lib-mail: message_header_hash_more() now allows input in any slices.

There wasn't necessarily any guarantees that the input would be sliced in
such a way that the repeating '?' would be dropped the same way every time.

src/doveadm/dsync/dsync-mail.c

@@ -30,6 +30,7 @@ int dsync_mail_get_hdr_hash(struct mail *mail, unsigned int version,
 {
 	struct istream *hdr_input, *input;
 	struct mailbox_header_lookup_ctx *hdr_ctx;
+	struct message_header_hash_context hash_ctx;
 	struct md5_context md5_ctx;
 	unsigned char md5_result[MD5_RESULTLEN];
 	const unsigned char *data;
@@ -45,13 +46,14 @@ int dsync_mail_get_hdr_hash(struct mail *mail, unsigned int version,
 	input = i_stream_create_lf(hdr_input);
 
 	md5_init(&md5_ctx);
+	memset(&hash_ctx, 0, sizeof(hash_ctx));
 	while (!i_stream_is_eof(input)) {
 		if (i_stream_read_more(input, &data, &size) == -1)
 			break;
 		if (size == 0)
 			break;
-		message_header_hash_more(&hash_method_md5, &md5_ctx, version,
-					 data, size);
+		message_header_hash_more(&hash_ctx, &hash_method_md5, &md5_ctx,
+					 version, data, size);
 		i_stream_skip(input, size);
 	}
 	if (input->stream_errno != 0)

src/lib-mail/message-header-hash.c

@@ -4,7 +4,8 @@
 #include "hash-method.h"
 #include "message-header-hash.h"
 
-void message_header_hash_more(const struct hash_method *method, void *context,
+void message_header_hash_more(struct message_header_hash_context *ctx,
+			      const struct hash_method *method, void *context,
 			      unsigned int version,
 			      const unsigned char *data, size_t size)
 {
@@ -32,12 +33,13 @@ void message_header_hash_more(const struct hash_method *method, void *context,
 		if ((data[i] < 0x20 || data[i] >= 0x7f || data[i] == '?') &&
 		    (data[i] != '\t' && data[i] != '\n')) {
 			/* remove repeated '?' */
-			if (start < i || i == 0) {
+			if (start < i || (i == 0 && !ctx->prev_was_questionmark)) {
 				method->loop(context, data + start, i-start);
 				method->loop(context, "?", 1);
 			}
 			start = i+1;
 		}
 	}
+	ctx->prev_was_questionmark = start == i;
 	method->loop(context, data + start, i-start);
 }

src/lib-mail/message-header-hash.h

@@ -3,7 +3,13 @@
 
 struct hash_method;
 
-void message_header_hash_more(const struct hash_method *method, void *context,
+struct message_header_hash_context {
+	bool prev_was_questionmark;
+};
+
+/* Initialize ctx with zeros. */
+void message_header_hash_more(struct message_header_hash_context *ctx,
+			      const struct hash_method *method, void *context,
 			      unsigned int version,
 			      const unsigned char *data, size_t size);
 

src/lib-mail/test-message-header-hash.c

@@ -14,12 +14,14 @@ static const unsigned char test_output[] =
 
 static void test_dsync_mail_hash_more(void)
 {
+	struct message_header_hash_context ctx;
 	struct md5_context md5_ctx;
 	unsigned char md5_input[MD5_RESULTLEN], md5_output[MD5_RESULTLEN];
 
 	test_begin("dsync_mail_hash_more v2");
 	md5_init(&md5_ctx);
-	message_header_hash_more(&hash_method_md5, &md5_ctx, 2,
+	memset(&ctx, 0, sizeof(ctx));
+	message_header_hash_more(&ctx, &hash_method_md5, &md5_ctx, 2,
 				 test_input, sizeof(test_input)-1);
 	md5_final(&md5_ctx, md5_input);
 
@@ -28,6 +30,17 @@ static void test_dsync_mail_hash_more(void)
 	md5_final(&md5_ctx, md5_output);
 
 	test_assert(memcmp(md5_input, md5_output, MD5_RESULTLEN) == 0);
+
+	/* single byte at a time */
+	md5_init(&md5_ctx);
+	memset(&ctx, 0, sizeof(ctx));
+	for (unsigned int i = 0; i < sizeof(test_input)-1; i++) {
+		message_header_hash_more(&ctx, &hash_method_md5, &md5_ctx, 2,
+					 test_input + i, 1);
+	}
+	md5_final(&md5_ctx, md5_input);
+	test_assert(memcmp(md5_input, md5_output, MD5_RESULTLEN) == 0);
+
 	test_end();
 }
 

src/plugins/pop3-migration/pop3-migration-plugin.c

@@ -178,6 +178,7 @@ int pop3_migration_get_hdr_sha1(uint32_t mail_seq, struct istream *input,
 {
 	const unsigned char *data;
 	size_t size;
+	struct message_header_hash_context hash_ctx;
 	struct sha1_ctxt sha1_ctx;
 	struct pop3_hdr_context hdr_ctx;
 
@@ -190,8 +191,9 @@ int pop3_migration_get_hdr_sha1(uint32_t mail_seq, struct istream *input,
 				pop3_header_filter_callback, &hdr_ctx);
 
 	sha1_init(&sha1_ctx);
+	memset(&hash_ctx, 0, sizeof(hash_ctx));
 	while (i_stream_read_more(input, &data, &size) > 0) {
-		message_header_hash_more(&hash_method_sha1, &sha1_ctx, 2,
+		message_header_hash_more(&hash_ctx, &hash_method_sha1, &sha1_ctx, 2,
 					 data, size);
 		i_stream_skip(input, size);
 	}