1 # Copyright (C) 2015-2016 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
3 # based on notmuch, but with no concept of folders, files or flags
5 # Wraps a document inside our Xapian search index.
6 package PublicInbox::SearchMsg;
10 use Email::Address qw//;
12 use Date::Parse qw/str2time/;
13 use PublicInbox::MID qw/mid_clean/;
14 use Encode qw/find_encoding/;
15 my $enc_utf8 = find_encoding('UTF-8');
16 our $PFX2TERM_RE = undef;
17 use constant EPOCH_822 => 'Thu, 01 Jan 1970 00:00:00 +0000';
18 use POSIX qw(strftime);
21 my ($class, $mime) = @_;
22 my $doc = Search::Xapian::Document->new;
23 $doc->add_term(PublicInbox::Search::xpfx('type') . 'mail');
25 bless { type => 'mail', doc => $doc, mime => $mime }, $class;
29 my ($class, $doc, $mid) = @_;
30 bless { doc => $doc, mime => undef, mid => $mid }, $class;
35 Search::Xapian::sortable_unserialise($doc->get_value($col));
39 my ($class, $doc) = @_;
40 my $data = $doc->get_data;
41 my $ts = get_val($doc, &PublicInbox::Search::TS);
42 $data = $enc_utf8->decode($data);
43 my ($subj, $from, $refs, $to, $cc) = split(/\n/, $data);
55 # :bytes and :lines metadata in RFC 3977
56 sub bytes ($) { get_val($_[0]->{doc}, &PublicInbox::Search::BYTES) }
57 sub lines ($) { get_val($_[0]->{doc}, &PublicInbox::Search::LINES) }
58 sub num ($) { get_val($_[0]->{doc}, &PublicInbox::Search::NUM) }
61 my ($self, $field) = @_;
62 my $val = $self->{$field};
63 return $val if defined $val;
65 my $mime = $self->{mime} or return;
66 $val = $mime->header($field);
67 $val = '' unless defined $val;
70 $self->{$field} = $val;
73 sub subject ($) { __hdr($_[0], 'subject') }
74 sub to ($) { __hdr($_[0], 'to') }
75 sub cc ($) { __hdr($_[0], 'cc') }
79 my $date = __hdr($self, 'date');
80 return $date if defined $date;
82 return unless defined $ts;
83 $self->{date} = strftime('%a, %d %b %Y %T %z', gmtime($ts));
88 my $from = __hdr($self, 'from');
89 if (defined $from && !defined $self->{from_name}) {
90 $from =~ tr/\t\r\n/ /;
91 my @from = Email::Address->parse($from);
92 $self->{from_name} = $from[0]->name;
99 my $from_name = $self->{from_name};
100 return $from_name if defined $from_name;
107 $self->{ts} ||= eval { str2time($self->mime->header('Date')) } || 0;
112 join("\n", $self->subject, $self->from, $self->references,
113 $self->to, $self->cc);
118 my $x = $self->{references};
119 defined $x ? $x : '';
122 sub ensure_metadata {
124 my $doc = $self->{doc};
125 my $end = $doc->termlist_end;
127 unless (defined $PFX2TERM_RE) {
128 my $or = join('|', keys %PublicInbox::Search::PFX2TERM_RMAP);
129 $PFX2TERM_RE = qr/\A($or)/;
132 while (my ($pfx, $field) = each %PublicInbox::Search::PFX2TERM_RMAP) {
133 # ideally we'd move this out of the loop:
134 my $i = $doc->termlist_begin;
138 my $val = $i->get_termname;
140 if ($val =~ s/$PFX2TERM_RE//o) {
141 $self->{$field} = $val;
150 $self->ensure_metadata;
152 'Subject' => $self->subject,
153 'X-PI-From' => $self->from_name,
157 # prevent Email::Simple::Creator from running,
158 # this header is useless for threading as we use X-PI-TS
159 # for sorting and display:
161 'Message-ID' => "<$self->{mid}>",
162 'X-PI-TS' => $self->ts,
164 if (my $refs = $self->{references}) {
165 push @h, References => $refs;
167 my $mime = Email::MIME->create(header_str => \@hs, header => \@h);
168 my $h = $mime->header_obj;
170 # set these headers manually since Encode::encode('MIME-Q', ...)
171 # will add spaces to long values when using header_str above.
173 # drop useless headers Email::MIME set for us
174 $h->header_set('Date');
175 $h->header_set('MIME-Version');
180 my ($self, $mid) = @_;
184 } elsif (my $rv = $self->{mid}) {
187 $self->ensure_metadata; # needed for ghosts
188 $self->{mid} ||= $self->_extract_mid;
192 sub _extract_mid { mid_clean(mid_mime($_[0]->mime)) }
195 my ($self, $mime) = @_;
197 $self->{mime} = $mime;
205 my ($self, $doc_id) = @_;
206 if (defined $doc_id) {
207 $self->{doc_id} = $doc_id;
209 # TODO load from xapian
216 my $tid = $self->{thread};
217 return $tid if defined $tid;
218 $self->ensure_metadata;
224 my $path = $self->{path};
225 return $path if defined $path;
226 $self->ensure_metadata;