From: Arnt Gulbrandsen <arnt@gulbrandsen.priv.no>
Date: Fri, 1 Mar 2024 16:47:28 +0000 (+0100)
Subject: lib-mail: Make FETCH ENVELOPE and SEARCH work on the human-readable domains
X-Git-Tag: 2.4.0~1555
X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=a15333dd35768ea1878217b7f3eec080adff91b0;p=thirdparty%2Fdovecot%2Fcore.git

lib-mail: Make FETCH ENVELOPE and SEARCH work on the human-readable domains

This stores the unicode form of domains in all indexes, meaning that
searching uses and serverside parsing shows the human-readable form of all
addresses.
---

diff --git a/configure.ac b/configure.ac
index ba6e1bd5cf..66e3f5eaba 100644
--- a/configure.ac
+++ b/configure.ac
@@ -50,6 +50,13 @@ AC_ARG_ENABLE([pro-build],
   [is_pro_build=no]
 )
 
+AC_ARG_ENABLE(experimental-mail-utf8,
+AS_HELP_STRING([--enable-experimental-mail-utf8], [Enable experimental support for SMTPUTF8 and UTF8=ACCEPT]),
+	AS_IF([test x$enableval = xyes], [
+	       AC_DEFINE([EXPERIMENTAL_MAIL_UTF8],, [Build with SMTPUTF8 and UTF8=ACCEPT support])
+	])
+)
+
 AC_ARG_WITH(shared-libs,
 AS_HELP_STRING([--with-shared-libs], [Link binaries using shared Dovecot libraries (default)]),
 	want_shared_libs=$withval,
diff --git a/src/lib-mail/rfc822-parser.c b/src/lib-mail/rfc822-parser.c
index c8595b4187..606ede8715 100644
--- a/src/lib-mail/rfc822-parser.c
+++ b/src/lib-mail/rfc822-parser.c
@@ -2,6 +2,7 @@
 
 #include "lib.h"
 #include "str.h"
+#include "punycode.h"
 #include "strescape.h"
 #include "rfc822-parser.h"
 
@@ -412,6 +413,34 @@ rfc822_parse_domain_literal(struct rfc822_parser_context *ctx, string_t *str)
 	return -1;
 }
 
+void rfc822_decode_punycode(const char *input, size_t len, string_t *result)
+{
+	string_t *decoded = t_str_new(64);
+	const char *pos = input;
+	const char *end = CONST_PTR_OFFSET(input, len);
+
+	while (pos < end) {
+		const char *value;
+		const char *delim = strchr(pos, '.');
+		if (delim == NULL)
+			delim = end;
+		if (str_begins(pos, "xn--", &value)) {
+			str_truncate(decoded, 0);
+			if (punycode_decode(value, delim - value, result) < 0)
+				/* Consider it as data */
+				str_append_data(result, pos, delim - pos + 1);
+			else if (*delim == '.')
+				str_append_c(result, *delim);
+		} else {
+			/* No punycode prefix */
+			str_append_data(result, pos, delim - pos + 1);
+		}
+		pos = delim + 1;
+	}
+	if (pos < end)
+		str_append_data(result, pos, end - pos);
+}
+
 int rfc822_parse_domain(struct rfc822_parser_context *ctx, string_t *str)
 {
 	/*
@@ -428,8 +457,21 @@ int rfc822_parse_domain(struct rfc822_parser_context *ctx, string_t *str)
 
 	if (*ctx->data == '[')
 		return rfc822_parse_domain_literal(ctx, str);
-	else
-		return rfc822_parse_dot_atom(ctx, str);
+	else {
+		int ret = rfc822_parse_dot_atom(ctx, str);
+#ifdef EXPERIMENTAL_MAIL_UTF8
+		if (ret == 0) {
+			size_t start_pos = str_len(str);
+			string_t *u = t_str_new(64);
+			const char *data = t_strndup(str_data(str) + start_pos,
+						     str_len(str) - start_pos);
+			rfc822_decode_punycode(data, strlen(data), u);
+			str_truncate(str, start_pos);
+			str_append_str(str, u);
+		}
+#endif
+		return ret;
+	}
 }
 
 int rfc822_parse_content_type(struct rfc822_parser_context *ctx, string_t *str)
diff --git a/src/lib-mail/rfc822-parser.h b/src/lib-mail/rfc822-parser.h
index c001f761f0..84e9bda671 100644
--- a/src/lib-mail/rfc822-parser.h
+++ b/src/lib-mail/rfc822-parser.h
@@ -68,4 +68,8 @@ int rfc822_parse_content_type(struct rfc822_parser_context *ctx, string_t *str);
 int rfc822_parse_content_param(struct rfc822_parser_context *ctx,
 			       const char **key_r, string_t *value);
 
+/* Decode a punycode-encoded domain name and return the UTF8
+   form in result. Returns 0 on success and -1 on failure. */
+void rfc822_decode_punycode(const char *input, size_t len, string_t *result);
+
 #endif
diff --git a/src/lib-mail/test-rfc822-parser.c b/src/lib-mail/test-rfc822-parser.c
index a0e7ad04b8..4a95df9a5d 100644
--- a/src/lib-mail/test-rfc822-parser.c
+++ b/src/lib-mail/test-rfc822-parser.c
@@ -199,6 +199,30 @@ static void test_rfc822_parse_domain_literal(void)
 	test_end();
 }
 
+static void test_rfc822_decode_punycode(void)
+{
+	const struct test_case {
+		const char *in;
+		const char *out;
+	} cases[] = {
+		{ .in = "xn--gr-zia.org", .out = "gr\xc3\xa5.org" },
+		{ .in = "xn--gr-zia", "gr\xc3\xa5" },
+		{ .in = "org.xn--gr-zia", "org.gr\xc3\xa5" },
+		{ .in = "org.xn--gr-zia.org", "org.gr\xc3\xa5.org" },
+		{ .in = "org.xn--zz-zzzz.org", "org.xn--zz-zzzz.org" },
+	};
+	string_t *res = t_str_new(64);
+
+	test_begin("rfc822 decode punycode");
+	for (size_t i = 0; i < N_ELEMENTS(cases); i++) {
+		str_truncate(res, 0);
+		rfc822_decode_punycode(cases[i].in, strlen(cases[i].in), res);
+		test_assert_strcmp_idx(str_c(res),
+				       cases[i].out, i);
+	}
+	test_end();
+}
+
 #undef TEST_STRING
 #define TEST_STRING(a) .input = (const unsigned char*)a, .input_len = sizeof(a)-1
 
@@ -436,6 +460,7 @@ int main(void)
 		test_rfc822_parse_quoted_string,
 		test_rfc822_parse_dot_atom,
 		test_rfc822_parse_domain_literal,
+		test_rfc822_decode_punycode,
 		test_rfc822_parse_content_type,
 		test_rfc822_parse_content_param,
 		test_rfc822_parse_content_type_param,