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;
9 use PublicInbox::MID qw/mid_clean mid_mime/;
10 use PublicInbox::Address;
11 use PublicInbox::MsgTime qw(msg_timestamp msg_datestamp);
12 use Time::Local qw(timegm);
15 my ($class, $mime) = @_;
16 my $doc = Search::Xapian::Document->new;
17 bless { doc => $doc, mime => $mime }, $class;
21 my ($class, $doc, $mid) = @_;
22 bless { doc => $doc, mime => undef, mid => $mid }, $class;
26 my ($class, $head, $db, $mid) = @_;
27 my $doc_id = $head->get_docid;
28 my $doc = $db->get_document($doc_id);
29 load_expand(wrap($class, $doc, $mid))
34 Search::Xapian::sortable_unserialise($doc->get_value($col));
38 my ($self, $oid, $mid0) = @_;
39 $oid = '' unless defined $oid;
54 sub load_from_data ($$) {
55 my ($self) = $_[0]; # data = $_[1]
61 # To: and Cc: are stored to optimize HDR/XHDR in NNTP since
62 # some NNTP clients will use that for message displays.
70 ) = split(/\n/, $_[1]);
75 my $doc = $self->{doc};
76 my $data = $doc->get_data or return;
77 $self->{ts} = get_val($doc, PublicInbox::Search::TS());
78 my $dt = get_val($doc, PublicInbox::Search::DT());
79 my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $dt);
80 $self->{ds} = timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
82 load_from_data($self, $data);
87 my ($class, $doc) = @_;
88 my $self = bless { doc => $doc }, $class;
92 # :bytes and :lines metadata in RFC 3977
93 sub bytes ($) { $_[0]->{bytes} }
94 sub lines ($) { $_[0]->{lines} }
95 sub num ($) { $_[0]->{num} ||= _get_term_val($_[0], 'XNUM', qr/\AXNUM/) }
98 my ($self, $field) = @_;
99 my $val = $self->{$field};
100 return $val if defined $val;
102 my $mime = $self->{mime} or return;
103 $val = $mime->header($field);
104 $val = '' unless defined $val;
107 $self->{$field} = $val;
110 sub subject ($) { __hdr($_[0], 'subject') }
111 sub to ($) { __hdr($_[0], 'to') }
112 sub cc ($) { __hdr($_[0], 'cc') }
114 # no strftime, that is locale-dependent and not for RFC822
115 my @DoW = qw(Sun Mon Tue Wed Thu Fri Sat);
116 my @MoY = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
120 my $ds = $self->{ds};
121 return unless defined $ds;
122 my ($sec, $min, $hour, $mday, $mon, $year, $wday) = gmtime($ds);
123 "$DoW[$wday], " . sprintf("%02d $MoY[$mon] %04d %02d:%02d:%02d +0000",
124 $mday, $year+1900, $hour, $min, $sec);
130 my $from = __hdr($self, 'from');
131 if (defined $from && !defined $self->{from_name}) {
132 my @n = PublicInbox::Address::names($from);
133 $self->{from_name} = join(', ', @n);
140 my $from_name = $self->{from_name};
141 return $from_name if defined $from_name;
148 $self->{ts} ||= eval { msg_timestamp($self->{mime}->header_obj) } || 0;
153 $self->{ds} ||= eval { msg_datestamp($self->{mime}->header_obj); } || 0;
158 my $x = $self->{references};
159 defined $x ? $x : '';
162 sub _get_term_val ($$$) {
163 my ($self, $pfx, $re) = @_;
164 my $doc = $self->{doc};
165 my $end = $doc->termlist_end;
166 my $i = $doc->termlist_begin;
169 my $val = $i->get_termname;
170 $val =~ s/$re// and return $val;
176 my ($self, $mid) = @_;
180 } elsif (my $rv = $self->{mid}) {
182 } elsif ($self->{doc}) {
183 $self->{mid} = _get_term_val($self, 'Q', qr/\AQ/);
185 $self->_extract_mid; # v1 w/o Xapian
189 sub _extract_mid { mid_clean(mid_mime($_[0]->{mime})) }