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;
9 use parent qw(Exporter);
10 our @EXPORT_OK = qw(retry_reopen int_val get_pct xap_terms);
11 use List::Util qw(max);
12 use POSIX qw(strftime);
15 # values for searching, changing the numeric value breaks
16 # compatibility with old indices (so don't change them it)
18 TS => 0, # Received: in Unix time (IMAP INTERNALDATE, JMAP receivedAt)
19 YYYYMMDD => 1, # redundant with DT below
20 DT => 2, # Date: YYYYMMDDHHMMSS (IMAP SENT*, JMAP sentAt)
22 # added for public-inbox 1.6.0+
23 BYTES => 3, # IMAP RFC822.SIZE
24 UID => 4, # IMAP UID == NNTP article number == Xapian docid
25 THREADID => 5, # RFC 8474, RFC 8621
28 # REPLYCNT => ?, # IMAP ANSWERED
30 # SCHEMA_VERSION history
32 # 1 - subject_path is lower-cased
33 # 2 - subject_path is id_compress in the index, only
34 # 3 - message-ID is compressed if it includes '%' (hack!)
35 # 4 - change "Re: " normalization, avoid circular Reference ghosts
36 # 5 - subject_path drops trailing '.'
37 # 6 - preserve References: order in document data
38 # 7 - remove references and inreplyto terms
39 # 8 - remove redundant/unneeded document data
40 # 9 - disable Message-ID compression (SHA-1)
41 # 10 - optimize doc for NNTP overviews
42 # 11 - merge threads when vivifying ghosts
43 # 12 - change YYYYMMDD value column to numeric
44 # 13 - fix threading for empty References/In-Reply-To
45 # (commit 83425ef12e4b65cdcecd11ddcb38175d4a91d5a0)
46 # 14 - fix ghost root vivification
47 # 15 - see public-inbox-v2-format(5)
48 # further bumps likely unnecessary, we'll suggest in-place
49 # "--reindex" use for further fixes and tweaks:
51 # public-inbox v1.5.0 adds (still SCHEMA_VERSION=15):
52 # * "lid:" and "l:" for List-Id searches
54 # v1.6.0 adds BYTES, UID and THREADID values
58 use PublicInbox::Smsg;
59 use PublicInbox::Over;
61 our %X = map { $_ => 0 } qw(BoolWeight Database Enquire QueryParser Stem Query);
62 our $Xap; # 'Search::Xapian' or 'Xapian'
63 our $NVRP; # '$Xap::'.('NumberValueRangeProcessor' or 'NumberRangeProcessor')
65 # ENQ_DESCENDING and ENQ_ASCENDING weren't in SWIG Xapian.pm prior to 1.4.16,
66 # let's hope the ABI is stable
67 our $ENQ_DESCENDING = 0;
68 our $ENQ_ASCENDING = 1;
71 return 1 if defined $Xap;
72 # n.b. PI_XAPIAN is intended for development use only. We still
73 # favor Search::Xapian since that's what's available in current
74 # Debian stable (10.x) and derived distros.
75 for my $x (($ENV{PI_XAPIAN} // 'Search::Xapian'), 'Xapian') {
79 $x->import(qw(:standard));
82 # `version_string' was added in Xapian 1.1
83 my $xver = eval('v'.eval($x.'::version_string()')) //
84 eval('v'.eval($x.'::xapian_version_string()'));
86 # NumberRangeProcessor was added in Xapian 1.3.6,
87 # NumberValueRangeProcessor was removed for 1.5.0+,
88 # favor the older /Value/ variant since that's what our
89 # (currently) preferred Search::Xapian supports
90 $NVRP = $x.'::'.($x eq 'Xapian' && $xver ge v1.5 ?
91 'NumberRangeProcessor' : 'NumberValueRangeProcessor');
92 $X{$_} = $Xap.'::'.$_ for (keys %X);
94 *sortable_serialise = $x.'::sortable_serialise';
95 *sortable_unserialise = $x.'::sortable_unserialise';
96 # n.b. FLAG_PURE_NOT is expensive not suitable for a public
97 # website as it could become a denial-of-service vector
98 # FLAG_PHRASE also seems to cause performance problems chert
99 # (and probably earlier Xapian DBs). glass seems fine...
100 # TODO: make this an option, maybe?
101 # or make indexlevel=medium as default
102 $QP_FLAGS = FLAG_PHRASE() | FLAG_BOOLEAN() | FLAG_LOVEHATE() |
109 # This is English-only, everything else is non-standard and may be confused as
110 # a prefix common in patch emails
111 our $LANG = 'english';
113 # note: the non-X term prefix allocations are shared with
114 # Xapian omega, see xapian-applications/omega/docs/termprefixes.rst
115 my %bool_pfx_external = (
116 mid => 'Q', # Message-ID (full/exact), this is mostly uniQue
117 lid => 'G', # newsGroup (or similar entity), just inside <>
120 dfblob => 'XDFPRE XDFPOST',
123 my $non_quoted_body = 'XNQ XDFN XDFA XDFB XDFHH XDFCTX XDFPRE XDFPOST';
125 # for mairix compatibility
127 m => 'XM', # 'mid:' (bool) is exact, 'm:' (prob) can do partial
128 l => 'XL', # 'lid:' (bool) is exact, 'l:' (prob) can do partial
135 b => $non_quoted_body . ' XQUOT',
136 bs => $non_quoted_body . ' XQUOT S',
140 nq => $non_quoted_body,
148 '' => 'XM S A XQUOT XFN ' . $non_quoted_body,
151 # not documenting m: and mid: for now, the using the URLs works w/o Xapian
152 # not documenting lid: for now, either, it is probably redundant with l:,
153 # especially since we don't offer boolean searches for To/Cc/From
156 's:' => 'match within Subject e.g. s:"a quick brown fox"',
158 match date-time range, git "approxidate" formats supported
159 Open-ended ranges such as `d:last.week..' and
160 `d:..2.days.ago' are supported
162 'b:' => 'match within message body, including text attachments',
163 'nq:' => 'match non-quoted text within message body',
164 'q:' => 'match quoted text within message body',
165 'n:' => 'match filename of attachment(s)',
166 't:' => 'match within the To header',
167 'c:' => 'match within the Cc header',
168 'f:' => 'match within the From header',
169 'a:' => 'match within the To, Cc, and From headers',
170 'tc:' => 'match within the To and Cc headers',
171 'l:' => 'match contents of the List-Id header',
172 'bs:' => 'match within the Subject and body',
173 'dfn:' => 'match filename from diff',
174 'dfa:' => 'match diff removed (-) lines',
175 'dfb:' => 'match diff added (+) lines',
176 'dfhh:' => 'match diff hunk header context (usually a function name)',
177 'dfctx:' => 'match diff context lines',
178 'dfpre:' => 'match pre-image git blob ID',
179 'dfpost:' => 'match post-image git blob ID',
180 'dfblob:' => 'match either pre or post-image git blob ID',
182 match received time, like `d:' if sender's clock was correct
188 my ($self, $rdonly) = @_;
189 if ($rdonly || !defined($self->{shard})) {
191 } else { # v2 + extindex only:
192 "$self->{xpfx}/$self->{shard}";
196 # returns all shards as separate Xapian::Database objects w/o combining
197 sub xdb_shards_flat ($) {
199 my $xpfx = $self->{xpfx};
200 my (@xdb, $slow_phrase);
202 $self->{qp_flags} //= $QP_FLAGS;
203 if ($xpfx =~ m!/xapian[0-9]+\z!) {
204 @xdb = ($X{Database}->new($xpfx));
205 $self->{qp_flags} |= FLAG_PHRASE() if !-f "$xpfx/iamchert";
207 opendir(my $dh, $xpfx) or return (); # not initialized yet
208 # We need numeric sorting so shard[0] is first for reading
209 # Xapian metadata, if needed
210 my $last = max(grep(/\A[0-9]+\z/, readdir($dh))) // return ();
212 my $shard_dir = "$self->{xpfx}/$_";
213 push @xdb, $X{Database}->new($shard_dir);
214 $slow_phrase ||= -f "$shard_dir/iamchert";
216 $self->{qp_flags} |= FLAG_PHRASE() if !$slow_phrase;
221 # v2 Xapian docids don't conflict, so they're identical to
222 # NNTP article numbers and IMAP UIDs.
223 # https://trac.xapian.org/wiki/FAQ/MultiDatabaseDocumentID
225 my ($nshard, $mitem) = @_;
226 my $docid = $mitem->get_docid;
227 int(($docid - 1) / $nshard) + 1;
230 sub mset_to_artnums {
231 my ($self, $mset) = @_;
232 my $nshard = $self->{nshard};
233 [ map { mdocid($nshard, $_) } $mset->items ];
239 my @xdb = $self->xdb_shards_flat or return;
240 $self->{nshard} = scalar(@xdb);
241 my $xdb = shift @xdb;
242 $xdb->add_database($_) for @xdb;
248 my ($class, $ibx) = @_;
249 ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
250 my $xap = $ibx->version > 1 ? 'xap' : 'public-inbox/xapian';
251 my $xpfx = "$ibx->{inboxdir}/$xap".SCHEMA_VERSION;
252 my $self = bless { xpfx => $xpfx }, $class;
253 $self->{altid} = $ibx->{altid} if defined($ibx->{altid});
259 if (my $xdb = $self->{xdb}) {
262 $self; # make chaining easier
265 # Convert git "approxidate" ranges to something usable with our
266 # Xapian indices. At the moment, Xapian only offers a C++-only API
267 # and neither the SWIG nor XS bindings allow us to use custom code
268 # to parse dates (and libgit2 doesn't expose git__date_parse, either,
269 # so we're running git-rev-parse(1)).
270 # This replaces things we need to send to $git->date_parse with
271 # "\0".$strftime_format.['+'|$idx]."\0" placeholders
272 sub date_parse_prepare {
273 my ($to_parse, $pfx, $range) = @_;
274 # are we inside a parenthesized statement?
275 my $end = $range =~ s/([\)\s]*)\z// ? $1 : '';
276 my @r = split(/\.\./, $range, 2);
278 # expand "d:20101002" => "d:20101002..20101003" and like
279 # n.b. git doesn't do YYYYMMDD w/o '-', it needs YYYY-MM-DD
280 # We upgrade "d:" to "dt:" to iff using approxidate
282 my $fmt = "\0%Y%m%d";
283 if (!defined($r[1])) {
284 if ($r[0] =~ /\A([0-9]{4})([0-9]{2})([0-9]{2})\z/) {
285 push @$to_parse, "$1-$2-$3";
286 # we could've handled as-is, but we need
287 # to parse anyways for "d+" below
289 push @$to_parse, $r[0];
290 if ($r[0] !~ /\A[0-9]{4}-[0-9]{2}-[0-9]{2}\z/) {
292 $fmt = "\0%Y%m%d%H%M%S";
295 $r[0] = "$fmt+$#$to_parse\0";
299 next if $x eq '' || $x =~ /\A[0-9]{8}\z/;
301 if ($x !~ /\A[0-9]{4}-[0-9]{2}-[0-9]{2}\z/) {
304 $x = "$fmt$#$to_parse\0";
308 s/\0%Y%m%d/\0%Y%m%d%H%M%S/;
309 s/\A([0-9]{8})\z/${1}000000/;
313 } elsif ($pfx eq 'dt') {
314 if (!defined($r[1])) { # git needs gaps and not /\d{14}/
315 if ($r[0] =~ /\A([0-9]{4})([0-9]{2})([0-9]{2})
316 ([0-9]{2})([0-9]{2})([0-9]{2})\z/x) {
317 push @$to_parse, "$1-$2-$3 $4:$5:$6";
319 push @$to_parse, $r[0];
321 $r[0] = "\0%Y%m%d%H%M%S$#$to_parse\0";
322 $r[1] = "\0%Y%m%d%H%M%S+\0";
325 next if $x eq '' || $x =~ /\A[0-9]{14}\z/;
327 $x = "\0%Y%m%d%H%M%S$#$to_parse\0";
330 } else { # "rt", let git interpret "YYYY", deal with Y10K later :P
332 next if $x eq '' || $x =~ /\A[0-9]{5,}\z/;
334 $x = "\0%s$#$to_parse\0";
336 $r[1] //= "\0%s+\0"; # add 1 day
338 "$pfx:".join('..', @r).$end;
341 sub date_parse_finalize {
342 my ($git, $to_parse) = @_;
343 # git-rev-parse can handle any number of args up to system
344 # limits (around (4096*32) bytes on Linux).
345 my @r = $git->date_parse(@$to_parse);
346 # n.b. git respects TZ, times stored in SQLite/Xapian are always UTC,
347 # and gmtime doesn't seem to do the right thing when TZ!=UTC
349 $_[2] =~ s/\0(%[%YmdHMSs]+)([0-9\+]+)\0/
350 $t = $2 eq '+' ? ($r[$i]+86400) : $r[$i=$2+0];
351 $1 eq '%s' ? $t : strftime($1, gmtime($t))/sge;
354 # n.b. argv never has NUL, though we'll need to filter it out
355 # if this $argv isn't from a command execution
356 sub query_argv_to_string {
357 my (undef, $git, $argv) = @_;
359 my $tmp = join(' ', map {;
360 if (s!\b(d|rt|dt):(\S+)\z!date_parse_prepare(
361 $to_parse //= [], $1, $2)!sge) {
364 s/(.*?)\b(\w+:)// ? qq{$1$2"$_"} : qq{"$_"};
369 date_parse_finalize($git, $to_parse, $tmp) if $to_parse;
373 # this is for the WWW "q=" query parameter and "lei q --stdin"
374 # it can't do d:"5 days ago", but it will do d:5.days.ago
375 sub query_approxidate {
376 my (undef, $git) = @_; # $_[2] = $query_string (modified in-place)
377 my $DQ = qq<"\x{201c}\x{201d}>; # Xapian can use curly quotes
378 $_[2] =~ tr/\x00/ /; # Xapian doesn't do NUL, we use it as a placeholder
379 my ($terms, $phrase, $to_parse);
380 $_[2] =~ s{([^$DQ]*)([$DQ][^$DQ]*[$DQ])?}{
381 ($terms, $phrase) = ($1, $2);
382 $terms =~ s!\b(d|rt|dt):(\S+)!
383 date_parse_prepare($to_parse //= [], $1, $2)!sge;
384 $terms.($phrase // '');
386 date_parse_finalize($git, $to_parse, $_[2]) if $to_parse;
391 my ($self, $query_string, $opts) = @_;
393 my $qp = $self->{qp} //= $self->qparse_new;
394 my $query = $qp->parse_query($query_string, $self->{qp_flags});
395 _do_enquire($self, $query, $opts);
399 my ($self, $cb, @arg) = @_;
402 my @ret = eval { $cb->($self, @arg) };
403 return @ret unless $@;
405 my $ret = eval { $cb->($self, @arg) };
406 return $ret unless $@;
408 # Exception: The revision being read has been discarded -
409 # you should call Xapian::Database::reopen()
410 if (ref($@) =~ /\bDatabaseModifiedError\b/) {
413 # let caller decide how to spew, because ExtMsg queries
414 # get wonky and trigger:
415 # "something terrible happened at .../Xapian/Enquire.pm"
419 Carp::croak("Too many Xapian database modifications in progress\n");
423 my ($self, $query, $opts) = @_;
424 retry_reopen($self, \&_enquire_once, $query, $opts);
427 # returns true if all docs have the THREADID value
428 sub has_threadid ($) {
430 (xdb($self)->get_metadata('has_threadid') // '') eq '1';
433 sub _enquire_once { # retry_reopen callback
434 my ($self, $query, $opts) = @_;
435 my $xdb = xdb($self);
436 if (defined(my $eidx_key = $opts->{eidx_key})) {
437 $query = $X{Query}->new(OP_FILTER(), $query, 'O'.$eidx_key);
439 if (defined(my $uid_range = $opts->{uid_range})) {
440 my $range = $X{Query}->new(OP_VALUE_RANGE(), UID,
441 sortable_serialise($uid_range->[0]),
442 sortable_serialise($uid_range->[1]));
443 $query = $X{Query}->new(OP_FILTER(), $query, $range);
445 my $enquire = $X{Enquire}->new($xdb);
446 $enquire->set_query($query);
448 my $rel = $opts->{relevance} // 0;
449 if ($rel == -2) { # ORDER BY docid/UID (highest first)
450 $enquire->set_weighting_scheme($X{BoolWeight}->new);
451 $enquire->set_docid_order($ENQ_DESCENDING);
452 } elsif ($rel == -1) { # ORDER BY docid/UID (lowest first)
453 $enquire->set_weighting_scheme($X{BoolWeight}->new);
454 $enquire->set_docid_order($ENQ_ASCENDING);
455 } elsif ($rel == 0) {
456 $enquire->set_sort_by_value_then_relevance(TS, !$opts->{asc});
458 $enquire->set_sort_by_relevance_then_value(TS, !$opts->{asc});
461 # `mairix -t / --threads' or JMAP collapseThreads
462 if ($opts->{threads} && has_threadid($self)) {
463 $enquire->set_collapse_key(THREADID);
465 $enquire->get_mset($opts->{offset} || 0, $opts->{limit} || 50);
469 my ($self, $ibx, $mset) = @_;
470 my $nshard = $self->{nshard};
472 my %order = map { mdocid($nshard, $_) => ++$i } $mset->items;
474 $order{$a->{num}} <=> $order{$b->{num}}
475 } @{$ibx->over->get_all(keys %order)};
476 wantarray ? ($mset->get_matches_estimated, \@msgs) : \@msgs;
480 sub stemmer { $X{Stem}->new($LANG) }
486 my $xdb = xdb($self);
487 my $qp = $X{QueryParser}->new;
488 $qp->set_default_op(OP_AND());
489 $qp->set_database($xdb);
490 $qp->set_stemmer(stemmer($self));
491 $qp->set_stemming_strategy(STEM_SOME());
492 my $cb = $qp->can('set_max_wildcard_expansion') //
493 $qp->can('set_max_expansion'); # Xapian 1.5.0+
495 $cb = $qp->can('add_valuerangeprocessor') //
496 $qp->can('add_rangeprocessor'); # Xapian 1.5.0+
497 $cb->($qp, $NVRP->new(YYYYMMDD, 'd:'));
498 $cb->($qp, $NVRP->new(DT, 'dt:'));
500 # for IMAP, undocumented for WWW and may be split off go away
501 $cb->($qp, $NVRP->new(BYTES, 'z:'));
502 $cb->($qp, $NVRP->new(TS, 'rt:'));
503 $cb->($qp, $NVRP->new(UID, 'uid:'));
505 while (my ($name, $prefix) = each %bool_pfx_external) {
506 $qp->add_boolean_prefix($name, $_) foreach split(/ /, $prefix);
509 # we do not actually create AltId objects,
510 # just parse the spec to avoid the extra DB handles for now.
511 if (my $altid = $self->{altid}) {
512 my $user_pfx = $self->{-user_pfx} = [];
514 # $_ = 'serial:gmane:/path/to/gmane.msgmap.sqlite3'
515 # note: Xapian supports multibyte UTF-8, /^[0-9]+$/,
516 # and '_' with prefixes matching \w+
517 /\Aserial:(\w+):/ or next;
519 push @$user_pfx, "$pfx:", <<EOF;
520 alternate serial number e.g. $pfx:12345 (boolean)
523 $qp->add_boolean_prefix($pfx, 'X'.uc($pfx));
528 while (my ($name, $prefix) = each %prob_prefix) {
529 $qp->add_prefix($name, $_) foreach split(/ /, $prefix);
536 $self->{qp} //= $self->qparse_new; # parse altids
538 if (my $user_pfx = $self->{-user_pfx}) {
539 push @ret, @$user_pfx;
545 my ($doc, $col) = @_;
546 my $val = $doc->get_value($col) or return; # undefined is '' in Xapian
547 sortable_unserialise($val) + 0; # PV => IV conversion
550 sub get_pct ($) { # mset item
551 # Capped at "99%" since "100%" takes an extra column in the
552 # thread skeleton view. <xapian/mset.h> says the value isn't
553 # very meaningful, anyways.
554 my $n = $_[0]->get_percent;
558 sub xap_terms ($$;@) {
559 my ($pfx, $xdb_or_doc, @docid) = @_; # @docid may be empty ()
561 my $end = $xdb_or_doc->termlist_end(@docid);
562 my $cur = $xdb_or_doc->termlist_begin(@docid);
563 for (; $cur != $end; $cur++) {
565 last if $cur == $end;
566 my $tn = $cur->get_termname;
567 $ret{substr($tn, length($pfx))} = undef if !index($tn, $pfx);
569 wantarray ? sort(keys(%ret)) : \%ret;
572 # get combined docid from over.num:
573 # (not generic Xapian, only works with our sharding scheme)
575 my ($self, $num) = @_;
576 my $nshard = $self->{nshard};
577 ($num - 1) * $nshard + $num % $nshard + 1;