2 # Copyright (C) 2020-2021 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 $sock = tcp_server();
17 my $host_port = $sock->sockhost . ':' . $sock->sockport;
18 my ($home, $for_destroy) = tmpdir();
19 local $ENV{HOME} = $home;
20 mkdir "$home/.public-inbox" or BAIL_OUT $!;
21 my $cfg_path = "$home/.public-inbox/config";
22 open my $fh, '>', $cfg_path or BAIL_OUT $!;
23 print $fh <<EOF or BAIL_OUT $!;
27 close $fh or BAIL_OUT $!;
28 my $v2addr = 'v2test@example.com';
29 my $v1addr = 'v1test@example.com';
30 ok(run_script([qw(-init -Lbasic -V2 v2test --newsgroup v2.example),
31 "$home/v2test", 'http://example.com/v2test', $v2addr ]), 'v2test init');
32 my $env = { ORIGINAL_RECIPIENT => $v2addr };
33 my $eml = eml_load('t/utf8.eml');
35 $eml->header_set('List-Id', '<v2.example.com>');
36 open($fh, '+>', undef) or BAIL_OUT $!;
38 print $fh $eml->as_string or BAIL_OUT $!;
39 seek($fh, 0, SEEK_SET) or BAIL_OUT $!;
41 run_script(['-mda', '--no-precheck'], $env, { 0 => $fh }) or BAIL_OUT '-mda';
43 ok(run_script([qw(-init -V1 v1test --newsgroup v1.example), "$home/v1test",
44 'http://example.com/v1test', $v1addr ]), 'v1test init');
46 $eml->header_set('List-Id', '<v1.example.com>');
47 seek($fh, 0, SEEK_SET) or BAIL_OUT $!;
48 truncate($fh, 0) or BAIL_OUT $!;
49 print $fh $eml->as_string or BAIL_OUT $!;
50 seek($fh, 0, SEEK_SET) or BAIL_OUT $!;
52 $env = { ORIGINAL_RECIPIENT => $v1addr };
53 run_script(['-mda', '--no-precheck'], $env, { 0 => $fh }) or BAIL_OUT '-mda';
55 run_script([qw(-index -Lbasic), "$home/v1test"]) or BAIL_OUT "index $?";
57 ok(run_script([qw(-extindex --all), "$home/extindex"]), 'extindex init');
59 my $es = PublicInbox::ExtSearch->new("$home/extindex");
60 ok($es->has_threadid, '->has_threadid');
63 { # TODO: -extindex should write this to config
64 open $fh, '>>', $cfg_path or BAIL_OUT $!;
65 print $fh <<EOF or BAIL_OUT $!;
68 topdir = $home/extindex
70 close $fh or BAIL_OUT $!;
72 my $pi_cfg = PublicInbox::Config->new;
74 ok($pi_cfg->ALL, '->ALL');
75 my $ibx = $pi_cfg->{-by_newsgroup}->{'v2.example'};
76 my $ret = $pi_cfg->ALL->nntp_xref_for($ibx, $ibx->over->get_art(1));
77 is_deeply($ret, { 'v1.example' => 1, 'v2.example' => 1 },
82 require_mods(qw(Net::NNTP), 1);
83 my ($out, $err) = ("$home/nntpd.out.log", "$home/nntpd.err.log");
84 my $cmd = [ '-nntpd', '-W0', "--stdout=$out", "--stderr=$err" ];
85 my $td = start_script($cmd, undef, { 3 => $sock });
86 my $n = Net::NNTP->new($host_port);
87 my @xp = $n->xpath('<testmessage@example.com>');
88 is_deeply(\@xp, [ qw(v1.example/1 v2.example/1) ]);
89 $n->group('v1.example');
90 my $res = $n->head(1);
91 @$res = grep(/^Xref: /, @$res);
92 like($res->[0], qr/ v1\.example:1 v2\.example:1/, 'nntp_xref works');
95 my $es = PublicInbox::ExtSearch->new("$home/extindex");
97 my $smsg = $es->over->get_art(1);
98 ok($smsg, 'got first article');
99 is($es->over->get_art(2), undef, 'only one added');
100 my $xref3 = $es->over->get_xref3(1);
101 like($xref3->[0], qr/\A\Qv2.example\E:1:/, 'order preserved 1');
102 like($xref3->[1], qr/\A\Qv1.example\E:1:/, 'order preserved 2');
103 is(scalar(@$xref3), 2, 'only to entries');
106 if ('inbox edited') {
107 my ($in, $out, $err);
108 $in = $out = $err = '';
109 my $opt = { 0 => \$in, 1 => \$out, 2 => \$err };
110 my $env = { MAIL_EDITOR => "$^X -i -p -e 's/test message/BEST MSG/'" };
111 my $cmd = [ qw(-edit -Ft/utf8.eml), "$home/v2test" ];
112 ok(run_script($cmd, $env, $opt), '-edit');
113 ok(run_script([qw(-extindex --all), "$home/extindex"], undef, $opt),
115 like($err, qr/discontiguous range/, 'warned about discontiguous range');
116 my $msg1 = $es->over->get_art(1) or BAIL_OUT 'msg1 missing';
117 my $msg2 = $es->over->get_art(2) or BAIL_OUT 'msg2 missing';
118 is($msg1->{mid}, $msg2->{mid}, 'edited message indexed');
119 isnt($msg1->{blob}, $msg2->{blob}, 'blobs differ');
120 my $eml2 = $es->smsg_eml($msg2);
121 like($eml2->body, qr/BEST MSG/, 'edited body in #2');
122 unlike($eml2->body, qr/test message/, 'old body discarded in #2');
123 my $eml1 = $es->smsg_eml($msg1);
124 like($eml1->body, qr/test message/, 'original body in #1');
125 my $x1 = $es->over->get_xref3(1);
126 my $x2 = $es->over->get_xref3(2);
127 is(scalar(@$x1), 1, 'original only has one xref3');
128 is(scalar(@$x2), 1, 'new message has one xref3');
129 isnt($x1->[0], $x2->[0], 'xref3 differs');
131 my $mset = $es->mset('b:"BEST MSG"');
132 is($mset->size, 1, 'new message found');
133 $mset = $es->mset('b:"test message"');
134 is($mset->size, 1, 'old message found');
135 delete @$es{qw(git over xdb)}; # fork preparation
137 my $pi_cfg = PublicInbox::Config->new;
139 is(scalar($pi_cfg->ALL->mset('s:Testing')->items), 2,
140 '2 results in ->ALL');
143 $pi_cfg->each_inbox(sub {
146 local $SIG{__WARN__} = sub {}; # FIXME support --reindex
147 my $mset = $ibx->isrch->mset('s:Testing');
148 $res->{$ibx->eidx_key} = $ibx->isrch->mset_to_smsg($ibx, $mset);
150 is($nr, 2, 'two inboxes');
152 for my $v (qw(v1 v2)) {
153 my $ibx = $pi_cfg->lookup_newsgroup("$v.example");
154 my $smsg = $ibx->over->get_art(1);
156 $exp->{"$v.example"} = [ $smsg ];
158 is_deeply($res, $exp, 'isearch limited results');
159 $pi_cfg = $res = $exp = undef;
161 open my $rmfh, '+>', undef or BAIL_OUT $!;
163 print $rmfh $eml2->as_string or BAIL_OUT $!;
164 seek($rmfh, 0, SEEK_SET) or BAIL_OUT $!;
166 ok(run_script([qw(-learn rm --all)], undef, $opt), '-learn rm');
168 ok(run_script([qw(-extindex --all), "$home/extindex"], undef, undef),
169 'extindex after rm');
170 is($es->over->get_art(2), undef, 'doc #2 gone');
171 $mset = $es->mset('b:"BEST MSG"');
172 is($mset->size, 0, 'new message gone');
175 my $misc = $es->misc;
176 my @it = $misc->mset('')->items;
177 is(scalar(@it), 2, 'two inboxes');
178 like($it[0]->get_document->get_data, qr/v2test/, 'docdata matched v2');
179 like($it[1]->get_document->get_data, qr/v1test/, 'docdata matched v1');
181 my $cfg = PublicInbox::Config->new;
182 my $schema_version = PublicInbox::Search::SCHEMA_VERSION();
183 my $f = "$home/extindex/ei$schema_version/over.sqlite3";
184 my $oidx = PublicInbox::OverIdx->new($f);
185 if ('inject w/o indexing') {
186 use PublicInbox::Import;
187 my $v1ibx = $cfg->lookup_name('v1test');
188 my $last_v1_commit = $v1ibx->mm->last_commit;
189 my $v2ibx = $cfg->lookup_name('v2test');
190 my $last_v2_commit = $v2ibx->mm->last_commit_xap($schema_version, 0);
191 my $git0 = PublicInbox::Git->new("$v2ibx->{inboxdir}/git/0.git");
192 chomp(my $cmt = $git0->qx(qw(rev-parse HEAD^0)));
193 is($last_v2_commit, $cmt, 'v2 index up-to-date');
195 my $v2im = PublicInbox::Import->new($git0, undef, undef, $v2ibx);
196 $v2im->{lock_path} = undef;
197 $v2im->{path_type} = 'v2';
198 $v2im->add(eml_load('t/mda-mime.eml'));
200 chomp(my $tip = $git0->qx(qw(rev-parse HEAD^0)));
201 isnt($tip, $cmt, '0.git v2 updated');
203 # inject a message w/o updating index
204 rename("$home/v1test/public-inbox", "$home/v1test/skip-index") or
206 open(my $eh, '<', 't/iso-2202-jp.eml') or BAIL_OUT $!;
207 run_script(['-mda', '--no-precheck'], $env, { 0 => $eh}) or
209 rename("$home/v1test/skip-index", "$home/v1test/public-inbox") or
212 my ($in, $out, $err);
213 $in = $out = $err = '';
214 my $opt = { 0 => \$in, 1 => \$out, 2 => \$err };
215 ok(run_script([qw(-extindex -v -v --all), "$home/extindex"],
216 undef, undef), 'extindex noop');
218 my $mset = $es->mset('mid:199707281508.AAA24167@hoyogw.example');
219 is($mset->size, 0, 'did not attempt to index unindexed v1 message');
220 $mset = $es->mset('mid:multipart-html-sucks@11');
221 is($mset->size, 0, 'did not attempt to index unindexed v2 message');
222 ok(run_script([qw(-index --all)]), 'indexed v1 and v2 inboxes');
224 isnt($v1ibx->mm->last_commit, $last_v1_commit, '-index v1 worked');
225 isnt($v2ibx->mm->last_commit_xap($schema_version, 0),
226 $last_v2_commit, '-index v2 worked');
227 ok(run_script([qw(-extindex --all), "$home/extindex"]),
231 $mset = $es->mset('mid:199707281508.AAA24167@hoyogw.example');
232 is($mset->size, 1, 'got v1 message');
233 $mset = $es->mset('mid:multipart-html-sucks@11');
234 is($mset->size, 1, 'got v2 message');
237 if ('reindex catches missed messages') {
238 my $v2ibx = $cfg->lookup_name('v2test');
239 my $im = PublicInbox::InboxWritable->new($v2ibx)->importer(0);
240 my $cmt_a = $v2ibx->mm->last_commit_xap($schema_version, 0);
241 my $eml = eml_load('t/data/0001.patch');
244 my $cmt_b = $v2ibx->mm->last_commit_xap($schema_version, 0);
245 isnt($cmt_a, $cmt_b, 'v2 0.git HEAD updated');
247 my $uv = $v2ibx->uidvalidity;
248 my $lc_key = "lc-v2:v2.example//$uv;0";
249 is($oidx->eidx_meta($lc_key, $cmt_b), $cmt_a,
250 'update lc-v2 meta, old is as expected');
251 my $max = $oidx->max;
253 ok(run_script([qw(-extindex), "$home/extindex", $v2ibx->{inboxdir}]),
255 is($oidx->max, $max, '->max unchanged');
256 is($oidx->eidx_meta($lc_key), $cmt_b, 'lc-v2 unchanged');
258 my $opt = { 2 => \(my $err = '') };
259 ok(run_script([qw(-extindex --reindex), "$home/extindex",
260 $v2ibx->{inboxdir}], undef, $opt),
261 '--reindex for unseen');
262 is($oidx->max, $max + 1, '->max bumped');
263 is($oidx->eidx_meta($lc_key), $cmt_b, 'lc-v2 stays unchanged');
264 my @err = split(/^/, $err);
265 is(scalar(@err), 1, 'only one warning') or diag "err=$err";
266 like($err[0], qr/I: reindex_unseen/, 'got reindex_unseen message');
267 my $new = $oidx->get_art($max + 1);
268 is($new->{subject}, $eml->header('Subject'), 'new message added');
271 my $mset = $es->mset("mid:$new->{mid}");
272 is($mset->size, 1, 'previously unseen, now indexed in Xapian');
274 ok($im->remove($eml), 'remove new message from v2 inbox');
276 my $cmt_c = $v2ibx->mm->last_commit_xap($schema_version, 0);
277 is($oidx->eidx_meta($lc_key, $cmt_c), $cmt_b,
278 'bump lc-v2 meta again to skip v2 remove');
281 ok(run_script([qw(-extindex --reindex), "$home/extindex",
282 $v2ibx->{inboxdir}], undef, $opt),
283 '--reindex for stale');
284 @err = split(/^/, $err);
285 is(scalar(@err), 1, 'only one warning') or diag "err=$err";
286 like($err[0], qr/\(#$new->{num}\): stale/, 'got stale message warning');
287 is($oidx->get_art($new->{num}), undef,
288 'stale message gone from over');
289 is_deeply($oidx->get_xref3($new->{num}), [],
290 'stale message has no xref3');
292 $mset = $es->mset("mid:$new->{mid}");
293 is($mset->size, 0, 'stale mid gone Xapian');
296 if ('reindex catches content bifurcation') {
297 use PublicInbox::MID qw(mids);
298 my $v2ibx = $cfg->lookup_name('v2test');
299 my $im = PublicInbox::InboxWritable->new($v2ibx)->importer(0);
300 my $eml = eml_load('t/data/message_embed.eml');
301 my $cmt_a = $v2ibx->mm->last_commit_xap($schema_version, 0);
304 my $cmt_b = $v2ibx->mm->last_commit_xap($schema_version, 0);
305 my $uv = $v2ibx->uidvalidity;
306 my $lc_key = "lc-v2:v2.example//$uv;0";
308 is($oidx->eidx_meta($lc_key, $cmt_b), $cmt_a,
309 'update lc-v2 meta, old is as expected');
310 my $mid = mids($eml)->[0];
311 my $smsg = $v2ibx->over->next_by_mid($mid, \(my $id), \(my $prev));
312 my $oldmax = $oidx->max;
313 my $x3_orig = $oidx->get_xref3(3);
314 is(scalar(@$x3_orig), 1, '#3 has one xref');
315 $oidx->add_xref3(3, $smsg->{num}, $smsg->{blob}, 'v2.example');
316 my $x3 = $oidx->get_xref3(3);
317 is(scalar(@$x3), 2, 'injected xref3');
319 my $opt = { 2 => \(my $err = '') };
320 ok(run_script([qw(-extindex --all), "$home/extindex"], undef, $opt),
321 'extindex --all is noop');
322 is($err, '', 'no warnings in index');
324 is($oidx->max, $oldmax, 'oidx->max unchanged');
326 ok(run_script([qw(-extindex --reindex --all), "$home/extindex"],
327 undef, $opt), 'extindex --reindex');
329 ok($oidx->max > $oldmax, 'oidx->max bumped');
330 like($err, qr/split into 2 due to deduplication change/,
331 'bifurcation noted');
332 my $added = $oidx->get_art($oidx->max);
333 is($added->{blob}, $smsg->{blob}, 'new blob indexed');
334 is_deeply(["v2.example:$smsg->{num}:$smsg->{blob}"],
335 $oidx->get_xref3($added->{num}),
336 'xref3 corrected for bifurcated message');
337 is_deeply($oidx->get_xref3(3), $x3_orig, 'xref3 restored for #3');
340 if ('--reindex --rethread') {
341 my $before = $oidx->dbh->selectrow_array(<<'');
342 SELECT MAX(tid) FROM over WHERE num > 0
345 ok(run_script([qw(-extindex --reindex --rethread --all),
346 "$home/extindex"], undef, $opt),
348 my $after = $oidx->dbh->selectrow_array(<<'');
349 SELECT MIN(tid) FROM over WHERE num > 0
351 # actual rethread logic is identical to v1/v2 and tested elsewhere
352 ok($after > $before, '--rethread updates MIN(tid)');
355 if ('remove v1test and test gc') {
356 xsys([qw(git config --unset publicinbox.v1test.inboxdir)],
357 { GIT_CONFIG => $cfg_path });
358 my $opt = { 2 => \(my $err = '') };
359 ok(run_script([qw(-extindex --gc), "$home/extindex"], undef, $opt),
361 like($err, qr/^I: remove #1 v1\.example /ms, 'removed v1 message');
362 is(scalar(grep(!/^I:/, split(/^/m, $err))), 0,
363 'no non-informational messages');
364 $misc->{xdb}->reopen;
365 @it = $misc->mset('')->items;
366 is(scalar(@it), 1, 'only one inbox left');