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 # 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; # Received: header in Unix time
12 use constant YYYYMMDD => 1; # Date: header for searching in the WWW UI
13 use constant DT => 2; # Date: YYYYMMDDHHMMSS
15 use PublicInbox::SearchMsg;
16 use PublicInbox::MIME;
17 use PublicInbox::MID qw/id_compress/;
18 use PublicInbox::Over;
22 require Search::Xapian;
23 Search::Xapian->import(qw(:standard));
25 # n.b. FLAG_PURE_NOT is expensive not suitable for a public
26 # website as it could become a denial-of-service vector
27 # FLAG_PHRASE also seems to cause performance problems chert
28 # (and probably earlier Xapian DBs). glass seems fine...
29 # TODO: make this an option, maybe?
30 # or make indexlevel=medium as default
31 FLAG_PHRASE()|FLAG_BOOLEAN()|FLAG_LOVEHATE()|FLAG_WILDCARD();
35 # This is English-only, everything else is non-standard and may be confused as
36 # a prefix common in patch emails
37 our $LANG = 'english';
40 # SCHEMA_VERSION history
42 # 1 - subject_path is lower-cased
43 # 2 - subject_path is id_compress in the index, only
44 # 3 - message-ID is compressed if it includes '%' (hack!)
45 # 4 - change "Re: " normalization, avoid circular Reference ghosts
46 # 5 - subject_path drops trailing '.'
47 # 6 - preserve References: order in document data
48 # 7 - remove references and inreplyto terms
49 # 8 - remove redundant/unneeded document data
50 # 9 - disable Message-ID compression (SHA-1)
51 # 10 - optimize doc for NNTP overviews
52 # 11 - merge threads when vivifying ghosts
53 # 12 - change YYYYMMDD value column to numeric
54 # 13 - fix threading for empty References/In-Reply-To
55 # (commit 83425ef12e4b65cdcecd11ddcb38175d4a91d5a0)
56 # 14 - fix ghost root vivification
60 my %bool_pfx_external = (
61 mid => 'Q', # Message-ID (full/exact), this is mostly uniQue
64 dfblob => 'XDFPRE XDFPOST',
67 my $non_quoted_body = 'XNQ XDFN XDFA XDFB XDFHH XDFCTX XDFPRE XDFPOST';
69 # for mairix compatibility
71 m => 'XM', # 'mid:' (bool) is exact, 'm:' (prob) can do partial
78 b => $non_quoted_body . ' XQUOT',
79 bs => $non_quoted_body . ' XQUOT S',
83 nq => $non_quoted_body,
91 '' => 'XM S A XQUOT XFN ' . $non_quoted_body,
94 # not documenting m: and mid: for now, the using the URLs works w/o Xapian
96 's:' => 'match within Subject e.g. s:"a quick brown fox"',
98 date range as YYYYMMDD e.g. d:19931002..20101002
99 Open-ended ranges such as d:19931002.. and d:..20101002
103 date-time range as YYYYMMDDhhmmss (e.g. dt:19931002011000..19931002011200)
105 'b:' => 'match within message body, including text attachments',
106 'nq:' => 'match non-quoted text within message body',
107 'q:' => 'match quoted text within message body',
108 'n:' => 'match filename of attachment(s)',
109 't:' => 'match within the To header',
110 'c:' => 'match within the Cc header',
111 'f:' => 'match within the From header',
112 'a:' => 'match within the To, Cc, and From headers',
113 'tc:' => 'match within the To and Cc headers',
114 'bs:' => 'match within the Subject and body',
115 'dfn:' => 'match filename from diff',
116 'dfa:' => 'match diff removed (-) lines',
117 'dfb:' => 'match diff added (+) lines',
118 'dfhh:' => 'match diff hunk header context (usually a function name)',
119 'dfctx:' => 'match diff context lines',
120 'dfpre:' => 'match pre-image git blob ID',
121 'dfpost:' => 'match post-image git blob ID',
122 'dfblob:' => 'match either pre or post-image git blob ID',
127 my ($self, $rdonly) = @_;
128 if ($self->{version} == 1) {
129 "$self->{mainrepo}/public-inbox/xapian" . SCHEMA_VERSION;
131 my $dir = "$self->{mainrepo}/xap" . SCHEMA_VERSION;
132 return $dir if $rdonly;
134 my $shard = $self->{shard};
135 defined $shard or die "shard not given";
142 my $dir = xdir($self, 1);
143 my ($xdb, $slow_phrase);
144 my $qpf = \($self->{qp_flags} ||= $QP_FLAGS);
145 if ($self->{version} >= 2) {
146 foreach my $shard (<$dir/*>) {
147 -d $shard && $shard =~ m!/[0-9]+\z! or next;
148 my $sub = Search::Xapian::Database->new($shard);
150 $xdb->add_database($sub);
154 $slow_phrase ||= -f "$shard/iamchert";
157 $slow_phrase = -f "$dir/iamchert";
158 $xdb = Search::Xapian::Database->new($dir);
160 $$qpf |= FLAG_PHRASE() unless $slow_phrase;
166 $self->{xdb} ||= do {
173 my ($class, $ibx) = @_;
174 ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
176 mainrepo => $ibx->{mainrepo},
177 altid => $ibx->{altid},
178 version => $ibx->{version} // 1,
180 my $dir = xdir($self, 1);
181 $self->{over_ro} = PublicInbox::Over->new("$dir/over.sqlite3");
187 if (my $xdb = $self->{xdb}) {
190 $self; # make chaining easier
195 my ($self, $query_string, $opts) = @_;
197 if ($query_string eq '' && !$opts->{mset}) {
198 $self->{over_ro}->recent($opts);
201 my $qp_flags = $self->{qp_flags};
202 my $query = $qp->parse_query($query_string, $qp_flags);
203 $opts->{relevance} = 1 unless exists $opts->{relevance};
204 _do_enquire($self, $query, $opts);
209 my ($self, $cb) = @_;
213 eval { @ret = $cb->() };
214 return @ret unless $@;
217 eval { $ret = $cb->() };
218 return $ret unless $@;
220 # Exception: The revision being read has been discarded -
221 # you should call Xapian::Database::reopen()
222 if (ref($@) eq 'Search::Xapian::DatabaseModifiedError') {
223 warn "reopen try #$i on $@\n";
226 # let caller decide how to spew, because ExtMsg queries
227 # get wonky and trigger:
228 # "something terrible happened at .../Xapian/Enquire.pm"
232 die "Too many Xapian database modifications in progress\n";
236 my ($self, $query, $opts) = @_;
237 retry_reopen($self, sub { _enquire_once($self, $query, $opts) });
241 my ($self, $query, $opts) = @_;
242 my $xdb = xdb($self);
243 my $enquire = Search::Xapian::Enquire->new($xdb);
244 $enquire->set_query($query);
246 my $desc = !$opts->{asc};
247 if (($opts->{mset} || 0) == 2) {
248 $enquire->set_docid_order(Search::Xapian::ENQ_ASCENDING());
249 $enquire->set_weighting_scheme(Search::Xapian::BoolWeight->new);
250 } elsif ($opts->{relevance}) {
251 $enquire->set_sort_by_relevance_then_value(TS, $desc);
253 $enquire->set_sort_by_value_then_relevance(TS, $desc);
255 my $offset = $opts->{offset} || 0;
256 my $limit = $opts->{limit} || 50;
257 my $mset = $enquire->get_mset($offset, $limit);
258 return $mset if $opts->{mset};
260 PublicInbox::SearchMsg->load_doc($_->get_document);
262 return \@msgs unless wantarray;
264 ($mset->get_matches_estimated, \@msgs)
268 sub stemmer { Search::Xapian::Stem->new($LANG) }
274 my $qp = $self->{query_parser};
276 my $xdb = xdb($self);
278 $qp = Search::Xapian::QueryParser->new;
279 $qp->set_default_op(OP_AND());
280 $qp->set_database($xdb);
281 $qp->set_stemmer($self->stemmer);
282 $qp->set_stemming_strategy(STEM_SOME());
283 $qp->set_max_wildcard_expansion(100);
284 $qp->add_valuerangeprocessor(
285 Search::Xapian::NumberValueRangeProcessor->new(YYYYMMDD, 'd:'));
286 $qp->add_valuerangeprocessor(
287 Search::Xapian::NumberValueRangeProcessor->new(DT, 'dt:'));
289 while (my ($name, $prefix) = each %bool_pfx_external) {
290 $qp->add_boolean_prefix($name, $_) foreach split(/ /, $prefix);
293 # we do not actually create AltId objects,
294 # just parse the spec to avoid the extra DB handles for now.
295 if (my $altid = $self->{altid}) {
296 my $user_pfx = $self->{-user_pfx} ||= [];
298 # $_ = 'serial:gmane:/path/to/gmane.msgmap.sqlite3'
299 /\Aserial:(\w+):/ or next;
301 push @$user_pfx, "$pfx:", <<EOF;
302 alternate serial number e.g. $pfx:12345 (boolean)
305 $qp->add_boolean_prefix($pfx, 'X'.uc($pfx));
310 while (my ($name, $prefix) = each %prob_prefix) {
311 $qp->add_prefix($name, $_) foreach split(/ /, $prefix);
314 $self->{query_parser} = $qp;
318 my ($self, $num) = @_;
319 $self->{over_ro}->get_art($num);
324 $self->qp; # parse altids
326 if (my $user_pfx = $self->{-user_pfx}) {
327 push @ret, @$user_pfx;