1 # Copyright (C) 2015 all contributors <meta@public-inbox.org>
2 # License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
6 eval { require PublicInbox::SearchIdx; };
7 plan skip_all => "Xapian missing for search" if $@;
8 use File::Temp qw/tempdir/;
10 my $tmpdir = tempdir('pi-search-XXXXXX', TMPDIR => 1, CLEANUP => 1);
11 my $git_dir = "$tmpdir/a.git";
12 my ($root_id, $last_id);
14 is(0, system(qw(git init -q --bare), $git_dir), "git init (main)");
15 eval { PublicInbox::Search->new($git_dir) };
16 ok($@, "exception raised on non-existent DB");
19 my $orig = "FOO " x 30;
20 my $summ = PublicInbox::Search::subject_summary($orig);
22 $summ = length($summ);
23 $orig = length($orig);
24 ok($summ < $orig && $summ > 0, "summary shortened ($orig => $summ)");
27 $summ = PublicInbox::Search::subject_summary($orig);
29 $summ = length($summ);
30 $orig = length($orig);
31 ok($summ < $orig && $summ > 0,
32 "summary shortened but not empty: $summ");
35 my $rw = PublicInbox::SearchIdx->new($git_dir, 1);
39 my $ro = PublicInbox::Search->new($git_dir);
41 $rw->{xdb}->commit_transaction if $rw && $rw->{xdb};
42 $rw = PublicInbox::SearchIdx->new($git_dir, 1);
43 $rw->_xdb_acquire->begin_transaction;
47 # git repository perms
48 is(PublicInbox::SearchIdx->_git_config_perm(undef),
49 &PublicInbox::SearchIdx::PERM_GROUP,
50 "undefined permission is group");
51 is(PublicInbox::SearchIdx::_umask_for(
52 PublicInbox::SearchIdx->_git_config_perm('0644')),
53 0022, "644 => umask(0022)");
54 is(PublicInbox::SearchIdx::_umask_for(
55 PublicInbox::SearchIdx->_git_config_perm('0600')),
56 0077, "600 => umask(0077)");
57 is(PublicInbox::SearchIdx::_umask_for(
58 PublicInbox::SearchIdx->_git_config_perm('0640')),
59 0027, "640 => umask(0027)");
60 is(PublicInbox::SearchIdx::_umask_for(
61 PublicInbox::SearchIdx->_git_config_perm('group')),
62 0007, 'group => umask(0007)');
63 is(PublicInbox::SearchIdx::_umask_for(
64 PublicInbox::SearchIdx->_git_config_perm('everybody')),
65 0002, 'everybody => umask(0002)');
66 is(PublicInbox::SearchIdx::_umask_for(
67 PublicInbox::SearchIdx->_git_config_perm('umask')),
68 umask, 'umask => existing umask');
72 my $root = Email::MIME->create(
74 Date => 'Fri, 02 Oct 1993 00:00:00 +0000',
75 Subject => 'Hello world',
76 'Message-ID' => '<root@s>',
77 From => 'John Smith <js@example.com>',
78 To => 'list@example.com',
81 my $last = Email::MIME->create(
83 Date => 'Sat, 02 Oct 2010 00:00:00 +0000',
84 Subject => 'Re: Hello world',
85 'In-Reply-To' => '<root@s>',
86 'Message-ID' => '<last@s>',
87 From => 'John Smith <js@example.com>',
88 To => 'list@example.com',
89 Cc => 'foo@example.com',
91 body => "goodbye forever :<\n");
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 { $_->mid } @{$res->{msgs}});
109 my $found = $ro->lookup_message('<root@s>');
110 ok($found, "message found");
111 is($root_id, $found->{doc_id}, 'doc_id set correctly');
112 $found->ensure_metadata;
113 is($found->mid, 'root@s', 'mid set correctly');
114 ok(int($found->thread_id) > 0, 'thread_id is an integer');
116 my @exp = sort qw(root@s last@s);
117 my $res = $ro->query("path:hello_world");
118 my @res = filter_mids($res);
119 is_deeply(\@res, \@exp, 'got expected results for path: match');
121 foreach my $p (qw(hello hello_ hello_world2 hello_world_)) {
122 $res = $ro->query("path:$p");
123 is($res->{total}, 0, "path variant `$p' does not match");
126 $res = $ro->query('s:(Hello world)');
127 @res = filter_mids($res);
128 is_deeply(\@res, \@exp, 'got expected results for s:() match');
130 $res = $ro->query('s:"Hello world"');
131 @res = filter_mids($res);
132 is_deeply(\@res, \@exp, 'got expected results for s:"" match');
134 $res = $ro->query('s:"Hello world"', {limit => 1});
135 is(scalar @{$res->{msgs}}, 1, "limit works");
136 my $first = $res->{msgs}->[0];
138 $res = $ro->query('s:"Hello world"', {offset => 1});
139 is(scalar @{$res->{msgs}}, 1, "offset works");
140 my $second = $res->{msgs}->[0];
142 isnt($first, $second, "offset returned different result from limit");
148 my $rmid = '<ghost-message@s>';
149 my $reply_to_ghost = Email::MIME->create(
151 Date => 'Sat, 02 Oct 2010 00:00:00 +0000',
152 Subject => 'Re: ghosts',
153 'Message-ID' => '<ghost-reply@s>',
154 'In-Reply-To' => $rmid,
155 From => 'Time Traveler <tt@example.com>',
156 To => 'list@example.com',
161 my $reply_id = $rw->add_message($reply_to_ghost);
162 is($reply_id, int($reply_id), "reply_id is an integer: $reply_id");
164 my $was_ghost = Email::MIME->create(
166 Date => 'Sat, 02 Oct 2010 00:00:01 +0000',
168 'Message-ID' => $rmid,
169 From => 'Laggy Sender <lag@example.com>',
170 To => 'list@example.com',
172 body => "are real\n");
174 my $ghost_id = $rw->add_message($was_ghost);
175 is($ghost_id, int($ghost_id), "ghost_id is an integer: $ghost_id");
176 ok($ghost_id < $reply_id, "ghost vivified from earlier message");
179 # search thread on ghost
185 my $res = $ro->query('ghost');
186 my @exp = sort qw(ghost-message@s ghost-reply@s);
187 my @res = filter_mids($res);
188 is_deeply(\@res, \@exp, 'got expected results for Subject match');
191 $res = $ro->query('goodbye');
192 is($res->{msgs}->[0]->mid, 'last@s', 'got goodbye message body');
199 my $long_mid = 'last' . ('x' x 60). '@s';
201 my $long = Email::MIME->create(
203 Date => 'Sat, 02 Oct 2010 00:00:00 +0000',
204 Subject => 'long message ID',
205 'References' => '<root@s> <last@s>',
206 'In-Reply-To' => '<last@s>',
207 'Message-ID' => "<$long_mid>",
208 From => '"Long I.D." <long-id@example.com>',
209 To => 'list@example.com',
212 my $long_id = $rw->add_message($long);
213 is($long_id, int($long_id), "long_id is an integer: $long_id");
220 my $long_reply_mid = 'reply-to-long@1';
221 my $long_reply = Email::MIME->create(
223 Subject => 'I break references',
224 Date => 'Sat, 02 Oct 2010 00:00:00 +0000',
225 'Message-ID' => "<$long_reply_mid>",
227 # 'References' => '<root@s> <last@s> <'.$long_mid.'>',
228 'In-Reply-To' => "<$long_mid>",
229 From => '"no1 <no1@example.com>',
230 To => 'list@example.com',
232 body => "no References\n");
233 ok($rw->add_message($long_reply) > $long_id, "inserted long reply");
237 my $t = $ro->get_thread('root@s');
238 is($t->{total}, 4, "got all 4 mesages 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
247 $rw->add_message(Email::MIME->create(
249 Date => 'Sat, 02 Oct 2010 00:00:01 +0000',
251 'Message-ID' => '<quote@a>',
252 From => 'Quoter <quoter@example.com>',
253 To => 'list@example.com',
255 body => "> theatre illusions\nfade\n"));
257 $rw->add_message(Email::MIME->create(
259 Date => 'Sat, 02 Oct 2010 00:00:02 +0000',
261 'Message-ID' => '<nquote@a>',
262 From => 'Non-Quoter<non-quoter@example.com>',
263 To => 'list@example.com',
265 body => "theatre\nfade\n"));
266 my $res = $rw->query("theatre");
267 is($res->{total}, 2, "got both matches");
268 is($res->{msgs}->[0]->mid, 'nquote@a', "non-quoted scores higher");
269 is($res->{msgs}->[1]->mid, 'quote@a', "quoted result still returned");
271 $res = $rw->query("illusions");
272 is($res->{total}, 1, "got a match for quoted text");
273 is($res->{msgs}->[0]->mid, 'quote@a',
274 "quoted result returned if nothing else");
277 # circular references
279 my $s = 'foo://'. ('Circle' x 15).'/foo';
280 my $doc_id = $rw->add_message(Email::MIME->create(
281 header => [ Subject => $s ],
283 Date => 'Sat, 02 Oct 2010 00:00:01 +0000',
284 'Message-ID' => '<circle@a>',
285 'References' => '<circle@a>',
286 'In-Reply-To' => '<circle@a>',
287 From => 'Circle <circle@example.com>',
288 To => 'list@example.com',
291 ok($doc_id > 0, "doc_id defined with circular reference");
292 my $smsg = $rw->lookup_message('circle@a');
293 $smsg->ensure_metadata;
294 is($smsg->references, '', "no references created");
295 my $msg = PublicInbox::SearchMsg->load_doc($smsg->{doc});
296 is($s, $msg->mini_mime->header('Subject'), 'long subject not rewritten');
301 my $mbox = 't/utf8.mbox';
302 open(my $fh, '<', $mbox) or die "failed to open mbox: $mbox\n";
306 $str =~ s/\AFrom [^\n]+\n//s;
307 my $mime = Email::MIME->new($str);
308 my $doc_id = $rw->add_message($mime);
309 ok($doc_id > 0, 'message indexed doc_id with UTF-8');
310 my $smsg = $rw->lookup_message('testmessage@example.com');
311 my $msg = PublicInbox::SearchMsg->load_doc($smsg->{doc});
313 # mini_mime technically not valid (I think),
314 # but good enough for displaying HTML:
315 is($mime->header('Subject'), $msg->mini_mime->header('Subject'),
316 'UTF-8 subject preserved');
320 my $res = $ro->query('d:19931002..20101002');
321 ok(scalar @{$res->{msgs}} > 0, 'got results within range');
322 $res = $ro->query('d:20101003..');
323 is(scalar @{$res->{msgs}}, 0, 'nothing after 20101003');
324 $res = $ro->query('d:..19931001');
325 is(scalar @{$res->{msgs}}, 0, 'nothing before 19931001');
328 # names and addresses
330 my $res = $ro->query('t:list@example.com');
331 is(scalar @{$res->{msgs}}, 6, 'searched To: successfully');
332 foreach my $smsg (@{$res->{msgs}}) {
333 like($smsg->to, qr/\blist\@example\.com\b/, 'to appears');
336 $res = $ro->query('tc:list@example.com');
337 is(scalar @{$res->{msgs}}, 6, 'searched To+Cc: successfully');
338 foreach my $smsg (@{$res->{msgs}}) {
339 my $tocc = join("\n", $smsg->to, $smsg->cc);
340 like($tocc, qr/\blist\@example\.com\b/, 'tocc appears');
343 foreach my $pfx ('tcf:', 'c:') {
344 $res = $ro->query($pfx . 'foo@example.com');
345 is(scalar @{$res->{msgs}}, 1,
346 "searched $pfx successfully for Cc:");
347 foreach my $smsg (@{$res->{msgs}}) {
348 like($smsg->cc, qr/\bfoo\@example\.com\b/,
353 foreach my $pfx ('', 'tcf:', 'f:') {
354 $res = $ro->query($pfx . 'Laggy');
355 is(scalar @{$res->{msgs}}, 1,
356 "searched $pfx successfully for From:");
357 foreach my $smsg (@{$res->{msgs}}) {
358 like($smsg->from, qr/Laggy Sender/,
359 "From appears with $pfx");