1 # Copyright (C) 2018-2021 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 use File::Path qw(remove_tree);
10 local $ENV{HOME} = abs_path('t');
12 # Integration tests for HTTP cloning + mirroring
13 require_mods(qw(Plack::Util Plack::Builder
14 HTTP::Date HTTP::Status Search::Xapian DBD::SQLite));
17 use_ok 'PublicInbox::V2Writable';
18 use PublicInbox::InboxWritable;
20 use PublicInbox::Config;
21 # FIXME: too much setup
22 my ($tmpdir, $for_destroy) = tmpdir();
23 my $pi_config = "$tmpdir/config";
25 open my $fh, '>', $pi_config or die "open($pi_config): $!";
26 print $fh <<"" or die "print $pi_config: $!";
29 address = test\@example.com
31 close $fh or die "close($pi_config): $!";
33 local $ENV{PI_CONFIG} = $pi_config;
35 my $cfg = PublicInbox::Config->new($pi_config);
36 my $ibx = $cfg->lookup('test@example.com');
37 ok($ibx, 'inbox found');
39 $ibx->{-no_fsync} = 1;
40 my $v2w = PublicInbox::V2Writable->new($ibx, 1);
41 ok $v2w, 'v2w loaded';
43 my $mime = PublicInbox::Eml->new(<<'');
44 From: Me <me@example.com>
45 To: You <you@example.com>
47 Date: Thu, 01 Jan 1970 00:00:00 +0000
49 my $old_rotate_bytes = $v2w->{rotate_bytes};
50 $v2w->{rotate_bytes} = 500; # force rotating
52 $mime->header_set('Message-ID', "<$i\@example.com>");
53 $mime->header_set('Subject', "subject = $i");
54 ok($v2w->add($mime), "add msg $i OK");
57 my $epoch_max = $v2w->{epoch_max};
58 ok($epoch_max > 0, "multiple epochs");
61 my $smsg = $ibx->over->get_art(1);
62 like($smsg->{lines}, qr/\A[0-9]+\z/, 'lines is a digit');
63 like($smsg->{bytes}, qr/\A[0-9]+\z/, 'bytes is a digit');
67 my $sock = tcp_server();
68 my @cmd = ('-httpd', '-W0', "--stdout=$tmpdir/out", "--stderr=$tmpdir/err");
69 my $td = start_script(\@cmd, undef, { 3 => $sock });
70 my ($host, $port) = tcp_host_port($sock);
73 @cmd = (qw(-clone -q), "http://$host:$port/v2/", "$tmpdir/m");
74 run_script(\@cmd) or xbail '-clone';
76 for my $i (0..$epoch_max) {
77 ok(-d "$tmpdir/m/git/$i.git", "epoch $i cloned");
80 @cmd = ("-init", '-j1', '-V2', 'm', "$tmpdir/m", 'http://example.com/m',
82 ok(run_script(\@cmd), 'initialized public-inbox -V2');
83 my @shards = glob("$tmpdir/m/xap*/?");
84 is(scalar(@shards), 1, 'got a single shard on init');
86 ok(run_script([qw(-index -j0), "$tmpdir/m"]), 'indexed');
88 my $mibx = { inboxdir => "$tmpdir/m", address => 'alt@example.com' };
89 $mibx = PublicInbox::Inbox->new($mibx);
90 is_deeply([$mibx->mm->minmax], [$ibx->mm->minmax], 'index synched minmax');
93 $mime->header_set('Message-ID', "<$i\@example.com>");
94 $mime->header_set('Subject', "subject = $i");
95 ok($v2w->add($mime), "add msg $i OK");
101 my $fetch_each_epoch = sub {
102 my %before = map { $_ => 1 } glob("$tmpdir/m/git/*");
103 run_script([qw(-fetch -q)], undef, {-C => "$tmpdir/m"}) or
105 my @after = grep { !$before{$_} } glob("$tmpdir/m/git/*");
106 push @new_epochs, @after;
109 $fetch_each_epoch->();
111 my $mset = $mibx->search->reopen->mset('m:15@example.com');
112 is(scalar($mset->items), 0, 'new message not found in mirror, yet');
113 ok(run_script([qw(-index -j0), "$tmpdir/m"]), 'index updated');
114 is_deeply([$mibx->mm->minmax], [$ibx->mm->minmax], 'index synched minmax');
115 $mset = $mibx->search->reopen->mset('m:15@example.com');
116 is(scalar($mset->items), 1, 'found message in mirror');
119 $mime->header_set('Message-ID', '<10@example.com>');
120 $mime->header_set('Subject', 'subject = 10');
123 local $SIG{__WARN__} = sub { push @warn, @_ };
124 ok($v2w->purge($mime), 'purge a message');
125 my $warn = join('', @warn);
126 like($warn, qr/purge rewriting/);
127 my @subj = ($warn =~ m/^# subject .*$/mg);
128 is_deeply(\@subj, ["# subject = 10"], "only rewrote one");
133 my $msgs = $mibx->over->get_thread('10@example.com');
134 my $to_purge = $msgs->[0]->{blob};
135 like($to_purge, qr/\A[a-f0-9]{40,}\z/, 'read blob to be purged');
136 $mset = $ibx->search->reopen->mset('m:10@example.com');
137 is(scalar($mset->items), 0, 'purged message gone from origin');
139 $fetch_each_epoch->();
142 PublicInbox::InboxWritable::cleanup($mibx);
144 my $cmd = [ qw(-index --prune -j0), "$tmpdir/m" ];
145 my ($out, $err) = ('', '');
146 my $opt = { 1 => \$out, 2 => \$err };
147 ok(run_script($cmd, undef, $opt), '-index --prune');
148 like($err, qr/discontiguous range/, 'warned about discontiguous range');
149 unlike($err, qr/fatal/, 'no scary fatal error shown');
152 $mset = $mibx->search->reopen->mset('m:10@example.com');
153 is(scalar($mset->items), 0, 'purged message not found in mirror');
154 is_deeply([$mibx->mm->minmax], [$ibx->mm->minmax], 'minmax still synced');
155 for my $i ((1..9),(11..15)) {
156 $mset = $mibx->search->mset("m:$i\@example.com");
157 is(scalar($mset->items), 1, "$i\@example.com remains visible");
159 is($mibx->git->check($to_purge), undef, 'unindex+prune successful in mirror');
163 local $SIG{__WARN__} = sub { push @warn, @_ };
165 is_deeply(\@warn, [], 'no warnings from index_sync after purge');
168 # deletes happen in a different fetch window
170 $mset = $mibx->search->reopen->mset('m:1@example.com');
171 is(scalar($mset->items), 1, '1@example.com visible in mirror');
172 $mime->header_set('Message-ID', '<1@example.com>');
173 $mime->header_set('Subject', 'subject = 1');
174 ok($v2w->remove($mime), 'removed <1@example.com> from source');
177 $fetch_each_epoch->();
178 PublicInbox::InboxWritable::cleanup($mibx);
180 my $cmd = [ qw(-index -j0), "$tmpdir/m" ];
181 my ($out, $err) = ('', '');
182 my $opt = { 1 => \$out, 2 => \$err };
183 ok(run_script($cmd, undef, $opt), 'index ran');
184 is($err, '', 'no errors reported by index');
185 $mset = $mibx->search->reopen->mset('m:1@example.com');
186 is(scalar($mset->items), 0, '1@example.com no longer visible in mirror');
189 if ('sequential-shard') {
190 $mset = $mibx->search->mset('m:15@example.com');
191 is(scalar($mset->items), 1, 'large message not indexed');
192 remove_tree(glob("$tmpdir/m/xap*"), glob("$tmpdir/m/msgmap.*"));
193 my $cmd = [ qw(-index -j9 --sequential-shard), "$tmpdir/m" ];
194 ok(run_script($cmd), '--sequential-shard works');
195 my @shards = glob("$tmpdir/m/xap*/?");
196 is(scalar(@shards), 8, 'got expected shard count');
197 PublicInbox::InboxWritable::cleanup($mibx);
198 $mset = $mibx->search->mset('m:15@example.com');
199 is(scalar($mset->items), 1, 'search works after --sequential-shard');
203 $mime->header_set('Message-ID', '<2big@a>');
205 $mime->body_str_set("z\n" x 1024);
206 ok($v2w->add($mime), "add big message");
209 $fetch_each_epoch->();
210 PublicInbox::InboxWritable::cleanup($mibx);
211 my $cmd = [qw(-index -j0), "$tmpdir/m", "--max-size=$max" ];
212 my $opt = { 2 => \(my $err) };
213 ok(run_script($cmd, undef, $opt), 'indexed with --max-size');
214 like($err, qr/skipping [a-f0-9]{40,}/, 'warned about skipping message');
215 $mset = $mibx->search->reopen->mset('m:2big@a');
216 is(scalar($mset->items), 0, 'large message not indexed');
219 open my $fh, '>>', $pi_config or die;
220 print $fh <<EOF or die;
226 $cmd = [ qw(-index -j0 --reindex), "$tmpdir/m" ];
227 ok(run_script($cmd, undef, $opt), 'reindexed w/ indexMaxSize in file');
228 like($err, qr/skipping [a-f0-9]{40,}/, 'warned about skipping message');
229 $mset = $mibx->search->reopen->mset('m:2big@a');
230 is(scalar($mset->items), 0, 'large message not re-indexed');
232 ok(scalar(@new_epochs), 'new epochs were created and fetched');
234 ok($td->kill, 'killed httpd');