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 Date::Parse qw/str2time/;
11 use PublicInbox::MID qw/mid_clean mid_mime/;
12 use PublicInbox::Address;
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) = split(/\n/, $_[1]);
35 $self->{subject} = $subj;
36 $self->{from} = $from;
37 $self->{references} = $refs;
40 $self->{blob} = $blob;
45 my $doc = $self->{doc};
46 my $data = $doc->get_data or return;
47 $self->{ts} = get_val($doc, &PublicInbox::Search::TS);
49 load_from_data($self, $data);
54 my ($class, $doc) = @_;
55 my $data = $doc->get_data or return;
56 my $ts = get_val($doc, &PublicInbox::Search::TS);
58 my $self = bless { doc => $doc, ts => $ts }, $class;
59 load_from_data($self, $data);
63 # :bytes and :lines metadata in RFC 3977
64 sub bytes ($) { get_val($_[0]->{doc}, &PublicInbox::Search::BYTES) }
65 sub lines ($) { get_val($_[0]->{doc}, &PublicInbox::Search::LINES) }
66 sub num ($) { get_val($_[0]->{doc}, &PublicInbox::Search::NUM) }
69 my ($self, $field) = @_;
70 my $val = $self->{$field};
71 return $val if defined $val;
73 my $mime = $self->{mime} or return;
74 $val = $mime->header($field);
75 $val = '' unless defined $val;
78 $self->{$field} = $val;
81 sub subject ($) { __hdr($_[0], 'subject') }
82 sub to ($) { __hdr($_[0], 'to') }
83 sub cc ($) { __hdr($_[0], 'cc') }
85 # no strftime, that is locale-dependent and not for RFC822
86 my @DoW = qw(Sun Mon Tue Wed Thu Fri Sat);
87 my @MoY = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
92 return unless defined $ts;
93 my ($sec, $min, $hour, $mday, $mon, $year, $wday) = gmtime($ts);
94 "$DoW[$wday], " . sprintf("%02d $MoY[$mon] %04d %02d:%02d:%02d +0000",
95 $mday, $year+1900, $hour, $min, $sec);
101 my $from = __hdr($self, 'from');
102 if (defined $from && !defined $self->{from_name}) {
103 my @n = PublicInbox::Address::names($from);
104 $self->{from_name} = join(', ', @n);
111 my $from_name = $self->{from_name};
112 return $from_name if defined $from_name;
119 $self->{ts} ||= eval { str2time($self->{mime}->header('Date')) } || 0;
123 my ($self, $blob) = @_;
124 my @rows = ($self->subject, $self->from, $self->references,
125 $self->to, $self->cc);
126 push @rows, $blob if defined $blob;
132 my $x = $self->{references};
133 defined $x ? $x : '';
136 sub _get_term_val ($$$) {
137 my ($self, $pfx, $re) = @_;
138 my $doc = $self->{doc};
139 my $end = $doc->termlist_end;
140 my $i = $doc->termlist_begin;
143 my $val = $i->get_termname;
144 $val =~ s/$re// and return $val;
150 my ($self, $mid) = @_;
154 } elsif (my $rv = $self->{mid}) {
157 $self->{mid} = _get_term_val($self, 'Q', qr/\AQ/) ||
162 sub _extract_mid { mid_clean(mid_mime($_[0]->{mime})) }
166 my $tid = $self->{thread};
167 return $tid if defined $tid;
168 $self->{thread} = _get_term_val($self, 'G', qr/\AG/); # *G*roup
171 # XXX: consider removing this, we can phrase match subject
174 my $path = $self->{path};
175 return $path if defined $path;
176 $self->{path} = _get_term_val($self, 'XPATH', qr/\AXPATH/); # path
181 my $type = $self->{type};
182 return $type if defined $type;
183 $self->{type} = _get_term_val($self, 'T', qr/\AT/);