer user_data)
 							part));
 				mime_part->mime = part;
 
+				if (mime_part->content->len > 0) {
+					rspamd_cryptobox_hash (mime_part->digest,
+							mime_part->content->data, mime_part->content->len,
+							hash_key, sizeof (hash_key));
+				}
+
 				debug_task ("found part with content-type: %s/%s",
 					type->type,
 					type->subtype);
diff --git a/src/libmime/message.h b/src/libmime/message.h
index 3fe26e685..b6f7062dc 100644
--- a/src/libmime/message.h
+++ b/src/libmime/message.h
@@ -9,6 +9,7 @@
 #include "config.h"
 #include "email_addr.h"
 #include "addr.h"
+#include "cryptobox.h"
 #include <gmime/gmime.h>
 
 struct rspamd_task;
@@ -29,11 +30,11 @@ struct rspamd_mime_part {
 	GMimeObject *mime;
 	GHashTable *raw_headers;
 	gchar *raw_headers_str;
-	gchar *checksum;
+	guchar digest[rspamd_cryptobox_HASHBYTES];
 	const gchar *filename;
 	const gchar *boundary;
-	enum rspamd_mime_part_flags flags;
 	gpointer specific_data;
+	enum rspamd_mime_part_flags flags;
 };
 
 #define RSPAMD_MIME_TEXT_PART_FLAG_UTF (1 << 0)
diff --git a/src/plugins/fuzzy_check.c b/src/plugins/fuzzy_check.c
index 82800e714..429360e1c 100644
--- a/src/plugins/fuzzy_check.c
+++ b/src/plugins/fuzzy_check.c
@@ -1288,13 +1288,11 @@ fuzzy_cmd_from_data_part (struct fuzzy_rule *rule,
 		gint flag,
 		guint32 weight,
 		rspamd_mempool_t *pool,
-		const guchar *data,
-		gsize datalen)
+		guchar digest[rspamd_cryptobox_HASHBYTES])
 {
 	struct rspamd_fuzzy_cmd *cmd;
 	struct rspamd_fuzzy_encrypted_cmd *enccmd = NULL;
 	struct fuzzy_cmd_io *io;
-	rspamd_cryptobox_hash_state_t st;
 
 	if (rule->peer_key) {
 		enccmd = rspamd_mempool_alloc0 (pool, sizeof (*enccmd));
@@ -1312,10 +1310,7 @@ fuzzy_cmd_from_data_part (struct fuzzy_rule *rule,
 	}
 	cmd->shingles_count = 0;
 	cmd->tag = ottery_rand_uint32 ();
-	/* Use blake2b for digest */
-	rspamd_cryptobox_hash_init (&st, rule->hash_key->str, rule->hash_key->len);
-	rspamd_cryptobox_hash_update (&st, data, datalen);
-	rspamd_cryptobox_hash_final (&st, cmd->digest);
+	memcpy (cmd->digest, digest, sizeof (cmd->digest));
 
 	io = rspamd_mempool_alloc (pool, sizeof (*io));
 	io->flags = 0;
@@ -2059,7 +2054,7 @@ fuzzy_generate_commands (struct rspamd_task *task, struct fuzzy_rule *rule,
 							fuzzy_module_ctx->min_width) {
 						io = fuzzy_cmd_from_data_part (rule, c, flag, value,
 								task->task_pool,
-								image->data->data, image->data->len);
+								image->parent->digest);
 						if (io) {
 							g_ptr_array_add (res, io);
 						}
@@ -2074,7 +2069,7 @@ fuzzy_generate_commands (struct rspamd_task *task, struct fuzzy_rule *rule,
 				fuzzy_module_ctx->min_bytes) {
 				io = fuzzy_cmd_from_data_part (rule, c, flag, value,
 						task->task_pool,
-						mime_part->content->data, mime_part->content->len);
+						mime_part->digest);
 				if (io) {
 					g_ptr_array_add (res, io);
 				}