1 # Copyright (C) 2015-2018 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 PublicInbox::MID qw/mid_clean mid_mime/;
11 use PublicInbox::Address;
12 use PublicInbox::MsgTime qw(msg_timestamp);
15 my ($class, $mime) = @_;
16 my $doc = Search::Xapian::Document->new;
17 $doc->add_boolean_term('T' . 'mail');
19 bless { type => 'mail', doc => $doc, mime => $mime }, $class;
23 my ($class, $doc, $mid) = @_;
24 bless { doc => $doc, mime => undef, mid => $mid }, $class;
29 Search::Xapian::sortable_unserialise($doc->get_value($col));
32 sub load_from_data ($$) {
33 my ($self) = $_[0]; # data = $_[1]
34 my ($subj, $from, $refs, $to, $cc, $blob, $mid0) = split(/\n/, $_[1]);
35 $self->{subject} = $subj;
36 $self->{from} = $from;
37 $self->{references} = $refs;
40 $self->{blob} = $blob;
46 my $doc = $self->{doc};
47 my $data = $doc->get_data or return;
48 $self->{ts} = get_val($doc, &PublicInbox::Search::TS);
50 load_from_data($self, $data);
55 my ($class, $doc) = @_;
56 my $data = $doc->get_data or return;
57 my $ts = get_val($doc, &PublicInbox::Search::TS);
59 my $self = bless { doc => $doc, ts => $ts }, $class;
60 load_from_data($self, $data);
64 # :bytes and :lines metadata in RFC 3977
65 sub bytes ($) { get_val($_[0]->{doc}, &PublicInbox::Search::BYTES) }
66 sub lines ($) { get_val($_[0]->{doc}, &PublicInbox::Search::LINES) }
68 $_[0]->{num} ||= get_val($_[0]->{doc}, PublicInbox::Search::NUM)
72 my ($self, $field) = @_;
73 my $val = $self->{$field};
74 return $val if defined $val;
76 my $mime = $self->{mime} or return;
77 $val = $mime->header($field);
78 $val = '' unless defined $val;
81 $self->{$field} = $val;
84 sub subject ($) { __hdr($_[0], 'subject') }
85 sub to ($) { __hdr($_[0], 'to') }
86 sub cc ($) { __hdr($_[0], 'cc') }
88 # no strftime, that is locale-dependent and not for RFC822
89 my @DoW = qw(Sun Mon Tue Wed Thu Fri Sat);
90 my @MoY = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
95 return unless defined $ts;
96 my ($sec, $min, $hour, $mday, $mon, $year, $wday) = gmtime($ts);
97 "$DoW[$wday], " . sprintf("%02d $MoY[$mon] %04d %02d:%02d:%02d +0000",
98 $mday, $year+1900, $hour, $min, $sec);
104 my $from = __hdr($self, 'from');
105 if (defined $from && !defined $self->{from_name}) {
106 my @n = PublicInbox::Address::names($from);
107 $self->{from_name} = join(', ', @n);
114 my $from_name = $self->{from_name};
115 return $from_name if defined $from_name;
122 $self->{ts} ||= eval {
123 msg_timestamp($self->{mime}->header_obj);
128 my ($self, $oid, $mid0) = @_;
129 my @rows = ($self->subject, $self->from, $self->references,
130 $self->to, $self->cc);
131 $oid = '' unless defined $oid;
132 join("\n", @rows, $oid, $mid0);
137 my $x = $self->{references};
138 defined $x ? $x : '';
141 sub _get_term_val ($$$) {
142 my ($self, $pfx, $re) = @_;
143 my $doc = $self->{doc};
144 my $end = $doc->termlist_end;
145 my $i = $doc->termlist_begin;
148 my $val = $i->get_termname;
149 $val =~ s/$re// and return $val;
155 my ($self, $mid) = @_;
159 } elsif (my $rv = $self->{mid}) {
162 $self->{mid} = _get_term_val($self, 'Q', qr/\AQ/) ||
167 sub _extract_mid { mid_clean(mid_mime($_[0]->{mime})) }
171 my $tid = $self->{thread};
172 return $tid if defined $tid;
173 $self->{thread} = _get_term_val($self, 'G', qr/\AG/); # *G*roup
176 # XXX: consider removing this, we can phrase match subject
179 my $path = $self->{path};
180 return $path if defined $path;
181 $self->{path} = _get_term_val($self, 'XPATH', qr/\AXPATH/); # path
186 my $type = $self->{type};
187 return $type if defined $type;
188 $self->{type} = _get_term_val($self, 'T', qr/\AT/);