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;
11 use Date::Parse qw/str2time/;
12 use PublicInbox::MID qw/mid_clean/;
13 use PublicInbox::Address;
14 our $PFX2TERM_RE = undef;
15 use constant EPOCH_822 => 'Thu, 01 Jan 1970 00:00:00 +0000';
16 use POSIX qw(strftime);
19 my ($class, $mime) = @_;
20 my $doc = Search::Xapian::Document->new;
21 $doc->add_term(PublicInbox::Search::xpfx('type') . 'mail');
23 bless { type => 'mail', doc => $doc, mime => $mime }, $class;
27 my ($class, $doc, $mid) = @_;
28 bless { doc => $doc, mime => undef, mid => $mid }, $class;
33 Search::Xapian::sortable_unserialise($doc->get_value($col));
37 my ($class, $doc) = @_;
38 my $data = $doc->get_data or return;
39 my $ts = get_val($doc, &PublicInbox::Search::TS);
41 my ($subj, $from, $refs, $to, $cc) = split(/\n/, $data);
53 # :bytes and :lines metadata in RFC 3977
54 sub bytes ($) { get_val($_[0]->{doc}, &PublicInbox::Search::BYTES) }
55 sub lines ($) { get_val($_[0]->{doc}, &PublicInbox::Search::LINES) }
56 sub num ($) { get_val($_[0]->{doc}, &PublicInbox::Search::NUM) }
59 my ($self, $field) = @_;
60 my $val = $self->{$field};
61 return $val if defined $val;
63 my $mime = $self->{mime} or return;
64 $val = $mime->header($field);
65 $val = '' unless defined $val;
68 $self->{$field} = $val;
71 sub subject ($) { __hdr($_[0], 'subject') }
72 sub to ($) { __hdr($_[0], 'to') }
73 sub cc ($) { __hdr($_[0], 'cc') }
77 my $date = __hdr($self, 'date');
78 return $date if defined $date;
80 return unless defined $ts;
81 $self->{date} = strftime('%a, %d %b %Y %T +0000', gmtime($ts));
86 my $from = __hdr($self, 'from');
87 if (defined $from && !defined $self->{from_name}) {
88 my @n = PublicInbox::Address::names($from);
89 $self->{from_name} = join(', ', @n);
96 my $from_name = $self->{from_name};
97 return $from_name if defined $from_name;
104 $self->{ts} ||= eval { str2time($self->mime->header('Date')) } || 0;
109 join("\n", $self->subject, $self->from, $self->references,
110 $self->to, $self->cc);
115 my $x = $self->{references};
116 defined $x ? $x : '';
119 sub ensure_metadata {
121 my $doc = $self->{doc};
122 my $end = $doc->termlist_end;
124 unless (defined $PFX2TERM_RE) {
125 my $or = join('|', keys %PublicInbox::Search::PFX2TERM_RMAP);
126 $PFX2TERM_RE = qr/\A($or)/;
129 while (my ($pfx, $field) = each %PublicInbox::Search::PFX2TERM_RMAP) {
130 # ideally we'd move this out of the loop:
131 my $i = $doc->termlist_begin;
135 my $val = $i->get_termname;
137 if ($val =~ s/$PFX2TERM_RE//o) {
138 $self->{$field} = $val;
147 $self->ensure_metadata;
149 'Subject' => $self->subject,
150 'X-PI-From' => $self->from_name,
151 # prevent Email::Simple::Creator from running,
152 # this header is useless for threading as we use X-PI-TS
153 # for sorting and display:
155 'Message-ID' => "<$self->{mid}>",
156 'X-PI-TS' => $self->ts,
158 if (my $refs = $self->{references}) {
159 push @h, References => $refs;
161 my $mime = Email::MIME->create(header => \@h);
162 my $h = $mime->header_obj;
164 # set these headers manually since Encode::encode('MIME-Q', ...)
165 # will add spaces to long values when using header_str above.
167 # drop useless headers Email::MIME set for us
168 $h->header_set('Date');
169 $h->header_set('MIME-Version');
174 my ($self, $mid) = @_;
178 } elsif (my $rv = $self->{mid}) {
181 $self->ensure_metadata; # needed for ghosts
182 $self->{mid} ||= $self->_extract_mid;
186 sub _extract_mid { mid_clean(mid_mime($_[0]->mime)) }
189 my ($self, $mime) = @_;
191 $self->{mime} = $mime;
199 my ($self, $doc_id) = @_;
200 if (defined $doc_id) {
201 $self->{doc_id} = $doc_id;
203 # TODO load from xapian
210 my $tid = $self->{thread};
211 return $tid if defined $tid;
212 $self->ensure_metadata;
218 my $path = $self->{path};
219 return $path if defined $path;
220 $self->ensure_metadata;