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 Date::Parse qw/str2time/;
11 use PublicInbox::MID qw/mid_clean/;
12 use PublicInbox::Address;
13 our $PFX2TERM_RE = undef;
16 my ($class, $mime) = @_;
17 my $doc = Search::Xapian::Document->new;
18 $doc->add_term(PublicInbox::Search::xpfx('type') . 'mail');
20 bless { type => 'mail', doc => $doc, mime => $mime }, $class;
24 my ($class, $doc, $mid) = @_;
25 bless { doc => $doc, mime => undef, mid => $mid }, $class;
30 Search::Xapian::sortable_unserialise($doc->get_value($col));
34 my ($class, $doc) = @_;
35 my $data = $doc->get_data or return;
36 my $ts = get_val($doc, &PublicInbox::Search::TS);
38 my ($subj, $from, $refs, $to, $cc, $blob) = split(/\n/, $data);
51 # :bytes and :lines metadata in RFC 3977
52 sub bytes ($) { get_val($_[0]->{doc}, &PublicInbox::Search::BYTES) }
53 sub lines ($) { get_val($_[0]->{doc}, &PublicInbox::Search::LINES) }
54 sub num ($) { get_val($_[0]->{doc}, &PublicInbox::Search::NUM) }
57 my ($self, $field) = @_;
58 my $val = $self->{$field};
59 return $val if defined $val;
61 my $mime = $self->{mime} or return;
62 $val = $mime->header($field);
63 $val = '' unless defined $val;
66 $self->{$field} = $val;
69 sub subject ($) { __hdr($_[0], 'subject') }
70 sub to ($) { __hdr($_[0], 'to') }
71 sub cc ($) { __hdr($_[0], 'cc') }
73 # no strftime, that is locale-dependent
74 my @DoW = qw(Sun Mon Tue Wed Thu Fri Sat);
75 my @MoY = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
79 my $date = __hdr($self, 'date');
80 return $date if defined $date;
82 return unless defined $ts;
83 my ($sec, $min, $hour, $mday, $mon, $year, $wday) = gmtime($ts);
84 $self->{date} = "$DoW[$wday], ".
85 sprintf("%02d $MoY[$mon] %04d %02d:%02d:%02d +0000",
86 $mday, $year+1900, $hour, $min, $sec);
92 my $from = __hdr($self, 'from');
93 if (defined $from && !defined $self->{from_name}) {
94 my @n = PublicInbox::Address::names($from);
95 $self->{from_name} = join(', ', @n);
102 my $from_name = $self->{from_name};
103 return $from_name if defined $from_name;
110 $self->{ts} ||= eval { str2time($self->mime->header('Date')) } || 0;
114 my ($self, $blob) = @_;
115 my @rows = ($self->subject, $self->from, $self->references,
116 $self->to, $self->cc);
117 push @rows, $blob if defined $blob;
123 my $x = $self->{references};
124 defined $x ? $x : '';
127 sub ensure_metadata {
129 my $doc = $self->{doc};
130 my $end = $doc->termlist_end;
132 unless (defined $PFX2TERM_RE) {
133 my $or = join('|', keys %PublicInbox::Search::PFX2TERM_RMAP);
134 $PFX2TERM_RE = qr/\A($or)/;
137 while (my ($pfx, $field) = each %PublicInbox::Search::PFX2TERM_RMAP) {
138 # ideally we'd move this out of the loop:
139 my $i = $doc->termlist_begin;
143 my $val = $i->get_termname;
145 if ($val =~ s/$PFX2TERM_RE//o) {
146 $self->{$field} = $val;
153 my ($self, $mid) = @_;
157 } elsif (my $rv = $self->{mid}) {
160 $self->ensure_metadata; # needed for ghosts
161 $self->{mid} ||= $self->_extract_mid;
165 sub _extract_mid { mid_clean(mid_mime($_[0]->mime)) }
168 my ($self, $x40) = @_;
170 $self->{blob} = $x40;
177 my ($self, $mime) = @_;
179 $self->{mime} = $mime;
187 my ($self, $doc_id) = @_;
188 if (defined $doc_id) {
189 $self->{doc_id} = $doc_id;
191 # TODO load from xapian
198 my $tid = $self->{thread};
199 return $tid if defined $tid;
200 $self->ensure_metadata;
206 my $path = $self->{path};
207 return $path if defined $path;
208 $self->ensure_metadata;