1 # Copyright (C) 2016 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # base class for creating per-list or per-project filters
5 package PublicInbox::Filter::Base;
8 use PublicInbox::MsgIter;
9 use constant MAX_MID_SIZE => 244; # max term size - 1 in Xapian
11 my $NO_HTML = '*** We only accept plain-text mail, no HTML ***';
13 reject_suffix => [ qw(exe bat cmd com pif scr vbs cpl zip) ],
14 reject_type => [ "text/html:$NO_HTML", "text/xhtml:$NO_HTML",
15 'application/vnd.ms-*:No proprietary data formats' ],
17 our $INVALID_FN = qr/\0/;
20 sub ACCEPT { scalar @_ > 1 ? $_[1] : 1 }
23 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
26 $glob =~ s!(.)!$patmap{$1} || "\Q$1"!ge;
31 my ($class, %opts) = @_;
32 my $self = bless { err => '', %opts }, $class;
33 foreach my $f (qw(reject_suffix reject_type)) {
35 $self->{$f} = $DEFAULTS{$f} unless exists $self->{$f};
37 if (defined $self->{reject_suffix}) {
38 my $tmp = $self->{reject_suffix};
39 $tmp = join('|', map { glob2pat($_) } @$tmp);
40 $self->{reject_suffix} = qr/\.($tmp)\s*\z/i;
43 if (defined $self->{reject_type}) {
44 my $tmp = $self->{reject_type};
46 my ($type, $msg) = split(':', $_, 2);
48 $msg ||= "Unacceptable Content-Type: $type";
49 my $re = glob2pat($type);
50 [ qr/\b$re\b/i, $msg ];
53 $self->{reject_type} = $rt;
58 my ($self, $reason) = @_;
59 $self->{err} = $reason;
63 sub err ($) { $_[0]->{err} }
65 # by default, scrub is a no-op, see PublicInbox::Filter::Vger::scrub
66 # for an example of the override
68 my ($self, $mime) = @_;
74 my ($self, $mime) = @_;
76 my $rt = $self->{reject_type};
77 my $reject_suffix = $self->{reject_suffix} || $INVALID_FN;
81 my ($part, $depth, @idx) = @{$_[0]};
83 my $ct = $part->content_type || 'text/plain';
84 foreach my $p (@$rt) {
90 my $fn = $part->filename;
91 if (defined($fn) && $fn =~ $reject_suffix) {
98 push @r, sort keys %type;
101 push @r, 'Rejected suffixes(s): '.join(', ', sort keys %sfx);
104 @r ? $self->reject(join("\n", @r)) : $self->scrub($mime);