lib-mail: rfc822-parser: Add nul_replacement_char

src/lib-mail/rfc822-parser.c

@@ -75,6 +75,16 @@ int rfc822_skip_comment(struct rfc822_parser_context *ctx)
 	start = ++ctx->data;
 	for (; ctx->data < ctx->end; ctx->data++) {
 		switch (*ctx->data) {
+		case '\0':
+			if (ctx->last_comment != NULL &&
+			    ctx->nul_replacement_char != '\0') {
+				str_append_data(ctx->last_comment, start,
+						ctx->data - start);
+				str_append_c(ctx->last_comment,
+					     ctx->nul_replacement_char);
+				start = ctx->data + 1;
+			}
+			break;
 		case '(':
 			level++;
 			break;
@@ -103,8 +113,9 @@ int rfc822_skip_comment(struct rfc822_parser_context *ctx)
 			if (ctx->data >= ctx->end)
 				return -1;
 
-			if (*ctx->data == '\r' || *ctx->data == '\n') {
-				/* quoted-pair doesn't allow CR/LF.
+			if (*ctx->data == '\r' || *ctx->data == '\n' ||
+			    *ctx->data == '\0') {
+				/* quoted-pair doesn't allow CR/LF/NUL.
 				   They are part of the obs-qp though, so don't
 				   return them as error. */
 				ctx->data--;
@@ -235,6 +246,13 @@ int rfc822_parse_quoted_string(struct rfc822_parser_context *ctx, string_t *str)
 
 	for (start = ctx->data; ctx->data < ctx->end; ctx->data++) {
 		switch (*ctx->data) {
+		case '\0':
+			if (ctx->nul_replacement_char != '\0') {
+				str_append_data(str, start, ctx->data - start);
+				str_append_c(str, ctx->nul_replacement_char);
+				start = ctx->data + 1;
+			}
+			break;
 		case '"':
 			str_append_data(str, start, ctx->data - start);
 			ctx->data++;
@@ -345,6 +363,13 @@ rfc822_parse_domain_literal(struct rfc822_parser_context *ctx, string_t *str)
 
 	for (start = ctx->data++; ctx->data < ctx->end; ctx->data++) {
 		switch (*ctx->data) {
+		case '\0':
+			if (ctx->nul_replacement_char != '\0') {
+				str_append_data(str, start, ctx->data - start);
+				str_append_c(str, ctx->nul_replacement_char);
+				start = ctx->data + 1;
+			}
+			break;
 		case '[':
 			/* not allowed */
 			return -1;

src/lib-mail/rfc822-parser.h

@@ -1,9 +1,15 @@
 #ifndef RFC822_PARSER_H
 #define RFC822_PARSER_H
 
+/* This can be used as a common NUL replacement character */
+#define RFC822_NUL_REPLACEMENT_CHAR 0x80
+
 struct rfc822_parser_context {
 	const unsigned char *data, *end;
 	string_t *last_comment;
+
+	/* Replace NULs with this character */
+	char nul_replacement_char;
 };
 
 #define IS_ATEXT(c) \

src/lib-mail/test-rfc822-parser.c

@@ -50,6 +50,32 @@ static void test_rfc822_parse_comment(void)
 	test_end();
 }
 
+static void test_rfc822_parse_comment_nuls(void)
+{
+	const unsigned char input[] = "(\000a\000\000b\\\000c(\000d)\000)";
+	const char output[] = "!a!!b\\!c(!d)!";
+	struct rfc822_parser_context parser;
+	string_t *str = t_str_new(64);
+
+	test_begin("rfc822 parse comment with NULs");
+
+	rfc822_parser_init(&parser, input, sizeof(input)-1, str);
+	test_assert(rfc822_skip_comment(&parser) == 0);
+	/* should be same as input, except the outer () removed */
+	test_assert(str_len(str) == sizeof(input)-1-2 &&
+		    memcmp(input+1, str_data(str), str_len(str)) == 0);
+	rfc822_parser_deinit(&parser);
+
+	str_truncate(str, 0);
+	rfc822_parser_init(&parser, input, sizeof(input)-1, str);
+	parser.nul_replacement_char = '!';
+	test_assert(rfc822_skip_comment(&parser) == 0);
+	test_assert(strcmp(str_c(str), output) == 0);
+	rfc822_parser_deinit(&parser);
+
+	test_end();
+}
+
 static void test_rfc822_parse_quoted_string(void)
 {
 	static const struct {
@@ -154,6 +180,7 @@ int main(void)
 {
 	static void (*const test_functions[])(void) = {
 		test_rfc822_parse_comment,
+		test_rfc822_parse_comment_nuls,
 		test_rfc822_parse_quoted_string,
 		test_rfc822_parse_domain_literal,
 		test_rfc822_parse_content_param,