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;
11 use Date::Parse qw/str2time/;
12 use PublicInbox::MID qw/mid_clean/;
13 use PublicInbox::Address;
14 our $PFX2TERM_RE = undef;
15 use POSIX qw(strftime);
18 my ($class, $mime) = @_;
19 my $doc = Search::Xapian::Document->new;
20 $doc->add_term(PublicInbox::Search::xpfx('type') . 'mail');
22 bless { type => 'mail', doc => $doc, mime => $mime }, $class;
26 my ($class, $doc, $mid) = @_;
27 bless { doc => $doc, mime => undef, mid => $mid }, $class;
32 Search::Xapian::sortable_unserialise($doc->get_value($col));
36 my ($class, $doc) = @_;
37 my $data = $doc->get_data or return;
38 my $ts = get_val($doc, &PublicInbox::Search::TS);
40 my ($subj, $from, $refs, $to, $cc, $blob) = split(/\n/, $data);
53 # :bytes and :lines metadata in RFC 3977
54 sub bytes ($) { get_val($_[0]->{doc}, &PublicInbox::Search::BYTES) }
55 sub lines ($) { get_val($_[0]->{doc}, &PublicInbox::Search::LINES) }
56 sub num ($) { get_val($_[0]->{doc}, &PublicInbox::Search::NUM) }
59 my ($self, $field) = @_;
60 my $val = $self->{$field};
61 return $val if defined $val;
63 my $mime = $self->{mime} or return;
64 $val = $mime->header($field);
65 $val = '' unless defined $val;
68 $self->{$field} = $val;
71 sub subject ($) { __hdr($_[0], 'subject') }
72 sub to ($) { __hdr($_[0], 'to') }
73 sub cc ($) { __hdr($_[0], 'cc') }
77 my $date = __hdr($self, 'date');
78 return $date if defined $date;
80 return unless defined $ts;
81 $self->{date} = strftime('%a, %d %b %Y %T +0000', gmtime($ts));
86 my $from = __hdr($self, 'from');
87 if (defined $from && !defined $self->{from_name}) {
88 my @n = PublicInbox::Address::names($from);
89 $self->{from_name} = join(', ', @n);
96 my $from_name = $self->{from_name};
97 return $from_name if defined $from_name;
104 $self->{ts} ||= eval { str2time($self->mime->header('Date')) } || 0;
108 my ($self, $blob) = @_;
109 my @rows = ($self->subject, $self->from, $self->references,
110 $self->to, $self->cc);
111 push @rows, $blob if defined $blob;
117 my $x = $self->{references};
118 defined $x ? $x : '';
121 sub ensure_metadata {
123 my $doc = $self->{doc};
124 my $end = $doc->termlist_end;
126 unless (defined $PFX2TERM_RE) {
127 my $or = join('|', keys %PublicInbox::Search::PFX2TERM_RMAP);
128 $PFX2TERM_RE = qr/\A($or)/;
131 while (my ($pfx, $field) = each %PublicInbox::Search::PFX2TERM_RMAP) {
132 # ideally we'd move this out of the loop:
133 my $i = $doc->termlist_begin;
137 my $val = $i->get_termname;
139 if ($val =~ s/$PFX2TERM_RE//o) {
140 $self->{$field} = $val;
147 my ($self, $mid) = @_;
151 } elsif (my $rv = $self->{mid}) {
154 $self->ensure_metadata; # needed for ghosts
155 $self->{mid} ||= $self->_extract_mid;
159 sub _extract_mid { mid_clean(mid_mime($_[0]->mime)) }
162 my ($self, $x40) = @_;
164 $self->{blob} = $x40;
171 my ($self, $mime) = @_;
173 $self->{mime} = $mime;
181 my ($self, $doc_id) = @_;
182 if (defined $doc_id) {
183 $self->{doc_id} = $doc_id;
185 # TODO load from xapian
192 my $tid = $self->{thread};
193 return $tid if defined $tid;
194 $self->ensure_metadata;
200 my $path = $self->{path};
201 return $path if defined $path;
202 $self->ensure_metadata;