1 # Copyright (C) 2015-2021 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;
8 use parent qw(Exporter);
9 our @EXPORT_OK = qw(retry_reopen int_val get_pct xap_terms);
10 use List::Util qw(max);
11 use POSIX qw(strftime);
13 # values for searching, changing the numeric value breaks
14 # compatibility with old indices (so don't change them it)
16 TS => 0, # Received: in Unix time (IMAP INTERNALDATE, JMAP receivedAt)
17 YYYYMMDD => 1, # Date: header for searching in the WWW UI
18 DT => 2, # Date: YYYYMMDDHHMMSS (IMAP SENT*, JMAP sentAt)
20 # added for public-inbox 1.6.0+
21 BYTES => 3, # IMAP RFC822.SIZE
22 UID => 4, # IMAP UID == NNTP article number == Xapian docid
23 THREADID => 5, # RFC 8474, RFC 8621
26 # REPLYCNT => ?, # IMAP ANSWERED
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
45 # 15 - see public-inbox-v2-format(5)
46 # further bumps likely unnecessary, we'll suggest in-place
47 # "--reindex" use for further fixes and tweaks:
49 # public-inbox v1.5.0 adds (still SCHEMA_VERSION=15):
50 # * "lid:" and "l:" for List-Id searches
52 # v1.6.0 adds BYTES, UID and THREADID values
56 use PublicInbox::Smsg;
57 use PublicInbox::Over;
59 our %X = map { $_ => 0 } qw(BoolWeight Database Enquire QueryParser Stem Query);
60 our $Xap; # 'Search::Xapian' or 'Xapian'
61 our $NVRP; # '$Xap::'.('NumberValueRangeProcessor' or 'NumberRangeProcessor')
63 # ENQ_DESCENDING and ENQ_ASCENDING weren't in SWIG Xapian.pm prior to 1.4.16,
64 # let's hope the ABI is stable
65 our $ENQ_DESCENDING = 0;
66 our $ENQ_ASCENDING = 1;
69 return 1 if defined $Xap;
70 # n.b. PI_XAPIAN is intended for development use only. We still
71 # favor Search::Xapian since that's what's available in current
72 # Debian stable (10.x) and derived distros.
73 for my $x (($ENV{PI_XAPIAN} // 'Search::Xapian'), 'Xapian') {
77 $x->import(qw(:standard));
80 # `version_string' was added in Xapian 1.1
81 my $xver = eval('v'.eval($x.'::version_string()')) //
82 eval('v'.eval($x.'::xapian_version_string()'));
84 # NumberRangeProcessor was added in Xapian 1.3.6,
85 # NumberValueRangeProcessor was removed for 1.5.0+,
86 # favor the older /Value/ variant since that's what our
87 # (currently) preferred Search::Xapian supports
88 $NVRP = $x.'::'.($x eq 'Xapian' && $xver ge v1.5 ?
89 'NumberRangeProcessor' : 'NumberValueRangeProcessor');
90 $X{$_} = $Xap.'::'.$_ for (keys %X);
92 *sortable_serialise = $x.'::sortable_serialise';
93 *sortable_unserialise = $x.'::sortable_unserialise';
94 # n.b. FLAG_PURE_NOT is expensive not suitable for a public
95 # website as it could become a denial-of-service vector
96 # FLAG_PHRASE also seems to cause performance problems chert
97 # (and probably earlier Xapian DBs). glass seems fine...
98 # TODO: make this an option, maybe?
99 # or make indexlevel=medium as default
100 $QP_FLAGS = FLAG_PHRASE() | FLAG_BOOLEAN() | FLAG_LOVEHATE() |
107 # This is English-only, everything else is non-standard and may be confused as
108 # a prefix common in patch emails
109 our $LANG = 'english';
111 # note: the non-X term prefix allocations are shared with
112 # Xapian omega, see xapian-applications/omega/docs/termprefixes.rst
113 my %bool_pfx_external = (
114 mid => 'Q', # Message-ID (full/exact), this is mostly uniQue
115 lid => 'G', # newsGroup (or similar entity), just inside <>
118 dfblob => 'XDFPRE XDFPOST',
121 my $non_quoted_body = 'XNQ XDFN XDFA XDFB XDFHH XDFCTX XDFPRE XDFPOST';
123 # for mairix compatibility
125 m => 'XM', # 'mid:' (bool) is exact, 'm:' (prob) can do partial
126 l => 'XL', # 'lid:' (bool) is exact, 'l:' (prob) can do partial
133 b => $non_quoted_body . ' XQUOT',
134 bs => $non_quoted_body . ' XQUOT S',
138 nq => $non_quoted_body,
146 '' => 'XM S A XQUOT XFN ' . $non_quoted_body,
149 # not documenting m: and mid: for now, the using the URLs works w/o Xapian
150 # not documenting lid: for now, either, it is probably redundant with l:,
151 # especially since we don't offer boolean searches for To/Cc/From
154 's:' => 'match within Subject e.g. s:"a quick brown fox"',
156 date range as YYYYMMDD e.g. d:19931002..20101002
157 Open-ended ranges such as d:19931002.. and d:..20101002
161 date-time range as YYYYMMDDhhmmss (e.g. dt:19931002011000..19931002011200)
163 'b:' => 'match within message body, including text attachments',
164 'nq:' => 'match non-quoted text within message body',
165 'q:' => 'match quoted text within message body',
166 'n:' => 'match filename of attachment(s)',
167 't:' => 'match within the To header',
168 'c:' => 'match within the Cc header',
169 'f:' => 'match within the From header',
170 'a:' => 'match within the To, Cc, and From headers',
171 'tc:' => 'match within the To and Cc headers',
172 'l:' => 'match contents of the List-Id header',
173 'bs:' => 'match within the Subject and body',
174 'dfn:' => 'match filename from diff',
175 'dfa:' => 'match diff removed (-) lines',
176 'dfb:' => 'match diff added (+) lines',
177 'dfhh:' => 'match diff hunk header context (usually a function name)',
178 'dfctx:' => 'match diff context lines',
179 'dfpre:' => 'match pre-image git blob ID',
180 'dfpost:' => 'match post-image git blob ID',
181 'dfblob:' => 'match either pre or post-image git blob ID',
186 my ($self, $rdonly) = @_;
187 if ($rdonly || !defined($self->{shard})) {
190 "$self->{xpfx}/$self->{shard}";
194 # returns all shards as separate Xapian::Database objects w/o combining
195 sub xdb_shards_flat ($) {
197 my $xpfx = $self->{xpfx};
198 my (@xdb, $slow_phrase);
200 $self->{qp_flags} //= $QP_FLAGS;
201 if ($xpfx =~ m/xapian${\SCHEMA_VERSION}\z/) {
202 @xdb = ($X{Database}->new($xpfx));
203 $self->{qp_flags} |= FLAG_PHRASE() if !-f "$xpfx/iamchert";
205 opendir(my $dh, $xpfx) or return (); # not initialized yet
206 # We need numeric sorting so shard[0] is first for reading
207 # Xapian metadata, if needed
208 my $last = max(grep(/\A[0-9]+\z/, readdir($dh))) // return ();
210 my $shard_dir = "$self->{xpfx}/$_";
211 push @xdb, $X{Database}->new($shard_dir);
212 $slow_phrase ||= -f "$shard_dir/iamchert";
214 $self->{qp_flags} |= FLAG_PHRASE() if !$slow_phrase;
219 # v2 Xapian docids don't conflict, so they're identical to
220 # NNTP article numbers and IMAP UIDs.
221 # https://trac.xapian.org/wiki/FAQ/MultiDatabaseDocumentID
223 my ($nshard, $mitem) = @_;
224 my $docid = $mitem->get_docid;
225 int(($docid - 1) / $nshard) + 1;
228 sub mset_to_artnums {
229 my ($self, $mset) = @_;
230 my $nshard = $self->{nshard};
231 [ map { mdocid($nshard, $_) } $mset->items ];
236 $self->{xdb} //= do {
237 my @xdb = $self->xdb_shards_flat or return;
238 $self->{nshard} = scalar(@xdb);
239 my $xdb = shift @xdb;
240 $xdb->add_database($_) for @xdb;
246 my ($class, $ibx) = @_;
247 ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
248 my $xap = $ibx->version > 1 ? 'xap' : 'public-inbox/xapian';
250 xpfx => "$ibx->{inboxdir}/$xap" . SCHEMA_VERSION,
251 altid => $ibx->{altid},
257 if (my $xdb = $self->{xdb}) {
260 $self; # make chaining easier
263 # Convert git "approxidate" ranges to something usable with our
264 # Xapian indices. At the moment, Xapian only offers a C++-only API
265 # and neither the SWIG nor XS bindings allow us to use custom code
266 # to parse dates (and libgit2 doesn't expose git__date_parse, either,
267 # so we're running git-rev-parse(1)).
268 # This replaces things we need to send to $git->date_parse with
269 # "\0".$strftime_format.['+'|$idx]."\0" placeholders
270 sub date_parse_prepare {
271 my ($to_parse, $pfx, $range) = @_;
272 # are we inside a parenthesized statement?
273 my $end = $range =~ s/([\)\s]*)\z// ? $1 : '';
274 my @r = split(/\.\./, $range, 2);
276 # expand "d:20101002" => "d:20101002..20101003" and like
277 # n.b. git doesn't do YYYYMMDD w/o '-', it needs YYYY-MM-DD
278 # We upgrade "d:" to "dt:" to iff using approxidate
280 my $fmt = "\0%Y%m%d";
281 if (!defined($r[1])) {
282 if ($r[0] =~ /\A([0-9]{4})([0-9]{2})([0-9]{2})\z/) {
283 push @$to_parse, "$1-$2-$3";
284 # we could've handled as-is, but we need
285 # to parse anyways for "d+" below
287 push @$to_parse, $r[0];
288 if ($r[0] !~ /\A[0-9]{4}-[0-9]{2}-[0-9]{2}\z/) {
290 $fmt = "\0%Y%m%d%H%M%S";
293 $r[0] = "$fmt+$#$to_parse\0";
297 next if $x eq '' || $x =~ /\A[0-9]{8}\z/;
299 if ($x !~ /\A[0-9]{4}-[0-9]{2}-[0-9]{2}\z/) {
302 $x = "$fmt$#$to_parse\0";
306 s/\0%Y%m%d/\0%Y%m%d%H%M%S/;
307 s/\A([0-9]{8})\z/${1}000000/;
311 } elsif ($pfx eq 'dt') {
312 if (!defined($r[1])) { # git needs gaps and not /\d{14}/
313 if ($r[0] =~ /\A([0-9]{4})([0-9]{2})([0-9]{2})
314 ([0-9]{2})([0-9]{2})([0-9]{2})\z/x) {
315 push @$to_parse, "$1-$2-$3 $4:$5:$6";
317 push @$to_parse, $r[0];
319 $r[0] = "\0%Y%m%d%H%M%S$#$to_parse\0";
320 $r[1] = "\0%Y%m%d%H%M%S+\0";
323 next if $x eq '' || $x =~ /\A[0-9]{14}\z/;
325 $x = "\0%Y%m%d%H%M%S$#$to_parse\0";
328 } else { # "rt", let git interpret "YYYY", deal with Y10K later :P
330 next if $x eq '' || $x =~ /\A[0-9]{5,}\z/;
332 $x = "\0%s$#$to_parse\0";
336 "$pfx:".join('..', @r).$end;
339 sub date_parse_finalize {
340 my ($git, $to_parse) = @_;
341 # git-rev-parse can handle any number of args up to system
342 # limits (around (4096*32) bytes on Linux).
343 my @r = $git->date_parse(@$to_parse);
345 $_[2] =~ s/\0(%[%YmdHMSs]+)([0-9\+]+)\0/strftime($1,
346 gmtime($2 eq '+' ? ($r[$i]+86400) : $r[$i=$2+0]))/sge;
349 # n.b. argv never has NUL, though we'll need to filter it out
350 # if this $argv isn't from a command execution
351 sub query_argv_to_string {
352 my (undef, $git, $argv) = @_;
354 my $tmp = join(' ', map {;
355 if (s!\b(d|rt|dt):(\S+)\z!date_parse_prepare(
356 $to_parse //= [], $1, $2)!sge) {
359 s/(.*?)\b(\w+:)// ? qq{$1$2"$_"} : qq{"$_"};
364 date_parse_finalize($git, $to_parse, $tmp) if $to_parse;
368 # this is for the WWW "q=" query parameter and "lei q --stdin"
369 # it can't do d:"5 days ago", but it will do d:5.days.ago
370 sub query_approxidate {
371 my (undef, $git) = @_; # $_[2] = $query_string (modified in-place)
372 my $DQ = qq<"\x{201c}\x{201d}>; # Xapian can use curly quotes
373 $_[2] =~ tr/\x00/ /; # Xapian doesn't do NUL, we use it as a placeholder
374 my ($terms, $phrase, $to_parse);
375 $_[2] =~ s{([^$DQ]*)([$DQ][^$DQ]*[$DQ])?}{
376 ($terms, $phrase) = ($1, $2);
377 $terms =~ s!\b(d|rt|dt):(\S+)!
378 date_parse_prepare($to_parse //= [], $1, $2)!sge;
379 $terms.($phrase // '');
381 date_parse_finalize($git, $to_parse, $_[2]) if $to_parse;
386 my ($self, $query_string, $opts) = @_;
388 my $qp = $self->{qp} //= $self->qparse_new;
389 my $query = $qp->parse_query($query_string, $self->{qp_flags});
390 _do_enquire($self, $query, $opts);
394 my ($self, $cb, @arg) = @_;
398 eval { @ret = $cb->($self, @arg) };
399 return @ret unless $@;
402 eval { $ret = $cb->($self, @arg) };
403 return $ret unless $@;
405 # Exception: The revision being read has been discarded -
406 # you should call Xapian::Database::reopen()
407 if (ref($@) =~ /\bDatabaseModifiedError\b/) {
408 warn "reopen try #$i on $@\n";
411 # let caller decide how to spew, because ExtMsg queries
412 # get wonky and trigger:
413 # "something terrible happened at .../Xapian/Enquire.pm"
417 die "Too many Xapian database modifications in progress\n";
421 my ($self, $query, $opts) = @_;
422 retry_reopen($self, \&_enquire_once, $query, $opts);
425 # returns true if all docs have the THREADID value
426 sub has_threadid ($) {
428 (xdb($self)->get_metadata('has_threadid') // '') eq '1';
431 sub _enquire_once { # retry_reopen callback
432 my ($self, $query, $opts) = @_;
433 my $xdb = xdb($self);
434 if (defined(my $eidx_key = $opts->{eidx_key})) {
435 $query = $X{Query}->new(OP_FILTER(), $query, 'O'.$eidx_key);
437 if (defined(my $uid_range = $opts->{uid_range})) {
438 my $range = $X{Query}->new(OP_VALUE_RANGE(), UID,
439 sortable_serialise($uid_range->[0]),
440 sortable_serialise($uid_range->[1]));
441 $query = $X{Query}->new(OP_FILTER(), $query, $range);
443 my $enquire = $X{Enquire}->new($xdb);
444 $enquire->set_query($query);
446 my $rel = $opts->{relevance} // 0;
447 if ($rel == -2) { # ORDER BY docid/UID (highest first)
448 $enquire->set_weighting_scheme($X{BoolWeight}->new);
449 $enquire->set_docid_order($ENQ_DESCENDING);
450 } elsif ($rel == -1) { # ORDER BY docid/UID (lowest first)
451 $enquire->set_weighting_scheme($X{BoolWeight}->new);
452 $enquire->set_docid_order($ENQ_ASCENDING);
453 } elsif ($rel == 0) {
454 $enquire->set_sort_by_value_then_relevance(TS, !$opts->{asc});
456 $enquire->set_sort_by_relevance_then_value(TS, !$opts->{asc});
459 # `mairix -t / --threads' or JMAP collapseThreads
460 if ($opts->{threads} && has_threadid($self)) {
461 $enquire->set_collapse_key(THREADID);
463 $enquire->get_mset($opts->{offset} || 0, $opts->{limit} || 50);
467 my ($self, $ibx, $mset) = @_;
468 my $nshard = $self->{nshard};
470 my %order = map { mdocid($nshard, $_) => ++$i } $mset->items;
472 $order{$a->{num}} <=> $order{$b->{num}}
473 } @{$ibx->over->get_all(keys %order)};
474 wantarray ? ($mset->get_matches_estimated, \@msgs) : \@msgs;
478 sub stemmer { $X{Stem}->new($LANG) }
484 my $xdb = xdb($self);
485 my $qp = $X{QueryParser}->new;
486 $qp->set_default_op(OP_AND());
487 $qp->set_database($xdb);
488 $qp->set_stemmer(stemmer($self));
489 $qp->set_stemming_strategy(STEM_SOME());
490 my $cb = $qp->can('set_max_wildcard_expansion') //
491 $qp->can('set_max_expansion'); # Xapian 1.5.0+
493 $cb = $qp->can('add_valuerangeprocessor') //
494 $qp->can('add_rangeprocessor'); # Xapian 1.5.0+
495 $cb->($qp, $NVRP->new(YYYYMMDD, 'd:'));
496 $cb->($qp, $NVRP->new(DT, 'dt:'));
498 # for IMAP, undocumented for WWW and may be split off go away
499 $cb->($qp, $NVRP->new(BYTES, 'z:'));
500 $cb->($qp, $NVRP->new(TS, 'rt:'));
501 $cb->($qp, $NVRP->new(UID, 'uid:'));
503 while (my ($name, $prefix) = each %bool_pfx_external) {
504 $qp->add_boolean_prefix($name, $_) foreach split(/ /, $prefix);
507 # we do not actually create AltId objects,
508 # just parse the spec to avoid the extra DB handles for now.
509 if (my $altid = $self->{altid}) {
510 my $user_pfx = $self->{-user_pfx} = [];
512 # $_ = 'serial:gmane:/path/to/gmane.msgmap.sqlite3'
513 # note: Xapian supports multibyte UTF-8, /^[0-9]+$/,
514 # and '_' with prefixes matching \w+
515 /\Aserial:(\w+):/ or next;
517 push @$user_pfx, "$pfx:", <<EOF;
518 alternate serial number e.g. $pfx:12345 (boolean)
521 $qp->add_boolean_prefix($pfx, 'X'.uc($pfx));
526 while (my ($name, $prefix) = each %prob_prefix) {
527 $qp->add_prefix($name, $_) foreach split(/ /, $prefix);
534 $self->{qp} //= $self->qparse_new; # parse altids
536 if (my $user_pfx = $self->{-user_pfx}) {
537 push @ret, @$user_pfx;
543 my ($doc, $col) = @_;
544 my $val = $doc->get_value($col) or return; # undefined is '' in Xapian
545 sortable_unserialise($val) + 0; # PV => IV conversion
548 sub get_pct ($) { # mset item
549 # Capped at "99%" since "100%" takes an extra column in the
550 # thread skeleton view. <xapian/mset.h> says the value isn't
551 # very meaningful, anyways.
552 my $n = $_[0]->get_percent;
556 sub xap_terms ($$;@) {
557 my ($pfx, $xdb_or_doc, @docid) = @_; # @docid may be empty ()
560 my $end = $xdb_or_doc->termlist_end(@docid);
561 my $cur = $xdb_or_doc->termlist_begin(@docid);
562 for (; $cur != $end; $cur++) {
564 last if $cur == $end;
565 my $tn = $cur->get_termname;
566 if (index($tn, $pfx) == 0) {
567 $ret{substr($tn, length($pfx))} = undef;