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;
15 my ($class, $mime) = @_;
16 my $doc = Search::Xapian::Document->new;
17 $doc->add_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));
34 my $doc = $self->{doc};
35 my $data = $doc->get_data or return;
36 $self->{ts} = get_val($doc, &PublicInbox::Search::TS);
38 my ($subj, $from, $refs, $to, $cc, $blob) = split(/\n/, $data);
39 $self->{subject} = $subj;
40 $self->{from} = $from;
41 $self->{references} = $refs;
44 $self->{blob} = $blob;
49 my ($class, $doc) = @_;
50 my $data = $doc->get_data or return;
51 my $ts = get_val($doc, &PublicInbox::Search::TS);
53 my ($subj, $from, $refs, $to, $cc, $blob) = split(/\n/, $data);
66 # :bytes and :lines metadata in RFC 3977
67 sub bytes ($) { get_val($_[0]->{doc}, &PublicInbox::Search::BYTES) }
68 sub lines ($) { get_val($_[0]->{doc}, &PublicInbox::Search::LINES) }
69 sub num ($) { 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 $self->{ts} ||= eval { str2time($self->{mime}->header('Date')) } || 0;
126 my ($self, $blob) = @_;
127 my @rows = ($self->subject, $self->from, $self->references,
128 $self->to, $self->cc);
129 push @rows, $blob if defined $blob;
135 my $x = $self->{references};
136 defined $x ? $x : '';
139 sub _get_term_val ($$$) {
140 my ($self, $pfx, $re) = @_;
141 my $doc = $self->{doc};
142 my $end = $doc->termlist_end;
143 my $i = $doc->termlist_begin;
146 my $val = $i->get_termname;
147 $val =~ s/$re// and return $val;
153 my ($self, $mid) = @_;
157 } elsif (my $rv = $self->{mid}) {
160 $self->{mid} = _get_term_val($self, 'Q', qr/\AQ/) ||
165 sub _extract_mid { mid_clean(mid_mime($_[0]->{mime})) }
169 my $tid = $self->{thread};
170 return $tid if defined $tid;
171 $self->{thread} = _get_term_val($self, 'G', qr/\AG/); # *G*roup
174 # XXX: consider removing this, we can phrase match subject
177 my $path = $self->{path};
178 return $path if defined $path;
179 $self->{path} = _get_term_val($self, 'XPATH', qr/\AXPATH/); # path