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);
14 my ($class, $mime) = @_;
15 my $doc = Search::Xapian::Document->new;
16 $doc->add_boolean_term('T' . 'mail');
18 bless { type => 'mail', doc => $doc, mime => $mime }, $class;
22 my ($class, $doc, $mid) = @_;
23 bless { doc => $doc, mime => undef, mid => $mid }, $class;
27 my ($class, $head, $db, $mid) = @_;
28 my $doc_id = $head->get_docid;
29 my $doc = $db->get_document($doc_id);
30 load_expand(wrap($class, $doc, $mid))
35 Search::Xapian::sortable_unserialise($doc->get_value($col));
38 sub load_from_data ($$) {
39 my ($self) = $_[0]; # data = $_[1]
40 my ($subj, $from, $refs, $to, $cc, $blob, $mid0) = split(/\n/, $_[1]);
41 $self->{subject} = $subj;
42 $self->{from} = $from;
43 $self->{references} = $refs;
45 # To: and Cc: are stored to optimize HDR/XHDR in NNTP since
46 # some NNTP clients will use that for message displays.
50 $self->{blob} = $blob;
56 my $doc = $self->{doc};
57 my $data = $doc->get_data or return;
58 $self->{ts} = get_val($doc, &PublicInbox::Search::TS);
59 $self->{ds} = get_val($doc, &PublicInbox::Search::DS);
61 load_from_data($self, $data);
66 my ($class, $doc) = @_;
67 my $self = bless { doc => $doc }, $class;
71 # :bytes and :lines metadata in RFC 3977
72 sub bytes ($) { get_val($_[0]->{doc}, &PublicInbox::Search::BYTES) }
73 sub lines ($) { get_val($_[0]->{doc}, &PublicInbox::Search::LINES) }
75 $_[0]->{num} ||= get_val($_[0]->{doc}, PublicInbox::Search::NUM())
79 my ($self, $field) = @_;
80 my $val = $self->{$field};
81 return $val if defined $val;
83 my $mime = $self->{mime} or return;
84 $val = $mime->header($field);
85 $val = '' unless defined $val;
88 $self->{$field} = $val;
91 sub subject ($) { __hdr($_[0], 'subject') }
92 sub to ($) { __hdr($_[0], 'to') }
93 sub cc ($) { __hdr($_[0], 'cc') }
95 # no strftime, that is locale-dependent and not for RFC822
96 my @DoW = qw(Sun Mon Tue Wed Thu Fri Sat);
97 my @MoY = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
101 my $ds = $self->{ds};
102 return unless defined $ds;
103 my ($sec, $min, $hour, $mday, $mon, $year, $wday) = gmtime($ds);
104 "$DoW[$wday], " . sprintf("%02d $MoY[$mon] %04d %02d:%02d:%02d +0000",
105 $mday, $year+1900, $hour, $min, $sec);
111 my $from = __hdr($self, 'from');
112 if (defined $from && !defined $self->{from_name}) {
113 my @n = PublicInbox::Address::names($from);
114 $self->{from_name} = join(', ', @n);
121 my $from_name = $self->{from_name};
122 return $from_name if defined $from_name;
129 $self->{ts} ||= eval { msg_timestamp($self->{mime}->header_obj) } || 0;
134 $self->{ds} ||= eval { msg_datestamp($self->{mime}->header_obj); } || 0;
138 my ($self, $oid, $mid0) = @_;
139 my @rows = ($self->subject, $self->from, $self->references,
140 $self->to, $self->cc);
141 $oid = '' unless defined $oid;
142 join("\n", @rows, $oid, $mid0);
147 my $x = $self->{references};
148 defined $x ? $x : '';
151 sub _get_term_val ($$$) {
152 my ($self, $pfx, $re) = @_;
153 my $doc = $self->{doc};
154 my $end = $doc->termlist_end;
155 my $i = $doc->termlist_begin;
158 my $val = $i->get_termname;
159 $val =~ s/$re// and return $val;
165 my ($self, $mid) = @_;
169 } elsif (my $rv = $self->{mid}) {
171 } elsif ($self->{doc}) {
172 $self->{mid} = _get_term_val($self, 'Q', qr/\AQ/);
174 $self->_extract_mid; # v1 w/o Xapian
178 sub _extract_mid { mid_clean(mid_mime($_[0]->{mime})) }
182 my $tid = $self->{thread};
183 return $tid if defined $tid;
184 $self->{thread} = _get_term_val($self, 'G', qr/\AG/); # *G*roup
187 # XXX: consider removing this, we can phrase match subject
190 my $path = $self->{path};
191 return $path if defined $path;
192 $self->{path} = _get_term_val($self, 'XPATH', qr/\AXPATH/); # path
197 my $type = $self->{type};
198 return $type if defined $type;
199 $self->{type} = _get_term_val($self, 'T', qr/\AT/);