1 # Copyright (C) 2015 all contributors <meta@public-inbox.org>
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
3 # based on notmuch, but with no concept of folders, files or flags
5 # Read-only search interface for use by the web and NNTP interfaces
6 package PublicInbox::Search;
10 # values for searching
11 use constant TS => 0; # timestamp
12 use constant NUM => 1; # NNTP article number
13 use constant BYTES => 2; # :bytes as defined in RFC 3977
14 use constant LINES => 3; # :lines as defined in RFC 3977
15 use constant YYYYMMDD => 4; # for searching in the WWW UI
17 use Search::Xapian qw/:standard/;
18 use PublicInbox::SearchMsg;
19 use PublicInbox::MIME;
20 use PublicInbox::MID qw/mid_clean id_compress/;
22 # This is English-only, everything else is non-standard and may be confused as
23 # a prefix common in patch emails
24 our $REPLY_RE = qr/^re:\s+/i;
25 our $LANG = 'english';
28 # SCHEMA_VERSION history
30 # 1 - subject_path is lower-cased
31 # 2 - subject_path is id_compress in the index, only
32 # 3 - message-ID is compressed if it includes '%' (hack!)
33 # 4 - change "Re: " normalization, avoid circular Reference ghosts
34 # 5 - subject_path drops trailing '.'
35 # 6 - preserve References: order in document data
36 # 7 - remove references and inreplyto terms
37 # 8 - remove redundant/unneeded document data
38 # 9 - disable Message-ID compression (SHA-1)
39 # 10 - optimize doc for NNTP overviews
40 # 11 - merge threads when vivifying ghosts
41 # 12 - change YYYYMMDD value column to numeric
42 # 13 - fix threading for empty References/In-Reply-To
43 # (commit 83425ef12e4b65cdcecd11ddcb38175d4a91d5a0)
44 # 14 - fix ghost root vivification
47 # n.b. FLAG_PURE_NOT is expensive not suitable for a public website
48 # as it could become a denial-of-service vector
49 QP_FLAGS => FLAG_PHRASE|FLAG_BOOLEAN|FLAG_LOVEHATE|FLAG_WILDCARD,
53 my %bool_pfx_internal = (
54 type => 'T', # "mail" or "ghost"
55 thread => 'G', # newsGroup (or similar entity - e.g. a web forum name)
58 my %bool_pfx_external = (
59 mid => 'Q', # uniQue id (Message-ID)
63 # for mairix compatibility
65 m => 'XMID', # 'mid:' (bool) is exact, 'm:' (prob) can do partial
80 '' => 'XMID S A XNQ XQUOT XFN',
83 # not documenting m: and mid: for now, the using the URLs works w/o Xapian
85 's:' => 'match within Subject e.g. s:"a quick brown fox"',
87 date range as YYYYMMDD e.g. d:19931002..20101002
88 Open-ended ranges such as d:19931002.. and d:..20101002
91 'b:' => 'match within message body, including text attachments',
92 'nq:' => 'match non-quoted text within message body',
93 'q:' => 'match quoted text within message body',
94 'n:' => 'match filename of attachment(s)',
95 't:' => 'match within the To header',
96 'c:' => 'match within the Cc header',
97 'f:' => 'match within the From header',
98 'a:' => 'match within the To, Cc, and From headers',
99 'tc:' => 'match within the To and Cc headers',
100 'bs:' => 'match within the Subject and body',
104 # df (filenames from diff)
105 # da (diff a/ removed lines)
106 # db (diff b/ added lines)
108 my $mail_query = Search::Xapian::Query->new('T' . 'mail');
111 my (undef, $git_dir) = @_;
112 "$git_dir/public-inbox/xapian" . SCHEMA_VERSION;
116 my ($class, $git_dir, $altid) = @_;
117 my $dir = $class->xdir($git_dir);
118 my $db = Search::Xapian::Database->new($dir);
119 bless { xdb => $db, git_dir => $git_dir, altid => $altid }, $class;
122 sub reopen { $_[0]->{xdb}->reopen }
126 my ($self, $query_string, $opts) = @_;
130 unless ($query_string eq '') {
131 $query = $self->qp->parse_query($query_string, QP_FLAGS);
132 $opts->{relevance} = 1 unless exists $opts->{relevance};
135 _do_enquire($self, $query, $opts);
139 my ($self, $mid, $opts) = @_;
140 my $smsg = eval { $self->lookup_message($mid) };
142 return { total => 0, msgs => [] } unless $smsg;
143 my $qtid = Search::Xapian::Query->new('G' . $smsg->thread_id);
144 my $path = $smsg->path;
145 if (defined $path && $path ne '') {
146 my $path = id_compress($smsg->path);
147 my $qsub = Search::Xapian::Query->new('XPATH' . $path);
148 $qtid = Search::Xapian::Query->new(OP_OR, $qtid, $qsub);
151 $opts->{limit} ||= 1000;
153 # always sort threads by timestamp, this makes life easier
154 # for the threading algorithm (in SearchThread.pm)
157 _do_enquire($self, $qtid, $opts);
161 my ($self, $cb) = @_;
164 eval { $ret = $cb->() };
165 return $ret unless $@;
166 # Exception: The revision being read has been discarded -
167 # you should call Xapian::Database::reopen()
168 if (ref($@) eq 'Search::Xapian::DatabaseModifiedError') {
177 my ($self, $query, $opts) = @_;
178 retry_reopen($self, sub { _enquire_once($self, $query, $opts) });
182 my ($self, $query, $opts) = @_;
183 my $enquire = $self->enquire;
184 if (defined $query) {
185 $query = Search::Xapian::Query->new(OP_AND,$query,$mail_query);
187 $query = $mail_query;
189 $enquire->set_query($query);
191 my $desc = !$opts->{asc};
192 if ($opts->{relevance}) {
193 $enquire->set_sort_by_relevance_then_value(TS, $desc);
194 } elsif ($opts->{num}) {
195 $enquire->set_sort_by_value(NUM, 0);
197 $enquire->set_sort_by_value_then_relevance(TS, $desc);
199 my $offset = $opts->{offset} || 0;
200 my $limit = $opts->{limit} || 50;
201 my $mset = $enquire->get_mset($offset, $limit);
202 return $mset if $opts->{mset};
204 PublicInbox::SearchMsg->load_doc($_->get_document);
207 { total => $mset->get_matches_estimated, msgs => \@msgs }
211 sub stemmer { Search::Xapian::Stem->new($LANG) }
217 my $qp = $self->{query_parser};
221 $qp = Search::Xapian::QueryParser->new;
222 $qp->set_default_op(OP_AND);
223 $qp->set_database($self->{xdb});
224 $qp->set_stemmer($self->stemmer);
225 $qp->set_stemming_strategy(STEM_SOME);
226 $qp->add_valuerangeprocessor(
227 Search::Xapian::NumberValueRangeProcessor->new(YYYYMMDD, 'd:'));
229 while (my ($name, $prefix) = each %bool_pfx_external) {
230 $qp->add_boolean_prefix($name, $prefix);
233 # we do not actually create AltId objects,
234 # just parse the spec to avoid the extra DB handles for now.
235 if (my $altid = $self->{altid}) {
236 my $user_pfx = $self->{-user_pfx} ||= [];
238 # $_ = 'serial:gmane:/path/to/gmane.msgmap.sqlite3'
239 /\Aserial:(\w+):/ or next;
241 push @$user_pfx, "$pfx:", <<EOF;
242 alternate serial number e.g. $pfx:12345 (boolean)
245 $qp->add_boolean_prefix($pfx, 'X'.uc($pfx));
250 while (my ($name, $prefix) = each %prob_prefix) {
251 $qp->add_prefix($name, $_) foreach split(/ /, $prefix);
254 $self->{query_parser} = $qp;
257 sub num_range_processor {
258 $_[0]->{nrp} ||= Search::Xapian::NumberValueRangeProcessor->new(NUM);
261 # only used for NNTP server
263 my ($self, $beg, $end, $offset) = @_;
264 my $qp = Search::Xapian::QueryParser->new;
265 $qp->set_database($self->{xdb});
266 $qp->add_valuerangeprocessor($self->num_range_processor);
267 my $query = $qp->parse_query("$beg..$end", QP_FLAGS);
269 _do_enquire($self, $query, {num => 1, limit => 200, offset => $offset});
273 my ($self, $mid) = @_;
274 $mid = mid_clean($mid);
276 my $doc_id = $self->find_unique_doc_id('Q' . $mid);
278 if (defined $doc_id) {
280 my $doc = $self->{xdb}->get_document($doc_id);
281 $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
282 $smsg->{doc_id} = $doc_id;
287 sub lookup_mail { # no ghosts!
288 my ($self, $mid) = @_;
289 retry_reopen($self, sub {
290 my $smsg = lookup_message($self, $mid) or return;
291 PublicInbox::SearchMsg->load_doc($smsg->{doc});
295 sub find_unique_doc_id {
296 my ($self, $termval) = @_;
298 my ($begin, $end) = $self->find_doc_ids($termval);
300 return undef if $begin->equal($end); # not found
302 my $rv = $begin->get_docid;
306 $begin->equal($end) or die "Term '$termval' is not unique\n";
310 # returns begin and end PostingIterator
312 my ($self, $termval) = @_;
313 my $db = $self->{xdb};
315 ($db->postlist_begin($termval), $db->postlist_end($termval));
318 # normalize subjects so they are suitable as pathnames for URLs
319 # XXX: consider for removal
322 $subj = subject_normalized($subj);
323 $subj =~ s![^a-zA-Z0-9_\.~/\-]+!_!g;
327 sub subject_normalized {
329 $subj =~ s/\A\s+//s; # no leading space
330 $subj =~ s/\s+\z//s; # no trailing space
331 $subj =~ s/\s+/ /gs; # no redundant spaces
332 $subj =~ s/\.+\z//; # no trailing '.'
333 $subj =~ s/$REPLY_RE//igo; # remove reply prefix
339 $self->{enquire} ||= Search::Xapian::Enquire->new($self->{xdb});
344 $self->qp; # parse altids
346 if (my $user_pfx = $self->{-user_pfx}) {
347 push @ret, @$user_pfx;