1 # Copyright (C) 2018 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
7 use PublicInbox::ContentId qw(content_digest);
8 use File::Temp qw/tempdir/;
9 use File::Path qw(remove_tree);
10 require './t/common.perl';
13 foreach my $mod (qw(DBD::SQLite Search::Xapian)) {
15 plan skip_all => "$mod missing for v2reindex.t" if $@;
17 use_ok 'PublicInbox::V2Writable';
18 my $mainrepo = tempdir('pi-v2reindex-XXXXXX', TMPDIR => 1, CLEANUP => 1);
20 mainrepo => $mainrepo,
21 name => 'test-v2writable',
23 -primary_address => 'test@example.com',
27 open my $fh, '<', 'COPYING' or die "can't open COPYING: $!";
31 $agpl or die "AGPL or die :P\n";
32 my $phrase = q("defending all users' freedom");
33 my $mime = PublicInbox::MIME->create(
35 From => 'a@example.com',
36 To => 'test@example.com',
37 Subject => 'this is a subject',
38 Date => 'Fri, 02 Oct 1993 00:00:00 +0000',
44 my ($mark1, $mark2, $mark3, $mark4);
46 my %config = %$ibx_config;
47 my $ibx = PublicInbox::Inbox->new(\%config);
48 my $im = PublicInbox::V2Writable->new($ibx, {nproc => 1});
49 my $im0 = $im->importer();
50 foreach my $i (1..10) {
51 $mime->header_set('Message-Id', "<$i\@example.com>");
52 ok($im->add($mime), "message $i added");
54 $mark1 = $im0->get_mark($im0->{tip});
56 $mark2 = $im0->get_mark($im0->{tip});
60 if ('test remove later') {
61 $mark3 = $im0->get_mark($im0->{tip});
62 $mime->header_set('Message-Id', "<5\@example.com>");
64 $mark4 = $im0->get_mark($im0->{tip});
68 $minmax = [ $ibx->mm->minmax ];
69 ok(defined $minmax->[0] && defined $minmax->[1], 'minmax defined');
70 is_deeply($minmax, [ 1, 10 ], 'minmax as expected');
71 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
73 my ($min, $max) = @$minmax;
74 $msgmap = $ibx->mm->msg_range(\$min, $max);
76 [1, '1@example.com' ],
77 [2, '2@example.com' ],
78 [3, '3@example.com' ],
79 [6, '6@example.com' ],
80 [7, '7@example.com' ],
81 [8, '8@example.com' ],
82 [9, '9@example.com' ],
83 [10, '10@example.com' ],
84 ], 'msgmap as expected');
88 my %config = %$ibx_config;
89 my $ibx = PublicInbox::Inbox->new(\%config);
90 my $im = PublicInbox::V2Writable->new($ibx, 1);
91 eval { $im->index_sync({reindex => 1}) };
92 is($@, '', 'no error from reindexing');
96 is_deeply([ $ibx->mm->minmax ], $minmax, 'minmax unchanged');
97 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
99 my ($min, $max) = $ibx->mm->minmax;
100 is_deeply($ibx->mm->msg_range(\$min, $max), $msgmap, 'msgmap unchanged');
103 my $xap = "$mainrepo/xap".PublicInbox::Search::SCHEMA_VERSION();
105 ok(!-d $xap, 'Xapian directories removed');
107 my %config = %$ibx_config;
108 my $ibx = PublicInbox::Inbox->new(\%config);
109 my $im = PublicInbox::V2Writable->new($ibx, 1);
110 eval { $im->index_sync({reindex => 1}) };
111 is($@, '', 'no error from reindexing');
113 ok(-d $xap, 'Xapian directories recreated');
116 is_deeply([ $ibx->mm->minmax ], $minmax, 'minmax unchanged');
117 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
119 my ($min, $max) = $ibx->mm->minmax;
120 is_deeply($ibx->mm->msg_range(\$min, $max), $msgmap, 'msgmap unchanged');
123 ok(unlink "$mainrepo/msgmap.sqlite3", 'remove msgmap');
125 ok(!-d $xap, 'Xapian directories removed again');
128 local $SIG{__WARN__} = sub { push @warn, @_ };
129 my %config = %$ibx_config;
130 my $ibx = PublicInbox::Inbox->new(\%config);
131 my $im = PublicInbox::V2Writable->new($ibx, 1);
132 eval { $im->index_sync({reindex => 1}) };
133 is($@, '', 'no error from reindexing without msgmap');
134 is(scalar(@warn), 0, 'no warnings from reindexing');
136 ok(-d $xap, 'Xapian directories recreated');
138 is_deeply([ $ibx->mm->minmax ], $minmax, 'minmax unchanged');
139 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
141 my ($min, $max) = $ibx->mm->minmax;
142 is_deeply($ibx->mm->msg_range(\$min, $max), $msgmap, 'msgmap unchanged');
146 ok(unlink "$mainrepo/msgmap.sqlite3", 'remove msgmap');
148 ok(!-d $xap, 'Xapian directories removed again');
151 local $SIG{__WARN__} = sub { push @warn, @_ };
152 my %config = %$ibx_config;
153 my $ibx = PublicInbox::Inbox->new(\%config);
154 my $im = PublicInbox::V2Writable->new($ibx, 1);
155 eval { $im->index_sync({reindex => 1}) };
156 is($@, '', 'no error from reindexing without msgmap');
157 is_deeply(\@warn, [], 'no warnings');
159 ok(-d $xap, 'Xapian directories recreated');
161 is_deeply([ $ibx->mm->minmax ], $minmax, 'minmax unchanged');
162 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
163 my $mset = $ibx->search->query($phrase, {mset=>1});
164 isnt($mset->size, 0, "phrase search succeeds on indexlevel=full");
165 for (<"$xap/*/*">) { $sizes{$ibx->{indexlevel}} += -s _ if -f $_ }
167 my ($min, $max) = $ibx->mm->minmax;
168 is_deeply($ibx->mm->msg_range(\$min, $max), $msgmap, 'msgmap unchanged');
171 ok(unlink "$mainrepo/msgmap.sqlite3", 'remove msgmap');
173 ok(!-d $xap, 'Xapian directories removed again');
176 local $SIG{__WARN__} = sub { push @warn, @_ };
177 my %config = %$ibx_config;
178 $config{indexlevel} = 'medium';
179 my $ibx = PublicInbox::Inbox->new(\%config);
180 my $im = PublicInbox::V2Writable->new($ibx);
181 eval { $im->index_sync({reindex => 1}) };
182 is($@, '', 'no error from reindexing without msgmap');
183 is_deeply(\@warn, [], 'no warnings');
185 ok(-d $xap, 'Xapian directories recreated');
187 is_deeply([ $ibx->mm->minmax ], $minmax, 'minmax unchanged');
188 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
191 # not sure why, but Xapian seems to fallback to terms and
192 # phrase searches still work
193 delete $ibx->{search};
194 my $mset = $ibx->search->query($phrase, {mset=>1});
195 is($mset->size, 0, 'phrase search does not work on medium');
199 my $mset = $ibx->search->query($words, {mset=>1});
200 isnt($mset->size, 0, "normal search works on indexlevel=medium");
201 for (<"$xap/*/*">) { $sizes{$ibx->{indexlevel}} += -s _ if -f $_ }
203 ok($sizes{full} > $sizes{medium}, 'medium is smaller than full');
206 my ($min, $max) = $ibx->mm->minmax;
207 is_deeply($ibx->mm->msg_range(\$min, $max), $msgmap, 'msgmap unchanged');
210 ok(unlink "$mainrepo/msgmap.sqlite3", 'remove msgmap');
212 ok(!-d $xap, 'Xapian directories removed again');
215 local $SIG{__WARN__} = sub { push @warn, @_ };
216 my %config = %$ibx_config;
217 $config{indexlevel} = 'basic';
218 my $ibx = PublicInbox::Inbox->new(\%config);
219 my $im = PublicInbox::V2Writable->new($ibx);
220 eval { $im->index_sync({reindex => 1}) };
221 is($@, '', 'no error from reindexing without msgmap');
222 is_deeply(\@warn, [], 'no warnings');
224 ok(-d $xap, 'Xapian directories recreated');
226 is_deeply([ $ibx->mm->minmax ], $minmax, 'minmax unchanged');
227 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
228 my $mset = $ibx->search->query('freedom', {mset=>1});
229 is($mset->size, 0, "search fails on indexlevel='basic'");
230 for (<"$xap/*/*">) { $sizes{$ibx->{indexlevel}} += -s _ if -f $_ }
231 ok($sizes{medium} > $sizes{basic}, 'basic is smaller than medium');
233 my ($min, $max) = $ibx->mm->minmax;
234 is_deeply($ibx->mm->msg_range(\$min, $max), $msgmap, 'msgmap unchanged');
238 # An incremental indexing test
239 ok(unlink "$mainrepo/msgmap.sqlite3", 'remove msgmap');
241 ok(!-d $xap, 'Xapian directories removed again');
244 local $SIG{__WARN__} = sub { push @warn, @_ };
245 my %config = %$ibx_config;
246 my $ibx = PublicInbox::Inbox->new(\%config);
247 # mark1 4 simple additions in the same index_sync
248 $ibx->{ref_head} = $mark1;
249 my $im = PublicInbox::V2Writable->new($ibx);
250 eval { $im->index_sync() };
251 is($@, '', 'no error from reindexing without msgmap');
252 is_deeply(\@warn, [], 'no warnings');
254 my ($min, $max) = $ibx->mm->minmax;
255 is($min, 1, 'min as expected');
256 is($max, 4, 'max as expected');
257 is($ibx->mm->num_highwater, 4, 'num_highwater as expected');
258 is_deeply($ibx->mm->msg_range(\$min, $max),
260 [1, '1@example.com' ],
261 [2, '2@example.com' ],
262 [3, '3@example.com' ],
263 [4, '4@example.com' ],
264 ], 'msgmap as expected' );
268 local $SIG{__WARN__} = sub { push @warn, @_ };
269 my %config = %$ibx_config;
270 my $ibx = PublicInbox::Inbox->new(\%config);
271 # mark2 A delete separated from an add in the same index_sync
272 $ibx->{ref_head} = $mark2;
273 my $im = PublicInbox::V2Writable->new($ibx);
274 eval { $im->index_sync() };
275 is($@, '', 'no error from reindexing without msgmap');
276 is_deeply(\@warn, [], 'no warnings');
278 my ($min, $max) = $ibx->mm->minmax;
279 is($min, 1, 'min as expected');
280 is($max, 3, 'max as expected');
281 is($ibx->mm->num_highwater, 4, 'num_highwater as expected');
282 is_deeply($ibx->mm->msg_range(\$min, $max),
284 [1, '1@example.com' ],
285 [2, '2@example.com' ],
286 [3, '3@example.com' ],
287 ], 'msgmap as expected' );
291 local $SIG{__WARN__} = sub { push @warn, @_ };
292 my %config = %$ibx_config;
293 my $ibx = PublicInbox::Inbox->new(\%config);
294 # mark3 adds following the delete at mark2
295 $ibx->{ref_head} = $mark3;
296 my $im = PublicInbox::V2Writable->new($ibx);
297 eval { $im->index_sync() };
298 is($@, '', 'no error from reindexing without msgmap');
299 is_deeply(\@warn, [], 'no warnings');
301 my ($min, $max) = $ibx->mm->minmax;
302 is($min, 1, 'min as expected');
303 is($max, 10, 'max as expected');
304 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
305 is_deeply($ibx->mm->msg_range(\$min, $max),
307 [1, '1@example.com' ],
308 [2, '2@example.com' ],
309 [3, '3@example.com' ],
310 [5, '5@example.com' ],
311 [6, '6@example.com' ],
312 [7, '7@example.com' ],
313 [8, '8@example.com' ],
314 [9, '9@example.com' ],
315 [10, '10@example.com' ],
316 ], 'msgmap as expected' );
320 local $SIG{__WARN__} = sub { push @warn, @_ };
321 my %config = %$ibx_config;
322 my $ibx = PublicInbox::Inbox->new(\%config);
323 # mark4 A delete of an older message
324 $ibx->{ref_head} = $mark4;
325 my $im = PublicInbox::V2Writable->new($ibx);
326 eval { $im->index_sync() };
327 is($@, '', 'no error from reindexing without msgmap');
328 is_deeply(\@warn, [], 'no warnings');
330 my ($min, $max) = $ibx->mm->minmax;
331 is($min, 1, 'min as expected');
332 is($max, 10, 'max as expected');
333 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
334 is_deeply($ibx->mm->msg_range(\$min, $max),
336 [1, '1@example.com' ],
337 [2, '2@example.com' ],
338 [3, '3@example.com' ],
339 [6, '6@example.com' ],
340 [7, '7@example.com' ],
341 [8, '8@example.com' ],
342 [9, '9@example.com' ],
343 [10, '10@example.com' ],
344 ], 'msgmap as expected' );
348 # Another incremental indexing test
349 ok(unlink "$mainrepo/msgmap.sqlite3", 'remove msgmap');
351 ok(!-d $xap, 'Xapian directories removed again');
354 local $SIG{__WARN__} = sub { push @warn, @_ };
355 my %config = %$ibx_config;
356 my $ibx = PublicInbox::Inbox->new(\%config);
357 # mark2 an add and it's delete in the same index_sync
358 $ibx->{ref_head} = $mark2;
359 my $im = PublicInbox::V2Writable->new($ibx);
360 eval { $im->index_sync() };
361 is($@, '', 'no error from reindexing without msgmap');
362 is_deeply(\@warn, [], 'no warnings');
364 my ($min, $max) = $ibx->mm->minmax;
365 is($min, 1, 'min as expected');
366 is($max, 3, 'max as expected');
367 is($ibx->mm->num_highwater, 4, 'num_highwater as expected');
368 is_deeply($ibx->mm->msg_range(\$min, $max),
370 [1, '1@example.com' ],
371 [2, '2@example.com' ],
372 [3, '3@example.com' ],
373 ], 'msgmap as expected' );
377 local $SIG{__WARN__} = sub { push @warn, @_ };
378 my %config = %$ibx_config;
379 my $ibx = PublicInbox::Inbox->new(\%config);
380 # mark3 adds following the delete at mark2
381 $ibx->{ref_head} = $mark3;
382 my $im = PublicInbox::V2Writable->new($ibx);
383 eval { $im->index_sync() };
384 is($@, '', 'no error from reindexing without msgmap');
385 is_deeply(\@warn, [], 'no warnings');
387 my ($min, $max) = $ibx->mm->minmax;
388 is($min, 1, 'min as expected');
389 is($max, 10, 'max as expected');
390 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
391 is_deeply($ibx->mm->msg_range(\$min, $max),
393 [1, '1@example.com' ],
394 [2, '2@example.com' ],
395 [3, '3@example.com' ],
396 [5, '5@example.com' ],
397 [6, '6@example.com' ],
398 [7, '7@example.com' ],
399 [8, '8@example.com' ],
400 [9, '9@example.com' ],
401 [10, '10@example.com' ],
402 ], 'msgmap as expected' );
406 local $SIG{__WARN__} = sub { push @warn, @_ };
407 my %config = %$ibx_config;
408 my $ibx = PublicInbox::Inbox->new(\%config);
409 # mark4 A delete of an older message
410 $ibx->{ref_head} = $mark4;
411 my $im = PublicInbox::V2Writable->new($ibx);
412 eval { $im->index_sync() };
413 is($@, '', 'no error from reindexing without msgmap');
414 is_deeply(\@warn, [], 'no warnings');
416 my ($min, $max) = $ibx->mm->minmax;
417 is($min, 1, 'min as expected');
418 is($max, 10, 'max as expected');
419 is($ibx->mm->num_highwater, 10, 'num_highwater as expected');
420 is_deeply($ibx->mm->msg_range(\$min, $max),
422 [1, '1@example.com' ],
423 [2, '2@example.com' ],
424 [3, '3@example.com' ],
425 [6, '6@example.com' ],
426 [7, '7@example.com' ],
427 [8, '8@example.com' ],
428 [9, '9@example.com' ],
429 [10, '10@example.com' ],
430 ], 'msgmap as expected' );