www: $MSGID/raw: set charset in HTTP response

By using the charset specified in the message, web browsers are
more likely to display the raw text properly for human readers.

Inspired by a patch by Thomas Weißschuh:
  https://public-inbox.org/meta/20211024214337.161779-3-thomas@t-8ch.de/

Cc: Thomas Weißschuh <thomas@t-8ch.de>

4 files changed, 53 insertions, 21 deletions

diff --git a/lib/PublicInbox/GzipFilter.pm b/lib/PublicInbox/GzipFilter.pm
index c4858a97..e37f1f76 100644
--- a/lib/PublicInbox/GzipFilter.pm
+++ b/lib/PublicInbox/GzipFilter.pm
@@ -46,11 +46,10 @@ sub gz_or_noop {
 sub gzf_maybe ($$) { bless { gz => gz_or_noop(@_) }, __PACKAGE__ }
 
 sub psgi_response {
+        # $code may be an HTTP response code (e.g. 200) or a CODE ref (mbox_hdr)
         my ($self, $code, $res_hdr) = @_;
-        my $env = $self->{env};
-        $self->{gz} //= gz_or_noop($res_hdr, $env);
-        if ($env->{'pi-httpd.async'}) {
-                my $http = $env->{'psgix.io'}; # PublicInbox::HTTP
+        if ($self->{env}->{'pi-httpd.async'}) {
+                my $http = $self->{env}->{'psgix.io'}; # PublicInbox::HTTP
                 $http->{forward} = $self;
                 sub {
                         my ($wcb) = @_; # -httpd provided write callback
@@ -58,6 +57,9 @@ sub psgi_response {
                         $self->can('async_next')->($http); # start stepping
                 };
         } else { # generic PSGI code path
+                ref($code) eq 'CODE' and
+                        ($code, $res_hdr) = @{$code->($self)};
+                $self->{gz} //= gz_or_noop($res_hdr, $self->{env});
                 [ $code, $res_hdr, $self ];
         }
 }
@@ -116,9 +118,13 @@ sub translate ($$) {
 
 sub http_out ($) {
         my ($self) = @_;
-        $self->{http_out} //= do {
+        $self->{http_out} // do {
                 my $args = delete $self->{wcb_args} // return undef;
-                pop(@$args)->($args); # $wcb->([$code, $hdr_ary])
+                my $wcb = pop @$args; # from PublicInbox:HTTP async
+                # $args->[0] may be \&mbox_hdr or similar
+                $args = $args->[0]->($self) if ref($args->[0]) eq 'CODE';
+                $self->{gz} //= gz_or_noop($args->[1], $self->{env});
+                $self->{http_out} = $wcb->($args); # $wcb->([$code, $hdr_ary])
         };
 }
 
@@ -131,6 +137,7 @@ sub write {
 # more data to buffer after this
 sub zmore {
         my $self = $_[0]; # $_[1] => input
+        http_out($self);
         my $err = $self->{gz}->deflate($_[1], $self->{zbuf});
         die "gzip->deflate: $err" if $err != Z_OK;
         undef;
diff --git a/lib/PublicInbox/Mbox.pm b/lib/PublicInbox/Mbox.pm
index 4f84eea6..b977308d 100644
--- a/lib/PublicInbox/Mbox.pm
+++ b/lib/PublicInbox/Mbox.pm
@@ -18,7 +18,7 @@ sub getline {
         my ($ctx) = @_; # ctx
         my $smsg = $ctx->{smsg} or return;
         my $ibx = $ctx->{ibx};
-        my $eml = $ibx->smsg_eml($smsg) or return;
+        my $eml = delete($ctx->{eml}) // $ibx->smsg_eml($smsg) // return;
         my $n = $ctx->{smsg} = $ibx->over->next_by_mid(@{$ctx->{next_arg}});
         $ctx->zmore(msg_hdr($ctx, $eml));
         if ($n) {
@@ -45,14 +45,15 @@ sub async_eml { # for async_blob_cb
         my $smsg = delete $ctx->{smsg};
         # next message
         $ctx->{smsg} = $ctx->{ibx}->over->next_by_mid(@{$ctx->{next_arg}});
-
+        local $ctx->{eml} = $eml; # for mbox_hdr
         $ctx->zmore(msg_hdr($ctx, $eml));
         $ctx->write(msg_body($eml));
 }
 
-sub res_hdr ($$) {
-        my ($ctx, $subject) = @_;
-        my $fn = $subject // '';
+sub mbox_hdr ($) {
+        my ($ctx) = @_;
+        my $eml = $ctx->{eml} //= $ctx->{ibx}->smsg_eml($ctx->{smsg});
+        my $fn = $eml->header_str('Subject') // '';
         $fn =~ s/^re:\s+//i;
         $fn = to_filename($fn) // 'no-subject';
         my @hdr = ('Content-Type');
@@ -64,17 +65,19 @@ sub res_hdr ($$) {
                 push @hdr, 'text/plain';
                 $fn .= '.txt';
         }
+        my $cs = $ctx->{eml}->ct->{attributes}->{charset} // 'UTF-8';
+        $cs = 'UTF-8' if $cs =~ /[^a-zA-Z0-9\-\_]/; # avoid header injection
+        $hdr[-1] .= "; charset=$cs";
         push @hdr, 'Content-Disposition', "inline; filename=$fn";
-        \@hdr;
+        [ 200, \@hdr ];
 }
 
 # for rare cases where v1 inboxes aren't indexed w/ ->over at all
 sub no_over_raw ($) {
         my ($ctx) = @_;
         my $mref = $ctx->{ibx}->msg_by_mid($ctx->{mid}) or return;
-        my $eml = PublicInbox::Eml->new($mref);
-        [ 200, res_hdr($ctx, $eml->header_str('Subject')),
-                [ msg_hdr($ctx, $eml) . msg_body($eml) ] ]
+        my $eml = $ctx->{eml} = PublicInbox::Eml->new($mref);
+        [ @{mbox_hdr($ctx)}, [ msg_hdr($ctx, $eml) . msg_body($eml) ] ]
 }
 
 # /$INBOX/$MESSAGE_ID/raw
@@ -85,9 +88,8 @@ sub emit_raw {
         my ($id, $prev);
         my $mip = $ctx->{next_arg} = [ $ctx->{mid}, \$id, \$prev ];
         my $smsg = $ctx->{smsg} = $over->next_by_mid(@$mip) or return;
-        my $res_hdr = res_hdr($ctx, $smsg->{subject});
         bless $ctx, __PACKAGE__;
-        $ctx->psgi_response(200, $res_hdr);
+        $ctx->psgi_response(\&mbox_hdr);
 }
 
 sub msg_hdr ($$) {
diff --git a/t/plack.t b/t/plack.t
index 40ff2baa..e4dedce6 100644
--- a/t/plack.t
+++ b/t/plack.t
@@ -10,17 +10,24 @@ require_mods(@mods);
 foreach my $mod (@mods) { use_ok $mod; }
 ok(-f $psgi, "psgi example file found");
 my $pfx = 'http://example.com/test';
-# ensure successful message delivery
-my $ibx = create_inbox('test', sub {
+my $eml = eml_load('t/iso-2202-jp.eml');
+# ensure successful message deliveries
+my $ibx = create_inbox('test-1', sub {
         my ($im, $ibx) = @_;
         my $addr = $ibx->{-primary_address};
-        $im->add(PublicInbox::Eml->new(<<EOF)) or BAIL_OUT '->add';
+        $im->add($eml) or xbail '->add';
+        $eml->header_set('Content-Type',
+                "text/plain; charset=\rso\rb\0gus\rithurts");
+        $eml->header_set('Message-ID', '<broken@example.com>');
+        $im->add($eml) or xbail '->add';
+        $im->add(PublicInbox::Eml->new(<<EOF)) or xbail '->add';
 From: Me <me\@example.com>
 To: You <you\@example.com>
 Cc: $addr
 Message-Id: <blah\@example.com>
 Subject: hihi
 Date: Fri, 02 Oct 1993 00:00:00 +0000
+Content-Type: text/plain; charset=iso-8859-1
 
 > quoted text
 zzzzzz
@@ -195,6 +202,19 @@ test_psgi($app, sub {
         my $res = $cb->(GET($pfx . '/blah@example.com/raw'));
         is(200, $res->code, 'success response received for /*/raw');
         like($res->content, qr!^From !sm, "mbox returned");
+        is($res->header('Content-Type'), 'text/plain; charset=iso-8859-1',
+                'charset from message used');
+
+        $res = $cb->(GET($pfx . '/broken@example.com/raw'));
+        is($res->header('Content-Type'), 'text/plain; charset=UTF-8',
+                'broken charset ignored');
+
+        $res = $cb->(GET($pfx . '/199707281508.AAA24167@hoyogw.example/raw'));
+        is($res->header('Content-Type'), 'text/plain; charset=ISO-2022-JP',
+                'ISO-2002-JP returned');
+        chomp(my $body = $res->content);
+        my $raw = PublicInbox::Eml->new(\$body);
+        is($raw->body_raw, $eml->body_raw, 'ISO-2022-JP body unmodified');
 
         $res = $cb->(GET($pfx . '/blah@example.com/t.mbox.gz'));
         is(501, $res->code, '501 when overview missing');
diff --git a/t/psgi_v2.t b/t/psgi_v2.t
index 64c1a8d3..7d73b606 100644
--- a/t/psgi_v2.t
+++ b/t/psgi_v2.t
@@ -20,11 +20,12 @@ To: test@example.com
 Subject: this is a subject
 Message-ID: <a-mid@b>
 Date: Fri, 02 Oct 1993 00:00:00 +0000
+Content-Type: text/plain; charset=iso-8859-1
 
 hello world
 EOF
 my $new_mid;
-my $ibx = create_inbox 'v2', version => 2, indexlevel => 'medium',
+my $ibx = create_inbox 'v2-1', version => 2, indexlevel => 'medium',
                         tmpdir => "$tmpdir/v2", sub {
         my ($im, $ibx) = @_;
         $im->add($eml) or BAIL_OUT;
@@ -68,6 +69,8 @@ my $client0 = sub {
         like($res->content, qr!\$INBOX_DIR/description missing!,
                 'got v2 description missing message');
         $res = $cb->(GET('/v2test/a-mid@b/raw'));
+        is($res->header('Content-Type'), 'text/plain; charset=iso-8859-1',
+                'charset from message used');
         $raw = $res->content;
         unlike($raw, qr/^From oldbug/sm, 'buggy "From_" line omitted');
         like($raw, qr/^hello world$/m, 'got first message');