Merge pull request #4831 from cyrusimap/parseaddr_utf8_domain

Parse email addresses with non-ASCII UTF8 domain parts

cassandane/data/mime/utf8-domain.bin

@@ -0,0 +1,9 @@
+From: J. Besteiro <jb@julián.example.com>
+To: to@local
+Subject: test
+Date: Mon, 13 Apr 2020 15:34:03 +0200
+MIME-Version: 1.0
+Content-Type: text/plain;charset=us-ascii
+Content-Transfer-Encoding: 7bit
+
+hello

cassandane/tiny-tests/JMAPEmail/email_get_utf8_domain

@@ -0,0 +1,38 @@
+#!perl
+use Cassandane::Tiny;
+
+sub test_email_get_utf8_domain
+    :min_version_3_9 :needs_component_jmap :NoMunge8Bit :RFC2047_UTF8
+{
+    my ($self) = @_;
+    my $jmap = $self->{jmap};
+    my $imap = $self->{store}->get_client();
+
+    open(my $F, "data/mime/utf8-domain.bin") || die $!;
+    $imap->append('INBOX', $F) || die $@;
+    close($F);
+
+    my $res = $jmap->CallMethods([
+        ['Email/query', { }, 'R1'],
+        ['Email/get', {
+            '#ids' => {
+                resultOf => 'R1',
+                name => 'Email/query',
+                path => '/ids'
+            },
+            properties => ['from', 'subject'],
+        }, 'R2'],
+    ]);
+
+use utf8;
+    $self->assert_deep_equals([{
+        name =>  'J. Besteiro',
+        email => 'jb@julián.example.com',
+    }], $res->[1][1]{list}[0]{from});
+no utf8;
+
+    $imap->select('INBOX');
+    $res = $imap->fetch('1:*', 'ENVELOPE');
+    $self->assert_str_equals('"J. Besteiro" <jb@julián.example.com>',
+        $res->{1}{envelope}{From});
+}

changes/next/parseaddr_utf8_domain

@@ -0,0 +1,11 @@
+Description:
+Updates the email address parser to preserve non-ASCII characters
+in the domain part.
+
+
+Config changes:
+None.
+
+
+Upgrade instructions:
+Reconstruct mailboxes with the -G option to force reparsing email headers.

cunit/parseaddr.testc

@@ -367,6 +367,20 @@ static void test_rfc2047_text(void)
     parseaddr_free(a);
 }
 
+static void test_utf8_domain(void)
+{
+    struct address *a;
+
+    a = NULL;
+    parseaddr_list("J. Besteiro <jb@julián.example.com>", &a);
+    CU_ASSERT_PTR_NOT_NULL_FATAL(a);
+    CU_ASSERT_STRING_EQUAL(a->name, "J. Besteiro");
+    CU_ASSERT_STRING_EQUAL(a->mailbox, "jb");
+    CU_ASSERT_STRING_EQUAL(a->domain, "julián.example.com");
+    CU_ASSERT_PTR_NULL(a->next);
+
+    parseaddr_free(a);
+}
 
 static void test_group(void)
 {

imap/mailbox.h

@@ -79,7 +79,7 @@
  * changes to backend_version() in backend.c.
  */
 #define MAILBOX_MINOR_VERSION   19
-#define MAILBOX_CACHE_MINOR_VERSION 11
+#define MAILBOX_CACHE_MINOR_VERSION 12
 
 #define FNAME_HEADER "/cyrus.header"
 #define FNAME_INDEX "/cyrus.index"

lib/parseaddr.c

@@ -301,7 +301,7 @@ static int parseaddr_phrase(char **inp, char **phrasep, const char *specials)
  */
 static int parseaddr_domain(char **inp, char **domainp, char **commentp, int *invalid)
 {
-    int c;
+    u_char c;
     char *src = *inp;
     char *dst;
     char *cdst;
@@ -314,7 +314,7 @@ static int parseaddr_domain(char **inp, char **domainp, char **commentp, int *in
 
     for (;;) {
         c = *src++;
-        if (Uisalnum(c) || c == '-' || c == '[' || c == ']' || c == ':') {
+        if (Uisalnum(c) || c == '-' || c == '[' || c == ']' || c == ':' || c > 127) {
             *dst++ = c;
             if (commentp) *commentp = 0;
         }