1 # Copyright (C) 2015-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
6 use PublicInbox::TestCommon;
7 require_mods(qw(DBD::SQLite Search::Xapian));
8 require PublicInbox::SearchIdx;
9 require PublicInbox::Inbox;
10 require PublicInbox::InboxWritable;
12 my ($tmpdir, $for_destroy) = tmpdir();
13 my $git_dir = "$tmpdir/a.git";
14 my $ibx = PublicInbox::Inbox->new({ inboxdir => $git_dir });
15 my ($root_id, $last_id);
17 is(0, xsys(qw(git init --shared -q --bare), $git_dir), "git init (main)")
18 or BAIL_OUT("`git init --shared' failed, weird FS or seccomp?");
19 eval { PublicInbox::Search->new($ibx)->xdb };
20 ok($@, "exception raised on non-existent DB");
22 my $rw = PublicInbox::SearchIdx->new($ibx, 1);
23 $ibx->with_umask(sub {
29 $rw->commit_txn_lazy if $rw;
30 $rw = PublicInbox::SearchIdx->new($ibx, 1);
31 $rw->{qp_flags} = 0; # quiet a warning
37 my ($got, $exp, $msg) = @_;
38 is(sprintf('0%03o', $got), sprintf('0%03o', $exp), $msg);
42 # git repository perms
43 oct_is($ibx->_git_config_perm(),
44 &PublicInbox::InboxWritable::PERM_GROUP,
45 'undefined permission is group');
47 [ '0644', 0022, '644 => umask(0022)' ],
48 [ '0600', 0077, '600 => umask(0077)' ],
49 [ '0640', 0027, '640 => umask(0027)' ],
50 [ 'group', 0007, 'group => umask(0007)' ],
51 [ 'everybody', 0002, 'everybody => umask(0002)' ],
52 [ 'umask', umask, 'umask => existing umask' ],
55 my ($perm, $exp, $msg) = @$_;
56 my $got = PublicInbox::InboxWritable::_umask_for(
57 PublicInbox::InboxWritable->_git_config_perm($perm));
58 oct_is($got, $exp, $msg);
63 my $crlf_adjust = \&PublicInbox::SearchIdx::crlf_adjust;
64 is($crlf_adjust->("hi\r\nworld\r\n"), 0, 'no adjustment needed');
65 is($crlf_adjust->("hi\nworld\n"), 2, 'LF-only counts two CR');
66 is($crlf_adjust->("hi\r\nworld\n"), 1, 'CRLF/LF-mix 1 counts 1 CR');
67 is($crlf_adjust->("hi\nworld\r\n"), 1, 'CRLF/LF-mix 2 counts 1 CR');
70 $ibx->with_umask(sub {
71 my $root = PublicInbox::Eml->new(<<'EOF');
72 Date: Fri, 02 Oct 1993 00:00:00 +0000
75 From: John Smith <js@example.com>
77 List-Id: I'm not mad <i.m.just.bored>
81 my $last = PublicInbox::Eml->new(<<'EOF');
82 Date: Sat, 02 Oct 2010 00:00:00 +0000
83 Subject: Re: Hello world
86 From: John Smith <js@example.com>
89 List-Id: there's nothing <left.for.me.to.do>
95 $root_id = $rw->add_message($root);
96 is($root_id, int($root_id), "root_id is an integer: $root_id");
97 $last_id = $rw->add_message($last);
98 is($last_id, int($last_id), "last_id is an integer: $last_id");
103 sort(map { $_->{mid} } @$msgs);
107 my ($query_string, $opt) = @_;
108 my $mset = $ibx->search->mset($query_string, $opt);
109 $ibx->search->mset_to_smsg($ibx, $mset);
114 my $found = $query->('m:root@s');
115 is(scalar(@$found), 1, "message found");
116 is($found->[0]->{mid}, 'root@s', 'mid set correctly') if @$found;
119 my @exp = sort qw(root@s last@s);
121 $res = $query->('s:(Hello world)');
122 @res = filter_mids($res);
123 is_deeply(\@res, \@exp, 'got expected results for s:() match');
125 $res = $query->('s:"Hello world"');
126 @res = filter_mids($res);
127 is_deeply(\@res, \@exp, 'got expected results for s:"" match');
129 $res = $query->('s:"Hello world"', {limit => 1});
130 is(scalar @$res, 1, "limit works");
131 my $first = $res->[0];
133 $res = $query->('s:"Hello world"', {offset => 1});
134 is(scalar @$res, 1, "offset works");
135 my $second = $res->[0];
137 isnt($first, $second, "offset returned different result from limit");
141 $ibx->with_umask(sub {
143 my $rmid = '<ghost-message@s>';
144 my $reply_to_ghost = PublicInbox::Eml->new(<<"EOF");
145 Date: Sat, 02 Oct 2010 00:00:00 +0000
147 Message-ID: <ghost-reply\@s>
149 From: Time Traveler <tt\@example.com>
150 To: list\@example.com
155 my $reply_id = $rw->add_message($reply_to_ghost);
156 is($reply_id, int($reply_id), "reply_id is an integer: $reply_id");
158 my $was_ghost = PublicInbox::Eml->new(<<"EOF");
159 Date: Sat, 02 Oct 2010 00:00:01 +0000
162 From: Laggy Sender <lag\@example.com>
163 To: list\@example.com
167 my $ghost_id = $rw->add_message($was_ghost);
168 is($ghost_id, int($ghost_id), "ghost_id is an integer: $ghost_id");
169 my $msgs = $rw->{oidx}->get_thread('ghost-message@s');
170 is(scalar(@$msgs), 2, 'got both messages in ghost thread');
171 foreach (qw(sid tid)) {
172 is($msgs->[0]->{$_}, $msgs->[1]->{$_}, "{$_} match");
174 isnt($msgs->[0]->{num}, $msgs->[1]->{num}, "num do not match");
175 ok($_->{num} > 0, 'positive art num') foreach @$msgs
178 # search thread on ghost
183 my $res = $query->('ghost');
184 my @exp = sort qw(ghost-message@s ghost-reply@s);
185 my @res = filter_mids($res);
186 is_deeply(\@res, \@exp, 'got expected results for Subject match');
189 $res = $query->('goodbye');
190 is(scalar(@$res), 1, "goodbye message found");
191 is($res->[0]->{mid}, 'last@s', 'got goodbye message body') if @$res;
194 $res = $query->('dt:20101002000001..20101002000001');
195 @res = filter_mids($res);
196 is_deeply(\@res, ['ghost-message@s'], 'exact Date: match works');
197 $res = $query->('dt:20101002000002..20101002000002');
198 is_deeply($res, [], 'exact Date: match down to the second');
202 $ibx->with_umask(sub {
204 my $long_mid = 'last' . ('x' x 60). '@s';
205 my $long = PublicInbox::Eml->new(<<EOF);
206 Date: Sat, 02 Oct 2010 00:00:00 +0000
207 Subject: long message ID
208 References: <root\@s> <last\@s>
209 In-Reply-To: <last\@s>
210 Message-ID: <$long_mid>,
211 From: "Long I.D." <long-id\@example.com>
212 To: list\@example.com
216 my $long_id = $rw->add_message($long);
217 is($long_id, int($long_id), "long_id is an integer: $long_id");
223 my $long_reply_mid = 'reply-to-long@1';
224 my $long_reply = PublicInbox::Eml->new(<<EOF);
225 Subject: I break references
226 Date: Sat, 02 Oct 2010 00:00:00 +0000
227 Message-ID: <$long_reply_mid>
228 In-Reply-To: <$long_mid>
229 From: no1 <no1\@example.com>
230 To: list\@example.com
234 ok($rw->add_message($long_reply) > $long_id, "inserted long reply");
237 my $t = $ibx->over->get_thread('root@s');
238 is(scalar(@$t), 4, "got all 4 messages in thread");
239 my @exp = sort($long_reply_mid, 'root@s', 'last@s', $long_mid);
240 @res = filter_mids($t);
241 is_deeply(\@res, \@exp, "get_thread works");
244 # quote prioritization
245 $ibx->with_umask(sub {
247 $rw->add_message(PublicInbox::Eml->new(<<'EOF'));
248 Date: Sat, 02 Oct 2010 00:00:01 +0000
250 Message-ID: <quote@a>
251 From: Quoter <quoter@example.com>
257 $rw->add_message(PublicInbox::Eml->new(<<'EOF'));
258 Date: Sat, 02 Oct 2010 00:00:02 +0000
260 Message-ID: <nquote@a>
261 From: Non-Quoter<non-quoter@example.com>
268 my $res = $query->("theatre");
269 is(scalar(@$res), 2, "got both matches");
271 is($res->[0]->{mid}, 'nquote@a', 'non-quoted scores higher');
272 is($res->[1]->{mid}, 'quote@a', 'quoted result still returned');
274 $res = $query->("illusions");
275 is(scalar(@$res), 1, "got a match for quoted text");
276 is($res->[0]->{mid}, 'quote@a',
277 "quoted result returned if nothing else") if scalar(@$res);
280 # circular references
281 $ibx->with_umask(sub {
282 my $s = 'foo://'. ('Circle' x 15).'/foo';
283 my $doc_id = $rw->add_message(PublicInbox::Eml->new(<<EOF));
285 Date: Sat, 02 Oct 2010 00:00:01 +0000
286 Message-ID: <circle\@a>
287 References: <circle\@a>
288 In-Reply-To: <circle\@a>
289 From: Circle <circle\@example.com>
290 To: list\@example.com
294 ok($doc_id > 0, "doc_id defined with circular reference");
296 my $smsg = $query->('m:circle@a', {limit=>1})->[0];
297 is(defined($smsg), 1, 'found m:circl@a');
299 is($smsg->{references}, '', "no references created");
300 is($smsg->{subject}, $s, 'long subject not rewritten');
305 my $msgs = $query->('d:19931002..20101002');
306 ok(scalar(@$msgs) > 0, 'got results within range');
307 $msgs = $query->('d:20101003..');
308 is(scalar(@$msgs), 0, 'nothing after 20101003');
309 $msgs = $query->('d:..19931001');
310 is(scalar(@$msgs), 0, 'nothing before 19931001');
313 $ibx->with_umask(sub {
314 my $mime = eml_load 't/utf8.eml';
315 my $doc_id = $rw->add_message($mime);
316 ok($doc_id > 0, 'message indexed doc_id with UTF-8');
318 my $msg = $query->('m:testmessage@example.com', {limit => 1})->[0];
319 is(defined($msg), 1, 'found testmessage@example.com');
321 is($mime->header('Subject'), $msg->{subject},
322 'UTF-8 subject preserved');
326 # names and addresses
328 my $mset = $ibx->search->mset('t:list@example.com');
329 is($mset->size, 9, 'searched To: successfully');
330 foreach my $m ($mset->items) {
331 my $smsg = $ibx->over->get_art($m->get_docid);
332 like($smsg->{to}, qr/\blist\@example\.com\b/, 'to appears');
333 my $doc = $m->get_document;
334 my $col = PublicInbox::Search::BYTES();
335 my $bytes = PublicInbox::SearchIdx::get_val($doc, $col);
336 like($bytes, qr/\A[0-9]+\z/, '$bytes stored as digit');
337 ok($bytes > 0, '$bytes is > 0');
338 is($bytes, $smsg->{bytes}, 'bytes Xapian value matches Over');
340 $col = PublicInbox::Search::UID();
341 my $uid = PublicInbox::SearchIdx::get_val($doc, $col);
342 is($uid, $smsg->{num}, 'UID column matches {num}');
343 is($uid, $m->get_docid, 'UID column matches docid');
346 $mset = $ibx->search->mset('tc:list@example.com');
347 is($mset->size, 9, 'searched To+Cc: successfully');
348 foreach my $m ($mset->items) {
349 my $smsg = $ibx->over->get_art($m->get_docid);
350 my $tocc = join("\n", $smsg->{to}, $smsg->{cc});
351 like($tocc, qr/\blist\@example\.com\b/, 'tocc appears');
354 foreach my $pfx ('tcf:', 'c:') {
355 my $mset = $ibx->search->mset($pfx . 'foo@example.com');
356 is($mset->items, 1, "searched $pfx successfully for Cc:");
357 foreach my $m ($mset->items) {
358 my $smsg = $ibx->over->get_art($m->get_docid);
359 like($smsg->{cc}, qr/\bfoo\@example\.com\b/,
364 foreach my $pfx ('', 'tcf:', 'f:') {
365 my $res = $query->($pfx . 'Laggy');
367 "searched $pfx successfully for From:");
368 foreach my $smsg (@$res) {
369 like($smsg->{from_name}, qr/Laggy Sender/,
370 "From appears with $pfx");
377 my $res = $query->('b:hello');
378 is(scalar(@$res), 0, 'no match on body search only');
379 $res = $query->('bs:smith');
381 'no match on body+subject search for From');
383 $res = $query->('q:theatre');
384 is(scalar(@$res), 1, 'only one quoted body');
385 like($res->[0]->{from_name}, qr/\AQuoter/,
386 'got quoted body') if (scalar(@$res));
388 $res = $query->('nq:theatre');
389 is(scalar @$res, 1, 'only one non-quoted body');
390 like($res->[0]->{from_name}, qr/\ANon-Quoter/,
391 'got non-quoted body') if (scalar(@$res));
393 foreach my $pfx (qw(b: bs:)) {
394 $res = $query->($pfx . 'theatre');
395 is(scalar @$res, 2, "searched both bodies for $pfx");
396 like($res->[0]->{from_name}, qr/\ANon-Quoter/,
397 "non-quoter first for $pfx") if scalar(@$res);
401 $ibx->with_umask(sub {
402 my $amsg = eml_load 't/search-amsg.eml';
404 my $smsg = bless { blob => $oid }, 'PublicInbox::Smsg';
405 ok($rw->add_message($amsg, $smsg), 'added attachment');
407 my $n = $query->('n:attached_fart.txt');
408 is(scalar @$n, 1, 'got result for n:');
409 my $res = $query->('part_deux.txt');
410 is(scalar @$res, 1, 'got result without n:');
411 is($n->[0]->{mid}, $res->[0]->{mid},
412 'same result with and without') if scalar(@$res);
413 my $txt = $query->('"inside another"');
414 is(scalar @$txt, 1, 'found inside another');
415 is($txt->[0]->{mid}, $res->[0]->{mid},
416 'search inside text attachments works') if scalar(@$txt);
419 if (scalar(@$n) >= 1) {
420 my $mid = $n->[0]->{mid};
422 $art = $ibx->over->next_by_mid($mid, \$id, \$prev);
423 ok($art, 'article exists in OVER DB');
426 $rw->unindex_eml($oid, $amsg);
427 $rw->commit_txn_lazy;
429 skip('$art not defined', 1) unless defined $art;
430 is($ibx->over->get_art($art->{num}), undef,
431 'gone from OVER DB');
435 my $all_mask = 07777;
436 my $dir_mask = 02770;
438 # FreeBSD and apparently OpenBSD does not allow non-root users to set S_ISGID,
439 # so git doesn't set it, either (see DIR_HAS_BSD_GROUP_SEMANTICS in git.git)
440 if ($^O =~ /(?:free|open)bsd/i) {
445 foreach my $f ("$git_dir/public-inbox/msgmap.sqlite3",
446 "$git_dir/public-inbox",
447 glob("$git_dir/public-inbox/xapian*/"),
448 glob("$git_dir/public-inbox/xapian*/*")) {
450 my ($bn) = (split(m!/!, $f))[-1];
451 oct_is($st[2] & $all_mask, -f _ ? 0660 : $dir_mask,
452 "sharedRepository respected for $bn");
455 $ibx->with_umask(sub {
457 my $digits = '10010260936330';
458 my $ua = 'Pine.LNX.4.10';
459 my $mid = "$ua.$digits.2460-100000\@penguin.transmeta.com";
460 is($ibx->search->mset("m:$digits")->size, 0, 'no results yet');
461 my $pine = PublicInbox::Eml->new(<<EOF);
464 From: torvalds\@transmeta
465 To: list\@example.com
468 my $x = $rw->add_message($pine);
469 $rw->commit_txn_lazy;
470 $ibx->search->reopen;
471 is($ibx->search->mset("m:$digits")->size, 1,
472 'searching only digit yielded result');
477 is($ibx->search->mset("m:$wild*")->size, 1,
478 "searching chopped($i) digit yielded result $wild ");
480 is($ibx->search->mset('m:Pine m:LNX m:10010260936330')->size, 1);
483 { # List-Id searching
484 my $found = $query->('lid:i.m.just.bored');
485 is_deeply([ filter_mids($found) ], [ 'root@s' ],
486 'got expected mid on exact lid: search');
488 $found = $query->('lid:just.bored');
489 is_deeply($found, [], 'got nothing on lid: search');
491 $found = $query->('lid:*.just.bored');
492 is_deeply($found, [], 'got nothing on lid: search');
494 $found = $query->('l:i.m.just.bored');
495 is_deeply([ filter_mids($found) ], [ 'root@s' ],
496 'probabilistic search works on full List-Id contents');
498 $found = $query->('l:just.bored');
499 is_deeply([ filter_mids($found) ], [ 'root@s' ],
500 'probabilistic search works on partial List-Id contents');
502 $found = $query->('lid:mad');
503 is_deeply($found, [], 'no match on phrase with lid:');
505 $found = $query->('lid:bored');
506 is_deeply($found, [], 'no match on partial List-Id with lid:');
508 $found = $query->('l:nothing');
509 is_deeply($found, [], 'matched on phrase with l:');
512 $ibx->with_umask(sub {
514 my $doc_id = $rw->add_message(eml_load('t/data/message_embed.eml'));
515 ok($doc_id > 0, 'messages within messages');
516 $rw->commit_txn_lazy;
517 $ibx->search->reopen;
518 my $n_test_eml = $query->('n:test.eml');
519 is(scalar(@$n_test_eml), 1, 'got a result');
520 my $n_embed2x_eml = $query->('n:embed2x.eml');
521 is_deeply($n_test_eml, $n_embed2x_eml, '.eml filenames searchable');
522 for my $m (qw(20200418222508.GA13918@dcvr 20200418222020.GA2745@dcvr
523 20200418214114.7575-1-e@yhbt.net)) {
524 is($query->("m:$m")->[0]->{mid},
525 '20200418222508.GA13918@dcvr', 'probabilistic m:'.$m);
526 is($query->("mid:$m")->[0]->{mid},
527 '20200418222508.GA13918@dcvr', 'boolean mid:'.$m);
529 is($query->('dfpost:4dc62c50')->[0]->{mid},
530 '20200418222508.GA13918@dcvr',
531 'diff search reaches inside message/rfc822');
532 is($query->('s:"mail header experiments"')->[0]->{mid},
533 '20200418222508.GA13918@dcvr',
534 'Subject search reaches inside message/rfc822');