2 # Copyright (C) all contributors <meta@public-inbox.org>
3 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
6 use PublicInbox::TestCommon;
7 use PublicInbox::Config;
8 use PublicInbox::InboxWritable;
11 require_mods(qw(json DBD::SQLite Search::Xapian));
12 require PublicInbox::Search;
13 use_ok 'PublicInbox::ExtSearch';
14 use_ok 'PublicInbox::ExtSearchIdx';
15 use_ok 'PublicInbox::OverIdx';
16 my ($home, $for_destroy) = tmpdir();
17 local $ENV{HOME} = $home;
18 mkdir "$home/.public-inbox" or BAIL_OUT $!;
19 my $cfg_path = "$home/.public-inbox/config";
20 open my $fh, '>', $cfg_path or BAIL_OUT $!;
21 print $fh <<EOF or BAIL_OUT $!;
25 close $fh or BAIL_OUT $!;
26 my $v2addr = 'v2test@example.com';
27 my $v1addr = 'v1test@example.com';
28 ok(run_script([qw(-init -Lbasic -V2 v2test --newsgroup v2.example),
29 "$home/v2test", 'http://example.com/v2test', $v2addr ]), 'v2test init');
30 my $env = { ORIGINAL_RECIPIENT => $v2addr };
31 my $eml = eml_load('t/utf8.eml');
33 $eml->header_set('List-Id', '<v2.example.com>');
34 open($fh, '+>', undef) or BAIL_OUT $!;
36 print $fh $eml->as_string or BAIL_OUT $!;
37 seek($fh, 0, SEEK_SET) or BAIL_OUT $!;
39 run_script(['-mda', '--no-precheck'], $env, { 0 => $fh }) or BAIL_OUT '-mda';
41 ok(run_script([qw(-init -V1 v1test --newsgroup v1.example), "$home/v1test",
42 'http://example.com/v1test', $v1addr ]), 'v1test init');
44 $eml->header_set('List-Id', '<v1.example.com>');
45 seek($fh, 0, SEEK_SET) or BAIL_OUT $!;
46 truncate($fh, 0) or BAIL_OUT $!;
47 print $fh $eml->as_string or BAIL_OUT $!;
48 seek($fh, 0, SEEK_SET) or BAIL_OUT $!;
50 $env = { ORIGINAL_RECIPIENT => $v1addr };
51 run_script(['-mda', '--no-precheck'], $env, { 0 => $fh }) or BAIL_OUT '-mda';
53 run_script([qw(-index -Lbasic), "$home/v1test"]) or BAIL_OUT "index $?";
55 ok(run_script([qw(-extindex --dangerous --all), "$home/extindex"]),
58 my $es = PublicInbox::ExtSearch->new("$home/extindex");
59 ok($es->has_threadid, '->has_threadid');
63 xsys([qw(git config publicinbox.v1test.boost), 10],
64 { GIT_CONFIG => $cfg_path });
65 ok(run_script([qw(-extindex --all), "$home/extindex-b"]),
66 'extindex init with boost');
67 my $es = PublicInbox::ExtSearch->new("$home/extindex-b");
68 my $smsg = $es->over->get_art(1);
69 ok($smsg, 'got first article');
70 my $xref3 = $es->over->get_xref3($smsg->{num});
71 my @v1 = grep(/\Av1/, @$xref3);
72 my @v2 = grep(/\Av2/, @$xref3);
73 like($v1[0], qr/\Av1\.example.*?\b\Q$smsg->{blob}\E\b/,
74 'smsg->{blob} respected boost');
75 is(scalar(@$xref3), 2, 'only to entries');
78 xsys([qw(git config publicinbox.v2test.boost), 20],
79 { GIT_CONFIG => $cfg_path });
80 ok(run_script([qw(-extindex --all --reindex), "$home/extindex-b"]),
81 'extindex --reindex with altered boost');
83 $es = PublicInbox::ExtSearch->new("$home/extindex-b");
84 $smsg = $es->over->get_art(1);
85 like($v2[0], qr/\Av2\.example.*?\b\Q$smsg->{blob}\E\b/,
86 'smsg->{blob} respects boost after reindex');
88 # high boost added later
89 my $b2 = "$home/extindex-bb";
90 ok(run_script([qw(-extindex), $b2, "$home/v1test"]),
91 'extindex with low boost inbox only');
92 ok(run_script([qw(-extindex), $b2, "$home/v2test"]),
93 'extindex with high boost inbox only');
94 $es = PublicInbox::ExtSearch->new($b2);
95 $smsg = $es->over->get_art(1);
96 $xref3 = $es->over->get_xref3($smsg->{num});
97 like($v2[0], qr/\Av2\.example.*?\b\Q$smsg->{blob}\E\b/,
98 'smsg->{blob} respected boost across 2 index runs');
100 xsys([qw(git config --unset publicinbox.v1test.boost)],
101 { GIT_CONFIG => $cfg_path });
102 xsys([qw(git config --unset publicinbox.v2test.boost)],
103 { GIT_CONFIG => $cfg_path });
106 { # TODO: -extindex should write this to config
107 open $fh, '>>', $cfg_path or BAIL_OUT $!;
108 print $fh <<EOF or BAIL_OUT $!;
111 topdir = $home/extindex
113 close $fh or BAIL_OUT $!;
115 my $pi_cfg = PublicInbox::Config->new;
117 ok($pi_cfg->ALL, '->ALL');
118 my $ibx = $pi_cfg->{-by_newsgroup}->{'v2.example'};
119 my $ret = $pi_cfg->ALL->nntp_xref_for($ibx, $ibx->over->get_art(1));
120 is_deeply($ret, { 'v1.example' => 1, 'v2.example' => 1 },
125 require_mods(qw(Net::NNTP), 1);
126 my $sock = tcp_server();
127 my $host_port = tcp_host_port($sock);
128 my ($out, $err) = ("$home/nntpd.out.log", "$home/nntpd.err.log");
129 my $cmd = [ '-nntpd', '-W0', "--stdout=$out", "--stderr=$err" ];
130 my $td = start_script($cmd, undef, { 3 => $sock });
131 my $n = Net::NNTP->new($host_port);
132 my @xp = $n->xpath('<testmessage@example.com>');
133 is_deeply(\@xp, [ qw(v1.example/1 v2.example/1) ]);
134 $n->group('v1.example');
135 my $res = $n->head(1);
136 @$res = grep(/^Xref: /, @$res);
137 like($res->[0], qr/ v1\.example:1 v2\.example:1/, 'nntp_xref works');
140 my $es = PublicInbox::ExtSearch->new("$home/extindex");
142 my $smsg = $es->over->get_art(1);
143 ok($smsg, 'got first article');
144 is($es->over->get_art(2), undef, 'only one added');
145 my $xref3 = $es->over->get_xref3(1);
146 like($xref3->[0], qr/\A\Qv2.example\E:1:/, 'order preserved 1');
147 like($xref3->[1], qr/\A\Qv1.example\E:1:/, 'order preserved 2');
148 is(scalar(@$xref3), 2, 'only to entries');
151 if ('inbox edited') {
152 my ($in, $out, $err);
153 $in = $out = $err = '';
154 my $opt = { 0 => \$in, 1 => \$out, 2 => \$err };
155 my $env = { MAIL_EDITOR => "$^X -i -p -e 's/test message/BEST MSG/'" };
156 my $cmd = [ qw(-edit -Ft/utf8.eml), "$home/v2test" ];
157 ok(run_script($cmd, $env, $opt), '-edit');
158 ok(run_script([qw(-extindex --all), "$home/extindex"], undef, $opt),
160 like($err, qr/discontiguous range/, 'warned about discontiguous range');
161 my $msg1 = $es->over->get_art(1) or BAIL_OUT 'msg1 missing';
162 my $msg2 = $es->over->get_art(2) or BAIL_OUT 'msg2 missing';
163 is($msg1->{mid}, $msg2->{mid}, 'edited message indexed');
164 isnt($msg1->{blob}, $msg2->{blob}, 'blobs differ');
165 my $eml2 = $es->smsg_eml($msg2);
166 like($eml2->body, qr/BEST MSG/, 'edited body in #2');
167 unlike($eml2->body, qr/test message/, 'old body discarded in #2');
168 my $eml1 = $es->smsg_eml($msg1);
169 like($eml1->body, qr/test message/, 'original body in #1');
170 my $x1 = $es->over->get_xref3(1);
171 my $x2 = $es->over->get_xref3(2);
172 is(scalar(@$x1), 1, 'original only has one xref3');
173 is(scalar(@$x2), 1, 'new message has one xref3');
174 isnt($x1->[0], $x2->[0], 'xref3 differs');
176 my $mset = $es->mset('b:"BEST MSG"');
177 is($mset->size, 1, 'new message found');
178 $mset = $es->mset('b:"test message"');
179 is($mset->size, 1, 'old message found');
180 delete @$es{qw(git over xdb qp)}; # fork preparation
182 my $pi_cfg = PublicInbox::Config->new;
184 is(scalar($pi_cfg->ALL->mset('s:Testing')->items), 2,
185 '2 results in ->ALL');
188 $pi_cfg->each_inbox(sub {
191 local $SIG{__WARN__} = sub {}; # FIXME support --reindex
192 my $mset = $ibx->isrch->mset('s:Testing');
193 $res->{$ibx->eidx_key} = $ibx->isrch->mset_to_smsg($ibx, $mset);
195 is($nr, 2, 'two inboxes');
197 for my $v (qw(v1 v2)) {
198 my $ibx = $pi_cfg->lookup_newsgroup("$v.example");
199 my $smsg = $ibx->over->get_art(1);
201 $exp->{"$v.example"} = [ $smsg ];
203 is_deeply($res, $exp, 'isearch limited results');
204 $pi_cfg = $res = $exp = undef;
206 open my $rmfh, '+>', undef or BAIL_OUT $!;
208 print $rmfh $eml2->as_string or BAIL_OUT $!;
209 seek($rmfh, 0, SEEK_SET) or BAIL_OUT $!;
211 ok(run_script([qw(-learn rm --all)], undef, $opt), '-learn rm');
213 ok(run_script([qw(-extindex --all), "$home/extindex"], undef, undef),
214 'extindex after rm');
215 is($es->over->get_art(2), undef, 'doc #2 gone');
216 $mset = $es->mset('b:"BEST MSG"');
217 is($mset->size, 0, 'new message gone');
220 my $misc = $es->misc;
221 my @it = $misc->mset('')->items;
222 is(scalar(@it), 2, 'two inboxes');
223 like($it[0]->get_document->get_data, qr/v2test/, 'docdata matched v2');
224 like($it[1]->get_document->get_data, qr/v1test/, 'docdata matched v1');
226 my $cfg = PublicInbox::Config->new;
227 my $schema_version = PublicInbox::Search::SCHEMA_VERSION();
228 my $f = "$home/extindex/ei$schema_version/over.sqlite3";
229 my $oidx = PublicInbox::OverIdx->new($f);
230 if ('inject w/o indexing') {
231 use PublicInbox::Import;
232 my $v1ibx = $cfg->lookup_name('v1test');
233 my $last_v1_commit = $v1ibx->mm->last_commit;
234 my $v2ibx = $cfg->lookup_name('v2test');
235 my $last_v2_commit = $v2ibx->mm->last_commit_xap($schema_version, 0);
236 my $git0 = PublicInbox::Git->new("$v2ibx->{inboxdir}/git/0.git");
237 chomp(my $cmt = $git0->qx(qw(rev-parse HEAD^0)));
238 is($last_v2_commit, $cmt, 'v2 index up-to-date');
240 my $v2im = PublicInbox::Import->new($git0, undef, undef, $v2ibx);
241 $v2im->{lock_path} = undef;
242 $v2im->{path_type} = 'v2';
243 $v2im->add(eml_load('t/mda-mime.eml'));
245 chomp(my $tip = $git0->qx(qw(rev-parse HEAD^0)));
246 isnt($tip, $cmt, '0.git v2 updated');
248 # inject a message w/o updating index
249 rename("$home/v1test/public-inbox", "$home/v1test/skip-index") or
251 open(my $eh, '<', 't/iso-2202-jp.eml') or BAIL_OUT $!;
252 run_script(['-mda', '--no-precheck'], $env, { 0 => $eh}) or
254 rename("$home/v1test/skip-index", "$home/v1test/public-inbox") or
257 my ($in, $out, $err);
258 $in = $out = $err = '';
259 my $opt = { 0 => \$in, 1 => \$out, 2 => \$err };
260 ok(run_script([qw(-extindex -v -v --all), "$home/extindex"],
261 undef, undef), 'extindex noop');
263 my $mset = $es->mset('mid:199707281508.AAA24167@hoyogw.example');
264 is($mset->size, 0, 'did not attempt to index unindexed v1 message');
265 $mset = $es->mset('mid:multipart-html-sucks@11');
266 is($mset->size, 0, 'did not attempt to index unindexed v2 message');
267 ok(run_script([qw(-index --all)]), 'indexed v1 and v2 inboxes');
269 isnt($v1ibx->mm->last_commit, $last_v1_commit, '-index v1 worked');
270 isnt($v2ibx->mm->last_commit_xap($schema_version, 0),
271 $last_v2_commit, '-index v2 worked');
272 ok(run_script([qw(-extindex --all), "$home/extindex"]),
276 $mset = $es->mset('mid:199707281508.AAA24167@hoyogw.example');
277 is($mset->size, 1, 'got v1 message');
278 $mset = $es->mset('mid:multipart-html-sucks@11');
279 is($mset->size, 1, 'got v2 message');
282 if ('reindex catches missed messages') {
283 my $v2ibx = $cfg->lookup_name('v2test');
284 $v2ibx->{-no_fsync} = 1;
285 my $im = PublicInbox::InboxWritable->new($v2ibx)->importer(0);
286 my $cmt_a = $v2ibx->mm->last_commit_xap($schema_version, 0);
287 my $eml = eml_load('t/data/0001.patch');
290 my $cmt_b = $v2ibx->mm->last_commit_xap($schema_version, 0);
291 isnt($cmt_a, $cmt_b, 'v2 0.git HEAD updated');
293 my $uv = $v2ibx->uidvalidity;
294 my $lc_key = "lc-v2:v2.example//$uv;0";
295 is($oidx->eidx_meta($lc_key, $cmt_b), $cmt_a,
296 'update lc-v2 meta, old is as expected');
297 my $max = $oidx->max;
299 ok(run_script([qw(-extindex), "$home/extindex", $v2ibx->{inboxdir}]),
301 is($oidx->max, $max, '->max unchanged');
302 is($oidx->eidx_meta($lc_key), $cmt_b, 'lc-v2 unchanged');
304 my $opt = { 2 => \(my $err = '') };
305 ok(run_script([qw(-extindex --reindex), "$home/extindex",
306 $v2ibx->{inboxdir}], undef, $opt),
307 '--reindex for unseen');
308 is($oidx->max, $max + 1, '->max bumped');
309 is($oidx->eidx_meta($lc_key), $cmt_b, 'lc-v2 stays unchanged');
310 my @err = split(/^/, $err);
311 is(scalar(@err), 1, 'only one warning') or diag "err=$err";
312 like($err[0], qr/# reindex_unseen/, 'got reindex_unseen message');
313 my $new = $oidx->get_art($max + 1);
314 is($new->{subject}, $eml->header('Subject'), 'new message added');
317 # git patch-id --stable <t/data/0001.patch | awk '{print $1}'
318 my $patchid = '91ee6b761fc7f47cad9f2b09b10489f313eb5b71';
319 my $mset = $es->search->mset("patchid:$patchid");
320 is($mset->size, 1, 'patchid search works');
322 $mset = $es->mset("mid:$new->{mid}");
323 is($mset->size, 1, 'previously unseen, now indexed in Xapian');
325 ok($im->remove($eml), 'remove new message from v2 inbox');
327 my $cmt_c = $v2ibx->mm->last_commit_xap($schema_version, 0);
328 is($oidx->eidx_meta($lc_key, $cmt_c), $cmt_b,
329 'bump lc-v2 meta again to skip v2 remove');
332 ok(run_script([qw(-extindex --reindex), "$home/extindex",
333 $v2ibx->{inboxdir}], undef, $opt),
334 '--reindex for stale');
335 @err = split(/^/, $err);
336 is(scalar(@err), 1, 'only one warning') or diag "err=$err";
337 like($err[0], qr/\(#$new->{num}\): stale/, 'got stale message warning');
338 is($oidx->get_art($new->{num}), undef,
339 'stale message gone from over');
340 is_deeply($oidx->get_xref3($new->{num}), [],
341 'stale message has no xref3');
343 $mset = $es->mset("mid:$new->{mid}");
344 is($mset->size, 0, 'stale mid gone Xapian');
346 ok(run_script([qw(-extindex --reindex --all --fast), "$home/extindex"],
347 undef, $opt), '--reindex w/ --fast');
348 ok(!run_script([qw(-extindex --all --fast), "$home/extindex"],
349 undef, $opt), '--fast alone makes no sense');
352 if ('reindex catches content bifurcation') {
353 use PublicInbox::MID qw(mids);
354 my $v2ibx = $cfg->lookup_name('v2test');
355 $v2ibx->{-no_fsync} = 1;
356 my $im = PublicInbox::InboxWritable->new($v2ibx)->importer(0);
357 my $eml = eml_load('t/data/message_embed.eml');
358 my $cmt_a = $v2ibx->mm->last_commit_xap($schema_version, 0);
361 my $cmt_b = $v2ibx->mm->last_commit_xap($schema_version, 0);
362 my $uv = $v2ibx->uidvalidity;
363 my $lc_key = "lc-v2:v2.example//$uv;0";
365 is($oidx->eidx_meta($lc_key, $cmt_b), $cmt_a,
366 'update lc-v2 meta, old is as expected');
367 my $mid = mids($eml)->[0];
368 my $smsg = $v2ibx->over->next_by_mid($mid, \(my $id), \(my $prev));
369 my $oldmax = $oidx->max;
370 my $x3_orig = $oidx->get_xref3(3);
371 is(scalar(@$x3_orig), 1, '#3 has one xref');
372 $oidx->add_xref3(3, $smsg->{num}, $smsg->{blob}, 'v2.example');
373 my $x3 = $oidx->get_xref3(3);
374 is(scalar(@$x3), 2, 'injected xref3');
376 my $opt = { 2 => \(my $err = '') };
377 ok(run_script([qw(-extindex --all), "$home/extindex"], undef, $opt),
378 'extindex --all is noop');
379 is($err, '', 'no warnings in index');
381 is($oidx->max, $oldmax, 'oidx->max unchanged');
383 ok(run_script([qw(-extindex --reindex --all), "$home/extindex"],
384 undef, $opt), 'extindex --reindex') or diag explain($opt);
386 ok($oidx->max > $oldmax, 'oidx->max bumped');
387 like($err, qr/split into 2 due to deduplication change/,
388 'bifurcation noted');
389 my $added = $oidx->get_art($oidx->max);
390 is($added->{blob}, $smsg->{blob}, 'new blob indexed');
391 is_deeply(["v2.example:$smsg->{num}:$smsg->{blob}"],
392 $oidx->get_xref3($added->{num}),
393 'xref3 corrected for bifurcated message');
394 is_deeply($oidx->get_xref3(3), $x3_orig, 'xref3 restored for #3');
397 if ('--reindex --rethread') {
398 my $before = $oidx->dbh->selectrow_array(<<'');
399 SELECT MAX(tid) FROM over WHERE num > 0
402 ok(run_script([qw(-extindex --reindex --rethread --all),
403 "$home/extindex"], undef, $opt),
405 my $after = $oidx->dbh->selectrow_array(<<'');
406 SELECT MIN(tid) FROM over WHERE num > 0
408 # actual rethread logic is identical to v1/v2 and tested elsewhere
409 ok($after > $before, '--rethread updates MIN(tid)');
412 if ('remove v1test and test gc') {
413 xsys([qw(git config --unset publicinbox.v1test.inboxdir)],
414 { GIT_CONFIG => $cfg_path });
415 my $opt = { 2 => \(my $err = '') };
416 ok(run_script([qw(-extindex --gc), "$home/extindex"], undef, $opt),
418 like($err, qr/^# remove #1 v1\.example /ms, 'removed v1 message');
419 is(scalar(grep(!/^#/, split(/^/m, $err))), 0,
420 'no non-informational messages');
421 $misc->{xdb}->reopen;
422 @it = $misc->mset('')->items;
423 is(scalar(@it), 1, 'only one inbox left');
426 if ('dedupe + dry-run') {
427 my @cmd = ('-extindex', "$home/extindex");
428 my $opt = { 2 => \(my $err = '') };
429 ok(run_script([@cmd, '--dedupe'], undef, $opt), '--dedupe');
430 ok(run_script([@cmd, qw(--dedupe --dry-run)], undef, $opt),
431 '--dry-run --dedupe');
432 is $err, '', 'no errors';
433 ok(!run_script([@cmd, qw(--dry-run)], undef, $opt),
434 '--dry-run alone fails');
437 # chmod 0755, $home or xbail "chmod: $!";
438 for my $j (1, 3, 6) {
439 my $o = { 2 => \(my $err = '') };
440 my $d = "$home/extindex-j$j";
441 ok(run_script(['-extindex', "-j$j", '--all', $d], undef, $o),
444 $max = 0 if $max < 0;
445 my @dirs = glob("$d/ei*/?");
446 like($dirs[-1], qr!/ei[0-9]+/$max\z!, '-j works');
450 my $d = "$home/extindex-j1";
451 my $es = PublicInbox::ExtSearch->new($d);
452 ok(my $nresult0 = $es->mset('z:0..')->size, 'got results');
453 ok(ref($es->{xdb}), '{xdb} created');
454 my $nshards1 = $es->{nshard};
455 is($nshards1, 1, 'correct shard count');
457 my @ei_dir = glob("$d/ei*/");
458 chmod 0755, $ei_dir[0] or xbail "chmod: $!";
459 my $mode = sprintf('%04o', 07777 & (stat($ei_dir[0]))[2]);
460 is($mode, '0755', 'mode set on ei*/ dir');
461 my $o = { 2 => \(my $err = '') };
462 ok(run_script([qw(-xcpdb -R4), $d]), 'xcpdb R4');
463 my @dirs = glob("$d/ei*/?");
465 is(grep(m!/ei[0-9]+/$i\z!, @dirs), 1, "shard [$i] created");
466 my $m = sprintf('%04o', 07777 & (stat($dirs[$i]))[2]);
467 is($m, $mode, "shard [$i] mode");
469 delete @$es{qw(xdb qp)};
470 is($es->mset('z:0..')->size, $nresult0, 'new shards, same results');
473 is(grep(m!/ei[0-9]+/$i\z!, @dirs), 0, "no shard [$i]");
476 ok(run_script([qw(-xcpdb -R2), $d]), 'xcpdb -R2');
477 @dirs = glob("$d/ei*/?");
479 is(grep(m!/ei[0-9]+/$i\z!, @dirs), 1, "shard [$i] kept");
482 is(grep(m!/ei[0-9]+/$i\z!, @dirs), 0, "no shard [$i]");
484 skip 'xapian-compact missing', 4 unless have_xapian_compact;
485 ok(run_script([qw(-compact), $d], undef, $o), 'compact');
486 # n.b. stderr contains xapian-compact output
488 my @d2 = glob("$d/ei*/?");
489 is_deeply(\@d2, \@dirs, 'dirs consistent after compact');
490 ok(run_script([qw(-extindex --dedupe --all), $d]),
491 '--dedupe works after compact');
492 ok(run_script([qw(-extindex --gc), $d], undef, $o),
493 '--gc works after compact');
496 { # ensure --gc removes non-xposted messages
497 my $old_size = -s $cfg_path // xbail "stat $cfg_path $!";
498 my $tmp_addr = 'v2tmp@example.com';
499 run_script([qw(-init v2tmp --indexlevel basic
500 --newsgroup v2tmp.example),
501 "$home/v2tmp", 'http://example.com/v2tmp', $tmp_addr ])
503 $env = { ORIGINAL_RECIPIENT => $tmp_addr };
504 open $fh, '+>', undef or xbail "open $!";
506 my $mid = 'tmpmsg@example.com';
507 print $fh <<EOM or xbail "print $!";
512 Date: Tue, 19 Jan 2038 03:14:07 +0000
515 seek $fh, 0, SEEK_SET or xbail "seek $!";
516 run_script([qw(-mda --no-precheck)], $env, {0 => $fh}) or xbail '-mda';
517 ok(run_script([qw(-extindex --all), "$home/extindex"]), 'update');
520 my $es = PublicInbox::ExtSearch->new("$home/extindex");
522 my $smsg = $es->over->next_by_mid($mid, \$id, \$prv);
523 ok($smsg, 'tmpmsg indexed');
524 my $mset = $es->search->mset("mid:$mid");
525 is($mset->size, 1, 'new message found');
526 $mset = $es->search->mset('z:0..');
529 truncate($cfg_path, $old_size) or xbail "truncate $!";
530 my $rdr = { 2 => \(my $err) };
531 ok(run_script([qw(-extindex --gc), "$home/extindex"], undef, $rdr),
532 'gc to get rid of removed inbox');
533 is_deeply([ grep(!/^(?:I:|#)/, split(/^/m, $err)) ], [],
534 'no non-informational errors in stderr');
536 my $es = PublicInbox::ExtSearch->new("$home/extindex");
537 my $mset = $es->search->mset("mid:$mid");
538 is($mset->size, 0, 'tmpmsg gone from search');
540 is($es->over->next_by_mid($mid, \$id, \$prv), undef,
541 'tmpmsg gone from over');
543 is($es->over->next_by_mid('testmessage@example.com', \$id, \$prv),
544 undef, 'remaining message not indavderover');
545 $mset = $es->search->mset('z:0..');
546 is($mset->size, $nr - 1, 'existing messages not clobbered from search');
547 my $o = $es->over->{dbh}->selectall_arrayref(<<EOM);
548 SELECT num FROM over ORDER BY num
550 is(scalar(@$o), $mset->size, 'over row count matches Xapian');
551 my $x = $es->over->{dbh}->selectall_arrayref(<<EOM);
552 SELECT DISTINCT(docid) FROM xref3 ORDER BY docid
554 is_deeply($x, $o, 'xref3 and over docids match');