1 # Copyright (C) all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Unstable internal API.
5 # Used for on-the-fly duplicate detection in V2 inboxes.
6 # This is not stored in any database anywhere and may change
7 # as changes in duplicate detection are needed.
8 # See L<public-inbox-v2-format(5)> manpage for more details.
9 package PublicInbox::ContentHash;
12 use parent qw(Exporter);
13 our @EXPORT_OK = qw(content_hash content_digest git_sha);
14 use PublicInbox::MID qw(mids references);
15 use PublicInbox::MsgIter;
17 # not sure if less-widely supported hash families are worth bothering with
18 use PublicInbox::SHA; # faster, but no ->clone
19 use Digest::SHA; # we still need this for ->clone
21 sub digest_addr ($$$) {
22 my ($dig, $h, $v) = @_;
25 $v =~ s/@([a-z0-9\_\.\-\(\)]*([A-Z])\S*)/'@'.lc($1)/ge;
27 $dig->add("$h\0$v\0");
32 my ($part, $depth, @idx) = @{$_[0]};
33 $dig->add("\0$depth:".join('.', @idx)."\0");
34 my $fn = $part->filename;
37 $dig->add("fn\0$fn\0");
39 my @d = $part->header('Content-Description');
45 my $ct = $part->content_type || 'text/plain';
46 my ($s, undef) = msg_part_text($part, $ct);
57 sub content_digest ($;$) {
59 $dig //= Digest::SHA->new(256);
61 # References: and In-Reply-To: get used interchangeably
62 # in some "duplicates" in LKML. We treat them the same
63 # in SearchIdx, so treat them the same for this:
64 # do NOT consider the Message-ID as part of the content_hash
65 # if we got here, we've already got Message-ID reuse
66 my %seen = map { $_ => 1 } @{mids($eml)};
67 for (grep { !$seen{$_}++ } @{references($eml)}) {
69 $dig->add("ref\0$_\0");
72 # Only use Sender: if From is not present
73 foreach my $h (qw(From Sender)) {
74 my @v = $eml->header($h) or next;
75 digest_addr($dig, $h, $_) foreach @v;
78 foreach my $h (qw(Subject Date)) {
79 my @v = $eml->header($h);
82 $dig->add("$h\0$v\0");
85 # Some mail processors will add " to unquoted names that were
86 # not in the original message. For the purposes of deduplication,
87 # do not take it into account:
88 foreach my $h (qw(To Cc)) {
89 my @v = $eml->header($h);
90 digest_addr($dig, $h, $_) foreach @v;
92 msg_iter($eml, \&content_dig_i, $dig);
96 sub content_hash ($) {
97 content_digest($_[0], PublicInbox::SHA->new(256))->digest;
100 # don't clone the result of this
103 my $dig = PublicInbox::SHA->new($n);
104 my $bref = ref($eml) eq 'SCALAR' ? $eml : \($eml->as_string);
105 $dig->add('blob '.length($$bref)."\0", $$bref);