disallow "\t" and "\n" in OVER headers

For Subject/To/Cc/From headers, we squeeze them to a space (' ').

For Message-IDs (including References/In-Reply-To), '\t', '\n', '\r'
are deleted since some MUAs might screw them up:

  https://public-inbox.org/git/656C30A1EFC89F6B2082D9B6@localhost/raw

4 files changed, 23 insertions, 8 deletions

diff --git a/lib/PublicInbox/MID.pm b/lib/PublicInbox/MID.pm
index c82e8401..cd56f272 100644
--- a/lib/PublicInbox/MID.pm
+++ b/lib/PublicInbox/MID.pm
@@ -87,6 +87,7 @@ sub uniq_mids ($) {
         my @ret;
         my %seen;
         foreach my $mid (@$mids) {
+                $mid =~ tr/\n\t\r//d;
                 if (length($mid) > MAX_MID_SIZE) {
                         warn "Message-ID: <$mid> too long, truncating\n";
                         $mid = substr($mid, 0, MAX_MID_SIZE);
diff --git a/lib/PublicInbox/SearchMsg.pm b/lib/PublicInbox/SearchMsg.pm
index ab971e00..c7787ea1 100644
--- a/lib/PublicInbox/SearchMsg.pm
+++ b/lib/PublicInbox/SearchMsg.pm
@@ -100,7 +100,7 @@ sub __hdr ($$) {
         my $mime = $self->{mime} or return;
         $val = $mime->header($field);
         $val = '' unless defined $val;
-        $val =~ tr/\n/ /;
+        $val =~ tr/\t\n/  /;
         $val =~ tr/\r//d;
         $self->{$field} = $val;
 }
diff --git a/t/mid.t b/t/mid.t
index 223be798..8c307c82 100644
--- a/t/mid.t
+++ b/t/mid.t
@@ -25,6 +25,17 @@ is(mid_escape('foo%!@(bar)'), 'foo%25!@(bar)');
         $mime->header_set('In-Reply-To', '<weld>');
         is_deeply(['hello', 'world', 'weld'], references($mime->header_obj),
                 'references combines with In-Reply-To');
+
+        $mime->header_set('References', "<hello>\n\t<world>");
+        $mime->header_set('In-Reply-To');
+        is_deeply(references($mime->header_obj), ['hello', 'world'],
+                'multiline References OK');
+        $mime->header_set('References', "<hello\tworld>");
+        is_deeply(references($mime->header_obj), ['helloworld'],
+                'drop \t in References <656C30A1EFC89F6B2082D9B6@localhost>');
+        $mime->header_set('Message-ID', "<hello\tworld>");
+        is_deeply(mids($mime->header_obj), ['helloworld'],
+                'drop \t in Message-ID');
 }
 
 done_testing();
diff --git a/t/nntpd.t b/t/nntpd.t
index 3698f98b..960e83c1 100644
--- a/t/nntpd.t
+++ b/t/nntpd.t
@@ -80,9 +80,10 @@ From: =?utf-8?Q?El=C3=A9anor?= <me\@example.com>
 Cc: $addr
 Message-Id: <nntp\@example.com>
 Content-Type: text/plain; charset=utf-8
-Subject: Testing for =?utf-8?Q?El=C3=A9anor?=
+Subject: Testing for        =?utf-8?Q?El=C3=A9anor?=
 Date: Thu, 01 Jan 1970 06:06:06 +0000
 Content-Transfer-Encoding: 8bit
+References: <ref        tab        squeezed>
 
 This is a test message for El\xc3\xa9anor
 EOF
@@ -139,7 +140,8 @@ EOF
                 'from' => "El\xc3\xa9anor <me\@example.com>",
                 'to' => "El\xc3\xa9anor <you\@example.com>",
                 'cc' => $addr,
-                'xref' => "example.com $group:1"
+                'xref' => "example.com $group:1",
+                'references' => '<reftabsqueezed>',
         );
 
         my $s = IO::Socket::INET->new(%opts);
@@ -189,7 +191,7 @@ EOF
                         "El\xc3\xa9anor <me\@example.com>",
                         'Thu, 01 Jan 1970 06:06:06 +0000',
                         '<nntp@example.com>',
-                        '',
+                        '<reftabsqueezed>',
                         $len,
                         '1' ] }, "XOVER range works");
 
@@ -198,7 +200,7 @@ EOF
                         "El\xc3\xa9anor <me\@example.com>",
                         'Thu, 01 Jan 1970 06:06:06 +0000',
                         '<nntp@example.com>',
-                        '',
+                        '<reftabsqueezed>',
                         $len,
                         '1' ] }, "XOVER by article works");
 
@@ -220,14 +222,15 @@ EOF
                 is($r[1], "0\tTesting for El\xc3\xa9anor\t" .
                         "El\xc3\xa9anor <me\@example.com>\t" .
                         "Thu, 01 Jan 1970 06:06:06 +0000\t" .
-                        "$mid\t\t$len\t1", 'OVER by Message-ID works');
+                        "$mid\t<reftabsqueezed>\t$len\t1",
+                        'OVER by Message-ID works');
                 is($r[2], '.', 'correctly terminated response');
         }
 
         is_deeply($n->xhdr(qw(Cc 1-)), { 1 => 'test-nntpd@example.com' },
                  'XHDR Cc 1- works');
-        is_deeply($n->xhdr(qw(References 1-)), { 1 => '' },
-                 'XHDR References 1- works (empty string)');
+        is_deeply($n->xhdr(qw(References 1-)), { 1 => '<reftabsqueezed>' },
+                 'XHDR References 1- works)');
         is_deeply($n->xhdr(qw(list-id 1-)), {},
                  'XHDR on invalid header returns empty');